1 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8 to improve monte carlo in scimark.
10 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
13 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
14 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
15 pentium4_cost, nocona_cost): Set reassociation width to 1.
16 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
17 width to 2 for fp operations and 1 otherwise.
18 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
20 (atom_cost): Set reassociation width to 2.
21 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
22 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
23 (core_cost): Set fp reassociation width to 4 and vector to 2.
24 (ix86_reassociation_width): Rewrite using cost table; special case
25 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
26 and TARGET_AVX128_OPTIMAL.
27 * i386.h (processor_costs): Add
28 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
29 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
30 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
31 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
32 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
33 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
35 2017-10-05 Nathan Sidwell <nathan@acm.org>
37 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
39 2017-10-05 Tamar Christina <tamar.christina@arm.com>
41 * config/arm/arm.c (arm_test_fpu_data): New.
42 (arm_run_selftests): Call arm_test_fpu_data.
44 2017-10-04 Nathan Sidwell <nathan@acm.org>
46 * toplev.c (toplev::main): Remove excess parens on pretty_printer
48 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
50 2017-10-04 Sudakshina Das <sudi.das@arm.com>
52 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
53 check type for aarch64_simd_valid_immediate.
54 (aarch64_output_simd_mov_immediate): Update prototype.
55 (aarch64_simd_valid_immediate): Update prototype.
56 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
57 support for ORR-immediate.
58 (and<mode>3): modified pattern to add support for BIC-immediate.
59 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
60 now checks for valid immediate for BIC and ORR based on new enum
62 (aarch64_output_simd_mov_immediate): Function now used to output
63 BIC/ORR imm as well based on new enum argument.
64 * config/aarch64/constraints.md (Do): New vector immediate constraint.
66 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
67 (aarch64_reg_or_bic_imm): Likewise.
69 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
71 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
72 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
74 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
78 PR rtl-optimization/82396
79 * haifa-sched.c (autopref_multipass_init): Simplify
81 (autopref_rank_data): Simplify sort order.
82 * sched-int.h (autopref_multipass_data_): Remove
83 multi_mem_insn_p, min_offset and max_offset.
85 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
87 * doc/sourcebuild.texi: Document vect_peeling_profitable.
89 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
91 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
94 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
96 * doc/sourcebuild.texi: Document vect_long_mult.
98 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
100 PR tree-optimization/82413
101 * fold-const.c (build_range_check): Use widest_int when comparing
102 the maximum ETYPE value with HIGH.
104 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
106 PR rtl-optimization/82396
107 * haifa-sched.c (autopref_multipass_init): Simplify
109 (autopref_rank_data): Simplify sort order.
110 * sched-int.h (autopref_multipass_data_): Remove
111 multi_mem_insn_p, min_offset and max_offset.
113 2017-10-04 Jakub Jelinek <jakub@redhat.com>
115 PR tree-optimization/82381
116 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
117 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
120 PR tree-optimization/82374
121 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
122 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
123 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
124 current_function_decl to the new decl.
126 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
128 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
129 helper macro for IEEE float128 hardware built-in functions.
130 (SQRTF128_ODD): Add built-in functions with the round-to-odd
132 (TRUNCF128_ODD): Likewise.
133 (ADDF128_ODD): Likewise.
134 (SUBF128_ODD): Likewise.
135 (MULF128_ODD): Likewise.
136 (DIVF128_ODD): Likewise.
137 (FMAF128_ODD): Likewise.
138 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
139 UNSPEC_TRUNC_ROUND_TO_ODD.
140 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
141 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
142 floating point round to odd instructions.
143 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
144 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
145 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
146 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
147 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
148 (trunc<mode>sf2_hw): Change the truncate with round to odd
149 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
150 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
151 to odd hardware instructions.
152 (sub<mode>3_odd): Likewise.
153 (mul<mode>3_odd): Likewise.
154 (div<mode>3_odd): Likewise.
155 (sqrt<mode>2_odd): Likewise.
156 (fma<mode>4_odd): Likewise.
157 (fms<mode>4_odd): Likewise.
158 (nfma<mode>4_odd): Likewise.
159 (nfms<mode>4_odd): Likewise.
160 (trunc<mode>df2_odd): Change the truncate with round to odd
161 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
163 * doc/extend.texi (PowerPC built-in functions): Update documentation
164 for existing IEEE float128-bit built-in functions. Add built-in
165 functions that generate the IEEE 128-bit floating point round to
168 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
170 PR rtl-optimization/77729
171 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
172 to (X&(C1&~C2))|C2 transformations.
174 2017-10-03 Martin Jambor <mjambor@suse.cz>
176 PR tree-optimization/82363
177 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
178 mismatch, mark lacc written regardless of racc.
180 2017-10-03 Jakub Jelinek <jakub@redhat.com>
182 PR tree-optimization/82381
183 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
184 stmt_to_insert nor wheather SSA_NAMEs are default defs.
185 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
186 fallthrough into reassoc_stmt_dominates_stmt_p.
189 * combine.c (combine_instructions): Don't combine in unreachable
192 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
195 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
196 function to not use the have_cpu variable. Do not set cpu_index,
197 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
199 (rs6000_valid_attribute_p): Remove duplicate initializations of
200 old_optimize and func_optimize.
201 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
202 (rs6000_activate_target_options): Make global.
203 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
206 2017-10-02 Jakub Jelinek <jakub@redhat.com>
208 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
209 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
210 Set *poffset to 0 and *psize and *pmax_size to -1 if
211 *poffset + *psize overflows in HOST_WIDE_INT.
213 PR tree-optimization/82387
214 PR tree-optimization/82388
215 PR tree-optimization/82389
216 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
217 instead of live_bytes non-NULL.
219 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
222 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
225 2017-10-02 Richard Biener <rguenther@suse.de>
227 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
228 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
229 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
231 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
233 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
234 requested precision matches the type's.
235 * calls.c (alloc_max_size): Calculate the new candidate size as
236 a widest_int and use wi::to_widest when comparing it with the
237 current candidate size.
238 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
239 zero rather than integer_zero_node.
240 * match.pd: Check for a no-op conversion before using wi::add
241 rather than after. Use tree_to_uhwi when summing small shift
242 counts into an unsigned int.
244 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
245 Alan Hayward <alan.hayward@arm.com>
246 David Sherwood <david.sherwood@arm.com>
249 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
250 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
251 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
254 2017-10-02 Richard Biener <rguenther@suse.de>
256 PR tree-optimization/82355
257 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
258 a mapping for the enclosing loop but avoid generating one for
260 (copy_bb_and_scalar_dependences): Remove premature codegen
261 error on PHIs in blocks duplicated into multiple places.
262 * graphite-scop-detection.c
263 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
264 in the region use it as loop and nest to analyze the DR in.
265 (try_generate_gimple_bb): Likewise.
266 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
267 (add_loop_constraints): For blocks in a loop not in the region
268 create a dimension with a single iteration.
269 * sese.h (gbb_loop_at_index): Remove assert.
271 2017-10-01 Kevin Buettner <kevinb@redhat.com>
273 * omp-expand.c (adjust_context_scope): New function.
274 (expand_parallel_call): Call adjust_context_scope.
276 2017-10-01 Jeff Law <law@redhat.com>
278 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
279 dom_opt_dom_walker class with direct access to private members.
280 Add comments. Call test_for_singularity.
281 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
282 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
283 m_dummy_cond anymore.
284 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
286 (pass_dominator:execute): Build the dummy_cond here and pass it
287 to the dom_opt_dom_walker ctor.
288 (test_for_singularity): New function.
290 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
291 Maya Rashish <coypu@sdf.org>
293 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
294 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
295 (alpha*-*-netbsd*) Use nbsd_tm_file.
296 (arm*-*-netbsdelf*) Likewise.
297 (i[34567]86-*-netbsdelf*) Likewise.
298 (x86_64-*-netbsd*) Likewise.
299 (mips*-*-netbsd*) Likewise.
300 (powerpc-*-netbsd*) Likewise.
301 (sh*-*-netbsd*) Likewise.
302 (sparc-*-netbsdelf*) Likewise.
303 (sparc64-*-netbsd*) Likewise.
304 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
306 (vax-*-netbsdelf*) Likewise.
307 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
308 (UINT_FAST8_TYPE) Likewise.
309 (INT_FAST16_TYPE) Check CHAR_FAST16.
310 (UINT_FAST16_TYPE) Likewise.
312 2017-09-30 Jakub Jelinek <jakub@redhat.com>
315 * config/i386/i386.md
316 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
317 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
318 *divmodsi4_zext_2): New define_insn_and_split.
319 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
320 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
321 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
322 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
323 New define_insn_and_split.
324 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
325 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
326 operands[1] having DImode when mode is SImode.
328 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
329 always SImode for DIV and MOD in REG_EQUAL notes.
331 2017-09-29 Yury Gribov <tetra2005@gmail.com>
334 * match.pd: Fix handling of NaNs in pattern.
336 2017-09-29 Jeff Law <law@redhat.com>
338 * sbitmap.c (bitmap_bit_in_range_p): New function.
339 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
340 * tree-ssa-dse.c (live_bytes_read): New function.
341 (dse_classify_store): Ignore reads of dead bytes.
343 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
344 typos and whitespace errors.
345 * config/i386/predicates.md (address_no_seg_operand): Likewise.
346 * config/s390/s390.c (s390_emit_prologue): Likewise.
348 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
351 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
352 with a symbol for LRA.
354 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
356 PR rtl-optimization/82338
357 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
359 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
361 * genmodes.c (calc_wider_mode): Suppress qsort macro.
362 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
363 (qsort_chk): Declare.
364 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
365 (qsort_chk): New function.
367 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
369 PR tree-optimization/82337
370 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
371 phi definition if the PHI result appears in an abnormal PHI.
372 (find_basis_for_base_expr): Don't record a basis if the LHS of the
373 basis appears in an abnormal PHI.
375 2017-09-29 Richard Biener <rguenther@suse.de>
377 * graphite-isl-ast-to-gimple.c
378 (translate_isl_ast_to_gimple::set_codegen_error): New function.
379 (binary_op_to_tree): Use it.
380 (get_rename_from_scev): Likewise.
381 (copy_loop_phi_nodes): Likewise.
382 (copy_bb_and_scalar_dependences): Likewise.
383 (translate_pending_phi_nodes): Likewise.
385 2017-09-29 Jakub Jelinek <jakub@redhat.com>
388 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
389 for movabsq $(i32 << shift), r64.
391 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
393 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
394 index when encoding %esp as %rsp to avoid 0x67 prefix.
396 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
398 * config/i386/i386.md (*movsf_internal, *movdf_internal):
399 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
401 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
403 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
404 Extensions with more than 16 double VFP registers.
405 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
406 to_clear_mask and all code related to it. Replace the remaining
407 entry by a sbitmap and adapt code accordingly.
409 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
411 * brig-builtins.def: Change pure attributes to const.
413 2017-09-28 Joseph Myers <joseph@codesourcery.com>
415 * config.gcc (default_gnu_indirect_function): Default to yes for
416 sparc*-*-linux* with glibc.
418 2017-09-28 Joseph Myers <joseph@codesourcery.com>
420 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
421 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
422 when creating .init_array and .fini_array sections with priority
425 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
428 * config/aarch64/aarch64.c
429 (aarch64_builtin_support_vector_misalignment): Always return false
430 when misalignment is unknown.
432 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
434 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
435 this function to return false if the definition used by the swap
436 instruction is artificial, or if the memory address from which the
437 constant value is loaded is not represented by a base address held
438 in a register or if the base address register is a frame or stack
439 pointer. Additionally, return false if the base address of the
440 loaded constant is a SYMBOL_REF but is not considered to be a
442 (replace_swapped_load_constant): New function.
443 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
444 loaded constant vector with a load of a swapped constant vector.
446 2017-09-27 Carl Love <cel@us.ibm.com>
448 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
449 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
450 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
453 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
455 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
456 first, always call autopref_rank_data otherwise.
458 2017-09-27 Richard Biener <rguenther@suse.de>
460 * graphite-scop-detection.c (find_scop_parameters): Move
461 loop bound handling ...
462 (gather_bbs::before_dom_children): ... here, avoiding the need
463 to build scop_info->loop_nest.
464 (record_loop_in_sese): Remove.
465 * sese.h (sese_info_t::loop_nest): Remove.
466 * sese.c (new_sese_info): Do not allocate loop_nest.
467 (free_sese_info): Do not free loop_nest.
469 2017-09-27 Jakub Jelinek <jakub@redhat.com>
472 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
473 lhs from calls if the lhs has addressable type.
475 2017-09-27 Richard Biener <rguenther@suse.de>
477 * graphite.h (scop::max_alias_set): New member.
478 * graphite-scop-detection.c: Remove references to non-existing
480 (build_alias_sets): Record the maximum alias set used for drs.
481 (build_scops): Support zero as unlimited for
482 --param graphite-max-arrays-per-scop.
483 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
485 (build_poly_sr_1): ... here. Compute alias set based on the
486 maximum alias set used for drs rather than
487 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
489 2017-09-27 Richard Biener <rguenther@suse.de>
491 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
492 --param loop-block-tile-size=0 to disable tiling.
494 2017-09-27 Richard Biener <rguenther@suse.de>
496 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
497 (graphite-max-nb-scop-params): Document special value zero.
498 * domwalk.h (dom_walker::STOP): New symbolical constant.
499 (dom_walker::dom_walker): Add optional parameter for bb to
501 (dom_walker::~dom_walker): Declare.
502 (dom_walker::before_dom_children): Document STOP return value.
503 (dom_walker::m_user_bb_to_rpo): New member.
504 (dom_walker::m_bb_to_rpo): Likewise.
505 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
506 mapping here if not provided by the user.
507 (dom_walker::~dom_walker): Free bb to RPO mapping if not
508 provided by the user.
509 (dom_walker::STOP): Define.
510 (dom_walker::walk): Do not compute bb to RPO mapping here.
511 Support STOP return value from before_dom_children to stop
513 * graphite-optimize-isl.c (optimize_isl): If the schedule
514 is the same still generate code if -fgraphite-identity
515 or -floop-parallelize-all are given.
516 * graphite-scop-detection.c: Include cfganal.h.
517 (gather_bbs::gather_bbs): Get and pass through bb to RPO
519 (gather_bbs::before_dom_children): Return STOP for BBs
521 (build_scops): Compute bb to RPO mapping and pass it to
522 the domwalk. Treat --param graphite-max-nb-scop-params=0
523 as not limiting the number of params.
524 * graphite.c (graphite_initialize): Remove limit on the
525 number of basic-blocks in a function.
526 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
527 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
530 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
532 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
533 Adjust code to eliminate needing to do the shift right 32-bits
534 operation after XSCVDPSPN.
536 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
538 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
539 ((X / Y) != 0 -> X >= Y): Likewise.
541 2017-09-26 Carl Love <cel@us.ibm.com>
543 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
544 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
545 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
546 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
547 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
548 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
549 definitions and overloading.
550 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
551 statement for P9V_BUILTIN_XST_LEN_R.
552 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
553 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
554 define_expand and define_insn for the instructions and builtins.
555 * doc/extend.texi: Update the built-in documentation file for the new
557 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
558 define_insn for the instructions
560 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
563 * gcc/config/netbsd-protos.h: New file.
564 * gcc/config/netbsd.c: New file.
565 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
566 * gcc/config/t-netbsd: New file.
567 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
568 (tmake_file) Add t-netbsd.
569 (extra_objs) Add netbsd.o.
571 2017-09-26 Janus Weil <janus@gcc.gnu.org>
575 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
577 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
579 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
580 sign extension from a vector register to a GPR by doing a 32-bit
581 direct move and then an EXTSW.
582 (extendsi<mode>2 splitter): Likewise.
583 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
584 right or vector extract after doing XSCVDPSPN. Use
585 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
587 (movdi_from_sf_zero_ext): Likewise.
588 (reload_gpr_from_vsxsf): Likewise.
589 (p8_mfvsrd_4_disf): Delete, no longer used.
590 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
591 then needing to move the SFmode to a GPR to use the XSCVDPSP
592 instruction instead of FRSP and XSCVDPSPN.
593 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
594 it is adjacent to the other XSCVSPDP insns.
595 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
596 SFmode to be in traditional Altivec registers.
597 (vsx_xscvdpspn): Eliminate useless alternative constraint.
598 (vsx_xscvspdpn): Likewise.
599 (vsx_xscvspdpn_scalar): Likewise.
601 2017-09-26 Martin Jambor <mjambor@suse.cz>
603 * tree-sra.c (compare_access_positions): Put integral types first,
604 stabilize sorting of integral types, remove conditions putting
605 non-full-precision integers last.
606 (sort_and_splice_var_accesses): Disable scalarization if a
607 non-integert would be represented by a non-full-precision integer.
609 2017-09-26 Joseph Myers <joseph@codesourcery.com>
611 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
612 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
613 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
614 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
615 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
616 conditionals inside the function instead of around it. Call
617 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
618 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
620 2017-09-26 Richard Biener <rguenther@suse.de>
622 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
624 (scop_detection::build_scop_breadth): ... this. Removed.
625 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
626 (scop_detection::harmful_stmt_in_bb): Likewise.
627 (scop_detection::graphite_can_represent_stmt): Likewise.
628 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
629 (scop_detection::can_represent_loop): Remove recursion, fold in ...
630 (scop_detection::can_represent_loop_1): ... this. Removed.
631 (scop_detection::harmful_loop_in_region): Simplify after inlining
632 the above and remove more quadraticness.
633 (build_scops): Adjust.
634 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
637 2017-09-26 Jakub Jelinek <jakub@redhat.com>
640 * config/i386/i386.c (ix86_print_operand_address_as): Only test
641 REGNO (base) == SP_REG if base is a REG.
644 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
645 if it is different SSA_NAME.
646 (optimize_range_tests_cmp_bitwise): New function.
647 (optimize_range_tests): Call it.
649 2017-09-26 Richard Biener <rguenther@suse.de>
651 PR tree-optimization/82321
652 * graphite.c (canonicalize_loop_closed_ssa): Properly check
653 for the def being inside the loop.
655 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
657 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
659 * config/s390/s390-builtins.def: Fix constraint on op4.
661 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
663 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
664 independent expanders.
665 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
666 ("vec_ordered", "vec_unordered"): New expanders.
668 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
670 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
673 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
675 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
676 vec_unpacks_lo_v16qi.
677 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
679 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
681 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
682 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
683 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
685 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
687 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
689 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
691 ("*vec_slb<mode>"): New insn pattern.
692 ("vec_shr_<mode>"): New expander.
693 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
694 and force the shift count operand to V16QImode.
695 ("vec_srb<mode>"): Set shift count mode to V16QI.
697 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
699 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
700 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
701 ("vec_widen_smult_hi_<mode>"): New expander definitions.
703 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
706 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
708 2017-09-26 Richard Biener <rguenther@suse.de>
710 PR tree-optimization/82320
711 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
714 2017-09-25 Jeff Law <law@redhat.com>
716 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
717 prototype for new argument.
718 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
719 mostly extracted from rs6000_emit_allocate_stack.
720 (rs6000_emit_probe_stack_range_stack_clash): New function.
721 (rs6000_emit_allocate_stack): Call
722 rs6000_emit_probe_stack_range_stack_clash as needed.
723 (rs6000_emit_probe_stack_range): Add additional argument
724 to call to gen_probe_stack_range{si,di}.
725 (output_probe_stack_range): New.
726 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
727 (output_probe_stack_range_stack_clash): New.
728 (rs6000_emit_prologue): Emit notes into dump file as requested.
729 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
730 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
731 Add additional operand and pass it to output_probe_stack_range.
733 2017-09-25 Bin Cheng <bin.cheng@arm.com>
735 PR tree-optimization/82163
736 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
737 (checking_verify_loop_closed_ssa): New parameter.
738 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
739 (check_loop_closed_ssa_stmt): Delete.
740 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
741 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
742 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
745 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
747 * brig-builtins.def: Treat HSAIL barrier builtins as
748 setjmp/longjump style functions.
750 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
752 * target.def (constant_alignment): New hook.
753 * defaults.h (CONSTANT_ALIGNMENT): Delete.
754 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
755 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
756 * doc/tm.texi: Regenerate.
757 * targhooks.h (default_constant_alignment): Declare.
758 (constant_alignment_word_strings): Likewise.
759 * targhooks.c (default_constant_alignment): New function.
760 (constant_alignment_word_strings): Likewise.
761 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
762 instead of CONSTANT_ALIGNMENT.
763 * varasm.c (align_variable, get_variable_align, build_constant_desc)
764 (force_const_mem): Likewise.
765 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
766 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
767 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
768 (TARGET_CONSTANT_ALIGNMENT): Redefine.
769 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
771 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
772 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
773 constant_alignment_word_strings.
774 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
775 (CONSTANT_ALIGNMENT): Likewise.
776 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
777 (arm_constant_alignment): New function.
778 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
779 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
780 constant_alignment_word_strings.
781 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
782 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
783 constant_alignment_word_strings.
784 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
785 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
786 (cris_constant_alignment): New function.
787 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
788 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
789 (epiphany_constant_alignment): New function.
790 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
791 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
792 constant_alignment_word_strings.
793 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
794 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
795 constant_alignment_word_strings.
796 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
797 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
798 constant_alignment_word_strings.
799 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
800 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
801 * config/i386/i386.c (ix86_constant_alignment): Make static.
802 Use the same interface as the target hook.
803 (TARGET_CONSTANT_ALIGNMENT): Redefine.
804 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
805 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
806 constant_alignment_word_strings.
807 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
808 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
809 (TARGET_CONSTANT_ALIGNMENT): Redefine.
810 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
811 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
812 constant_alignment_word_strings.
813 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
814 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
815 constant_alignment_word_strings.
816 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
817 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
818 constant_alignment_word_strings.
819 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
820 * config/microblaze/microblaze.c (microblaze_constant_alignment):
822 (TARGET_CONSTANT_ALIGNMENT): Redefine.
823 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
824 * config/mips/mips.c (mips_constant_alignment): New function.
825 (TARGET_CONSTANT_ALIGNMENT): Redefine.
826 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
827 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
828 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
829 (mmix_constant_alignment): Make static. Use the same interface
831 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
832 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
833 constant_alignment_word_strings.
834 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
835 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
836 constant_alignment_word_strings.
837 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
838 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
839 constant_alignment_word_strings.
840 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
841 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
842 (rs6000_constant_alignment): New function.
843 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
844 * config/riscv/riscv.c (riscv_constant_alignment): New function.
845 (TARGET_CONSTANT_ALIGNMENT): Redefine.
846 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
847 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
848 (rs6000_constant_alignment): New function.
849 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
850 * config/s390/s390.c (s390_constant_alignment): New function.
851 (TARGET_CONSTANT_ALIGNMENT): Redefine.
852 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
853 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
854 constant_alignment_word_strings.
855 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
856 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
857 (sparc_constant_alignment): New function.
858 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
859 * config/spu/spu.c (spu_constant_alignment): New function.
860 (TARGET_CONSTANT_ALIGNMENT): Redefine.
861 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
862 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
863 constant_alignment_word_strings.
864 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
865 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
866 constant_alignment_word_strings.
867 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
868 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
869 constant_alignment_word_strings.
870 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
871 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
872 (visium_constant_alignment): New function.
873 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
874 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
875 (xtensa_constant_alignment): New function.
876 * system.h (CONSTANT_ALIGNMENT): Poison.
878 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
880 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
881 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
882 (rs6000_builtin_valid_without_lhs): New helper function.
883 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
884 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
886 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
888 * target.h (vec_perm_indices): Use unsigned short rather than
890 (auto_vec_perm_indices): Likewise.
891 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
892 Use unsigned int rather than unsigned char.
893 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
895 2017-09-25 Richard Biener <rguenther@suse.de>
897 * cfgloop.h (sort_sibling_loops): Declare.
898 * cfgloop.c (sort_sibling_loops_cmp): New helper.
899 (sort_sibling_loops): New function sorting the sibling loop list
901 * graphite.c (graphite_transform_loops): Sort sibling loops.
903 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
905 * target.def (vec_perm_const_ok): Change sel parameter to
907 * optabs-query.c (can_vec_perm_p): Update accordingly.
908 * doc/tm.texi: Regenerate.
909 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
910 auto_vec_perm_indices and remove separate nelt field.
911 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
912 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
913 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
914 (aarch64_expand_vec_perm_const): Update accordingly.
915 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
917 * config/arm/arm.c (expand_vec_perm_d): Change perm to
918 auto_vec_perm_indices and remove separate nelt field.
919 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
920 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
921 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
923 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
925 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
926 sel to vec_perm_indices.
927 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
928 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
929 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
931 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
934 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
937 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
938 on the FUNCTION_DECL function context if it has a DIE that is a
941 2017-09-25 Richard Biener <rguenther@suse.de>
943 PR tree-optimization/82285
944 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
947 2017-09-25 Tom de Vries <tom@codesourcery.com>
951 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
954 2017-09-25 Richard Biener <rguenther@suse.de>
956 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
957 ISL errors other than isl_error_quota happen. Dump if the
958 schedule is the same.
959 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
960 errors instead of aborting inside ISL.
962 2017-09-25 Iain Sandoe <iain@codesourcery.com>
965 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
966 of libgcc_eh for m64.
967 * config/i386/darwin64.h: Likewise.
969 2017-09-25 Richard Biener <rguenther@suse.de>
972 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
973 attribute for incomplete types nor twice for complete ones.
975 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
978 * config/i386/i386.c (ix86_print_operand_address_as): Encode
979 %esp as %rsp to avoid 0x67 prefix if there is no index or base
982 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
985 * config/i386/i386.opt (mprefer-avx256): Use
986 ix86_target_flags variable.
987 * config/i386/i386.c (ix86_target_string): Move
988 -mprefer-avx256 to flag2_opts.
990 2017-09-22 Jakub Jelinek <jakub@redhat.com>
993 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
994 and x != -1 | y != -1 into (x & y) != -1.
996 2017-09-22 Steve Ellcey <sellcey@cavium.com>
998 * config.gcc: Add new case statement to set
999 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1000 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1001 s390x-*-linux* case statements. Added aarch64 to the list of
1002 supported architectures.
1004 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1006 PR tree-optimization/82289
1007 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1008 STMT_VINFO_RELEVANT_P.
1010 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1011 Alan Hayward <alan.hayward@arm.com>
1012 David Sherwood <david.sherwood@arm.com>
1014 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1015 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1016 (extract_range_from_binary_expr_1): Don't call
1017 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1019 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1020 Alan Hayward <alan.hayward@arm.com>
1021 David Sherwood <david.sherwood@arm.com>
1023 * target.def (preferred_vector_alignment): New hook.
1024 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1026 * doc/tm.texi: Regenerate.
1027 * targhooks.h (default_preferred_vector_alignment): Declare.
1028 * targhooks.c (default_preferred_vector_alignment): New function.
1029 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1031 (DR_TARGET_ALIGNMENT): New macro.
1032 (aligned_access_p): Update commentary.
1033 (vect_known_alignment_in_bytes): New function.
1034 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1036 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1037 Calculate the misalignment based on the target alignment rather than
1039 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1040 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1041 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1042 the target alignment, rather than masking the element misalignment
1043 with the number of elements in a vector. Also use the target
1044 alignment when calculating the maximum number of peels.
1045 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1046 instead of TYPE_ALIGN_UNIT.
1047 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1048 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1049 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1050 (vect_create_data_ref_ptr): Likewise.
1051 (vect_setup_realignment): Realign by ANDing with
1052 -DR_TARGET_MISALIGNMENT.
1053 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1054 the number of peels based on DR_TARGET_ALIGNMENT.
1055 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1056 with the guaranteed alignment boundary when deciding whether
1058 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1059 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1060 (ensure_base_align): Remove stmt_info parameter. Get the
1061 target base alignment from DR_TARGET_ALIGNMENT.
1062 (vectorizable_store): Update call accordingly. Interpret
1063 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1065 (vectorizable_load): Likewise.
1067 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1068 Alan Hayward <alan.hayward@arm.com>
1069 David Sherwood <david.sherwood@arm.com>
1071 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1072 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1073 (vect_enhance_data_refs_alignment): Likewise.
1075 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1077 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1078 error. Only quit immediately if parsing is complete.
1079 (BEGIN): Initialize fatal_err and parse_done.
1080 (begin fpu, end fpu): Check number of arguments.
1081 (begin arch, end arch): Likewise.
1082 (begin cpu, end cpu): Likewise.
1083 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1084 (optalias): Likewise.
1086 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1088 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1089 * config/arm/arm-isa.h: Delete. Move definitions to ...
1090 * arm-cpus.in: ... here. Use new feature and fgroup values.
1091 * config/arm/arm.c (arm_option_override): Use lower case for feature
1093 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1094 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1095 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1097 (print_isa_bits_for): New function.
1098 (gen_isa): New function.
1099 (gen_comm_data): Use print_isa_bits_for.
1100 (define feature): New keyword.
1101 (define fgroup): New keyword.
1102 * config/arm/t-arm (TM_H): Remove.
1103 (GTM_H): Add arm-isa.h.
1104 (arm-isa.h): Add rule to generate file.
1105 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1106 case for feature bit names.
1108 2017-09-22 Richard Biener <rguenther@suse.de>
1110 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1112 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1113 print of no dependency loops ...
1114 * graphite.c (graphite_transform_loops): ... here.
1115 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1117 (same_close_phi_node, remove_duplicate_close_phi,
1118 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1119 (canonicalize_loop_closed_ssa): ... here and simplify.
1120 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1121 (optimize_isl): Use dump_printf_loc to tell when we stopped
1122 optimizing because of an ISL timeout.
1124 2017-09-22 Richard Biener <rguenther@suse.de>
1126 PR tree-optimization/82291
1127 * tree-if-conv.c (predicate_mem_writes): Make sure to
1128 remove writes in blocks predicated with false.
1130 2017-09-22 Richard Biener <rguenther@suse.de>
1132 * sese.c: Include cfganal.h.
1133 (if_region_set_false_region): Remove.
1134 (create_if_region_on_edge): Likewise.
1135 (move_sese_in_condition): Re-implement without destroying
1138 2017-09-22 Richard Biener <rguenther@suse.de>
1140 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1141 Verify both BBs contain loop PHI nodes before dispatching to
1143 (graphite_regenerate_ast_isl): Do not recompute dominators,
1144 do not verify three times. Restructure for clarity.
1145 * graphite-scop-detection.c (same_close_phi_node,
1146 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1147 defined_in_loop_p, canonicalize_loop_closed_ssa,
1148 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1149 checking and SSA rewrite, move to ...
1150 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1151 (graphite_initialize): Do not pass in ctx, do not reset the
1152 SCEV cache, compute only dominators.
1153 (graphite_transform_loops): Allocate ISL ctx after
1154 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1155 Maintain post-dominators only around build_scops.
1156 * sese.c (if_region_set_false_region): Make static. Free
1157 and recompute dominators.
1158 (move_sese_in_condition): Assert we don't get called with
1159 post-dominators computed.
1160 * sese.h (if_region_set_false_region): Remove.
1162 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1164 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1165 mode attribute for TARGET_AVX512VL.
1167 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1169 * config/i386/i386.opt (mprefer-avx256): New option.
1170 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1172 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1173 for TARGET_PREFER_AVX256.
1174 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1176 2017-09-21 Jeff Law <law@redhat.com>
1178 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1179 Fix dump output if the only stack space is for pushed registers.
1181 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1183 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1186 2017-09-21 Martin Sebor <msebor@redhat.com>
1189 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1190 code (in C++) or code that triggers warnings.
1192 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1194 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1196 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1198 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1199 * sched-rgn.c: Ditto.
1200 * sel-sched-ir.c: Ditto.
1202 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1204 * toplev.h (set_random_seed): Adjust return type.
1205 * toplev.c (init_local_tick): Move eager initialization of random_seed
1206 to get_random_seed. Adjust comment.
1207 (init_random_seed): Inline to get_random_seed, delete.
1208 (get_random_seed): Initialize random_seed lazily.
1209 (set_random_seed): Do not return previous value.
1210 (print_switch_value): Do not call get_random_seed.
1212 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1214 * cgraph.c (delete_function_version): New, broken out from...
1215 (cgraph_node::delete_function_version): ...here. Rename to
1216 cgraph_node::delete_function_version_by_decl. Update all uses.
1217 (cgraph_node::remove): Call delete_function_version.
1219 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1222 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1223 VAR_DECLs to which addressable non-volatile parameters are mapped
1224 and for id->retvar after the return value assignment. Clear
1225 id->retval and id->retbnd after inlining.
1227 2017-09-21 Richard Biener <rguenther@suse.de>
1229 PR tree-optimization/82276
1230 PR tree-optimization/82244
1231 * tree-vrp.c (build_assert_expr_for): Set
1232 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1234 (remove_range_assertions): Revert earlier change.
1236 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1239 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1241 2017-09-21 Richard Biener <rguenther@suse.de>
1243 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1244 Restore valid IL after code generation errors.
1245 * graphite.c (graphite_transform_loops): Diagnose code
1246 generation issues as MSG_MISSED_OPTIMIZATION and continue
1247 with processing SCOPs.
1249 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1250 Alan Hayward <alan.hayward@arm.com>
1251 David Sherwood <david.sherwood@arm.com>
1253 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1254 rather than choosing between plus_constant and gen_rtx_<CODE>.
1255 * expr.c (emit_push_insn): Likewise.
1256 (expand_expr_real_2): Likewise.
1258 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1259 Alan Hayward <alan.hayward@arm.com>
1260 David Sherwood <david.sherwood@arm.com>
1262 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1264 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1265 Alan Hayward <alan.hayward@arm.com>
1266 David Sherwood <david.sherwood@arm.com>
1268 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1269 calling tree_to_uhwi.
1271 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1272 Alan Hayward <alan.hayward@arm.com>
1273 David Sherwood <david.sherwood@arm.com>
1275 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1276 INTEGER_CST rather than a negative test for ADDR_EXPR.
1278 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1279 Alan Hayward <alan.hayward@arm.com>
1280 David Sherwood <david.sherwood@arm.com>
1282 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1283 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1285 2017-09-21 Richard Biener <rguenther@suse.de>
1287 PR tree-optimization/71351
1288 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1289 graphite_create_new_loop_guard): Remove, fold remaining parts
1291 (translate_isl_ast_node_for): ... here and simplify.
1293 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1296 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1297 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1298 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1299 alternative always use QI mode, for -Os imov (=R,R) alternative
1300 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1303 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1304 Jeff Law <law@redhat.com>
1306 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1307 (allocate_stack_space): New function, partially extracted from
1309 (s390_emit_prologue): Track offset to most recent stack probe.
1310 Code to allocate space moved into allocate_stack_space.
1311 Dump actions when no stack is allocated.
1312 (s390_prologue_plus_offset): New function.
1313 (s390_emit_stack_probe): Likewise.
1315 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1317 * common.opt (Wa, Wl, Wp, g, gz=): Add
1319 (gno-column-info): Remove.
1320 (gcolumn-info): Drop RejectNegative.
1322 (gno-record-gcc-switches): Remove.
1323 (grecord-gcc-switches): Drop RejectNegative.
1324 (gno-split-dwarf): Remove.
1325 (gsplit-dwarf): Drop RejectNegative.
1326 (gno-strict-dwarf): Remove.
1327 (gstrict-dwarf): Drop RejectNegative.
1328 * config/darwin.opt (gfull, gused): Add RejectNegative.
1329 * dwarf2out.c (gen_producer_string): Drop
1330 gno-record-gcc-switches handler.
1331 * optc-gen.awk: Add g to prefixes with negative forms.
1332 * opts-common.c (remapping_prefix_p): New.
1333 (find_opt): Check it.
1334 (generate_canonical_option): Test g prefix.
1335 (option_map): Add -gno- mapping.
1336 (add_misspelling_candidates): Check remapping_prefix_p.
1338 2017-09-20 Jeff Law <law@redhat.com>
1340 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1341 thinko in stack clash protection support.
1343 * explow.c (compute_stack_clash_protection_loop_data): Use
1344 CONST_INT_P instead of explicit test. Verify object is a
1345 CONST_INT_P before looking at INTVAL.
1346 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1347 instead of explicit test.
1349 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1352 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1353 address instead of to r1 and r11.
1355 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1357 * config.gcc: Support "knm".
1358 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1359 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1361 * config/i386/i386.c (m_KNM): Define.
1362 (processor_target_table): Add "knm".
1364 (ix86_option_override_internal): Add "knm".
1365 (ix86_issue_rate): Add PROCESSOR_KNM.
1366 (ix86_adjust_cost): Ditto.
1367 (ia32_multipass_dfa_lookahead): Ditto.
1368 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1369 (fold_builtin_cpu): Add M_INTEL_KNM.
1370 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1371 (processor_type): Add PROCESSOR_KNM.
1372 * config/i386/x86-tune.def: Add m_KNM.
1373 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1375 2017-09-20 Richard Biener <rguenther@suse.de>
1377 PR tree-optimization/80213
1378 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1379 are allowed in empty BBs as well.
1380 (canonicalize_loop_closed_ssa): Also look for other complex
1382 (scop_detection::get_sese): Include the loop-closed PHI block
1384 (scop_detection::merge_sese): Remove code adding extra blocks.
1385 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1386 (build_scops): Assert the final returned scop is invalid.
1388 2017-09-20 Richard Biener <rguenther@suse.de>
1390 PR tree-optimization/82264
1391 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1393 (vn_phi_lookup): Likewise.
1394 (vn_phi_insert): Likewise.
1396 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1398 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1399 that fits into uhwi or shwi, add DW_AT_const_value regardless
1400 of early_dwarf without going through RTL, using add_AT_unsigned
1403 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1404 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1405 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1406 (DEBUG_MACRO_SECTION): Likewise.
1407 (DEBUG_LTO_MACRO_SECTION): Likewise.
1408 (DEBUG_STR_DWO_SECTION): Likewise.
1409 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1410 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1411 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1412 (DEBUG_LTO_LINE_STR_SECTION): Define.
1413 (init_sections_and_labels): Initialize debug_line_str_section
1414 variable. Initialize debug_loc_section for -gdwarf-5 to
1415 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1417 2017-09-20 Richard Biener <rguenther@suse.de>
1419 * graphite-sese-to-poly.c (extract_affine): Properly handle
1420 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1422 2017-09-20 Richard Biener <rguenther@suse.de>
1424 PR tree-optimization/81373
1425 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1426 Force SESE live-out defs to be handled even if they are
1429 2017-09-19 Jeff Law <law@redhat.com>
1431 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1432 nothing for stack adjustments with REG_STACK_CHECK.
1433 * sched-deps.c (parse_add_or_inc): Reject insns with
1434 REG_STACK_CHECK from dependency breaking.
1435 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1436 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1437 * reg-notes.def (STACK_CHECK): New note.
1439 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1440 (ix86_expand_prologue): Dump stack clash info as needed.
1441 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1443 * function.c (dump_stack_clash_frame_info): New function.
1444 * function.h (dump_stack_clash_frame_info): Prototype.
1445 (enum stack_clash_probes): New enum.
1447 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1448 flag_stack_clash_protection.
1449 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1450 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1451 (arm_frame_pointer_required): Likewise.
1452 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1453 (ia64_expand_prologue): Likewise.
1454 * config/mips/mips.c (mips_expand_prologue): Likewise.
1455 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1456 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1457 (sparc_flat_expand_prologue): Likewise.
1458 * config/spu/spu.c (spu_expand_prologue): Likewise.
1460 * explow.c: Include "params.h".
1461 (anti_adjust_stack_and_probe_stack_clash): New function.
1462 (get_stack_check_protect): Likewise.
1463 (compute_stack_clash_protection_loop_data): Likewise.
1464 (emit_stack_clash_protection_loop_start): Likewise.
1465 (emit_stack_clash_protection_loop_end): Likewise.
1466 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1467 Use anti_adjust_stack_and_probe_stack_clash.
1468 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1469 (emit_stack_clash_protection_loop_start): Likewise.
1470 (emit_stack_clash_protection_loop_end): Likewise.
1471 * rtl.h (get_stack_check_protect): Prototype.
1472 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1473 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1474 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1476 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1478 * doc/tm.texi: Rebuilt.
1479 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1480 get_stack_check_protect.
1481 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1482 * config/arm/arm.c (arm_expand_prologue): Likewise.
1483 (arm_frame_pointer_required): Likewise.
1484 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1485 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1486 * config/mips/mips.c (mips_expand_prologue): Likewise.
1487 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1488 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1489 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1490 (sparc_flat_expand_prologue): Likewise.
1492 * common.opt (-fstack-clash-protection): New option.
1493 * flag-types.h (enum stack_check_type): Note difference between
1494 -fstack-check= and -fstack-clash-protection.
1495 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1496 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1497 * toplev.c (process_options): Issue warnings/errors for cases
1498 not handled with -fstack-clash-protection.
1499 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1500 (-fstack-check): Note additional problem with -fstack-check=generic.
1501 Note that -fstack-check is primarily for Ada and refer users
1502 to -fstack-clash-protection for stack-clash-protection.
1503 Document new params for stack clash protection.
1505 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1507 * config/i386/i386.c (ix86_split_long_move): Do not handle
1508 address used for LEA in a special way.
1510 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1512 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1514 2017-09-19 Martin Sebor <msebor@redhat.com>
1517 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1518 of incompatible types.
1520 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1522 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1523 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1524 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1525 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1527 2017-09-19 Richard Biener <rguenther@suse.de>
1529 PR tree-optimization/82244
1530 * tree-vrp.c (remove_range_assertions): Do not propagate
1531 a constant to abnormals but replace the assert with a copy.
1533 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1535 PR rtl-optimization/57878
1536 PR rtl-optimization/68988
1537 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1538 avoidance test involving non_reload_pseudos. Move frequency test
1539 below the general fragmentation avoidance test.
1541 2017-09-19 Richard Biener <rguenther@suse.de>
1543 PR tree-optimization/69728
1544 * graphite-sese-to-poly.c (schedule_error): New global.
1545 (add_loop_schedule): Handle empty domain by failing the
1547 (build_original_schedule): Handle schedule_error.
1549 2017-09-19 Richard Biener <rguenther@suse.de>
1551 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1552 Do not iterate to sibling loops but only to siblings of inner
1555 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1558 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1559 that the registers are different.
1561 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1563 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1564 to processor_model and "amdfam17h" to arch_names_table.
1565 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1567 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1570 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1572 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1573 Alan Hayward <alan.hayward@arm.com>
1574 David Sherwood <david.sherwood@arm.com>
1576 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1578 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1580 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1581 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1582 vect_schedule_slp_instance.
1583 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1584 Update call to vect_slp_analyze_node_operations. Simplify return
1586 (vect_slp_analyze_bb_1): Update call accordingly.
1587 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1588 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1589 (vect_schedule_slp): Update call accordingly.
1591 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1592 Alan Hayward <alan.hayward@arm.com>
1593 David Sherwood <david.sherwood@arm.com>
1595 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1596 with types that aren't in fact scalar.
1598 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1600 * tree-vect-slp.c (vect_record_max_nunits): New function,
1602 (vect_build_slp_tree_1): ...here.
1603 (vect_build_slp_tree_2): Call it for phis too.
1605 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1607 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1608 to vect_get_vec_def_for_operand when getting the mask operand.
1610 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1611 Alan Hayward <alan.hayward@arm.com>
1612 David Sherwood <david.sherwood@arm.com>
1614 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1617 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1618 Alan Hayward <alan.hayward@arm.com>
1619 David Sherwood <david.sherwood@arm.com>
1621 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1622 calculation for vector booleans.
1624 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1625 Alan Hayward <alan.hayward@arm.com>
1626 David Sherwood <david.sherwood@arm.com>
1628 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1630 (vect_transform_stmt): ...here.
1631 (vect_analyze_stmt): Use it instead of calling
1632 vectorizable_live_operation directly.
1634 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1636 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1637 non-SIMT targets in acc vector loops.
1639 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1641 * configure.ac: Add arc and check if assembler supports gdwarf2.
1642 * configure: Regenerate.
1644 2017-09-18 Richard Biener <rguenther@suse.de>
1646 PR tree-optimization/82220
1647 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1648 epilogue niters from the min_profitable_iters compute.
1650 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1653 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1654 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1655 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1657 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1660 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1661 switching to a new text section.
1663 2017-09-18 Richard Biener <rguenther@suse.de>
1665 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1667 (build_alias_set): Reject aliases with no access function.
1669 2017-09-18 Richard Biener <rguenther@suse.de>
1671 PR tree-optimization/79622
1672 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1674 (build_cross_bb_scalars_use): Likewise.
1676 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1678 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1680 2017-09-18 Alan Modra <amodra@gmail.com>
1683 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1684 stack_pointer_rtx for count 0. Update comments. Break up
1685 large rtl expression.
1687 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1689 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1690 Increase to 20 bytes.
1691 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1692 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1693 or avx version of the stub.
1695 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1698 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1699 compute the minimum stack alignment. Also update preferred stack
1700 boundary for leaf functions.
1702 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1704 PR tree-optimization/82228
1705 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1708 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1710 * common/config/nds32/nds32-common.c
1711 (nds32_option_optimization_table): Refine formatting.
1712 (nds32_option_optimization_table): Use -fsched-pressure and
1713 -fomit-frame-pointer for specific optimization level.
1715 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1717 * config/nds32/nds32.c: Refine formatting and comments.
1718 * config/nds32/nds32.h: Likewise.
1719 * config/nds32/nds32.md: Likewise.
1720 * config/nds32/nds32-cost.c: Likewise.
1721 * config/nds32/nds32-isr.c: Likewise.
1722 * config/nds32/nds32-md-auxiliary.c: Likewise.
1723 * config/nds32/nds32-multiple.md: Likewise.
1724 * config/nds32/nds32-predicates.c: Likewise.
1726 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1727 Jakub Jelinek <jakub@redhat.com>
1729 Add support for -std=c++2a.
1730 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1732 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1734 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1737 * doc/extend.texi (Common Function Attributes): Add
1738 references to ARM, AArch64, and S/390 specific attributes.
1739 (Function Specific Option Pragmas): Add AArch64 and S/390
1740 to list of back ends that support the target pragma.
1742 2017-09-15 Nathan Sidwell <nathan@acm.org>
1744 * doc/standards.texi: Fix C++17 description. Update URLs for
1747 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1749 * common.opt (Wcast-align=strict): New warning option.
1750 * doc/invoke.texi: Document -Wcast-align=strict.
1752 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1754 * cgraph.h (cgraph_thunk_info): Add comments.
1755 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1756 assert for VIRTUAL_* arguments stricter.
1758 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1760 PR tree-optimization/71026
1761 * match.pd: Move RDIV patterns from fold-const.c
1762 * fold-const.c (distribute_real_division): Removed.
1763 (fold_binary_loc): Remove calls to distribute_real_divison.
1765 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1767 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1768 c++1z and gnu++1z as deprecated. Change other references to
1769 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1770 Change -Wc++1z-compat to -Wc++17-compat.
1771 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1772 * dwarf2out.c (highest_c_language): Handle C++17.
1773 (gen_compile_unit_die): Likewise.
1775 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1777 PR rtl-optimization/82192
1778 * combine.c (make_extraction): Don't look through non-paradoxical
1779 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1782 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1783 Alan Hayward <alan.hayward@arm.com>
1784 David Sherwood <david.sherwood@arm.com>
1786 * target.def (function_arg_offset): New hook.
1787 * targhooks.h (default_function_arg_offset): Declare.
1788 * targhooks.c (default_function_arg_offset): New function.
1789 * function.c (locate_and_pad_parm): Use
1790 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1791 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1792 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1793 * doc/tm.texi: Regenerate.
1794 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1795 * config/spu/spu.c (spu_function_arg_offset): New function.
1796 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1797 * system.h (FUNCTION_ARG_OFFSET): Poison.
1799 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayard <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * target.def (truly_noop_truncation): New hook.
1804 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1805 than TRULY_NOOP_TRUNCATION.
1806 * hooks.h (hook_bool_uint_uint_true): Declare.
1807 * hooks.c (hook_bool_uint_uint_true): New function.
1808 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1809 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1810 * doc/tm.texi: Regenerate.
1811 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1812 rather than TRULY_NOOP_TRUNCATION in comments.
1813 (simplify_comparison): Likewise.
1814 (record_truncated_value): Likewise.
1815 * expmed.c (extract_bit_field_1): Likewise.
1816 (extract_split_bit_field): Likewise.
1817 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1818 instead of TRULY_NOOP_TRUNCATION.
1819 * function.c (assign_parm_setup_block): Likewise.
1820 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1821 * rtlhooks.c: Include target.h.
1822 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1823 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1824 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1825 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1826 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1827 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1828 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1829 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1830 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1831 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1832 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1833 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1834 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1835 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1836 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1837 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1838 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1839 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1840 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1841 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1842 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1843 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1844 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1845 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1846 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1847 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1848 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1849 rather than TRULY_NOOP_TRUNCATION in comments.
1850 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1851 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1852 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1853 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1854 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1855 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1856 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1857 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1858 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1859 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1860 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1861 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1862 rather than TRULY_NOOP_TRUNCATION in comments.
1863 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1864 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1865 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1866 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1867 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1868 TRULY_NOOP_TRUNCATION condition.
1869 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1870 (TRULY_NOOP_TRUNCATION): Delete.
1871 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1872 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1873 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1874 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1875 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1876 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1877 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1878 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1879 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1880 rather than TRULY_NOOP_TRUNCATION in comments.
1881 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1882 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1883 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1884 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1885 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1886 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1888 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1891 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1892 (*cmp_ior): Likewise.
1893 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1894 (*ior_scc_scc_cmp): Likewise.
1895 (*and_scc_scc): Likewise.
1896 (*and_scc_scc_cmp): Likewise.
1898 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1899 Alan Hayard <alan.hayward@arm.com>
1900 David Sherwood <david.sherwood@arm.com>
1902 * target.def (can_change_mode_class): New hook.
1903 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1904 (hard_regno_nregs): Likewise.
1905 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1906 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1907 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1908 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1909 (LOAD_EXTEND_OP): Update accordingly.
1910 * doc/tm.texi: Regenerate.
1911 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1912 CANNOT_CHANGE_MODE_CLASS.
1913 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1914 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1915 * combine.c (simplify_set): Update accordingly.
1916 * emit-rtl.c (validate_subreg): Likewise.
1917 * recog.c (general_operand): Likewise.
1918 * regcprop.c (mode_change_ok): Likewise.
1919 * reload1.c (choose_reload_regs): Likewise.
1920 (inherit_piecemeal_p): Likewise.
1921 * rtlanal.c (simplify_subreg_regno): Likewise.
1922 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1923 instead of CANNOT_CHANGE_MODE_CLASS.
1924 (reload_cse_simplify_operands): Likewise.
1925 * reload.c (push_reload): Use targetm.can_change_mode_class
1926 instead of CANNOT_CHANGE_MODE_CLASS.
1927 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1928 REG_CANNOT_CHANGE_MODE_P.
1929 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1930 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1931 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1932 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1933 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1934 (arm_can_change_mode_class): New function.
1935 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1936 than CANNOT_CHANGE_MODE_CLASS in comments.
1937 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1938 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1939 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1940 (ix86_can_change_mode_class): ...this new function, inverting the
1941 sense of the return value.
1942 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1943 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1944 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1945 (ia64_can_change_mode_class): New function.
1946 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1947 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1948 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1949 (m32c_can_change_mode_class): ...this new function, inverting the
1950 sense of the return value.
1951 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1952 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1953 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1954 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1955 (mips_can_change_mode_class): ...this new function, inverting the
1956 sense of the return value.
1957 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1958 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1959 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1960 (msp430_can_change_mode_class): New function.
1961 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1962 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1963 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1964 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1965 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1966 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1967 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1968 (pa_cannot_change_mode_class): Replace with...
1969 (pa_can_change_mode_class): ...this new function, inverting the
1970 sense of the return value.
1971 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1972 than CANNOT_CHANGE_MODE_CLASS in comments.
1973 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1974 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1975 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1976 (pdp11_cannot_change_mode_class): Replace with...
1977 (pdp11_can_change_mode_class): ...this new function, inverting the
1978 sense of the return value.
1979 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1980 * config/powerpcspe/powerpcspe-protos.h
1981 (rs6000_cannot_change_mode_class_ptr): Delete.
1982 * config/powerpcspe/powerpcspe.c
1983 (rs6000_cannot_change_mode_class_ptr): Delete.
1984 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1985 (rs6000_option_override_internal): Assign to
1986 targetm.can_change_mode_class instead of
1987 rs6000_cannot_change_mode_class_ptr.
1988 (rs6000_cannot_change_mode_class): Replace with...
1989 (rs6000_can_change_mode_class): ...this new function, inverting the
1990 sense of the return value.
1991 (rs6000_debug_cannot_change_mode_class): Replace with...
1992 (rs6000_debug_can_change_mode_class): ...this new function.
1993 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1994 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1995 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1996 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1997 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1999 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2000 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2001 (rs6000_option_override_internal): Assign to
2002 targetm.can_change_mode_class instead of
2003 rs6000_cannot_change_mode_class_ptr.
2004 (rs6000_cannot_change_mode_class): Replace with...
2005 (rs6000_can_change_mode_class): ...this new function, inverting the
2006 sense of the return value.
2007 (rs6000_debug_cannot_change_mode_class): Replace with...
2008 (rs6000_debug_can_change_mode_class): ...this new function.
2009 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2010 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2011 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2012 (s390_can_change_mode_class): ...this new function, inverting the
2013 sense of the return value.
2014 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2015 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2016 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2017 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2018 (sh_cannot_change_mode_class): Replace with...
2019 (sh_can_change_mode_class): ...this new function, inverting the
2020 sense of the return value.
2021 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2022 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2023 (sparc_can_change_mode_class): New function.
2024 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2025 * config/spu/spu.c (spu_can_change_mode_class): New function.
2026 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2027 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2028 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2029 (visium_can_change_mode_class): New function.
2030 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2032 2017-09-15 Richard Biener <rguenther@suse.de>
2034 PR tree-optimization/82217
2035 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2036 but not undefined case.
2038 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2041 * postreload.c (reload_cse_simplify_operands): Skip
2042 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2044 2017-09-15 Richard Biener <rguenther@suse.de>
2046 PR tree-optimization/68823
2047 * graphite-scop-detection.c (build_alias_set): If we have a
2048 possible dependence check whether we can handle them by just
2049 looking at the DRs DR_ACCESS_FNs.
2050 (build_scops): If build_alias_set fails, fail the SCOP.
2052 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2054 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2055 to support float128 built-in functions that require the ISA 3.0
2057 (BU_FLOAT128_3_HW): Likewise.
2058 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2060 (FMAF128): Likewise.
2062 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2063 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2064 floating point instructions.
2065 (rs6000_invalid_builtin): Likewise.
2066 (rs6000_builtin_mask_names): Likewise.
2067 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2068 (RS6000_BTM_FLOAT128_HW): Likewise.
2069 (RS6000_BTM_COMMON): Likewise.
2070 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2072 * doc/extend.texi (RS/6000 built-in functions): Document the
2073 IEEE 128-bit floating point square root and fused multiply-add
2076 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2078 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2079 reg (r2) isn't in the set of registers defined in the prologue.
2081 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2082 Alan Hayward <alan.hayward@arm.com>
2083 David Sherwood <david.sherwood@arm.com>
2085 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2086 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2087 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2088 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2089 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2091 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2092 max_vectorization_factor.
2093 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2095 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2096 Alan Hayward <alan.hayward@arm.com>
2097 David Sherwood <david.sherwood@arm.com>
2099 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2100 (vect_worthwhile_without_simd_p): Declare.
2101 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2102 (vectorizable_reduction): Use it.
2103 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2104 (vectorizable_operation): Likewise.
2106 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2107 Alan Hayward <alan.hayward@arm.com>
2108 David Sherwood <david.sherwood@arm.com>
2110 * tree-vectorizer.h (vect_get_num_copies): New function.
2111 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2112 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2113 (vectorizable_induction): Likewise.
2114 (vectorizable_live_operation): Likewise.
2115 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2116 (vectorizable_bswap): Likewise.
2117 (vectorizable_call): Likewise.
2118 (vectorizable_conversion): Likewise.
2119 (vectorizable_assignment): Likewise.
2120 (vectorizable_shift): Likewise.
2121 (vectorizable_operation): Likewise.
2122 (vectorizable_store): Likewise.
2123 (vectorizable_load): Likewise.
2124 (vectorizable_condition): Likewise.
2125 (vectorizable_comparison): Likewise.
2126 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2128 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2129 Alan Hayward <alan.hayward@arm.com>
2130 David Sherwood <david.sherwood@arm.com>
2132 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2133 of vect_init_vector.
2135 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2136 Alan Hayward <alan.hayward@arm.com>
2137 David Sherwood <david.sherwood@arm.com>
2139 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2140 an inline wrapper that provides a location.
2141 (gimple_build_vector): Likewise.
2142 * gimple-fold.c (gimple_build_vector_from_val): New function.
2143 (gimple_build_vector): Likewise.
2144 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2145 functions to build the initial value. Always return a gimple value.
2146 (get_initial_defs_for_reduction): Likewise. Only compute
2148 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2149 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2150 (vectorizable_induction): Use gimple_build_vector rather than
2153 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2154 Alan Hayward <alan.hayward@arm.com>
2155 David Sherwood <david.sherwood@arm.com>
2157 * target.h (vec_perm_indices): New typedef.
2158 (auto_vec_perm_indices): Likewise.
2159 * optabs-query.h: Include target.h
2160 (can_vec_perm_p): Take a vec_perm_indices *.
2161 * optabs-query.c (can_vec_perm_p): Likewise.
2162 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2163 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2164 * tree-vect-generic.c (lower_vec_perm): Likewise.
2165 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2166 (vect_grouped_load_supported): Likewise.
2167 (vect_shift_permute_load_chain): Likewise.
2168 (vect_permute_store_chain): Use auto_vec_perm_indices.
2169 (vect_permute_load_chain): Likewise.
2170 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2171 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2172 Update uses of can_vec_perm_p.
2173 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2174 mode with a number of elements. Take a vec_perm_indices *.
2175 (vect_create_epilog_for_reduction): Update accordingly.
2176 Use auto_vec_perm_indices.
2177 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2178 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2179 (vect_transform_slp_perm_load): Likewise.
2180 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2181 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2182 (vect_gen_perm_mask_checked): Likewise.
2183 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2184 (vect_gen_perm_mask_checked): Likewise.
2185 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2186 (vectorizable_store): Likewise.
2187 (vectorizable_load): Likewise.
2188 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2189 (vectorizable_bswap): Likewise.
2191 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2192 Alan Hayward <alan.hayward@arm.com>
2193 David Sherwood <david.sherwood@arm.com>
2195 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2196 * tree.c (build_vector): Likewise.
2197 (build_vector_from_ctor): Update accordingly.
2198 (build_vector_from_val): Likewise.
2199 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2200 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2201 * tree-vect-generic.c (add_rshift): Likewise.
2202 (expand_vector_divmod): Likewise.
2203 (optimize_vector_constructor): Likewise.
2204 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2205 (vect_transform_slp_perm_load): Likewise.
2206 (vect_schedule_slp_instance): Likewise.
2207 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2208 (vectorizable_call): Likewise.
2209 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2210 * expmed.c (make_tree): Likewise.
2211 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2212 a vector passed to build_vector.
2213 (fold_convert_const): Likewise.
2214 (exact_inverse): Likewise.
2215 (fold_ternary_loc): Likewise.
2216 (fold_relational_const): Likewise.
2217 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2218 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2219 (const_unop): Likewise. Store the reduction accumulator in a
2220 variable rather than an array.
2221 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2222 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2223 the new vector, rather than constructing it after the input arrays.
2224 (native_interpret_vector): Use auto_vec<tree> when building
2225 a vector passed to build_vector. Add elements in order.
2226 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2227 auto_vec<tree> when building a vector passed to build_vector.
2228 (vect_create_epilog_for_reduction): Likewise.
2229 (vectorizable_induction): Likewise.
2230 (get_initial_def_for_reduction): Likewise. Fix indentation of
2232 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2234 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2235 passed to build_vector.
2237 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2238 Alan Hayward <alan.hayward@arm.com>
2239 David Sherwood <david.sherwood@arm.com>
2241 * tree-core.h (tree_base::u): Add an "nelts" field.
2242 (tree_vector): Use VECTOR_CST_NELTS as the length.
2243 * tree.c (tree_size): Likewise.
2244 (make_vector): Initialize VECTOR_CST_NELTS.
2245 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2246 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2247 TYPE_VECTOR_SUBPARTS.
2248 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2249 as the number of units, setting it from VECTOR_CST_NELTS.
2250 (const_vector_from_tree): Likewise.
2251 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2252 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2253 (fold_negate_expr_1): Likewise.
2254 (fold_convert_const): Likewise.
2255 (const_binop): Likewise. Differentiate the number of output and
2257 (const_unop): Likewise.
2258 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2259 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2260 in cases that did the opposite.
2262 2017-09-14 Richard Biener <rguenther@suse.de>
2264 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2267 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2269 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2271 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2274 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2275 if and where to split a bb, except for splitting before debug insn
2276 sequences followed by non-label real insn. Delete debug insns
2277 in between basic blocks.
2279 * combine.c (make_compound_operation_int): Formatting fixes.
2281 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2282 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2283 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2284 * config/sol2.h (LINK_EH_SPEC): Likewise.
2285 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2286 * config/s390/linux.h (LINK_SPEC): Likewise.
2287 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2288 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2289 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2290 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2291 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2292 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2293 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2294 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2296 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2298 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2300 (ENDFILE_LINUX_SPEC): Likewise.
2301 (LINK_EH_SPEC): Likewise.
2302 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2303 (LINK_OS_LINUX_SPEC32): Likewise.
2304 (LINK_OS_LINUX_SPEC64): Likewise.
2305 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2306 (LINK_OS_LINUX_SPEC): Likewise.
2308 2017-09-13 Martin Liska <mliska@suse.cz>
2311 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2312 CASE_HIGH is NULL_TREE.
2314 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2315 Alan Hayward <alan.hayward@arm.com>
2316 David Sherwood <david.sherwood@arm.com>
2318 * target.def (secondary_memory_needed): New hook.
2319 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2320 instead of SECONDARY_MEMORY_NEEDED.
2321 (secondary_memory_needed_mode): Likewise.
2322 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2323 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2324 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2325 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2326 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2327 * doc/tm.texi: Regenerate.
2328 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2329 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2330 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2331 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2332 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2333 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2334 mode argument first and change the reg_class arguments to reg_class_t.
2335 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2336 Make static. Update the call to inline_secondary_memory_needed.
2337 (ix86_register_move_cost): Update the call to
2338 inline_secondary_memory_needed.
2339 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2340 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2342 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2343 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2345 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2346 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2347 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2348 and match hook interface. Add comment from mips.h.
2349 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2350 * config/mmix/mmix.md (truncdfsf2): Refer to
2351 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2353 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2354 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2355 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2356 (pa_secondary_memory_needed): New function.
2357 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2358 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2359 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2360 (pdp11_secondary_memory_needed): Make static and match hook interface.
2361 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2362 * config/powerpcspe/powerpcspe-protos.h
2363 (rs6000_secondary_memory_needed_ptr): Delete.
2364 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2366 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2367 (rs6000_option_override_internal): Assign to
2368 targetm.secondary_memory_needed rather than
2369 rs6000_secondary_memory_needed_ptr.
2370 (rs6000_secondary_memory_needed): Match hook interface.
2371 (rs6000_debug_secondary_memory_needed): Likewise.
2372 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2373 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2374 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2375 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2376 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2377 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2379 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2380 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2381 (rs6000_option_override_internal): Assign to
2382 targetm.secondary_memory_needed rather than
2383 rs6000_secondary_memory_needed_ptr.
2384 (rs6000_secondary_memory_needed): Match hook interface.
2385 (rs6000_debug_secondary_memory_needed): Likewise.
2386 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2387 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2388 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2389 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2390 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2391 (sparc_secondary_memory_needed): New function.
2392 * lra-constraints.c (check_and_process_move): Refer to
2393 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2395 (curr_insn_transform): Likewise.
2396 (process_alt_operands): Use targetm.secondary_memory_needed
2397 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2398 (check_secondary_memory_needed_p): Likewise.
2399 (choose_split_class): Likewise.
2400 * reload.c: Unconditionally include code that was previously
2401 conditional on SECONDARY_MEMORY_NEEDED.
2402 (push_secondary_reload): Use targetm.secondary_memory_needed
2403 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2404 (push_reload): Likewise.
2405 * reload1.c: Unconditionally include code that was previously
2406 conditional on SECONDARY_MEMORY_NEEDED.
2407 (choose_reload_regs): Use targetm.secondary_memory_needed
2408 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2409 (gen_reload): Likewise.
2410 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2412 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2413 Alan Hayward <alan.hayward@arm.com>
2414 David Sherwood <david.sherwood@arm.com>
2416 * target.def (secondary_memory_needed_mode): New hook:
2417 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2418 * targhooks.h (default_secondary_memory_needed_mode): New function.
2419 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2420 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2421 * doc/tm.texi: Regenerate.
2422 * lra-constraints.c (check_and_process_move): Use
2423 targetm.secondary_memory_needed_mode instead of
2424 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2425 (curr_insn_transform): Likewise.
2426 * reload.c (get_secondary_mem): Likewise.
2427 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2428 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2430 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2431 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2432 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2433 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2434 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2436 * config/powerpcspe/powerpcspe-protos.h
2437 (rs6000_secondary_memory_needed_mode): Delete.
2438 * config/powerpcspe/powerpcspe.c
2439 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2440 (rs6000_secondary_memory_needed_mode): Make static.
2441 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2442 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2444 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2446 (rs6000_secondary_memory_needed_mode): Make static.
2447 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2448 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2449 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2450 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2451 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2453 (sparc_secondary_memory_needed_mode): New function.
2454 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2456 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2458 * config/aarch64/constraints.md (Umq): New constraint.
2459 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2461 (mov<mode>): Update condition.
2463 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2465 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2466 when bitposition is the same.
2468 2017-09-13 Richard Biener <rguenther@suse.de>
2470 * dwarf2out.c (output_die_symbol): Remove.
2471 (output_die): Do not output a DIEs symbol.
2473 2017-09-13 Richard Biener <rguenther@suse.de>
2476 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2477 default-def to avoid breaking iterator update with the weird
2478 interaction with cgraph_update_edges_for_call_stmt_node.
2480 2017-09-13 Richard Biener <rguenther@suse.de>
2482 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2483 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2484 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2485 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2487 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2489 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2490 Disable pc relative literal load irrespective of
2491 TARGET_FIX_ERR_A53_84341 for default.
2493 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2495 * config/sparc/sparc.c (output_return): Output the source location of
2496 the insn in the delay slot, if any.
2497 (output_sibcall): Likewise.
2499 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2502 * common.opt (-static-pie): New alias.
2503 (shared): Negate static-pie.
2504 (-no-pie): Update help text.
2506 (static-pie): New option.
2507 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2508 -static-pie support.
2509 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2510 (LINK_EH_SPEC): Likewise.
2511 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2512 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2513 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2514 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2515 (init_gcc_specs): Likewise.
2516 (init_spec): Likewise.
2517 (display_help): Update help message for -pie.
2518 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2521 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2523 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2524 (movdi_aarch64): Likewise.
2525 (movti_aarch64): Likewise.
2527 2017-09-12 Simon Wright <simon@pushface.org>
2530 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2531 calculation of the minor version, always output as 0.
2533 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2536 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2537 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2538 on it early, rather than manual conversion late. For
2539 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2540 instead of performing manual conversion.
2542 2017-09-12 Carl Love <cel@us.ibm.com>
2544 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2545 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2547 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2548 VMULOSW): Add definitions.
2549 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2550 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2551 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2552 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2553 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2555 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2557 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2559 (movti_aarch64): Likewise.
2560 (movdf_aarch64): Likewise.
2561 (movtf_aarch64): Likewise.
2562 (load_pairdi): Likewise.
2563 (store_pairdi): Likewise.
2564 (load_pairdf): Likewise.
2565 (store_pairdf): Likewise.
2566 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2567 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2568 (ldr_got_small_<mode>): Likewise.
2569 (ldr_got_small_28k_<mode>): Likewise.
2570 (ldr_got_tiny): Likewise.
2571 * config/aarch64/iterators.md (ldst_sz): New.
2572 (ldpstp_sz): Likewise.
2573 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2575 (thunderx_load): Split load_8 to load_16.
2576 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2578 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2579 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2580 (xgene1_store_pair): Split store_8 to store_16.
2581 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2582 (falkor_st_0_st_sd): Split store_8 to store_16.
2584 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2586 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2587 and store1/2/3/4 to store_4/8/12/16.
2588 * config/aarch64/aarch64.md: Update for rename.
2589 * config/arm/arm.md: Likewise.: Likewise.
2590 * config/arm/arm.c: Likewise.
2591 * config/arm/thumb1.md: Likewise.
2592 * config/arm/thumb2.md: Likewise.
2593 * config/arm/vfp.md: Likewise.
2594 * config/arm/arm-generic.md: Likewise.
2595 * config/arm/arm1020e.md: Likewise.
2596 * config/arm/arm1026ejs.md: Likewise.
2597 * config/arm/arm1136jfs.md: Likewise.
2598 * config/arm/arm926ejs.md: Likewise.
2599 * config/arm/cortex-a15.md: Likewise.
2600 * config/arm/cortex-a17.md: Likewise.
2601 * config/arm/cortex-a5.md: Likewise.
2602 * config/arm/cortex-a53.md: Likewise.
2603 * config/arm/cortex-a57.md: Likewise.
2604 * config/arm/cortex-a7.md: Likewise.
2605 * config/arm/cortex-a8.md: Likewise.
2606 * config/arm/cortex-a9.md: Likewise.
2607 * config/arm/cortex-m4.md: Likewise.
2608 * config/arm/cortex-m7.md: Likewise.
2609 * config/arm/cortex-r4.md: Likewise.
2610 * config/arm/exynos-m1.md: Likewise.
2611 * config/arm/fa526.md: Likewise.
2612 * config/arm/fa606te.md: Likewise.
2613 * config/arm/fa626te.md: Likewise.
2614 * config/arm/fa726te.md: Likewise.
2615 * config/arm/fmp626.md: Likewise.
2616 * config/arm/iwmmxt.md: Likewise.
2617 * config/arm/ldmstm.md: Likewise.
2618 * config/arm/marvell-pj4.md: Likewise.
2619 * config/arm/xgene1.md: Likewise.
2620 * config/aarch64/thunderx.md: Likewise.
2621 * config/aarch64/thunderx2t99.md: Likewise.
2622 * config/aarch64/falkor.md: Likewise.
2624 2017-09-12 Martin Liska <mliska@suse.cz>
2626 * attribs.c (private_lookup_attribute): New function.
2627 * attribs.h (private_lookup_attribute): Declared here.
2628 (lookup_attribute): Called from this place.
2630 2017-09-12 Richard Biener <rguenther@suse.de>
2632 PR tree-optimization/82157
2633 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2634 stmts with side-effects.
2636 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2637 Alan Hayward <alan.hayward@arm.com>
2638 David Sherwood <david.sherwood@arm.com>
2640 * target.def (hard_regno_nregs): New hook.
2641 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2642 * targhooks.h (default_hard_regno_nregs): Declare.
2643 * targhooks.c (default_hard_regno_nregs): New function.
2644 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2645 (TARGET_HARD_REGNO_NREGS): ...this hook.
2646 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2647 (CLASS_MAX_NREGS): Likewise.
2648 * doc/tm.texi: Regenerate.
2649 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2650 instead of HARD_REGNO_NREGS.
2651 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2652 HARD_REGNO_NREGS in the comment.
2653 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2654 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2655 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2656 Return an unsigned int.
2657 (TARGET_HARD_REGNO_NREGS): Redefine.
2658 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2659 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2660 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2661 (arc_hard_regno_nregs): New function.
2662 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2663 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2664 (arm_hard_regno_nregs): New function.
2665 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2666 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2667 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2668 (TARGET_HARD_REGNO_NREGS): Redefine.
2669 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2670 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2671 (HARD_REGNO_NREGS): Delete.
2672 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2673 (cr16_hard_regno_nregs): New function.
2674 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2675 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2676 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2677 (cris_hard_regno_nregs): New function.
2678 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2679 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2680 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2681 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2682 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2683 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2684 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2685 (frv_hard_regno_nregs): Make static. Take and return an
2687 (frv_class_max_nregs): Remove outdated copy of documentation.
2688 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2689 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2690 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2691 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2692 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2693 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2694 (TARGET_HARD_REGNO_NREGS): Redefine.
2695 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2696 (CLASS_MAX_NREGS): Update comment.
2697 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2698 (ia64_hard_regno_nregs): New function.
2699 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2700 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2701 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2702 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2703 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2705 (m32c_hard_regno_nregs): Likewise. Make static.
2706 (TARGET_HARD_REGNO_NREGS): Redefine.
2707 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2708 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2709 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2710 (m68k_hard_regno_nregs): New function.
2711 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2712 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2713 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2714 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2715 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2716 Take and return an unsigned int.
2717 (TARGET_HARD_REGNO_NREGS): Redefine.
2718 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2719 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2720 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2721 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2722 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2723 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2724 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2725 (msp430_hard_regno_nregs): Make static. Take and return an
2727 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2728 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2729 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2730 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2731 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2732 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2733 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2734 (TARGET_HARD_REGNO_NREGS): Redefine.
2735 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2736 (PA_HARD_REGNO_NREGS): ...this.
2737 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2738 (PA_HARD_REGNO_NREGS): ...this.
2739 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2740 (pa_hard_regno_nregs): New function.
2741 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2742 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2743 (pdp11_hard_regno_nregs): New function.
2744 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2745 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2746 (rs6000_hard_regno_nregs_hook): New function.
2747 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2748 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2749 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2750 Take and return an unsigned int. Move earlier in file.
2751 (TARGET_HARD_REGNO_NREGS): Redefine.
2752 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2753 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2754 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2755 (rl78_hard_regno_nregs): Make static. Take and return an
2757 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2758 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2759 (rs6000_hard_regno_nregs_hook): New function.
2760 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2761 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2762 (TARGET_HARD_REGNO_NREGS): Redefine.
2763 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2764 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2765 instead of HARD_REGNO_NREGS.
2766 (s390_hard_regno_nregs): New function.
2767 (s390_hard_regno_mode_ok): Add comment from s390.h.
2768 (TARGET_HARD_REGNO_NREGS): Redefine.
2769 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2770 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2771 (sh_hard_regno_nregs): New function.
2772 (sh_pass_in_reg_p): Use it.
2773 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2774 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2775 (sparc_hard_regno_nregs): New function.
2776 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2777 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2778 (spu_function_arg_advance): Use it, supplying a valid register number.
2779 (TARGET_HARD_REGNO_NREGS): Redefine.
2780 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2781 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2782 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2783 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2784 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2785 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2786 (CLASS_MAX_NREGS): Remove copy of old documentation.
2787 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2788 (visium_hard_regno_nregs): New function.
2789 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2790 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2791 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2792 xtensa_hard_regno_nregs): New function.
2793 * system.h (HARD_REGNO_NREGS): Poison.
2795 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2797 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2798 hard_regno_nregs instead of HARD_REGNO_NREGS.
2799 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2800 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2801 (c6x_expand_epilogue): Likewise.
2802 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2803 (frv_read_iacc_argument): Likewise.
2804 * config/sh/sh.c: Include regs.h.
2805 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2806 (regs_used): Likewise.
2807 (output_stack_adjust): Likewise.
2808 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2809 * expmed.c: Include regs.h.
2810 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2811 * ree.c: Include regs.h.
2812 (combine_reaching_defs): Use hard_regno_nregs instead of
2814 (add_removable_extension): Likewise.
2816 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2818 * regs.h (hard_regno_nregs): Turn into a function.
2819 (end_hard_regno): Update accordingly.
2820 * caller-save.c (setup_save_areas): Likewise.
2821 (save_call_clobbered_regs): Likewise.
2822 (replace_reg_with_saved_mem): Likewise.
2823 (insert_restore): Likewise.
2824 (insert_save): Likewise.
2825 * combine.c (can_change_dest_mode): Likewise.
2826 (move_deaths): Likewise.
2827 (distribute_notes): Likewise.
2828 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2829 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2830 (rs6000_split_multireg_move): Likewise.
2831 (rs6000_register_move_cost): Likewise.
2832 (rs6000_memory_move_cost): Likewise.
2833 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2834 (rs6000_split_multireg_move): Likewise.
2835 (rs6000_register_move_cost): Likewise.
2836 (rs6000_memory_move_cost): Likewise.
2837 * cselib.c (cselib_reset_table): Likewise.
2838 (cselib_lookup_1): Likewise.
2839 * emit-rtl.c (set_mode_and_regno): Likewise.
2840 * function.c (aggregate_value_p): Likewise.
2841 * ira-color.c (setup_profitable_hard_regs): Likewise.
2842 (check_hard_reg_p): Likewise.
2843 (calculate_saved_nregs): Likewise.
2844 (assign_hard_reg): Likewise.
2845 (improve_allocation): Likewise.
2846 (calculate_spill_cost): Likewise.
2847 * ira-emit.c (modify_move_list): Likewise.
2848 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2849 (ira_hard_reg_in_set_p): Likewise.
2850 * ira.c (setup_reg_mode_hard_regset): Likewise.
2851 (clarify_prohibited_class_mode_regs): Likewise.
2852 (check_allocation): Likewise.
2853 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2854 (lra_setup_reg_renumber): Likewise.
2855 (setup_try_hard_regno_pseudos): Likewise.
2856 (spill_for): Likewise.
2857 (assign_hard_regno): Likewise.
2858 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2859 * lra-constraints.c (in_class_p): Likewise.
2860 (lra_constraint_offset): Likewise.
2861 (simplify_operand_subreg): Likewise.
2862 (lra_constraints): Likewise.
2863 (split_reg): Likewise.
2864 (split_if_necessary): Likewise.
2865 (invariant_p): Likewise.
2866 (inherit_in_ebb): Likewise.
2867 * lra-lives.c (process_bb_lives): Likewise.
2868 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2869 (get_hard_regs): Likewise.
2870 (do_remat): Likewise.
2871 * lra-spills.c (assign_spill_hard_regs): Likewise.
2872 * mode-switching.c (create_pre_exit): Likewise.
2873 * postreload.c (reload_combine_recognize_pattern): Likewise.
2874 * recog.c (peep2_find_free_register): Likewise.
2875 * regcprop.c (kill_value_regno): Likewise.
2876 (set_value_regno): Likewise.
2877 (copy_value): Likewise.
2878 (maybe_mode_change): Likewise.
2879 (find_oldest_value_reg): Likewise.
2880 (copyprop_hardreg_forward_1): Likewise.
2881 * regrename.c (check_new_reg_p): Likewise.
2882 (regrename_do_replace): Likewise.
2883 * reload.c (push_reload): Likewise.
2884 (combine_reloads): Likewise.
2885 (find_dummy_reload): Likewise.
2886 (operands_match_p): Likewise.
2887 (find_reloads): Likewise.
2888 (find_equiv_reg): Likewise.
2889 (reload_adjust_reg_for_mode): Likewise.
2890 * reload1.c (count_pseudo): Likewise.
2891 (count_spilled_pseudo): Likewise.
2892 (find_reg): Likewise.
2893 (clear_reload_reg_in_use): Likewise.
2894 (free_for_value_p): Likewise.
2895 (allocate_reload_reg): Likewise.
2896 (choose_reload_regs): Likewise.
2897 (reload_adjust_reg_for_temp): Likewise.
2898 (emit_reload_insns): Likewise.
2899 (delete_output_reload): Likewise.
2900 * rtlanal.c (subreg_get_info): Likewise.
2901 * sched-deps.c (sched_analyze_reg): Likewise.
2902 * sel-sched.c (init_regs_for_mode): Likewise.
2903 (mark_unavailable_hard_regs): Likewise.
2904 (choose_best_reg_1): Likewise.
2905 (verify_target_availability): Likewise.
2906 * valtrack.c (dead_debug_insert_temp): Likewise.
2907 * var-tracking.c (track_loc_p): Likewise.
2908 (emit_note_insn_var_location): Likewise.
2909 * varasm.c (make_decl_rtl): Likewise.
2910 * reginfo.c (choose_hard_reg_mode): Likewise.
2911 (init_reg_modes_target): Refer directly to
2912 this_target_regs->x_hard_regno_nregs.
2914 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2916 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2917 instead of hard_regno_nregs.
2919 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2921 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2922 end_hard_regno instead of hard_regno_nregs.
2923 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2924 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2925 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2926 * ira-color.c (improve_allocation): Likewise.
2927 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2928 * lra-lives.c (mark_regno_live): Likewise.
2929 (mark_regno_dead): Likewise.
2930 * lra-remat.c (operand_to_remat): Likewise.
2931 * lra.c (collect_non_operand_hard_regs): Likewise.
2932 * postreload.c (reload_combine_note_store): Likewise.
2933 (move2add_valid_value_p): Likewise.
2934 * reload.c (regno_clobbered_p): Likewise.
2936 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2938 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2940 * config/v850/v850.c (v850_reorg): Likewise.
2941 * reload.c (refers_to_regno_for_reload_p): Likewise.
2942 (find_equiv_reg): Likewise.
2943 * reload1.c (reload_reg_reaches_end_p): Likewise.
2945 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2947 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2949 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2950 * config/arm/arm.c (output_move_neon): Likewise.
2951 (arm_attr_length_move_neon): Likewise.
2952 (neon_split_vcombine): Likewise.
2953 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2954 (c6x_mark_reg_written): Likewise.
2955 (c6x_dwarf_register_span): Likewise.
2956 * config/i386/i386.c (ix86_save_reg): Likewise.
2957 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2958 (rws_access_reg): Likewise.
2959 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2960 * mode-switching.c (create_pre_exit): Likewise.
2961 * ree.c (combine_reaching_defs): Likewise.
2962 (add_removable_extension): Likewise.
2963 * regcprop.c (find_oldest_value_reg): Likewise.
2964 (copyprop_hardreg_forward_1): Likewise.
2965 * reload.c (reload_inner_reg_of_subreg): Likewise.
2966 (push_reload): Likewise.
2967 (combine_reloads): Likewise.
2968 (find_dummy_reload): Likewise.
2969 (reload_adjust_reg_for_mode): Likewise.
2970 * reload1.c (find_reload_regs): Likewise.
2971 (forget_old_reloads_1): Likewise.
2972 (reload_reg_free_for_value_p): Likewise.
2973 (reload_adjust_reg_for_temp): Likewise.
2974 (emit_reload_insns): Likewise.
2975 (delete_output_reload): Likewise.
2976 * sel-sched.c (choose_best_reg_1): Likewise.
2977 (choose_best_pseudo_reg): Likewise.
2979 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2980 Alan Hayward <alan.hayward@arm.com>
2981 David Sherwood <david.sherwood@arm.com>
2983 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2984 * target.def (slow_unaligned_access): New hook.
2985 * targhooks.h (default_slow_unaligned_access): Declare.
2986 * targhooks.c (default_slow_unaligned_access): New function.
2987 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2988 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2989 * doc/tm.texi: Regenerate.
2990 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2991 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2992 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2994 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2995 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2997 (rs6000_slow_unaligned_access): New function.
2998 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2999 (expand_block_compare): Likewise.
3000 (expand_strn_compare): Likewise.
3001 (rs6000_rtx_costs): Likewise.
3002 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3003 (riscv_slow_unaligned_access): Likewise.
3004 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3005 (riscv_slow_unaligned_access_p): ...this and make static.
3006 (riscv_option_override): Update accordingly.
3007 (riscv_slow_unaligned_access): New function.
3008 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3009 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3010 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3011 (rs6000_slow_unaligned_access): New function.
3012 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3013 (rs6000_rtx_costs): Likewise.
3014 * config/rs6000/rs6000-string.c (expand_block_compare)
3015 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3016 of SLOW_UNALIGNED_ACCESS.
3017 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3018 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3019 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3020 of SLOW_UNALIGNED_ACCESS.
3021 * expmed.c (simple_mem_bitfield_p): Likewise.
3022 * expr.c (alignment_for_piecewise_move): Likewise.
3023 (emit_group_load_1): Likewise.
3024 (emit_group_store): Likewise.
3025 (copy_blkmode_from_reg): Likewise.
3026 (emit_push_insn): Likewise.
3027 (expand_assignment): Likewise.
3028 (store_field): Likewise.
3029 (expand_expr_real_1): Likewise.
3030 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3031 * lra-constraints.c (simplify_operand_subreg): Likewise.
3032 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3033 * gimple-ssa-store-merging.c: Likewise in block comment at start
3035 * tree-ssa-strlen.c: Include target.h.
3036 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3037 of SLOW_UNALIGNED_ACCESS.
3038 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3040 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3042 PR rtl-optimization/82185
3043 * expmed.c (emit_store_flag_int): Only test tem if it has been
3046 2017-09-12 Richard Biener <rguenther@suse.de>
3049 * match.pd ((FTYPE) N CMP CST): Fix typo.
3051 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3053 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3055 (mips_near_type_p): Add 'short_call' attribute as a synonym
3057 * doc/extend.texi (short_call): Document new function attribute.
3059 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3062 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3063 assertion check that in the condition.
3064 (get_atomic_generic_size): Likewise. Before testing if parameter
3065 has pointer type, if it has array type, call for C++
3066 default_conversion to perform array-to-pointer conversion.
3068 2017-09-12 Richard Biener <rguenther@suse.de>
3070 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3071 for operations we cannot scalarize.
3073 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3075 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3076 vectors heap vectors. Clean up comments.
3077 Make visited_bbs a reference.
3078 (profitable_jump_thread_path): Make GC
3079 vectors heap vectors. Clean up comments.
3081 (convert_and_register_jump_thread_path): Make GC vectors heap
3083 (check_subpath_and_update_thread_path): Same. Clean up comments.
3084 Make visited_bbs a reference.
3085 (handle_phi): Abstract common code to to
3086 register_jump_thread_path_if_profitable.
3087 Rename VAR_BB to DEF_BB.
3089 Make GC vectors heap vectors.
3090 Make visited_bbs a reference.
3091 (handle_assignment): Same.
3092 (register_jump_thread_path_if_profitable): New.
3093 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3095 Make GC vectors heap vectors. Clean up comments.
3096 Make visited_bbs a reference.
3097 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3098 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3101 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3104 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3105 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3107 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3109 Revert r251800 and r251799.
3111 2017-09-11 Martin Jambor <mjambor@suse.cz>
3114 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3115 arguments in advance.
3116 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3117 use it to find predecessor edges.
3118 (naive_outof_ssa): Collect vector of predecessors.
3120 2017-09-08 Jason Merrill <jason@redhat.com>
3122 PR c++/70029 - ICE with ref-qualifier and -flto
3123 * langhooks.h (struct lang_hooks_for_types): Add
3124 copy_lang_qualifiers.
3125 * attribs.c (build_type_attribute_qual_variant): Use it.
3126 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3128 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3129 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3131 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3134 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3135 (*mulsi3_sp64): New instruction.
3136 (mulsi3): New expander.
3138 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3140 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3142 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3144 * sancov.c: Include memmodel.h.
3146 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3149 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3152 2017-09-07 Carl Love <cel@us.ibm.com>
3154 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3155 the sldi instruction.
3157 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3159 * sancov.c: Include tm_p.h.
3161 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3164 * output.h (switch_to_other_text_partition): New declaration.
3165 * varasm.c (switch_to_other_text_partition): New function.
3166 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3167 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3168 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3169 to the other text partition before emitting LCL label and switch back
3170 after emitting the word after it.
3172 2017-09-07 Richard Biener <rguenther@suse.de>
3174 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3175 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3176 critical edges here, after loop init.
3177 (pass_data_pre): Remove PROP_no_crit_edges flags.
3178 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3179 for valueization of call args to avoid leaking VN_TOP.
3180 (visit_use): Assert we do not visit default defs.
3181 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3182 Use error_mark_node to more easily detect leaking VN_TOP.
3183 All default-defs are varying, not VN_TOP. Mark them visited.
3184 (run_scc_vn): Make code match comment.
3186 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3188 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3189 OPTION_MASK_FLOAT128_KEYWORD.
3190 (POWERPC_MASKS): Likewise.
3191 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3192 support for the -mfloat128-type option, and make -mfloat128
3193 default on PowerPC Linux systems. Define or undefine
3194 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3195 Define __float128 to be __ieee128 if IEEE 128-bit support is
3196 enabled, or undefine it.
3197 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3198 Delete defining __FLOAT128_TYPE__.
3199 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3200 -mfloat128-type option and make -mfloat128 default on PowerPC
3202 (TARGET_FLOAT128_TYPE): Likewise.
3203 (-mfloat128-type): Likewise.
3204 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3205 Delete the -mfloat128-type option and make -mfloat128 default on
3206 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3207 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3208 128-bit floating point is enabled. Change tests from using
3209 -mfloat128-type to -mfloat128.
3210 (rs6000_mangle_type): Use the correct mangling for the __float128
3211 type even if normal long double is restricted to 64-bits.
3212 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3214 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3215 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3216 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3218 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3219 documentation for -mfloat128.
3221 2017-09-06 Olivier Hainque <hainque@adacore.com>
3223 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3225 2017-09-06 Wish Wu <wishwu007@gmail.com>
3226 Jakub Jelinek <jakub@redhat.com>
3228 * asan.c (initialize_sanitizer_builtins): Add
3229 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3230 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3231 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3232 BT_FN_VOID_UINT64_PTR variables.
3233 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3234 (BT_FN_VOID_UINT16_UINT16): Likewise.
3235 (BT_FN_VOID_UINT32_UINT32): Likewise.
3236 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3237 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3238 (BT_FN_VOID_UINT64_PTR): Likewise.
3239 * common.opt (flag_sanitize_coverage): New variable.
3240 (fsanitize-coverage=trace-pc): Remove.
3241 (fsanitize-coverage=): Add.
3242 * flag-types.h (enum sanitize_coverage_code): New enum.
3243 * fold-const.c (fold_range_test): Disable non-short-circuit
3244 optimization if flag_sanitize_coverage.
3245 (fold_truth_andor): Likewise.
3246 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3247 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3248 (coverage_sanitizer_opts): New array.
3249 (get_closest_sanitizer_option): Add OPTS argument, handle also
3250 OPT_fsanitize_coverage_.
3251 (parse_sanitizer_options): Adjusted to also handle
3252 OPT_fsanitize_coverage_.
3253 (common_handle_option): Add OPT_fsanitize_coverage_.
3254 * sancov.c (instrument_comparison, instrument_switch): New function.
3255 (sancov_pass): Add trace-cmp support.
3256 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3257 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3258 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3259 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3260 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3261 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3262 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3263 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3264 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3265 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3267 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3269 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3270 error. Only quit immediately if parsing is complete.
3271 (BEGIN): Initialize fatal_err and parse_done.
3272 (begin fpu, end fpu): Check number of arguments.
3273 (begin arch, end arch): Likewise.
3274 (begin cpu, end cpu): Likewise.
3275 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3276 (optalias): Likewise.
3278 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3280 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3281 * config/arm/arm-isa.h: Delete. Move definitions to ...
3282 * arm-cpus.in: ... here. Use new feature and fgroup values.
3283 * config/arm/arm.c (arm_option_override): Use lower case for feature
3285 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3286 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3287 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3289 (print_isa_bits_for): New function.
3290 (gen_isa): New function.
3291 (gen_comm_data): Use print_isa_bits_for.
3292 (define feature): New keyword.
3293 (define fgroup): New keyword.
3294 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3295 (arm-isa.h): Add rule to generate file.
3296 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3297 case for feature bit names.
3299 2017-09-06 Richard Biener <rguenther@suse.de>
3301 * tree-ssa-pre.c (NECESSARY): Remove.
3302 (create_expression_by_pieces): Do not touch pass-local flags.
3303 (insert_into_preds_of_block): Likewise.
3304 (do_pre_regular_insertion): Likewise.
3305 (eliminate_insert): Likewise.
3306 (eliminate_dom_walker::before_dom_children): Likewise.
3307 (fini_eliminate): Do not look at inserted_exprs.
3308 (mark_operand_necessary): Remove.
3309 (remove_dead_inserted_code): Replace with simple work-list
3310 algorithm based on inserted_exprs and SSA uses.
3311 (pass_pre::execute): Re-order fini_eliminate and
3312 remove_dead_inserted_code.
3314 2017-09-06 Olivier Hainque <hainque@adacore.com>
3316 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3317 for VxWorks 7. Adjust surrounding comments.
3319 2017-09-06 Richard Biener <rguenther@suse.de>
3321 * gimple-ssa-strength-reduction.c
3322 (find_candidates_dom_walker::before_dom_children): Also allow
3325 2017-09-06 Richard Biener <rguenther@suse.de>
3327 PR tree-optimization/82108
3328 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3329 for gap in the non-permutation SLP case.
3331 2017-09-06 Martin Jambor <mjambor@suse.cz>
3333 PR tree-optimization/82078
3334 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3335 add_access_to_work_queue...
3336 (build_accesses_from_assign): ...here.
3337 (propagate_all_subaccesses): Make sure racc is the group
3338 representative, if there is one.
3340 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3343 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3346 2017-09-06 Richard Biener <rguenther@suse.de>
3348 * gimple-ssa-strength-reduction.c
3349 (find_candidates_dom_walker::before_doom_children): Use a
3350 type and not a mode check.
3352 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3355 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3357 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3359 2017-09-05 Jeff Law <law@redhat.com>
3361 PR tree-optimization/64910
3362 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3363 cases where we have 3 or more operands.
3365 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3368 * omp-low.c (lower_omp_for): Recompute tree invariant if
3369 gimple_omp_for_initial/final is ADDR_EXPR.
3372 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3373 into gimple val before gimplification fo the COND_EXPR.
3375 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3377 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3378 REGION_COPY argument.
3379 (thread_through_all_blocks): Remove unused argument to
3380 duplicate_thread_path.
3382 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3383 Alan Hayward <alan.hayward@arm.com>
3384 David Sherwood <david.sherwood@arm.com>
3386 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3387 Take a scalar_mode rather than a machine_mode.
3388 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3389 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3390 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3391 (aarch64_gen_adjusted_ldpstp): Likewise.
3392 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3394 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3395 Alan Hayward <alan.hayward@arm.com>
3396 David Sherwood <david.sherwood@arm.com>
3398 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3399 Take a scalar_int_mode instead of a machine_mode.
3400 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3401 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3402 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3403 (aarch64_simd_attr_length_rglist): Delete.
3404 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3405 a scalar_int_mode instead of a machine_mode.
3406 (aarch64_add_offset): Likewise.
3407 (aarch64_internal_mov_immediate): Likewise
3408 (aarch64_add_constant_internal): Likewise.
3409 (aarch64_add_constant): Likewise.
3410 (aarch64_movw_imm): Likewise.
3411 (aarch64_rtx_arith_op_extract_p): Likewise.
3412 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3413 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3414 Remove assert that the mode isn't a vector.
3415 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3416 (aarch64_expand_mov_immediate): Update calls after above changes.
3417 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3418 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3419 (aarch64_move_imm): Likewise.
3420 (aarch64_can_const_movi_rtx_p): Likewise.
3421 (aarch64_strip_extend): Likewise.
3422 (aarch64_extr_rtx_p): Likewise.
3423 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3424 a CONST_INT when the mode parameter is VOIDmode.
3425 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3427 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3429 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3430 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3431 (bitwise_type_for_mode): Update accordingly.
3433 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3435 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3436 * stor-layout.c (mode_for_size_tree): Likewise.
3437 (mode_for_array): Update accordingly.
3438 (layout_decl): Likewise.
3439 (compute_record_mode): Likewise. Only set the mode once.
3441 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3443 * target.def (get_mask_mode): Change return type to opt_mode.
3445 * doc/tm.texi: Regenerate.
3446 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3447 * targhooks.c (default_get_mask_mode): Likewise.
3448 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3449 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3450 targetm.get_mask_mode.
3451 * tree.c (build_truth_vector_type): Likewise.
3453 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3455 * machmode.h (mode_for_vector): Return an opt_mode.
3456 * stor-layout.c (mode_for_vector): Likewise.
3457 (mode_for_int_vector): Update accordingly.
3458 (layout_type): Likewise.
3459 * config/i386/i386.c (emit_memmov): Likewise.
3460 (ix86_expand_set_or_movmem): Likewise.
3461 (ix86_expand_vector_init): Likewise.
3462 (ix86_get_mask_mode): Likewise.
3463 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3465 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3466 * expmed.c (extract_bit_field_1): Likewise.
3467 * expr.c (expand_expr_real_2): Likewise.
3468 * optabs-query.c (can_vec_perm_p): Likewise.
3469 (can_vec_mask_load_store_p): Likewise.
3470 * optabs.c (expand_vec_perm): Likewise.
3471 * targhooks.c (default_get_mask_mode): Likewise.
3472 * tree-vect-stmts.c (vectorizable_store): Likewise.
3473 (vectorizable_load): Likewise.
3474 (get_vectype_for_scalar_type_and_size): Likewise.
3476 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3478 * machmode.h (mode_for_int_vector): New function.
3479 * stor-layout.c (mode_for_int_vector): Likewise.
3480 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3481 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3482 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3483 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3484 (s390_expand_vcond): Likewise.
3486 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3488 * machmode.h (opt_machine_mode): New type.
3489 (opt_mode<T>): Allow construction from anything that can be
3491 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3492 (mode_for_size): Return an opt_machine_mode.
3493 * stor-layout.c (mode_for_size): Likewise.
3494 (mode_for_size_tree): Update call accordingly.
3495 (bitwise_mode_for_mode): Likewise.
3496 (make_fract_type): Likewise.
3497 (make_accum_type): Likewise.
3498 * caller-save.c (replace_reg_with_saved_mem): Update call
3500 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3501 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3502 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3503 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3504 * expmed.c (extract_bit_field_1): Likewise.
3505 * reload.c (get_secondary_mem): Likewise.
3506 * varasm.c (assemble_integer): Likewise.
3507 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3510 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3512 * machmode.h (decimal_float_mode_for_size): New function.
3513 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3514 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3515 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3516 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3517 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3518 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3520 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3522 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3523 (get_builtin_sync_mode): Likewise.
3524 (expand_ifn_atomic_compare_exchange): Likewise.
3525 (expand_builtin_atomic_clear): Likewise.
3526 (expand_builtin_atomic_test_and_set): Likewise.
3527 (fold_builtin_atomic_always_lock_free): Likewise.
3528 * calls.c (compute_argument_addresses): Likewise.
3529 (emit_library_call_value_1): Likewise.
3530 (store_one_arg): Likewise.
3531 * combine.c (combine_instructions): Likewise.
3532 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3533 * config/arm/arm.c (arm_function_value): Likewise.
3534 (aapcs_allocate_return_reg): Likewise.
3535 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3536 * config/i386/i386.c (construct_container): Likewise.
3537 (ix86_gimplify_va_arg): Likewise.
3538 (ix86_expand_sse_cmp): Likewise.
3539 (emit_memmov): Likewise.
3540 (emit_memset): Likewise.
3541 (expand_small_movmem_or_setmem): Likewise.
3542 (ix86_expand_pextr): Likewise.
3543 (ix86_expand_pinsr): Likewise.
3544 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3545 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3547 * config/mips/mips.c (mips_function_value_1) Likewise.
3548 (mips_block_move_straight): Likewise.
3549 (mips_expand_ins_as_unaligned_store): Likewise.
3550 * config/powerpcspe/powerpcspe.c
3551 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3552 (rs6000_darwin64_record_arg_flush): Likewise.
3553 * config/rs6000/rs6000.c
3554 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3555 (rs6000_darwin64_record_arg_flush): Likewise.
3556 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3557 (sparc_function_value_1): Likewise.
3558 * config/spu/spu.c (adjust_operand): Likewise.
3559 (spu_emit_branch_or_set): Likewise.
3560 (arith_immediate_p): Likewise.
3561 * emit-rtl.c (gen_lowpart_common): Likewise.
3562 * expr.c (expand_expr_real_1): Likewise.
3563 * function.c (assign_parm_setup_block): Likewise.
3564 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3565 * reload1.c (alter_reg): Likewise.
3566 * stor-layout.c (mode_for_vector): Likewise.
3567 (layout_type): Likewise.
3569 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3571 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3572 (spu_convert_move): Likewise.
3573 * lower-subreg.c (resolve_simple_move): Likewise.
3575 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3578 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3579 define_insn to a define_expand.
3580 (altivec_vsum2sws_direct): New define_insn.
3581 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3583 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3585 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3588 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3593 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3594 realign stack if stack alignment needed is less than incoming
3597 2017-09-05 Marek Polacek <polacek@redhat.com>
3600 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3603 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3605 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3607 2017-09-05 Richard Biener <rguenther@suse.de>
3609 PR tree-optimization/82084
3610 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3612 2017-09-05 Richard Biener <rguenther@suse.de>
3614 PR tree-optimization/82102
3615 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3617 2017-09-05 Martin Liska <mliska@suse.cz>
3619 PR tree-optimization/82032
3620 * tree-cfg.c (generate_range_test): New function.
3621 * tree-cfg.h (generate_range_test): Declared here.
3622 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3623 (cleanup_control_expr_graph): Use it.
3624 * tree-switch-conversion.c (try_switch_expansion): Remove
3626 (emit_case_nodes): Use generate_range_test.
3628 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3631 * config/i386/i386.md (*<btsc><mode>_mask): Add
3632 TARGET_USE_BT to insn constraint.
3633 (*btr<mode>_mask): Ditto.
3635 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3637 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3638 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3640 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3643 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3644 TARGET_NEON and TARGET_IWMMXT.
3645 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3646 TARGET_NEON and TARGET_IWMMXT.
3647 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3649 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3651 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3652 (ix86_rewrite_tls_address): Ditto.
3653 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3654 (ix86_rewrite_tls_address_1): Ditto.
3655 (ix86_rewrite_tls_address): Ditto.
3656 * config/i386/predicates.md (tls_address_pattern): New predicate.
3657 * config/i386/i386.md (TLS address splitter): New splitter.
3659 2017-09-04 Richard Biener <rguenther@suse.de>
3661 PR tree-optimization/82084
3662 * fold-const.h (can_native_encode_string_p): Declare.
3663 * fold-const.c (can_native_encode_string_p): Factor out from ...
3664 (native_encode_string): ... here.
3665 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3666 vectorizing stores from constants we later cannot handle.
3668 2017-09-04 Marek Polacek <polacek@redhat.com>
3671 * doc/invoke.texi: Update -Wtautological-compare documentation.
3673 2017-09-04 Jeff Law <law@redhat.com>
3675 PR tree-optimization/64910
3676 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3677 swap the first and last operand if the last is a constant.
3679 2017-09-04 Marek Polacek <polacek@redhat.com>
3682 * convert.c (do_narrow): When sanitizing signed integer overflows,
3683 bail out for signed types.
3684 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3686 2017-09-04 Richard Biener <rguenther@suse.de>
3688 PR tree-optimization/82060
3689 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3690 Move devirtualization after stmt folding and before EH/AB/noreturn
3691 cleanup to get the stmt refs canonicalized. Use a bool instead
3692 of gimple_modified_p since that doesn't work for NOPs. Schedule
3693 NOPs generated by folding for removal.
3695 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3696 Alan Hayward <alan.hayward@arm.com>
3697 David Sherwood <david.sherwood@arm.com>
3699 * coretypes.h (pad_direction): New enum.
3700 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3701 (FUNCTION_ARG_PADDING): Likewise.
3702 * target.def (function_arg_padding): New hook.
3703 * targhooks.h (default_function_arg_padding): Declare.
3704 * targhooks.c (default_function_arg_padding): New function.
3705 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3706 (TARGET_FUNCTION_ARG_PADDING): ...this.
3707 * doc/tm.texi: Regenerate.
3708 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3709 instead of direction.
3710 (compute_argument_addresses): Likewise.
3711 (load_register_parameters): Likewise.
3712 (emit_library_call_value_1): Likewise.
3713 (store_one_arg): Use targetm.calls.function_arg_padding instead
3714 of FUNCTION_ARG_PADDING.
3715 (must_pass_in_stack_var_size_or_pad): Likewise.
3716 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3717 (emit_group_store): Likewise.
3718 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3719 instead of FUNCTION_ARG_PADDING.
3720 (emit_push_insn): Likewise, and propagate enum change throughout
3722 * function.h (direction): Delete.
3723 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3725 * function.c (assign_parm_find_stack_rtl): Likewise.
3726 (assign_parm_setup_block_p): Likewise.
3727 (assign_parm_setup_block): Likewise.
3728 (gimplify_parameters): Likewise.
3729 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3730 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3732 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3733 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3734 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3735 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3736 (aarch64_function_arg_padding): ...this new function.
3737 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3738 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3739 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3740 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3741 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3742 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3743 (arm_pad_arg_upward): Replace with...
3744 (arm_function_arg_padding): ...this new function.
3745 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3747 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3748 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3749 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3750 (ia64_hpux_function_arg_padding): Replace with...
3751 (ia64_function_arg_padding): ...this new function. Use pad_direction
3752 instead of direction. Check for TARGET_HPUX.
3753 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3754 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3755 (iq2000_function_arg_padding): New function.
3756 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3757 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3758 (mips_function_arg_padding): ...this new function.
3759 (mips_pad_reg_upward): Update accordingly.
3760 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3761 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3762 targetm.calls.function_arg_padding.
3763 (FUNCTION_ARG_PADDING): Delete.
3764 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3765 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3766 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3767 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3768 (nios2_block_reg_padding): Return pad_direction instead of direction.
3769 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3770 instead of direction.
3771 (nios2_function_arg_padding): Likewise. Make static.
3772 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3773 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3774 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3775 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3776 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3777 (pa_function_arg_padding): Make static. Return pad_direction instead
3779 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3780 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3781 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3782 instead of direction. Use targetm.calls.function_arg_padding.
3783 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3784 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3785 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3786 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3787 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3789 (function_arg_padding): Rename to...
3790 (rs6000_function_arg_padding): ...this. Make static. Return
3791 pad_direction instead of direction.
3792 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3793 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3794 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3795 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3796 instead of direction. Use targetm.calls.function_arg_padding.
3797 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3798 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3799 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3800 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3801 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3802 (function_arg_padding): Rename to...
3803 (rs6000_function_arg_padding): ...this. Make static. Return
3804 pad_direction instead of direction.
3805 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3806 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3807 * config/s390/s390.c (s390_function_arg_padding): New function.
3808 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3809 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3810 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3811 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3812 (function_arg_padding): Rename to...
3813 (sparc_function_arg_padding): ...this. Make static. Return
3814 pad_direction instead of direction.
3815 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3816 * config/spu/spu.c (spu_function_arg_padding): New function.
3817 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3818 * system.h (FUNCTION_ARG_PADDING): Poison.
3820 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3821 Alan Hayward <alan.hayward@arm.com>
3822 David Sherwood <david.sherwood@arm.com>
3824 * target.def (modes_tieable_p): New hook.
3825 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3826 (TARGET_MODES_TIEABLE_P): ...this.
3827 * doc/tm.texi.in: Regenerate.
3828 * hooks.h (hook_bool_mode_mode_true): Declare.
3829 * hooks.c (hook_bool_mode_mode_true): New function.
3830 * combine.c (subst): Use targetm.modes_tieable_p instead of
3832 * dse.c (find_shift_sequence): Likewise.
3833 * expmed.c (extract_low_bits): Likewise.
3834 * lower-subreg.c: Include target.h.
3835 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3837 * rtlanal.c (rtx_cost): Likewise.
3838 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3839 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3840 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3841 (TARGET_MODES_TIEABLE_P): Redefine.
3842 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3843 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3844 (TARGET_MODES_TIEABLE_P): Redefine.
3845 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3846 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3847 (arc_modes_tieable_p): New function.
3848 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3849 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3850 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3851 (arm_modes_tieable_p): Make static.
3852 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3853 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3854 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3855 (TARGET_MODES_TIEABLE_P): Redefine.
3856 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3857 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3858 (TARGET_MODES_TIEABLE_P): Redefine.
3859 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3860 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3861 (cr16_modes_tieable_p): New function.
3862 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3863 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3864 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3865 (TRULY_NOOP_TRUNCATION): Update comment.
3866 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3867 (TRULY_NOOP_TRUNCATION): Update comment.
3868 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3869 (frv_modes_tieable_p): New function.
3870 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3871 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3872 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3873 (TARGET_MODES_TIEABLE_P): Redefine.
3874 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3875 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3876 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3877 (TARGET_MODES_TIEABLE_P): Redefine.
3878 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3879 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3880 (ia64_modes_tieable_p): New function.
3881 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3882 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3883 (iq2000_modes_tieable_p): New function.
3884 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3885 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3886 (lm32_modes_tieable_p): New function.
3887 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3888 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3889 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3890 (TARGET_MODES_TIEABLE_P): Redefine.
3891 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3892 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3893 (m32r_modes_tieable_p): New function.
3894 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3895 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3896 (m68k_modes_tieable_p): New function.
3897 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3898 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3899 (mcore_modes_tieable_p): New function.
3900 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3901 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3903 (TARGET_MODES_TIEABLE_P): Redefine.
3904 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3905 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3906 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3907 (TARGET_MODES_TIEABLE_P): Redefine.
3908 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3909 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3910 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3911 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3912 (mn10300_modes_tieable_p): ...this and make static.
3913 (TARGET_MODES_TIEABLE_P): Redefine.
3914 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3915 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3916 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3917 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3918 (msp430_modes_tieable_p): Make static.
3919 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3920 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3921 (TARGET_MODES_TIEABLE_P): Redefine.
3922 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3923 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3924 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3925 (TARGET_MODES_TIEABLE_P): Redefine.
3926 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3927 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3928 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3929 (TARGET_MODES_TIEABLE_P): Redefine.
3930 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3931 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3932 (pdp11_modes_tieable_p): New function.
3933 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3934 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3935 (rs6000_modes_tieable_p): New function.
3936 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3937 * config/powerpcspe/powerpcspe.md: Update comment.
3938 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3939 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3940 (TARGET_MODES_TIEABLE_P): Redefine.
3941 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3942 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3943 (rl78_modes_tieable_p): New function.
3944 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3945 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3946 (rs6000_modes_tieable_p): New function.
3947 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3948 * config/rs6000/rs6000.md: Update comment.
3949 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3950 * config/rx/rx.c (rx_modes_tieable_p): New function.
3951 (TARGET_MODES_TIEABLE_P): Redefine.
3952 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3953 * config/s390/s390.c (s390_modes_tieable_p): New function.
3954 (TARGET_MODES_TIEABLE_P): Redefine.
3955 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3956 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3957 (sh_modes_tieable_p): New function.
3958 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3959 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3960 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3961 (sparc_modes_tieable_p): Make static.
3962 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3963 * config/spu/spu.c (spu_modes_tieable_p): New function.
3964 (TARGET_MODES_TIEABLE_P): Redefine.
3965 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3966 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3967 (TARGET_MODES_TIEABLE_P): Redefine.
3968 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3969 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3970 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3971 * config/v850/v850.c (v850_modes_tieable_p): New function.
3972 (TARGET_MODES_TIEABLE_P): Redefine.
3973 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3974 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3975 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3976 (visium_modes_tieable_p): New function.
3977 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3978 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3979 (xtensa_modes_tieable_p): New function.
3980 * system.h (MODES_TIEABLE_P): Poison.
3982 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3983 Alan Hayward <alan.hayward@arm.com>
3984 David Sherwood <david.sherwood@arm.com>
3986 * target.def (hard_regno_mode_ok): New hook.
3987 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3988 (TARGET_HARD_REGNO_MODE_OK): ...this.
3989 * doc/tm.texi.in: Regenerate.
3990 * hooks.h (hook_bool_uint_mode_true): Declare.
3991 * hooks.c (hook_bool_uint_mode_true): New function.
3992 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3994 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3995 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3996 instead of HARD_REGNO_MODE_OK.
3997 * caller-save.c: Include target.h.
3998 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4000 * combine.c (can_combine_p): Likewise.
4001 (combinable_i3pat): Likewise.
4002 (can_change_dest_mode): Likewise.
4003 * expr.c (init_expr_target): Likewise.
4004 (convert_move): Likewise.
4005 (convert_modes): Likewise.
4006 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4007 (setup_prohibited_mode_move_regs): Likewise.
4008 * ira.h (target_ira): Likewise.
4009 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4010 * lra-constraints.c (process_alt_operands): Likewise.
4011 (split_reg): Likewise.
4012 * recog.c (peep2_find_free_register): Likewise.
4013 * ree.c (combine_reaching_defs): Likewise.
4014 * regcprop.c (maybe_mode_change): Likewise.
4015 * reginfo.c (init_reg_sets_1): Likewise.
4016 (choose_hard_reg_mode): Likewise.
4017 (simplifiable_subregs): Likewise.
4018 * regrename.c (check_new_reg_p): Likewise.
4019 * reload.c (find_valid_class): Likewise.
4020 (find_valid_class_1): Likewise.
4021 (reload_inner_reg_of_subreg): Likewise.
4022 (push_reload): Likewise.
4023 (combine_reloads): Likewise.
4024 (find_dummy_reload): Likewise.
4025 (find_reloads): Likewise.
4026 * reload1.c (find_reg): Likewise.
4027 (set_reload_reg): Likewise.
4028 (allocate_reload_reg): Likewise.
4029 (choose_reload_regs): Likewise.
4030 (reload_adjust_reg_for_temp): Likewise.
4031 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4032 (simplify_subreg_regno): Likewise.
4033 * sel-sched.c (init_regs_for_mode): Likewise.
4034 * varasm.c (make_decl_rtl): Likewise.
4035 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4036 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4038 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4039 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4040 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4041 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4042 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4043 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4044 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4045 (arc_mode_class): Delete.
4046 (HARD_REGNO_MODE_OK): Delete.
4047 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4048 (arc_hard_regno_mode_ok): Rename old array to...
4049 (arc_hard_regno_mode_ok_modes): ...this.
4050 (arc_conditional_register_usage): Update accordingly.
4051 (arc_mode_class): Make static.
4052 (arc_hard_regno_mode_ok): New function.
4053 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4054 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4055 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4056 (arm_hard_regno_mode_ok): Make static.
4057 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4059 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4060 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4061 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4063 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4064 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4065 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4066 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4067 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4068 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4069 * config/bfin/predicates.md (valid_reg_operand): Use
4070 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4071 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4072 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4073 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4074 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4075 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4076 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4077 (cr16_hard_regno_mode_ok): Make static and return a bool.
4078 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4079 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4080 (cris_hard_regno_mode_ok): New function.
4081 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4082 (epiphany_mode_class): Delete.
4083 (HARD_REGNO_MODE_OK): Delete.
4084 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4085 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4086 (hard_regno_mode_ok): Rename to...
4087 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4088 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4089 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4091 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4092 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4093 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4094 (frv_hard_regno_mode_ok): Make static and return a bool.
4095 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4097 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4098 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4099 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4100 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4102 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4103 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4104 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4105 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4107 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4108 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4109 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4110 (ia64_hard_regno_mode_ok): New function.
4111 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4112 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4113 (iq2000_hard_regno_mode_ok): New function.
4114 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4115 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4116 (lm32_hard_regno_mode_ok): New function.
4117 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4118 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4119 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4120 instead of HARD_REGNO_MODE_OK.
4121 (m32c_hard_regno_ok): Rename to...
4122 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4123 (m32c_cannot_change_mode_class): Update accordingly.
4124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4125 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4126 (m32r_mode_class): Delete.
4127 (HARD_REGNO_MODE_OK): Delete.
4128 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4129 (m32r_hard_regno_mode_ok): Rename to...
4130 (m32r_hard_regno_modes): ...this.
4131 (m32r_mode_class): Make static.
4132 (m32r_hard_regno_mode_ok): New function.
4133 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4134 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4135 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4136 (m68k_hard_regno_mode_ok): Make static.
4137 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4138 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4139 (mcore_hard_regno_mode_ok): New function.
4140 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4141 (HARD_REGNO_MODE_OK): Delete.
4142 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4144 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4145 (microblaze_hard_regno_mode_ok): New function.
4146 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4147 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4148 (mips_hard_regno_mode_ok): Delete.
4149 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4150 (mips_hard_regno_mode_ok_p): ...this and make static.
4151 (mips_hard_regno_mode_ok_p): Rename to...
4152 (mips_hard_regno_mode_ok_uncached): ...this.
4153 (mips_hard_regno_mode_ok): New function.
4154 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4155 of HARD_REGNO_MODE_OK.
4156 (mips_option_override): Update after above name changes.
4157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4158 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4159 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4160 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4161 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4162 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4163 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4164 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4165 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4166 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4167 (msp430_hard_regno_mode_ok): Make static and return a bool.
4168 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4169 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4170 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4172 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4173 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4174 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4175 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4176 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4177 (PA_HARD_REGNO_MODE_OK): ...this
4178 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4179 (PA_HARD_REGNO_MODE_OK): ...this.
4180 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4181 (pa_hard_regno_mode_ok): New function.
4182 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4183 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4184 (pdp11_hard_regno_mode_ok): New function.
4185 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4186 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4188 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4190 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4191 (rs6000_hard_regno_mode_ok): Rename to...
4192 (rs6000_hard_regno_mode_ok_uncached): ...this.
4193 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4194 (rs6000_hard_regno_mode_ok): New function.
4195 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4196 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4197 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4198 (riscv_hard_regno_mode_ok): ...this and make static.
4199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4200 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4201 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4202 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4203 (rl78_hard_regno_mode_ok): Make static and return bool.
4204 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4205 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4207 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4208 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4209 (rs6000_hard_regno_mode_ok): Rename to...
4210 (rs6000_hard_regno_mode_ok_uncached): ...this.
4211 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4212 (rs6000_hard_regno_mode_ok): New function.
4213 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4214 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4215 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4216 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4217 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4218 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4219 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4220 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4221 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4222 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4223 (sh_hard_regno_mode_ok): Make static.
4224 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4225 instead of HARD_REGNO_MODE_OK.
4226 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4227 (sparc_mode_class): Delete.
4228 (HARD_REGNO_MODE_OK): Delete.
4229 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4230 (hard_regno_mode_classes): Make static.
4231 (sparc_mode_class): Likewise.
4232 (sparc_hard_regno_mode_ok): New function.
4233 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4234 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4235 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4237 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4238 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4239 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4240 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4241 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4243 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4244 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4245 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4246 (visium_hard_regno_mode_ok): New function.
4247 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4248 instead of HARD_REGNO_MODE_OK.
4249 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4250 (HARD_REGNO_MODE_OK): Delete.
4251 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4252 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4253 (xtensa_option_override): Update accordingly.
4254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4255 (xtensa_hard_regno_mode_ok): New function.
4256 * system.h (HARD_REGNO_MODE_OK): Poison.
4258 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4259 Alan Hayward <alan.hayward@arm.com>
4260 David Sherwood <david.sherwood@arm.com>
4262 * target.def (hard_regno_call_part_clobbered): New hook.
4263 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4264 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4265 * doc/tm.texi: Regenerate.
4266 * hooks.h (hook_bool_uint_mode_false): Declare.
4267 * hooks.c (hook_bool_uint_mode_false): New function.
4268 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4269 * cselib.c (cselib_process_insn): Use
4270 targetm.hard_regno_call_part_clobbered instead of
4271 HARD_REGNO_CALL_PART_CLOBBERED.
4272 * ira-conflicts.c (ira_build_conflicts): Likewise.
4273 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4274 * lra-constraints.c (need_for_call_save_p): Likewise.
4275 * lra-lives.c: Include target.h.
4276 (check_pseudos_live_through_calls): Use
4277 targetm.hard_regno_call_part_clobbered instead of
4278 HARD_REGNO_CALL_PART_CLOBBERED.
4279 * regcprop.c: Include target.h.
4280 (copyprop_hardreg_forward_1): Use
4281 targetm.hard_regno_call_part_clobbered instead of
4282 HARD_REGNO_CALL_PART_CLOBBERED.
4283 * reginfo.c (choose_hard_reg_mode): Likewise.
4284 * regrename.c (check_new_reg_p): Likewise.
4285 * reload.c (find_equiv_reg): Likewise.
4286 * reload1.c (emit_reload_insns): Likewise.
4287 * sched-deps.c (deps_analyze_insn): Likewise.
4288 * sel-sched.c (init_regs_for_mode): Likewise.
4289 (mark_unavailable_hard_regs): Likewise.
4290 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4291 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4292 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4294 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4295 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4296 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4298 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4300 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4301 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4302 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4304 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4305 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4306 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4308 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4309 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4311 * config/powerpcspe/powerpcspe.c
4312 (rs6000_hard_regno_call_part_clobbered): New function.
4313 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4314 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4315 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4317 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4318 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4319 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4321 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4322 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4323 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4325 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4326 Alan Hayward <alan.hayward@arm.com>
4327 David Sherwood <david.sherwood@arm.com>
4329 * rtl.h (subreg_memory_offset): Declare.
4330 * emit-rtl.c (subreg_memory_offset): New function.
4331 * expmed.c (store_bit_field_1): Use it.
4332 * expr.c (undefined_operand_subword_p): Likewise.
4333 * simplify-rtx.c (simplify_subreg): Likewise.
4335 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4337 PR rtl-optimization/57448
4340 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4341 using atomic_load pattern.
4342 (expand_atomic_store): Likewise.
4344 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4347 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4348 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4349 handling. Use replace_call_with_value with NULL instead of
4350 gsi_replace, unlink_stmt_vdef and release_defs.
4352 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4355 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4357 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4360 * rtl.h (emit_library_call_value_1): Declare.
4361 (emit_library_call): Replace declaration with a series of overloads.
4362 Remove the parameter count argument.
4363 (emit_library_call_value): Likewise.
4364 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4365 with an "rtx_mode_t *".
4366 (emit_library_call_value): Delete.
4367 (emit_library_call): Likewise.
4368 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4369 (asan_emit_allocas_unpoison): Likewise.
4370 * builtins.c (expand_builtin_powi): Likewise.
4371 (expand_asan_emit_allocas_unpoison): Likewise.
4372 * cfgexpand.c (expand_main_function): Likewise.
4373 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4374 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4375 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4376 * config/arm/arm.c (arm_trampoline_init): Likewise.
4377 (arm_call_tls_get_addr): Likewise.
4378 (arm_expand_divmod_libfunc): Likewise.
4379 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4380 (smulsi3_highpart): Likewise.
4381 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4382 (c6x_expand_compare): Likewise.
4383 (c6x_expand_movmem): Likewise.
4384 * config/frv/frv.c (frv_trampoline_init): Likewise.
4385 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4386 (ix86_expand_divmod_libfunc): Likewise.
4387 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4388 (ia64_expand_compare): Likewise.
4389 (ia64_profile_hook): Likewise.
4390 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4391 (nonlocal_goto): Likewise.
4392 (restore_stack_nonlocal): Likewise.
4393 * config/m32r/m32r.c (block_move_call): Likewise.
4394 (m32r_trampoline_init): Likewise.
4395 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4396 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4397 (m68k_call_m68k_read_tp): Likewise.
4398 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4399 (microblaze_expand_divide): Likewise.
4400 * config/mips/mips.h (mips_args): Likewise.
4401 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4402 (MIPS_ICACHE_SYNC): Likewise.
4403 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4404 (nios2_trampoline_init): Likewise.
4405 * config/pa/pa.c (hppa_tls_call): Likewise.
4406 (pa_trampoline_init): Likewise.
4407 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4408 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4409 (expand_strn_compare): Likewise.
4410 (rs6000_generate_compare): Likewise.
4411 (rs6000_expand_float128_convert): Likewise.
4412 (output_profile_hook): Likewise.
4413 (rs6000_trampoline_init): Likewise.
4414 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4415 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4416 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4417 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4418 (rs6000_generate_compare): Likewise.
4419 (rs6000_expand_float128_convert): Likewise.
4420 (output_profile_hook): Likewise.
4421 (rs6000_trampoline_init): Likewise.
4422 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4423 * config/sh/sh.c (sh_trampoline_init): Likewise.
4424 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4425 (sparc_emit_float_lib_cmp): Likewise.
4426 (sparc32_initialize_trampoline): Likewise.
4427 (sparc64_initialize_trampoline): Likewise.
4428 (sparc_profile_hook): Likewise.
4429 * config/spu/spu.c (ea_load_store): Likewise.
4430 * config/spu/spu.md (floatunssidf2): Likewise.
4431 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4432 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4433 * config/visium/visium.c (expand_block_move_4): Likewise.
4434 (expand_block_move_2): Likewise.
4435 (expand_block_move_1): Likewise.
4436 (expand_block_set_4): Likewise.
4437 (expand_block_set_2): Likewise.
4438 (expand_block_set_1): Likewise.
4439 (visium_trampoline_init): Likewise.
4440 (visium_profile_hook): Likewise.
4441 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4442 (xtensa_setup_frame_addresses): Likewise.
4443 (xtensa_trampoline_init): Likewise.
4444 * except.c (sjlj_emit_function_enter): Likewise.
4445 (sjlj_emit_function_exit): Likewise.
4446 * explow.c (allocate_dynamic_stack_space): Likewise.
4447 (probe_stack_range): Likewise.
4448 * expr.c (convert_mode_scalar): Likewise.
4449 * optabs.c (expand_binop): Likewise.
4450 (expand_twoval_binop_libfunc): Likewise.
4451 (expand_unop): Likewise.
4452 (prepare_cmp_insn): Likewise.
4453 (prepare_float_lib_cmp): Likewise.
4454 (expand_float): Likewise.
4455 (expand_fix): Likewise.
4456 (expand_fixed_convert): Likewise.
4457 (maybe_emit_sync_lock_test_and_set): Likewise.
4458 (expand_atomic_compare_and_swap): Likewise.
4459 (expand_mem_thread_fence): Likewise.
4460 (expand_atomic_fetch_op): Likewise.
4462 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4464 * doc/generic.texi (OpenACC): Adjust URL.
4465 * doc/invoke.texi (C Dialect Options): Ditto.
4467 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4469 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4470 predicate for operand 1. Add (m,<S>) constraint.
4471 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4472 Prevent memory operand 1 with register operand 2.
4474 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4476 PR rtl-optimization/82024
4477 * combine.c (try_combine): If the combination result is a PARALLEL,
4478 and we only need to retain the SET in there that would be placed
4479 at I2, check that we can place that at I3 instead, before doing so.
4481 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4484 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4486 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4487 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4490 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4491 Jeff Law <law@redhat.com>
4493 * varasm.c (bss_initializer_p): Do not put constants into .bss
4494 (categorize_decl_for_section): Handle bss_initializer_p returning
4495 false when DECL_INITIAL is NULL.
4497 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4500 * config/s390/s390.c (s390_can_inline_p): New function.
4502 2017-09-01 Jeff Law <law@redhat.com>
4504 PR tree-optimization/82052
4505 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4506 Always initialize the returned slot after a hash table miss
4507 when INSERT is true.
4509 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4511 * config/s390/s390.md (mem_signal_fence): Remove.
4512 * doc/md.texi (mem_signal_fence): Remove.
4513 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4515 * target-insns.def (mem_signal_fence): Remove.
4517 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4520 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4523 * asan.c (create_odr_indicator): Strip name encoding from assembler
4524 name before appending it after __odr_asan_.
4526 2017-09-01 Martin Liska <mliska@suse.cz>
4528 PR tree-optimization/82059
4529 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4530 frequency only when an edge is redirected.
4532 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4534 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4535 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4536 (arc_conditional_register_usage): Remove ARC600 lp_count
4538 (arc_file_start): Emit Tag_ARC_CPU_variation.
4539 (arc_can_use_doloop_p): New conditions to use ZOLs.
4540 (hwloop_fail): New function.
4541 (hwloop_optimize): Likewise.
4542 (hwloop_pattern_reg): Likewise.
4543 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4544 (arc_reorg_loops): New function, calls reorg_loops.
4545 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4546 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4548 (arc_loop_hazard): Remove function, functionality moved into
4550 (arc_hazard): Remove arc_loop_hazard call.
4551 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4552 into hwloop_optimize.
4553 (arc_label_align): Remove ZOL handling.
4554 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4555 * config/arc/arc.md (doloop_begin): Remove pattern.
4556 (doloop_begin_i): Likewise.
4557 (doloop_end_i): Likewise.
4558 (doloop_fallback): Likewise.
4559 (doloop_fallback_m): Likewise.
4560 (doloop_end): Reimplement expand.
4561 (arc_lp): New pattern for LP instruction.
4562 (loop_end): New pattern.
4563 (loop_fail): Likewise.
4564 (decrement_and_branch_until_zero): Likewise.
4565 * config/arc/arc.opt (mlpc-width): New option.
4566 * doc/invoke.texi (mlpc-width): Document option.
4568 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4570 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4571 (arc_ccfsm_advance): Fix checking for delay slots.
4572 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4574 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4576 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4578 (movhi_insn): Update store instruction constraint which are saving
4579 6-bit short immediates.
4580 (movsi_insn): Consider also short scaled load operations.
4581 (zero_extendhisi2_i): Use Usd constraint instead of T.
4582 (extendhisi2_i): Add q constraint.
4583 (arc_clzsi2): Add type and length attributes.
4584 (arc_ctzsi2): Likewise.
4585 * config/arc/constraints.md (Usc): Update constraint, the
4586 assembler can parse two relocations for a single instruction.
4588 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4590 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4591 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4593 2017-08-31 Olivier Hainque <hainque@adacore.com>
4595 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4596 match as powerpc-wrs-vxworks*.
4598 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4600 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4601 register constraint for by-element operand.
4602 (aarch64_mls_elt_merge<mode>): Likewise.
4604 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4606 * config/arc/arc.c (arc_can_follow_jump): Check for short
4609 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4611 * config.gcc: Use g.opt for arc.
4612 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4613 functionality moved to ...
4614 (legitimate_scaled_address_p): New function, ...here.
4615 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4616 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4617 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4619 (arc_override_options): Handle G option.
4620 (arc_output_pic_addr_const): Correct function definition.
4621 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4622 (arc_decl_anon_ns_mem_p): Delete.
4623 (arc_in_small_data_p): Overhaul this function to take into
4624 consideration the value given via G option.
4625 (arc_rewrite_small_data_1): Renamed and corrected old
4626 arc_rewrite_small_data function.
4627 (arc_rewrite_small_data): New function.
4628 (small_data_pattern): Don't use pic_offset_table_rtx.
4629 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4630 * config/arc/simdext.md (movmisalignv2hi): Use
4631 prepare_move_operands function.
4633 (movmisalign*): Likewise.
4634 * doc/invoke.texi (ARC options): Document -G option.
4636 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4638 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4640 * config/arc/arc.c (arc_print_operand): Output scalled address for
4641 sdata whenever is possible.
4642 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4643 load/stores are available.
4644 (compact_sda_memory_operand): Check for the alignment required by
4645 code density instructions.
4646 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4648 * config/arc/constraints.md (Usd): Update constraint.
4649 (Us0): New constraint.
4650 (Usc): Update constraint.
4652 2017-08-31 Richard Biener <rguenther@suse.de>
4655 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4658 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4660 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4661 Resize type_signature.
4663 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4664 Alan Hayward <alan.hayward@arm.com>
4665 David Sherwood <david.sherwood@arm.com>
4667 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4668 subregs whose inner modes can be stored in GPRs.
4669 (aarch64_classify_index): Likewise.
4671 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4672 Alan Hayward <alan.hayward@arm.com>
4673 David Sherwood <david.sherwood@arm.com>
4675 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4676 (V_INT_EQUIV): ...this.
4677 (v_cmp_result): Rename to...
4678 (v_int_equiv): ...this.
4679 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4680 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4681 (copysign<mode>3): Likewise.
4682 (aarch64_simd_bsl<mode>_internal): Likewise.
4683 (aarch64_simd_bsl<mode>): Likewise.
4684 (vec_cmp<mode><mode>): Likewise.
4685 (vcond<mode><mode>): Likewise.
4686 (vcond<v_cmp_mixed><mode>): Likewise.
4687 (vcondu<mode><v_cmp_mixed>): Likewise.
4688 (aarch64_cm<optab><mode>): Likewise.
4689 (aarch64_cmtst<mode>): Likewise.
4690 (aarch64_fac<optab><mode>): Likewise.
4691 (vec_perm_const<mode>): Likewise.
4692 (vcond_mask_<mode><v_cmp_result>): Rename to...
4693 (vcond_mask_<mode><v_int_equiv>): ...this.
4694 (vec_cmp<mode><v_cmp_result>): Rename to...
4695 (vec_cmp<mode><v_int_equiv>): ...this.
4697 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4698 Alan Hayward <alan.hayward@arm.com>
4699 David Sherwood <david.sherwood@arm.com>
4701 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4703 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4704 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4705 (UNSPEC_LD4_DREG): New unspecs.
4706 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4707 (aarch64_ld2<mode>_dreg_be): Replace with...
4708 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4710 (aarch64_ld3<mode>_dreg_le)
4711 (aarch64_ld3<mode>_dreg_be): Replace with...
4712 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4714 (aarch64_ld4<mode>_dreg_le)
4715 (aarch64_ld4<mode>_dreg_be): Replace with...
4716 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4719 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4721 PR tree-optimization/81987
4722 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4723 insert an initializer in a location not dominated by the stride
4726 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4728 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4729 on the entire header of the finally block in the fallthru case.
4731 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4733 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4735 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4737 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4738 rs6000_emit_move_from_cr and call renamed function.
4739 (rs6000_emit_prologue): Call renamed functions.
4740 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4741 movesi_from_cr, remove volatile CRs.
4743 2017-08-30 Jon Beniston <jon@beniston.com>
4744 Richard Biener <rguenther@suse.de>
4746 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4747 of VECTOR_MODE_P check.
4748 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4749 element vector types.
4751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4753 * df.h (df_read_modify_subreg_p): Remove in favor of...
4754 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4755 const_rtx instead of an rtx.
4756 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4757 * df-problems.c (df_word_lr_mark_ref): Likewise.
4758 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4759 (mark_pseudo_reg_dead): Likewise.
4760 (mark_ref_dead): Likewise.
4761 * reginfo.c (init_subregs_of_mode): Likewise.
4762 * sched-deps.c (sched_analyze_1): Likewise.
4763 * df-scan.c (df_def_record_1): Likewise.
4764 (df_uses_record): Likewise.
4765 (df_read_modify_subreg_p): Remove in favor of...
4766 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4767 const_rtx instead of an rtx.
4769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4770 Alan Hayward <alan.hayward@arm.com>
4771 David Sherwood <david.sherwood@arm.com>
4773 * rtl.h (partial_subreg_p): New function.
4774 * caller-save.c (save_call_clobbered_regs): Use it.
4775 * calls.c (expand_call): Likewise.
4776 * combine.c (combinable_i3pat): Likewise.
4777 (simplify_set): Likewise.
4778 (make_extraction): Likewise.
4779 (make_compound_operation_int): Likewise.
4780 (gen_lowpart_or_truncate): Likewise.
4781 (force_to_mode): Likewise.
4782 (make_field_assignment): Likewise.
4783 (reg_truncated_to_mode): Likewise.
4784 (record_truncated_value): Likewise.
4785 (move_deaths): Likewise.
4786 * cse.c (record_jump_cond): Likewise.
4787 (cse_insn): Likewise.
4788 * cselib.c (cselib_lookup_1): Likewise.
4789 * expmed.c (extract_bit_field_using_extv): Likewise.
4790 * function.c (assign_parm_setup_reg): Likewise.
4791 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4792 * ira-build.c (create_insn_allocnos): Likewise.
4793 * lra-coalesce.c (merge_pseudos): Likewise.
4794 * lra-constraints.c (match_reload): Likewise.
4795 (simplify_operand_subreg): Likewise.
4796 (curr_insn_transform): Likewise.
4797 * lra-lives.c (process_bb_lives): Likewise.
4798 * lra.c (new_insn_reg): Likewise.
4799 (lra_substitute_pseudo): Likewise.
4800 * regcprop.c (mode_change_ok): Likewise.
4801 (maybe_mode_change): Likewise.
4802 (copyprop_hardreg_forward_1): Likewise.
4803 * reload.c (push_reload): Likewise.
4804 (find_reloads): Likewise.
4805 (find_reloads_subreg_address): Likewise.
4806 * reload1.c (alter_reg): Likewise.
4807 (eliminate_regs_1): Likewise.
4808 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4810 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4812 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4813 back to if statements, including unpack.
4815 2017-08-30 Martin Liska <mliska@suse.cz>
4818 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4820 (func_checker::compare_asm_inputs_outputs): ... this function.
4821 (func_checker::compare_gimple_asm): Use the function to compare
4822 also ASM constrains.
4823 * ipa-icf-gimple.h: Rename the function.
4825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4826 Alan Hayward <alan.hayward@arm.com>
4827 David Sherwood <david.sherwood@arm.com>
4829 * coretypes.h (complex_mode): New type.
4830 * gdbhooks.py (build_pretty_printer): Handle it.
4831 * machmode.h (complex_mode): New class.
4832 (complex_mode::includes_p): New function.
4833 (is_complex_int_mode): Likewise.
4834 (is_complex_float_mode): Likewise.
4835 * genmodes.c (get_mode_class): Handle complex mode classes.
4836 * function.c (expand_function_end): Use is_complex_int_mode.
4838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4839 Alan Hayward <alan.hayward@arm.com>
4840 David Sherwood <david.sherwood@arm.com>
4842 * coretypes.h (scalar_mode_pod): New typedef.
4843 * gdbhooks.py (build_pretty_printer): Handle it.
4844 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4845 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4846 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4847 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4851 Alan Hayward <alan.hayward@arm.com>
4852 David Sherwood <david.sherwood@arm.com>
4854 * machmode.h (mode_for_vector): Take a scalar_mode instead
4856 * stor-layout.c (mode_for_vector): Likewise.
4857 * explow.c (promote_mode): Use as_a <scalar_mode>.
4858 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4861 Alan Hayward <alan.hayward@arm.com>
4862 David Sherwood <david.sherwood@arm.com>
4864 * target.def (preferred_simd_mode): Take a scalar_mode
4865 instead of a machine_mode.
4866 * targhooks.h (default_preferred_simd_mode): Likewise.
4867 * targhooks.c (default_preferred_simd_mode): Likewise.
4868 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4869 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4870 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4871 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4872 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4873 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4874 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4875 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4877 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4878 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4879 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4880 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4881 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4882 * doc/tm.texi: Regenerate.
4883 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4887 Alan Hayward <alan.hayward@arm.com>
4888 David Sherwood <david.sherwood@arm.com>
4890 * target.def (scalar_mode_supported_p): Take a scalar_mode
4891 instead of a machine_mode.
4892 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4893 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4894 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4895 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4896 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4897 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4898 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4899 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4900 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4901 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4902 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4903 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4904 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4905 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4907 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4908 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4909 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4910 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4911 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4913 * doc/tm.texi: Regenerate.
4915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4916 Alan Hayward <alan.hayward@arm.com>
4917 David Sherwood <david.sherwood@arm.com>
4919 * coretypes.h (opt_scalar_mode): New typedef.
4920 * gdbhooks.py (build_pretty_printers): Handle it.
4921 * machmode.h (mode_iterator::get_2xwider): Add overload for
4923 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4925 * expr.c (convert_mode_scalar): Likewise.
4926 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4927 * optabs.c (expand_float): Likewise.
4928 (expand_fix): Likewise.
4929 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
4936 using a branching expansion.
4937 (expand_fix): Likewise.
4939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4940 Alan Hayward <alan.hayward@arm.com>
4941 David Sherwood <david.sherwood@arm.com>
4943 * expr.c (convert_mode): Split scalar handling out into...
4944 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4953 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4956 Alan Hayward <alan.hayward@arm.com>
4957 David Sherwood <david.sherwood@arm.com>
4959 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4960 rather than a machine_mode.
4961 (fixed_from_string): Likewise.
4962 (fixed_convert): Likewise.
4963 (fixed_convert_from_int): Likewise.
4964 (fixed_convert_from_real): Likewise.
4965 (real_convert_from_fixed): Likewise.
4966 * fixed-value.c (fixed_from_double_int): Likewise.
4967 (fixed_from_string): Likewise.
4968 (fixed_convert): Likewise.
4969 (fixed_convert_from_int): Likewise.
4970 (fixed_convert_from_real): Likewise.
4971 (real_convert_from_fixed): Likewise.
4972 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4975 Alan Hayward <alan.hayward@arm.com>
4976 David Sherwood <david.sherwood@arm.com>
4978 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4979 of separate mode class checks. Do not allow vector modes here.
4980 (immed_wide_int_const): Use as_a <scalar_mode>.
4981 * explow.c (trunc_int_for_mode): Likewise.
4982 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4983 (wi::shwi): Likewise.
4984 (wi::min_value): Likewise.
4985 (wi::max_value): Likewise.
4986 * dwarf2out.c (loc_descriptor): Likewise.
4987 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4991 Alan Hayward <alan.hayward@arm.com>
4992 David Sherwood <david.sherwood@arm.com>
4994 * tree.h (SCALAR_TYPE_MODE): New macro.
4995 * expr.c (expand_expr_addr_expr_1): Use it.
4996 (expand_expr_real_2): Likewise.
4997 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4998 (fold_convert_const_fixed_from_int): Likewise.
4999 (fold_convert_const_fixed_from_real): Likewise.
5000 (native_encode_fixed): Likewise
5001 (native_encode_complex): Likewise
5002 (native_encode_vector): Likewise.
5003 (native_interpret_fixed): Likewise.
5004 (native_interpret_real): Likewise.
5005 (native_interpret_complex): Likewise.
5006 (native_interpret_vector): Likewise.
5007 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5008 (simd_clone_adjust_argument_types): Likewise.
5009 (simd_clone_init_simd_arrays): Likewise.
5010 (simd_clone_adjust): Likewise.
5011 * stor-layout.c (layout_type): Likewise.
5012 * tree.c (build_minus_one_cst): Likewise.
5013 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5014 * tree-inline.c (estimate_move_cost): Likewise.
5015 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5016 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5017 (vectorizable_reduction): Likewise.
5018 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5019 (vect_recog_mixed_size_cond_pattern): Likewise.
5020 (check_bool_pattern): Likewise.
5021 (adjust_bool_pattern): Likewise.
5022 (search_type_for_mask_1): Likewise.
5023 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5024 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5025 (vectorizable_load): Likewise.
5026 (vectorizable_store): Likewise.
5027 * ubsan.c (ubsan_encode_value): Likewise.
5028 * varasm.c (output_constant): Likewise.
5030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5031 Alan Hayward <alan.hayward@arm.com>
5032 David Sherwood <david.sherwood@arm.com>
5034 * coretypes.h (scalar_mode): New class.
5035 * machmode.h (scalar_mode): Likewise.
5036 (scalar_mode::includes_p): New function.
5037 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5038 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5039 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5040 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5041 * expmed.c (store_bit_field_1): Likewise.
5042 (extract_bit_field_1): Likewise.
5043 * expr.c (write_complex_part): Likewise.
5044 (read_complex_part): Likewise.
5045 (emit_move_complex_push): Likewise.
5046 (expand_expr_real_2): Likewise.
5047 * function.c (assign_parm_setup_reg): Likewise.
5048 (assign_parms_unsplit_complex): Likewise.
5049 * optabs.c (expand_binop): Likewise.
5050 * rtlanal.c (subreg_get_info): Likewise.
5051 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5052 * varasm.c (output_constant_pool_2): Likewise.
5054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5055 Alan Hayward <alan.hayward@arm.com>
5056 David Sherwood <david.sherwood@arm.com>
5058 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5060 (expmed_mult_highpart_optab): Likewise.
5061 (expmed_mult_highpart): Likewise.
5063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5064 Alan Hayward <alan.hayward@arm.com>
5065 David Sherwood <david.sherwood@arm.com>
5067 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5068 instead of a machine_mode.
5069 (builtin_memset_read_str): Likewise.
5070 * builtins.c (c_readstr): Likewise.
5071 (builtin_memcpy_read_str): Likewise.
5072 (builtin_strncpy_read_str): Likewise.
5073 (builtin_memset_read_str): Likewise.
5074 (builtin_memset_gen_str): Likewise.
5075 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5076 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5077 instead of a machine_mode.
5078 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5080 (make_extraction): Likewise.
5081 (try_widen_shift_mode): Take and return scalar_int_modes instead
5083 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5084 a scalar_int_mode instead of a machine_mode.
5085 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5086 (avr_addr_space_pointer_mode): Likewise.
5087 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5088 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5089 (msp430_unwind_word_mode): Likewise.
5090 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5091 (spu_addr_space_pointer_mode): Likewise.
5092 (spu_addr_space_address_mode): Likewise.
5093 (spu_libgcc_cmp_return_mode): Likewise.
5094 (spu_libgcc_shift_count_mode): Likewise.
5095 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5096 (rl78_addr_space_pointer_mode): Likewise.
5097 (fl78_unwind_word_mode): Likewise.
5098 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5100 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5101 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5102 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5103 (mips_valid_pointer_mode): Likewise.
5104 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5105 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5106 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5108 (ft32_addr_space_address_mode): Likewise.
5109 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5110 scalar_int_mode instead of a machine_mode.
5111 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5113 (m32c_addr_space_address_mode): Likewise.
5114 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5115 (rs6000_eh_return_filter_mode): Likewise.
5116 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5117 (rs6000_eh_return_filter_mode): Likewise.
5118 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5119 (s390_libgcc_shift_count_mode): Likewise.
5120 (s390_unwind_word_mode): Likewise.
5121 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5123 * target.def (mode_rep_extended): Likewise.
5124 (valid_pointer_mode): Likewise.
5125 (addr_space.valid_pointer_mode): Likewise.
5126 (eh_return_filter_mode): Return a scalar_int_mode rather than
5128 (libgcc_cmp_return_mode): Likewise.
5129 (libgcc_shift_count_mode): Likewise.
5130 (unwind_word_mode): Likewise.
5131 (addr_space.pointer_mode): Likewise.
5132 (addr_space.address_mode): Likewise.
5133 * doc/tm.texi: Regenerate.
5134 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5136 (do_jump): Use scalar_int_mode for local variables.
5137 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5138 rather than a machine_mode.
5139 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5140 (scompare_loc_descriptor_wide): Likewise.
5141 (scompare_loc_descriptor_narrow): Likewise.
5142 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5144 * except.c (sjlj_emit_dispatch_table): Likewise.
5145 (expand_builtin_eh_copy_values): Likewise.
5146 * explow.c (convert_memory_address_addr_space_1): Likewise.
5147 Take a scalar_int_mode rather than a machine_mode.
5148 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5149 than a machine_mode.
5150 (memory_address_addr_space): Use scalar_int_mode for local variables.
5151 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5152 rather than a machine_mode.
5153 * expmed.c (mask_rtx): Likewise.
5154 (init_expmed_one_conv): Likewise.
5155 (expand_mult_highpart_adjust): Likewise.
5156 (extract_high_half): Likewise.
5157 (expmed_mult_highpart_optab): Likewise.
5158 (expmed_mult_highpart): Likewise.
5159 (expand_smod_pow2): Likewise.
5160 (expand_sdiv_pow2): Likewise.
5161 (emit_store_flag_int): Likewise.
5162 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5164 (extract_low_bits): Likewise.
5165 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5167 * expr.c (pieces_addr::adjust): Likewise.
5168 (can_store_by_pieces): Likewise.
5169 (store_by_pieces): Likewise.
5170 (clear_by_pieces_1): Likewise.
5171 (expand_expr_addr_expr_1): Likewise.
5172 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5173 (expand_expr_real_1): Likewise.
5174 (try_casesi): Likewise.
5175 * final.c (shorten_branches): Likewise.
5176 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5177 type of "mode" to machine_mode.
5178 * internal-fn.c (expand_arith_overflow_result_store): Take a
5179 scalar_int_mode rather than a machine_mode.
5180 (expand_mul_overflow): Use scalar_int_mode for local variables.
5181 * loop-doloop.c (doloop_modify): Likewise.
5182 (doloop_optimize): Likewise.
5183 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5184 than a machine_mode.
5185 (expand_doubleword_shift_condmove): Likewise.
5186 (expand_doubleword_shift): Likewise.
5187 (expand_doubleword_clz): Likewise.
5188 (expand_doubleword_popcount): Likewise.
5189 (expand_doubleword_parity): Likewise.
5190 (expand_absneg_bit): Use scalar_int_mode for local variables.
5191 (prepare_float_lib_cmp): Likewise.
5192 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5193 rather than a machine_mode.
5194 (convert_memory_address_addr_space): Likewise.
5195 (get_mode_bounds): Likewise.
5196 (get_address_mode): Return a scalar_int_mode rather than a
5198 * rtlanal.c (get_address_mode): Likewise.
5199 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5200 than a machine_mode.
5201 * targhooks.c (default_mode_rep_extended): Likewise.
5202 (default_valid_pointer_mode): Likewise.
5203 (default_addr_space_valid_pointer_mode): Likewise.
5204 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5205 than a machine_mode.
5206 (default_libgcc_cmp_return_mode): Likewise.
5207 (default_libgcc_shift_count_mode): Likewise.
5208 (default_unwind_word_mode): Likewise.
5209 (default_addr_space_pointer_mode): Likewise.
5210 (default_addr_space_address_mode): Likewise.
5211 * targhooks.h (default_eh_return_filter_mode): Likewise.
5212 (default_libgcc_cmp_return_mode): Likewise.
5213 (default_libgcc_shift_count_mode): Likewise.
5214 (default_unwind_word_mode): Likewise.
5215 (default_addr_space_pointer_mode): Likewise.
5216 (default_addr_space_address_mode): Likewise.
5217 (default_mode_rep_extended): Take a scalar_int_mode rather than
5219 (default_valid_pointer_mode): Likewise.
5220 (default_addr_space_valid_pointer_mode): Likewise.
5221 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5223 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5224 rather than a machine_mode.
5225 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5226 for local variables.
5227 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5228 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5229 than a machine_mode.
5231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5232 Alan Hayward <alan.hayward@arm.com>
5233 David Sherwood <david.sherwood@arm.com>
5235 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5236 the mode argument to scalar_int_mode.
5237 (do_jump_by_parts_zero_rtx): Likewise.
5238 (do_jump_by_parts_equality_rtx): Likewise.
5239 (do_jump_by_parts_greater): Take a mode argument.
5240 (do_jump_by_parts_equality): Likewise.
5241 (do_jump_1): Update calls accordingly.
5243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5244 Alan Hayward <alan.hayward@arm.com>
5245 David Sherwood <david.sherwood@arm.com>
5247 * is-a.h (safe_dyn_cast): New function.
5248 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5249 (jump_table_for_label): Likewise.
5250 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5251 instead of an rtx_insn *.
5252 (shorten_branches): Use dyn_cast instead of LABEL_P and
5253 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5254 rtx_jump_table_data::get_data_mode.
5255 (final_scan_insn): Likewise.
5257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5258 Alan Hayward <alan.hayward@arm.com>
5259 David Sherwood <david.sherwood@arm.com>
5261 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5262 trying to combine a full-register integer set with a subreg
5265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5266 Alan Hayward <alan.hayward@arm.com>
5267 David Sherwood <david.sherwood@arm.com>
5269 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5270 that is always either address_mode or pointer_mode.
5272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5273 Alan Hayward <alan.hayward@arm.com>
5274 David Sherwood <david.sherwood@arm.com>
5276 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5277 when the two are known to be equal.
5279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5280 Alan Hayward <alan.hayward@arm.com>
5281 David Sherwood <david.sherwood@arm.com>
5283 * simplify-rtx.c (simplify_const_unary_operation): Use
5284 is_a <scalar_int_mode> instead of checking for a nonzero
5285 precision. Forcibly convert op_mode to a scalar_int_mode
5286 in that case. More clearly differentiate the operand and
5287 result modes and use the former when deciding what the value
5288 of a count-bits operation should be. Use is_int_mode instead
5289 of checking for a MODE_INT. Remove redundant check for whether
5290 this mode has a zero precision.
5292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5293 Alan Hayward <alan.hayward@arm.com>
5294 David Sherwood <david.sherwood@arm.com>
5296 * optabs.c (widen_leading): Change the type of the mode argument
5297 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5298 (widen_bswap): Likewise.
5299 (expand_parity): Likewise.
5300 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5301 (expand_ffs): Likewise.
5302 (epand_unop): Check for scalar integer modes before calling the
5305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5306 Alan Hayward <alan.hayward@arm.com>
5307 David Sherwood <david.sherwood@arm.com>
5309 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5311 (expand_expr_real_1): Update call accordingly.
5313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5314 Alan Hayward <alan.hayward@arm.com>
5315 David Sherwood <david.sherwood@arm.com>
5317 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5318 value_mode arguments. Use scalar_int_mode internally.
5319 (store_bit_field_1): Rename the new integer mode from imode
5320 to op0_mode and use it instead of GET_MODE (op0). Update calls
5321 to store_split_bit_field, store_bit_field_using_insv and
5322 store_fixed_bit_field.
5323 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5324 Use scalar_int_mode internally. Use a bit count rather than a mode
5325 when calculating the largest bit size for get_best_mode.
5326 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5327 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5328 Remove assertion that OP0 has a scalar integer mode.
5329 (store_split_bit_field): Add op0_mode and value_mode arguments.
5330 Update calls to extract_fixed_bit_field.
5331 (extract_bit_field_using_extv): Add an op0_mode argument.
5332 Use scalar_int_mode internally.
5333 (extract_bit_field_1): Rename the new integer mode from imode to
5334 op0_mode and use it instead of GET_MODE (op0). Update calls to
5335 extract_split_bit_field, extract_bit_field_using_extv and
5336 extract_fixed_bit_field.
5337 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5338 to extract_split_bit_field and extract_fixed_bit_field_1.
5339 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5340 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5342 (extract_split_bit_field): Add an op0_mode argument. Update call
5343 to extract_fixed_bit_field.
5345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5346 Alan Hayward <alan.hayward@arm.com>
5347 David Sherwood <david.sherwood@arm.com>
5349 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5350 * explow.c (hard_function_value): Likewise.
5351 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5352 convert_to_mode call outside the loop.
5353 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5354 for the mode iterator. Require the mode specified by max_pieces
5356 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5358 (copy_blkmode_to_reg): Likewise.
5359 (set_storage_via_setmem): Likewise.
5360 * optabs.c (prepare_cmp_insn): Likewise.
5361 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5362 * stor-layout.c (finish_bitfield_representative): Likewise.
5364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5365 Alan Hayward <alan.hayward@arm.com>
5366 David Sherwood <david.sherwood@arm.com>
5368 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5369 * expr.c (convert_move): Use them.
5370 (convert_modes): Likewise.
5371 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5378 parameter for the mode of "x". Remove the "known_x", "known_mode"
5379 and "known_ret" arguments. Change the type of the mode argument
5381 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5382 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5383 (reg_num_sign_bit_copies_for_combine): Likewise.
5384 * rtlanal.c (nonzero_bits1): Likewise.
5385 (num_sign_bit_copies1): Likewise.
5386 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5387 (reg_num_sign_bit_copies_general): Likewise.
5388 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5389 (reg_nonzero_bits_general): Likewise.
5391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5392 Alan Hayward <alan.hayward@arm.com>
5393 David Sherwood <david.sherwood@arm.com>
5395 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5396 than in subroutines. Return 1 for non-integer modes.
5397 (cached_num_sign_bit_copies): Change the type of the mode parameter
5399 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5400 classes. Handle CONST_INT_P first and then check whether X also
5401 has a scalar integer mode. Check the same thing for inner registers
5402 of a SUBREG and for values that are being extended or truncated.
5404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5405 Alan Hayward <alan.hayward@arm.com>
5406 David Sherwood <david.sherwood@arm.com>
5408 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5409 in subroutines. Return the mode mask for non-integer modes.
5410 (cached_nonzero_bits): Change the type of the mode parameter
5412 (nonzero_bits1): Likewise. Remove early exit for other mode
5413 classes. Handle CONST_INT_P first and then check whether X
5414 also has a scalar integer mode.
5416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5417 Alan Hayward <alan.hayward@arm.com>
5418 David Sherwood <david.sherwood@arm.com>
5420 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5421 Return a scalar_int_mode and assert that the size is greater than
5423 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5424 (op_by_pieces_d::op_by_pieces_d): Likewise.
5425 (op_by_pieces_d::run): Likewise.
5426 (can_store_by_pieces): Likewise.
5428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5429 Alan Hayward <alan.hayward@arm.com>
5430 David Sherwood <david.sherwood@arm.com>
5432 * combine.c (extract_left_shift): Add a mode argument and update
5434 (make_compound_operation_int): Change the type of the mode parameter
5435 to scalar_int_mode and update the call to extract_left_shift.
5437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5438 Alan Hayward <alan.hayward@arm.com>
5439 David Sherwood <david.sherwood@arm.com>
5441 * combine.c (simplify_and_const_int): Change the type of the mode
5442 parameter to scalar_int_mode.
5443 (simplify_and_const_int_1): Likewise. Update recursive call.
5445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * combine.c (simplify_compare_const): Check that the mode is a
5450 scalar_int_mode (rather than VOIDmode) before testing its
5452 (simplify_comparison): Move COMPARISON_P handling out of the
5453 loop and restrict the latter part of the loop to scalar_int_modes.
5454 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5455 and when considering SUBREG_REGs. Use is_int_mode instead of
5456 checking GET_MODE_CLASS against MODE_INT.
5458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5459 Alan Hayward <alan.hayward@arm.com>
5460 David Sherwood <david.sherwood@arm.com>
5462 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5463 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5464 shift_unit_mode and for modes involved in scalar shifts.
5466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5467 Alan Hayward <alan.hayward@arm.com>
5468 David Sherwood <david.sherwood@arm.com>
5470 * combine.c (force_int_to_mode): New function, split out from...
5471 (force_to_mode): ...here. Keep xmode up-to-date and use it
5472 instead of GET_MODE (x).
5474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5475 Alan Hayward <alan.hayward@arm.com>
5476 David Sherwood <david.sherwood@arm.com>
5478 * optabs-query.h (extraction_insn::struct_mode): Change type to
5479 opt_scalar_int_mode and update comment.
5480 (extraction_insn::field_mode): Change type to scalar_int_mode.
5481 (extraction_insn::pos_mode): Likewise.
5482 * combine.c (make_extraction): Update accordingly.
5483 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5484 (get_optab_extraction_insn): Likewise.
5485 * recog.c (simplify_while_replacing): Likewise.
5486 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5487 parameter to opt_scalar_int_mode.
5489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5490 Alan Hayward <alan.hayward@arm.com>
5491 David Sherwood <david.sherwood@arm.com>
5493 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5494 to a scalar_int_mode instead of a machine_mode.
5495 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5496 (get_best_mode): Return a boolean and use a pointer argument to store
5497 the selected mode. Replace the limit mode parameter with a bit limit.
5498 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5499 for the values returned by bit_field_mode_iterator::next_mode.
5500 (store_bit_field): Update call to get_best_mode.
5501 (store_fixed_bit_field): Likewise.
5502 (extract_fixed_bit_field): Likewise.
5503 * expr.c (optimize_bitfield_assignment_op): Likewise.
5504 * fold-const.c (optimize_bit_field_compare): Likewise.
5505 (fold_truth_andor_1): Likewise.
5506 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5507 Update for new type of m_mode.
5508 (get_best_mode): As above.
5510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5511 Alan Hayward <alan.hayward@arm.com>
5512 David Sherwood <david.sherwood@arm.com>
5514 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5515 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5516 (store_bit_field): Check is_a <scalar_int_mode> before calling
5517 strict_volatile_bitfield_p.
5518 (extract_bit_field): Likewise.
5520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5521 Alan Hayward <alan.hayward@arm.com>
5522 David Sherwood <david.sherwood@arm.com>
5524 * target.def (cstore_mode): Return a scalar_int_mode.
5525 * doc/tm.texi: Regenerate.
5526 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5527 * targhooks.h (default_cstore_mode): Likewise.
5528 * targhooks.c (default_cstore_mode): Likewise, using a forced
5530 * expmed.c (emit_cstore): Expect the target of the cstore to be
5533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5534 Alan Hayward <alan.hayward@arm.com>
5535 David Sherwood <david.sherwood@arm.com>
5537 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5539 (niter_desc): Likewise mode.
5540 (iv_analyze): Add a mode parameter.
5542 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5543 and change its type to scalar_int_mode.
5544 * loop-iv.c: Update commentary at head of file.
5545 (iv_constant): Pass the mode paraeter before the rtx it describes
5546 and change its type to scalar_int_mode. Remove VOIDmode handling.
5547 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5548 (iv_extend): Likewise.
5549 (shorten_into_mode): Likewise.
5550 (iv_add): Use scalar_int_mode.
5551 (iv_mult): Likewise.
5552 (iv_shift): Likewise.
5553 (canonicalize_iv_subregs): Likewise.
5554 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5555 it describes and change its mode to scalar_int_mode. Also change
5556 the type of the returned inner_mode to scalar_int_mode.
5557 (get_biv_step): Likewise, turning outer_mode from a pointer
5558 into a direct parameter. Update call to get_biv_step_1.
5559 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5560 iv_constant and get_biv_step.
5561 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5562 and change its type to scalar_int_mode. Don't initialise iv->mode
5563 to VOIDmode and remove later checks for its still being VOIDmode.
5564 Update calls to iv_analyze_op and iv_analyze_expr. Check
5565 is_a <scalar_int_mode> when changing the mode under consideration.
5566 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5567 Update call to iv_analyze_expr.
5568 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5569 inner register is not also a scalar_int_mode. Update call to
5571 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5572 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5573 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5574 separate mode class checks. Update calls to iv_analyze. Remove
5575 fix-up of VOIDmodes after iv_analyze_biv.
5576 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5577 don't have a scalar_int_mode. Update call to biv_p.
5579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5580 Alan Hayward <alan.hayward@arm.com>
5581 David Sherwood <david.sherwood@arm.com>
5583 * cfgexpand.c (convert_debug_memory_address): Use
5584 as_a <scalar_int_mode>.
5585 * combine.c (expand_compound_operation): Likewise.
5586 (make_extraction): Likewise.
5587 (change_zero_ext): Likewise.
5588 (simplify_comparison): Likewise.
5589 * cse.c (cse_insn): Likewise.
5590 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5591 (mem_loc_descriptor): Likewise.
5592 (loc_descriptor): Likewise.
5593 * expmed.c (init_expmed_one_mode): Likewise.
5594 (synth_mult): Likewise.
5595 (emit_store_flag_1): Likewise.
5596 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5597 of a comparison with size.
5598 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5599 (reduce_to_bit_field_precision): Likewise.
5600 * function.c (expand_function_end): Likewise.
5601 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5602 * loop-doloop.c (doloop_modify): Likewise.
5603 * optabs.c (expand_binop): Likewise.
5604 (expand_unop): Likewise.
5605 (expand_copysign_absneg): Likewise.
5606 (prepare_cmp_insn): Likewise.
5607 (maybe_legitimize_operand): Likewise.
5608 * recog.c (const_scalar_int_operand): Likewise.
5609 * rtlanal.c (get_address_mode): Likewise.
5610 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5611 (simplify_cond_clz_ctz): Likewise.
5612 * tree-nested.c (get_nl_goto_field): Likewise.
5613 * tree.c (build_vector_type_for_mode): Likewise.
5614 * var-tracking.c (use_narrower_mode): 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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5621 * builtins.c (expand_builtin_signbit): Use it.
5622 * cfgexpand.c (expand_debug_expr): Likewise.
5623 * dojump.c (do_jump): Likewise.
5624 (do_compare_and_jump): Likewise.
5625 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5626 * expmed.c (make_tree): Likewise.
5627 * expr.c (expand_expr_real_2): Likewise.
5628 (expand_expr_real_1): Likewise.
5629 (try_casesi): Likewise.
5630 * fold-const-call.c (fold_const_call_ss): Likewise.
5631 * fold-const.c (unextend): Likewise.
5632 (extract_muldiv_1): Likewise.
5633 (fold_single_bit_test): Likewise.
5634 (native_encode_int): Likewise.
5635 (native_encode_string): Likewise.
5636 (native_interpret_int): Likewise.
5637 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5638 * internal-fn.c (expand_addsub_overflow): Likewise.
5639 (expand_neg_overflow): Likewise.
5640 (expand_mul_overflow): Likewise.
5641 (expand_arith_overflow): Likewise.
5642 * match.pd: Likewise.
5643 * stor-layout.c (layout_type): Likewise.
5644 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5645 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5646 * tree-ssanames.c (get_range_info): Likewise.
5647 * tree-switch-conversion.c (array_value_type) Likewise.
5648 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5649 (vect_recog_divmod_pattern): Likewise.
5650 (vect_recog_mixed_size_cond_pattern): Likewise.
5651 * tree-vrp.c (extract_range_basic): Likewise.
5652 (simplify_float_conversion_using_ranges): Likewise.
5653 * tree.c (int_fits_type_p): Likewise.
5654 * ubsan.c (instrument_bool_enum_load): Likewise.
5655 * varasm.c (mergeable_string_section): Likewise.
5656 (narrowing_initializer_constant_valid_p): Likewise.
5657 (output_constant): Likewise.
5659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5660 Alan Hayward <alan.hayward@arm.com>
5661 David Sherwood <david.sherwood@arm.com>
5663 * machmode.h (NARROWEST_INT_MODE): New macro.
5664 * expr.c (alignment_for_piecewise_move): Use it instead of
5665 GET_CLASS_NARROWEST_MODE (MODE_INT).
5666 (push_block): Likewise.
5667 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5669 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5672 Alan Hayward <alan.hayward@arm.com>
5673 David Sherwood <david.sherwood@arm.com>
5675 * postreload.c (move2add_valid_value_p): Change the type of the
5676 mode parameter to scalar_int_mode.
5677 (move2add_use_add2_insn): Add a mode parameter and use it instead
5679 (move2add_use_add3_insn): Likewise.
5680 (reload_cse_move2add): Update accordingly.
5682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5683 Alan Hayward <alan.hayward@arm.com>
5684 David Sherwood <david.sherwood@arm.com>
5686 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5688 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5689 * optabs.c (expand_unop): Likewise.
5691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5692 Alan Hayward <alan.hayward@arm.com>
5693 David Sherwood <david.sherwood@arm.com>
5695 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5696 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5697 (popcount_loc_descriptor): Likewise.
5698 (bswap_loc_descriptor): Likewise.
5699 (rotate_loc_descriptor): Likewise.
5700 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5701 calling the functions above.
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 (sign_extend_short_imm): Add is_a <scalar_int_mode>
5709 (try_combine): Likewise.
5710 (simplify_if_then_else): Likewise.
5711 * cse.c (cse_insn): Likewise.
5712 * dwarf2out.c (mem_loc_descriptor): Likewise.
5713 * emit-rtl.c (gen_lowpart_common): Likewise.
5714 * simplify-rtx.c (simplify_truncation): Likewise.
5715 (simplify_binary_operation_1): Likewise.
5716 (simplify_const_relational_operation): Likewise.
5717 (simplify_ternary_operation): Likewise.
5718 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5721 Alan Hayward <alan.hayward@arm.com>
5722 David Sherwood <david.sherwood@arm.com>
5724 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5725 * reload.c (push_reload): Likewise.
5726 (find_reloads): Likewise.
5728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5729 Alan Hayward <alan.hayward@arm.com>
5730 David Sherwood <david.sherwood@arm.com>
5732 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5733 (make_compound_operation_int): Likewise.
5734 (change_zero_ext): Likewise.
5735 * expr.c (convert_move): Likewise.
5736 (convert_modes): Likewise.
5737 * fwprop.c (forward_propagate_subreg): Likewise.
5738 * loop-iv.c (get_biv_step_1): Likewise.
5739 * optabs.c (widen_operand): Likewise.
5740 * postreload.c (move2add_valid_value_p): Likewise.
5741 * recog.c (simplify_while_replacing): Likewise.
5742 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5743 (simplify_binary_operation_1): Likewise. Remove redundant
5744 mode equality check.
5746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5747 Alan Hayward <alan.hayward@arm.com>
5748 David Sherwood <david.sherwood@arm.com>
5750 * combine.c (combine_simplify_rtx): Add checks for
5751 is_a <scalar_int_mode>.
5752 (simplify_if_then_else): Likewise.
5753 (make_field_assignment): Likewise.
5754 (simplify_comparison): Likewise.
5755 * ifcvt.c (noce_try_bitop): Likewise.
5756 * loop-invariant.c (canonicalize_address_mult): Likewise.
5757 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5760 Alan Hayward <alan.hayward@arm.com>
5761 David Sherwood <david.sherwood@arm.com>
5763 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5764 is_a <scalar_int_mode> instead of != BLKmode.
5766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5767 Alan Hayward <alan.hayward@arm.com>
5768 David Sherwood <david.sherwood@arm.com>
5770 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5771 instead of != VOIDmode.
5772 * combine.c (if_then_else_cond): Likewise.
5773 (change_zero_ext): Likewise.
5774 * dwarf2out.c (mem_loc_descriptor): Likewise.
5775 (loc_descriptor): Likewise.
5776 * rtlanal.c (canonicalize_condition): Likewise.
5777 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5780 Alan Hayward <alan.hayward@arm.com>
5781 David Sherwood <david.sherwood@arm.com>
5783 * simplify-rtx.c (simplify_binary_operation_1): Use
5784 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5787 Alan Hayward <alan.hayward@arm.com>
5788 David Sherwood <david.sherwood@arm.com>
5790 * wide-int.h (int_traits<unsigned char>) New class.
5791 (int_traits<unsigned short>) Likewise.
5792 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5793 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5795 * combine.c (set_nonzero_bits_and_sign_copies): Use
5796 is_a <scalar_int_mode>.
5797 (find_split_point): Likewise.
5798 (combine_simplify_rtx): Likewise.
5799 (simplify_logical): Likewise.
5800 (expand_compound_operation): Likewise.
5801 (expand_field_assignment): Likewise.
5802 (make_compound_operation): Likewise.
5803 (extended_count): Likewise.
5804 (change_zero_ext): Likewise.
5805 (simplify_comparison): Likewise.
5806 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5807 (ucompare_loc_descriptor): Likewise.
5808 (minmax_loc_descriptor): Likewise.
5809 (mem_loc_descriptor): Likewise.
5810 (loc_descriptor): Likewise.
5811 * expmed.c (init_expmed_one_mode): Likewise.
5812 * lra-constraints.c (lra_constraint_offset): Likewise.
5813 * optabs.c (prepare_libcall_arg): Likewise.
5814 * postreload.c (move2add_note_store): Likewise.
5815 * reload.c (operands_match_p): Likewise.
5816 * rtl.h (load_extend_op): Likewise.
5817 * rtlhooks.c (gen_lowpart_general): Likewise.
5818 * simplify-rtx.c (simplify_truncation): Likewise.
5819 (simplify_unary_operation_1): Likewise.
5820 (simplify_binary_operation_1): Likewise.
5821 (simplify_const_binary_operation): Likewise.
5822 (simplify_const_relational_operation): Likewise.
5823 (simplify_subreg): Likewise.
5824 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5825 * var-tracking.c (adjust_mems): Likewise.
5826 (prepare_call_arguments): Likewise.
5828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5829 Alan Hayward <alan.hayward@arm.com>
5830 David Sherwood <david.sherwood@arm.com>
5832 * machmode.h (is_int_mode): New fuction.
5833 * combine.c (find_split_point): Use it.
5834 (combine_simplify_rtx): Likewise.
5835 (simplify_if_then_else): Likewise.
5836 (simplify_set): Likewise.
5837 (simplify_shift_const_1): Likewise.
5838 (simplify_comparison): Likewise.
5839 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5840 * cse.c (notreg_cost): Likewise.
5841 (cse_insn): Likewise.
5842 * cselib.c (cselib_lookup_1): Likewise.
5843 * dojump.c (do_jump_1): Likewise.
5844 (do_compare_rtx_and_jump): Likewise.
5845 * dse.c (get_call_args): Likewise.
5846 * dwarf2out.c (rtl_for_decl_init): Likewise.
5847 (native_encode_initializer): Likewise.
5848 * expmed.c (emit_store_flag_1): Likewise.
5849 (emit_store_flag): Likewise.
5850 * expr.c (convert_modes): Likewise.
5851 (store_field): Likewise.
5852 (expand_expr_real_1): Likewise.
5853 * fold-const.c (fold_read_from_constant_string): Likewise.
5854 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5855 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5856 * optabs.c (expand_binop): Likewise.
5857 (expand_unop): Likewise.
5858 (expand_abs_nojump): Likewise.
5859 (expand_one_cmpl_abs_nojump): Likewise.
5860 * simplify-rtx.c (mode_signbit_p): Likewise.
5861 (val_signbit_p): Likewise.
5862 (val_signbit_known_set_p): Likewise.
5863 (val_signbit_known_clear_p): Likewise.
5864 (simplify_relational_operation_1): Likewise.
5865 * tree.c (vector_type_mode): Likewise.
5867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5868 Alan Hayward <alan.hayward@arm.com>
5869 David Sherwood <david.sherwood@arm.com>
5871 * machmode.h (smallest_mode_for_size): Fix formatting.
5872 (smallest_int_mode_for_size): New function.
5873 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5874 instead of smallest_mode_for_size.
5875 * combine.c (make_extraction): Likewise.
5876 * config/arc/arc.c (arc_expand_movmem): Likewise.
5877 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5878 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5879 * config/s390/s390.c (s390_expand_insv): Likewise.
5880 * config/sparc/sparc.c (assign_int_registers): Likewise.
5881 * config/spu/spu.c (spu_function_value): Likewise.
5882 (spu_function_arg): Likewise.
5883 * coverage.c (get_gcov_type): Likewise.
5884 (get_gcov_unsigned_t): Likewise.
5885 * dse.c (find_shift_sequence): Likewise.
5886 * expmed.c (store_bit_field_1): Likewise.
5887 * expr.c (convert_move): Likewise.
5888 (store_field): Likewise.
5889 * internal-fn.c (expand_arith_overflow): Likewise.
5890 * optabs-query.c (get_best_extraction_insn): Likewise.
5891 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5892 * stor-layout.c (layout_type): Likewise.
5893 (initialize_sizetypes): Likewise.
5894 * targhooks.c (default_get_mask_mode): Likewise.
5895 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): 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 (opt_mode::else_blk): New function.
5902 (int_mode_for_mode): Declare.
5903 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5904 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5906 * cfgexpand.c (expand_debug_expr): Likewise.
5907 * combine.c (gen_lowpart_or_truncate): Likewise.
5908 (gen_lowpart_for_combine): Likewise.
5909 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5910 * config/avr/avr.c (avr_to_int_mode): Likewise.
5911 (avr_out_plus_1): Likewise.
5912 (avr_out_plus): Likewise.
5913 (avr_out_round): Likewise.
5914 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5915 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5916 (s390_expand_vcond): Likewise.
5917 * config/spu/spu.c (spu_split_immediate): Likewise.
5918 (spu_expand_mov): Likewise.
5919 * dse.c (get_stored_val): Likewise.
5920 * expmed.c (store_bit_field_1): Likewise.
5921 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5923 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5924 (extract_low_bits): Likewise.
5925 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5926 handling rather than repeating the check.
5927 (emit_group_store): Likewise.
5928 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5929 * optabs.c (expand_absneg_bit): Likewise.
5930 (expand_copysign_absneg): Likewise.
5931 (expand_copysign_bit): Likewise.
5932 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5933 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5934 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5935 * var-tracking.c (prepare_call_arguments): Likewise.
5936 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5937 int_mode_for_mode instead of mode_for_size.
5938 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5941 Alan Hayward <alan.hayward@arm.com>
5942 David Sherwood <david.sherwood@arm.com>
5944 * machmode.h (int_mode_for_size): New function.
5945 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5946 instead of mode_for_size.
5947 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5949 * combine.c (expand_field_assignment): Use int_mode_for_size
5950 instead of mode_for_size.
5951 (make_extraction): Likewise.
5952 (simplify_shift_const_1): Likewise.
5953 (simplify_comparison): Likewise.
5954 * dojump.c (do_jump): Likewise.
5955 * dwarf2out.c (mem_loc_descriptor): Likewise.
5956 * emit-rtl.c (init_derived_machine_modes): Likewise.
5957 * expmed.c (flip_storage_order): Likewise.
5958 (convert_extracted_bit_field): Likewise.
5959 * expr.c (copy_blkmode_from_reg): Likewise.
5960 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5961 * internal-fn.c (expand_mul_overflow): Likewise.
5962 * lower-subreg.c (simple_move): Likewise.
5963 * optabs-libfuncs.c (init_optabs): Likewise.
5964 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5965 * tree.c (vector_type_mode): Likewise.
5966 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5967 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5968 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5969 * tree-vect-stmts.c (vectorizable_load): Likewise.
5970 (vectorizable_store): Likewise.
5972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5973 Alan Hayward <alan.hayward@arm.com>
5974 David Sherwood <david.sherwood@arm.com>
5976 * coretypes.h (pod_mode): New type.
5977 (scalar_int_mode_pod): New typedef.
5978 * machmode.h (pod_mode): New class.
5979 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5980 * genmodes.c (emit_mode_int_n): Update accordingly.
5981 * lower-subreg.h (target_lower_subreg): Change type to
5982 scalar_int_mode_pod.
5983 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5984 scalar_int_mode_pod.
5986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5987 Alan Hayward <alan.hayward@arm.com>
5988 David Sherwood <david.sherwood@arm.com>
5990 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5991 machine_mode to scalar_int_mode.
5992 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5993 (rs6000_option_override_internal): Remove cast to int.
5994 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5995 machine_mode to scalar_int_mode.
5996 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5997 (rs6000_option_override_internal): Remove cast to int.
5998 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5999 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6001 * config/s390/s390.c (s390_expand_builtin): Likewise.
6002 * coretypes.h (scalar_int_mode): New type.
6003 (opt_scalar_int_mode): New typedef.
6004 * machmode.h (scalar_int_mode): New class.
6005 (scalar_int_mode::includes_p): New function.
6006 (byte_mode): Change type to scalar_int_mode.
6007 (word_mode): Likewise.
6008 (ptr_mode): Likewise.
6009 * emit-rtl.c (byte_mode): Likewise.
6010 (word_mode): Likewise.
6011 (ptr_mode): Likewise.
6012 (init_derived_machine_modes): Update accordingly.
6013 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6014 and MODE_PARTIAL_INT.
6015 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6016 opt_scalar_int_mode.
6018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6019 Alan Hayward <alan.hayward@arm.com>
6020 David Sherwood <david.sherwood@arm.com>
6022 * target.def (libgcc_floating_mode_supported_p): Take a
6024 * doc/tm.texi: Regenerate.
6025 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6027 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6028 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6032 Alan Hayward <alan.hayward@arm.com>
6033 David Sherwood <david.sherwood@arm.com>
6035 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6036 * doc/tm.texi: Regenerate.
6037 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6038 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6039 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6040 * targhooks.h (default_floatn_mode): Likewise.
6041 * targhooks.c (default_floatn_mode): Likewise.
6042 * tree.c (build_common_tree_nodes): Update accordingly.
6044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6045 Alan Hayward <alan.hayward@arm.com>
6046 David Sherwood <david.sherwood@arm.com>
6048 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6049 (mode_iterator::iterate_p): Likewise.
6050 (mode_iterator::get_wider): Likewise.
6051 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6054 Alan Hayward <alan.hayward@arm.com>
6055 David Sherwood <david.sherwood@arm.com>
6057 * coretypes.h (opt_scalar_float_mode): New typedef.
6058 * machmode.h (float_mode_for_size): New function.
6059 * emit-rtl.c (double_mode): Delete.
6060 (init_emit_once): Use float_mode_for_size.
6061 * stor-layout.c (layout_type): Likewise.
6062 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6065 Alan Hayward <alan.hayward@arm.com>
6066 David Sherwood <david.sherwood@arm.com>
6068 * output.h (assemble_real): Take a scalar_float_mode.
6069 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6070 * config/arm/arm.md (consttable_4): Likewise.
6071 (consttable_8): Likewise.
6072 (consttable_16): Likewise.
6073 * config/mips/mips.md (consttable_float): Likewise.
6074 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6075 * varasm.c (assemble_real): Take a scalar_float_mode.
6076 (output_constant_pool_2): Update accordingly.
6077 (output_constant): Likewise.
6079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6080 Alan Hayward <alan.hayward@arm.com>
6081 David Sherwood <david.sherwood@arm.com>
6083 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6084 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6085 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6086 (native_encode_real): Likewise.
6087 (native_interpret_real): Likewise.
6088 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6089 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6092 Alan Hayward <alan.hayward@arm.com>
6093 David Sherwood <david.sherwood@arm.com>
6095 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6096 <scalar_float_mode>. Simplify.
6097 (gen_extend_conv_libfunc): Likewise.
6099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6100 Alan Hayward <alan.hayward@arm.com>
6101 David Sherwood <david.sherwood@arm.com>
6103 * coretypes.h (scalar_float_mode): New type.
6104 * machmode.h (mode_traits::from_int): Use machine_mode if
6105 USE_ENUM_MODES is defined.
6106 (is_a): New function.
6108 (dyn_cast): Likewise.
6109 (scalar_float_mode): New class.
6110 (scalar_float_mode::includes_p): New function.
6111 (is_float_mode): Likewise.
6112 * gdbhooks.py (MachineModePrinter): New class.
6113 (build_pretty_printer): Use it for scalar_float_mode.
6114 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6115 (format_helper::format_helper): Turn into a template.
6116 * genmodes.c (get_mode_class): New function.
6117 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6118 or machine_mode if none.
6119 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6120 as_a <scalar_float_mode>.
6121 * dwarf2out.c (mem_loc_descriptor): Likewise.
6122 (insert_float): Likewise.
6123 (add_const_value_attribute): Likewise.
6124 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6125 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6126 (expand_unop): Update accordingly.
6127 (expand_abs_nojump): Likewise.
6128 (expand_copysign_absneg): Take a scalar_float_mode.
6129 (expand_copysign_bit): Likewise.
6130 (expand_copysign): Update accordingly.
6132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6133 Alan Hayward <alan.hayward@arm.com>
6134 David Sherwood <david.sherwood@arm.com>
6136 * coretypes.h (opt_mode): New class.
6137 * machmode.h (opt_mode): Likewise.
6138 (opt_mode::else_void): New function.
6139 (opt_mode::require): Likewise.
6140 (opt_mode::exists): Likewise.
6141 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6142 (GET_MODE_2XWIDER_MODE): Likewise.
6143 (mode_iterator::get_wider): Update accordingly.
6144 (mode_iterator::get_2xwider): Likewise.
6145 (mode_iterator::get_known_wider): Likewise, turning into a template.
6146 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6147 forcing a wider mode to exist.
6148 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6149 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6150 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6151 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6152 * lower-subreg.c (init_lower_subreg): Likewise.
6153 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6154 on the final iteration.
6155 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6156 a wider mode exists before asking for a move pattern.
6157 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6158 forcing a wider mode to exist.
6159 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6160 returning false if no such mode exists.
6161 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6162 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6163 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6164 Avoid checking for a MODE_INT if we already know the mode is not a
6166 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6167 forcing a wider mode to exist.
6168 (expmed_mult_highpart_optab): Likewise.
6169 (expmed_mult_highpart): Likewise.
6170 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6172 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6173 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6174 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6175 * internal-fn.c (expand_mul_overflow): Update use of
6176 GET_MODE_2XWIDER_MODE.
6177 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6178 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6179 GET_MODE_WIDER_MODE.
6180 (convert_plusminus_to_widen): Likewise.
6181 * tree-switch-conversion.c (array_value_type): Likewise.
6182 * var-tracking.c (emit_note_insn_var_location): Likewise.
6183 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6184 Return false inside rather than outside the loop if no wider mode
6186 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6187 and GET_MODE_2XWIDER_MODE
6188 (can_compare_p): Use else_void.
6189 * gdbhooks.py (OptMachineModePrinter): New class.
6190 (build_pretty_printer): Use it for opt_mode.
6192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6193 Alan Hayward <alan.hayward@arm.com>
6194 David Sherwood <david.sherwood@arm.com>
6196 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6197 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6200 Alan Hayward <alan.hayward@arm.com>
6201 David Sherwood <david.sherwood@arm.com>
6203 * machmode.h (mode_traits): New structure.
6204 (get_narrowest_mode): New function.
6205 (mode_iterator::start): Likewise.
6206 (mode_iterator::iterate_p): Likewise.
6207 (mode_iterator::get_wider): Likewise.
6208 (mode_iterator::get_known_wider): Likewise.
6209 (mode_iterator::get_2xwider): Likewise.
6210 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6211 (FOR_EACH_MODE): Likewise.
6212 (FOR_EACH_MODE_FROM): Likewise.
6213 (FOR_EACH_MODE_UNTIL): Likewise.
6214 (FOR_EACH_WIDER_MODE): Likewise.
6215 (FOR_EACH_2XWIDER_MODE): Likewise.
6216 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6217 * combine.c (simplify_comparison): Likewise
6218 * config/i386/i386.c (type_natural_mode): Likewise.
6219 * cse.c (cse_insn): Likewise.
6220 * dse.c (find_shift_sequence): Likewise.
6221 * emit-rtl.c (init_derived_machine_modes): Likewise.
6222 (init_emit_once): Likewise.
6223 * explow.c (hard_function_value): Likewise.
6224 * expmed.c (extract_fixed_bit_field_1): Likewise.
6225 (extract_bit_field_1): Likewise.
6226 (expand_divmod): Likewise.
6227 (emit_store_flag_1): Likewise.
6228 * expr.c (init_expr_target): Likewise.
6229 (convert_move): Likewise.
6230 (alignment_for_piecewise_move): Likewise.
6231 (widest_int_mode_for_size): Likewise.
6232 (emit_block_move_via_movmem): Likewise.
6233 (copy_blkmode_to_reg): Likewise.
6234 (set_storage_via_setmem): Likewise.
6235 (compress_float_constant): Likewise.
6236 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6237 * optabs-query.c (get_best_extraction_insn): Likewise.
6238 * optabs.c (expand_binop): Likewise.
6239 (expand_twoval_unop): Likewise.
6240 (expand_twoval_binop): Likewise.
6241 (widen_leading): Likewise.
6242 (widen_bswap): Likewise.
6243 (expand_parity): Likewise.
6244 (expand_unop): Likewise.
6245 (prepare_cmp_insn): Likewise.
6246 (prepare_float_lib_cmp): Likewise.
6247 (expand_float): Likewise.
6248 (expand_fix): Likewise.
6249 (expand_sfix_optab): Likewise.
6250 * postreload.c (move2add_use_add2_insn): Likewise.
6251 * reg-stack.c (reg_to_stack): Likewise.
6252 * reginfo.c (choose_hard_reg_mode): Likewise.
6253 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6254 * stor-layout.c (mode_for_size): Likewise.
6255 (smallest_mode_for_size): Likewise.
6256 (mode_for_vector): Likewise.
6257 (finish_bitfield_representative): Likewise.
6258 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6259 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6260 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6261 * var-tracking.c (prepare_call_arguments): Likewise.
6263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6264 Alan Hayward <alan.hayward@arm.com>
6265 David Sherwood <david.sherwood@arm.com>
6267 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6268 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6269 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6270 * machmode.h (mode_size): Move earlier in file.
6271 (mode_precision): Likewise.
6272 (mode_inner): Likewise.
6273 (mode_nunits): Likewise.
6274 (mode_unit_size): Likewise.
6275 (unit_unit_precision): Likewise.
6276 (mode_wider): Likewise.
6277 (mode_2xwider): Likewise.
6278 (machine_mode): New class.
6279 (mode_to_bytes): New function.
6280 (mode_to_bits): Likewise.
6281 (mode_to_precision): Likewise.
6282 (mode_to_inner): Likewise.
6283 (mode_to_unit_size): Likewise.
6284 (mode_to_unit_precision): Likewise.
6285 (mode_to_nunits): Likewise.
6286 (GET_MODE_SIZE): Use mode_to_bytes.
6287 (GET_MODE_BITSIZE): Use mode_to_bits.
6288 (GET_MODE_PRECISION): Use mode_to_precision.
6289 (GET_MODE_INNER): Use mode_to_inner.
6290 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6291 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6292 (GET_MODE_NUNITS): Use mode_to_nunits.
6293 * system.h (ALWAYS_INLINE): New macro.
6294 * config/powerpcspe/powerpcspe-c.c
6295 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6296 int for arg1_mode and arg2_mode.
6298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6299 Alan Hayward <alan.hayward@arm.com>
6300 David Sherwood <david.sherwood@arm.com>
6302 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6303 Prefix mode names with E_ in case statements.
6304 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6305 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6306 (aarch64_split_simd_move): Likewise.
6307 (aarch64_gen_storewb_pair): Likewise.
6308 (aarch64_gen_loadwb_pair): Likewise.
6309 (aarch64_gen_store_pair): Likewise.
6310 (aarch64_gen_load_pair): Likewise.
6311 (aarch64_get_condition_code_1): Likewise.
6312 (aarch64_constant_pool_reload_icode): Likewise.
6313 (get_rsqrte_type): Likewise.
6314 (get_rsqrts_type): Likewise.
6315 (get_recpe_type): Likewise.
6316 (get_recps_type): Likewise.
6317 (aarch64_gimplify_va_arg_expr): Likewise.
6318 (aarch64_simd_container_mode): Likewise.
6319 (aarch64_emit_load_exclusive): Likewise.
6320 (aarch64_emit_store_exclusive): Likewise.
6321 (aarch64_expand_compare_and_swap): Likewise.
6322 (aarch64_gen_atomic_cas): Likewise.
6323 (aarch64_emit_bic): Likewise.
6324 (aarch64_emit_atomic_swap): Likewise.
6325 (aarch64_emit_atomic_load_op): Likewise.
6326 (aarch64_evpc_trn): Likewise.
6327 (aarch64_evpc_uzp): Likewise.
6328 (aarch64_evpc_zip): Likewise.
6329 (aarch64_evpc_ext): Likewise.
6330 (aarch64_evpc_rev): Likewise.
6331 (aarch64_evpc_dup): Likewise.
6332 (aarch64_gen_ccmp_first): Likewise.
6333 (aarch64_gen_ccmp_next): Likewise.
6334 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6335 (alpha_emit_xfloating_libcall): Likewise.
6336 (emit_insxl): Likewise.
6337 (alpha_arg_type): Likewise.
6338 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6339 (arc_preferred_simd_mode): Likewise.
6340 (arc_secondary_reload): Likewise.
6341 (get_arc_condition_code): Likewise.
6342 (arc_print_operand): Likewise.
6343 (arc_legitimate_constant_p): Likewise.
6344 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6345 * config/arc/arc.md (casesi_load): Likewise.
6346 (casesi_compact_jump): Likewise.
6347 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6348 (cc_use_register): Likewise.
6349 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6350 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6351 (arm_init_iwmmxt_builtins): Likewise.
6352 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6353 (neon_expand_vector_init): Likewise.
6354 (arm_attr_length_move_neon): Likewise.
6355 (maybe_get_arm_condition_code): Likewise.
6356 (arm_emit_vector_const): Likewise.
6357 (arm_preferred_simd_mode): Likewise.
6358 (arm_output_iwmmxt_tinsr): Likewise.
6359 (thumb1_output_casesi): Likewise.
6360 (thumb2_output_casesi): Likewise.
6361 (arm_emit_load_exclusive): Likewise.
6362 (arm_emit_store_exclusive): Likewise.
6363 (arm_expand_compare_and_swap): Likewise.
6364 (arm_evpc_neon_vuzp): Likewise.
6365 (arm_evpc_neon_vzip): Likewise.
6366 (arm_evpc_neon_vrev): Likewise.
6367 (arm_evpc_neon_vtrn): Likewise.
6368 (arm_evpc_neon_vext): Likewise.
6369 (arm_validize_comparison): Likewise.
6370 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6371 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6372 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6373 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6374 (c6x_preferred_simd_mode): Likewise.
6375 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6376 (epiphany_rtx_costs): Likewise.
6377 * config/epiphany/predicates.md (proper_comparison_operator):
6379 * config/frv/frv.c (condexec_memory_operand): Likewise.
6380 (frv_emit_move): Likewise.
6381 (output_move_single): Likewise.
6382 (output_condmove_single): Likewise.
6383 (frv_hard_regno_mode_ok): Likewise.
6384 (frv_matching_accg_mode): Likewise.
6385 * config/h8300/h8300.c (split_adds_subs): Likewise.
6386 (h8300_rtx_costs): Likewise.
6387 (h8300_print_operand): Likewise.
6388 (compute_mov_length): Likewise.
6389 (output_logical_op): Likewise.
6390 (compute_logical_op_length): Likewise.
6391 (compute_logical_op_cc): Likewise.
6392 (h8300_shift_needs_scratch_p): Likewise.
6393 (output_a_shift): Likewise.
6394 (compute_a_shift_length): Likewise.
6395 (compute_a_shift_cc): Likewise.
6396 (expand_a_rotate): Likewise.
6397 (output_a_rotate): Likewise.
6398 * config/i386/i386.c (classify_argument): Likewise.
6399 (function_arg_advance_32): Likewise.
6400 (function_arg_32): Likewise.
6401 (function_arg_64): Likewise.
6402 (function_value_64): Likewise.
6403 (ix86_gimplify_va_arg): Likewise.
6404 (ix86_legitimate_constant_p): Likewise.
6405 (put_condition_code): Likewise.
6406 (split_double_mode): Likewise.
6407 (ix86_avx256_split_vector_move_misalign): Likewise.
6408 (ix86_expand_vector_logical_operator): Likewise.
6409 (ix86_split_idivmod): Likewise.
6410 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6411 (ix86_build_const_vector): Likewise.
6412 (ix86_build_signbit_mask): Likewise.
6413 (ix86_match_ccmode): Likewise.
6414 (ix86_cc_modes_compatible): Likewise.
6415 (ix86_expand_branch): Likewise.
6416 (ix86_expand_sse_cmp): Likewise.
6417 (ix86_expand_sse_movcc): Likewise.
6418 (ix86_expand_int_sse_cmp): Likewise.
6419 (ix86_expand_vec_perm_vpermi2): Likewise.
6420 (ix86_expand_vec_perm): Likewise.
6421 (ix86_expand_sse_unpack): Likewise.
6422 (ix86_expand_int_addcc): Likewise.
6423 (ix86_split_to_parts): Likewise.
6424 (ix86_vectorize_builtin_gather): Likewise.
6425 (ix86_vectorize_builtin_scatter): Likewise.
6426 (avx_vpermilp_parallel): Likewise.
6427 (inline_memory_move_cost): Likewise.
6428 (ix86_tieable_integer_mode_p): Likewise.
6429 (x86_maybe_negate_const_int): Likewise.
6430 (ix86_expand_vector_init_duplicate): Likewise.
6431 (ix86_expand_vector_init_one_nonzero): Likewise.
6432 (ix86_expand_vector_init_one_var): Likewise.
6433 (ix86_expand_vector_init_concat): Likewise.
6434 (ix86_expand_vector_init_interleave): Likewise.
6435 (ix86_expand_vector_init_general): Likewise.
6436 (ix86_expand_vector_set): Likewise.
6437 (ix86_expand_vector_extract): Likewise.
6438 (emit_reduc_half): Likewise.
6439 (ix86_emit_i387_round): Likewise.
6440 (ix86_mangle_type): Likewise.
6441 (ix86_expand_round_sse4): Likewise.
6442 (expand_vec_perm_blend): Likewise.
6443 (canonicalize_vector_int_perm): Likewise.
6444 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6445 (expand_vec_perm_1): Likewise.
6446 (expand_vec_perm_interleave3): Likewise.
6447 (expand_vec_perm_even_odd_pack): Likewise.
6448 (expand_vec_perm_even_odd_1): Likewise.
6449 (expand_vec_perm_broadcast_1): Likewise.
6450 (ix86_vectorize_vec_perm_const_ok): Likewise.
6451 (ix86_expand_vecop_qihi): Likewise.
6452 (ix86_expand_mul_widen_hilo): Likewise.
6453 (ix86_expand_sse2_abs): Likewise.
6454 (ix86_expand_pextr): Likewise.
6455 (ix86_expand_pinsr): Likewise.
6456 (ix86_preferred_simd_mode): Likewise.
6457 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6458 * config/i386/sse.md (*andnot<mode>3): Likewise.
6459 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6460 (*<code><mode>3): Likewise.
6461 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6462 (ia64_expand_atomic_op): Likewise.
6463 (ia64_arg_type): Likewise.
6464 (ia64_mode_to_int): Likewise.
6465 (ia64_scalar_mode_supported_p): Likewise.
6466 (ia64_vector_mode_supported_p): Likewise.
6467 (expand_vec_perm_broadcast): Likewise.
6468 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6469 (iq2000_function_arg_advance): Likewise.
6470 (iq2000_function_arg): Likewise.
6471 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6472 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6473 (m68k_libcall_value): Likewise.
6474 (m68k_function_value): Likewise.
6475 (sched_attr_op_type): Likewise.
6476 * config/mcore/mcore.c (mcore_output_move): Likewise.
6477 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6479 (microblaze_function_arg): Likewise.
6480 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6481 (mips_print_operand): Likewise.
6482 (mips_mode_ok_for_mov_fmt_p): Likewise.
6483 (mips_vector_mode_supported_p): Likewise.
6484 (mips_preferred_simd_mode): Likewise.
6485 (mips_expand_vpc_loongson_even_odd): Likewise.
6486 (mips_expand_vec_unpack): Likewise.
6487 (mips_expand_vi_broadcast): Likewise.
6488 (mips_expand_vector_init): Likewise.
6489 (mips_expand_vec_reduc): Likewise.
6490 (mips_expand_msa_cmp): Likewise.
6491 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6492 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6493 (cc_flags_for_mode): Likewise.
6494 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6495 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6496 (nds32_output_casesi_pc_relative): Likewise.
6497 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6498 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6499 (nvptx_gen_unpack): Likewise.
6500 (nvptx_gen_pack): Likewise.
6501 (nvptx_gen_shuffle): Likewise.
6502 (nvptx_gen_wcast): Likewise.
6503 (nvptx_preferred_simd_mode): Likewise.
6504 * config/pa/pa.c (pa_secondary_reload): Likewise.
6505 * config/pa/predicates.md (base14_operand): Likewise.
6506 * config/powerpcspe/powerpcspe-c.c
6507 (altivec_resolve_overloaded_builtin): Likewise.
6508 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6510 (rs6000_preferred_simd_mode): Likewise.
6511 (output_vec_const_move): Likewise.
6512 (rs6000_expand_vector_extract): Likewise.
6513 (rs6000_split_vec_extract_var): Likewise.
6514 (reg_offset_addressing_ok_p): Likewise.
6515 (rs6000_legitimate_offset_address_p): Likewise.
6516 (rs6000_legitimize_address): Likewise.
6517 (rs6000_emit_set_const): Likewise.
6518 (rs6000_const_vec): Likewise.
6519 (rs6000_emit_move): Likewise.
6520 (spe_build_register_parallel): Likewise.
6521 (rs6000_darwin64_record_arg_recurse): Likewise.
6522 (swap_selector_for_mode): Likewise.
6523 (spe_init_builtins): Likewise.
6524 (paired_init_builtins): Likewise.
6525 (altivec_init_builtins): Likewise.
6526 (do_load_for_compare): Likewise.
6527 (rs6000_generate_compare): Likewise.
6528 (rs6000_expand_float128_convert): Likewise.
6529 (emit_load_locked): Likewise.
6530 (emit_store_conditional): Likewise.
6531 (rs6000_output_function_epilogue): Likewise.
6532 (rs6000_handle_altivec_attribute): Likewise.
6533 (rs6000_function_value): Likewise.
6534 (emit_fusion_gpr_load): Likewise.
6535 (emit_fusion_p9_load): Likewise.
6536 (emit_fusion_p9_store): Likewise.
6537 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6538 (fusion_gpr_mem_load): Likewise.
6539 (fusion_addis_mem_combo_load): Likewise.
6540 (fusion_addis_mem_combo_store): Likewise.
6541 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6542 (fusion_gpr_mem_load): Likewise.
6543 (fusion_addis_mem_combo_load): Likewise.
6544 (fusion_addis_mem_combo_store): Likewise.
6545 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6547 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6548 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6549 (rs6000_preferred_simd_mode): Likewise.
6550 (output_vec_const_move): Likewise.
6551 (rs6000_expand_vector_extract): Likewise.
6552 (rs6000_split_vec_extract_var): Likewise.
6553 (reg_offset_addressing_ok_p): Likewise.
6554 (rs6000_legitimate_offset_address_p): Likewise.
6555 (rs6000_legitimize_address): Likewise.
6556 (rs6000_emit_set_const): Likewise.
6557 (rs6000_const_vec): Likewise.
6558 (rs6000_emit_move): Likewise.
6559 (rs6000_darwin64_record_arg_recurse): Likewise.
6560 (swap_selector_for_mode): Likewise.
6561 (paired_init_builtins): Likewise.
6562 (altivec_init_builtins): Likewise.
6563 (rs6000_expand_float128_convert): Likewise.
6564 (emit_load_locked): Likewise.
6565 (emit_store_conditional): Likewise.
6566 (rs6000_output_function_epilogue): Likewise.
6567 (rs6000_handle_altivec_attribute): Likewise.
6568 (rs6000_function_value): Likewise.
6569 (emit_fusion_gpr_load): Likewise.
6570 (emit_fusion_p9_load): Likewise.
6571 (emit_fusion_p9_store): Likewise.
6572 * config/rx/rx.c (rx_gen_move_template): Likewise.
6573 (flags_from_mode): Likewise.
6574 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6575 (s390_slb_comparison): Likewise.
6576 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6577 (s390_vector_mode_supported_p): Likewise.
6578 (s390_cc_modes_compatible): Likewise.
6579 (s390_match_ccmode_set): Likewise.
6580 (s390_canonicalize_comparison): Likewise.
6581 (s390_emit_compare_and_swap): Likewise.
6582 (s390_branch_condition_mask): Likewise.
6583 (s390_rtx_costs): Likewise.
6584 (s390_secondary_reload): Likewise.
6585 (__SECONDARY_RELOAD_CASE): Likewise.
6586 (s390_expand_cs): Likewise.
6587 (s390_preferred_simd_mode): Likewise.
6588 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6589 * config/sh/sh.c (sh_print_operand): Likewise.
6590 (dump_table): Likewise.
6591 (sh_secondary_reload): Likewise.
6592 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6593 * config/sh/sh.md (casesi_worker_1): Likewise.
6594 (casesi_worker_2): Likewise.
6595 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6596 (fcc_comparison_operator): Likewise.
6597 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6598 (emit_soft_tfmode_cvt): Likewise.
6599 (sparc_preferred_simd_mode): Likewise.
6600 (output_cbranch): Likewise.
6601 (sparc_print_operand): Likewise.
6602 (sparc_expand_vec_perm_bmask): Likewise.
6603 (vector_init_bshuffle): Likewise.
6604 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6605 (spu_vector_mode_supported_p): Likewise.
6606 (spu_expand_insv): Likewise.
6607 (spu_emit_branch_or_set): Likewise.
6608 (spu_handle_vector_attribute): Likewise.
6609 (spu_builtin_splats): Likewise.
6610 (spu_builtin_extract): Likewise.
6611 (spu_builtin_promote): Likewise.
6612 (spu_expand_sign_extend): Likewise.
6613 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6614 (tilegx_simd_int): Likewise.
6615 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6616 (tilepro_simd_int): Likewise.
6617 * config/v850/v850.c (const_double_split): Likewise.
6618 (v850_print_operand): Likewise.
6619 (ep_memory_offset): Likewise.
6620 * config/vax/vax.c (vax_rtx_costs): Likewise.
6621 (vax_output_int_move): Likewise.
6622 (vax_output_int_add): Likewise.
6623 (vax_output_int_subtract): Likewise.
6624 * config/visium/predicates.md (visium_branch_operator): Likewise.
6625 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6626 (visium_print_operand_address): Likewise.
6627 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6628 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6629 (xtensa_expand_conditional_branch): Likewise.
6630 (xtensa_copy_incoming_a7): Likewise.
6631 (xtensa_output_literal): Likewise.
6632 * dfp.c (decimal_real_maxval): Likewise.
6633 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6636 Alan Hayward <alan.hayward@arm.com>
6637 David Sherwood <david.sherwood@arm.com>
6639 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6640 (mode_nunits_inline): Likewise.
6641 (mode_inner_inline): Likewise.
6642 (mode_unit_size_inline): Likewise.
6643 (mode_unit_precision_inline): Likewise.
6644 (emit_insn_modes_h): Likewise. Also emit a #define of the
6646 (emit_mode_wider): Add an E_ prefix to mode names.
6647 (emit_mode_complex): Likewise.
6648 (emit_mode_inner): Likewise.
6649 (emit_mode_adjustments): Likewise.
6650 (emit_mode_int_n): Likewise.
6651 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6652 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6653 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6654 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6655 (CRC32_BUILTIN, ENTRY): Likewise.
6656 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6657 (aarch64_pop_regs): Likewise.
6658 (aarch64_process_components): Likewise.
6659 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6660 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6661 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6662 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6663 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6664 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6666 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6667 (pcmpistr): Likewise.
6668 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6669 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6670 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6672 * config/rl78/rl78.c (mduc_regs): Likewise.
6673 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6674 (htm_expand_builtin): Likewise.
6675 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6676 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6677 * config/xtensa/xtensa.c (print_operand): Likewise.
6678 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6679 (NUM_MODE_VECTOR_INT): Likewise.
6680 * genoutput.c (null_operand): Likewise.
6681 (output_operand_data): Likewise.
6682 * genrecog.c (print_parameter_value): Likewise.
6683 * lra.c (debug_operand_data): Likewise.
6685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6687 * dwarf2out.c (scompare_loc_descriptor_wide)
6688 (scompare_loc_descriptor_narrow): New functions, split out from...
6689 (scompare_loc_descriptor): ...here.
6690 * expmed.c (emit_store_flag_int): New function, split out from...
6691 (emit_store_flag): ...here.
6693 2017-08-30 Richard Biener <rguenther@suse.de>
6695 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6696 (dwarf2out_early_finish): Move setting of AT_pubnames from
6697 early debug output to early finish.
6699 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6701 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6702 and -mdata-region to the assembler.
6704 2017-08-30 Richard Biener <rguenther@suse.de>
6706 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6708 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6709 (dwarf2out_early_global_decl): Only generate a DIE for the
6710 abstract origin if it doesn't already exist or is a declaration DIE.
6711 (resolve_addr): Do not add the linkage name twice when
6712 generating a stub DIE for the DW_TAG_GNU_call_site target.
6714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6716 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6717 Use machine_mode rather than int for arg1_mode.
6719 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6722 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6723 that the second argument of the built-in functions to unpack
6724 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6725 switch statement instead a lot of if statements.
6726 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6727 Allow 64-bit values to be in Altivec registers as well as
6728 traditional floating point registers.
6729 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6731 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6733 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6734 MAX_REGS_PER_ADDRESS == 1.
6736 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6738 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6739 * config/i386/i386.c (ix86_option_override_internal): Simplify
6740 setting of opts->x_flag_entry.
6742 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6743 Jakub Jelinek <jakub@redhat.com>
6744 Richard Biener <rguenther@suse.de>
6746 PR tree-optimization/81503
6747 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6748 folded constant fits in the target type; reorder tests for clarity.
6750 2017-08-29 Martin Liska <mliska@suse.cz>
6752 * passes.def: Include pass_lower_switch.
6753 * stmt.c (dump_case_nodes): Remove and move to
6754 tree-switch-conversion.
6755 (case_values_threshold): Likewise.
6756 (expand_switch_as_decision_tree_p): Likewise.
6757 (emit_case_decision_tree): Likewise.
6758 (expand_case): Likewise.
6759 (balance_case_nodes): Likewise.
6760 (node_has_low_bound): Likewise.
6761 (node_has_high_bound): Likewise.
6762 (node_is_bounded): Likewise.
6763 (emit_case_nodes): Likewise.
6764 (struct simple_case_node): New struct.
6765 (add_case_node): Remove.
6766 (emit_case_dispatch_table): Use vector instead of case_list.
6767 (reset_out_edges_aux): Remove.
6768 (compute_cases_per_edge): Likewise.
6769 (expand_case): Build list of simple_case_node.
6770 (expand_sjlj_dispatch_table): Use it.
6771 * tree-switch-conversion.c (struct case_node): Moved from
6772 stmt.c and adjusted.
6773 (emit_case_nodes): Likewise.
6774 (node_has_low_bound): Likewise.
6775 (node_has_high_bound): Likewise.
6776 (node_is_bounded): Likewise.
6777 (case_values_threshold): Likewise.
6778 (reset_out_edges_aux): Likewise.
6779 (compute_cases_per_edge): Likewise.
6780 (add_case_node): Likewise.
6781 (dump_case_nodes): Likewise.
6782 (balance_case_nodes): Likewise.
6783 (expand_switch_as_decision_tree_p): Likewise.
6784 (emit_jump): Likewise.
6785 (emit_case_decision_tree): Likewise.
6786 (try_switch_expansion): Likewise.
6787 (do_jump_if_equal): Likewise.
6788 (emit_cmp_and_jump_insns): Likewise.
6789 (fix_phi_operands_for_edge): New function.
6790 (record_phi_operand_mapping): Likewise.
6791 (class pass_lower_switch): New pass.
6792 (pass_lower_switch::execute): New function.
6793 (make_pass_lower_switch): Likewise.
6794 (conditional_probability):
6795 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6796 * tree-pass.h: Add make_pass_lower_switch.
6798 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6801 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6804 2017-08-29 Richard Biener <rguenther@suse.de>
6806 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6807 we do not add a DW_AT_inline attribute twice.
6808 (gen_subprogram_die): Remove code setting DW_AT_inline on
6809 DECL_ABSTRACT_P nodes.
6811 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6813 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6814 calls to internal functions.
6815 (gimplify_modify_expr): Likewise.
6816 * tree-call-cdce.c (use_internal_fn): Likewise.
6817 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6818 (convert_to_divmod): Set the nothrow flag.
6819 * tree-if-conv.c (predicate_mem_writes): Likewise.
6820 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6821 (vectorizable_call): Likewise.
6822 (vectorizable_store): Likewise.
6823 (vectorizable_load): Likewise.
6824 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6825 (vect_recog_mask_conversion_pattern): Likewise.
6827 2017-08-29 Martin Liska <mliska@suse.cz>
6830 * gcc.c (driver_handle_option): Add new argument.
6831 * opts-common.c (handle_option): Pass
6832 target_option_override_hook.
6833 * opts-global.c (lang_handle_option): Add new option.
6834 (set_default_handlers): Add new argument.
6835 (decode_options): Likewise.
6836 * opts.c (target_handle_option): Likewise.
6837 (common_handle_option): Call target_option_override_hook.
6838 * opts.h (struct cl_option_handler_func): Add hook for
6839 target option override.
6840 (struct cl_option_handlers): Likewise.
6841 (set_default_handlers): Add new argument.
6842 (decode_options): Likewise.
6843 (common_handle_option): Likewise.
6844 (target_handle_option): Likewise.
6845 * toplev.c (toplev::main): Pass targetm.target_option.override
6848 2017-08-29 Richard Biener <rguenther@suse.de>
6849 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6851 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6852 life to the active subtree.
6854 2017-08-28 Jeff Law <law@redhat.com>
6856 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6857 derive_equivalences.
6858 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6860 (edge_info::derive_equivalences): New private member function
6862 * tree-ssa-dom.c (class edge_info): Changed from a struct
6863 to a class. Add ctor/dtor, methods and data members.
6864 (edge_info::edge_info): Renamed from allocate_edge_info.
6865 Initialize additional members.
6866 (edge_info::~edge_info): New.
6867 (free_dom_edge_info): Delete the edge info.
6868 (record_edge_info): Use new class & associated member functions.
6869 Tighten forms for testing for edge equivalences.
6870 (record_temporary_equivalences): Iterate over the simple
6871 equivalences rather than assuming there's only one per edge.
6872 (cprop_into_successor_phis): Iterate over the simple
6873 equivalences rather than assuming there's only one per edge.
6874 (optimize_stmt): Use operand_equal_p rather than pointer
6875 equality for mini-DSE code.
6877 2017-08-28 Nathan Sidwell <nathan@acm.org>
6879 * gcc.c (execute): Fold SIGPIPE handling into switch
6880 statement. Adjust internal error message.
6882 2017-08-28 Richard Biener <rguenther@suse.de>
6885 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6886 Do nothing for removed DIEs.
6888 2017-08-28 Richard Biener <rguenther@suse.de>
6890 PR tree-optimization/81977
6891 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6894 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6897 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6898 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6899 using targetm.gen_mem_thread_fence.
6901 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6904 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6905 predicate to register_operand. Reorder operands.
6906 (*btr<mode>): Ditto.
6907 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6908 (*btr<mode>_mask): Ditto.
6910 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6912 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6913 * config/rs6000/xmmintrin.h: New file.
6914 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6916 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6919 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6920 parameter and_insn and return it.
6921 (recombine_lvx_pattern): Insert a copy to ensure availability of
6922 the base register of the copied masking operation at the point of
6923 the instruction replacement.
6924 (recombine_stvx_pattern): Likewise.
6926 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6928 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6929 undocumented switches.
6930 (-mpower9-dform-vector): Likewise.
6931 (-mpower9-dform): Likewise.
6932 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6933 comments to delete references to -mpower9-dform* switches.
6934 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6935 Delete reference to -mpower9-dform* switches, test for
6936 -mpower9-vector instead.
6937 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6938 (OTHER_P9_VECTOR_MASKS): Likewise.
6939 (POWERPC_MASKS): Likewise.
6940 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6941 tests against -mpower9-dform* to -mpower9-vector. Delete code
6942 that checked for -mpower9-dform* consistancy with other options.
6943 Add test for -mpower9-misc to enable other power9 switches.
6944 (rs6000_init_hard_regno_mode_ok): Likewise.
6945 (rs6000_option_override_internal): Likewise.
6946 (rs6000_emit_prologue): Likewise.
6947 (rs6000_emit_epilogue): Likewise.
6948 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6949 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6950 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6952 (emit_fusion_p9_store): Likewise.
6953 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6954 resetting these macros if the assembler does not support ISA 3.0
6956 (TARGET_P9_DFORM_VECTOR): Likewise.
6957 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6958 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6960 2017-08-25 Alan Modra <amodra@gmail.com>
6962 PR rtl-optimization/81747
6963 * cse.c (cse_extended_basic_block): Don't attempt to record
6964 equivalences for degenerate conditional branches that branch
6965 to their fall-through.
6967 2017-08-24 Martin Sebor <msebor@redhat.com>
6970 * gimple-fold.c (size_must_be_zero_p): New function.
6971 (gimple_fold_builtin_memory_op): Call it.
6973 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6975 * config/rs6000/mm_malloc.h: New file.
6977 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6979 PR tree-optimization/81913
6980 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6981 analysis when either IVs in condition can wrap.
6983 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6985 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6986 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6988 2017-08-24 Richard Biener <rguenther@suse.de>
6991 * targhooks.c (default_target_can_inline_p): Properly
6992 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6993 is present and always compare.
6994 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6995 infer -mfpmath=sse from TARGET_SSE_P.
6996 (ix86_can_inline_p): Properly use target_option_default_node when
6997 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6999 2017-08-24 Richard Biener <rguenther@suse.de>
7002 * dwarf2out.c (output_die): Handle flag_generate_offload like
7004 (output_comp_unit): Likewise.
7005 (gen_array_type_die): Likewise.
7006 (dwarf2out_early_finish): Likewise.
7007 (note_variable_value_in_expr): Likewise.
7008 (dwarf2out_finish): Likewise. Adjust assert.
7009 * cgraphunit.c (symbol_table::compile): Move setting of
7010 flag_generate_offload earlier ...
7011 (symbol_table::finalize_compilation_unit): ... here, before
7012 early debug finalization.
7014 2017-08-24 Richard Biener <rguenther@suse.de>
7016 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7017 and ipa-fnsummary.h.
7018 (ix86_can_inline_p): When ix86_fpmath flags do not match
7019 check whether the callee uses FP math at all.
7021 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7024 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7025 instead of TYPE_PRECISION.
7027 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7028 Alan Hayward <alan.hayward@arm.com>
7029 David Sherwood <david.sherwood@arm.com>
7031 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7033 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7035 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7037 (ix86_frame::stack_realign_allocate): New field.
7038 (struct machine_frame_state): Modify comments.
7039 (machine_frame_state::sp_realigned_fp_end): New field.
7040 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7042 (sp_valid_at): Add assertion to assure no attempt to access invalid
7043 offset of a realigned stack.
7044 (fp_valid_at): Likewise.
7045 (choose_baseaddr): Modify comments.
7046 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7047 ix86_expand_prologue.
7048 (ix86_expand_prologue): Modify stack realignment and allocation.
7049 (ix86_expand_epilogue): Modify comments.
7050 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7051 avx2_runtime, avx512f, and avx512f_runtime.
7053 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7055 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7056 (mstackrealign): Do not init to -1.
7057 * config/i386/i386.c (ix86_option_override_internal):
7058 Check opts_set, not opts when setting default value of
7059 opts->x_ix86_force_align_arg_pointer.
7061 2017-08-23 Richard Biener <rguenther@suse.de>
7063 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7064 lang_hooks.decl_printable_name.
7065 * print-rtl-function.c (print_rtx_function): Likewise.
7066 * tree-pretty-print.c (dump_function_header): Likewise.
7068 2017-08-23 Richard Biener <rguenther@suse.de>
7071 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7072 -g0 at compile-time.
7074 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7077 * doc/sourcebuild.texi (Other hardware attributes):
7080 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7083 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7084 Fix single-use check.
7086 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7088 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7090 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7092 * doc/install.texi: Modify to add more details on running selected
7095 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7097 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7098 is combined with -mabi=ms.
7099 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7102 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7104 PR tree-optimization/81488
7105 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7106 and cached_basis fields.
7107 (MAX_SPREAD): New constant.
7108 (alloc_cand_and_find_basis): Initialize new fields.
7109 (clear_visited): New function.
7110 (create_phi_basis_1): Rename from create_phi_basis, set visited
7111 and cached_basis fields.
7112 (create_phi_basis): New wrapper function.
7113 (phi_add_costs_1): Rename from phi_add_costs, add spread
7114 parameter, set visited field, short-circuit when limits reached.
7115 (phi_add_costs): New wrapper function.
7116 (record_phi_increments_1): Rename from record_phi_increments, set
7118 (record_phi_increments): New wrapper function.
7119 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7120 (phi_incr_cost): New wrapper function.
7121 (all_phi_incrs_profitable_1): Rename from
7122 all_phi_incrs_profitable, set visited field.
7123 (all_phi_incrs_profitable): New wrapper function.
7125 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7126 Alan Hayward <alan.hayward@arm.com>
7127 David Sherwood <david.sherwood@arm.com>
7129 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7130 that takes the outer and inner modes.
7131 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7132 comparison as the canonical test for a paradoxical subreg.
7133 * combine.c (simplify_set): Use paradoxical_subreg_p.
7134 (make_extraction): Likewise.
7135 (force_to_mode): Likewise.
7136 (rtx_equal_for_field_assignment_p): Likewise.
7137 (gen_lowpart_for_combine): Likewise.
7138 (simplify_comparison): Likewise.
7139 * cse.c (equiv_constant): Likewise.
7140 * expmed.c (store_bit_field_1): Likewise.
7141 * final.c (alter_subreg): Likewise.
7142 * fwprop.c (propagate_rtx): Likewise.
7143 (forward_propagate_subreg): Likewise.
7144 * ira-conflicts.c (ira_build_conflicts): Likewise.
7145 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7146 * lra-constraints.c (curr_insn_transform): Likewise.
7147 (split_reg): Likewise.
7148 * lra-eliminations.c (move_plus_up): Likewise.
7149 (lra_eliminate_regs_1): Likewise.
7150 * recog.c (general_operand): Likewise.
7151 * ree.c (combine_reaching_defs): Likewise.
7152 * reload.c (push_reload): Likewise.
7153 (find_reloads): Likewise.
7154 * reload1.c (elimination_effects): Likewise.
7155 (compute_reload_subreg_offset): Likewise.
7156 (choose_reload_regs): Likewise.
7157 * rtlanal.c (subreg_lsb_1): Likewise.
7158 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7159 (simplify_subreg): Likewise.
7160 * var-tracking.c (track_loc_p): Likewise.
7161 * emit-rtl.c (byte_lowpart_offset): Likewise.
7162 (paradoxical_subreg_p): Delete out-of-line definition.
7164 2017-08-22 Jeff Law <law@redhat.com>
7166 PR tree-optimization/81741
7167 PR tree-optimization/71947
7168 * tree-ssa-dom.c: Include tree-inline.h.
7169 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7170 equivalences if one is more expensive to compute than the other.
7171 * tree-ssa-scopedtables.h (class const_or_copies): Make
7172 record_const_or_copy_raw method private.
7173 (class avail_exprs_stack): New method simplify_binary_operation.
7174 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7175 avail_exprs_stack::simplify_binary_operation as needed.
7176 (avail_exprs_stack::simplify_binary_operation): New function.
7178 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7180 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7181 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7182 (DOT_SYMBOLS): Likewise.
7183 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7184 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7185 (RS6000_ABI_NAME): Likewise.
7186 (TARGET_CMODEL): Likewise.
7187 (TOC_SECTION_ASM_OP): Likewise.
7188 (SET_CMODEL): New macro.
7189 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7191 2017-08-22 Richard Biener <rguenther@suse.de>
7193 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7194 to allow for free-lang-data replacements similar to verify_type_variant.
7196 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7197 Alan Hayward <alan.hayward@arm.com>
7198 David Sherwood <david.sherwood@arm.com>
7200 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7201 VOIDmode for the LABEL_REF.
7203 2017-08-22 Richard Biener <rguenther@suse.de>
7205 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7207 2017-08-22 Richard Biener <rguenther@suse.de>
7209 * common.opt (feliminate-dwarf2-dups): Ignore.
7210 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7211 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7212 same_die_p_wrap, compute_section_prefix,
7213 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7214 (comdat_symbol_id, comdat_symbol_number): Likewise.
7215 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7217 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7218 (output_die): Mark unreachable path unreachable.
7219 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7220 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7221 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7222 (dwarf2out_early_finish): Likewise.
7224 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7226 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7228 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7231 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7232 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7233 (avr_attribute_table) <io, io_low, address>: Initialize
7234 .decl_required with true.
7236 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7238 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7239 undocumented debugging options.
7240 (-mvsx-scalar-double): Likewise.
7241 (-mallow-df-permute): Likewise.
7242 (-mvectorize-builtins): Likewise.
7243 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7244 (rs6000_builtin_vectorized_function): Likewise.
7245 (rs6000_builtin_md_vectorized_function): Likewise.
7246 (rs6000_opt_vars): Likewise.
7248 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7251 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7252 (*btrq_imm): Rename from *btrq.
7253 (*btcq_imm): Rename from *btcq.
7254 (btsc): New code attribute.
7255 (*<btsc><mode>): New insn pattern.
7256 (*btr<mode>): Ditto.
7257 (*<btsc><mode>_mask): New insn_and_split pattern.
7258 (*btr<mode>_mask): Ditto.
7260 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7261 Alan Hayward <alan.hayward@arm.com>
7262 David Sherwood <david.sherwood@arm.com>
7264 * function.c (pad_below): Simplify padding calculation.
7266 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7267 Alan Hayward <alan.hayward@arm.com>
7268 David Sherwood <david.sherwood@arm.com>
7270 * target.def (function_prologue): Remove frame size argument.
7271 (function_epilogue): Likewise.
7272 * doc/tm.texi: Regenerate.
7273 * final.c (final_start_function): Update call to function_prologue.
7274 (final_end_function): Update call to function_epilogue.
7275 (default_function_pro_epilogue): Remove frame size argument.
7276 * output.h (default_function_pro_epilogue): Likewise.
7277 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7278 (arm_output_function_prologue): Likewise.
7279 * config/frv/frv.c (frv_function_prologue): Likewise.
7280 (frv_function_epilogue): Likewise.
7281 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7282 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7283 (ia64_output_function_epilogue): Likewise.
7284 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7285 (m32r_output_function_epilogue): Likewise.
7286 * config/microblaze/microblaze.c (microblaze_function_prologue)
7287 (microblaze_function_epilogue): Likewise.
7288 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7289 (mips_output_function_epilogue): Likewise.
7290 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7291 (mmix_target_asm_function_epilogue): Likewise.
7292 * config/msp430/msp430.c (msp430_start_function): Likewise.
7293 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7294 (nds32_asm_function_epilogue): Likewise.
7295 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7296 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7297 (pa_output_function_epilogue): Likewise.
7298 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7299 (rs6000_output_function_epilogue): Likewise.
7300 * config/rl78/rl78.c (rl78_start_function): Likewise.
7301 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7302 (rs6000_output_function_epilogue): Likewise.
7303 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7304 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7305 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7306 (sparc_asm_function_epilogue): Likewise.
7308 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7310 * tree.h (type_has_mode_precision_p): New function.
7311 * convert.c (convert_to_integer_1): Use it.
7312 * expr.c (expand_expr_real_2): Likewise.
7313 (expand_expr_real_1): Likewise.
7314 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7315 * match.pd: Likewise.
7316 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7317 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7318 * tree-tailcall.c (process_assignment): Likewise.
7319 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7320 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7321 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7322 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7323 (vectorizable_assignment): Likewise.
7324 (vectorizable_shift): Likewise.
7325 (vectorizable_operation): Likewise.
7326 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7328 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7330 * match.pd: Add pow (C, x) simplification.
7332 2017-08-21 Richard Biener <rguenther@suse.de>
7334 PR tree-optimization/81900
7335 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7336 for blocks with abnormal predecessors.
7337 (compute_antic): Do not set visited flag prematurely.
7339 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7342 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7344 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7346 * stor-layout.h (vector_type_mode): Move to...
7347 * tree.h (vector_type_mode): ...here.
7348 * stor-layout.c (vector_type_mode): Move to...
7349 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7351 2017-08-21 Richard Biener <rguenther@suse.de>
7353 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7354 register_external_die hooks.
7355 (debug_false_tree_charstarstar_uhwistar): Declare.
7356 (debug_nothing_tree_charstar_uhwi): Likewise.
7357 * debug.c (do_nothing_debug_hooks): Adjust.
7358 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7359 (debug_nothing_tree_charstar_uhwi): Likewise.
7360 * dbxout.c (dbx_debug_hooks): Adjust.
7361 (xcoff_debug_hooks): Likewise.
7362 * sdbout.c (sdb_debug_hooks): Likewise.
7363 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7364 * dwarf2out.c (macinfo_label_base): New global.
7365 (dwarf2out_register_external_die): New function for the
7366 register_external_die hook.
7367 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7368 (dwarf2_debug_hooks): Use them.
7369 (dwarf2_lineno_debug_hooks): Adjust.
7370 (struct die_struct): Add with_offset flag.
7371 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7372 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7373 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7374 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7375 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7376 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7377 defining section names for the early LTO debug variants.
7378 (reset_indirect_string): New helper.
7379 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7380 (print_dw_val): Add support for offsetted symbol references.
7381 (get_ultimate_context): Split out from is_cxx.
7382 (is_cxx): Use get_ultimate_context.
7383 (is_fortran): Add decl overload.
7384 (compute_comp_unit_symbol): Split out worker from
7385 compute_section_prefix.
7386 (compute_section_prefix): Call compute_comp_unit_symbol and
7387 set comdat_type_p here.
7388 (output_die): Skip DIE symbol output for the LTO added one.
7389 Handle DIE symbol references with offset.
7390 (output_comp_unit): Guard section name mangling properly.
7391 For LTO debug sections emit a symbol at the section beginning
7392 which we use to refer to its DIEs.
7393 (add_abstract_origin_attribute): For DIEs registered via
7394 dwarf2out_register_external_die directly refer to the early
7395 DIE rather than indirectly through the shadow one we created.
7396 Remove obsolete call to dwarf2out_abstract_function for
7397 non-function/block origins.
7398 (gen_array_type_die): When generating early LTO debug do
7399 not emit DW_AT_string_length.
7400 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7401 late when in LTO. As suggested place a gcc_unreachable for
7402 the DECL_ABSTRACT_P case.
7403 (gen_subprogram_die): Avoid another specification DIE
7404 for early built declarations/definitions for the late LTO case.
7405 (gen_variable_die): Add type references for late duplicated VLA dies
7407 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7408 we have the abstract instance already.
7409 (process_scope_var): Adjust decl DIE contexts in LTO which
7410 first puts them in limbo.
7411 (gen_decl_die): Do not generate type DIEs late apart from
7412 types for VLAs or for decls we do not yet have a DIE. Do not
7413 call dwarf2out_abstract_function late.
7414 (dwarf2out_early_global_decl): Make sure to create DIEs
7415 for abstract instances of a decl first.
7416 (dwarf2out_late_global_decl): Adjust comment.
7417 (output_macinfo_op): With multiple macro sections use
7418 macinfo_label_base to distinguish labels.
7419 (output_macinfo): Likewise. Update macinfo_label_base.
7420 Pass in the line info label.
7421 (note_variable_value_in_expr): When generating LTO resolve
7422 all variable values here by generating DIEs as needed.
7423 (init_sections_and_labels): Add early LTO debug flag parameter
7424 and generate different sections and names if set. Add generation
7425 counter for the labels so we can have multiple of them.
7426 (reset_dies): Helper to allow DIEs to be output multiple times.
7427 (dwarf2out_finish): When outputting DIEs to the fat part of an
7428 LTO object first reset DIEs.
7429 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7430 (modified_type_die): Check for decl_ultimate_origin being self
7432 (gen_type_die_with_usage): Likewise.
7433 (gen_typedef_die): Allow decl_ultimate_origin being self.
7434 (set_decl_abstract_flags): Remove.
7435 (set_block_abstract_flags): Likewise.
7436 (dwarf2out_abstract_function): Treat the early generated DIEs
7437 as the abstract copy and only add DW_AT_inline and
7438 DW_AT_artificial here and call set_decl_origin_self.
7439 If the DIE has an abstract origin don't do anything.
7440 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7441 if we have none yet (Go fails to build one, PR78628).
7442 (variably_modified_type_p): Prevent endless recursion for Ada
7443 cyclic pointer types.
7444 * lto-streamer-in.c: Include debug.h.
7445 (dref_queue): New global.
7446 (lto_read_tree_1): Stream in DIE references.
7447 (lto_input_tree): Register DIE references.
7448 (input_function): Stream DECL_DEBUG_ARGS.
7449 * lto-streamer-out.c: Include debug.h.
7450 (lto_write_tree_1): Output DIE references.
7451 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7452 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7453 (output_function): Stream DECL_DEBUG_ARGS.
7454 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7455 Stream DECL_ABSTRACT_ORIGIN.
7456 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7457 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7458 DECL_CONTEXT for file-scope decls.
7459 * lto-streamer.h (struct dref_entry): Declare.
7460 (dref_queue): Likewise.
7461 * cfgexpand.c (pass_expand::execute): Do not call the
7462 outlining_inline_function hook here.
7463 * lto-wrapper.c (debug_obj): New global.
7464 (tool_cleanup): Unlink it if required.
7465 (debug_objcopy): New function.
7466 (run_gcc): Handle early debug sections in the IL files by
7467 extracting them to separate files, partially linkin them and
7468 feeding the result back as result to the linker.
7469 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7470 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7471 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7472 sections into a separate segment.
7473 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7475 (darwin_asm_dwarf_section): Likewise.
7476 (darwin_asm_output_dwarf_offset): Likewise.
7477 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7479 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7480 Alan Hayward <alan.hayward@arm.com>
7481 David Sherwood <david.sherwood@arm.com>
7483 * read-md.h (md_reader::record_potential_iterator_use): Replace
7484 pointer argument with an rtx and an index.
7485 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7486 (apply_mode_iterator): Likewise.
7487 (apply_code_iterator): Likewise.
7488 (apply_int_iterator): Likewise.
7489 (apply_subst_iterator): Likewise.
7490 (record_iterator_use): Likewise.
7491 (record_attribute_use): Likewise.
7492 (md_reader::record_potential_iterator_use): Likewise. Update calls
7493 to record_iterator_use and apply_iterator.
7494 (iterator_use): Replace ptr with x and index.
7495 (attribute_use): Likewise.
7496 (apply_attribute_uses): Update calls to apply_iterator.
7497 (apply_iterators): Likewise. Update initialization of iterator_use.
7498 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7499 and record_potential_iterator_use.
7500 (rtx_reader::read_rtx_operand): Likewise.
7502 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7503 Alan Hayward <alan.hayward@arm.com>
7504 David Sherwood <david.sherwood@arm.com>
7506 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7509 2017-08-21 Richard Biener <rguenther@suse.de>
7512 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7513 at struct end conservatively when comparing common bases.
7515 2017-08-21 Richard Biener <rguenther@suse.de>
7517 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7518 (mem_ref_in_stmt): Remove.
7519 (determine_max_movement): Use ref index to get at the reference.
7520 (invariantness_dom_walker::before_dom_children): Deal with
7521 lim data already initialized.
7522 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7524 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7526 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7527 (TARGET_ISA_ROUND): Ditto.
7528 (TARGET_ROUND): Ditto.
7529 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7530 * config/i386/i386.md: Ditto.
7531 * config/i386/sse.md: Ditto.
7532 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7533 with OPTION_MASK_ISA_SSE4_1.
7535 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7538 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7539 __builtin_ia32_lzcnt_u16.
7541 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7544 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7545 (rs6000_set_current_function): Rewrite function to use it.
7547 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7550 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7551 and TYPE_WARN_IF_NOT_ALIGN.
7552 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7553 (handle_warn_if_not_align): New.
7554 (place_union_field): Call handle_warn_if_not_align.
7555 (place_field): Call handle_warn_if_not_align.
7556 Copy TYPE_WARN_IF_NOT_ALIGN.
7557 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7558 (layout_type): Likewise.
7559 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7561 (tree_decl_common): Add warn_if_not_align.
7562 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7563 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7564 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7565 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7566 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7567 * doc/extend.texi: Document warn_if_not_aligned attribute.
7568 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7570 2017-08-17 Martin Liska <mliska@suse.cz>
7573 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7574 (get_data_dependence): Use it as pointer type.
7575 (distribute_loop): Likewise.
7577 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7579 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7580 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7581 (p8_vmrgow_<mode>_direct): New define_insn.
7582 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7583 handle endianness for vmrgew and vmrgow permute patterns.
7585 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7587 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7588 * config/rs6000/rs6000-cpus.def: Remove comment.
7589 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7590 (POWERPC_MASKS): Likewise.
7591 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7592 use of TARGET_VSX_TIMODE.
7593 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7594 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7595 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7596 (rs6000_option_override_internal): Remove dead code.
7597 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7598 (rs6000_legitimize_reload_address): Likewise.
7599 (rs6000_legitimate_address_p): Likewise.
7600 (rs6000_opt_masks): Delete "vsx-timode".
7601 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7602 from function comment.
7603 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7604 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7605 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7607 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7608 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7609 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7612 (define_peephole2): Likewise.
7614 2017-08-17 Martin Sebor <msebor@redhat.com>
7617 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7618 past the end of an array.
7619 (test_pp_format): Add test cases.
7621 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7623 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7624 missing ECF_NOTHROW flags.
7626 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7629 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7630 operands residing in integer registers.
7631 (*vsx_le_perm_load_<mode>): Likewise.
7632 (*vsx_le_perm_store_<mode>): Likewise.
7633 (define_peephole2): Add peepholes to optimize the above.
7635 2017-08-17 Marek Polacek <polacek@redhat.com>
7638 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7639 to mimic what shorten_compare did. Change the return type to bool.
7640 (fold_cond_expr_with_comparison): Update call to
7641 operand_equal_for_comparison_p.
7642 (fold_ternary_loc): Likewise.
7644 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7646 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7648 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7650 2017-08-17 Richard Biener <rguenther@suse.de>
7652 * tree-ssa-structalias.c (solve_graph): When propagating
7653 to successors update the graphs succ edges and avoid duplicate work.
7655 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7658 * config/i386/i386.c (ix86_option_override_internal): Save target
7659 specific options after ix86_stack_protector_guard_reg was changed.
7661 2017-08-17 Richard Biener <rguenther@suse.de>
7663 PR tree-optimization/81827
7664 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7665 (new_var_info): Initialize it conservatively.
7666 (get_call_vi): Mark register vars.
7667 (new_scalar_tmp_constraint_exp): Likewise.
7668 (handle_rhs_call): Likewise.
7669 (handle_const_call): Likewise.
7670 (create_function_info_for): Likewise.
7671 (solve_constraints): Sort varinfos to separate register from
7672 non-register vars to pack points-to solution bitmaps during
7675 2017-08-17 Marek Polacek <polacek@redhat.com>
7677 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7679 2017-08-17 Richard Biener <rguenther@suse.de>
7681 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7682 to true when overflow is undefined and we saturated the result.
7684 2017-08-17 Alan Modra <amodra@gmail.com>
7687 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7688 Don't use store multiple if only one reg needs saving.
7689 (interesting_frame_related_regno): New function.
7690 (rs6000_frame_related): Don't emit frame info for regs that
7692 (rs6000_emit_epilogue): Likewise.
7694 2017-08-16 Nathan Sidwell <nathan@acm.org>
7696 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7697 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7698 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7699 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7700 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7701 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7702 (verify_type): Adjust for TYPE_BINFO move.
7703 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7704 process TYPE_BINFO directly.
7705 (hash_tree): Likewise.
7706 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7708 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7711 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7713 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7715 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7718 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7719 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7720 Add ix86_binary_operator_ok to insn constraint.
7721 (*iordi_1_bts): Ditto.
7722 (*xordi_1_btc): Ditto.
7723 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7724 Update corresponding peephole2 pattern.
7728 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7730 PR tree-optimization/81832
7731 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7732 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7734 2017-08-16 Marek Polacek <polacek@redhat.com>
7737 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7738 regarding size_zero_node.
7740 2017-08-16 Martin Liska <mliska@suse.cz>
7743 * config.gcc: Respect previously set extra_objs in case
7746 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7748 PR tree-optimization/81835
7749 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7750 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7751 not depend on the phi.
7753 2017-08-16 Alan Modra <amodra@gmail.com>
7755 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7758 2017-08-16 Alan Modra <amodra@gmail.com>
7760 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7761 (save_reg_p): ..into this. Update all callers.
7762 (first_reg_to_save): Simplify.
7764 2017-08-16 Alan Modra <amodra@gmail.com>
7766 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7769 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7773 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7774 constant count if that count is less than 32.
7776 2017-08-15 Nathan Sidwell <nathan@acm.org>
7778 * gcc.c (execute): Emit friendlier message if inferior is killed
7779 by an external cause.
7781 2017-08-15 Richard Biener <rguenther@suse.de>
7783 PR tree-optimization/81790
7784 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7785 CONSTRUCTORs from simplifying and VN.
7787 2017-08-14 Martin Sebor <msebor@redhat.com>
7789 * builtin-attrs.def: Add comments.
7791 2017-08-14 Martin Sebor <msebor@redhat.com>
7794 * doc/extend.texi (attribute nonstring): Document new attribute.
7796 2017-08-14 Martin Sebor <msebor@redhat.com>
7799 * tree-diagnostic.c (default_tree_printer): Handle %G.
7800 * gimple-pretty-print.h (percent_G_format): Declare new function.
7801 * gimple-pretty-print.c (percent_G_format): Define.
7802 * tree-pretty-print.c (percent_K_format): Add argument.
7804 2017-08-14 Martin Sebor <msebor@redhat.com>
7806 PR translation/79998
7807 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7808 Remove a stray space.
7810 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7813 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7814 (*iordi_1_bts): Ditto.
7815 (*xordi_1_btc): Ditto.
7817 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7820 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7821 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7823 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7824 quoted strings, and make more translator-friendly.
7825 (darwin_rs6000_override_options): Likewise.
7826 (rs6000_option_override_internal): Likewise.
7827 (rs6000_return_in_memory): Fix overlong line.
7828 (init_cmulative_args): Use quoted strings, and make more
7829 translator-friendly.
7830 (rs6000_pass_by_reference): Fix overlong line.
7831 (def_builtin): Use quoted strings.
7832 (altivec_expand_predicate_builtin): Use quoted strings, and make
7833 more translator-friendly.
7834 (htm_expand_builtin): Use quoted strings.
7835 (cpu_expand_builtin): Use quoted strings, and make more
7836 translator-friendly.
7837 (altivec_expand_builtin): Likewise.
7838 (paired_expand_predicate_builtin): Likewise.
7839 (rs6000_invalid_builtin): Likewise.
7840 (builtin_function_type): Use quoted strings.
7841 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7842 more translator-friendly.
7843 (rs6000_trampoline_init): Likewise.
7844 (rs6000_handle_altivec_attribute): Likewise.
7845 (rs6000_inner_target_options): Use quoted strings.
7846 (rs6000_disable_incompatible_switches): Likewise.
7847 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7848 strings, and make more translator-friendly.
7849 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7851 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7853 PR tree-optimization/81799
7854 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7855 cond_expr to simple gimple operand.
7857 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7860 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7862 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7866 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7867 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7868 TARGET_GASISR_PROLOGUES.
7869 * config/avr/avr.c (avr_option_override): Same.
7870 (avr_pass_pre_proep::execute): Same.
7872 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7875 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7876 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7878 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7880 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7881 define_insn_and_split. Split to a memory load from 0 in
7882 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7883 using PTR mode iterator.
7884 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7885 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7886 (*add_tp_<mode>): Redefine as define_insn_and_split.
7887 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7888 address space. Merge with *add_tp_x32 using PTR mode iterator.
7889 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7890 Split to an add with a memory load from 0 in
7891 DEFAULT_TLS_SEG_REG address space.
7893 2017-08-12 Andrew Pinski <apinski@cavium.com>
7895 * config/aarch64/aarch64-option-extensions.def (rdma):
7896 Fix feature string to what Linux prints out in /proc/cpuinfo.
7898 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7901 * dwarf2out.c (modified_type_die): For C typedef types that have
7902 an ultimate origin, process the ultimate origin instead of the
7904 (gen_typedef_die): Assert that input DECLs have no ultimate
7906 (gen_type_die_with_usage): For typedef variants that have an
7907 ultimate origin, just call gen_decl_die on the original DECL.
7908 (process_scope_var): Avoid creating DIEs for local typedefs and
7909 concrete static variables.
7911 2017-08-12 Alan Modra <amodra@gmail.com>
7915 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7916 match gnu-user.h startfile.
7917 (ENDFILE_LINUX_SPEC): Similarly.
7919 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7922 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7924 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7926 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7927 * config/aarch64/aarch64.md (mov<mode>): Change.
7928 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7929 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7930 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7932 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7934 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7935 for storage order barriers.
7937 2017-08-11 Martin Liska <mliska@suse.cz>
7940 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7941 variables of void type.
7943 2017-08-11 Martin Liska <mliska@suse.cz>
7945 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7946 TARGET_SUPPORTS_ALIASES.
7947 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7948 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7949 (optimize_weakref): Likewise.
7950 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7951 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7952 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7954 2017-08-11 Martin Liska <mliska@suse.cz>
7957 * config/i386/i386.c (make_resolver_func): Do complete
7958 refactoring of the function.
7960 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7963 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7964 * config/i386/i386.c (ix86_stack_protect_guard): Use
7965 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7966 * doc/invoke.texi (x86 Options): Document
7967 -mstack-protector-guard-symbol= option.
7969 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7971 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7972 * config/i386/i386.c (ix86_split_stack_guard): New function.
7973 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7974 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7975 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7976 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7977 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7978 (split_stack_space_check): Call ix86_split_stack_guard.
7980 2017-08-10 Martin Sebor <msebor@redhat.com>
7982 * print-tree.c (print_node): Print location using the established
7984 Replace spaces with colons.
7985 (debug_raw, debug): Ditto.
7987 2017-08-10 Martin Sebor <msebor@redhat.com>
7990 * pretty-print.c (pp_format): Correct the handling of %s precision.
7992 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7995 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7997 (ix86_finalize_stack_frame_flags): This. Also clear
7998 frame_pointer_needed if -fno-omit-frame-pointer is used without
8000 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8001 with ix86_finalize_stack_frame_flags.
8002 (ix86_expand_epilogue): Likewise.
8003 (ix86_expand_split_stack_prologue): Likewise.
8004 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8006 2017-08-10 Martin Liska <mliska@suse.cz>
8009 * c-attribs.c (handle_target_attribute):
8010 Report warning for an empty string argument of target attribute.
8012 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8015 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8017 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8018 or DECL_NONLOCAL labels.
8019 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8020 or DECL_NONLOCAL labels here.
8022 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8024 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8025 to indicate when early gimple folding has been disabled.
8026 (rs6000_gimple_fold_builtin): Add debug content.
8027 (rs6000_invalid_builtin): Fix whitespace.
8028 (rs6000_expand_builtin): Fix whitespace.
8029 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8031 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8034 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8035 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8037 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8039 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8040 (qdf24xx): Likewise.
8041 * config/aarch64/aarch64.md: Include falkor.md.
8042 * config/aarch64/falkor.md: New.
8044 2017-08-09 Marek Polacek <polacek@redhat.com>
8047 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8048 * diagnostic.c (emit_diagnostic): Add a comment.
8049 (emit_diagnostic_valist): New function.
8051 2017-08-09 Marek Polacek <polacek@redhat.com>
8054 * input.c (make_location): New overload.
8055 * input.h (make_location): Declare.
8057 2017-08-08 Alan Modra <amodra@gmail.com>
8058 H.J. Lu <hongjiu.lu@intel.com>
8061 * gcc.c (NO_PIE_SPEC): Delete.
8062 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8064 (LINK_PIE_SPEC): ..to here.
8065 (LINK_COMMAND_SPEC): Support -no-pie.
8066 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8067 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8069 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8070 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8071 (ENDFILE_CRTEND_SPEC): Similarly.
8073 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8076 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8077 (mstack-protector-guard-offset=): Ditto.
8078 * config/i386/i386.c (ix86_option_override): Handle
8079 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8081 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8082 ix86_stack_protect_guard_offset variables.
8083 (TARGET_STACK_PROTECT_GUARD): Always define.
8084 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8085 and -mstack-protector-guard-offset= options.
8087 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8089 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8090 boundary case for the last candidate.
8092 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8094 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8095 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8097 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8100 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8101 * config/aarch64/aarch64-builtins.c
8102 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8103 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8104 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8106 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8107 Andrew Pinski <pinskia@gmail.com>
8110 * internal-fn.def (XORSIGN): New.
8111 * optabs.def (xorsign_optab): New.
8112 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8113 (convert_expand_mult_copysign): New.
8114 (pass_optimize_widening_mul::execute): Call
8115 convert_expand_mult_copysign.
8117 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8119 PR tree-optimization/81354
8120 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8121 Insert on edges rather than explicitly creating landing pads.
8122 (analyze_candidates_and_replace): Commit edge inserts.
8124 2017-08-08 Richard Biener <rguenther@suse.de>
8127 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8128 (expand_simple_operations): Also look through ADDR_EXPRs with
8129 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8131 2017-08-08 Richard Biener <rguenther@suse.de>
8133 PR tree-optimization/81723
8134 * tree-vect-slp.c (struct bst_traits): New hash traits.
8135 (bst_fail): New global.
8136 (vect_build_slp_tree_2): New worker, split out from ...
8137 (vect_build_slp_tree): ... this now wrapping it with using
8138 bst_fail set to cache SLP tree build fails. Properly handle
8140 (vect_analyze_slp_instance): Allocate and free bst_fail.
8142 2017-08-08 Martin Liska <mliska@suse.cz>
8145 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8146 LABEL_DECLs that can be from a different function.
8148 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8150 PR tree-optimization/81744
8151 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8152 loop's number of iterations.
8154 2017-08-08 Martin Liska <mliska@suse.cz>
8156 * asan.c: Include header files.
8157 * attribs.c (build_decl_attribute_variant): New function moved
8159 (build_type_attribute_qual_variant): Likewise.
8160 (cmp_attrib_identifiers): Likewise.
8161 (simple_cst_list_equal): Likewise.
8162 (omp_declare_simd_clauses_equal): Likewise.
8163 (attribute_value_equal): Likewise.
8164 (comp_type_attributes): Likewise.
8165 (build_type_attribute_variant): Likewise.
8166 (lookup_ident_attribute): Likewise.
8167 (remove_attribute): Likewise.
8168 (merge_attributes): Likewise.
8169 (merge_type_attributes): Likewise.
8170 (merge_decl_attributes): Likewise.
8171 (merge_dllimport_decl_attributes): Likewise.
8172 (handle_dll_attribute): Likewise.
8173 (attribute_list_equal): Likewise.
8174 (attribute_list_contained): Likewise.
8175 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8176 (lookup_attribute_by_prefix): Likewise.
8177 * bb-reorder.c: Include header files.
8178 * builtins.c: Likewise.
8179 * calls.c: Likewise.
8180 * cfgexpand.c: Likewise.
8181 * cgraph.c: Likewise.
8182 * cgraphunit.c: Likewise.
8183 * convert.c: Likewise.
8184 * dwarf2out.c: Likewise.
8185 * final.c: Likewise.
8186 * fold-const.c: Likewise.
8187 * function.c: Likewise.
8188 * gimple-expr.c: Likewise.
8189 * gimple-fold.c: Likewise.
8190 * gimple-pretty-print.c: Likewise.
8191 * gimple.c: Likewise.
8192 * gimplify.c: Likewise.
8193 * hsa-common.c: Likewise.
8194 * hsa-gen.c: Likewise.
8195 * internal-fn.c: Likewise.
8196 * ipa-chkp.c: Likewise.
8197 * ipa-cp.c: Likewise.
8198 * ipa-devirt.c: Likewise.
8199 * ipa-fnsummary.c: Likewise.
8200 * ipa-inline.c: Likewise.
8201 * ipa-visibility.c: Likewise.
8203 * lto-cgraph.c: Likewise.
8204 * omp-expand.c: Likewise.
8205 * omp-general.c: Likewise.
8206 * omp-low.c: Likewise.
8207 * omp-offload.c: Likewise.
8208 * omp-simd-clone.c: Likewise.
8209 * opts-global.c: Likewise.
8210 * passes.c: Likewise.
8211 * predict.c: Likewise.
8212 * sancov.c: Likewise.
8213 * sanopt.c: Likewise.
8214 * symtab.c: Likewise.
8215 * toplev.c: Likewise.
8216 * trans-mem.c: Likewise.
8217 * tree-chkp.c: Likewise.
8218 * tree-eh.c: Likewise.
8219 * tree-into-ssa.c: Likewise.
8220 * tree-object-size.c: Likewise.
8221 * tree-parloops.c: Likewise.
8222 * tree-profile.c: Likewise.
8223 * tree-ssa-ccp.c: Likewise.
8224 * tree-ssa-live.c: Likewise.
8225 * tree-ssa-loop.c: Likewise.
8226 * tree-ssa-sccvn.c: Likewise.
8227 * tree-ssa-structalias.c: Likewise.
8228 * tree-ssa.c: Likewise.
8229 * tree-streamer-in.c: Likewise.
8230 * tree-vectorizer.c: Likewise.
8231 * tree-vrp.c: Likewise.
8233 * ubsan.c: Likewise.
8234 * varasm.c: Likewise.
8235 * varpool.c: Likewise.
8236 * tree.c: Remove functions moved to attribs.[ch].
8238 * config/aarch64/aarch64.c: Add attrs.h header file.
8239 * config/alpha/alpha.c: Likewise.
8240 * config/arc/arc.c: Likewise.
8241 * config/arm/arm.c: Likewise.
8242 * config/avr/avr.c: Likewise.
8243 * config/bfin/bfin.c: Likewise.
8244 * config/c6x/c6x.c: Likewise.
8245 * config/cr16/cr16.c: Likewise.
8246 * config/cris/cris.c: Likewise.
8247 * config/darwin.c: Likewise.
8248 * config/epiphany/epiphany.c: Likewise.
8249 * config/fr30/fr30.c: Likewise.
8250 * config/frv/frv.c: Likewise.
8251 * config/ft32/ft32.c: Likewise.
8252 * config/h8300/h8300.c: Likewise.
8253 * config/i386/winnt.c: Likewise.
8254 * config/ia64/ia64.c: Likewise.
8255 * config/iq2000/iq2000.c: Likewise.
8256 * config/lm32/lm32.c: Likewise.
8257 * config/m32c/m32c.c: Likewise.
8258 * config/m32r/m32r.c: Likewise.
8259 * config/m68k/m68k.c: Likewise.
8260 * config/mcore/mcore.c: Likewise.
8261 * config/microblaze/microblaze.c: Likewise.
8262 * config/mips/mips.c: Likewise.
8263 * config/mmix/mmix.c: Likewise.
8264 * config/mn10300/mn10300.c: Likewise.
8265 * config/moxie/moxie.c: Likewise.
8266 * config/msp430/msp430.c: Likewise.
8267 * config/nds32/nds32-isr.c: Likewise.
8268 * config/nds32/nds32.c: Likewise.
8269 * config/nios2/nios2.c: Likewise.
8270 * config/nvptx/nvptx.c: Likewise.
8271 * config/pa/pa.c: Likewise.
8272 * config/pdp11/pdp11.c: Likewise.
8273 * config/powerpcspe/powerpcspe.c: Likewise.
8274 * config/riscv/riscv.c: Likewise.
8275 * config/rl78/rl78.c: Likewise.
8276 * config/rx/rx.c: Likewise.
8277 * config/s390/s390.c: Likewise.
8278 * config/sh/sh.c: Likewise.
8279 * config/sol2.c: Likewise.
8280 * config/sparc/sparc.c: Likewise.
8281 * config/spu/spu.c: Likewise.
8282 * config/stormy16/stormy16.c: Likewise.
8283 * config/tilegx/tilegx.c: Likewise.
8284 * config/tilepro/tilepro.c: Likewise.
8285 * config/v850/v850.c: Likewise.
8286 * config/vax/vax.c: Likewise.
8287 * config/visium/visium.c: Likewise.
8288 * config/xtensa/xtensa.c: Likewise.
8290 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8293 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8294 constraints since the -mupper-regs-* switches have been
8296 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8297 into a vector from a double word element that was extracted from
8298 another vector, and eliminate extra XXPERMDI instructions.
8299 (vsx_concat_<mode>_2): Likewise.
8300 (vsx_concat_<mode>_3): Likewise.
8301 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8302 concat to allow optimizing inserts from previous extracts.
8304 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8306 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8307 memory reference to a SSP offset in TLS address space.
8308 (ix86_print_operand) <case '@'>: Remove.
8309 (ix86_print_operand_punct_valid_p): Remove '@' code.
8310 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8312 (stack_tls_protect_set_<mode>): Remove.
8313 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8314 (stack_tls_protect_test_<mode>): Remove.
8315 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8317 2017-08-07 Olivier Hainque <hainque@adacore.com>
8320 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8322 2017-08-07 Douglas Rupp <rupp@adacore.com>
8324 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8325 variable was referenced as multidir in command.
8327 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8330 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8333 2017-08-07 Martin Liska <mliska@suse.cz>
8335 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8336 * config/rl78/rl78.c: Add include of attribs.h.
8337 * config/sh/sh.c: Likewise.
8338 * config/v850/v850.c: Likewise.
8340 2017-08-07 Tom de Vries <tom@codesourcery.com>
8343 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8345 2017-08-07 Martin Liska <mliska@suse.cz>
8347 * config/mips/mips.c: Include attribs.h.
8349 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8352 * doc/invoke.texi: Document change in behvaior for -Ofast for
8355 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8357 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8359 (aarch64_pop_regs): Likewise.
8360 (aarch64_gen_load_pair): Likewise.
8361 (aarch64_save_callee_saves): Likewise.
8362 (aarch64_restore_callee_saves): Likewise.
8364 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8366 * config/i386/i386.c: Revert the last change.
8368 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8371 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8373 (ix86_finalize_stack_frame_flags): This. Also clear
8374 frame_pointer_needed if -fno-omit-frame-pointer is used without
8376 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8377 with ix86_finalize_stack_frame_flags.
8378 (ix86_expand_epilogue): Likewise.
8379 (ix86_expand_split_stack_prologue): Likewise.
8381 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8384 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8385 to P_AES for Westmere.
8387 2017-08-07 Jonathan Yong <10walls@gmail.com>
8389 * config/i386/mingw.opt (fset-stack-executable): Removed.
8390 * config/i386/cygming.opt (fset-stack-executable): Moved
8392 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8394 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8396 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8398 2017-08-07 Marek Polacek <polacek@redhat.com>
8401 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8403 2017-08-07 Martin Liska <mliska@suse.cz>
8405 * attribs.h (canonicalize_attr_name): New function.
8406 (cmp_attribs): Move from c-format.c and adjusted.
8407 (is_attribute_p): Moved from tree.h.
8408 * tree-inline.c: Add new includes.
8409 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8410 (private_is_attribute_p): Remove.
8411 (private_lookup_attribute): Likewise.
8412 (private_lookup_attribute_by_prefix): Simplify.
8413 (remove_attribute): Use is_attribute_p.
8414 * tree.h: Remove removed declarations.
8416 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8419 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8420 instead of computing it in the function. Formatting fix.
8421 (expand_case): Don't rely on default_edge being the first edge,
8422 clear it if removing it, pass default_edge to
8423 emit_case_dispatch_table.
8424 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8427 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8429 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8430 insn in the function, emit NOP after the insn.
8432 2017-08-06 Tom de Vries <tom@codesourcery.com>
8434 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8437 2017-08-06 Tom de Vries <tom@codesourcery.com>
8439 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8442 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8444 PR tree-optimization/57371
8445 * match.pd: New pattern.
8447 2017-08-04 Marek Polacek <polacek@redhat.com>
8450 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8451 perform the computation in offset_int.
8453 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8455 PR tree-optimization/81136
8456 * tree-vectorizer.h: Include tree-hash-traits.h.
8457 (vec_base_alignments): New typedef.
8458 (vec_info): Add a base_alignments field.
8459 (vect_record_base_alignments): Declare.
8460 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8462 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8463 (create_data_ref): Add an is_conditional_in_stmt argument.
8464 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8465 the is_conditional_in_stmt field.
8466 (data_ref_loc): Add an is_conditional_in_stmt field.
8467 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8468 (find_data_references_in_stmt): Update call to create_data_ref.
8469 (graphite_find_data_references_in_stmt): Likewise.
8470 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8471 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8472 (vect_record_base_alignment): New function.
8473 (vect_record_base_alignments): Likewise.
8474 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8475 for nested statements even if we fail to compute a misalignment.
8476 Use pooled base alignments for unconditional references.
8477 (vect_find_same_alignment_drs): Compare base addresses instead
8479 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8480 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8482 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8484 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8485 Add an explicit name for the enum. Use auto_vec for slp_instances
8487 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8489 (_bb_vec_info): Add a constructor and destructor.
8490 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8491 (destroy_loop_vec_info): Delete.
8492 (vect_destroy_datarefs): Likewise.
8493 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8494 (vec_info::vec_info): New function.
8495 (vec_info::~vec_info): Likewise.
8496 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8497 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8498 destroy_loop_vec_info.
8499 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8500 (_loop_vec_info::_loop_vec_info): ...this.
8501 (destroy_loop_vec_info): Replace with...
8502 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8503 the stmt_vec_infos. Leave handling of vec_info information to its
8504 destructor. Remove explicit vector releases.
8505 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8506 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8507 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8508 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8509 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8510 (destroy_bb_vec_info): Replace with...
8511 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8512 information to its destructor.
8513 (vect_slp_analyze_bb_1): Use new and delete instead of
8514 new_bb_vec_info and destroy_bb_vec_info.
8515 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8518 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8520 * tree-data-ref.h (subscript): Add access_fn field.
8521 (data_dependence_relation): Add could_be_independent_p.
8522 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8523 (same_access_functions): Move to tree-data-ref.c.
8524 * tree-data-ref.c (ref_contains_union_access_p): New function.
8525 (access_fn_component_p): Likewise.
8526 (access_fn_components_comparable_p): Likewise.
8527 (dr_analyze_indices): Add a reference to access_fn_component_p.
8528 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8530 (constant_access_functions): Likewise.
8531 (add_other_self_distances): Likewise.
8532 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8533 (initialize_data_dependence_relation): Use XCNEW and remove
8534 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8535 of access functions that have the same type. Allow the
8536 subsequence to end with different bases in some circumstances.
8537 Record the chosen access functions in SUB_ACCESS_FN.
8538 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8539 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8540 (subscript_dependence_tester_1): Likewise dra and drb.
8541 (build_classic_dist_vector): Update calls accordingly.
8542 (subscript_dependence_tester): Likewise.
8543 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8544 DDR_COULD_BE_INDEPENDENT_P.
8545 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8546 comp_alias_ddrs instead of may_alias_ddrs.
8547 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8549 (vect_analyze_data_ref_dependence): Use it if
8550 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8551 distance vectors if that fails.
8552 (dependence_distance_ge_vf): New function.
8553 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8554 LOOP_VINFO_MAY_ALIAS_DDRS.
8556 2017-08-04 Richard Biener <rguenther@suse.de>
8559 * fold-const.c (fold_binary_loc): Properly restrict
8560 minus_var0 && minus_var1 case when associating undefined overflow
8563 2017-08-04 Tom de Vries <tom@codesourcery.com>
8565 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8567 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8569 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8570 Don't start diagnostic messages with a capital letter.
8571 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8573 (rs6000_invalid_builtin): Likewise.
8574 (rs6000_trampoline_init): Likewise.
8576 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8579 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8580 after setting changeable df flags.
8582 2017-08-03 Richard Biener <rguenther@suse.de>
8584 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8585 up if the use is in USE - X.
8587 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8589 * toplev.c (dumpfile.h): New include.
8590 (internal_error_reentered): New static function. Use it...
8591 (internal_error_function): ...here to handle reentered internal_error.
8593 2017-08-03 Richard Biener <rguenther@suse.de>
8596 * fold-const.c (split_tree): Add minus_var and minus_con
8597 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8598 here but always use minus_*.
8599 (associate_trees): Assert we never associate with MINUS_EXPR
8600 and NULL first operand. Do not recurse for PLUS_EXPR operands
8601 when associating as MINUS_EXPR either.
8602 (fold_binary_loc): Track minus_var and minus_con.
8604 2017-08-03 Tom de Vries <tom@codesourcery.com>
8607 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8608 ACCEL_COMPILER, apply finish_options on
8609 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8611 2017-08-03 Tom de Vries <tom@codesourcery.com>
8614 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8615 function_entry_patch_area_size > 0.
8617 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8620 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8621 instead of 10??LU, perform unit multiplication in wide_int,
8622 don't change alloc_object_size_limit if the limit is larger
8625 PR tree-optimization/81655
8626 PR tree-optimization/81588
8627 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8628 the case when ranges[i].low and high are 1 for unsigned type with
8632 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8633 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8635 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8637 * tree-vrp.h: Add include guard.
8639 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8642 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8643 (ud2): New insn pattern.
8644 * config/i386/i386.c (ix86_expand_epilogue):
8645 For naked functions, generate ud2 instead of trap insn.
8647 2017-08-02 Marek Polacek <polacek@redhat.com>
8650 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8652 2017-08-02 Tom de Vries <tom@codesourcery.com>
8653 Cesar Philippidis <cesar@codesourcery.com>
8655 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8656 Add missing edge probabilities.
8658 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8660 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8663 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8666 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8667 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8668 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8669 prologue_seq sequences - if any.
8671 2017-08-02 Richard Biener <rguenther@suse.de>
8673 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8674 via vectors if supported, integer extracts via punning if supported
8675 or otherwise vector extracts.
8677 2017-08-02 Richard Biener <rguenther@suse.de>
8679 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8681 (bitmap_insert_into_set): ... this.
8683 2017-08-02 Richard Biener <rguenther@suse.de>
8685 PR tree-optimization/81633
8687 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8689 PR tree-optimization/71752
8690 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8692 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8694 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8695 (machine_function::call_ms2sysv_pad_out): Remove field.
8696 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8697 (ix86_compute_frame_layout): Likewise.
8699 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8702 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8703 attribute with interrupt attribute.
8705 2017-08-01 Andrew Pinski <apinski@cavium.com>
8707 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8708 BIT_INSERT_EXPR's operand 1
8709 to see if the types precision matches.
8711 2017-08-01 Martin Liska <mliska@suse.cz>
8714 * builtins.c (expand_builtin_memcpy_args): Remove.
8715 (expand_builtin_memcpy): Call newly added function
8716 expand_builtin_memory_copy_args.
8717 (expand_builtin_memcpy_with_bounds): Likewise.
8718 (expand_builtin_mempcpy): Remove last argument.
8719 (expand_builtin_mempcpy_with_bounds): Likewise.
8720 (expand_builtin_memory_copy_args): New function created from
8721 expand_builtin_mempcpy_args with small modifications.
8722 (expand_builtin_mempcpy_args): Remove.
8723 (expand_builtin_stpcpy): Remove unused argument.
8724 (expand_builtin): Likewise.
8725 (expand_builtin_with_bounds): Likewise.
8727 2017-08-01 Martin Liska <mliska@suse.cz>
8730 Make mempcpy more optimal (PR middle-end/70140).
8732 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8735 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8736 __builtin_vec_cmpne verify both arguments are compatible vectors
8737 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8738 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8739 move computation of aligned to after checking the argument types.
8743 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8744 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8747 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8748 Doug Rupp <rupp@adacore.com>
8749 Olivier Hainque <hainque@adacore.com>
8751 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8752 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8754 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8755 for TARGET_OS_CPP_BUILTIN.
8756 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8757 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8759 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8760 passing required abi options to the assembler for EABI configurations.
8761 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8762 of .text.hot and .text.unlikely sections for kernel modules when
8763 using ARM style exceptions.
8764 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8765 options. Add EXTRA_CC1_SPEC.
8766 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8768 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8770 (ARM_TARGET2_DWARF_FORMAT): Define.
8771 * config/arm/t-vxworks: Adjust multilib control to removal of the
8772 Diab command line options.
8774 2017-08-01 Martin Liska <mliska@suse.cz>
8776 PR gcov-profile/81561
8777 * gcov.c (unblock): Make unblocking safe as we need to preserve
8778 index correspondence of blocks and block_lists.
8780 2017-08-01 Richard Biener <rguenther@suse.de>
8782 PR tree-optimization/81181
8783 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8784 (compute_antic): ... end of iteration here.
8786 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8788 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8789 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8790 (ftree-slp-vectorize): Likewise.
8791 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8792 can no longer be set independent of flag_tree_loop_vectorize.
8793 * omp-general.c (emp_max_vf): Likewise.
8794 * opts.c (enable_fdo_optimizations): Remove references to
8795 flag_tree_vectorize, these are now implicit.
8796 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8797 and leave it for the options machinery.
8799 2017-08-01 Martin Liska <mliska@suse.cz>
8802 * builtins.c (expand_builtin_memcpy_args): Remove.
8803 (expand_builtin_memcpy): Call newly added function
8804 expand_builtin_memory_copy_args.
8805 (expand_builtin_memcpy_with_bounds): Likewise.
8806 (expand_builtin_mempcpy): Remove last argument.
8807 (expand_builtin_mempcpy_with_bounds): Likewise.
8808 (expand_builtin_memory_copy_args): New function created from
8809 expand_builtin_mempcpy_args with small modifications.
8810 (expand_builtin_mempcpy_args): Remove.
8811 (expand_builtin_stpcpy): Remove unused argument.
8812 (expand_builtin): Likewise.
8813 (expand_builtin_with_bounds): Likewise.
8815 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8818 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8819 print "ds:" only for immediates in generic address space.
8821 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8824 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8825 (ix86_function_ok_for_sibcall): Return false for naked functions.
8827 2017-08-01 Richard Biener <rguenther@suse.de>
8829 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8830 (compute_antic): Seed worklist with exit block predecessors.
8831 * cfganal.c (dfs_find_deadend): For a cycle return the source
8832 of the edge closing it.
8834 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8836 * config/aarch64/aarch64.c
8837 (aarch64_can_const_movi_rtx_p): Move 0 check.
8839 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8841 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8842 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8844 * match.pd: Ditto in address comparison pattern.
8846 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8848 PR tree-optimization/81627
8849 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8850 closed ssa form for store-store chain.
8852 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8854 PR tree-optimization/81620
8855 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8856 for store-store chain.
8858 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8860 PR tree-optimization/81588
8861 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8862 ranges[i].in_p, invert comparison code ccode. For >/>=,
8863 swap rhs1 and rhs2 and comparison code unconditionally,
8864 for </<= don't do that. Don't swap rhs1/rhs2 again if
8865 ranges[i].in_p, instead invert comparison code ccode if
8866 opcode or oe->rank is BIT_IOR_EXPR.
8869 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8870 a direct optab to conversion optab.
8871 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8872 with GET_MODE_INNER as last argument instead of optab_handler.
8873 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8874 vector extraction if possible and optab is available.
8875 * expr.c (store_constructor): Use convert_optab_handler instead
8876 of optab_handler. Use vector initialization from smaller
8877 vectors if possible and optab is available.
8878 * tree-vect-stmts.c (vectorizable_load): Likewise.
8879 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8881 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8882 of vec_init from half-sized vectors with the same element mode.
8883 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8884 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8885 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8886 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8887 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8888 after mode in gen_vec_extract* calls.
8889 (vec_extract<mode>): Renamed to ...
8890 (vec_extract<mode><ssescalarmodelower>): ... this.
8891 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8892 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8893 element mode after mode in gen_vec_init* calls.
8894 (VEC_INIT_HALF_MODE): New mode iterator.
8895 (vec_init<mode>): Renamed to ...
8896 (vec_init<mode><ssescalarmodelower>): ... this.
8897 (vec_init<mode><ssehalfvecmodelower>): New expander.
8898 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8899 (vec_extractv2sfsf): ... this.
8900 (vec_initv2sf): Renamed to ...
8901 (vec_initv2sfsf): ... this.
8902 (vec_extractv2si): Renamed to ...
8903 (vec_extractv2sisi): ... this.
8904 (vec_initv2si): Renamed to ...
8905 (vec_initv2sisi): ... this.
8906 (vec_extractv4hi): Renamed to ...
8907 (vec_extractv4hihi): ... this.
8908 (vec_initv4hi): Renamed to ...
8909 (vec_initv4hihi): ... this.
8910 (vec_extractv8qi): Renamed to ...
8911 (vec_extractv8qiqi): ... this.
8912 (vec_initv8qi): Renamed to ...
8913 (vec_initv8qiqi): ... this.
8914 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8915 (vec_init<mode>): Renamed to ...
8916 (vec_init<mode><VEC_base_l>): ... this.
8917 (vec_extract<mode>): Renamed to ...
8918 (vec_extract<mode><VEC_base_l>): ... this.
8919 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8920 (vec_initv2sfsf): ... this.
8921 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8922 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8923 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8924 element mode after mode in gen_vec_init* calls.
8925 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8926 (vec_init<mode><Vel>): ... this.
8927 (vec_extract<mode>): Renamed to ...
8928 (vec_extract<mode><Vel>): ... this.
8929 * config/aarch64/iterators.md (Vel): New mode attribute.
8930 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8931 Add element mode after mode in gen_vec_extract* calls.
8932 * config/s390/vector.md (non_vec_l): New mode attribute.
8933 (vec_extract<mode>): Renamed to ...
8934 (vec_extract<mode><non_vec_l>): ... this.
8935 (vec_init<mode>): Renamed to ...
8936 (vec_init<mode><non_vec_l>): ... this.
8937 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8938 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8940 * config/arm/iterators.md (V_elem_l): New mode attribute.
8941 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8942 (vec_extract<mode><V_elem_l>): ... this.
8943 (vec_extractv2di): Renamed to ...
8944 (vec_extractv2didi): ... this.
8945 (vec_init<mode>): Renamed to ...
8946 (vec_init<mode><V_elem_l>): ... this.
8947 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8948 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8949 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8950 Add element mode after gen_vec_extract* calls.
8951 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8952 (vec_init<mode><unitmode>): ... this.
8953 (vec_extract<mode>): Renamed to ...
8954 (vec_extract<mode><unitmode>): ... this.
8955 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8956 (vec_init<mode><unitmode>): ... this.
8957 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8958 (vec_initv2sfsf): ... this.
8959 (vec_extractv2sf): Renamed to ...
8960 (vec_extractv2sfsf): ... this.
8961 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8962 Add element mode after gen_vec_extract* calls.
8963 * config/mips/mips.md (unitmode): New mode iterator.
8964 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8965 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8966 * config/spu/spu.md (inner_l): New mode attribute.
8967 (vec_init<mode>): Renamed to ...
8968 (vec_init<mode><inner_l>): ... this.
8969 (vec_extract<mode>): Renamed to ...
8970 (vec_extract<mode><inner_l>): ... this.
8971 * config/sparc/sparc.md (veltmode): New mode iterator.
8972 (vec_init<VMALL:mode>): Renamed to ...
8973 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8974 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8975 (vec_initv2sisi): ... this.
8976 (vec_initv2sf): Renamed to ...
8977 (vec_initv2sfsf): ... this.
8978 (vec_extractv2sf): Renamed to ...
8979 (vec_extractv2sfsf): ... this.
8980 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8981 (vec_init<mode>): Renamed to ...
8982 (vec_init<mode><VEC_base_l>): ... this.
8983 (vec_extract<mode>): Renamed to ...
8984 (vec_extract<mode><VEC_base_l>): ... this.
8985 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8986 (vec_initv2sfsf): ... this.
8987 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8988 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8989 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8990 gen_vec_init* calls.
8992 2017-08-01 Richard Biener <rguenther@suse.de>
8994 PR tree-optimization/81297
8995 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8996 TREE_OVERFLOW from INTEGER_CSTs.
8998 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9000 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9002 2017-07-31 Carl Love <cel@us.ibm.com>
9004 * config/rs6000/rs6000-c: Add support for built-in functions
9005 vector signed char vec_xl_be (signed long long, signed char *);
9006 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9007 vector signed int vec_xl_be (signed long long, signed int *);
9008 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9009 vector signed long long vec_xl_be (signed long long, signed long long *);
9010 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9011 vector signed short vec_xl_be (signed long long, signed short *);
9012 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9013 vector double vec_xl_be (signed long long, double *);
9014 vector float vec_xl_be (signed long long, float *);
9015 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9016 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9017 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9019 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9020 (altivec_expand_builtin): Add switch statement to call
9021 altivec_expand_xl_be for each builtin.
9022 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9023 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9024 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9025 __builtin_vsx_le_be_v16qi.
9026 * doc/extend.texi: Update the built-in documentation file for the
9027 new built-in functions.
9029 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9032 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9034 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9036 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9038 * config.gcc: Add z14.
9039 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9040 CPU model numbers for z13s and z14.
9041 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9043 * config/s390/s390-opts.h (enum processor_type): Rename
9044 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9045 * config/s390/s390.c (processor_table): Add field for CPU name to
9046 be passed to Binutils.
9047 (s390_asm_output_machine_for_arch): Use the new field in
9048 processor_table for Binutils.
9049 (s390_expand_builtin): Replace arch12 with z14.
9050 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9051 (s390_get_sched_attrmask): Likewise.
9052 (s390_get_unit_mask): Likewise.
9053 * config/s390/s390.opt: Add z14 to processor_type enum.
9055 2017-07-31 Martin Jambor <mjambor@suse.cz>
9058 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9059 regardless of optimization level.
9061 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9062 Martin Liska <mliska@suse.cz>
9064 * predict.def: Remove old comment and adjust probability.
9065 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9068 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9071 * config/i386/i386.c (ix86_function_naked): New function.
9072 (ix86_can_use_return_insn_p): Return false for naked functions.
9073 (ix86_expand_prologue): Skip prologue for naked functions.
9074 (ix86_expand_epilogue): Skip epilogue for naked functions
9075 and emit trap instruction.
9076 (ix86_warn_func_return): New function.
9077 (ix86_attribute_table): Add "naked" attribute specification.
9078 (TARGET_WARN_FUNC_RETURN): Define.
9079 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9081 2017-07-31 Martin Liska <mliska@suse.cz>
9083 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9084 (dump_gimple_bb_header): Always dump BB info.
9085 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9087 2017-07-31 Martin Liska <mliska@suse.cz>
9090 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9091 also with current_function_decl non-null equality.
9093 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9096 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9097 change type to the element type, instead add eltype variable and
9098 use it where we are interested in the element type.
9100 PR tree-optimization/81603
9101 * ipa-polymorphic-call.c
9102 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9103 offset arithmetic in offset_int, bail out if the resulting bit offset
9104 doesn't fit into shwi.
9106 2017-07-31 Martin Liska <mliska@suse.cz>
9108 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9109 (gimplify_save_expr): Fix comment.
9111 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9114 * config/i386/i386.c (ix86_function_arg): Update arguments for
9116 (ix86_compute_frame_layout): Set the initial stack offset to
9117 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9118 INCOMING_FRAME_SP_OFFSET.
9119 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9120 stack before exception handler returns.
9121 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9122 the 'ERROR_CODE' for exception handler.
9124 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9126 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9127 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9128 (ASM_OUTPUT_REG_POP): Ditto.
9129 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9130 instead of asm_fprintf to output pure string.
9132 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9134 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9135 to imported_module_or_decl hook.
9136 (debug_nothing_tree_tree_tree_bool): Remove.
9137 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9138 * debug.c (do_nothing_debug_hooks): Use
9139 debug_nothing_tree_tree_tree_bool_bool instead of
9140 debug_nothing_tree_tree_tree_bool.
9141 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9142 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9143 * sdbout.c (sdb_debug_hooks): Likewise.
9144 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9145 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9147 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9148 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9149 attribute, don't add anything.
9151 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9153 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9154 (fold_build2_stat_loc): Likewise.
9155 (fold_build3_stat_loc): Likewise.
9156 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9157 (fold_build1_loc): Remove macro.
9158 (fold_build2_loc): Likewise.
9159 (fold_build3_loc): Likewise.
9161 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9163 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9164 (gimple_build_debug_bind_source_stat): Likewise.
9165 * gimple.h (gimple_build_debug_bind): Remove macro.
9166 (gimple_build_debug_bind_source): Likewise.
9168 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9170 * bitmap.c (bitmap_alloc): Adjust.
9171 (bitmap_gc_alloc): Likewise.
9172 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9174 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9176 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9177 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9178 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9179 (bitmap_gc_alloc_stat): Likewise.
9180 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9182 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9184 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9185 * rtl.h (shallow_copy_rtx): Remove macro.
9187 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9189 * emit-rtl.c (gen_raw_REG): Adjust.
9190 * gengenrtl.c (gendef): Likewise.
9191 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9192 * rtl.h (rtx_alloc): Remove macro.
9194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9196 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9197 (build_tree_list_stat): Likewise.
9198 * tree.h (build_tree_list): Remove macro.
9199 (build_tree_list_vec): Likewise.
9201 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9203 * tree.c (make_vector_stat): Remove _stat from name.
9204 (build_vector_stat): Likewise.
9205 * tree.h (make_vector_stat): Remove macro.
9206 (build_vector_stat): Likewise.
9208 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9210 * tree.h (build_var_debug_value): Remove prototype.
9212 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9214 * tree.c (tree_cons_stat): Remove _stat from name.
9215 * tree.h (tree_cons): Remove macro.
9217 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9219 * tree.c (build_vl_exp_stat): Remove _stat from name.
9220 * tree.h (build_vl_exp): Remove macro.
9222 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9224 * tree.c (build_decl_stat): Remove _stat from name.
9225 * tree.h (build_decl): Remove macro.
9227 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9229 * gimple.c (gimple_build_with_ops_stat): Adjust.
9230 (gimple_alloc_stat): Remove _stat from name.
9231 * gimple.h (gimple_alloc): Remove macro.
9233 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9235 * tree.c (make_tree_vec_stat): Remove _stat from name.
9236 (grow_tree_vec_stat): Likewise.
9237 * tree.h (make_tree_vec_stat): Adjust prototype.
9238 (grow_tree_vec_stat): Likewise.
9239 (make_tree_vec): Remove macro.
9240 (grow_tree_vec): Likewise.
9242 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9244 * fold-const.c (fold_build1_stat_loc): Adjust.
9245 (fold_build2_stat_loc): Likewise.
9246 (fold_build3_stat_loc): Likewise.
9247 * tree.c (build0_stat): Remove _stat from name.
9248 (build1_stat): Likewise.
9249 (build2_stat): Likewise.
9250 (build3_stat): Likewise.
9251 (build4_stat): Likewise.
9252 (build5_stat): Likewise.
9253 * tree.h (build1_loc): Remove macro, and rename _stat function
9255 (build2_loc): Likewise.
9256 (build3_loc): Likewise.
9257 (build4_loc): Likewise.
9258 (build5_loc): Likewise.
9260 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9262 * tree.c (make_int_cst_stat): Remove _stat from name.
9263 * tree.h (make_int_cst_stat): Adjust prototype.
9264 (make_int_cst): Remove macro.
9266 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9268 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9269 * tree.h (make_tree_binfo_stat): Adjust prototype.
9270 (make_tree_binfo): Remove.
9272 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9274 * tree.c (copy_node_stat): Rename to copy_node.
9275 (build_distinct_type_copy): Adjust.
9276 * tree.h (copy_node_stat): Adjust prototype.
9277 (copy_node): Remove macro.
9279 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9281 * tree.c (make_node_stat): rename to make_node.
9282 (build_tree_list_stat): Adjust.
9283 (build0_stat): Likewise.
9284 (build2_stat): Likewise.
9285 (build3_stat): Likewise.
9286 (build4_stat): Likewise.
9287 (build5_stat): Likewise.
9288 (build_decl_stat): Likewise.
9289 * tree.h (make_node_stat): Adjust prototype.
9290 (make_node): remove macro.
9292 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9294 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9295 (PPC_FEATURE2_SCV): Likewise.
9296 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9298 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9300 * config/aarch64/aarch64.c
9301 (aarch64_internal_mov_immediate): Add new special pattern.
9302 * config/aarch64/aarch64.md (*movdi_aarch64):
9303 Add reg/32bit const mov case.
9305 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9306 Richard Sandiford <richard.sandiford@linaro.org>
9308 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9309 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9310 Add integer and movi cases.
9311 (movi-split-hf-df-sf split, fp16): New.
9312 (enabled): Added TARGET_FP_F16INST.
9313 * config/aarch64/iterators.md (GPF_HF): New.
9314 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9316 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9318 * config/aarch64/aarch64.c
9319 (aarch64_simd_container_mode): Add prototype.
9320 (aarch64_expand_mov_immediate): Add HI support.
9321 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9322 (aarch64_can_const_movi_rtx_p): New.
9323 (aarch64_preferred_reload_class):
9324 Remove restrictions of using FP registers for certain SIMD operations.
9325 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9326 (aarch64_valid_floating_const): Add integer move validation.
9327 (aarch64_simd_imm_scalar_p): Remove.
9328 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9329 (aarch64_legitimate_constant_p): Expand list of supported cases.
9330 * config/aarch64/aarch64-protos.h
9331 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9332 (aarch64_reinterpret_float_as_int): New.
9333 (aarch64_simd_imm_scalar_p): Remove.
9334 * config/aarch64/constraints.md (Uvi): New.
9335 (Dd): Split into Ds and new Dd.
9336 * config/aarch64/aarch64.md (*movsi_aarch64):
9338 (*movdi_aarch64): Add SIMD mov case.
9340 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9342 * tree-predcom.c: (struct chain): Handle store-store chain in which
9343 stores for elimination only store loop invariant values.
9344 (execute_pred_commoning_chain): Ditto.
9345 (prepare_initializers_chain_store_elim): Ditto.
9346 (prepare_finalizers): Ditto.
9347 (is_inv_store_elimination_chain): New function.
9348 (initialize_root_vars_store_elim_1): New function.
9350 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9352 * tree-predcom.c: Revise general description of the pass.
9353 (enum chain_type): New enum type for store elimination.
9354 (struct chain): New field supporting store elimination.
9355 (struct component): Ditto.
9356 (dump_chain): Dump store-stores chain.
9357 (release_chain): Release resources.
9358 (split_data_refs_to_components): Compute and create component
9359 contains only stores for elimination.
9360 (get_chain_last_ref_at): New function.
9361 (make_invariant_chain): Initialization.
9362 (make_rooted_chain): Specify chain type in parameter and record it.
9363 (add_looparound_copies): Skip for store-stores chain.
9364 (determine_roots_comp): Compute type of chain and pass it to
9366 (initialize_root_vars_store_elim_2): New function.
9367 (finalize_eliminated_stores): New function.
9368 (remove_stmt): Handle store for elimination.
9369 (execute_pred_commoning_chain): Execute predictive commoning on
9371 (determine_unroll_factor): Skip unroll for store-stores chain.
9372 (prepare_initializers_chain_store_elim): New function.
9373 (prepare_initializers_chain): Hanlde store-store chain.
9374 (prepare_finalizers_chain, prepare_finalizers): New function.
9375 (tree_predictive_commoning_loop): Return integer value indicating
9376 if loop is unrolled or lcssa form is corrupted.
9377 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9379 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9381 * tree-predcom.c (initialize_root): Delete.
9382 (execute_pred_commoning_chain): Initialize root vars and replace
9383 reference of non-combined chain directly, rather than call above
9386 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9388 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9389 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9391 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9393 * tree-predcom.c (struct chain): New field init_seq.
9394 (release_chain): Release init_seq.
9395 (prepare_initializers_chain): Record intialization stmts in above
9397 (insert_init_seqs): New function.
9398 (tree_predictive_commoning_loop): Call insert_init_seqs.
9400 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9402 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9404 2017-07-28 Richard Biener <rguenther@suse.de>
9406 * match.pd: Remove superfluous :c.
9407 * genmatch.c (simplify::id): Add member.
9408 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9410 (current_id): New global.
9411 (dt_node::parent): Move from ...
9412 (dt_operand::parent): ... here. Add for_id member.
9413 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9414 (decision_tree::find_node): Relax order requirement when
9415 merging DT_TRUE nodes to ones inbetween the current simplify
9416 and the one we try to merge with. Add diagnostic whenever
9417 we need to enforce pattern order by not merging.
9418 (decision_tree::insert): Set current_id.
9419 (decision_tree::print_node): Dump parent node and for_id.
9420 (parser::last_id): Add member.
9421 (parser::push_simplify): Assign unique id.
9422 (parser::parser): Initialize last_id.
9424 2017-07-28 Martin Liska <mliska@suse.cz>
9427 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9428 gimple_build_debug_bind.
9430 2017-07-28 Richard Biener <rguenther@suse.de>
9432 PR tree-optimization/81502
9433 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9435 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9437 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9438 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9439 for BIT_FIELD_REF args.
9440 * fold-const.c (make_bit_field_ref): Likewise.
9441 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9443 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9446 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9447 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9448 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9449 Or it into SANITIZER_UNDEFINED.
9450 * ubsan.c: Include gimple-fold.h and varasm.h.
9451 (ubsan_expand_ptr_ifn): New function.
9452 (instrument_pointer_overflow): New function.
9453 (maybe_instrument_pointer_overflow): New function.
9454 (instrument_object_size): Formatting fix.
9455 (pass_ubsan::execute): Call instrument_pointer_overflow
9456 and maybe_instrument_pointer_overflow.
9457 * internal-fn.c (expand_UBSAN_PTR): New function.
9458 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9459 * sanitizer.def (__ubsan_handle_pointer_overflow,
9460 __ubsan_handle_pointer_overflow_abort): New builtins.
9461 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9462 * internal-fn.def (UBSAN_PTR): New internal function.
9463 * opts.c (sanitizer_opts): Add pointer-overflow.
9464 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9465 * fold-const.c (build_range_check): Compute pointer range check in
9466 integral type if pointer arithmetics would be needed. Formatting
9469 2017-07-28 Martin Liska <mliska@suse.cz>
9472 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9473 parameters that are of a variable-length.
9475 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9477 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9479 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9480 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9481 (CRT_CALL_STATIC_FUNCTION): Likewise.
9482 (ASM_DEFAULT_SPEC): New define.
9483 (ASM_SPEC32): Likewise.
9484 (ASM_SPEC64): Likewise.
9485 (ASM_SPEC_COMMON): Likewise.
9486 (ASM_SPEC): Likewise.
9487 (INVALID_64BIT): Likewise.
9488 (LINK_OS_DEFAULT_SPEC): Likewise.
9489 (LINK_OS_SPEC32): Likewise.
9490 (LINK_OS_SPEC64): Likewise.
9491 (POWERPC_LINUX): Likewise.
9492 (PTRDIFF_TYPE): Likewise.
9493 (RESTORE_FP_PREFIX): Likewise.
9494 (RESTORE_FP_SUFFIX): Likewise.
9495 (SAVE_FP_PREFIX): Likewise.
9496 (SAVE_FP_SUFFIX): Likewise.
9497 (SIZE_TYPE): Likewise.
9498 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9499 (TARGET_64BIT): Likewise.
9500 (TARGET_64BIT): Likewise.
9501 (TARGET_AIX): Likewise.
9502 (WCHAR_TYPE_SIZE): Likewise.
9503 (WCHAR_TYPE): Undefine.
9504 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9505 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9506 (CPP_OS_RTEMS_SPEC): Delete.
9507 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9508 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9510 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9512 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9514 PR tree-optimization/81578
9515 * tree-parloops.c (build_new_reduction): Bail out if
9516 reduction_code isn't one of the standard OpenMP reductions.
9517 Move the details printing after that decision.
9519 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9521 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9522 related to reload_in_progress.
9523 (splat_input_operand): Likewise.
9524 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9526 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9528 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9529 (TARGET_INSTANTIATE_DECLS): Likewise.
9530 (legitimate_indexed_address_p): Delete reload_in_progress code.
9531 (rs6000_debug_legitimate_address_p): Likewise.
9532 (rs6000_eliminate_indexed_memrefs): Likewise.
9533 (rs6000_emit_le_vsx_store): Likewise.
9534 (rs6000_emit_move_si_sf_subreg): Likewise.
9535 (rs6000_emit_move): Likewise.
9536 (register_to_reg_type): Likewise.
9537 (rs6000_pre_atomic_barrier): Likewise.
9538 (rs6000_machopic_legitimize_pic_address): Likewise.
9539 (rs6000_allocate_stack_temp): Likewise.
9540 (rs6000_address_for_fpconvert): Likewise.
9541 (rs6000_address_for_altivec): Likewise.
9542 (rs6000_secondary_memory_needed_rtx): Delete function.
9543 (rs6000_check_sdmode): Likewise.
9544 (rs6000_alloc_sdmode_stack_slot): Likewise.
9545 (rs6000_instantiate_decls): Likewise.
9546 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9547 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9548 Delete reload_in_progress.
9549 (*vec_reload_and_plus_<mptrsize>): Likewise.
9550 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9551 (vsx_div_v2di): Likewise.
9552 (vsx_udiv_v2di): Likewise.
9554 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9556 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9557 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9558 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9559 (rs6000_debug_reg_global): Delete print of LRA status.
9560 (rs6000_option_override_internal): Delete dead LRA related code.
9561 (rs6000_lra_p): Delete function.
9562 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9564 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9566 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9567 * config/riscv/rtems.h: New file.
9569 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9570 Sudakshina Das <sudi.das@arm.com>
9572 * config/aarch64/aarch64.md
9573 (define_split for and<mode>3nr_compare): Move
9574 non aarch64_logical_operand to a register.
9575 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9576 register immediate operand to a register.
9577 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9579 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9582 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9584 2017-07-27 Richard Biener <rguenther@suse.de>
9586 PR tree-optimization/81573
9587 PR tree-optimization/81494
9588 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9589 multi defuse cycle case.
9591 2017-07-27 Richard Biener <rguenther@suse.de>
9593 PR tree-optimization/81571
9594 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9597 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9599 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9600 earlier and only if MASK_FPU is set. Adjust formatting.
9602 2017-07-27 Martin Liska <mliska@suse.cz>
9604 * opt-functions.awk: Add validation of value of Init.
9605 * optc-gen.awk: Pass new argument.
9607 2017-07-27 Martin Liska <mliska@suse.cz>
9609 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9610 Fix wrong condition.
9612 2017-07-27 Martin Liska <mliska@suse.cz>
9614 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9615 BBs and edges seen by autoFDO.
9617 2017-07-27 Richard Biener <rguenther@suse.de>
9619 PR tree-optimization/81502
9620 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9621 with incompatible but same sized type.
9622 (execute_update_addresses_taken): Likewise.
9624 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9626 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9627 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9629 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9632 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9633 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9634 Change s_operand to memory_operand.
9636 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9638 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9639 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9640 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9641 Emit instructions rather than returning an expression. Handle TFmode
9642 and KFmode by casting to TImode.
9643 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9644 (rs6000_emit_le_vsx_store): Likewise.
9645 * config/rs6000/vsx.md (VSX_TI): New iterator.
9646 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9647 (*vsx_le_undo_permute_<mode>): Likewise.
9648 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9649 emit the split sequence.
9650 (*vsx_le_perm_store_<mode>): Likewise.
9652 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9654 PR tree-optimization/81555
9655 PR tree-optimization/81556
9656 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9657 if true, force CHANGED for the recursive invocation.
9658 (reassociate_bb): Remember original length of ops array, pass
9659 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9661 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9662 attributes for noipa attribute. For naked attribute use
9663 lookup_attribute first before lookup_attribute_spec.
9664 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9666 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9667 for functions with noipa attribute.
9668 (cgraph_externally_visible_p): Return true for functions with noipa
9670 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9671 for functions with noipa attribute.
9672 * doc/extend.texi: Document noipa function attribute.
9673 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9674 also for functions with noipa attribute.
9675 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9677 2017-07-26 Andrew Pinski <apinski@cavium.com>
9679 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9680 vec_unalign_load_cost and vec_unalign_store_cost.
9682 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9684 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9685 -mvsx-small-integer option.
9686 (ISA_3_0_MASKS_IEEE): Likewise.
9687 (OTHER_VSX_VECTOR_MASKS): Likewise.
9688 (POWERPC_MASKS): Likewise.
9689 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9690 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9691 code, only testing for DImode being allowed in non-VSX floating
9693 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9694 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9696 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9697 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9698 TARGET_P8_VECTOR test.
9699 (rs6000_secondary_reload_simple_move): Likewise.
9700 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9701 since TARGET_P9_VECTOR was already tested.
9702 (rs6000_opt_masks): Remove -mvsx-small-integer.
9703 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9704 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9706 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9707 test for TARGET_VEXTRACTUB was used, and that uses
9709 (p9 extract splitter): Likewise.
9710 (vsx_extract_<mode>_di_p9): Likewise.
9711 (vsx_extract_<mode>_store_p9): Likewise.
9712 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9713 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9714 the elimination of TARGET_VSX_SMALL_INTEGER.
9715 (vsx_extract_<mode>_p8): Likewise.
9716 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9717 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9718 (vsx_set_<mode>_p9): Likewise.
9719 (vsx_set_v4sf_p9): Likewise.
9720 (vsx_set_v4sf_p9_zero): Likewise.
9721 (vsx_insert_extract_v4sf_p9): Likewise.
9722 (vsx_insert_extract_v4sf_p9_2): Likewise.
9723 * config/rs6000/rs6000.md (sign extend splitter): Change
9724 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9725 (floatsi<mode>2_lfiwax_mem): Likewise.
9726 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9727 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9728 since a test for TARGET_P9_VECTOR was used.
9729 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9730 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9731 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9732 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9733 TARGET_P8_VECTOR test.
9734 (fix_trunc<mode>si2_stfiwx): Likewise.
9735 (fix_trunc<mode>si2_internal): Likewise.
9736 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9737 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9739 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9740 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9741 TARGET_P8_VECTOR test.
9742 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9743 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9744 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9746 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9747 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9748 since a test for TARGET_P9_VECTOR was used.
9749 (splitter for loading small constants): Likewise.
9751 2017-07-26 Andrew Pinski <apinski@cavium.com>
9753 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9756 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9759 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9760 (fp_valid_at): Likewise.
9762 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9764 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9765 (qdf24xx_addrcost_table): Likewise.
9766 (cortexa57_tunings): Update to use generic_branch_cost.
9767 (cortexa72_tunings): Likewise.
9768 (cortexa73_tunings): Likewise.
9769 (qdf24xx_tunings): Likewise.
9771 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9773 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9774 (thunderx2t99_branch_cost): Likewise.
9775 (cortexa35_tunings): Update to use generic_branch_cost.
9776 (cortexa53_tunings): Likewise.
9777 (cortexa57_tunings): Likewise.
9778 (cortexa72_tunings): Likewise.
9779 (cortexa73_tunings): Likewise.
9780 (thunderx2t99_tunings): Likewise.
9782 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9784 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9785 (sparc_option_override): Honour MASK_FSMULD.
9786 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9787 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9788 * config/sparc/sparc.opt (mfsmuld): New option.
9789 * doc/invoke.texi (mfsmuld): Document option.
9791 2017-07-26 Marek Polacek <polacek@redhat.com>
9794 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9796 2017-07-26 Richard Biener <rguenther@suse.de>
9798 * gimple-match-head.c (do_valueize): Return OP if valueize
9800 (get_def): New helper to get at the def stmt of a SSA name
9802 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9803 do_valueize to get at the def stmt.
9804 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9806 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9809 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9811 2017-07-26 Martin Liska <mliska@suse.cz>
9814 * function.c (expand_function_start): Make expansion of
9815 nonlocal_goto_save_area after parm_birth_insn.
9817 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9819 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9820 from all CPU target flags enable members.
9822 2017-07-26 Richard Biener <rguenther@suse.de>
9824 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9825 (decision_tree::gen): Make 'type' const.
9826 (write_predicate): Likewise.
9828 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9830 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9831 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9832 (rs6000_option_override_internal): Likewise.
9833 (rs6000_expand_vector_set): Likewise.
9834 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9835 (TARGET_UPPER_REGS_SF): Likewise.
9836 (TARGET_UPPER_REGS_DI): Likewise.
9837 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9838 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9839 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9840 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9841 (Splitters for DI constants in Altivec registers): Likewise.
9842 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9843 (vsx_set_v4sf_p9): Likewise.
9844 (vsx_set_v4sf_p9_zero): Likewise.
9845 (vsx_insert_extract_v4sf_p9): Likewise.
9846 (vsx_insert_extract_v4sf_p9_2): Likewise.
9848 2017-07-25 Carl Love <cel@us.ibm.com>
9850 * doc/extend.texi: Update the built-in documentation file for the
9851 existing built-in functions
9852 vector signed char vec_cnttz (vector signed char);
9853 vector unsigned char vec_cnttz (vector unsigned char);
9854 vector signed short vec_cnttz (vector signed short);
9855 vector unsigned short vec_cnttz (vector unsigned short);
9856 vector signed int vec_cnttz (vector signed int);
9857 vector unsigned int vec_cnttz (vector unsigned int);
9858 vector signed long long vec_cnttz (vector signed long long);
9859 vector unsigned long long vec_cnttz (vector unsigned long long);
9861 2017-07-25 Andrew Pinski <apinski@cavium.com>
9863 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9864 accesses where the use is for the first operand of a BIT_INSERT.
9866 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9869 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9870 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9872 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9874 * config/i386/gstabs.h: Delete.
9875 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9877 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9879 * config/i386/i386.c (ix86_decompose_address): Do not check for
9880 register RTX when looking at index_reg or base_reg.
9881 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9883 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9885 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9886 to update EH info here.
9888 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9890 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9892 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9894 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9896 2017-07-25 Torsten Duwe <duwe@suse.de>
9898 * common.opt: Introduce -fpatchable-function-entry
9899 command line option, and its variables function_entry_patch_area_size
9900 and function_entry_patch_area_start.
9901 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9902 including a two-value parser.
9903 * target.def (print_patchable_function_entry): New target hook.
9904 * targhooks.h (default_print_patchable_function_entry): New function.
9905 * targhooks.c (default_print_patchable_function_entry): Likewise.
9906 * toplev.c (process_options): Switch off IPA-RA if
9907 patchable function entries are being generated.
9908 * varasm.c (assemble_start_function): Look at the
9909 patchable-function-entry command line switch and current
9910 function attributes and maybe generate NOP instructions by
9911 calling the print_patchable_function_entry hook.
9912 * doc/extend.texi: Document patchable_function_entry attribute.
9913 * doc/invoke.texi: Document -fpatchable_function_entry
9914 command line option.
9915 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9917 * doc/tm.texi: Re-generate.
9919 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9922 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9923 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9925 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9927 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9929 2017-07-25 Richard Biener <rguenther@suse.de>
9931 PR tree-optimization/81455
9932 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9933 not walk in cycles when looking for guards.
9935 2017-07-25 Richard Biener <rguenther@suse.de>
9937 PR tree-optimization/81529
9938 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9939 when optimizing backedge uses.
9941 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9943 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9945 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9946 to dl_section_ref. On AIX, append an expression to subtract
9947 the size of the section length to dl_section_ref.
9949 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9951 * configure.ac: If any of the config.* scripts fail, exit 1.
9952 * configure: Regenerate.
9954 2017-07-25 Richard Biener <rguenther@suse.de>
9957 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9958 of immediate uses, be more verbose on errors.
9960 2017-07-25 Richard Biener <rguenther@suse.de>
9962 PR tree-optimization/81510
9963 * tree-vect-loop.c (vect_is_simple_reduction): When the
9964 reduction stmt is not inside the loop bail out.
9966 2017-07-25 Richard Biener <rguenther@suse.de>
9968 PR tree-optimization/81303
9969 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9970 profitability check against LOOP_VINFO_NITERSM1.
9972 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9974 * domwalk.c (cmp_bb_postorder): Simplify.
9975 (sort_bbs_postorder): New function. Use it...
9976 (dom_walker::walk): ...here to optimize common cases.
9978 2017-07-25 Martin Liska <mliska@suse.cz>
9981 * ipa-visibility.c (function_and_variable_visibility): Make the
9982 redirection just on target that supports aliasing.
9983 Fix GNU coding style.
9985 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9988 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9990 * config/arm/rtems.h: Likewise.
9991 * config/bfin/rtems.h: Likewise.
9992 * config/i386/rtemself.h: Likewise.
9993 * config/lm32/rtems.h: Likewise.
9994 * config/m32c/rtems.h: Likewise.
9995 * config/m68k/rtemself.h: Likewise.
9996 * config/microblaze/rtems.h: Likewise.
9997 * config/mips/rtems.h: Likewise.
9998 * config/moxie/rtems.h: Likewise.
9999 * config/nios2/rtems.h: Likewise.
10000 * config/powerpcspe/rtems.h: Likewise.
10001 * config/rs6000/rtems.h: Likewise.
10002 * config/rtems.h: Likewise.
10003 * config/sh/rtems.h: Likewise.
10004 * config/sh/rtemself.h: Likewise.
10005 * config/sparc/rtemself.h: Likewise.
10007 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10010 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10011 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10012 * tree-ssa-structalias.c (alias_get_name): Same.
10014 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10017 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10018 instructions if no du chain is found.
10020 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10022 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10024 2017-07-25 Richard Biener <rguenther@suse.de>
10026 PR middle-end/81505
10027 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10030 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10032 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10033 upper-regs options.
10034 (ISA_2_7_MASKS_SERVER): Likewise.
10035 (ISA_3_0_MASKS_IEEE): Likewise.
10036 (OTHER_P8_VECTOR_MASKS): Likewise.
10037 (OTHER_VSX_VECTOR_MASKS): Likewise.
10038 (POWERPC_MASKS): Likewise.
10039 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10040 duplicate list of options.
10041 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10042 explicit -mupper-regs options.
10043 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10044 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10045 alias for -mupper-regs-df.
10046 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10047 (rs6000_init_hard_regno_mode_ok): Likewise.
10048 (rs6000_option_override_internal): Likewise.
10049 (rs6000_opt_masks): Likewise.
10050 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10051 options in terms of whether -mvsx or -mpower8-vector was used.
10052 (TARGET_UPPER_REGS_DI): Likewise.
10053 (TARGET_UPPER_REGS_SF): Likewise.
10054 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10055 -mupper-regs-* options.
10057 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10059 * passes.c (emergency_dump_function): Print some empty lines and a
10060 header before the RTL dump.
10062 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10064 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10066 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10069 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10070 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10072 2017-07-24 Carl Love <cel@us.ibm.com>
10074 * config/rs6000/rs6000-c.c: Add support for built-in functions
10075 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10076 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10077 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10078 vec_extract_fp_from_shortl): Add defines for the two builtins.
10079 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10080 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10082 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10083 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10084 * doc/extend.texi: Update the built-in documentation file for the
10085 new built-in function.
10087 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10090 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10092 * doc/generic.texi: Likewise.
10093 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10094 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10096 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10098 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10099 (aarch64_mls_elt_merge<mode>): Likewise.
10101 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10103 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10104 having __cxa_atexit.
10106 2017-07-23 Michael Collison <michael.collison@arm.com>
10108 * config/arm/arm.c (arm_option_override): Deprecate
10109 use of -mstructure-size-boundary.
10110 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10111 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10113 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10116 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10117 Reduce cost estimate for direct moves.
10119 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10122 * config/i386/i386.c (ix86_option_override_internal): Disable
10123 BMI, BMI2 and TBM instructions for -m16.
10125 2017-07-21 Carl Love <cel@us.ibm.com>
10127 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10128 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10129 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10130 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10131 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10132 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10133 VMULOSW): New enum "unspec" values.
10134 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10135 altivec_vmulosw): New patterns.
10136 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10137 VMULOSW): Add definitions.
10139 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10141 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10142 (qdf24xx): Likewise.
10143 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10144 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10145 (AARCH64_FL_V8_1): Renumber.
10146 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10147 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10148 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10149 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10150 rdma to feature modifiers list.
10152 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10154 PR middle-end/56727
10155 * ipa-visibility (function_and_variable_visibility): Convert
10156 recursive PLT call to direct call if appropriate.
10158 2017-07-21 Andrew Pinski <apinski@cavium.com>
10160 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10161 operand 1 to see if the types precision matches.
10162 * fold-const.c (operand_equal_p): Likewise.
10164 2017-07-21 Richard Biener <rguenther@suse.de>
10166 PR tree-optimization/81303
10167 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10168 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10169 (vect_peeling_hash_get_lowest_cost): Adjust.
10170 (vect_enhance_data_refs_alignment): Likewise. Use
10171 vect_get_peeling_costs_all_drs to compute the penalty for no
10172 peeling to match up costs.
10174 2017-07-21 Richard Biener <rguenther@suse.de>
10176 PR tree-optimization/81500
10177 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10178 we didn't identify a reduction path.
10180 2017-07-21 Tom de Vries <tom@codesourcery.com>
10181 Cesar Philippidis <cesar@codesourcery.com>
10183 PR gcov-profile/81442
10184 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10187 2017-07-21 Tom de Vries <tom@codesourcery.com>
10190 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10192 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10193 nvptx_override_options_after_change.
10195 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10197 * dwarf2out.c (output_file_names): Avoid double testing for
10198 dwarf_version >= 5.
10200 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10202 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10204 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10206 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10208 (try_crossjump_to_edge): Do not punt on partitioned functions.
10210 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10212 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10213 Put all BBs reachable only via paths crossing cold region to cold
10215 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10217 2016-07-21 Richard Biener <rguenther@suse.de>
10219 PR tree-optimization/81303
10220 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10221 into account prologue and epilogue iterations when raising
10222 min_profitable_iters to sth at least covering one vector iteration.
10224 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10226 * config/arm/arm.c (arm_test_cpu_arch_dat):
10229 2017-07-20 Nathan Sidwell <nathan@acm.org>
10231 Remove TYPE_METHODS.
10232 * tree.h (TYPE_METHODS): Delete.
10233 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10234 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10235 (dbxout_type_methods): Scan TYPE_FIELDS.
10236 (dbxout_type): Don't check TYPE_METHODS here.
10237 * function.c (use_register_for_decl): Always ignore register for
10238 class types when not optimizing.
10239 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10240 * tree.c (free_lang_data_in_type): Stitch out member functions and
10241 templates from TYPE_FIELDS.
10242 (build_distinct_type_copy, verify_type_variant,
10243 verify_type): Member fns are on TYPE_FIELDS.
10244 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10245 * tree-pretty-print.c (dump_generic_node): Likewise.
10247 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10250 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10251 V2TImode and V4TImode.
10252 (ix86_expand_vector_extract): Likewise.
10253 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10254 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10255 (ssescalarmode): Handle V4TImode and V2TImode.
10256 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10257 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10258 (VEXTRACTI128_MODE): New mode iterator.
10259 (splitter for *vec_extractv?ti first element): New.
10260 (VEC_INIT_MODE): New mode iterator.
10261 (vec_init<mode>): Consolidate 3 expanders into one using
10262 VEC_INIT_MODE mode iterator.
10264 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10266 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10267 non_spilled_static_chain_regno_p.
10269 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10271 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10273 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10275 * bb-reorder.c (connect_traces): Allow copying of blocks within
10278 2017-07-20 Richard Biener <rguenther@suse.de>
10280 * gimple.h (gimple_phi_result): Add gphi * overload.
10281 (gimple_phi_result_ptr): Likewise.
10282 (gimple_phi_arg): Likewise. Adjust index assert to only
10283 allow actual argument accesses rather than all slots available
10285 (gimple_phi_arg_def): Add gphi * overload.
10286 * tree-phinodes.c (make_phi_node): Initialize only actual
10288 (resize_phi_node): Clear memory not covered by old node,
10289 do not initialize excess argument slots.
10290 (reserve_phi_args_for_new_edge): Initialize new argument slot
10293 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10295 PR tree-optimization/81388
10297 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10299 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10300 by removing computation of may_be_zero.
10302 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10303 Tom de Vries <tom@codesourcery.com>
10305 PR middle-end/81030
10306 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10307 when gimple level profile disagrees with what RTL expander did.
10309 2017-07-20 Richard Biener <rguenther@suse.de>
10311 PR tree-optimization/61171
10312 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10313 (vect_analyze_stmt): Add slp instance parameter.
10314 (vectorizable_reduction): Likewise.
10315 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10316 (vect_is_simple_reduction): Deal with chains not detected
10317 as SLP reduction chain, specifically not properly associated
10318 chains containing a mix of plus/minus.
10319 (get_reduction_op): Remove.
10320 (get_initial_defs_for_reduction): Simplify, pass in whether
10321 this is a reduction chain, pass in the SLP node for the PHIs.
10322 (vect_create_epilog_for_reduction): Get the SLP instance as
10324 (vectorizable_reduction): Get the SLP instance as arg.
10325 During analysis remember the SLP node with the PHIs in the
10326 instance. Simplify getting at the vectorized reduction PHIs.
10327 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10328 through SLP instance.
10329 (vect_slp_analyze_operations): Likewise.
10330 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10331 (vect_transform_stmt): Likewise.
10333 2017-07-20 Tom de Vries <tom@codesourcery.com>
10335 PR tree-optimization/81489
10336 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10337 read of phi arg location to before loop that modifies phi.
10339 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10341 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10344 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10346 PR middle-end/81331
10347 * except.c (execute): Fix ordering issue.
10349 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10351 PR rtl-optimization/81423
10352 * combine.c (make_compound_operation_int): Don't try to optimize
10353 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10355 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10357 PR rtl-optimization/81423
10358 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10359 with a constant that is -1 in the truncated to mode.
10361 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10363 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10364 (determine_unlikely_bbs): ... here.
10365 * predict.h (propagate_unlikely_bbs_forward): Declare.
10366 * cfgexpand.c (pass_expand::execute): Use it.
10367 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10369 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10370 propagate_unlikely_bbs_forward.
10372 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10374 PR middle-end/81331
10375 * except.c (maybe_add_nop_after_section_switch): New function.
10378 2017-07-19 Tom de Vries <tom@codesourcery.com>
10380 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10382 2017-07-19 Tom de Vries <tom@codesourcery.com>
10384 * gimple.h (gimple_phi_arg): Make assert more strict.
10386 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10388 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10389 * config/rs6000/mmintrin.h: New file.
10390 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10392 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10394 PR tree-optimization/81346
10395 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10397 2017-07-19 Tom de Vries <tom@codesourcery.com>
10399 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10401 2017-07-19 Tom de Vries <tom@codesourcery.com>
10403 * config/nvptx/nvptx-modes.def: Add V2DImode.
10404 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10405 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10406 (nvptx_output_mov_insn): Handle lack of mov.b128.
10407 (nvptx_print_operand): Handle 'H' and 'L' codes.
10408 (nvptx_vector_mode_supported): Allow V2DImode.
10409 (nvptx_preferred_simd_mode): New function.
10410 (nvptx_data_alignment): New function.
10411 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10412 nvptx_preferred_simd_mode.
10413 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10415 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10417 2017-07-19 Tom de Vries <tom@codesourcery.com>
10419 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10420 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10421 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10422 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10423 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10424 (mov<VECIM>_insn): New define_insn.
10425 (define_expand "mov<VECIM>): New define_expand.
10427 2017-07-19 Tom de Vries <tom@codesourcery.com>
10429 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10431 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10433 PR tree-optimization/81346
10434 * fold-const.h (fold_div_compare, range_check_type): Declare.
10435 * fold-const.c (range_check_type): New function.
10436 (build_range_check): Use range_check_type.
10437 (fold_div_compare): No longer static, rewritten into
10438 a match.pd helper function.
10439 (fold_comparison): Don't call fold_div_compare here.
10440 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10441 as helper function.
10443 2017-07-19 Nathan Sidwell <nathan@acm.org>
10445 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10446 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10447 * tree.c (find_decls_types_r, verify_type): Use
10448 TYPE_{MIN,MAX}_VALUE_RAW.
10449 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10450 (hash_tree): Likewise.
10451 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10453 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10456 2017-07-18 Tom de Vries <tom@codesourcery.com>
10458 PR middle-end/81464
10459 * omp-expand.c (expand_omp_for_static_chunk): Handle
10460 equal-argument loop exit phi.
10462 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10465 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10466 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10467 operand 2 predicate.
10468 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10469 operand 2 predicate.
10470 (ror,rol -> rorx splitters): Use const_int_operand as
10471 operand 2 predicate.
10473 2017-06-18 Richard Biener <rguenther@suse.de>
10475 PR tree-optimization/81410
10476 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10477 the gap in the ! slp_perm SLP case after each group.
10479 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10481 PR middle-end/81463
10482 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10485 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10487 PR middle-end/81462
10488 * predict.c (set_even_probabilities): Cleanup; do not affect
10489 probabilities that are already known.
10490 (combine_predictions_for_bb): Call even when count is set.
10492 2017-07-18 Nathan Sidwell <nathan@acm.org>
10494 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10497 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10500 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10501 optimization for loop niter analysis.
10503 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10506 * config/avr/avr.c (avr_optimize_casesi): Don't use
10507 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10509 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10511 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10512 body_cost_vec from _vect_peel_extended_info.
10513 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10514 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10517 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10519 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10521 2017-07-18 Richard Biener <rguenther@suse.de>
10523 PR tree-optimization/80620
10524 PR tree-optimization/81403
10525 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10526 info when re-using a VN table entry.
10528 2017-07-18 Richard Biener <rguenther@suse.de>
10530 PR tree-optimization/81418
10531 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10532 vectype_in. Verify that with lane-reducing reduction operations
10533 we have a single def-use cycle.
10535 2017-07-17 Carl Love <cel@us.ibm.com>
10537 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10539 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10540 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10541 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10542 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10543 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10544 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10545 VMULOSW): New enum "unspec" values.
10546 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10547 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10548 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10549 altivec_vmulosw): New patterns.
10550 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10551 VMULOSW): Add definitions.
10552 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10554 * config/alpha/alpha.c: Include predict.h.
10556 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10558 * tree-vrp.c (compare_assert_loc): Fix comparison function
10559 to return predictable results.
10561 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10563 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10565 (subdi3): Likewise.
10566 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10567 * doc/invoke.texi (mexpand-adddi): Update text.
10569 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10571 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10572 that also clobbers the CC register. The old expand code is moved
10574 (*arc_clzsi2): ... here.
10575 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10576 the CC register. The old expand code is moved to ...
10577 (arc_ctzsi2): ... here.
10579 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10581 * config/arc/arc.opt (mindexed-loads): Use initial value
10582 TARGET_INDEXED_LOADS_DEFAULT.
10583 (mauto-modify-reg): Use initial value
10584 TARGET_AUTO_MODIFY_REG_DEFAULT.
10585 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10586 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10587 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10588 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10590 2017-07-17 Martin Liska <mliska@suse.cz>
10593 * opts.c (finish_options): Do not allow -fgnu-tm
10594 w/ -fsanitize={kernel-,}address. Say sorry.
10596 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10599 * tree-loop-distribution.c (classify_partition): Only assert on
10600 numer of iterations.
10601 (merge_dep_scc_partitions): Delete prameter. Update function call.
10602 (distribute_loop): Remove code handling loop with unknown niters.
10603 (pass_loop_distribution::execute): Skip loop with unknown niters.
10605 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10608 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10609 function sort_partitions_by_post_order.
10611 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10613 PR tree-optimization/81374
10614 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10615 the max index of basic blocks, rather than number of basic blocks.
10617 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10619 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10621 (arc_legitimate_pic_operand_p): Likewise.
10622 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10624 (arc_needs_pcl_p): Likewise.
10625 (arc_legitimate_pc_offset_p): Likewise.
10626 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10627 function is also used in constrains.md.
10628 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10629 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10630 PLUS. Only return true/false in known cases, otherwise assert.
10631 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10632 is already called in arc_legitimate_constant_p.
10633 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10635 (LEGITIMATE_PIC_OPERAND_P): Use
10636 arc_raw_symbolic_reference_mentioned_p function.
10637 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10642 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10643 Andrew Burgess <andrew.burgess@embecosm.com>
10645 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10646 (arc_return_address_register): New function.
10647 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10648 (arc_handle_fndecl_attribute): Add naked attribute.
10649 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10650 (TARGET_WARN_FUNC_RETURN): Likewise.
10651 (arc_allocate_stack_slots_for_args): New function.
10652 (arc_warn_func_return): Likewise.
10653 (machine_function): Change type fn_type.
10654 (arc_compute_function_type): Consider new naked function type,
10655 change function return type.
10656 (arc_must_save_register): Adapt to handle new
10657 arc_compute_function_type's return type.
10658 (arc_expand_prologue): Likewise.
10659 (arc_expand_epilogue): Likewise.
10660 (arc_return_address_regs): Delete.
10661 (arc_return_address_register): New function.
10662 (arc_epilogue_uses): Use above function.
10663 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10664 (arc_function_type): Change encoding, add naked type.
10665 (ARC_INTERRUPT_P): Change to handle the new encoding.
10666 (ARC_FAST_INTERRUPT_P): Likewise.
10667 (ARC_NORMAL_P): Define.
10668 (ARC_NAKED_P): Likewise.
10669 (arc_compute_function_type): Delete prototype.
10670 * config/arc/arc.md (in_ret_delay_slot): Use
10671 arc_return_address_register function.
10672 (simple_return): Likewise.
10673 (p_return_i): Likewise.
10675 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10677 PR tree-optimization/81428
10678 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10679 can't be built for those types.
10681 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10683 Remove stuff dead since r239246.
10685 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10686 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10687 (avr_inform_devices): Remove dead stuff.
10689 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10691 * config/arm/arm_neon.h: Fix softp typo.
10693 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10695 PR tree-optimization/81365
10696 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10697 aggregate moves onto bb predecessor edges, make sure there are no
10698 loads that could alias the lhs in between the start of bb and the
10701 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10704 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10705 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10706 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10708 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10710 PR tree-optimization/81396
10711 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10712 (init_symbolic_number): Initialize it to 1.
10713 (perform_symbolic_merge): Add n_ops from both operands into the new
10715 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10716 without base_addr as useless if they need more than one operation.
10717 (bswap_replace): Handle !bswap case for NULL base_addr.
10719 2017-07-17 Tom de Vries <tom@codesourcery.com>
10722 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10725 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10727 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10728 conditional builtin define __FIX_LEON3FT_B2BST.
10730 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10732 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10733 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10736 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10738 PR rtl-optimization/81424
10739 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10740 to remove potential trapping from operands if -fnon-call-exceptions.
10742 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10744 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10745 profile_proability for scalling.
10746 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10748 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10750 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10752 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10754 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10755 fixpoint arithmetics.
10757 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10759 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10760 fixpoint arithmetics.
10762 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10764 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10765 fixpoint arithmetics.
10767 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10769 * profile-count.h (profile_probability::from_reg_br_prob_note,
10770 profile_probability::to_reg_br_prob_note): New functions.
10771 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10772 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10773 * predict.c (probability_reliable_p): Update.
10774 (edge_probability_reliable_p): Update.
10775 (br_prob_note_reliable_p): Update.
10776 (invert_br_probabilities): Update.
10777 (add_reg_br_prob_note): New function.
10778 (combine_predictions_for_insn): Update.
10779 * asan.c (asan_clear_shadow): Update.
10780 * cfgbuild.c (compute_outgoing_frequencies): Update.
10781 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10782 (update_br_prob_note): Update.
10783 (rtl_verify_edges): Update.
10784 (purge_dead_edges): Update.
10785 (fixup_reorder_chain): Update.
10786 * emit-rtl.c (try_split): Update.
10787 * ifcvt.c (cond_exec_process_insns): Update.
10788 (cond_exec_process_if_block): Update.
10789 (dead_or_predicable): Update.
10790 * internal-fn.c (expand_addsub_overflow): Update.
10791 (expand_neg_overflow): Update.
10792 (expand_mul_overflow): Update.
10793 * loop-doloop.c (doloop_modify): Update.
10794 * loop-unroll.c (compare_and_jump_seq): Update.
10795 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10796 * predict.h: Update.
10797 * reorg.c (mostly_true_jump): Update.
10799 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10800 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10801 * config/arc/arc.c: (emit_unlikely_jump): Update.
10802 * config/arm/arm.c: (emit_unlikely_jump): Update.
10803 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10804 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10805 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10806 (ix86_print_operand): Update.
10807 (ix86_split_fp_branch): Update.
10808 (predict_jump): Update.
10809 * config/ia64/ia64.c (ia64_print_operand): Update.
10810 * config/mmix/mmix.c (mmix_print_operand): Update.
10811 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10812 (rs6000_expand_split_stack_prologue): Update.
10813 * config/rs6000/rs6000.c: Update.
10814 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10815 (s390_expand_vec_movstr): Update.
10816 (s390_expand_cs_tdsi): Update.
10817 (s390_expand_split_stack_prologue): Update.
10818 * config/sh/sh.c (sh_print_operand): Update.
10819 (expand_cbranchsi4): Update.
10820 (expand_cbranchdi4): Update.
10821 * config/sparc/sparc.c (output_v9branch): Update.
10822 * config/spu/spu.c (get_branch_target): Update.
10823 (ea_load_store_inline): Update.
10824 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10825 * config/tilepro/tilepro.c: Update.
10827 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10829 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10830 (gimplify_save_expr): Likewise.
10832 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10834 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10836 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10838 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10840 * timevar.def (TV_IPA_FNSUMMARY): Define.
10842 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10844 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10845 to back store errata sensitive sequence from being generated.
10846 (sqrtdf2_fix): Likewise.
10848 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10850 * tree-ssa-threadupdate.c (compute_path_counts,
10851 update_joiner_offpath_counts): Use profile_probability.
10853 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10856 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10858 * config/arm/arm-c.c (arm_cpu_builtins): Define
10859 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10861 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10863 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10864 array entries to represent __ieee128 versions of the
10865 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10866 scalar_extract_sig, and scalar_insert_exp built-in functions.
10867 (altivec_resolve_overloaded_builtin): Add special case handling
10868 for the __builtin_scalar_insert_exp function, as represented by
10869 the P9V_BUILTIN_VEC_VSIEDP constant.
10870 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10871 exponent support for __ieee128 argument.
10872 (VSESQP): Add scalar extract signature support for __ieee128
10874 (VSTDCNQP): Add scalar test negative support for __ieee128
10876 (VSIEQP): Add scalar insert exponent support for __int128 argument
10877 with __ieee128 result.
10878 (VSIEQPF): Add scalar insert exponent support for __ieee128
10879 argument with __ieee128 result.
10880 (VSTDCQP): Add scalar test data class support for __ieee128
10882 (VSTDCNQP): Add overload support for scalar test negative with
10883 __ieee128 argument.
10884 (VSTDCQP): Add overload support for scalar test data class
10885 __ieee128 argument.
10886 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10887 UNSPEC_VSX_SXSIGDP.
10888 (UNSPEC_VSX_SIEXPQP): New constant.
10889 (xsxexpqp): New insn for VSX scalar extract exponent quad
10891 (xsxsigqp): New insn for VSX scalar extract significand quad
10893 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10894 precision with floating point argument.
10895 (xststdcqp): New expand for VSX scalar test data class quad
10897 (xststdcnegqp): New expand for VSX scalar test negative quad
10899 (xststdcqp): New insn to match expansions for VSX scalar test data
10900 class quad precision and VSX scalar test negative quad precision.
10901 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10902 special case operand checking to enforce that second operand of
10903 VSX scalar test data class with quad precision argument is a 7-bit
10905 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10906 prototypes and descriptions of __ieee128 versions of
10907 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10908 scalar_test_data_class, and scalar_test_neg built-in functions.
10910 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10912 PR tree-optimization/81162
10913 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10914 replace a negate with an add.
10916 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10918 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10920 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10922 * config/arm/arm-c.c (arm_cpu_builtins): Define
10923 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10925 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10927 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10928 (armv8-r): Set ARM Cortex-R52 as default CPU.
10929 * config/arm/arm-tables.opt: Regenerate.
10930 * config/arm/arm-tune.md: Regenerate.
10931 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10933 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10934 extension for -mcpu=cortex-r52.
10936 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10938 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10939 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10940 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10941 (fp-armv8): Define it as FP_ARMv8 only.
10942 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10943 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10945 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10946 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10947 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10948 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10949 than TARGET_FPU_ARMV8.
10950 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10951 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10952 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10954 * config/arm/neon.md (neon_vrint): Likewise.
10955 (neon_vcvt): Likewise.
10956 (neon_<fmaxmin_op><mode>): Likewise.
10957 (<fmaxmin><mode>3): Likewise.
10958 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10959 * config/arm/predicates.md (arm_cond_move_operator): Check against
10960 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10962 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10964 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10965 to top of function.
10967 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10969 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10970 loop in comment with memset.
10972 2017-07-14 Martin Liska <mliska@suse.cz>
10974 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10975 * dwarf2out.c (is_java): Remove the function.
10976 (output_pubname): Remove usage of the function.
10977 (lower_bound_default): Remove usage of DW_LANG_Java.
10978 (gen_compile_unit_die): Likewise.
10979 * gcc.c: Remove compiler defaults for .java and .zip files.
10980 * gimple-expr.c (remove_suffix): Change as there's no longer
10981 extension than 4-letter one.
10982 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10983 (gimplify_save_expr): Likewise.
10984 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10985 as it's possible even for other languages than Java.
10986 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10987 * lto-opts.c (lto_write_options): Remove reference to Java.
10988 * opts.c (strip_off_ending): Update file extension handling.
10989 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10990 * tree-eh.c (lower_resx): Likewise.
10991 * tree.c (free_lang_data_in_type): Remove dead code.
10992 (find_decls_types_r): Likewise.
10993 (build_common_builtin_nodes): Remove Java from a comment.
10994 (verify_type): Remove dead code.
10995 * varasm.c (assemble_external): Remove Java from a comment.
10997 2017-07-14 Martin Liska <mliska@suse.cz>
10999 * opts.c (finish_options): Add quotes.
11000 (common_handle_option): Likewise.
11002 2017-07-14 Martin Liska <mliska@suse.cz>
11004 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11005 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11006 Remove N_SO_PASCAL.
11007 * dwarf2out.c (lower_bound_default): Do not handle
11009 (gen_compile_unit_die): Likewise.
11010 * gcc.c: Remove default extension binding for GNU Pascal.
11011 * stmt.c: Remove Pascal language from a comment.
11012 * xcoffout.c: Likewise.
11014 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11017 * diagnostic-show-locus.c (fixit_cmp): New function.
11018 (layout::layout): Sort m_fixit_hints.
11019 (column_range::column_range): Assert that the values are valid.
11020 (struct char_span): New struct.
11021 (correction::overwrite): New method.
11022 (struct source_line): New struct.
11023 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11024 calls in terms of classes source_line and char_span, and
11025 correction::overwrite.
11026 (selftest::test_overlapped_fixit_printing_2): New function.
11027 (selftest::diagnostic_show_locus_c_tests): Call it.
11029 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11031 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11032 early if there is no lhs.
11034 2017-07-13 Martin Liska <mliska@suse.cz>
11036 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11037 (gen_reference_type_die): Likewise.
11038 * stor-layout.c: Remove Pascal-related comment.
11040 2017-07-13 Martin Liska <mliska@suse.cz>
11042 * opts.c (finish_options): Add quotes to error messages.
11043 (parse_sanitizer_options): Likewise.
11045 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11047 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11049 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11051 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11053 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11055 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11057 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11059 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11062 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11063 provides the hardware capability bits, define the macro
11064 __BUILTIN_CPU_SUPPORTS__.
11065 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11066 if GLIBC does not provide the hardware capability bits. Add a
11067 gcc_unreachable call if the built-in cpu function is neither
11068 __builtin_cpu_is nor __builtin_cpu_supports.
11069 (rs6000_get_function_versions_dispatcher): Change the warning
11070 that an old GLIBC is used which does not export the capability
11071 bits to be an error.
11072 * doc/extend.texi (target_clones attribute): Document the
11073 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11074 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11075 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11076 the macros defined by GCC if the newer GLIBC is available.
11078 2017-07-12 Jeff Law <law@redhat.com>
11080 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11081 remaining includes slightly.
11082 * config/riscv/riscv-builtins.c: Include profile-count.h.
11084 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11087 * config/avr/avr.c (avr_set_current_function): In diagnostic
11088 messages: Quote keywords and (parts of) identifiers.
11089 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11092 2017-07-12 Carl Love <cel@us.ibm.com>
11094 * config/rs6000/rs6000-c.c: Add support for built-in functions
11095 vector bool char vec_revb (vector bool char);
11096 vector bool short vec_revb (vector short char);
11097 vector bool int vec_revb (vector bool int);
11098 vector bool long long vec_revb (vector bool long long);
11099 * doc/extend.texi: Update the built-in documentation file for the
11100 new built-in functions.
11102 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11104 * config/s390/s390.md: Remove movcc splitter.
11106 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11108 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11109 load/store on condition.
11111 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11114 * config/avr/avr.c (avr_encode_section_info)
11115 [progmem && !TREE_READONLY]: Error if progmem object needs
11118 2017-07-11 Michael Collison <michael.collison@arm.com>
11120 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11123 2017-07-11 Carl Love <cel@us.ibm.com>
11125 * config/rs6000/rs6000-c.c: Add support for builtins
11126 vector unsigned int vec_parity_lsbb (vector signed int);
11127 vector unsigned int vec_parity_lsbb (vector unsigned int);
11128 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11129 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11130 vector unsigned long long vec_parity_lsbb (vector signed long long);
11131 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11132 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11133 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11134 * doc/extend.texi: Update the built-in documentation file for the
11135 new built-in functions.
11137 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11139 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11140 (layout::m_primary_loc): New field.
11141 (layout::layout): Initialize new field. Move location filtering
11142 logic from here to...
11143 (layout::maybe_add_location_range): ...this new method. Add
11144 support for filtering to just the lines already specified by other
11146 (layout::will_show_line_p): New method.
11147 (gcc_rich_location::add_location_if_nearby): New method.
11148 (selftest::test_add_location_if_nearby): New test function.
11149 (selftest::diagnostic_show_locus_c_tests): Call it.
11150 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11153 2017-07-11 Tom de Vries <tom@codesourcery.com>
11155 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11156 (bb_first_real_insn): New function.
11157 (nvptx_single): Add extra initialization of broadcasted condition
11160 2017-07-11 Nathan Sidwell <nathan@acm.org>
11162 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11164 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11166 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11167 Binutils doc as TEXI will mess them up.
11168 * doc/invoke.texi (AVR Options): Same here.
11170 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11172 * config/sparc/sparc.opt (mfix-ut700): New option.
11173 (mfix-gr712rc): Likewise.
11174 (sparc_fix_b2bst): New variable.
11175 * doc/invoke.texi (SPARC options): Document them.
11176 (ARM options): Fix warnings.
11177 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11178 instructions to prevent sequences that can trigger the store-store
11179 errata for certain LEON3FT processors.
11180 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11181 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11182 * config/sparc/sparc.md (fix_b2bst): New attribute.
11183 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11185 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11188 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11190 (*rsqrtsf2_sse): Ditto.
11192 (div<mode>3): Macroize insn from divdf3 and divsf3
11193 using MODEF mode iterator.
11195 2017-07-10 Martin Sebor <msebor@redhat.com>
11197 PR tree-optimization/80397
11198 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11199 instead of testing for equality to INTEGER_TYPE.
11201 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11203 * config.gcc: Remove uclibc from arc target spec.
11205 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11207 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11209 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11212 * lto-wrapper.c (remove_option): New function.
11213 (merge_and_complain): Merge PIC/PIE options more realistically.
11215 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11217 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11221 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11222 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11223 * config.in: Regenerate.
11224 * configure: Regenerate.
11225 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11226 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11227 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11228 (TARGET_GASISR_PROLOGUES): ...target mask.
11229 * common/config/avr/avr-common.c
11230 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11231 Set -mgas-isr-prologues.
11232 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11233 INSERT_PASS_BEFORE for it.
11234 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11235 * config/avr/avr.c (avr_option_override)
11236 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11237 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11238 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11239 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11240 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11242 (make_avr_pass_pre_proep): New function.
11243 (emit_push_sfr) <treg>: Add argument to function and use it
11244 instead of TMP_REG.
11245 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11246 and set machine->gasisr.yes.
11247 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11248 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11249 __gcc_isr.n_pushed to .L__stack_usage.
11250 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11251 (avr_asm_final_postscan_insn): ...this new static function.
11252 * config/avr/avr.h (machine_function)
11253 <is_no_gccisr, use_L__stack_usage>: New fields.
11254 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11255 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11256 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11257 (gasisr, *gasisr): New expander and insn.
11258 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11259 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11260 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11262 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11264 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11267 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11269 Move jump-tables out of .text again.
11272 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11273 (ASM_OUTPUT_ADDR_VEC): New function.
11274 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11275 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11276 INSN_ADDRESSes as asm comment.
11277 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11278 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11279 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11280 * config/avr/avr.md (*tablejump): Adjust comment.
11281 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11282 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11284 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11285 (avr_output_addr_vec): New proto.
11286 (avr_log_t) <insn_addresses>: New field.
11288 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11291 * config/i386/i386.c (ix86_function_arg_advance): Set
11292 outgoing_args_on_stack to true if there are outgoing arguments
11294 (ix86_function_arg): Likewise.
11295 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11296 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11297 * config/i386/i386.h (machine_function): Add
11298 outgoing_args_on_stack.
11300 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11302 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11303 supporting pthreds.
11304 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11306 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11308 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11309 (REAL_H): Remove $(MACHMODE_H).
11310 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11312 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11313 $(MACHMODE_H) and double-int.h.
11314 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11316 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11319 2017-07-07 Andrew Pinski <apinski@cavium.com>
11321 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11322 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11324 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11326 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11327 Add warning if GCC was not configured to link against a GLIBC that
11328 exports the hardware capability bits.
11329 (make_resolver_func): Make resolver function private and not a
11330 COMDAT function. Create the name with clone_function_name instead
11331 of make_unique_name.
11334 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11335 correct operand in doing the split.
11337 2017-07-07 Carl Love <cel@us.ibm.com>
11339 * config/rs6000/rs6000-c: Add support for built-in function
11340 vector unsigned short vec_pack_to_short_fp32 (vector float,
11342 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11343 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11344 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11345 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11346 (convert_4f32_8i16): Add define_expand.
11347 * doc/extend.texi: Update the built-in documentation file for the
11348 new built-in function.
11350 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11352 * config/sparc/m8.md: New file.
11353 * config/sparc/sparc.md: Include m8.md.
11355 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11357 * config/sparc/sparc.opt: New option -mvis4b.
11358 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11359 (sparc_option_override): Handle VIS4B.
11360 (enum sparc_builtins): Define
11361 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11362 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11363 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11364 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11365 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11366 (check_constant_argument): New function.
11367 (sparc_vis_init_builtins): Define builtins
11368 __builtin_vis_dictunpack{8,16,32},
11369 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11370 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11371 __builtin_vis_fpcmpde{8,16,32}shl and
11372 __builtin_vis_fpcmpur{8,16,32}shl.
11373 (sparc_expand_builtin): Check that the constant operands to
11374 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11375 constant and in range.
11376 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11378 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11379 (SPARC_IMM5_P): Likewise.
11380 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11381 (enabled): Handle vis4b.
11382 (UNSPEC_DICTUNPACK): New unspec.
11383 (UNSPEC_FPCMPSHL): Likewise.
11384 (UNSPEC_FPUCMPSHL): Likewise.
11385 (UNSPEC_FPCMPDESHL): Likewise.
11386 (UNSPEC_FPCMPURSHL): Likewise.
11387 (cpu_feature): New CPU feature `vis4b'.
11388 (dictunpack{8,16,32}): New insns.
11389 (FPCSMODE): New mode iterator.
11390 (fpcscond): New code iterator.
11391 (fpcsucond): Likewise.
11392 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11393 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11394 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11395 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11396 * config/sparc/constraints.md: Define constraints `q' for unsigned
11397 2-bit integer constants and `t' for unsigned 5-bit integer
11399 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11401 (imm5_operand_dictunpack16): Likewise.
11402 (imm5_operand_dictunpack32): Likewise.
11403 (imm2_operand): Likewise.
11404 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11405 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11406 ditunpack* and fpcmp*shl builtins.
11408 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11410 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11411 * config.in: Add HAVE_AS_SPARC6 define.
11412 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11414 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11416 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11417 (CPP_CPU_SPEC): Handle m8.
11418 (ASM_CPU_SPEC): Likewise.
11419 * config/sparc/sparc-opts.h (enum processor_type): Add
11421 * config/sparc/sparc.c (m8_costs): New struct.
11422 (sparc_option_override): Handle TARGET_CPU_m8.
11423 (sparc32_initialize_trampoline): Likewise.
11424 (sparc64_initialize_trampoline): Likewise.
11425 (sparc_issue_rate): Likewise.
11426 (sparc_register_move_cost): Likewise.
11427 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11428 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11429 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11430 (CPP_CPU_SPEC): Handle M8.
11431 (ASM_CPU_SPEC): Likewise.
11432 (AS_M8_FLAG): Define.
11433 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11434 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11435 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11437 * configure: Regenerate.
11438 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11441 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11443 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11445 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11446 ("*movdi_insn_sp32"): Do not set v3pipe.
11447 ("*movsi_insn"): Likewise.
11448 ("*movdi_insn_sp64"): Likewise.
11449 ("*movsf_insn"): Likewise.
11450 ("*movdf_insn_sp32"): Likewise.
11451 ("*movdf_insn_sp64"): Likewise.
11452 ("*zero_extendsidi2_insn_sp64"): Likewise.
11453 ("*sign_extendsidi2_insn"): Likewise.
11454 ("*mov<VM32:mode>_insn"): Likewise.
11455 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11456 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11457 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11458 ("<vlop:code><VL:mode>3"): Likewise.
11459 ("*not_<vlop:code><VL:mode>3"): Likewise.
11460 ("*nand<VL:mode>_vis"): Likewise.
11461 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11462 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11463 ("one_cmpl<VL:mode>2"): Likewise.
11464 ("faligndata<VM64:mode>_vis"): Likewise.
11465 ("alignaddrsi_vis"): Likewise.
11466 ("alignaddrdi_vis"): Likweise.
11467 ("alignaddrlsi_vis"): Likewise.
11468 ("alignaddrldi_vis"): Likewise.
11469 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11470 ("bmaskdi_vis"): Likewise.
11471 ("bmasksi_vis"): Likewise.
11472 ("bshuffle<VM64:mode>_vis"): Likewise.
11473 ("cmask8<P:mode>_vis"): Likewise.
11474 ("cmask16<P:mode>_vis"): Likewise.
11475 ("cmask32<P:mode>_vis"): Likewise.
11476 ("pdistn<P:mode>_vis"): Likewise.
11477 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11479 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11481 * config/sparc/sparc.md ("subtype"): New insn attribute.
11482 ("*wrgsr_sp64"): Set insn subtype.
11483 ("*rdgsr_sp64"): Likewise.
11484 ("alignaddrsi_vis"): Likewise.
11485 ("alignaddrdi_vis"): Likewise.
11486 ("alignaddrlsi_vis"): Likewise.
11487 ("alignaddrldi_vis"): Likewise.
11488 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11489 ("fexpand_vis"): Likewise.
11490 ("fpmerge_vis"): Likewise.
11491 ("faligndata<VM64:mode>_vis"): Likewise.
11492 ("bshuffle<VM64:mode>_vis"): Likewise.
11493 ("cmask8<P:mode>_vis"): Likewise.
11494 ("cmask16<P:mode>_vis"): Likewise.
11495 ("cmask32<P:mode>_vis"): Likewise.
11496 ("fchksm16_vis"): Likewise.
11497 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11498 ("fmean16_vis"): Likewise.
11499 ("fp<plusminus_insn>64_vis"): Likewise.
11500 ("<plusminus_insn>v8qi3"): Likewise.
11501 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11502 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11503 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11504 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11505 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11506 ("*movqi_insn"): Likewise.
11507 ("*movhi_insn"): Likewise.
11508 ("*movsi_insn"): Likewise.
11509 ("movsi_pic_gotdata_op"): Likewise.
11510 ("*movdi_insn_sp32"): Likewise.
11511 ("*movdi_insn_sp64"): Likewise.
11512 ("movdi_pic_gotdata_op"): Likewise.
11513 ("*movsf_insn"): Likewise.
11514 ("*movdf_insn_sp32"): Likewise.
11515 ("*movdf_insn_sp64"): Likewise.
11516 ("*zero_extendhisi2_insn"): Likewise.
11517 ("*zero_extendqihi2_insn"): Likewise.
11518 ("*zero_extendqisi2_insn"): Likewise.
11519 ("*zero_extendqidi2_insn"): Likewise.
11520 ("*zero_extendhidi2_insn"): Likewise.
11521 ("*zero_extendsidi2_insn_sp64"): Likewise.
11522 ("ldfsr"): Likewise.
11523 ("prefetch_64"): Likewise.
11524 ("prefetch_32"): Likewise.
11525 ("tie_ld32"): Likewise.
11526 ("tie_ld64"): Likewise.
11527 ("*tldo_ldub_sp32"): Likewise.
11528 ("*tldo_ldub1_sp32"): Likewise.
11529 ("*tldo_ldub2_sp32"): Likewise.
11530 ("*tldo_ldub_sp64"): Likewise.
11531 ("*tldo_ldub1_sp64"): Likewise.
11532 ("*tldo_ldub2_sp64"): Likewise.
11533 ("*tldo_ldub3_sp64"): Likewise.
11534 ("*tldo_lduh_sp32"): Likewise.
11535 ("*tldo_lduh1_sp32"): Likewise.
11536 ("*tldo_lduh_sp64"): Likewise.
11537 ("*tldo_lduh1_sp64"): Likewise.
11538 ("*tldo_lduh2_sp64"): Likewise.
11539 ("*tldo_lduw_sp32"): Likewise.
11540 ("*tldo_lduw_sp64"): Likewise.
11541 ("*tldo_lduw1_sp64"): Likewise.
11542 ("*tldo_ldx_sp64"): Likewise.
11543 ("*mov<VM32:mode>_insn"): Likewise.
11544 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11545 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11547 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11549 * config/sparc/sparc.md ("type"): New insn type viscmp.
11550 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11552 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11553 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11554 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11555 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11557 ("n7_vis_logical_11cycle"): Likewise.
11558 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11559 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11560 * config/sparc/niagara.md ("niag_vis"): Likewise.
11561 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11562 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11564 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11566 * config/sparc/sparc.md: New instruction type `bmask'.
11567 (bmaskdi_vis): Use the `bmask' type.
11568 (bmasksi_vis): Likewise.
11569 * config/sparc/ultra3.md (us3_array): Likewise.
11570 * config/sparc/niagara7.md (n7_array): Likewise.
11571 * config/sparc/niagara4.md (n4_array): Likewise.
11572 * config/sparc/niagara2.md (niag2_vis): Likewise.
11573 (niag3_vis): Likewise.
11574 * config/sparc/niagara.md (niag_vis): Likewise.
11576 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11578 * ipa-comdats.c: Remove optimize check from gate.
11579 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11580 for functions not optimized.
11581 (ipa_fn_summary_read): Skip optimize check.
11582 (ipa_fn_summary_write): Likewise.
11583 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11585 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11587 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11588 (check_callers): Give up on caller which is not optimized.
11589 (inline_small_functions): Likewise.
11590 (ipa_inline): Do not give up when not optimizing.
11591 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11592 away unoptimizes cdtors.
11593 (whole_program_function_and_variable_visibility): Do
11594 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11595 * ipa.c (process_references): Do not check optimize.
11596 (symbol_table::remove_unreachable_nodes): Update optimize check.
11597 (set_writeonly_bit): Update optimize check.
11598 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11599 (pass_ipa_single_use::gate): Remove.
11601 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11603 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11604 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11605 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11606 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11607 permute_load, permute_store, adjust_extract, adjust_splat,
11608 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11609 replace_swap_with_copy, dump_swap_insn_table,
11610 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11611 recombine_lvx_pattern, recombine_stvx_pattern,
11612 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11613 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11614 to file rs6000-p8swap.c.
11615 * config/rs6000/rs6000-p8swap.c: New file.
11616 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11617 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11618 and rs6000*-*-* targets.
11620 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11622 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11624 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11626 * lto-wrapper.c (merge_and_complain): Do not merge
11627 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11628 fsigned_zeros, ftrapping_math, fwrapv.
11629 (append_compiler_options): Do not track these options.
11630 (append_linker_options): Likewie
11632 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11634 * cgraphunit.c (cgraph_node::finalize_function): When
11635 !flag_toplevel_reorde set no_reorder flag.
11636 (varpool_node::finalize_decl): Likewise.
11637 (symbol_table::compile): Drop no toplevel reorder path.
11639 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11641 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11642 edges; zero probability is not better than uninitialized.
11644 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11646 * asan.h (asan_sanitize_allocas_p): Declare.
11647 * asan.c (asan_sanitize_allocas_p): New function.
11648 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11649 (handle_builtin_alloca): Likewise.
11650 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11651 if !asan_sanitize_allocas_p.
11652 * params.def (asan-instrument-allocas): Add new option.
11653 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11654 * opts.c (common_handle_option): Disable allocas sanitization for
11657 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11659 * asan.c: Include gimple-fold.h.
11660 (get_last_alloca_addr): New function.
11661 (handle_builtin_stackrestore): Likewise.
11662 (handle_builtin_alloca): Likewise.
11663 (asan_emit_allocas_unpoison): Likewise.
11664 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11665 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11666 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11667 (instrument_builtin_call): Pass gimple iterator to
11668 get_mem_refs_of_builtin_call.
11669 (last_alloca_addr): New global.
11670 * asan.h (asan_emit_allocas_unpoison): Declare.
11671 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11672 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11673 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11674 if function calls alloca.
11675 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11676 * gimple-fold.h (replace_call_with_value): Declare.
11677 * internal-fn.c: Include asan.h.
11678 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11679 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11681 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11683 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11684 (C_SELFTEST_FLAGS): New.
11685 (CPP_SELFTEST_FLAGS): New.
11686 (SELFTEST_DEPS): New, from deps of s-selftest.
11687 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11688 (CPP_SELFTEST_DEPS): New.
11689 (selftest): Add dependency on s-selftest-c++.
11690 (s-selftest): Rename to...
11691 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11692 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11693 than SELFTEST_FLAGS.
11694 (selftest-gdb): Rename to...
11695 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11697 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11698 (selftest-valgrind): Rename to...
11699 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11701 (selftest-valgrind): Reintroduce as an alias for
11702 selftest-c-valgrind.
11703 (s-selftest-c++): New.
11704 (selftest-c++-gdb): New.
11705 (selftest-c++-valgrind): New.
11707 2017-07-06 Olivier Hainque <hainque@adacore.com>
11709 * gcc.c (process_command): When deciding if undefined variables
11710 should be ignored when processing specs, accept "gcc -v" as well.
11712 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11714 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11715 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11717 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11719 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11720 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11721 * config/arm/arm-tables.opt: Regenerate.
11722 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11724 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11726 2017-07-06 Carl Love <cel@us.ibm.com>
11728 * ChangeLog: Clean up from mid air collision
11730 2017-07-06 Carl Love <cel@us.ibm.com>
11732 * config/rs6000/rs6000-c.c: Add support for built-in functions
11733 vector signed int vec_subc (vector signed int, vector signed int);
11734 vector signed __int128 vec_subc (vector signed __int128,
11735 vector signed __int128);
11736 vector unsigned __int128 vec_subc (vector unsigned __int128,
11737 vector unsigned __int128);
11738 vector signed int vec_sube (vector signed int, vector signed int,
11739 vector signed int);
11740 vector unsigned int vec_sube (vector unsigned int,
11741 vector unsigned int,
11742 vector unsigned int);
11743 vector signed __int128 vec_sube (vector signed __int128,
11744 vector signed __int128,
11745 vector signed__int128);
11746 vector unsigned __int128 vec_sube (vector unsigned __int128,
11747 vector unsigned __int128,
11748 vector unsigned __int128);
11749 vector signed int vec_subec (vector signed int, vector signed int,
11750 vector signed int);
11751 vector unsigned int vec_subec (vector unsigned int,
11752 vector unsigned int,
11753 vector unsigned int);
11754 vector signed __int128 vec_subec (vector signed __int128,
11755 vector signed __int128,
11756 vector signed__int128);
11757 vector unsigned __int128 vec_subec (vector unsigned __int128,
11758 vector unsigned __int128,
11759 vector unsigned __int128);
11760 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11761 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11762 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11763 BU_ALTIVEC_OVERLOAD_X definitions.
11764 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11765 * doc/extend.texi: Update the built-in documentation file for the new
11766 built-in functions.
11768 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11771 * diagnostic-show-locus.c (layout::layout): Use start and finish
11772 spelling location for the start and finish of each range.
11773 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11774 Add unused aspect param.
11775 * input.c (expand_location_1): Add "aspect" param, and use it
11776 to access the correct part of the location.
11777 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11779 (expand_location_to_spelling_point): Likewise.
11780 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11781 param, and pass it to expand_location_1.
11783 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11785 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11786 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11787 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11788 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11789 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11790 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11791 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11792 _mm_maskz_getmant_ss): New intrinsics.
11793 (__builtin_ia32_getexpss128_mask): Changed to ...
11794 __builtin_ia32_getexpss128_round ... this.
11795 (__builtin_ia32_getexpsd128_mask): Changed to ...
11796 __builtin_ia32_getexpsd128_round ... this.
11797 * config/i386/i386-builtin-types.def
11798 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11799 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11800 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11801 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11802 __builtin_ia32_getmantss_mask_round): New builtins.
11803 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11804 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11805 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11806 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11807 * config/i386/sse.md
11808 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11809 avx512f_sgetexp<mode><mask_scalar_name>
11810 <round_saeonly_scalar_name> ... this.
11811 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11812 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11813 vgetexp<ssescalarmodesuffix>
11814 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11815 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11816 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11817 avx512f_vgetmant<mode><mask_scalar_name>
11818 <round_saeonly_scalar_name> ... this.
11819 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11820 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11821 vgetmant<ssescalarmodesuffix>
11822 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11823 %0<mask_scalar_operand4>, %1, %2
11824 <round_saeonly_scalar_mask_op4>, %3} ... this.
11825 * config/i386/subst.md (mask_scalar_operand4,
11826 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11827 round_saeonly_scalar_nimm_predicate): New subst attributes.
11829 2017-07-06 Julia Koval <julia.koval@intel.com>
11831 * config/i386/i386.c (ix86_erase_embedded_rounding):
11832 Remove code for old rounding pattern.
11834 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11836 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11838 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11840 * doc/sourcebuild.texi (Test Directives, Variants of
11841 dg-require-support): Add documentation for dg-require-stack-check.
11843 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11845 * config/i386/subst.md (mask_scalar, round_scalar,
11846 round_saeonly_scalar): New meta-templates.
11847 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11848 round_scalar_mask_operand3, round_scalar_mask_op3,
11849 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11850 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11851 round_saeonly_scalar_constraint,
11852 round_saeonly_scalar_prefix): New subst attribute.
11853 * config/i386/sse.md
11854 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11855 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11856 <round_scalar_name> ... this.
11857 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11858 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11859 <round_scalar_name> ... this.
11860 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11861 <sse>_vm<code><mode>3<mask_scalar_name>
11862 <round_saeonly_scalar_name> ... this.
11863 (v<plusminus_mnemonic><ssescalarmodesuffix>
11864 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11865 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11866 v<plusminus_mnemonic><ssescalarmodesuffix>
11867 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11868 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11869 (v<multdiv_mnemonic><ssescalarmodesuffix>
11870 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11871 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11872 v<multdiv_mnemonic><ssescalarmodesuffix>
11873 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11874 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11875 (v<maxmin_float><ssescalarmodesuffix>
11876 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11877 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11878 v<maxmin_float><ssescalarmodesuffix>
11879 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11880 %0<mask_scalar_operand3>, %1, %<iptr>2
11881 <round_saeonly_scalar_mask_op3>} ... this.
11883 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11885 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11886 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11888 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11889 Alan Hayward <alan.hayward@arm.com>
11890 David Sherwood <david.sherwood@arm.com>
11892 * combine.c (simplify_if_then_else): Remove "enum" before
11894 * compare-elim.c (can_eliminate_compare): Likewise.
11895 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11897 (aarch64_lookup_simd_builtin_type): Likewise.
11898 (aarch64_simd_builtin_type): Likewise.
11899 (aarch64_init_simd_builtin_types): Likewise.
11900 (aarch64_simd_expand_args): Likewise.
11901 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11903 (aarch64_reverse_mask): Likewise.
11904 (aarch64_simd_emit_reg_reg_move): Likewise.
11905 (aarch64_gen_adjusted_ldpstp): Likewise.
11906 (aarch64_ccmp_mode_to_code): Likewise.
11907 (aarch64_operands_ok_for_ldpstp): Likewise.
11908 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11909 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11911 (aarch64_min_divisions_for_recip_mul): Likewise.
11912 (aarch64_reassociation_width): Likewise.
11913 (aarch64_get_condition_code_1): Likewise.
11914 (aarch64_simd_emit_reg_reg_move): Likewise.
11915 (aarch64_simd_attr_length_rglist): Likewise.
11916 (aarch64_reverse_mask): Likewise.
11917 (aarch64_operands_ok_for_ldpstp): Likewise.
11918 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11919 (aarch64_gen_adjusted_ldpstp): Likewise.
11920 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11922 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11923 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11924 (arm_lookup_simd_builtin_type): Likewise.
11925 (arm_simd_builtin_type): Likewise.
11926 (arm_init_simd_builtin_types): Likewise.
11927 (arm_expand_builtin_args): Likewise.
11928 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11929 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11930 (ft32_setup_incoming_varargs): Likewise.
11931 (ft32_function_arg): Likewise.
11932 (ft32_function_arg_advance): Likewise.
11933 (ft32_pass_by_reference): Likewise.
11934 (ft32_arg_partial_bytes): Likewise.
11935 (ft32_valid_pointer_mode): Likewise.
11936 (ft32_addr_space_pointer_mode): Likewise.
11937 (ft32_addr_space_legitimate_address_p): Likewise.
11938 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11940 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11941 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11942 (iamcu_alignment): Likewise.
11943 (canonicalize_vector_int_perm): Likewise.
11944 (ix86_noce_conversion_profitable_p): Likewise.
11945 (ix86_mpx_bound_mode): Likewise.
11946 (ix86_operands_ok_for_move_multiple): Likewise.
11947 * config/microblaze/microblaze-protos.h
11948 (microblaze_expand_conditional_branch_reg): Likewise.
11949 * config/microblaze/microblaze.c
11950 (microblaze_expand_conditional_branch_reg): Likewise.
11951 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11953 (rs6000_reassociation_width): Likewise.
11954 (rs6000_invalid_binary_op): Likewise.
11955 (fusion_p9_p): Likewise.
11956 (emit_fusion_p9_load): Likewise.
11957 (emit_fusion_p9_store): Likewise.
11958 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11960 (riscv_hard_regno_mode_ok_p): Likewise.
11961 (riscv_address_insns): Likewise.
11962 (riscv_split_symbol): Likewise.
11963 (riscv_legitimize_move): Likewise.
11964 (riscv_function_value): Likewise.
11965 (riscv_hard_regno_nregs): Likewise.
11966 (riscv_expand_builtin): Likewise.
11967 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11968 (riscv_build_integer): Likewise.
11969 (riscv_split_integer): Likewise.
11970 (riscv_legitimate_constant_p): Likewise.
11971 (riscv_cannot_force_const_mem): Likewise.
11972 (riscv_regno_mode_ok_for_base_p): Likewise.
11973 (riscv_valid_base_register_p): Likewise.
11974 (riscv_valid_offset_p): Likewise.
11975 (riscv_valid_lo_sum_p): Likewise.
11976 (riscv_classify_address): Likewise.
11977 (riscv_legitimate_address_p): Likewise.
11978 (riscv_address_insns): Likewise.
11979 (riscv_load_store_insns): Likewise.
11980 (riscv_force_binary): Likewise.
11981 (riscv_split_symbol): Likewise.
11982 (riscv_force_address): Likewise.
11983 (riscv_legitimize_address): Likewise.
11984 (riscv_move_integer): Likewise.
11985 (riscv_legitimize_const_move): Likewise.
11986 (riscv_legitimize_move): Likewise.
11987 (riscv_address_cost): Likewise.
11988 (riscv_subword): Likewise.
11989 (riscv_output_move): Likewise.
11990 (riscv_canonicalize_int_order_test): Likewise.
11991 (riscv_emit_int_order_test): Likewise.
11992 (riscv_function_arg_boundary): Likewise.
11993 (riscv_pass_mode_in_fpr_p): Likewise.
11994 (riscv_pass_fpr_single): Likewise.
11995 (riscv_pass_fpr_pair): Likewise.
11996 (riscv_get_arg_info): Likewise.
11997 (riscv_function_arg): Likewise.
11998 (riscv_function_arg_advance): Likewise.
11999 (riscv_arg_partial_bytes): Likewise.
12000 (riscv_function_value): Likewise.
12001 (riscv_pass_by_reference): Likewise.
12002 (riscv_setup_incoming_varargs): Likewise.
12003 (riscv_print_operand): Likewise.
12004 (riscv_elf_select_rtx_section): Likewise.
12005 (riscv_save_restore_reg): Likewise.
12006 (riscv_for_each_saved_reg): Likewise.
12007 (riscv_register_move_cost): Likewise.
12008 (riscv_hard_regno_mode_ok_p): Likewise.
12009 (riscv_hard_regno_nregs): Likewise.
12010 (riscv_class_max_nregs): Likewise.
12011 (riscv_memory_move_cost): Likewise.
12012 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12013 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12014 (rl78_addr_space_address_mode): Likewise.
12015 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12017 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12018 (rs6000_reassociation_width): Likewise.
12019 (rs6000_invalid_binary_op): Likewise.
12020 (fusion_p9_p): Likewise.
12021 (emit_fusion_p9_load): Likewise.
12022 (emit_fusion_p9_store): Likewise.
12023 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12024 (ok_for_simple_move_operands): Likewise.
12025 (ok_for_simple_move_strict_operands): Likewise.
12026 (ok_for_simple_arith_logic_operands): Likewise.
12027 (visium_legitimize_reload_address): Likewise.
12028 (visium_select_cc_mode): Likewise.
12029 (output_cbranch): Likewise.
12030 (visium_split_double_move): Likewise.
12031 (visium_expand_copysign): Likewise.
12032 (visium_expand_int_cstore): Likewise.
12033 (visium_expand_fp_cstore): Likewise.
12034 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12035 (visium_function_arg): Likewise.
12036 (visium_function_arg_advance): Likewise.
12037 (visium_libcall_value): Likewise.
12038 (visium_setup_incoming_varargs): Likewise.
12039 (visium_legitimate_constant_p): Likewise.
12040 (visium_legitimate_address_p): Likewise.
12041 (visium_legitimize_address): Likewise.
12042 (visium_secondary_reload): Likewise.
12043 (visium_register_move_cost): Likewise.
12044 (visium_memory_move_cost): Likewise.
12045 (prepare_move_operands): Likewise.
12046 (ok_for_simple_move_operands): Likewise.
12047 (ok_for_simple_move_strict_operands): Likewise.
12048 (ok_for_simple_arith_logic_operands): Likewise.
12049 (visium_function_value_1): Likewise.
12050 (rtx_ok_for_offset_p): Likewise.
12051 (visium_legitimize_reload_address): Likewise.
12052 (visium_split_double_move): Likewise.
12053 (visium_expand_copysign): Likewise.
12054 (visium_expand_int_cstore): Likewise.
12055 (visium_expand_fp_cstore): Likewise.
12056 (visium_split_cstore): Likewise.
12057 (visium_select_cc_mode): Likewise.
12058 (visium_split_cbranch): Likewise.
12059 (output_cbranch): Likewise.
12060 (visium_print_operand_address): Likewise.
12061 * expmed.c (flip_storage_order): Likewise.
12062 * expmed.h (emit_cstore): Likewise.
12063 (flip_storage_order): Likewise.
12064 * genrecog.c (validate_pattern): Likewise.
12065 * hsa-gen.c (gen_hsa_addr): Likewise.
12066 * internal-fn.c (expand_arith_overflow): Likewise.
12067 * ira-color.c (allocno_copy_cost_saving): Likewise.
12068 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12069 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12070 (process_invariant_for_inheritance): Likewise.
12071 * lra-eliminations.c (move_plus_up): Likewise.
12072 * omp-low.c (lower_oacc_reductions): Likewise.
12073 * simplify-rtx.c (simplify_subreg): Likewise.
12074 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12075 (TARGET_CHKP_BOUND_MODE): Likewise..
12076 * targhooks.c (default_chkp_bound_mode): Likewise.
12077 (default_setup_incoming_vararg_bounds): Likewise.
12078 * targhooks.h (default_chkp_bound_mode): Likewise.
12079 (default_setup_incoming_vararg_bounds): Likewise.
12080 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12081 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12082 (have_whole_vector_shift): Likewise.
12083 * tree-vect-stmts.c (vectorizable_load): Likewise.
12084 * doc/tm.texi: Regenerate.
12086 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12088 Graceful degrade if Binutils PR21472 is not available.
12091 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12092 .rodata in flash test fails.
12093 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12094 * confgure: Regenerate.
12095 * config.in: Regenerate.
12096 * config/avr/avr.c (avr_asm_named_section)
12097 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12098 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12099 (avr_asm_init_sections): Same.
12101 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12103 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12104 (fma<VH:mode>4_intrinsic): Likewise.
12105 (*fmsub<VCVTF:mode>4): Likewise.
12106 (*fmsub<VH:mode>4_intrinsic): Likewise.
12108 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12111 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12112 Don't depend on "optimize > 0".
12113 (out_movhi_r_mr, out_movqi_mr_r): Same.
12114 (out_movhi_mr_r, out_movqi_r_mr): Same.
12115 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12116 io_address_operand on "optimize > 0".
12118 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12120 * tree-loop-distribution.c: Add general explanantion on the pass.
12121 (generate_loops_for_partition): Mark distributed loop.
12122 (pg_add_dependence_edges): New parameter. Handle alias data
12123 dependence specially and record it in the parameter if asked.
12124 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12125 (init_partition_graph_vertices, add_partition_graph_edge): New.
12126 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12127 (free_partition_graph_vdata, build_partition_graph): New.
12128 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12129 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12130 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12131 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12132 (version_for_distribution_p, finalize_partitions): New.
12133 (distribute_loop): Handle alias data dependence specially. Factor
12134 out loop fusion code as functions and call these functions.
12136 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12138 * tree-loop-distribution.c (classify_partition): New parameter and
12139 better handle reduction statement.
12140 (rdg_build_partitions): Revise comment.
12141 (distribute_loop): Compute statements in all partitions and pass it
12142 to classify_partition.
12144 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12146 * tree-loop-distribution.c (enum partition_type): New.
12147 (struct partition): New field type.
12148 (partition_merge_into): Add parameter. Update partition type.
12149 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12150 (build_rdg_partition_for_vertex): Compute partition type.
12151 (rdg_build_partitions): Dump partition type.
12152 (distribute_loop): Update calls to partition_merge_into.
12154 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12156 * tree-loop-distribution.c (struct ddr_hasher): New.
12157 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12159 (classify_partition): Call get_data_dependence.
12160 (pg_add_dependence_edges): Ditto.
12161 (distribute_loop): Release data dependence hash table.
12163 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12165 * tree-loop-distribution.c (ref_base_address): Delete.
12166 (similar_memory_accesses): Rename ...
12167 (share_memory_accesses): ... to this. Check if partitions access
12168 the same memory reference.
12169 (distribute_loop): Call share_memory_accesses.
12171 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12173 * tree-loop-distribution.c (struct partition): New field recording
12174 its data reference.
12175 (partition_alloc, partition_free): Init and release data refs.
12176 (partition_merge_into): Merge data refs.
12177 (build_rdg_partition_for_vertex): Collect data refs for partition.
12178 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12180 (distribute_loop): Remve data refs from vertice data of partition
12183 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12185 * tree-loop-distribution.c (params.h): Include header file.
12186 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12187 (datarefs_vec): New global var.
12188 (create_rdg_vertices): Use datarefs_vec directly.
12189 (free_rdg): Don't free data references.
12190 (build_rdg): Update use. Don't free data references.
12191 (distribute_loop): Compute global variable for data references.
12192 Bail out if there are too many data references.
12194 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12196 * tree-loop-distribution.c (loop_nest): New global var.
12197 (build_rdg): Use loop directly, rather than loop nest.
12198 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12200 (distribute_loop): Compute global variable loop nest. Update use.
12202 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12204 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12205 (partition_merge_into): New parameter. Dump reason for fusion.
12206 (distribute_loop): Update use of partition_merge_into.
12208 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12210 * tree-loop-distribution.c (bb_top_order_index): New.
12211 (bb_top_order_index_size, bb_top_order_cmp): New.
12212 (stmts_from_loop): Use topological order.
12213 (pass_loop_distribution::execute): Compute and release topological
12214 order for basic blocks.
12216 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12218 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12221 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12223 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12224 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12225 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12226 * internal-fn.def (LOOP_DIST_ALIAS): New.
12227 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12228 (fold_loop_internal_call): ... this.
12229 (vect_loop_dist_alias_call): New function.
12230 (set_uid_loop_bbs): Call fold_loop_internal_call.
12231 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12234 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12237 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12238 Require dead FLAGS_REG at the beginning of a peephole.
12240 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12243 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12244 arguments in the call to __builtin_ia32_sbb_u32.
12245 (_subborrow_u64): Swap _X and _Y arguments in the call to
12246 __builtin_ia32_sbb_u64.
12248 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12251 * tree-vrp.c (compare_assert_loc): Turn into a function template
12252 with stable template parameter. Only test if a->e is NULL,
12253 !a->e == !b->e has been verified already. Use e == NULL or
12254 e != NULL instead of e or ! e tests. If stable is true, don't use
12255 iterative_hash_expr, on the other side allow a or b or both NULL
12256 and sort the NULLs last.
12257 (process_assert_insertions): Sort using compare_assert_loc<false>
12258 instead of compare_assert_loc, later sort using
12259 compare_assert_loc<true> before calling process_assert_insertions_for
12260 in a loop. Use break instead of continue once seen NULL pointer.
12262 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12264 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12265 Cortex-R7 and Cortex-R8 processors.
12267 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12269 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12270 uninitialized while src is not.
12272 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12274 * common/config/arm/arm-common.c: Adjust include path for
12276 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12277 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12278 (arm-cpu-data.h): Likewise.
12279 (arm-cpu-cdata.h): Likewise.
12280 * config/arm/arm-cpu.h: Delete.
12281 * config/arm/arm-cpu-cdata.h: Delete.
12282 * config/arm/arm-cpu-data.h: Delete.
12284 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12286 * config/arm/arm-cpus.in (cortex-a55): New.
12287 (cortex-a75): Likewise.
12288 (cortex-a75.cortex-a55): Likewise.
12289 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12291 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12292 * config/arm/arm-cpu-cdata.h: Regenerate.
12293 * config/arm/arm-cpu-data.h: Regenerate.
12294 * config/arm/arm-cpu.h: Regenerate.
12295 * config/arm/arm-tables.opt: Regenerate.
12296 * config/arm/arm-tune.md: Regenerate.
12298 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12300 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12302 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12304 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12307 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12309 PR tree-optimization/81292
12310 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12311 full_string_p, also call adjust_related_strinfos if the adjustment
12312 is simple, otherwise invalidate related strinfos.
12314 2017-07-04 Martin Liska <mliska@suse.cz>
12317 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12318 newly created variable as DECL_IGNORED_P.
12320 2017-07-04 Martin Liska <mliska@suse.cz>
12323 * ipa-inline.c (inline_small_functions):
12324 Use xstrdup_for_dump.
12326 2017-07-04 Tom de Vries <tom@codesourcery.com>
12328 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12330 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12333 * config/darwin.c (darwin_function_switched_text_sections):
12336 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12338 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12340 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12342 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12344 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12346 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12347 min_profitable_iters, and th as inclusive lower bounds.
12348 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12349 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12350 for min_profitable_iters and min_profitable_estimate.
12351 (vect_transform_loop): Treat th as an inclusive lower bound.
12352 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12354 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12357 * config/darwin.c (darwin_function_switched_text_sections):
12358 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12359 in two pieces, and suppress the use of buf.
12361 2017-07-03 Nathan Sidwell <nathan@acm.org>
12363 * hash-table.h (hash_table_mod1): Fix indentation.
12365 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12367 PR middle-end/81290
12368 * predict.c (force_edge_cold): Be more careful about propagation
12370 * profile-count.h (profile_probability::guessed,
12371 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12373 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12375 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12377 * doc/invoke.texi (rcpc architecture extension): Document it.
12379 2017-07-03 Richard Biener <rguenther@suse.de>
12381 PR tree-optimization/60510
12382 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12383 the scalar reduction PHI and use it.
12384 (vectorizable_reduction): Properly guard the single_defuse_cycle
12385 path for non-SLP reduction chains where we cannot use it.
12386 Rework reduc_def/index and vector type deduction. Rework
12387 vector operand gathering during reduction op code-gen.
12388 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12389 chains dissolve the chain and leave it to non-SLP reduction
12392 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12394 * tree-data-ref.h (dr_alignment): Declare.
12395 * tree-data-ref.c (dr_alignment): New function.
12396 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12397 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12399 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12401 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12403 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12404 and base_misalignment fields.
12405 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12406 * tree-data-ref.c: Include builtins.h.
12407 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12408 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12409 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12410 * tree-vect-data-refs.c: Include tree-cfg.h.
12411 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12412 fields instead of calculating an alignment here.
12413 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12414 innermost_loop_behavior fields.
12416 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12418 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12420 (DR_STEP_ALIGNMENT): New macro.
12421 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12422 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12423 (create_data_ref): Print it.
12424 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12425 to tell whether the step preserves vector (mis)alignment.
12426 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12427 Move the check for an integer step and generalise to all INTEGER_CST.
12428 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12429 Print the outer step alignment.
12431 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12433 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12434 with offset_alignment.
12435 (DR_ALIGNED_TO): Delete.
12436 (DR_OFFSET_ALIGNMENT): New macro.
12437 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12438 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12439 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12440 (create_data_ref): Likewise.
12441 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12442 (vect_analyze_data_refs): Likewise.
12443 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12444 creating dummy innermost behavior.
12446 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12448 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12449 with a "innermost_loop_behavior *" and refeence tree.
12450 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12451 (create_data_ref): Update call accordingly.
12452 * tree-predcom.c (find_looparound_phi): Likewise.
12454 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12456 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12457 fields with dr_wrt_vec_loop.
12458 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12459 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12460 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12461 (vect_dr_behavior): New function.
12462 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12463 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12464 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12465 track whether the step preserves the misalignment.
12466 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12467 Use vect_dr_behavior.
12468 (vect_setup_realignment): Update call accordingly.
12469 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12470 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12471 call to vect_create_addr_base_for_vector_ref.
12472 (vect_create_cond_for_align_checks): Likewise.
12473 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12474 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12475 (vect_recog_mask_conversion_pattern): Likewise.
12476 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12477 (new_stmt_vec_info): Remove redundant zeroing.
12479 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12481 * common/config/arm/arm-common.c (arm_be8_option): New function.
12482 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12483 (ISA_ARMv6): Add isa_bit_be8.
12484 * config/arm/arm.h (arm_be8_option): Add prototype.
12485 (BE8_SPEC_FUNCTION): New define.
12486 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12487 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12488 (mlittle-endian): Similarly.
12489 (mbe8, mbe32): New options.
12490 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12491 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12493 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12495 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12497 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12499 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12500 (cleanup_tree_cfg_bb): Use it.
12501 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12503 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12505 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12508 * loop-doloop.c (add_test): Update profile.
12510 2017-07-03 Martin Liska <mliska@suse.cz>
12513 * sanopt.c (rewrite_usage_of_param): New function.
12514 (sanitize_rewrite_addressable_params): Likewise.
12515 (pass_sanopt::execute): Call rewrite_usage_of_param.
12517 2017-07-03 Richard Biener <rguenther@suse.de>
12519 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12520 back to using VIEW_CONVERT_EXPR.
12522 2017-07-03 Martin Liska <mliska@suse.cz>
12525 * doc/extend.texi: Document when a resolver function is
12526 generated for target_clones.
12528 2017-07-03 Martin Liska <mliska@suse.cz>
12530 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12531 and shadow memory of auto variables which are subject of
12532 use-after-scope sanitization.
12533 (asan_expand_mark_ifn): Add do set only when is_poison.
12535 2016-07-03 Richard Biener <rguenther@suse.de>
12537 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12539 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12540 (vectorizable_reduction): Perform reduction PHI creation when
12541 visiting a reduction PHI and adjust and simplify code generation
12542 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12543 (vect_transform_loop): Visit reduction PHIs.
12544 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12545 defs into the SLP tree.
12546 (vect_build_slp_tree): Reduction defs terminate the recursion.
12547 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12549 (vect_get_vec_defs_for_stmt_copy): Export.
12550 (vect_get_vec_defs): Likewise.
12551 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12553 (vect_get_vec_defs_for_stmt_copy): Declare.
12554 (vect_get_vec_defs): Likewise.
12556 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12558 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12559 parameter with a "loop" parameter and use it instead of the
12560 loop containing DR_STMT. Don't check simple_iv when doing
12561 BB analysis. Describe the two analysis modes in the comment.
12563 2017-07-03 Tom de Vries <tom@codesourcery.com>
12565 PR tree-optimization/69468
12566 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12567 (find_same_succ_bb): Handle ignore_edge_flags.
12569 2017-07-03 Tom de Vries <tom@codesourcery.com>
12571 PR tree-optimization/81192
12572 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12574 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12576 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12578 2017-07-03 Tom de Vries <tom@codesourcery.com>
12580 PR tree-optimization/81192
12581 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12582 BB_SAME_SUCC (bb) == NULL.
12584 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12586 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12589 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12591 * dumpfile.c: Include profile-count.h
12592 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12594 (insert_cond_bb): Update profile.
12595 * tree-cfg.h (insert_cond_bb): Update prototype.
12596 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12597 * tree-dump.c: Do not include tree-cfg.
12599 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12601 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12603 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12605 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12608 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12610 * tree-complex.c (expand_complex_div_wide): update profile.
12612 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12613 Alan Hayward <alan.hayward@arm.com>
12614 David Sherwood <david.sherwood@arm.com>
12616 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12617 (CORETYPES_H): New define.
12618 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12619 (insn-modes-inline.h, s-modes-inline-h): New rules.
12620 (generated_files): Add insn-modes-inline.h.
12621 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12622 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12623 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12624 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12625 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12626 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12627 (build/gencodes.o, build/genconditions.o): Likewise.
12628 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12629 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12630 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12631 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12632 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12633 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12634 * coretypes.h: Include everything up to real.h for generators.
12635 Include insn-modes.h first. Include wide-int-print.h after
12636 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12637 * machmode.h: Don't include insn-modes.h here.
12638 * function-tests.c: Remove includes of signop.h, machmode.h,
12639 double-int.h and wide-int.h.
12641 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12643 * optc-save-gen.awk: Likewise.
12644 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12645 * godump.c: Remove include of wide-int-print.h.
12646 * pretty-print.h: Likewise.
12647 * wide-int-print.cc: Likewise.
12648 * wide-int.cc: Likewise.
12649 * hash-map-tests.c: Remove include of signop.h.
12650 * hash-set-tests.c: Likewise.
12651 * rtl-tests.c: Likewise.
12652 * mkconfig.sh: Remove include of machmode.h.
12653 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12655 (emit_insn_modes_inline_h): ...this new function. Emit the code
12656 into an insn-modes-inline.h header file, adding appropriate
12657 include guards and end comments.
12658 (emit_insn_modes_c_header): Remove include of machmode.h.
12659 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12661 (main): Handle -i flag and call emit_insn_modes_inline_h when
12664 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12666 * tree-ssa-strlen.c (strinfo): Rename the length field to
12667 nonzero_chars. Add a full_string_p field.
12668 (compare_nonzero_chars, zero_length_string_p): New functions.
12669 (get_addr_stridx): Add an offset_out parameter.
12670 Use compare_nonzero_chars.
12671 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12672 (new_strinfo): Update after above changes to strinfo.
12673 (set_endptr_and_length): Set full_string_p.
12674 (get_string_length): Update after above changes to strinfo.
12675 (unshare_strinfo): Update call to new_strinfo.
12676 (maybe_invalidate): Likewise.
12677 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12678 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12679 as a uhwi instead of an shwi. Update after above changes to
12680 strinfo and new_strinfo.
12681 (zero_length_string): Assert that chainsi contains full strings.
12682 Use zero_length_string_p. Update call to new_strinfo.
12683 (adjust_related_strinfos): Update after above changes to strinfo.
12684 Copy full_string_p from origsi.
12685 (adjust_last_stmt): Use zero_length_string_p.
12686 (handle_builtin_strlen): Update after above changes to strinfo and
12687 new_strinfo. Install the lhs as the string length if the previous
12688 entry didn't describe a full string.
12689 (handle_builtin_strchr): Update after above changes to strinfo
12691 (handle_builtin_strcpy): Likewise.
12692 (handle_builtin_strcat): Likewise.
12693 (handle_builtin_malloc): Likewise.
12694 (handle_pointer_plus): Likewise.
12695 (handle_builtin_memcpy): Likewise. Track nonzero characters
12696 that aren't necessarily followed by a nul terminator.
12697 (handle_char_store): Likewise.
12699 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12701 PR tree-optimization/80769
12702 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12703 for malloc and calloc. Document the new invariant that all related
12704 strinfos have delayed lengths or none do.
12705 (verify_related_strinfos): Move earlier in file.
12706 (set_endptr_and_length): New function, split out from...
12707 (get_string_length): ...here. Also set the lengths of related
12709 (zero_length_string): Assert that chainsi has known (rather than
12711 (adjust_related_strinfos): Likewise.
12713 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12715 PR tree-optimization/81136
12716 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12717 assert that two references with the same misalignment have the same
12718 compile-time misalignment if those compile-time misalignments
12721 2017-07-01 Andi Kleen <ak@linux.intel.com>
12723 * print-tree.c (print_node): Print all attributes.
12725 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12727 * cfg.c (scale_bbs_frequencies): New function.
12728 * cfg.h (scale_bbs_frequencies): Declare it.
12729 * cfgloopanal.c (single_likely_exit): Cleanup.
12730 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12732 (scale_loop_profile): Likewise.
12733 (loop_version): Likewise.
12734 (create_empty_loop_on_edge): Update.
12735 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12736 scale_loop_frequencies, scale_loop_profile, loopify,
12737 loop_version): Update prototypes.
12738 * modulo-sched.c (sms_schedule): Update.
12739 * predict.c (unlikely_executed_edge_p): Also check probability.
12740 (probably_never_executed_edge_p): Fix typo.
12741 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12742 * tree-parloops.c (gen_parallel_loop): Update.
12743 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12744 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12745 * tree-ssa-loop-split.c (split_loop): Update.
12746 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12747 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12748 (vect_loop_versioning): Update.
12749 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12751 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12753 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12755 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12757 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12758 to keep profile consistent.
12760 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12762 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12763 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12764 * profile-count.h (max_safe_multiplier): Make unsigned.
12765 (profile_count::guessed_zero): New.
12767 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12769 * bb-reorder.c (fix_up_crossing_landing_pad,
12770 fix_crossing_conditional_branches): Use make_single_succ_edge
12771 to keep profile consistent.
12773 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12775 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12778 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12781 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12782 the right scopes, make sure cond_jump isn't preserved between multiple
12783 iterations. Search for fallthru edge whenever there are 3+ edges and
12784 use find_fallthru_edge for it.
12786 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12788 Patch by Alexander Monakov <amonakov@ispras.ru>
12789 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12790 probabilities consistently.
12792 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12794 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12795 emit_cmp_and_jump_insns.
12797 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12800 * tree-inline.c (expand_call_inline): Combine profile statuses.
12802 2017-06-30 Andrew Pinski <apinski@cavium.com>
12804 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12805 fold_stmt returned true.
12807 2017-06-30 Nathan Sidwell <nathan@acm.org>
12809 * ggc.h (empty_string): Delete.
12810 * cfgexpand.c (expand_asm_stmt): Use plain "".
12811 * optabs.c (expand_asm_memory_barrier): Likewise.
12812 * stringpool.c (empty_string): Delete.
12813 (digit_vector, digit_string): Delete.
12814 (ggc_alloc_string): Use plain "", don't optimize single digit
12815 strings. Use ggc_alloc_atomic.
12817 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12819 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12820 comparison set and one other set, use the cost of the non-comparison
12823 2017-06-30 Nathan Sidwell <nathan@acm.org>
12825 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12828 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12830 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12831 loops. Remove now unneeded calls to gimple_switch_set_label() that
12832 just set removed labels to NULL_TREE.
12834 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12836 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12837 (set_range_info): ...here. Only call set_range_info_raw if domain
12839 (set_nonzero_bits): Call set_range_info_raw.
12840 * tree-ssanames.h (set_range_info_raw): New.
12842 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12845 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12846 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12847 of nonimmediate_operand and <store_mask_constraint> instead of m
12848 for the input operand. For V8FI iterator, always split if input
12849 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12850 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12851 <store_mask_predicate> instead of register_operand and
12852 <store_mask_constraint> instead of v for the input operand. Make
12853 sure both operands aren't MEMs for if not <mask_applied>.
12855 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12857 * lto-wrapper.c (copy_file) Close both file descriptors before
12860 2017-06-30 Martin Liska <mliska@suse.cz>
12863 * multiple_target.c (create_dispatcher_calls): Make ifunc
12864 also for function that don't have calls or are not referenced.
12866 2017-06-30 Richard Biener <rguenther@suse.de>
12868 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12869 analyze the first scalar stmt. Move vector type computation
12870 for the BB case here from ...
12871 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12872 live operation processing in the SLP case properly.
12874 2017-06-30 Richard Biener <rguenther@suse.de>
12876 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12878 2017-06-30 Martin Liska <mliska@suse.cz>
12881 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12882 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12884 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12886 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12889 2017-06-30 Richard Biener <rguenther@suse.de>
12891 PR tree-optimization/81249
12892 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12893 condition reduction result to original scalar type.
12895 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12897 * profile-count.h (enum profile_quality): Fix typos and whitespace
12900 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12902 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12903 type for branch probabilities.
12905 2017-06-29 Julian Brown <julian@codesourcery.com>
12906 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12908 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12909 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12910 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12911 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12913 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12915 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12916 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12917 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12918 CC usage from generic code to here.
12919 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12920 CC usage into the target macros.
12922 2017-06-29 Maya Rashish <coypu@sdf.org>
12924 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12927 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12929 * arm/arm-builtins.c: Include profile-count.h
12930 * except.c (sjlj_emit_function_enter): Use
12931 profile_probability::unlikely.
12933 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12935 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12936 and tocrel_offset be pointer args rather than implicitly using
12938 (legitimate_constant_pool_address_p, rs6000_emit_move,
12939 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12940 tocrel_offset and use in toc_relative_expr_p call.
12941 (print_operand, print_operand_address): Use static tocrel_base_oac
12942 and tocrel_offset_oac.
12943 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12946 2017-06-29 Maya Rashish <coypu@sdf.org>
12948 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12950 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12952 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12953 objects, take into account only the alignment of 'op0' and 'mode1' if
12956 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12958 * ccmp.c (ccmp_tree_comparison_p): New function.
12959 (ccmp_candidate_p): Update to use above function.
12960 (get_compare_parts): New function.
12961 (expand_ccmp_next): Update to use new functions.
12962 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12964 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12965 take mode as argument.
12966 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12967 * expr.c (expand_expr_real_1): Pass mode as argument.
12969 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12971 * combine.c (combine_instructions): Print insns to dump_file, together
12974 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12976 * asan.c (asan_emit_stack_protection): Update.
12977 (create_cond_insert_point): Update.
12978 * auto-profile.c (afdo_propagate_circuit): Update.
12979 * basic-block.h (struct edge_def): Turn probability to
12980 profile_probability.
12981 (EDGE_FREQUENCY): Update.
12982 * bb-reorder.c (find_traces_1_round): Update.
12983 (better_edge_p): Update.
12984 (sanitize_hot_paths): Update.
12985 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12986 (make_single_succ_edge): Update.
12987 (check_bb_profile): Update.
12988 (dump_edge_info): Update.
12989 (update_bb_profile_for_threading): Update.
12990 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12991 probabilitycount to 0.
12992 * cfgbuild.c (compute_outgoing_frequencies): Update.
12993 * cfgcleanup.c (try_forward_edges): Update.
12994 (outgoing_edges_match): Update.
12995 (try_crossjump_to_edge): Update.
12996 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12997 (expand_gimple_tailcall): Update.
12998 (construct_init_block): Use make_single_succ_edge.
12999 (construct_exit_block): Use make_single_succ_edge.
13000 * cfghooks.c (verify_flow_info): Update.
13001 (redirect_edge_succ_nodup): Update.
13002 (split_edge): Update.
13003 (account_profile_record): Update.
13004 * cfgloopanal.c (single_likely_exit): Update.
13005 * cfgloopmanip.c (scale_loop_profile): Update.
13006 (set_zero_probability): Remove.
13007 (duplicate_loop_to_header_edge): Update.
13008 * cfgloopmanip.h (loop_version): Update prototype.
13009 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13010 (force_nonfallthru_and_redirect): Update.
13011 (update_br_prob_note): Update.
13012 (rtl_verify_edges): Update.
13013 (purge_dead_edges): Update.
13014 (rtl_lv_add_condition_to_bb): Update.
13015 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13016 * cgraphunit.c (init_lowered_empty_function): Update.
13017 (cgraph_node::expand_thunk): Update.
13018 * cilk-common.c: Include profile-count.h
13019 * dojump.c (inv): Remove.
13020 (jumpifnot): Update.
13021 (jumpifnot_1): Update.
13022 (do_jump_1): Update.
13024 (do_jump_by_parts_greater_rtx): Update.
13025 (do_compare_rtx_and_jump): Update.
13026 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13027 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13028 * dwarf2cfi.c: Include profile-count.h
13029 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13030 (sjlj_emit_dispatch_table): Likewise.
13031 * explow.c: Include profile-count.h
13032 * expmed.c (emit_store_flag_force): Update.
13033 (do_cmp_and_jump): Update.
13034 * expr.c (compare_by_pieces_d::generate): Update.
13035 (compare_by_pieces_d::finish_mode): Update.
13036 (emit_block_move_via_loop): Update.
13037 (store_expr_with_bounds): Update.
13038 (store_constructor): Update.
13039 (expand_expr_real_2): Update.
13040 (expand_expr_real_1): Update.
13041 * expr.h (try_casesi, try_tablejump): Update prototypes.
13042 * gimple-pretty-print.c (dump_probability): Update.
13043 (dump_profile): New.
13044 (dump_gimple_label): Update.
13045 (dump_gimple_bb_header): Update.
13046 * graph.c (draw_cfg_node_succ_edges): Update.
13047 * hsa-gen.c (convert_switch_statements): Update.
13048 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13049 (find_if_case_1): Update.
13050 (find_if_case_2): Update.
13051 * internal-fn.c (expand_arith_overflow_result_store): Update.
13052 (expand_addsub_overflow): Update.
13053 (expand_neg_overflow): Update.
13054 (expand_mul_overflow): Update.
13055 (expand_vector_ubsan_overflow): Update.
13056 * ipa-cp.c (good_cloning_opportunity_p): Update.
13057 * ipa-split.c (split_function): Use make_single_succ_edge.
13058 * ipa-utils.c (ipa_merge_profiles): Update.
13059 * loop-doloop.c (add_test): Update.
13060 (doloop_modify): Update.
13061 * loop-unroll.c (compare_and_jump_seq): Update.
13062 (unroll_loop_runtime_iterations): Update.
13063 * lra-constraints.c (lra_inheritance): Update.
13064 * lto-streamer-in.c (input_cfg): Update.
13065 * lto-streamer-out.c (output_cfg): Update.
13066 * mcf.c (adjust_cfg_counts): Update.
13067 * modulo-sched.c (sms_schedule): Update.
13068 * omp-expand.c (expand_omp_for_init_counts): Update.
13069 (extract_omp_for_update_vars): Update.
13070 (expand_omp_ordered_sink): Update.
13071 (expand_omp_for_ordered_loops): Update.
13072 (expand_omp_for_generic): Update.
13073 (expand_omp_for_static_nochunk): Update.
13074 (expand_omp_for_static_chunk): Update.
13075 (expand_cilk_for): Update.
13076 (expand_omp_simd): Update.
13077 (expand_omp_taskloop_for_outer): Update.
13078 (expand_omp_taskloop_for_inner): Update.
13079 * omp-simd-clone.c (simd_clone_adjust): Update.
13080 * optabs.c (expand_doubleword_shift): Update.
13081 (expand_abs): Update.
13082 (emit_cmp_and_jump_insn_1): Update.
13083 (expand_compare_and_swap_loop): Update.
13084 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13085 * predict.c (predictable_edge_p): Update.
13086 (edge_probability_reliable_p): Update.
13087 (set_even_probabilities): Update.
13088 (combine_predictions_for_insn): Update.
13089 (combine_predictions_for_bb): Update.
13090 (propagate_freq): Update.
13091 (estimate_bb_frequencies): Update.
13092 (force_edge_cold): Update.
13093 * profile-count.c (profile_count::dump): Add missing space into dump.
13094 (profile_count::debug): Add newline.
13095 (profile_count::differs_from_p): Explicitly convert to unsigned.
13096 (profile_count::stream_in): Update.
13097 (profile_probability::dump): New member function.
13098 (profile_probability::debug): New member function.
13099 (profile_probability::differs_from_p): New member function.
13100 (profile_probability::differs_lot_from_p): New member function.
13101 (profile_probability::stream_in): New member function.
13102 (profile_probability::stream_out): New member function.
13103 * profile-count.h (profile_count_quality): Rename to ...
13104 (profile_quality): ... this one.
13105 (profile_probability): New.
13106 (profile_count): Update.
13107 * profile.c (compute_branch_probabilities): Update.
13108 * recog.c (peep2_attempt): Update.
13109 * sched-ebb.c (schedule_ebbs): Update.
13110 * sched-rgn.c (find_single_block_region): Update.
13111 (compute_dom_prob_ps): Update.
13112 (schedule_region): Update.
13113 * sel-sched-ir.c (compute_succs_info): Update.
13114 * stmt.c (struct case_node): Update.
13115 (do_jump_if_equal): Update.
13116 (get_outgoing_edge_probs): Update.
13117 (conditional_probability): Update.
13118 (emit_case_dispatch_table): Update.
13119 (expand_case): Update.
13120 (expand_sjlj_dispatch_table): Update.
13121 (emit_case_nodes): Update.
13122 * targhooks.c: Update.
13123 * tracer.c (better_p): Update.
13124 (find_best_successor): Update.
13125 * trans-mem.c (expand_transaction): Update.
13126 * tree-call-cdce.c: Update.
13127 * tree-cfg.c (gimple_split_edge): Upate.
13128 (move_sese_region_to_fn): Upate.
13129 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13130 * tree-eh.c (lower_resx): Upate.
13131 (cleanup_empty_eh_move_lp): Upate.
13132 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13133 * tree-inline.c (copy_edges_for_bb): Update.
13134 (copy_cfg_body): Update.
13135 * tree-parloops.c (gen_parallel_loop): Update.
13136 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13137 (gimple_gen_time_profiler): Update.
13138 * tree-ssa-dce.c (remove_dead_stmt): Update.
13139 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13140 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13141 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13142 (unloop_loops): Update.
13143 (try_peel_loop): Update.
13144 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13145 * tree-ssa-loop-split.c (connect_loops): Update.
13146 (split_loop): Update.
13147 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13148 (hoist_guard): Update.
13149 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13150 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13151 (value_replacement): Update.
13152 * tree-ssa-reassoc.c (branch_fixup): Update.
13153 * tree-ssa-tail-merge.c (replace_block_by): Update.
13154 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13155 (create_edge_and_update_destination_phis): Update.
13156 (compute_path_counts): Update.
13157 (recompute_probabilities): Update.
13158 (update_joiner_offpath_counts): Update.
13159 (freqs_to_counts_path): Update.
13160 (duplicate_thread_path): Update.
13161 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13162 (struct switch_conv_info): Update.
13163 (gen_inbound_check): Update.
13164 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13165 (vect_do_peeling): Update.
13166 (vect_loop_versioning): Update.
13167 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13168 (optimize_mask_stores): Update.
13169 * ubsan.c (ubsan_expand_null_ifn): Update.
13170 * value-prof.c (gimple_divmod_fixed_value): Update.
13171 (gimple_divmod_fixed_value_transform): Update.
13172 (gimple_mod_pow2): Update.
13173 (gimple_mod_pow2_value_transform): Update.
13174 (gimple_mod_subtract): Update.
13175 (gimple_mod_subtract_transform): Update.
13176 (gimple_ic): Update.
13177 (gimple_stringop_fixed_value): Update.
13178 (gimple_stringops_transform): Update.
13179 * value-prof.h: Update.
13181 2017-06-29 Carl Love <cel@us.ibm.com>
13183 * config/rs6000/rs6000-c.c: Add support for built-in functions
13184 vector signed int vec_signed (vector float);
13185 vector signed long long vec_signed (vector double);
13186 vector signed int vec_signed2 (vector double, vector double);
13187 vector signed int vec_signede (vector double);
13188 vector signed int vec_signedo (vector double);
13189 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13190 instruction generator.
13191 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13192 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13193 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13195 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13196 vunsignede_v2df): Add define_expands.
13197 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13198 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13199 VEC_UNSIGNEDO): Add definitions.
13200 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13201 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13202 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13203 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13204 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13205 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13206 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13207 vec_unsignede, vec_unsignedo): Add builtin defines.
13208 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13210 * doc/extend.texi: Update the built-in documentation file for the
13211 new built-in functions.
13213 2017-06-29 Richard Biener <rguenther@suse.de>
13215 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13216 reduction chains to LOOP_VINFO_REDUCTIONS.
13217 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13218 SLP reductions after processing reduction chains.
13220 2017-06-29 Nathan Sidwell <nathan@acm.org>
13222 * builtins.c (fold_builtin_FUNCTION): Use
13223 lang_hooks.decl_printable_name.
13225 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13227 PR middle-end/81194
13228 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13229 with only one label.
13230 * stmt.c (expand_case): Assert NCASES is greater than one.
13232 2017-06-29 Richard Biener <rguenther@suse.de>
13234 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13236 (group_case_labels): Likewise.
13237 (find_taken_edge): Push sanity checking on val to workers...
13238 (find_taken_edge_cond_expr): ... here
13239 (find_taken_edge_switch_expr): ... and here, handle cases
13240 with just a default label.
13241 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13242 (group_case_labels): Likewise.
13243 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13244 group_case_labels does anything cleanup the CFG again.
13246 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13248 PR tree-optimization/81196
13249 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13250 exit condition comparing two IVs.
13252 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13254 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13255 profile to the dummy entry at the end of the list of architectures.
13256 * config/arm/arm-cpu-cdata.h: Regenerated.
13258 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13259 Michael Collison <michael.collison@arm.com>
13262 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13264 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13265 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13266 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13267 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13268 of shift when the shift amount is masked with constant equal to
13269 the size of the mode.
13270 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13273 2017-06-29 Martin Liska <mliska@suse.cz>
13275 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13277 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13280 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13283 (edge_growth_cache_entry::edge_growth_cache_entry): New
13285 (reset_edge_growth_cache): Update to use constructor.
13287 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13289 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13290 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13291 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13293 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13295 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13296 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13298 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13300 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13301 (*-linux-uclibc*): Add t-uclibc tmake_file.
13302 * config/t-musl: New.
13303 * config/t-uclibc: New.
13305 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13307 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13309 (gen_comm_data): Emit architectural setting of arch_prof.
13310 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13312 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13313 (armv8-m.base, armv8-m.main): Likewise.
13314 * arm-protos.h (arm_build_target): Add profile field.
13315 (arch_option): Likewise.
13316 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13318 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13319 arm_active_target.profile.
13321 2017-06-28 Richard Biener <rguenther@suse.de>
13323 PR middle-end/81227
13324 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13325 TYPE_OVERFLOW_WRAPS.
13326 * match.pd (negate_expr_p): Likewise.
13327 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13328 fold_build2, not fold_binary.
13330 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13332 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13333 Convert memory address to Pmode.
13334 (aarch64_print_operand): Assert MEM operands are always Pmode.
13336 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13339 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13340 Remove redundant if.
13341 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13342 * config/arm/aarch-common-protos.h
13343 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13344 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13346 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13349 * multiple_target.c (create_dispatcher_calls): Set the default
13350 clone to be static, not public.
13352 2017-06-28 Richard Biener <rguenther@suse.de>
13354 * tree-vect-loop.c (vectorizable_reduction): Move special
13355 cond reduction IV var creation ...
13356 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13357 parameter. Use STMT_VINFO_VECTYPE.
13358 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13361 2017-06-28 Martin Liska <mliska@suse.cz>
13364 * ipa-visibility.c (non_local_p): Handle visibility.
13366 2017-06-28 Martin Liska <mliska@suse.cz>
13369 * common.opt: Add IntegerRange to various options.
13370 * opt-functions.awk (integer_range_info): New function.
13371 * optc-gen.awk: Add integer_range_info to cl_options struct.
13372 * opts-common.c (decode_cmdline_option): Handle
13373 CL_ERR_INT_RANGE_ARG.
13374 (cmdline_handle_error): Likewise.
13375 * opts.c (print_filtered_help): Show valid interval in
13376 when --help is provided.
13377 * opts.h (struct cl_option): Add range_min and range_max fields.
13378 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13380 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13382 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13383 (x * C EQ/NE y * C): New transformation.
13385 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13387 * genmultilib (combination_space): Accept '+' in option names.
13389 2017-06-28 Martin Liska <mliska@suse.cz>
13392 * asan.c (instrument_derefs): Bail out inner references
13393 that are hard register variables.
13395 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13398 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13399 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13401 2017-06-28 Richard Biener <rguenther@suse.de>
13403 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13404 (vect_get_slp_defs): Adjust.
13405 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13407 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13409 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13410 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13411 (vectorizable_reduction): Adjust.
13412 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13414 (vect_get_slp_defs): Likewise.
13415 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13416 (vectorizable_bswap): Adjust.
13417 (vectorizable_call): Likewise.
13418 (vectorizable_conversion): Likewise.
13419 (vectorizable_assignment): Likewise.
13420 (vectorizable_shift): Likewise.
13421 (vectorizable_operation): Likewise.
13422 (vectorizable_store): Likewise.
13423 (vectorizable_condition): Likewise.
13424 (vectorizable_comparison): Likewise.
13426 2017-06-28 Michael Collison <michael.collison@arm.com>
13429 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13431 (arm_gen_movmemqi): Removed unused variable 'i'.
13432 Convert 'for' loop into 'while' loop.
13433 (arm_expand_prologue): Remove last unnecessary set of insn.
13434 (thumb_pop): Remove unused variable 'pushed_words'.
13435 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13437 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13439 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13440 * config/s390/s390-protos.h: Add prototype of
13441 s390_rel_address_ok_p.
13442 * config/s390/s390.c (s390_got_symbol): New function.
13443 (s390_rel_address_ok_p): New function.
13444 (legitimize_pic_address): Use s390_rel_address_ok_p.
13445 (s390_load_got): Use s390_got_symbol.
13446 (s390_option_override): Issue error if
13447 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13448 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13450 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13452 2017-06-27 Andrew Pinski <apinski@cavium.com>
13454 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13455 (X * copysign (1.0, X)): New pattern.
13456 (X * copysign (1.0, -X)): New pattern.
13457 (copysign (-1.0, CST)): New pattern.
13459 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13461 * genmultilib (combination_space): Remove variable.
13462 Validate reuse rules against regular expression for any sequence
13463 of multilib options in any order.
13465 2017-06-27 Michael Collison <michael.collison@arm.com>
13467 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13468 call aarch64_split_simd_combine.
13469 * (aarch64_combine_internal<mode>): Delete pattern.
13470 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13471 Allow register and subreg operands.
13473 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13475 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13476 specific need, just fallback on defaults.
13477 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13479 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13480 Olivier Hainque <hainque@adacore.com>
13482 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13484 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13485 targets. Pick a default if no particular attempt applied.
13486 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13489 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13491 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13492 (x86_64-wrs-vxworks7): Likewise.
13494 2017-06-27 Marek Polacek <polacek@redhat.com>
13497 * ubsan.c (instrument_null): Check get_base_address's result for null.
13499 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13501 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13503 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13505 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13506 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13507 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13508 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13509 New function types.
13510 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13511 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13512 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13513 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13514 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13515 BUILT_IN_FEUPDATEENV): New builtins.
13516 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13517 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13518 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13519 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13521 (builtin_structptr_types): Adjust size.
13522 * tree.c (builtin_structptr_types): Add four entries.
13524 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13525 Olivier Hainque <hainque@adacore.com>
13527 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13528 (TLS_SYM): New local macro, forcing reference to __tls__ on
13529 link command lines for VxWorks 7 RTPs, triggering initialization
13531 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13532 OS features TLS support, true for RTPs on VxWorks 7.
13533 * config/vxworks.c (vxworks_override_options): Setup emutls
13536 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13538 * predict.c (test_prediction_value_range): Use -1U instead of -1
13539 to avoid narrowing conversion warning.
13540 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13541 to avoid narrowing conversion warning.
13542 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13544 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13546 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13548 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13549 64bit configurations.
13550 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13551 (SIZE_TYPE): Likewise.
13552 * config/vxworks.c (vxworks_emutls_var_fields): Use
13553 long_unsigned_type_node instead of unsigned_type_node as the offset
13554 field type, which is "pointer" mode in emutls.c.
13556 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13559 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13561 PR middle-end/81207
13562 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13563 gimple_vuse copying separately from gimple_vdef copying.
13565 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13567 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13570 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13571 Olivier Hainque <hainque@adacore.com>
13573 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13574 for all vxworks7 targets.
13575 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13576 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13577 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13578 variations for VX6/VX7 and 32/64bits later on in ...
13579 (VXWORKS_LIB_SPEC): Leverage new macros.
13580 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13581 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13583 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13585 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13586 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13588 2017-06-26 Carl Love <cel@us.ibm.com>
13590 * config/rs6000/rs6000-c.c: Add support for built-in functions
13591 vector bool char vec_reve (vector bool char);
13592 vector signed char vec_reve (vector signed char);
13593 vector unsigned char vec_reve (vector unsigned char);
13594 vector bool int vec_reve (vector bool int);
13595 vector signed int vec_reve (vector signed int);
13596 vector unsigned int vec_reve (vector unsigned int);
13597 vector bool long long vec_reve (vector bool long long);
13598 vector signed long long vec_reve (vector signed long long);
13599 vector unsigned long long vec_reve (vector unsigned long long);
13600 vector bool short vec_reve (vector bool short);
13601 vector signed short vec_reve (vector signed short);
13602 vector double vec_reve (vector double);
13603 vector float vec_reve (vector float);
13604 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13605 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13606 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13607 (altivec_vreve): New pattern.
13608 * config/rs6000/altivec.h (vec_reve): New define.
13609 * doc/extend.texi (vec_rev): Update the built-in documentation file
13610 for the new built-in functions.
13612 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13614 PR tree-optimization/71815
13615 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13617 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13619 (slsr_process_phi): Likewise.
13620 (replace_uncond_cands_and_profitable_phis): Don't replace a
13621 multiply candidate with a stride of 1 (copy or cast).
13622 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13624 (lowest_cost_path): Likewise.
13625 (total_savings): Likewise.
13627 2017-06-26 Richard Biener <rguenther@suse.de>
13630 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13631 Use def_builtin_pure for all gather builtins.
13633 2017-06-26 Richard Biener <rguenther@suse.de>
13635 PR tree-optimization/81203
13636 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13637 non-dominating BBs.
13639 2017-06-26 Marek Polacek <polacek@redhat.com>
13642 * doc/invoke.texi: Document -Wmultistatement-macros.
13644 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13646 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13647 arm_neon_ok_no_float_abi effective target.
13649 2017-06-26 Richard Biener <rguenther@suse.de>
13651 PR tree-optimization/80928
13652 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13653 (copy_bbs): Set BB_DUPLICATED flag early.
13654 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13656 (execute_on_shrinking_pred): Likewise.
13657 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13658 BB_DUPLICATED blocks.
13659 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13660 iterate over all PHIs considering removal of *gsi.
13662 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13664 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13667 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13669 * config/rs6000/rs6000-string.c: (expand_block_clear,
13670 do_load_for_compare, select_block_compare_mode,
13671 compute_current_alignment, expand_block_compare,
13672 expand_strncmp_align_check, expand_strn_compare,
13673 expand_block_move, rs6000_output_load_multiple)
13674 Move functions related to string/block move/compare
13675 to a separate file.
13676 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13677 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13678 for this function which is now used in two files.
13679 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13680 * config.gcc: Add rs6000-string.o to extra_objs for
13681 targets powerpc*-*-* and rs6000*-*-*.
13683 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13686 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13687 32-bit, since indexed is not valid for DImode.
13688 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13689 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13690 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13691 (define_peephole2 for Altivec d-form store): Likewise.
13694 * multiple_target.c (create_dispatcher_calls): Only create the
13695 dispatcher call if the function is the default clone of a
13696 versioned function.
13698 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13700 PR middle-end/80902
13701 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13702 a call, force the call to not be a tail call.
13704 2017-06-23 Jeff Law <law@redhat.com>
13706 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13709 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13711 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13712 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13713 handling for early expansion of vector shifts (sl,sr,sra,rl).
13714 (builtin_function_type): Add vector shift right instructions
13715 to the unsigned argument list.
13717 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13719 rtl-optimizatoin/79286
13720 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13721 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13722 trap. PIC register plus a const unspec without offset can never trap.
13724 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13726 * tree.h (builtin_structptr_type): New type.
13727 (builtin_structptr_types): Declare new array.
13728 * tree.c (builtin_structptr_types): New array.
13729 (free_lang_data, build_common_tree_nodes): Use it.
13731 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13734 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13737 2017-06-22 Matt Turner <mattst88@gmail.com>
13739 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13740 Lake models to skylake case. Assume skylake for unknown
13741 models with clflushopt.
13743 2017-06-22 Jeff Law <law@redhat.com>
13745 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13746 frame sizes that do not satisfy aarch64_uimm12_shift.
13748 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13750 * profile-count.h (apply_probability,
13751 apply_scale, probability_in): Fix checks for zero.
13753 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13755 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13756 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13758 2016-06-22 Richard Biener <rguenther@suse.de>
13760 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13761 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13762 REDUC_MAX_EXPR support.
13763 (vectorizable_reduction): Likewise.
13764 (vect_create_epilog_for_reduction): Likewise.
13766 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13768 * match.pd (A / (1 << B) -> A >> B): New.
13769 * generic-match-head.c: Include optabs-tree.h.
13770 * gimple-match-head.c: Likewise.
13771 * optabs-tree.h (target_supports_op_p): New.
13772 * optabs-tree.c (target_supports_op_p): New.
13774 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13776 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13777 $gcc_cv_ld --help output.
13778 (gcc_cv_ld_demangle): Likewise.
13779 (gcc_cv_ld_eh_frame_hdr): Likewise.
13780 (gcc_cv_ld_pie): Likewise.
13781 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13782 (gcc_cv_ld_buildid): Likewise.
13783 (gcc_cv_ld_sysroot): Likewise.
13784 (ld_bndplt_support): Likewise.
13785 (ld_pushpopstate_support): Likewise.
13786 * configure: Regenerate.
13787 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13789 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13792 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13793 operands indexes to avoid gap between operands and match_dups.
13795 2017-06-21 Andrew Pinski <apinski@cavium.com>
13797 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13798 Increment Arith_shift and Arith_shift_reg by 1.
13799 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13801 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13802 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13803 (aarch64_strip_extend): Add new argument and test for it.
13804 (aarch64_cheap_mult_shift_p): New function.
13805 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13806 add a cost if it is true.
13807 Update calls to aarch64_strip_extend.
13808 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13810 2017-06-21 Andrew Pinski <apinski@cavium.com>
13812 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13814 (thunderxt88): Likewise.
13815 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13816 (thunderx_prefetch_tune): New variable.
13817 (thunderx2t99_prefetch_tune): Update for the correct values.
13818 (thunderxt88_tunings): New variable.
13819 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13820 generic_prefetch_tune.
13821 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13823 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13825 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13826 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13827 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13828 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13829 (aarch64_atomic_cas<mode>, GPI): Likewise.
13831 2017-06-21 Martin Liska <mliska@suse.cz>
13833 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13834 statements on cold and hot labels.
13835 * predict.c (tree_estimate_probability_bb): Remove the
13836 prediction from this place.
13838 2017-06-21 Martin Liska <mliska@suse.cz>
13840 PR tree-optimization/79489
13841 * gimplify.c (maybe_add_early_return_predict_stmt): New
13843 (gimplify_return_expr): Call the function.
13844 * predict.c (tree_estimate_probability_bb): Remove handling
13846 * predict.def: Update comment about early return predictor.
13847 * gimple-predict.h (is_gimple_predict): New function.
13848 * predict.def: Change default value of early return to 66.
13849 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13851 * passes.def: Put pass_strip_predict_hints to the beginning of
13854 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13856 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13857 FUNCTION_DECL declarations.
13858 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13860 (dwaf2out_decl): Likewise.
13861 * godump.c (go_early_global_decl): Skip call to the real debug hook
13862 for FUNCTION_DECL declarations.
13863 * passes.c (rest_of_decl_compilation): Skip call to the
13864 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13865 -fdump-go-spec is passed.
13867 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13869 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13870 Reorder for compactness.
13871 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13872 (def_builtin_pure, def_builtin_pure2): New functions.
13873 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13875 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13877 * match.pd (nop_convert): New predicate.
13878 ((A +- CST1) +- CST2): Allow some NOP conversions.
13880 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13883 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13884 with ctors/dtors if GOVD_SHARED is set.
13886 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13888 * config/aarch64/aarch64.md (movti_aarch64):
13889 Emit mov rather than orr.
13890 (movtf_aarch64): Likewise.
13891 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13892 Emit mov rather than orr.
13894 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13896 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13897 Swap alternatives, make integer dup more expensive.
13899 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13901 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13902 Return true for non-tls symbols.
13904 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13906 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13907 (cortex-a75): Likewise.
13908 (cortex-a75.cortex-a55): Likewise.
13909 * config/aarch64/aarch64-tune.md: Regenerate.
13910 * doc/invoke.texi (-mtune): Document new values for -mtune.
13912 2017-06-21 Tom de Vries <tom@codesourcery.com>
13914 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13915 stack_size feature.
13916 (Effective-Target Keywords, Other attributes): Suggest using
13917 dg-add-options stack_size feature to get stack limit in stack_size
13918 effective target documentation.
13920 2017-06-21 Julian Brown <julian@codesourcery.com>
13921 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13923 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13924 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13925 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13927 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13928 attribute type list for neon_multiply.
13929 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13930 attribute type list for neon_multiply.
13931 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13932 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13933 attribute type list for neon_multiply.
13934 * config/arm/types.md (crypto_pmull): Add.
13935 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13936 attribute type list.
13938 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13940 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13943 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13945 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13946 to make sure not to dereference a NULL cost_classes_ptr pointer.
13948 2017-06-20 Carl Love <cel@us.ibm.com>
13950 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13951 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13952 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13953 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13954 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13955 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13956 VMULOSW): New enum "unspec" values.
13957 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13958 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13959 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13960 altivec_vmulosw): New patterns.
13961 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13962 VMULOSW): Add definitions.
13964 2017-06-20 Julia Koval <julia.koval@intel.com>
13966 * config/i386/i386.c: Fix rounding expand for new pattern.
13967 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13969 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13971 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13972 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13974 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13976 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13979 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13981 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13982 architecture, then by implementer ID.
13983 * config/aarch64/aarch64-tune.md: Regenerate.
13985 2017-06-20 Richard Biener <rguenther@suse.de>
13987 PR middle-end/81097
13988 * fold-const.c (split_tree): Fold to type before negating.
13990 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13992 * diagnostic-show-locus.c
13993 (selftest::test_fixit_deletion_affecting_newline): New function.
13994 (selftest::diagnostic_show_locus_c_tests): Call it.
13996 2017-06-20 Andreas Schwab <schwab@suse.de>
13999 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14002 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14004 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14005 __ARM_FEATURE_COPROC according to support.
14007 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14009 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14010 Rewritten to avoid overflow for > 32-bit pointers.
14013 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14014 by removing enum keyword.
14015 (ubsan_type_descriptor): Likewise. Formatting fix.
14018 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14019 splitter): Require TARGET_SSE2 in the condition.
14021 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14024 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14025 for doing vector set of SFmode on ISA 3.0.
14026 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14027 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14029 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14030 SFmode value into a V4SF variable that was extracted from another
14031 V4SF variable without converting the element to double precision
14032 and back to single precision vector format.
14033 (vsx_insert_extract_v4sf_p9_2): Likewise.
14035 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14037 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14038 in UWHI to avoid undefined overflow.
14041 * ubsan.h (enum ubsan_encode_value_phase): New.
14042 (ubsan_encode_value): Change second argument to
14043 enum ubsan_encode_value_phase with default value of
14044 UBSAN_ENCODE_VALUE_GENERIC.
14045 * ubsan.c (ubsan_encode_value): Change second argument to
14046 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14047 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14048 create_tmp_var_raw instead of create_tmp_var and use a
14050 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14051 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14052 ubsan_encode_value callers.
14055 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14056 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14057 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14059 2017-06-19 Richard Biener <rguenther@suse.de>
14061 PR middle-end/81118
14062 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14063 estimates if we changed anything.
14065 2017-06-19 Richard Biener <rguenther@suse.de>
14067 PR tree-optimization/80887
14068 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14069 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14070 simplified lookups, then reset mprts_hook.
14071 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14073 (try_to_simplify): Likewise.
14075 2017-06-19 Martin Liska <mliska@suse.cz>
14078 * gimplify.c (gimplify_switch_expr):
14079 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14081 2017-06-19 Martin Liska <mliska@suse.cz>
14083 * doc/install.texi: Document that PGO runs in 4 stages.
14085 2017-06-19 Martin Liska <mliska@suse.cz>
14088 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14089 to dispatcher function name.
14090 * multiple_target.c (replace_function_decl): New function.
14091 (create_dispatcher_calls): Redirect both edges and references.
14093 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14095 * profile-count.c (profile_count::dump): Dump quality.
14096 (profile_count::differs_from_p): Update for unsigned val.
14097 * profile-count.h (profile_count_quality): New enum.
14098 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14100 2017-06-19 Richard Biener <rguenther@suse.de>
14102 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14103 struct function as arg.
14104 (estimate_numbers_of_iterations): Export overload with loop arg.
14105 (free_numbers_of_iterations_estimates_loop): Use an overload of
14106 free_numbers_of_iterations_estimates instead.
14107 * tree-cfg.c (remove_bb): Adjust.
14108 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14109 * tree-parloops.c (gen_parallel_loop): Likewise.
14110 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14112 (tree_unroll_loops_completely): Likewise.
14113 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14114 Use an overload instead and export.
14115 (estimated_loop_iterations): Adjust.
14116 (max_loop_iterations): Likewise.
14117 (likely_max_loop_iterations): Likewise.
14118 (estimate_numbers_of_iterations): Take struct function as arg
14120 (loop_exits_before_overflow): Adjust.
14121 (free_numbers_of_iterations_estimates_loop): Use an overload.
14122 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14123 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14125 2017-06-19 Richard Biener <rguenther@suse.de>
14128 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14129 RANGE_EXPR conservatively.
14131 2017-06-16 Carl Love <cel@us.ibm.com>
14133 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14134 definitions for vec_float, vec_float2, vec_floato,
14135 vec_floate built-ins.
14136 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14137 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14139 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14140 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14141 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14142 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14143 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14144 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14145 vec_floato): Add builtin defines.
14146 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14147 Update the built-in documentation file for the new built-in
14150 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14152 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14153 (mthumb): Mark as the negative of -marm.
14155 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14157 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14159 (ARM Options, -mtune): Document that this accepts the same
14160 extension options as -mcpu.
14161 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14165 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14166 permitted extensions.
14168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14170 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14171 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14172 (armv8-m.main): Add option +nodsp.
14173 * config/arm/arm-cpu-cdata.h: Regenerated.
14175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14177 * config/arm/t-fuchsia: New file.
14178 * config.gcc (arm*-*-fuchsia*): Use it.
14180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14182 * config/arm/t-symbian: Rewrite for new option infrastructure.
14184 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14186 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14187 (MULTILIB_REQUIRED): Likewise.
14189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14191 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14192 (MULTILIB_RESUE): Likewise.
14193 (MULTILIB_MATCHES): Likewise.
14194 (MULTLIB_REQUIRED): Likewise.
14196 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14198 * config/arm/t-rtems: Rewrite for new option framework.
14200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14202 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14203 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14204 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14205 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14206 * config/arm/t-multilib: ... here.
14207 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14208 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14209 armv7-a and armv8*-a when A-profile libraries have not been built.
14210 * config/arm/t-rmprofile: Rewrite.
14212 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14214 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14215 with a backslash. Remove the backslash after substituting unescaped
14217 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14221 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14222 the driver to pass through the default setting of -mfloat-abi.
14223 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14225 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14226 (all_feat_combs): New rule.
14227 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14229 * config/arm/t-aprofile: Rewrite.
14231 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14233 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14234 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14235 fpu is not specified by the user/command-line.
14236 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14237 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14238 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14239 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14240 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14241 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14245 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14246 * config/arm/t-arm-elf: Rewritten.
14248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14250 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14251 have some floating-point instructions.
14252 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14253 (TARGET_MAYBE_HARD_FLOAT): New macro.
14254 * config/arm/arm-builtins.c (arm_init_builtins): Use
14255 TARGET_MAYBE_HARD_FLOAT.
14256 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14260 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14261 (configargs.h): Include it.
14262 (arm_print_hint_for_fpu_option): New function.
14263 (arm_parse_fpu_option): New function.
14264 (candidate_extension): New class.
14265 (arm_canon_for_multilib): New function.
14266 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14267 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14268 (ARCH_CANONICAL_SPECS): New macro.
14269 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14271 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14273 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14274 are set after handling multilib fragments. Set target_cpu_default2
14277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14279 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14281 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14282 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14284 (check_arch): Likewise.
14285 * config/arm/arm.c (arm_configure_build_target): Handle
14286 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14287 options in the default.
14289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14291 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14292 when an option is an alias of another.
14293 * config/arm/parsecpu.awk (optalias): New parser token.
14294 (gen_comm_data): Mark non-alias options as such. Emit entries
14295 for extension aliases.
14296 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14297 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14298 (armv6kz, armv6zk, armv6t2): Likewise.
14299 (armv7): Make vfpv3-d16 an alias.
14300 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14302 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14303 Sort in canonical order.
14304 (armv8-a): Sort in canonical order.
14305 (armv8.1-a, armv8.2-a): Likewise.
14306 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14308 (cortex-a9): Sort in canonical order.
14309 * config/arm/arm.c (selftests.h): Include it.
14310 (arm_test_cpu_arch_data): New function.
14311 (arm_run_self_tests): New function.
14312 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14313 (targetm): Move declaration to the end of the file.
14314 * arm-cpu-cdata.h: Regenerated.
14316 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14318 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14319 call to target_mode_check describing the type of option passed.
14320 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14321 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14322 arm_parse_cpu_option_name to match parameters against list of
14324 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14325 arm_arch_core_flags data structure.
14326 * config/arm/arm-cpu_cdata.h: Regenerated.
14328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14330 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14332 (arm_print_hint_for_cpu_option): Likewise.
14333 (arm_print_hint_for_arch_option): Likewise.
14334 (arm_parse_cpu_option_name): Likewise.
14335 (arm_parse_arch_option_name): Likewise.
14336 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14337 of entries in the all_fpus list.
14338 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14339 (arm_parse_cpu_option_name): Declare.
14340 (arm_parse_arch_option_name): Declare.
14341 (arm_parse_option_features): Declare.
14342 (arm_intialize_isa): Declare.
14343 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14345 (gen_comm_data): ... here. Make definitions non-static.
14346 * config/arm/arm-cpu-data.h: Regenerated.
14347 * config/arm/arm-cpu-cdata.h: Regenerated.
14349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14351 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14352 (cpu_arch_extension): New structure.
14353 (cpu_arch_option, arch_option, cpu_option): New structures.
14354 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14355 architecture types.
14356 (gen_data): Generate new format data tables.
14357 * config/arm/arm.c (cpu_tune): New structure.
14358 (cpu_option, processors): Delete.
14359 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14360 (arm_print_hint_for_cpu_option): ... this and ...
14361 (arm_print_hint_for_arch_option): ... this.
14362 (arm_parse_arch_cpu_name): Delete. Replace with ...
14363 (arm_parse_cpu_option_name): ... this and ...
14364 (arm_parse_arch_option_name): ... this.
14365 (arm_unrecognized_feature): Change type of target parameter to
14367 (arm_parse_arch_cpu_features): Delete. Replace with ...
14368 (arm_parse_option_features): ... this.
14369 (arm_configure_build_target): Rework to use new configuration data
14371 (arm_print_tune_info): Rework for new configuration data tables.
14372 * config/arm/arm-cpu-data.h: Regenerated.
14373 * config/arm/arm-cpu.h: Regenerated.
14375 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14377 * Makefile.in (OBJS): Move sbitmap.o from here ...
14378 (OBJS-libcommon): ... to here.
14380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14382 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14383 (ISA_ALL_CRYPTO): New macro.
14384 (ISA_ALL_SIMD): New macro
14385 (ISA_ALL_FP): New macro.
14386 * config/arm/arm.c (fpu_bitlist): Update initializer.
14387 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14389 (arm9e): Add fpu. Add option for nofp
14390 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14391 (arm926ej-s, arm1026ej-s): Likewise.
14392 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14393 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14394 neon-fp16, neon-vfpv4, nofp and nosimd.
14395 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14396 (cortex-a8): Add fpu. Add option for nofp.
14397 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14398 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14399 (cortex-r4f): Add fpu.
14400 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14401 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14403 (cortex-r8): Likewise.
14404 (cortex-m4): Add fpu. Add option for nofp.
14405 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14406 (cortex-a17.cortex-a7): Likewise.
14407 (cortex-a32): Add fpu. Add options for crypto and nofp.
14408 (cortex-a35, cortex-a53): Likewise.
14409 (cortex-a57): Add fpu. Add option for crypto.
14410 (cortex-a72, cortex-a73): Likewise.
14411 (exynos-m1): Likewise.
14412 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14413 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14414 (cortex-m33): Add fpu. Add option for nofp.
14415 * config/arm/arm-cpu-cdata.h: Regenerated
14416 * config/arm/arm-cpu-data.h: Regenerated.
14418 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14420 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14421 (armv5te, armv5tej): Likewise.
14422 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14423 (armv7): Add options fp and vfpv3-d16.
14424 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14425 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14427 (armv7ve): Likewise.
14428 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14429 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14430 (armv8-a): Add nocrypto option.
14431 (armv8.1-a, armv8.2-a): Likewise.
14432 (armv8-m.main): add options fp, fp.dp and nofp.
14434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14436 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14438 (armv8-a+crc): Delete.
14439 (armv8.1-a): Add options simd, crypto and nofp.
14440 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14441 (armv8.2-a+fp16): Delete.
14442 (armv8-m.main): Add option dsp.
14443 (armv8-m.main+dsp): Delete.
14444 (cortex-a8): Add fpu. Add nofp option.
14445 (cortex-a9): Add fpu. Add nofp and nosimd options.
14446 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14447 link to main cpu and architecture data structures.
14448 (gen_comm_data): Only put isa attributes from the main architecture
14450 (option): New statement for architecture and CPU entries.
14451 * arm.c (struct cpu_option): New structure.
14452 (struct processors): Add entry for options.
14453 (arm_unrecognized_feature): New function.
14454 (arm_parse_arch_cpu_name): Ignore any characters after the first
14456 (arm_parse_arch_cpu_feature): New function.
14457 (arm_configure_build_target): Separate out any CPU and architecture
14458 features and parse separately. Don't error out if -mfpu=auto is
14459 used with only an architecture string.
14460 (arm_print_asm_arch_directives): New function.
14461 (arm_file_start): Call it.
14462 * config/arm/arm-cpu-cdata.h: Regenerated.
14463 * config/arm/arm-cpu-data.h: Likewise.
14464 * config/arm/arm-tables.opt: Likewise.
14466 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14468 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14469 assembler when it is not -mfpu=auto.
14471 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14473 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14474 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14475 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14476 (ASM_CPU_SPEC): Rewrite.
14477 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14478 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14479 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14480 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14481 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14482 copied string is NUL-terminated. Also strip any characters prefixed
14484 (arm_rewrite_selected_arch): New function.
14485 (arm_rewrite_march): New function.
14487 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14489 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14490 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14491 (march, mcpu, mtune): Convert to string-based options.
14492 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14493 (arm_parse_arch_cpu_name): New function.
14494 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14495 identify selected architecture or CPU.
14496 (arm_option_save): New function.
14497 (TARGET_OPTION_SAVE): Redefine.
14498 (arm_option_restore): Restore string options.
14499 (arm_option_print): Print string options.
14501 2017-06-16 Martin Sebor <msebor@redhat.com>
14503 PR tree-optimization/80933
14504 PR tree-optimization/80934
14505 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14506 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14507 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14508 (gimple_fold_builtin): Call them.
14510 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14512 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14513 as unlikely; update profile.
14515 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14517 * predict.c (force_edge_cold): Handle declaring edges impossible
14520 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14522 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14524 (try_unroll_loop_completely): Fix reporting.
14526 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14528 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14530 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14533 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14534 if given a non-constant argument for an intrinsic which requires a
14537 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14539 * profile.c (compare_freqs): New function.
14540 (branch_prob): Sort edge list.
14541 (find_spanning_tree): Assume that the list is priority sorted.
14543 2017-06-16 Richard Biener <rguenther@suse.de>
14545 PR tree-optimization/81090
14546 * passes.def (pass_record_bounds): Remove.
14547 * tree-pass.h (make_pass_record_bounds): Likewise.
14548 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14549 make_pass_record_bounds): Likewise.
14550 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14551 not free niter estimates at the beginning but at the end.
14552 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14554 2017-06-16 Richard Biener <rguenther@suse.de>
14556 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14557 initializer to workaround ICE in host GCC 4.8.
14559 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14561 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14563 (clone_inlined_nodes): Update.
14565 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14567 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14568 prefetch settings, and enable prefetching by default at -O3.
14570 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14572 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14573 Set flag_prefetch_loop_arrays according to tuning data.
14575 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14577 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14578 New tune structure.
14579 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14580 [Unrelated to main purpose of the patch] Place the pointer field last
14581 to enable type checking errors when tune structure are wrongly merged.
14582 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14583 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14584 (thunderx2t99_prefetch_tune): New tune constants.
14585 (tune_params *_tunings): Update all tunings (no functional change).
14586 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14587 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14588 from tunings structures.
14590 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14593 * ubsan.c (instrument_null): Add T argument, use it instead
14594 of computing it based on IS_LHS.
14595 (instrument_object_size): Likewise.
14596 (pass_ubsan::execute): Adjust instrument_null and
14597 instrument_object_size callers to pass gimple_get_lhs or
14598 gimple_assign_rhs1 result to it. Use instrument_null instead of
14599 calling get_base_address and instrument_mem_ref. Handle
14600 aggregate call arguments for object-size sanitization.
14602 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14604 PR tree-optimization/81089
14605 * tree-vrp.c (is_masked_range_test): Validate operands of
14608 2017-06-15 Martin Sebor <msebor@redhat.com>
14611 * dumpfile.c (dump_register): Avoid calling memset to initialize
14612 a class with a default ctor.
14613 * gcc.c (struct compiler): Remove const qualification.
14614 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14615 * hash-table.h: Ditto.
14616 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14618 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14619 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14621 * params.h (struct param_info): Make struct members non-const.
14622 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14623 with default initialization.
14624 * vec.h (vec_copy_construct, vec_default_construct): New helper
14626 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14627 with vec_copy_construct.
14628 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14629 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14630 * doc/invoke.texi (-Wclass-memaccess): Document.
14632 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14634 * emit-rtl.h (is_leaf): Update comment about local
14635 register allocator.
14637 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14640 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14641 for a variable to have a section before checking if the section has a
14643 Set section to.persistent if persistent attribute is set.
14644 Warn if .persistent attribute is used on an automatic variable.
14646 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14648 PR rtl-optimization/80474
14649 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14651 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14653 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14656 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14658 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14659 (casesi): Emit bounds checking as RTL.
14660 (casesi_internal_mips16_<mode>): Remove bounds checking.
14662 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14664 * config/xtensa/xtensa.c (xtensa_option_override): Append
14665 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14666 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14667 xtensa_doloop_hooks): Define unconditionally.
14668 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14670 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14671 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14672 for it in xtensa_option_override.
14673 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14674 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14676 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14678 * doc/cppopts.texi: Document '-' special value to -MF.
14680 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14682 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14683 (cortex_a53_fconst): Likewise.
14684 (cortex_a53_fpmul): Likewise.
14685 (cortex_a53_f_load_64): Likewise.
14686 (cortex_a53_f_load_many): Likewise.
14687 (cortex_a53_advsimd_alu): Likewise.
14688 (cortex_a53_advsimd_alu_q): Likewise.
14689 (cortex_a53_advsimd_mul): Likewise.
14690 (cortex_a53_advsimd_mul_q): Likewise.
14691 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14692 Add missing fmul, r2f_cvt and fconst cases.
14694 2017-06-14 Richard Biener <rguenther@suse.de>
14696 PR middle-end/81088
14697 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14699 (fold_binary_loc): When associating do not treat pre-existing
14700 TREE_OVERFLOW on literal constants as a reason to allow
14701 TREE_OVERFLOW on associated literal constants.
14703 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14705 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14706 (MASK_FEATURES): New macro.
14707 * config/sparc/sparc.c (sparc_option_override): Remove the special
14708 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14710 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14712 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14713 a division of 0 if non-call exceptions are enabled.
14715 2017-06-14 Andrew Pinski <apinski@cavium.com>
14716 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14719 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14720 Improve vector initialization code gen for only variable case.
14722 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14724 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14726 2017-06-14 Richard Biener <rguenther@suse.de>
14728 PR tree-optimization/81083
14729 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14732 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14734 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14735 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14736 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14737 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14738 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14739 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14741 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14743 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14744 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14746 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14748 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14750 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14752 * config/rs6000/t-rtems: Don't handle SPE.
14754 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14756 * config/rs6000/t-linux: Don't handle SPE.
14758 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14760 * config/rs6000/eabispe.h: Delete file.
14762 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14764 * config/rs6000/t-spe: Delete file.
14766 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14768 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14769 (rs6000_legitimate_offset_address_p): Return false for anything in
14770 V2SImode or V2SFmode.
14772 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14774 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14775 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14776 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14778 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14779 (rs6000_legitimate_offset_address_p): Ditto.
14780 (rs6000_emit_move): Ditto.
14781 (rs6000_init_builtins): Remove V4HI_type_node.
14783 2017-06-13 Martin Liska <mliska@suse.cz>
14786 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14787 (gate_asan): Likewise.
14788 * asan.h (asan_no_sanitize_address_p): Remove the function.
14789 (sanitize_flags_p): New function.
14790 * builtins.def: Fix coding style.
14791 * common.opt: Use renamed enum value.
14792 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14793 * doc/extend.texi: Document no_sanitize attribute.
14794 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14795 to SANITIZE_UNDEFINED_NONDEFAULT.
14796 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14797 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14798 Use sanitize_flags_p.
14799 * gimplify.c (gimplify_function_tree): Likewise.
14800 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14801 * opts.c (parse_no_sanitize_attribute): New function.
14802 (common_handle_option): Use renamed enum value.
14803 * opts.h (parse_no_sanitize_attribute): Declare.
14804 * tree.c (sanitize_flags_p): New function.
14805 * tree.h: Declared here.
14806 * tsan.c: Use sanitize_flags_p.
14807 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14808 (instrument_mem_ref): Likewise.
14809 (instrument_bool_enum_load): Likewise.
14810 (do_ubsan_in_current_function): Remove the function.
14811 (pass_ubsan::execute): Use sanitize_flags_p.
14812 * ubsan.h: Remove do_ubsan_in_current_function
14813 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14814 (dump_function_to_file): Use it here.
14816 2017-06-13 Martin Jambor <mjambor@suse.cz>
14818 PR tree-optimization/80803
14819 PR tree-optimization/81063
14820 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14821 (propagate_subaccesses_across_link): Enqueue subtree whenever
14822 necessary instead of relying on the caller.
14824 2017-06-13 Martin Jambor <mjambor@suse.cz>
14826 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14827 that have a first_link.
14828 (sort_and_splice_var_accesses): Do not check first_link before
14830 (subtree_mark_written_and_enqueue): Likewise.
14831 (propagate_all_subaccesses): Likewise and do not stop at first
14832 parent with a first_link.
14834 2017-06-13 Martin Jambor <mjambor@suse.cz>
14836 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14839 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14841 * match.pd: New pattern.
14843 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14845 * tree-vrp.c (is_masked_range_test): New function.
14846 (register_edge_assert_for): Determine ranges for
14849 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14851 PR tree-optimization/67328
14852 * fold-const.c (maskable_range_p): New function.
14853 (build_range_check): Generate bittests if possible.
14855 2017-06-13 Martin Liska <mliska@suse.cz>
14857 * gimple-pretty-print.c (dump_probability): Add new argument.
14858 (dump_edge_probability): Dump both probability and count.
14859 (dump_gimple_label): Likewise.
14860 (dump_gimple_bb_header): Likewise.
14862 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14865 * config/avr/avr-devices.c: Fix indentation.
14866 * config/avr/gen-avr-mmcu-specs.c: Dito.
14868 2017-06-13 Richard Biener <rguenther@suse.de>
14870 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14871 instead get vector type from stmt_info.
14872 (vectorizable_reduction): Adjust. Remove dead code.
14874 2017-06-13 Richard Biener <rguenther@suse.de>
14876 PR middle-end/81065
14877 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14878 case of C * (x * C2 + C3).
14879 (fold_addr_of_array_ref_difference): Properly fold index difference.
14881 2017-06-12 David S. Miller <davem@davemloft.net>
14884 * config/sparc/sparc.md (return expander): Emit frame blockage if
14885 function uses alloca.
14887 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14889 * combine.c (make_field_assignment): Check len rather than the mode
14890 precision when calling force_to_mode.
14892 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14894 Support multilibs and devices that see flash in RAM address range.
14897 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14898 (avr_mcu_t) <flash_pm_offset>: New field.
14899 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14900 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14901 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14902 (AVR_TINY_PM_OFFSET): Remove macro.
14903 * config/avr/avr.opt (-mshort-calls): New option.
14904 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14905 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14906 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14907 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14908 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14909 instead of avr_arch->have_jmp_call.
14910 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14911 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14912 avr_arch->flash_pm_offset to define.
14913 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14914 new field flash_pm_offset. Add entry for avrxmega3.
14915 (avr_texinfo): Add entry for avrxmega3.
14916 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14917 attiny212, attiny214,
14918 attiny412, attiny414, attiny416, attiny417,
14919 attiny814, attiny816, attiny817,
14920 attiny1614, attiny1616, attiny1617,
14921 attiny3214, attiny3216, attiny3217.
14922 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14923 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14924 (avr_print_operand_address) [AVR_TINY]: Same.
14925 (avr_asm_init_sections) <readonly_data_section>: Only patch
14926 callback if avr_arch->flash_pm_offset = 0.
14927 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14928 for rodata if avr_arch->flash_pm_offset != 0.
14929 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14930 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14931 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14932 (m_options): Append opt_rcall.
14933 (m_dirnames): Append dir_rcall.
14934 * config/avr/t-multilib: Regenerate.
14936 * configure.ac [target=avr]: Check whether avrxmega3 default
14937 linker description file works as needed.
14938 * configure: Regenerate.
14939 * doc/avr-mmcu.texi: Regenerate.
14940 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14941 <__AVR_ARCH__>: Document avrxmega3 and 103.
14942 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14943 <__AVR_SHORT_CALLS__>: Document it.
14944 <__AVR_PM_BASE_ADDRESS__>: Document it.
14945 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14946 (AVR Variable Attributes) <progmem>: Document this is
14947 not needed for avrxmega3.
14948 (AVR Named Address Spaces) <__flash>: Dito.
14950 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14952 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14954 2017-06-12 Doug Rupp <rupp@adacore.com>
14956 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14957 Append vxworks-stdint.h to the tm_file list.
14958 * config/vxworks-stdint.h: New file.
14960 2017-06-12 Martin Liska <mliska@suse.cz>
14962 PR tree-optimization/81041
14963 * tree-profile.c (gimple_gen_ic_func_profiler):
14964 Create an extra BB in profile-generate
14965 (gimple_gen_time_profiler): Likewise.
14967 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14969 PR tree-optimization/81003
14970 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14971 (update_range_test): Use it instead of force_gimple_operand_gsi.
14973 2017-06-12 Richard Biener <rguenther@suse.de>
14975 PR tree-optimization/81053
14976 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14977 with backedge value not defined in loop. Simplify def stmt
14980 2017-06-11 Tom de Vries <tom@codesourcery.com>
14983 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14985 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14986 nvptx_cannot_force_const_mem.
14988 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14990 * opts.c (finish_options): Move test for flag_split_stack after
14991 it has been initialized.
14993 2017-06-11 Jason Merrill <jason@redhat.com>
14995 * tree.h (id_equal): New.
14996 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14997 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14998 instead of strcmp of IDENTIFIER_POINTER.
15000 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15002 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15003 (mark_all_inlined_calls_cdtor): Fix formating.
15004 (inline_transform): Rescale profile before inlining.
15006 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15008 * cgraph.h (cgraph_edge::clone): Update prototype.
15009 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15010 (cgraph_node::create_clone): Update.
15011 (cgraph_node::create_version_clone): Update.
15012 * tree-inline.c (copy_bb): Update.
15013 (expand_call_inline): Update.
15015 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15017 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15018 factored out from ...
15019 (rs6000_emit_prologue): ... here.
15021 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15023 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15024 factored out from ...
15025 (rs6000_emit_prologue): ... here.
15027 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15029 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15031 (handle_missing_profiles): Fix computation of tp_first_run.
15032 (counts_to_freqs): Do not touch freqs when count is 0.
15034 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15036 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15039 2017-06-10 Tom de Vries <tom@codesourcery.com>
15041 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15042 attributes): Document signal effective target.
15044 2017-06-10 Tom de Vries <tom@codesourcery.com>
15046 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15047 Document effective target stack_size.
15049 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15051 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15052 to the edit_context if they can be auto-applied.
15054 2017-06-9 Ian Lance Taylor <iant@golang.org>
15056 * opts.c (finish_options): If -fsplit-stack, disable implicit
15057 -forder-blocks-and-partition.
15058 * doc/invoke.texi (Optimize Options): Document that when using
15059 -fsplit-stack -forder-blocks-and-partition is not implicitly
15062 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15064 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15065 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15066 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15067 * builtins.def (abort, trap, unreachable): Declare cold.
15068 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15069 * tree-core.h (ECF_COLD): New.
15070 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15071 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15073 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15075 * predict.c (unlikely_executed_stmt_p): Cleanup.
15077 2017-06-09 Richard Biener <rguenther@suse.de>
15079 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15080 model if the ref is always written to.
15082 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15084 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15086 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15088 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15091 2017-06-09 Tom de Vries <tom@codesourcery.com>
15094 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15095 "target cannot support label values" when encountering LABEL_REF.
15097 2017-06-09 Martin Liska <mliska@suse.cz>
15099 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15100 (gimple_gen_ic_func_profiler): Emit direct comparison
15101 of __gcov_indirect_call_callee with NULL.
15102 (gimple_gen_time_profiler): Change probability from
15103 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15105 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15107 * profile.c (edge_gcov_counts): Turn to pointer.
15108 (compute_branch_probabilities, compute_branch_probabilities): Update.
15109 (branch_prob): Do not clear edge_gcov_count.
15110 * profile.h (edge_gcov_counts): Turn to pointer.
15111 (edge_gcov_count): Update.
15113 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15115 * gimple.h (gimple_check_failed): Mark cold.
15117 2017-06-09 Richard Biener <rguenther@suse.de>
15119 PR tree-optimization/66623
15120 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15121 refactor check_reduction into two parts, properly computing
15122 whether we have to check reduction validity for outer loop
15125 2017-06-09 Richard Biener <rguenther@suse.de>
15127 PR tree-optimization/79483
15128 * graphite-scop-detection.c (order): New global.
15129 (get_order): Compute bb to order mapping that satisfies code
15130 generation constraints.
15131 (cmp_pbbs): New helper.
15132 (build_scops): Start domwalk at entry block, sort generated
15135 2017-06-09 Richard Biener <rguenther@suse.de>
15137 PR middle-end/81007
15138 * ipa-polymorphic-call.c
15139 (ipa_polymorphic_call_context::restrict_to_inner_class):
15140 Skip FIELD_DECLs with error_mark_node type.
15141 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15144 2017-06-09 Martin Liska <mliska@suse.cz>
15146 * predict.c (struct branch_predictor): New struct.
15147 (test_prediction_value_range): New test.
15148 (predict_c_tests): New function.
15149 * selftest-run-tests.c (selftest::run_tests): Run the function.
15150 * selftest.h: Declare new tests.
15152 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15155 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15156 gen_add3_insn did not fail.
15157 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15158 r0, construct that number in a temporary reg and add that reg to r0.
15159 If asked to put the result in r0 as well, fail.
15161 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15163 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15164 for early expansion of vec_eqv.
15166 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15168 PR middle-end/81005
15169 * ubsan.c (instrument_null): Avoid pointless code temporary.
15170 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15172 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15176 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15179 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15182 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15184 * predict.c (unlikely_executed_edge_p): Move ahead.
15185 (probably_never_executed_edge_p): Use it.
15187 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15189 PR middle-end/79988
15190 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15191 gimple_call_builtin_p call.
15193 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15195 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15196 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15197 rtl_check_failed_type2, rtl_check_failed_code1,
15198 rtl_check_failed_code2, rtl_check_failed_code_mode,
15199 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15200 rtvec_check_failed_bounds, rtl_check_failed_flag,
15201 _fatal_insn_not_found, _fatal_insn): Likewise.
15202 * tree.h (tree_contains_struct_check_failed,
15203 tree_check_failed, tree_not_check_failed,
15204 tree_class_check_failed, tree_range_check_failed,
15205 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15206 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15207 tree_operand_check_failed, omp_clause_check_failed,
15208 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15211 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15213 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15214 flag_branch_probabilities.
15215 * ipa-inline.c (edge_badness): Likewise.
15216 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15217 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15218 * predict.c (maybe_hot_frequency_p): Likewise.
15219 (probably_never_executed): Likewise.
15220 * sched-ebb.c (schedule_ebbs): Likewise.
15221 * sched-rgn.c (find_single_block_region): Likewise.
15222 * tracer.c (tail_duplicate): Likewise.
15224 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15226 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15227 longer requires x_flag_profile_use.
15229 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15231 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15232 instead of flag_reorder_blocks_and_partition.
15233 * dbxout.c (dbxout_function_end): Likewise.
15234 * dwarf2out.c (gen_subprogram_die): Likewise.
15235 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15236 * hw-doloop.c (reorg_loops): Likewise.
15237 * varasm.c (assemble_start_function,
15238 assemble_end_function): Likewise.
15239 (decide_function_section): Do not check for
15240 flag_reorder_blocks_and_partition.
15242 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15244 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15246 (chkp_get_hard_register_fake_addr_expr): Ditto.
15247 (chkp_build_addr_expr): Add check for hard reg case.
15248 (chkp_parse_array_and_component_ref): Ditto.
15249 (chkp_find_bounds_1): Ditto.
15250 (chkp_process_stmt): Don't generate bounds store for
15253 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15255 * predict.c (maybe_hot_bb_p): Do not check profile status.
15256 (maybe_hot_edge_p): Likewise.
15257 (probably_never_executed): Check for zero counts even if profile
15259 (unlikely_executed_edge_p): New function.
15260 (unlikely_executed_stmt_p): New function.
15261 (unlikely_executed_bb_p): New function.
15262 (set_even_probabilities): Use unlikely predicates.
15263 (combine_predictions_for_bb): Likewise.
15264 (predict_paths_for_bb): Likewise.
15265 (predict_paths_leading_to_edge): Likewise.
15266 (determine_unlikely_bbs): New function.
15267 (estimate_bb_frequencies): Use it.
15268 (compute_function_frequency): Use zero counts even if profile is
15270 * profile-count.h: Fix typo.
15272 2017-08-08 Julia Koval <julia.koval@intel.com>
15274 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15275 _mm512_mask_cvtsepi16_storeu_epi8,
15276 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15277 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15278 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15279 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15280 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15281 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15282 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15283 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15284 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15285 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15286 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15287 __builtin_ia32_pmovuswb256mem_mask,
15288 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15289 __builtin_ia32_pmovwb512mem_mask): New builtins.
15291 2017-08-08 Julia Koval <julia.koval@intel.com>
15293 PR target/73350,80862
15294 * config/i386/subst.md (round): Fix round pattern.
15295 * config/i386/i386.c (ix86_erase_embedded_rounding):
15296 Fix erasing rounding for the fixed pattern.
15298 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15300 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15302 2017-06-08 Martin Liska <mliska@suse.cz>
15304 PR gcov-profile/80911
15305 * gcov.c (block_info::block_info): New constructor.
15307 2017-06-07 Carl Love <cel@us.ibm.com>
15309 * config/rs6000/rs6000-c: The return type of the following
15310 built-in functions was implemented as int not long long. Fix sign
15311 of return value for the unsigned version of vec_mulo and vec_mule.
15312 vector unsigned long long vec_bperm (vector unsigned long long,
15313 vector unsigned char)
15314 vector signed long long vec_mule (vector signed int,
15316 vector unsigned long long vec_mule (vector unsigned int,
15317 vector unsigned int)
15318 vector signed long long vec_mulo (vector signed int,
15320 vector unsigned long long vec_mulo (vector unsigned int,
15321 vector unsigned int)
15322 * doc/extend.texi: Fix the documentation for the built-in
15325 2017-06-07 Carl Love <cel@us.ibm.com>
15328 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15331 2017-06-07 Carl Love <cel@us.ibm.com>
15333 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15334 support, Generate doublehv for signed int/float for BE case only.
15336 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15338 * doc/invoke.texi (mcx16): Rewrite.
15340 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15342 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15343 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15344 *mov<mode>_softfloat, and an anonymous splitter): Use
15345 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15347 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15349 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15351 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15352 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15353 (rs6000_debug_reg_global): Adjust.
15354 (rs6000_init_hard_regno_mode_ok): Adjust.
15355 (rs6000_dbx_register_number): Adjust.
15356 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15357 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15358 Remove SPE_ACC and SPEFSCR.
15359 (REG_ALLOC_ORDER): Ditto.
15360 (FRAME_POINTER_REGNUM): Change to 111.
15361 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15362 (REG_CLASS_NAMES): Ditto.
15363 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15364 (REGISTER_NAMES): Ditto.
15365 (ADDITIONAL_REG_NAMES): Ditto.
15366 (rs6000_reg_names): Ditto.
15367 * config/rs6000/rs6000.md: Renumber some register number
15370 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15372 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15374 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15375 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15377 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15378 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15379 Delete the SPE high registers.
15380 (REG_ALLOC_ORDER): Ditto.
15381 (enum reg_class): Remove SPE_HIGH_REGS.
15382 (REG_CLASS_NAMES): Ditto.
15383 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15384 (REGISTER_NAMES): Ditto.
15385 (rs6000_reg_names): Ditto.
15386 * doc/tm.texi.in: Remove SPE as example.
15387 * doc/tm.texi: Regenerate.
15389 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15391 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15392 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15393 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15394 * config/rs6000/rs6000.md (type): Remove "brinc".
15396 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15398 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15399 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15400 * config/rs6000/linuxspe.h: Delete file.
15401 * config/rs6000/rs6000.md: Don't include spe.md.
15402 * config/rs6000/spe.h: Delete file.
15403 * config/rs6000/spe.md: Delete file.
15404 * config/rs6000/t-rs6000: Remove spe.md.
15406 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15408 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15409 (reg_or_none500mem_operand): Delete.
15410 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15411 instead of reg_or_none500mem_operand.
15413 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15415 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15416 handling of SPE flags.
15417 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15419 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15421 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15423 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15424 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15425 paired_divv2sf3): Similar.
15426 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15427 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15428 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15430 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15431 Rename the paired_* instruction patterns.
15432 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15434 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15435 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15436 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15437 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15438 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15439 PAIRED_VECTOR_MODE.
15440 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15441 (spe_func_has_64bit_regs_p): Delete.
15442 (spe_expand_predicate_builtin): Delete.
15443 (spe_expand_evsel_builtin): Delete.
15444 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15445 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15446 (invalid_e500_subreg): Delete.
15447 (rs6000_legitimize_address): Always force_reg op2 as well, for
15448 paired single memory accesses.
15449 (rs6000_member_type_forces_blk): Delete.
15450 (rs6000_spe_function_arg): Delete.
15451 (rs6000_expand_unop_builtin): Delete SPE handling.
15452 (rs6000_expand_binop_builtin): Ditto.
15453 (spe_expand_stv_builtin): Delete.
15454 (bdesc_2arg_spe): Delete.
15455 (spe_expand_builtin): Delete.
15456 (spe_expand_predicate_builtin): Delete.
15457 (spe_expand_evsel_builtin): Delete.
15458 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15459 (spe_init_builtins): Delete.
15460 (spe_func_has_64bit_regs_p): Delete.
15461 (savres_routine_name): Delete "info" parameter. Adjust callers.
15462 (rs6000_emit_stack_reset): Ditto.
15463 (rs6000_dwarf_register_span): Delete.
15464 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15465 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15466 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15468 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15470 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15471 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15472 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15473 mulv2sf3, divv2sf3): Delete expanders.
15475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15477 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15479 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15481 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15482 * config/rs6000/rs6000.c: Ditto.
15484 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15486 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15487 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15488 comparison_operator.
15490 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15492 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15493 * config/rs6000/rs6000.opt: Ditto.
15494 * config/rs6000/t-rtems: Ditto.
15496 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15498 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15499 TARGET_E500_SINGLE by 0, simplify.
15500 * config/rs6000/rs6000.c: Ditto.
15501 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15502 (spe_build_register_parallel): Delete.
15503 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15504 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15505 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15506 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15507 (E500_CONVERT): Delete.
15508 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15510 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15512 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15513 * config/rs6000/dfp.md: Ditto.
15514 (negdd2, *negdd2_fpr): Merge.
15515 (absdd2, *absdd2_fpr): Merge.
15516 (negtd2, *negtd2_fpr): Merge.
15517 (abstd2, *abstd2_fpr): Merge.
15518 * config/rs6000/e500.h: Delete file.
15519 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15520 TARGET_FPRS by 1 and simplify.
15521 * config/rs6000/rs6000-c.c: Ditto.
15522 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15523 TARGET_DF_SPE by 0.
15524 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15526 * config/rs6000/rs6000.md: Ditto.
15527 (floatdidf2, *floatdidf2_fpr): Merge.
15528 (move_from_CR_gt_bit): Delete.
15529 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15530 (E500_CR_IOR_COMPARE): Delete.
15531 (All patterns that require !TARGET_FPRS): Delete.
15532 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15534 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15536 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15538 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15540 * graphds.c (add_edge): Intitialize edge's attached data.
15541 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15542 pointer parameter. Call pointed function on each edge during
15543 graph traversing. Skip traversing the edge when the function
15545 (graphds_dfs, graphds_scc): Ditto.
15546 (for_each_edge): New parameter. Pass the new parameter to callback
15548 * graphds.h (skip_edge_callback): New function pointer type.
15549 (graphds_dfs, graphds_scc): New function pointer parameter.
15550 (graphds_edge_callback, for_each_edge): New parameter.
15552 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15554 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15555 out code checking if runtime alias check is possible to below ...
15556 Call the new function.
15557 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15558 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15560 2017-06-07 Marek Polacek <polacek@redhat.com>
15563 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15564 TYPE_OVERFLOW_WRAPS check.
15566 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15568 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15569 if versioning is required.
15570 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15571 peeling with the check for versioning.
15573 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15575 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15576 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15577 Set true to new parameter if new ssa variable is defined.
15578 (vect_gen_vector_loop_niters): Refactor. Set range information
15579 for the new vector loop bound variable.
15580 (vect_do_peeling): Ditto.
15582 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15584 * tree-affine.c (ssa.h): Include header file.
15585 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15586 has wrapping overflow behavior.
15588 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15590 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15592 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15594 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15595 (tree_to_aff_combination): ... here.
15597 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15599 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15600 reg_pressure model function.
15601 (ivopts_global_cost_for_size): Delete.
15602 (determine_set_costs, iv_ca_recount_cost): Call new model function
15603 ivopts_estimate_reg_pressure.
15605 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15607 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15608 expensive than udiv. Remove floating point cases from mod.
15610 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15612 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15613 Increase idiv cost.
15615 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15617 * config/aarch64/aarch64.md
15618 (copysignsf3): Fix mask generation.
15620 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15622 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15624 (class dump_manager): Add register_dumps method.
15625 * dumpfile.c: Include langhooks.h.
15626 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15627 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15628 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15629 (dump_manager::dump_register): Start with 512 entries instead of 32.
15630 (dump_manager::register_dumps): New method.
15631 * toplev.c (general_init): Instead of invoking register_dumps
15632 langhook, invoke register_dumps method on the dump manager.
15633 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15636 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15638 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15639 Say that # requires an associated define_split to exist, and that
15640 the define_split must be suitable for use after register allocation.
15642 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15644 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15645 (compute_outgoing_frequencies): Also initialize zero counts.
15646 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15647 around loops; preserve more of profile when nothing changes.
15649 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15651 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15653 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15654 * config/arm/arm-cpu-cdata.h: Regenerate.
15655 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15656 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15657 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15658 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15659 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15661 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15663 * config/arm/t-rmprofile: Likewise.
15664 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15666 2017-06-06 David S. Miller <davem@davemloft.net>
15669 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15670 blockage if function uses alloca.
15672 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15674 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15675 New "uid" fields to hold pretty-print IDs of group and ref.
15676 Memory references are now identified as <group_id>:<ref_id>
15677 instead of using [random] addresses.
15678 (dump_mem_details): Simplify, no functional change.
15679 (dump_mem_ref): Simplify and make output more concise.
15680 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15681 (find_or_create_group): Initialize group uid.
15682 (record_ref): Initialize ref uid. Improve debug output.
15683 (prune_group_by_reuse, should_issue_prefetch_p,)
15684 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15685 (mark_nontemporal_store, determine_loop_nest_reuse):
15686 Improve debug output.
15688 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15690 * dbgcnt.def (prefetch): New debug counter.
15691 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15692 (schedule_prefetches): Stop issueing prefetches if debug counter
15695 2017-06-06 Tom de Vries <tom@codesourcery.com>
15697 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15698 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15700 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15702 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15703 Use aarch64_reg_or_zero predicate for operand 4.
15704 (aarch64_compare_and_swap<mode> define_insn_and_split):
15705 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15706 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15708 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15710 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15711 (arm_compute_save_core_reg_mask): This.
15712 (thumb1_compute_save_reg_mask): Rename into ...
15713 (thumb1_compute_save_core_reg_mask): This.
15714 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15715 (arm_compute_frame_layout): Likewise.
15717 2017-06-06 Richard Biener <rguenther@suse.de>
15719 PR tree-optimization/80974
15720 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15721 keep or clear leaders SSA info.
15723 2017-06-06 Tom de Vries <tom@codesourcery.com>
15725 * config/nvptx/nvptx.c (split_mode_p): New function.
15726 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15728 2017-06-06 Tom de Vries <tom@codesourcery.com>
15730 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15732 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15735 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15738 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15740 * shrink-wrap.c (handle_simple_exit): Update profile.
15741 (try_shrink_wrapping): Upate profile.
15743 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15745 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15746 (tree_guess_outgoing_edge_probabilities): New.
15747 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15748 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15750 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15752 * ipa-split.c (split_function): Initialize return bb profile.
15754 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15756 * profile.c (compute_branch_probabilities): Also initialize
15757 EXIT_BLOCK profile.
15759 2017-06-06 Richard Biener <rguenther@suse.de>
15761 PR tree-optimization/80928
15762 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15763 (vect_analyze_loop_operations): Properly guard analysis for
15765 (vect_transform_loop): Likewise.
15766 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15767 (vect_model_induction_cost): Do not cost for pure SLP.
15768 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15769 of induction in inner loop vectorization.
15770 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15771 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15772 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15774 (vect_analyze_slp_cost_1): Cost induction.
15775 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15776 (vect_get_slp_vect_defs): Likewise.
15777 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15778 (vect_transform_stmt): Handle SLP reductions.
15779 * tree-vectorizer.h (vectorizable_induction): Adjust.
15781 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15783 * config/rs6000/rs6000.c (make_resolver_func): Update
15784 init_lowered_empty_function call.
15786 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15788 * doc/invoke.texi: Document the -fprofile-abs-path option.
15789 * common.opt (fprofile-abs-path): New option.
15790 * gcov-io.h (gcov_write_filename): Declare.
15791 * gcov-io.c (gcov_write_filename): New function.
15792 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15793 * profile.c (output_location): Likewise.
15795 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15797 * shring-wrap.c: Revert accidental commit.
15799 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15801 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15803 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15805 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15807 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15808 profile in callgraph edge.
15809 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15810 (apply_scale): Likewise.
15811 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15812 Also scale profile when inlining function with zero profile.
15813 (initialize_cfun): Update exit block profile even when it is zero.
15814 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15815 when profile is read.
15817 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15819 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15820 (CLONE_*): New constants to define the processors we can generate
15821 code for with the target_clone attribute.
15822 (rs6000_clone_map): New array to identify which clone processors
15823 the current program is running on.
15824 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15825 target_clone attribute.
15826 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15827 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15828 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15829 (cpu_expand_builtin): Add support for target_clone attribute.
15830 (rs6000_valid_attribute_p): Allow "default" attribute.
15831 (get_decl_name): New debug function to simplify printing the
15832 current function name in debugging statements.
15833 (rs6000_clone_priority): New functions to support the target_clone
15834 attribute, and be able to generate code to switch between ISA 2.05
15835 through ISA 3.0 (power6 through power9).
15836 (rs6000_compare_version_priority): Likewise.
15837 (rs6000_get_function_versions_dispatcher): Likewise.
15838 (make_resolver_func): Likewise.
15839 (add_condition_to_bb): Likewise.
15840 (dispatch_function_versions): Likewise.
15841 (rs6000_generate_version_dispatcher_body): Likewise.
15842 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15843 (fusion_gpr_load_p): Fix a spacing issue.
15844 * doc/extend.texi (Common Function Attributes): Document that the
15845 PowerPC supports the target_clone attribute.
15847 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15849 * config/arm/arm.h: explain F symbol found in description of ARM
15850 register allocation in its legend.
15852 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15854 * config/mips/frame-header-opt.c: Include profile-count.h.
15855 * config/riscv/riscv.c: Include profile-count.h
15857 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15859 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15861 (sm_set_flag_if_changed): Add bbs field.
15862 (execute_sm_if_changed_flag_set): Pass BBS.
15863 (execute_sm): Update.
15865 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15867 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15870 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15872 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15873 (peephole2): New peephole2 to emit the above.
15874 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15876 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15878 * config/aarch64/aarch64.c (define_peephole2 above
15879 *sub_<shift>_<mode>): New peephole.
15881 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15883 * config/i386/i386.c (make_resolver_func): Update.
15884 * Makefile.in: Add profile-count.h and profile-count.o
15885 * auto-profile.c (afdo_indirect_call): Update to new API.
15886 (afdo_set_bb_count): Update.
15887 (afdo_propagate_edge): Update.
15888 (afdo_propagate_circuit): Update.
15889 (afdo_calculate_branch_prob): Update.
15890 (afdo_annotate_cfg): Update.
15891 * basic-block.h: Include profile-count.h
15892 (struct edge_def): Turn count to profile_count.
15893 (struct basic_block_def): Likewie.
15894 (REG_BR_PROB_BASE): Move to profile-count.h
15895 (RDIV): Move to profile-count.h
15896 * bb-reorder.c (max_entry_count): Turn to profile_count.
15897 (find_traces): Update.
15898 (rotate_loop):Update.
15899 (connect_traces):Update.
15900 (sanitize_hot_paths):Update.
15901 * bt-load.c (migrate_btr_defs): Update.
15902 * cfg.c (RDIV): Remove.
15903 (init_flow): Use alloc_block.
15904 (alloc_block): Uninitialize count.
15905 (unchecked_make_edge): Uninitialize count.
15906 (check_bb_profile): Update.
15907 (dump_edge_info): Update.
15908 (dump_bb_info): Update.
15909 (update_bb_profile_for_threading): Update.
15910 (scale_bbs_frequencies_int): Update.
15911 (scale_bbs_frequencies_gcov_type): Update.
15912 (scale_bbs_frequencies_profile_count): New.
15913 * cfg.h (update_bb_profile_for_threading): Update.
15914 (scale_bbs_frequencies_profile_count): Declare.
15915 * cfgbuild.c (compute_outgoing_frequencies): Update.
15916 (find_many_sub_basic_blocks): Update.
15917 * cfgcleanup.c (try_forward_edges): Update.
15918 (try_crossjump_to_edge): Update.
15919 * cfgexpand.c (expand_gimple_tailcall): Update.
15920 (construct_exit_block): Update.
15921 * cfghooks.c (verify_flow_info): Update.
15922 (dump_bb_for_graph): Update.
15923 (split_edge): Update.
15924 (make_forwarder_block): Update.
15925 (duplicate_block): Update.
15926 (account_profile_record): Update.
15927 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15928 (get_estimated_loop_iterations): Update.
15929 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15930 (single_likely_exit): Update.
15931 * cfgloopmanip.c (scale_loop_profile): Update.
15933 (set_zero_probability): Update.
15934 (lv_adjust_loop_entry_edge): Update.
15935 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15936 (purge_dead_edges): Update.
15937 (rtl_account_profile_record): Update.
15938 * cgraph.c (cgraph_node::create): Uninitialize count.
15939 (symbol_table::create_edge): Uninitialize count.
15940 (cgraph_update_edges_for_call_stmt_node): Update.
15941 (cgraph_edge::dump_edge_flags): Update.
15942 (cgraph_node::dump): Update.
15943 (cgraph_edge::maybe_hot_p): Update.
15944 * cgraph.h: Include profile-count.h
15945 (create_clone), create_edge, create_indirect_edge): Update.
15946 (cgraph_node): Turn count to profile_count.
15947 (cgraph_edge0: Likewise.
15948 (make_speculative, clone): Update.
15949 (create_edge): Update.
15950 (init_lowered_empty_function): Update.
15951 * cgraphclones.c (cgraph_edge::clone): Update.
15952 (duplicate_thunk_for_node): Update.
15953 (cgraph_node::create_clone): Update.
15954 * cgraphunit.c (cgraph_node::analyze): Update.
15955 (cgraph_node::expand_thunk): Update.
15956 * final.c (dump_basic_block_info): Update.
15957 * gimple-streamer-in.c (input_bb): Update.
15958 * gimple-streamer-out.c (output_bb): Update.
15959 * graphite.c (print_global_statistics): Update.
15960 (print_graphite_scop_statistics): Update.
15961 * hsa-brig.c: Include basic-block.h.
15962 * hsa-dump.c: Include basic-block.h.
15963 * hsa-gen.c (T sum_slice): Update.
15964 (convert_switch_statements):Update.
15965 * hsa-regalloc.c: Include basic-block.h.
15966 * ipa-chkp.c (chkp_produce_thunks): Update.
15967 * ipa-cp.c (struct caller_statistics): Update.
15968 (init_caller_stats): Update.
15969 (gather_caller_stats): Update.
15970 (ipcp_cloning_candidate_p): Update.
15971 (good_cloning_opportunity_p): Update.
15972 (get_info_about_necessary_edges): Update.
15973 (dump_profile_updates): Update.
15974 (update_profiling_info): Update.
15975 (update_specialized_profile): Update.
15976 (perhaps_add_new_callers): Update.
15977 (decide_about_value): Update.
15978 (ipa_cp_c_finalize): Update.
15979 * ipa-devirt.c (struct odr_type_warn_count): Update.
15980 (struct decl_warn_count): Update.
15981 (struct final_warning_record): Update.
15982 (possible_polymorphic_call_targets): Update.
15983 (ipa_devirt): Update.
15984 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15985 * ipa-icf.c (sem_function::merge): Update.
15986 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15987 * ipa-inline.c (compute_uninlined_call_time): Update.
15988 (compute_inlined_call_time): Update.
15989 (want_inline_small_function_p): Update.
15990 (want_inline_self_recursive_call_p): Update.
15991 (edge_badness): Update.
15992 (lookup_recursive_calls): Update.
15993 (recursive_inlining): Update.
15994 (inline_small_functions): Update.
15995 (dump_overall_stats): Update.
15996 (dump_inline_stats): Update.
15997 * ipa-profile.c (ipa_profile_generate_summary): Update.
15998 (ipa_propagate_frequency): Update.
15999 (ipa_profile): Update.
16000 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16001 * ipa-utils.c (ipa_merge_profiles): Update.
16002 * loop-doloop.c (doloop_modify): Update.
16003 * loop-unroll.c (report_unroll): Update.
16004 (unroll_loop_runtime_iterations): Update.
16005 * lto-cgraph.c (lto_output_edge): Update.
16006 (lto_output_node): Update.
16007 (input_node): Update.
16008 (input_edge): Update.
16009 (merge_profile_summaries): Update.
16010 * lto-streamer-in.c (input_cfg): Update.
16011 * lto-streamer-out.c (output_cfg): Update.
16012 * mcf.c (create_fixup_graph): Update.
16013 (adjust_cfg_counts): Update.
16014 (sum_edge_counts): Update.
16015 * modulo-sched.c (sms_schedule): Update.
16016 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16017 * predict.c (maybe_hot_count_p): Update.
16018 (probably_never_executed): Update.
16019 (dump_prediction): Update.
16020 (combine_predictions_for_bb): Update.
16021 (propagate_freq): Update.
16022 (handle_missing_profiles): Update.
16023 (counts_to_freqs): Update.
16024 (rebuild_frequencies): Update.
16025 (force_edge_cold): Update.
16026 * predict.h: Include profile-count.h
16027 (maybe_hot_count_p, counts_to_freqs): UPdate.
16028 * print-rtl-function.c: Do not include cfg.h
16029 * print-rtl.c: Include basic-block.h
16030 * profile-count.c: New file.
16031 * profile-count.h: New file.
16032 * profile.c (is_edge_inconsistent): Update.
16033 (correct_negative_edge_counts): Update.
16034 (is_inconsistent): Update.
16035 (set_bb_counts): Update.
16036 (read_profile_edge_counts): Update.
16037 (compute_frequency_overlap): Update.
16038 (compute_branch_probabilities): Update; Initialize and deinitialize
16040 (branch_prob): Update.
16041 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16042 (edge_gcov_count): New.
16043 (bb_gcov_count): New.
16044 * shrink-wrap.c (try_shrink_wrapping): Update.
16045 * tracer.c (better_p): Update.
16046 * trans-mem.c (expand_transaction): Update.
16047 (ipa_tm_insert_irr_call): Update.
16048 (ipa_tm_insert_gettmclone_call): Update.
16049 * tree-call-cdce.c: Update.
16050 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16051 (gimple_duplicate_sese_tail): Update.
16052 (gimple_account_profile_record): Update.
16053 (execute_fixup_cfg): Update.
16054 * tree-inline.c (copy_bb): Update.
16055 (copy_edges_for_bb): Update.
16056 (initialize_cfun): Update.
16057 (freqs_to_counts): Update.
16058 (copy_cfg_body): Update.
16059 (expand_call_inline): Update.
16060 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16061 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16062 (try_unroll_loop_completely): Update.
16063 (try_peel_loop): Update.
16064 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16065 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16066 * tree-ssa-loop-split.c (connect_loops): Update.
16067 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16068 * tree-ssa-reassoc.c (branch_fixup): Update.
16069 * tree-ssa-tail-merge.c (replace_block_by): Update.
16070 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16071 (compute_path_counts): Update.
16072 (update_profile): Update.
16073 (recompute_probabilities): Update.
16074 (update_joiner_offpath_counts): Update.
16075 (estimated_freqs_path): Update.
16076 (freqs_to_counts_path): Update.
16077 (clear_counts_path): Update.
16078 (ssa_fix_duplicate_block_edges): Update.
16079 (duplicate_thread_path): Update.
16080 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16081 (struct switch_conv_info): Update.
16082 * tree-tailcall.c (decrease_profile): Update.
16083 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16084 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16085 * value-prof.c (check_counter): Update.
16086 (gimple_divmod_fixed_value): Update.
16087 (gimple_mod_pow2): Update.
16088 (gimple_mod_subtract): Update.
16089 (gimple_ic_transform): Update.
16090 (gimple_stringop_fixed_value): Update.
16091 * value-prof.h (gimple_ic): Update.
16093 2017-06-02 Carl Love <cel@us.ibm.com>
16095 * config/rs6000/rs6000-c: Add support for built-in functions
16096 vector double vec_doublee (vector signed int);
16097 vector double vec_doublee (vector unsigned int);
16098 vector double vec_doublee (vector float);
16099 vector double vec_doubleh (vector signed int);
16100 vector double vec_doubleh (vector unsigned int);
16101 vector double vec_doubleh (vector float);
16102 vector double vec_doublel (vector signed int);
16103 vector double vec_doublel (vector unsigned int);
16104 vector double vec_doublel (vector float);
16105 vector double vec_doubleo (vector signed int);
16106 vector double vec_doubleo (vector unsigned int);
16107 vector double vec_doubleo (vector float);.
16108 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16109 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16111 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16112 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16113 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16115 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16116 vec_doublel, vec_doubleh.
16117 * doc/extend.texi: Update the built-in documentation file for the
16118 new built-in functions.
16120 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16123 * ipa-inline-analysis.c (free_growth_caches): Set
16124 edge_removal_hook_holder to NULL after removing it.
16126 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16128 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16129 comparision with zero.
16131 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16132 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16133 for early expansion of vec_min and vec_max builtins.
16134 (builtin_function_type): Add min/max unsigned variants to those
16135 identified as having unsigned arguments.
16137 2017-06-02 Olivier Hainque <hainque@adacore.com>
16139 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16141 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16143 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16144 Use VALL_F16 iterator rather than VALL.
16146 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16148 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16149 Emit CBNZ inside loop when doing a strong exchange and comparing
16150 against zero. Generate the CC flags after the loop.
16152 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16154 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16155 (dl_section_ref): New.
16156 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16157 On AIX, append an expression to subtract the size of the
16158 section length to dl_section_ref.
16160 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16162 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16163 for early expansion of vector absolute builtins.
16165 2017-06-02 Richard Biener <rguenther@suse.de>
16167 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16168 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16170 2017-06-02 Richard Biener <rguenther@suse.de>
16172 PR tree-optimization/80948
16173 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16174 stmt order as well.
16176 2017-06-02 Richard Biener <rguenther@suse.de>
16178 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16180 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16181 for inductions as relevant.
16183 2017-06-02 Richard Biener <rguenther@suse.de>
16185 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16186 (vectorizable_induction): ... this. Remove dead code.
16188 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16190 * builtins. (expand_builtin_alloca): Remove second parameter and
16191 infer its value from the first parameter instead.
16192 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16194 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16196 PR rtl-optimization/80903
16197 * loop-doloop.c (add_test): Unshare sequence.
16199 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16201 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16203 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16205 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16207 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16208 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16209 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16210 (xlogue_layout::get_stub_rtx): Make static.
16211 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16212 (xlogue_layout::compute_stub_managed_regs): Rename to...
16213 (xlogue_layout::count_stub_managed_regs): ...this.
16214 (xlogue_layout::is_stub_managed_reg): New function.
16215 (xlogue_layout::m_stub_names): Rename to...
16216 (xlogue_layout::s_stub_names): ...this, make static.
16217 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16218 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16219 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16220 xlogue_layout::s_stub_names): Instantiate statics.
16221 (stub_managed_regs): Remove.
16222 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16223 (disable_call_ms2sysv_xlogues): Rename to...
16224 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16225 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16227 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16228 change after reload_completed.
16229 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16231 (ix86_expand_prologue): Likewise.
16232 (ix86_expand_epilogue): Likewise.
16233 (ix86_expand_split_stack_prologue): Likewise.
16234 (ix86_compute_frame_layout): Remove frame parameter ...
16235 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16236 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16238 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16239 (ix86_frame): Move from here ...
16240 * config/i386/i386.h (ix86_frame): ... to here.
16241 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16242 complete ix86_frame data structure instead. Remove some_ld_name.
16244 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16246 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16247 symbols that hold a DECL_VALUE_EXPR.
16249 2017-06-01 Martin Jambor <mjambor@suse.cz>
16251 PR tree-optimization/80898
16252 * tree-sra.c (process_subtree_disqualification): Removed.
16253 (disqualify_candidate): Do not acll
16254 process_subtree_disqualification.
16255 (subtree_mark_written_and_enqueue): New function.
16256 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16257 RHS has been disqualified and re-queue LHS if necessary. Apart
16258 from that, ignore disqualified RHS.
16260 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16262 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16263 address fetch for z10 or later.
16265 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16267 * config/arc/arc.md (tst_movb): Add guard when splitting.
16269 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16271 * config/arc/arc.c (arc_can_eliminate): Test against
16272 arc_frame_pointer_needed.
16274 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16276 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16277 to prevent store reordering.
16278 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16279 (type): Add block type.
16280 (stack_tie): Define special instruction to be used in
16283 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16285 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16286 constraint. It is not valid for the pattern.
16287 (noncommutative_binary_comparison): Likewise.
16289 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16291 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16294 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16296 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16297 be used by the reg-alloc.
16299 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16301 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16302 reg-alloc when having mul64 or mul32x16 instructions.
16303 (mulsidi3): Likewise.
16304 (umulsidi3): Likewise.
16305 (mulsi32x16): New pattern.
16306 (mulsi64): Likewise.
16307 (mulsidi64): Likewise.
16308 (umulsidi64): Likewise.
16309 (MUL32x16_REG): Define.
16310 (mul64_600): Use MUL32x16_REG.
16311 (mac64_600): Likewise.
16312 (umul64_600): Likewise.
16313 (umac64_600): Likewise.
16315 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16317 * config/arc/arc.md (mulsi3_700): Make it commutative.
16319 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16321 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16323 (*sign_extendsidi2_insn): Likewise for movstosw.
16325 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16327 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16328 the type of the input discriminant value. Convert the
16329 discriminant value of signedness vary.
16331 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16333 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16334 Add to -Wall section.
16336 2017-06-01 Richard Biener <rguenther@suse.de>
16338 PR middle-end/66313
16339 * fold-const.c (fold_plusminus_mult_expr): If the factored
16340 factor may be zero use a wrapping type for the inner operation.
16341 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16342 and handle moved defs.
16343 (process_assignment): Properly guard the unary op case. Return a
16344 tri-state indicating that moving the stmt before the call may allow
16345 to continue. Pass through to_move.
16346 (find_tail_calls): Handle moving unrelated defs before
16349 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16352 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16353 splitter result in the canonical way.
16355 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16357 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16358 also for 32bit target. Update insn attributes.
16359 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16361 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16363 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16364 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16365 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16366 (_mm_maskz_min_ss): New intrinsics.
16368 2017-05-31 Martin Liska <mliska@suse.cz>
16370 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16371 Change comment style to one we normally use.
16372 (vectorizable_reduction): Likewise.
16373 (vectorizable_induction): Likewise.
16374 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16375 (vectorizable_call): Likewise.
16376 (vectorizable_simd_clone_call): Likewise.
16377 (vectorizable_conversion): Likewise.
16378 (vectorizable_assignment): Likewise.
16379 (vectorizable_shift): Likewise.
16380 (vectorizable_operation): Likewise.
16381 (vectorizable_store): Likewise.
16382 (vectorizable_load): Likewise.
16383 * tree-vectorizer.h: Likewise.
16385 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16387 * passes.c (emergency_dump_function): New.
16388 * tree-pass.h (emergency_dump_function): Declare.
16389 * plugin.c (plugins_internal_error_function): Remove.
16390 * plugin.h (plugins_internal_error_function): Remove declaration.
16391 * toplev.c (internal_error_function): New static function. Use it...
16392 (general_init): ...here.
16394 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16396 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16397 (arc_rtx_costs): Add costs for new patterns.
16398 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16399 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16401 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16403 * tree-ssa-strlen.c (get_next_strinfo): New function.
16404 (get_stridx_plus_constant): Use it.
16405 (zero_length_string): Likewise.
16406 (adjust_related_strinfos): Likewise.
16407 (adjust_last_stmt): Likewise.
16409 2017-05-31 Richard Biener <rguenther@suse.de>
16412 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16413 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16415 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16417 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16418 loop_vinfo argument and use of dependence distance vectors.
16419 Check instead whether the two references differ only in their
16420 initial value and assume that they have the same alignment if the
16421 difference is a multiple of the vector alignment.
16422 (vect_analyze_data_refs_alignment): Update call accordingly.
16424 2017-05-31 Martin Liska <mliska@suse.cz>
16427 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16429 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16431 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16432 (create_intersect_range_checks): Move from ...
16433 * tree-data-ref.c (create_intersect_range_checks_index)
16434 (create_intersect_range_checks): ... to here.
16435 (create_runtime_alias_checks): New function factored from ...
16436 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16437 here. Call above function.
16438 * tree-data-ref.h (create_runtime_alias_checks): New function.
16440 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16442 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16443 segment length for dr_b and compute it in wide_int.
16445 2017-05-31 Richard Biener <rguenther@suse.de>
16447 PR tree-optimization/80906
16448 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16449 and pass through iv_map.
16450 (copy_bb_and_scalar_dependences): Adjust.
16451 (translate_pending_phi_nodes): Likewise.
16452 (copy_loop_close_phi_args): Handle code-generating IVs instead
16455 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16457 * diagnostic-color.c (color_dict): Add "type-diff".
16458 (parse_gcc_colors): Update comment.
16459 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16460 -fdiagnostics-show-template-tree and -fno-elide-type.
16461 (GCC_COLORS): Add type-diff to example.
16463 (-fdiagnostics-show-template-tree): New.
16464 (-fno-elide-type): New.
16465 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16466 the pp_format_decoder callback. Call any m_format_postprocessor's
16468 (pretty_printer::pretty_printer): Initialize
16469 m_format_postprocessor.
16470 (pretty_printer::~pretty_printer): Delete any
16471 m_format_postprocessor.
16472 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16473 (class format_postprocessor): New class.
16474 (struct pretty_printer::format_decoder): Document the new parameters.
16475 (struct pretty_printer::m_format_postprocessor): New field.
16476 * tree-diagnostic.c (default_tree_printer): Update for new
16477 bool and const char ** params.
16478 * tree-diagnostic.h (default_tree_printer): Likewise.
16480 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16482 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16483 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16484 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16485 rs6000_gen_cell_microcode code.
16486 (rs6000_final_prescan_insn): Delete.
16487 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16488 "warn-cell-microcode" entries.
16489 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16490 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16491 throughout. Change cc_reg_not_micro_cr0_operand to
16492 cc_reg_not_cr0_operand throughout.
16493 (*extendhi<mode>2_noload): Delete.
16494 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16495 (mwarn-cell-microcode): Delete.
16496 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16497 -mgen-cell-microcode and -mwarn-cell-microcode.
16499 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16502 * config/i386/constraints.md (Yd): New constraint.
16504 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16505 and (?Yd, r) alternatives. Update insn attributes.
16506 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16507 and (?*Yd, r) alternatives. Update insn attributes.
16508 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16510 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16512 * gimplify.c (gimplify_modify_expr): Don't create a
16513 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16516 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16518 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16520 2017-05-30 Richard Biener <rguenther@suse.de>
16522 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16523 and reduc_def fields.
16524 (STMT_VINFO_REDUC_TYPE): New define.
16525 (STMT_VINFO_REDUC_DEF): Likewise.
16526 (vect_force_simple_reduction): Adjust prototype.
16527 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16528 (vect_is_simple_reduction): Remove check_reduction argument.
16529 (vect_force_simple_reduction): Adjust and set
16530 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16531 (vectorizable_reduction): Do not re-do reduction analysis
16532 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16533 * tree-parloops.c (gather_scalar_reductions): Adjust.
16535 2017-05-30 Richard Biener <rguenther@suse.de>
16537 PR middle-end/80901
16538 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16541 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16543 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16544 Introduce unknown_misalignment parameter and remove vf.
16545 (vect_peeling_hash_get_lowest_cost):
16546 Pass unknown_misalignment parameter.
16547 (vect_enhance_data_refs_alignment):
16548 Fix unsupportable data ref treatment.
16550 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16552 * tree-vect-data-refs.c (vect_get_data_access_cost):
16553 Workaround for SLP handling.
16554 (vect_enhance_data_refs_alignment):
16555 Compute costs for doing no peeling at all, compare to the best
16556 peeling costs so far and avoid peeling if cheaper.
16558 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16560 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16561 Return peeling info and set costs to zero for unlimited cost
16563 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16564 with unknown misalignment. Compute and costs for unknown
16565 misalignment, compare them to the costs for known misalignment
16566 and choose the cheapest for peeling.
16568 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16570 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16571 (vect_get_peeling_costs_all_drs): Create function.
16572 (vect_peeling_hash_get_lowest_cost):
16573 Use vect_get_peeling_costs_all_drs.
16574 (vect_peeling_supportable): Create function.
16575 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16577 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16579 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16580 DR_HAS_NEGATIVE_STEP.
16581 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16582 (vect_enhance_data_refs_alignment): Use.
16583 (vect_duplicate_ssa_name_ptr_info): Use.
16584 * tree-vectorizer.h (dr_misalignment): Use.
16585 (known_alignment_for_access_p): Use.
16587 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16590 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16592 (has_section_name): New function.
16594 2017-05-30 Martin Liska <mliska@suse.cz>
16597 * auto-profile.c (get_function_decl_from_block): Fix
16600 2017-05-30 Richard Biener <rguenther@suse.de>
16602 PR middle-end/80876
16603 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16605 2017-05-30 Martin Liska <mliska@suse.cz>
16607 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16608 * dumpfile.h (struct dump_file_info): Remove ctors.
16610 2017-05-30 Martin Liska <mliska@suse.cz>
16612 * predict.def: Fix GNU coding style.
16614 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16616 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16617 Mark 'to' argument with ATTRIBUTE_UNUSED.
16619 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16621 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16622 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16623 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16626 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16628 * doc/install.texi (Options specification): Restore entry of
16629 --enable-sjlj-exceptions.
16631 2017-05-27 Michael Eager <eager@eagercon.com>
16634 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16636 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16638 * config/microblaze/microblaze.h
16639 (FIXED_REGISTERS): Update in macro.
16640 (CALL_USED_REGISTERS): Update in macro.
16642 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16644 * doc/install.texi: Add links to macOS binary distributions.
16646 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16650 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16652 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16654 2017-05-26 Martin Liska <mliska@suse.cz>
16656 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16658 2017-05-26 Martin Liska <mliska@suse.cz>
16660 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16661 always leading ';; '.
16662 (dump_bb_info): Likewise.
16663 (brief_dump_cfg): Likewise.
16664 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16665 * dumpfile.c: Remove usage of TDF_VERBOSE.
16666 * dumpfile.h (enum dump_kind): Likewise.
16667 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16668 * print-tree.c (debug_verbose): Remove.
16669 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16670 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16671 * tree-diagnostic.c (default_tree_printer): Replace
16672 TDF_DIAGNOSTIC with TDF_SLIM.
16674 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16676 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16677 in parameter loop, rather than loop_vinfo.
16678 (create_intersect_range_checks): Ditto.
16679 (vect_create_cond_for_alias_checks): Update call to above functions.
16681 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16683 PR tree-optimization/80815
16684 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16685 for merging runtime alias checks. Handle negative DR_STEPs.
16687 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16689 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16691 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16692 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16693 out code pruning runtime alias checks.
16694 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16695 factored out from above.
16696 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16698 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16700 (prune_runtime_alias_test_list): New decalaration.
16702 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16704 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16705 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16706 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16707 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16708 (operator==, comp_dr_with_seg_len_pair): Ditto.
16709 (vect_prune_runtime_alias_test_list): Ditto.
16711 2017-05-26 Martin Liska <mliska@suse.cz>
16714 * params.def: Bound partial-inlining-entry-probability param.
16716 2017-05-26 Marek Polacek <polacek@redhat.com>
16719 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16722 2017-05-26 Richard Biener <rguenther@suse.de>
16724 PR tree-optimization/80842
16725 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16728 2017-05-26 Richard Biener <rguenther@suse.de>
16730 PR tree-optimization/80844
16731 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16733 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16735 * doc/md.texi (Machine Constraints): Update x86 family
16736 machine constraints section to match 'config/i386/constraints.md'.
16738 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16740 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16742 2017-05-25 Nathan Sidwell <nathan@acm.org>
16744 * doc/invoke.texi (--enable-languages): Update documentation.
16746 2017-05-25 Martin Liska <mliska@suse.cz>
16748 * dumpfile.c: Add TDF_FOLDING.
16749 * dumpfile.h (enum dump_kind): Likewise.
16750 * genmatch.c (dt_simplify::gen_1): Use it.
16752 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16754 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16756 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16758 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16759 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16761 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16763 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16764 * match.pd (X == C): Rewrite it here.
16765 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16766 with_certain_nonzero_bits2): New predicates.
16767 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16769 2017-05-24 Nathan Sidwell <nathan@acm.org>
16771 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16774 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16777 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16779 * config/powerpcspe: New port. Files are copied from the rs6000
16780 port, with "rs6000" in filenames replaced by "powerpcspe".
16782 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16784 PR rtl-optimization/80754
16785 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16787 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16789 * config/sparc/sparc.md (length): Return the correct value for -mflat
16790 sibcalls to match output_sibcall.
16792 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16796 * config/rs6000/rs6000.c (struct machine_function): Add new field
16798 (rs6000_get_separate_components): Init that field, use it.
16799 (rs6000_components_for_bb): Use the field.
16801 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16803 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16805 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16807 PR middle-end/80823
16808 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16810 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16813 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16814 address against address_operand predicate.
16815 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16817 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16819 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16820 parameters passed indirectly.
16822 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16824 * config/i386/i386.md (*movdi_internal): Remove SSE4
16825 alternative 18 (?r, *v). Update insn attributes.
16826 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16827 Update insn attributes.
16828 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16829 Update insn attributes.
16830 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16831 alternative 1 (r, v). Remove isa attribute.
16832 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16833 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16834 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16836 2017-05-23 Tom de Vries <tom@codesourcery.com>
16838 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16841 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16843 * cgraphunit.c (symbol_table::process_new_functions): Update.
16844 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16845 (inline_generate_summary): Rename to ...
16846 (ipa_fn_summary_generate): ... this one.
16847 (inline_read_summary): Rename to ...
16848 (ipa_fn_summary_read): ... this one.
16849 (inline_write_summary): Rename to ...
16850 (ipa_fn_summary_write): ... this one.
16851 (inline_free_summary): Rename to ...
16852 (ipa_free_fn_summary): ... this one.
16853 (pass_data_local_fn_summary, pass_local_fn_summary,
16854 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16855 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16856 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16857 make_pass_ipa_fn_summary): New.
16858 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16859 inline_write_summary, inline_free_summary): Remove.
16860 (ipa_free_fn_summary) : New.
16861 * ipa-inline.c (ipa_inline): Update.
16862 (pass_ipa_inline): Do not generate summaries.
16863 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16865 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16866 and add pass_ipa_fn_summary.
16867 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16869 (make_pass_inline_parameters): Remove.
16871 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16873 * omp-low.c (struct omp_context): Remove "default_kind" member.
16876 * omp-offload.c (execute_oacc_device_lower): Remove the
16877 parallelism dimensions function attributes for unparallelized
16878 OpenACC kernels constructs.
16880 2017-05-23 Martin Liska <mliska@suse.cz>
16882 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16884 (cgraph_edge::make_speculative): Likewise.
16885 (cgraph_edge::resolve_speculation): Likewise.
16886 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16887 (cgraph_node::dump): Likewise.
16888 * cgraph.h: Likewise.
16889 * cgraphunit.c (analyze_functions): Likewise.
16890 (symbol_table::compile): Likewise.
16891 * ipa-cp.c (print_all_lattices): Likewise.
16892 (determine_versionability): Likewise.
16893 (initialize_node_lattices): Likewise.
16894 (ipcp_verify_propagated_values): Likewise.
16895 (estimate_local_effects): Likewise.
16896 (update_profiling_info): Likewise.
16897 (create_specialized_node): Likewise.
16898 (perhaps_add_new_callers): Likewise.
16899 (decide_about_value): Likewise.
16900 (decide_whether_version_node): Likewise.
16901 (identify_dead_nodes): Likewise.
16902 (ipcp_store_bits_results): Likewise.
16903 * ipa-devirt.c (dump_targets): Likewise.
16904 (ipa_devirt): Likewise.
16905 * ipa-icf.c (sem_item::dump): Likewise.
16906 (sem_function::equals): Likewise.
16907 (sem_variable::equals): Likewise.
16908 (sem_item_optimizer::read_section): Likewise.
16909 (sem_item_optimizer::execute): Likewise.
16910 (congruence_class::dump): Likewise.
16911 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16912 (dump_inline_summary): Likewise.
16913 (estimate_node_size_and_time): Likewise.
16914 (inline_analyze_function): Likewise.
16915 * ipa-inline-transform.c (inline_call): Likewise.
16916 * ipa-inline.c (report_inline_failed_reason): Likewise.
16917 (want_early_inline_function_p): Likewise.
16918 (edge_badness): Likewise.
16919 (update_edge_key): Likewise.
16920 (inline_small_functions): Likewise.
16921 * ipa-profile.c (ipa_profile): Likewise.
16922 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16923 (ipa_make_edge_direct_to_target): Likewise.
16924 (remove_described_reference): Likewise.
16925 (ipa_impossible_devirt_target): Likewise.
16926 (propagate_controlled_uses): Likewise.
16927 (ipa_print_node_params): Likewise.
16928 (ipcp_transform_function): Likewise.
16929 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16930 (propagate_pure_const): Likewise.
16931 * ipa-reference.c (generate_summary): Likewise.
16932 (read_write_all_from_decl): Likewise.
16933 (propagate): Likewise.
16934 (ipa_reference_read_optimization_summary): Likewise.
16935 * ipa-utils.c (ipa_merge_profiles): Likewise.
16936 * ipa.c (walk_polymorphic_call_targets): Likewise.
16937 (symbol_table::remove_unreachable_nodes): Likewise.
16938 (ipa_single_use): Likewise.
16939 * passes.c (execute_todo): Likewise.
16940 * predict.c (drop_profile): Likewise.
16941 * symtab.c (symtab_node::get_dump_name): New function.
16942 (symtab_node::dump_name): Likewise.
16943 (symtab_node::dump_asm_name): Likewise.
16944 (symtab_node::dump_references): Likewise.
16945 (symtab_node::dump_referring): Likewise.
16946 (symtab_node::dump_base): Likewise.
16947 (symtab_node::debug_symtab): Likewise.
16948 * tree-sra.c (convert_callers_for_node): Likewise.
16949 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16950 * value-prof.c (init_node_map): Likewise.
16952 2017-05-23 Martin Liska <mliska@suse.cz>
16954 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16955 and symtab_node::debug_symtab to symbol_table::debug.
16956 * cgraphunit.c (analyze_functions): Use the renamed function.
16957 (symbol_table::compile): Likewise.
16958 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16959 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16960 * passes.c (execute_todo): Likewise.
16961 * symtab.c (symbol_table::dump): New function.
16962 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16964 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16966 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16967 that nonconst implies exec.
16969 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16971 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16972 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16973 (inline_edge_summary_vec): Turn into ...
16974 (ipa_call_summaries): ... this one.
16975 (redirect_to_unreachable, edge_set_predicate,
16976 evaluate_properties_for_edge, inline_summary_alloc,
16977 reset_ipa_call_summary, reset_inline_summary,
16978 inline_summary_t::duplicate): Update.
16979 (inline_edge_duplication_hook): Turn to ...
16980 (ipa_call_summary_t::duplicate): ... this one.
16981 (inline_edge_removal_hook): Turn to ...
16982 (ipa_call_summary_t::remove): ... this one.
16983 (dump_inline_edge_summary): Turn to ...
16984 (dump_ipa_call_summary): ... this one.
16985 (estimate_function_body_sizes): Update.
16986 (inline_update_callee_summaries): Update.
16987 (remap_edge_change_prob): Update.
16988 (remap_edge_summaries): Update.
16989 (inline_merge_summary): Update.
16990 (do_estimate_edge_time): Update.
16991 (inline_generate_summary): Update.
16992 (inline_read_section): Update.
16993 (inline_read_summary): Update.
16994 (inline_free_summary): Update.
16995 * ipa-inline.c (can_inline_edge_p): Update.
16996 (compute_inlined_call_time): Update.
16997 (want_inline_small_function_p): Update.
16998 (edge_badness): Update.
16999 (early_inliner): Update.
17000 * ipa-inline.h (inline_edge_summary): Turn to ...
17001 (ipa_call_summary): ... this one.
17002 (ipa_call_summary_t): New class.
17003 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17004 (ipa_call_summaries): New.
17005 (inline_edge_summary): Remove.
17006 (estimate_edge_growth): Update.
17007 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17008 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17009 * ipa-split.c (execute_split_functions): Update.
17010 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17012 2017-05-23 Tom de Vries <tom@codesourcery.com>
17014 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17015 attributes): Document rdrand effective target.
17017 2017-05-23 Tom de Vries <tom@codesourcery.com>
17019 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17020 attributes): Sort alphabetically.
17022 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17024 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17026 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17029 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17030 V2DF/V2DI splat into two separate patterns, one that handles
17031 registers, and the other that only handles memory. Drop support
17032 for splatting from a GPR on ISA 2.07 and then splitting the
17033 splat into direct move and splat.
17034 (vsx_splat_<mode>_reg): Likewise.
17035 (vsx_splat_<mode>_mem): Likewise.
17037 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17039 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17041 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17043 PR middle-end/80809
17044 * omp-low.c (finish_taskreg_remap): New function.
17045 (finish_taskreg_scan): If unit size of ctx->record_type
17046 is non-constant, unshare the size expression and replace
17047 decls in it with possible outer var refs.
17049 PR middle-end/80809
17050 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17051 GOVD_SHARED rather than GOVD_PRIVATE with it.
17052 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17053 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17055 PR middle-end/80853
17056 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17057 as last argument to build_outer_var_ref for pointer bases of array
17058 section reductions.
17060 2017-05-19 Martin Sebor <msebor@redhat.com>
17062 * print-tree.c (print_node): Print DECL_READ_P flag.
17064 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17066 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17067 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17068 * cgraph.c: Likewise.
17069 * cgraphunit.c: Likewise.
17070 * gengtype.c: Likewise.
17071 * ipa-cp.c: Likewise.
17072 * ipa-devirt.c: Likewise.
17073 * ipa-icf.c: Likewise.
17074 * ipa-predicate.c: Likewise.
17075 * ipa-profile.c: Likewise.
17076 * ipa-prop.c: Likewise.
17077 * ipa-split.c: Likewise.
17079 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17080 edge_predicate_pool, dump_inline_hints,
17081 inline_summary::account_size_time, redirect_to_unreachable,
17082 edge_set_predicate, set_hint_predicate,
17083 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17084 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17085 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17086 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17087 ipa_call_summary_t::remove, initialize_growth_caches,
17088 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17089 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17090 mark_modified, unmodified_parm_1, unmodified_parm,
17091 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17092 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17093 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17094 will_be_nonconstant_predicate, record_modified_bb_info,
17095 get_minimal_bb, record_modified, param_change_prob,
17096 phi_result_unknown_predicate, predicate_for_phi_result,
17097 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17098 estimate_function_body_sizes, compute_inline_parameters,
17099 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17100 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17101 inline_update_callee_summaries, remap_edge_change_prob,
17102 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17103 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17104 inline_analyze_function, inline_summary_t::insert,
17105 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17106 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17107 inline_free_summary): Move to ipa-fnsummary.h
17108 (predicate_t): Remove.
17109 * ipa-fnsummary.c: New file.
17110 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17111 (enum inline_hints_vals, inline_hints, agg_position_info,
17112 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17113 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17114 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17115 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17116 inline_read_summary, inline_write_summary, inline_free_summary,
17117 inline_analyze_function, initialize_inline_failed,
17118 inline_merge_summary, inline_update_overall_summary,
17119 compute_inline_parameters): Move to ipa-fnsummary.h
17120 * ipa-fnsummary.h: New file.
17121 * ipa-inline-transform.h: Include ipa-inline.h.
17122 * ipa-inline.c: LIkewise.
17124 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17126 * ipa-inline.c (edge_badness): Use inlined_time instead of
17127 inline_summaries->get.
17129 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17131 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17133 2017-05-22 Nathan Sidwell <nathan@acm.org>
17135 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17136 (fdump-lang): Document 'raw' option.
17137 * dumpfile.h (TDI_tu): Delete.
17138 * dumpfile.c (dump_files): Remove translation-unit.
17139 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17141 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17143 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17144 command option from $(AWK) call.
17145 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17146 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17147 [FORMAT]: Remove handling of variable.
17148 * config/avr/t-multilib: Regenerate.
17150 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17152 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17154 (dump_inline_summary): Do not print self_time.
17155 (estimate_function_body_sizes): Do not set self_time.
17156 (compute_inline_parameters): Likewise.
17157 (inline_read_section, inline_write_summary): Do not stream self_time.
17158 * ipa-inline.h (inline_summary): Drop self_time.
17160 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17162 * ipa-inline-analysis.c (account_size_time): Rename to ...
17163 (inline_summary::account_size_time): ... this one.
17164 (reset_ipa_call_summary): Turn to ...
17165 (ipa_call_summary::reset): ... this one.
17166 (reset_inline_summary): Turn to ...
17167 (inline_summary::reset): ... this one.
17168 (inline_summary_t::remove): Update.
17169 (inline_summary_t::duplicate): Update.
17170 (ipa_call_summary_t::remove): Update.
17171 (dump_inline_summary): Update.
17172 (estimate_function_body_sizes): Update.
17173 (compute_inline_parameters): Update.
17174 (estimate_node_size_and_time): Update.
17175 (inline_merge_summary): Update.
17176 (inline_update_overall_summary): Update.
17177 (inline_read_section): Update.
17178 (inline_write_summary): Update.
17179 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17180 add account_size_time and reset member functions.
17181 (ipa_call_summary): Add reset function.
17182 * ipa-predicate.h (predicate::operator &): Constify.
17184 2017-05-22 Richard Biener <rguenther@suse.de>
17186 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17188 2017-05-19 Jason Merrill <jason@redhat.com>
17190 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17192 2017-05-19 Marek Polacek <polacek@redhat.com>
17195 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17196 TYPE_OVERFLOW_WRAPS checks.
17198 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17200 * tree-core.h (enum omp_clause_default_kind): Add
17201 "OMP_CLAUSE_DEFAULT_PRESENT".
17202 * tree-pretty-print.c (dump_omp_clause): Handle it.
17203 * gimplify.c (enum gimplify_omp_var_data): Add
17204 "GOVD_MAP_FORCE_PRESENT".
17205 (gimplify_adjust_omp_clauses_1): Map it to
17206 "GOMP_MAP_FORCE_PRESENT".
17207 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17209 * gimplify.c (oacc_default_clause): Clarify.
17211 2017-05-19 Nathan Sidwell <nathan@acm.org>
17213 LANG_HOOK_REGISTER_DUMPS
17214 * toplev.c (general_init): Call register dump lang hook.
17215 * doc/invoke.texi: Document -fdump-lang option family.
17216 * dumpfile.c (dump_files): Remove class dump here.
17217 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17218 * dumpfile.h (tree_dump_index): Remove TDI_class.
17219 * langhooks-def.h (lhd_register_dumps): Declare.
17220 (LANG_HOOKS_REGISTER_DUMPS): Define.
17221 (LANG_HOOKS_INITIALIZER): Add it.
17222 * langhooks.c (lhd_register_dumps): Define.
17223 * langhooks.h (struct lang_hooks): Add register_dumps.
17225 2017-05-19 Nathan Sidwell <nathan@acm.org>
17227 * context.h (context::set_passes): New.
17228 * context.c (context::context): Do not create pass manager.
17229 * toplev.c (general_init): Create pass manager here.
17231 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17233 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17234 use this splitter if two add or or instructions would also work for
17235 the constant we want to generate.
17237 2017-05-19 Richard Biener <rguenther@suse.de>
17240 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17241 predicate evaluation.
17243 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17245 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17247 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17250 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17252 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17253 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17254 (inline_edge_summary_vec): Turn into ...
17255 (ipa_call_summaries): ... this one.
17256 (redirect_to_unreachable, edge_set_predicate,
17257 evaluate_properties_for_edge, inline_summary_alloc,
17258 reset_ipa_call_summary, reset_inline_summary,
17259 inline_summary_t::duplicate): Update.
17260 (inline_edge_duplication_hook): Turn to ...
17261 (ipa_call_summary_t::duplicate): ... this one.
17262 (inline_edge_removal_hook): Turn to ...
17263 (ipa_call_summary_t::remove): ... this one.
17264 (dump_inline_edge_summary): Turn to ...
17265 (dump_ipa_call_summary): ... this one.
17266 (estimate_function_body_sizes): Update.
17267 (inline_update_callee_summaries): Update.
17268 (remap_edge_change_prob): Update.
17269 (remap_edge_summaries): Update.
17270 (inline_merge_summary): Update.
17271 (do_estimate_edge_time): Update.
17272 (inline_generate_summary): Update.
17273 (inline_read_section): Update.
17274 (inline_read_summary): Update.
17275 (inline_free_summary): Update.
17276 * ipa-inline.c (can_inline_edge_p): Update.
17277 (compute_inlined_call_time): Update.
17278 (want_inline_small_function_p): Update.
17279 (edge_badness): Update.
17280 (early_inliner): Update.
17281 * ipa-inline.h (inline_edge_summary): Turn to ...
17282 (ipa_call_summary): ... this one.
17283 (ipa_call_summary_t): New class.
17284 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17285 (ipa_call_summaries): New.
17286 (inline_edge_summary): Remove.
17287 (estimate_edge_growth): Update.
17288 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17289 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17290 * ipa-split.c (execute_split_functions): Update.
17291 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17293 2017-05-19 Richard Biener <rguenther@suse.de>
17295 PR middle-end/80764
17296 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17298 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17300 * config/rs6000/rs6000.c (struct machine_function): Add field
17301 fpr_is_wrapped_separately.
17302 (rs6000_get_separate_components): Use 64 components. Handle the
17303 new FPR components.
17304 (rs6000_components_for_bb): Handle the FPR components.
17305 (rs6000_emit_prologue_components): Handle the FPR components.
17306 (rs6000_emit_epilogue_components): Handle the FPR components.
17307 (rs6000_set_handled_components): Handle the FPR components.
17308 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17309 that are already separately shrink-wrapped.
17310 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17311 that are already separately shrink-wrapped.
17313 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17316 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17319 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17320 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17321 cases where the register allocator uses a move and an offsettable
17322 memory operation to/from a FPR register on ISA 2.06/2.07.
17323 (define_peephole2 for Altivec d-form store): Likewise.
17325 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17328 * config/i386/mmx.md (*mov<mode>_internal): Enable
17329 alternatives 11, 12, 13 and 14 also for 32bit targets.
17330 Remove alternatives 15, 16, 17 and 18.
17331 * config/i386/sse.md (vec_concatv2di): Change
17332 alternative (!x, *y) to (x, ?!*Yn).
17334 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17336 * dumpfile.h (enum dump_kind): Remove stray comma.
17338 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17340 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17341 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17342 predicate::num_conditions
17343 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17344 (CHANGED): turn into predicate::changed.
17345 (agg_position_info): Move to ipa-predicate.h
17346 (add_condition, predicate::add_clause, predicate::operator &=,
17347 predicate::or_with, predicate::evaluate, predicate::probability,
17348 dump_condition, dump_clause, predicate::dump,
17349 predicate::remap_after_duplication, predicate::remap_after_inlining,
17350 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17351 (evaluate_conditions_for_known_args): Update.
17352 (set_cond_stmt_execution_predicate): Update.
17353 * ipa-inline.h: Include ipa-predicate.h
17354 (condition, inline_param_summary, conditions, agg_position_info,
17355 predicate): Move to ipa-predicate.h
17356 * ipa-predicate.c: New file.
17357 * ipa-predicate.h: New file.
17359 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17361 * final.c (leaf_function_p): Check we are not in a sequence.
17363 2017-05-18 Martin Liska <mliska@suse.cz>
17365 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17366 * dumpfile.c (dump_register): Use new enum dump_kind.
17367 (get_dump_file_name): Likewise.
17368 (dump_enable_all): Likewise.
17369 (dump_switch_p_1): Likewise.
17370 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17371 * dumpfile.h (enum dump_kind): New enum type.
17372 (struct dump_file_info): Create constructor and
17373 format fields and comments.
17374 * passes.c (pass_manager::register_one_dump_file):
17376 * statistics.c (statistics_early_init): Likewise.
17377 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17378 TDF_TREE with TDF_SLIM.
17379 (gather_memory_references_ref): Likewise.
17381 2017-05-18 Martin Liska <mliska@suse.cz>
17383 * vec.h (struct vnull): Use it.
17385 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17387 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17388 (true_predicate, false_predicate, true_predicate_p,
17389 false_predicate_p): Remove.
17390 (single_cond_predicate, not_inlined_predicate): Turn to member function
17392 (add_condition): Update.
17393 (add_clause): Turn to...
17394 (predicate::add_clause): ... this one; update; allow passing NULL
17396 (and_predicates): Turn to ...
17397 (predicate::operator &=): ... this one.
17398 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17399 (or_predicates): Turn to ...
17400 (predicate::or_with): ... this one.
17401 (evaluate_predicate): Turn to ...
17402 (predicate::evaluate): ... this one.
17403 (predicate_probability): Turn to ...
17404 (predicate::probability): ... this one.
17405 (dump_condition): Update.
17406 (dump_predicate): Turn to ...
17407 (predicate::dump): ... this one.
17408 (account_size_time): Update.
17409 (edge_set_predicate): Update.
17410 (set_hint_predicate): UPdate.
17411 (evaluate_conditions_for_known_args): Update.
17412 (evaluate_properties_for_edge): Update.
17413 (remap_predicate_after_duplication): Turn to...
17414 (predicate::remap_after_duplication): ... this one.
17415 (remap_hint_predicate_after_duplication): Update.
17416 (inline_summary_t::duplicate): UPdate.
17417 (dump_inline_edge_summary): Update.
17418 (dump_inline_summary): Update.
17419 (set_cond_stmt_execution_predicate): Update.
17420 (set_switch_stmt_execution_predicate): Update.
17421 (compute_bb_predicates): Update.
17422 (will_be_nonconstant_expr_predicate): Update.
17423 (will_be_nonconstant_predicate): Update.
17424 (phi_result_unknown_predicate): Update.
17425 (predicate_for_phi_result): Update.
17426 (array_index_predicate): Update.
17427 (estimate_function_body_sizes): Update.
17428 (estimate_node_size_and_time): Update.
17429 (estimate_ipcp_clone_size_and_time): Update.
17430 (remap_predicate): Rename to ...
17431 (predicate::remap_after_inlining): ... this one.
17432 (remap_hint_predicate): Update.
17433 (inline_merge_summary): Update.
17434 (inline_update_overall_summary): Update.
17435 (estimate_size_after_inlining): Update.
17436 (read_predicate): Rename to ...
17437 (predicate::stream_in): ... this one.
17438 (read_inline_edge_summary): Update.
17439 (write_predicate): Rename to ...
17440 (predicate::stream_out): ... this one.
17441 (write_inline_edge_summary): Update.
17442 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17443 (clause_t): Turn to uint32_t
17444 (predicate): Turn to class; implement constructor and operators
17446 (size_time_entry): Update.
17447 (inline_summary): Update.
17448 (inline_edge_summary): Update.
17450 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17452 * fold-const.c (fold_binary_loc): Move transformation...
17453 * match.pd (C - X CMP X): ... here.
17455 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17457 * config/sparc/sparc.c (sparc_option_override): Set function
17458 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17459 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17461 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17463 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17465 2017-05-18 Marek Polacek <polacek@redhat.com>
17468 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17469 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17470 gimple_assign_load_p.
17472 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17474 PR middle-end/80692
17475 * real.c (do_compare): Give decimal_do_compare preference over
17476 comparing just the signs.
17478 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17480 * doc/md.texi (Canonicalization of Instructions): Describe the
17481 canonical form of instructions that inherently set a condition
17484 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17486 PR middle-end/80775
17487 * tree-cfg.c: Move deletion of unreachable case statements to after
17488 the merging of consecutive case labels.
17490 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17492 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17493 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17494 restoring of callee-saved registers.
17496 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17498 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17499 * config/visium/visium.c (single_set_and_flags): Likewise.
17500 * config/visium/visium.md (Substitutions): Likewise.
17502 2017-05-17 Martin Liska <mliska@suse.cz>
17504 * cfg.c: Introduce dump_flags_t type and
17505 use it instead of int type.
17507 * cfghooks.c: Likewise.
17508 * cfghooks.h (struct cfg_hooks): Likewise.
17509 * cfgrtl.c: Likewise.
17510 * cfgrtl.h: Likewise.
17511 * cgraph.c (cgraph_node::get_body): Likewise.
17512 * coretypes.h: Likewise.
17513 * domwalk.c: Likewise.
17514 * domwalk.h: Likewise.
17515 * dumpfile.c (struct dump_option_value_info): Likewise.
17516 (dump_enable_all): Likewise.
17517 (dump_switch_p_1): Likewise.
17518 (opt_info_switch_p): Likewise.
17519 * dumpfile.h (enum tree_dump_index): Likewise.
17520 (struct dump_file_info): Likewise.
17521 * genemit.c: Likewise.
17522 * generic-match-head.c: Likewise.
17523 * gengtype.c (open_base_files): Likewise.
17524 * gimple-pretty-print.c: Likewise.
17525 * gimple-pretty-print.h: Likewise.
17526 * graph.c (print_graph_cfg): Likewise.
17527 * graphite-scop-detection.c (dot_all_sese): Likewise.
17528 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17529 * loop-unroll.c (report_unroll): Likewise.
17530 * passes.c (pass_manager::register_one_dump_file): Likewise.
17531 * print-tree.c: Likewise.
17532 * statistics.c: Likewise.
17533 * tree-cfg.c: Likewise.
17534 * tree-cfg.h: Likewise.
17535 * tree-dfa.c: Likewise.
17536 * tree-dfa.h: Likewise.
17537 * tree-dump.c (dump_function): Likewise.
17538 * tree-dump.h (struct dump_info): Likewise.
17539 * tree-pretty-print.c: Likewise.
17540 * tree-pretty-print.h: Likewise.
17541 * tree-ssa-live.c: Likewise.
17542 * tree-ssa-live.h: Likewise.
17543 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17544 * tree-vect-loop.c: Likewise.
17545 * tree-vect-slp.c: Likewise.
17547 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17548 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17550 PR tree-optimization/80457
17551 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17552 of all arguments to a statement as scalar_to_vec operations.
17553 (vectorizable_call): Adjust call to vect_model_simple_cost for
17555 (vectorizable_conversion): Likewise.
17556 (vectorizable_assignment): Likewise.
17557 (vectorizable_shift): Likewise.
17558 (vectorizable_operation): Likewise.
17559 (vectorizable_comparison): Likewise.
17560 (vect_is_simple_cond): Record the def types for operands.
17561 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17562 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17563 for statement argument count.
17565 2017-05-16 Carl Love <cel@us.ibm.com>
17567 * config/rs6000/rs6000-c: Add support for built-in functions
17568 vector unsigned long long vec_bperm (vector unsigned long long,
17569 vector unsigned char)
17570 vector signed long long vec_mule (vector signed int,
17572 vector unsigned long long vec_mule (vector unsigned int,
17573 vector unsigned int)
17574 vector signed long long vec_mulo (vector signed int,
17576 vector unsigned long long vec_mulo (vector unsigned int,
17577 vector unsigned int)
17578 vector signed char vec_sldw (vector signed char,
17579 vector signed char,
17581 vector unsigned char vec_sldw (vector unsigned char,
17582 vector unsigned char,
17584 vector signed short vec_sldw (vector signed short,
17585 vector signed short,
17587 vector unsigned short vec_sldw (vector unsigned short,
17588 vector unsigned short,
17590 vector signed int vec_sldw (vector signed int,
17593 vector unsigned int vec_sldw (vector unsigned int,
17594 vector unsigned int,
17596 vector signed long long vec_sldw (vector signed long long,
17597 vector signed long long,
17599 vector unsigned long long vec_sldw (vector unsigned long long,
17600 vector unsigned long long,
17602 * config/rs6000/rs6000-c: Add support for built-in functions
17603 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17604 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17605 * doc/extend.texi: Update the built-in documentation for the
17606 new built-in functions.
17608 2017-05-16 Marek Polacek <polacek@redhat.com>
17612 * tree.c (save_expr): Don't fold the expression.
17614 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17616 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17617 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17618 and (?*y,m). Update insn attributes.
17620 2017-05-16 Martin Liska <mliska@suse.cz>
17622 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17623 flags argument of print_gimple_stmt, print_gimple_expr,
17624 print_generic_stmt and print_generic_expr.
17625 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17626 * coretypes.h: Likewise.
17627 * except.c (dump_eh_tree): Likewise.
17628 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17629 * gimple-pretty-print.h: Likewise.
17630 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17631 (backprop::push_to_worklist): Likewise.
17632 (backprop::pop_from_worklist): Likewise.
17633 (backprop::process_use): Likewise.
17634 (backprop::intersect_uses): Likewise.
17635 (note_replacement): Likewise.
17636 * gimple-ssa-store-merging.c
17637 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17638 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17639 (pass_store_merging::execute): Likewise.
17640 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17641 (ssa_base_cand_dump_callback): Likewise.
17642 (dump_incr_vec): Likewise.
17643 (replace_refs): Likewise.
17644 (replace_mult_candidate): Likewise.
17645 (create_add_on_incoming_edge): Likewise.
17646 (create_phi_basis): Likewise.
17647 (insert_initializers): Likewise.
17648 (all_phi_incrs_profitable): Likewise.
17649 (introduce_cast_before_cand): Likewise.
17650 (replace_one_candidate): Likewise.
17651 * gimplify.c (gimplify_expr): Likewise.
17652 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17653 (set_rename): Likewise.
17654 (rename_uses): Likewise.
17655 (copy_loop_phi_nodes): Likewise.
17656 (add_close_phis_to_merge_points): Likewise.
17657 (copy_loop_close_phi_args): Likewise.
17658 (copy_cond_phi_args): Likewise.
17659 (graphite_copy_stmts_from_block): Likewise.
17660 (translate_pending_phi_nodes): Likewise.
17661 * graphite-poly.c (print_pdr): Likewise.
17662 (dump_gbb_cases): Likewise.
17663 (dump_gbb_conditions): Likewise.
17664 (print_scop_params): Likewise.
17665 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17666 (build_cross_bb_scalars_use): Likewise.
17667 (gather_bbs::before_dom_children): Likewise.
17668 * hsa-dump.c (dump_hsa_immed): Likewise.
17669 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17670 (get_replacement_map): Likewise.
17671 * ipa-inline-analysis.c (dump_condition): Likewise.
17672 (estimate_function_body_sizes): Likewise.
17673 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17674 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17675 * ipa-prop.c (ipa_dump_param): Likewise.
17676 (ipa_print_node_jump_functions_for_edge): Likewise.
17677 (ipa_modify_call_arguments): Likewise.
17678 (ipa_modify_expr): Likewise.
17679 (ipa_dump_param_adjustments): Likewise.
17680 (ipa_dump_agg_replacement_values): Likewise.
17681 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17682 * ipa-pure-const.c (check_stmt): Likewise.
17683 (pass_nothrow::execute): Likewise.
17684 * ipa-split.c (execute_split_functions): Likewise.
17685 * omp-offload.c (dump_oacc_loop_part): Likewise.
17686 (dump_oacc_loop): Likewise.
17687 * trans-mem.c (tm_log_emit): Likewise.
17688 (tm_memopt_accumulate_memops): Likewise.
17689 (dump_tm_memopt_set): Likewise.
17690 (dump_tm_memopt_transform): Likewise.
17691 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17692 (print_loop): Likewise.
17693 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17694 (chkp_gather_checks_info): Likewise.
17695 (chkp_get_check_result): Likewise.
17696 (chkp_remove_check_if_pass): Likewise.
17697 (chkp_use_outer_bounds_if_possible): Likewise.
17698 (chkp_reduce_bounds_lifetime): Likewise.
17699 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17700 (chkp_mark_completed_bounds): Likewise.
17701 (chkp_register_incomplete_bounds): Likewise.
17702 (chkp_mark_invalid_bounds): Likewise.
17703 (chkp_maybe_copy_and_register_bounds): Likewise.
17704 (chkp_build_returned_bound): Likewise.
17705 (chkp_get_bound_for_parm): Likewise.
17706 (chkp_build_bndldx): Likewise.
17707 (chkp_get_bounds_by_definition): Likewise.
17708 (chkp_generate_extern_var_bounds): Likewise.
17709 (chkp_get_bounds_for_decl_addr): Likewise.
17710 * tree-chrec.c (chrec_apply): Likewise.
17711 * tree-data-ref.c (dump_data_reference): Likewise.
17712 (dump_subscript): Likewise.
17713 (dump_data_dependence_relation): Likewise.
17714 (analyze_overlapping_iterations): Likewise.
17715 * tree-inline.c (expand_call_inline): Likewise.
17716 (tree_function_versioning): Likewise.
17717 * tree-into-ssa.c (dump_defs_stack): Likewise.
17718 (dump_currdefs): Likewise.
17719 (dump_names_replaced_by): Likewise.
17720 (dump_update_ssa): Likewise.
17721 (update_ssa): Likewise.
17722 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17723 * tree-parloops.c (build_new_reduction): Likewise.
17724 (try_create_reduction_list): Likewise.
17725 (ref_conflicts_with_region): Likewise.
17726 (oacc_entry_exit_ok_1): Likewise.
17727 (oacc_entry_exit_single_gang): Likewise.
17728 * tree-pretty-print.h: Likewise.
17729 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17730 (get_scalar_evolution): Likewise.
17731 (add_to_evolution): Likewise.
17732 (get_loop_exit_condition): Likewise.
17733 (analyze_evolution_in_loop): Likewise.
17734 (analyze_initial_condition): Likewise.
17735 (analyze_scalar_evolution): Likewise.
17736 (instantiate_scev): Likewise.
17737 (number_of_latch_executions): Likewise.
17738 (gather_chrec_stats): Likewise.
17739 (final_value_replacement_loop): Likewise.
17740 (scev_const_prop): Likewise.
17741 * tree-sra.c (dump_access): Likewise.
17742 (disqualify_candidate): Likewise.
17743 (create_access): Likewise.
17744 (reject): Likewise.
17745 (maybe_add_sra_candidate): Likewise.
17746 (create_access_replacement): Likewise.
17747 (analyze_access_subtree): Likewise.
17748 (analyze_all_variable_accesses): Likewise.
17749 (sra_modify_assign): Likewise.
17750 (initialize_constant_pool_replacements): Likewise.
17751 (find_param_candidates): Likewise.
17752 (decide_one_param_reduction): Likewise.
17753 (replace_removed_params_ssa_names): Likewise.
17754 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17755 * tree-ssa-copy.c (dump_copy_of): Likewise.
17756 (copy_prop_visit_cond_stmt): Likewise.
17757 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17758 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17759 (record_equivalences_from_stmt): Likewise.
17760 * tree-ssa-dse.c (compute_trims): Likewise.
17761 (delete_dead_call): Likewise.
17762 (delete_dead_assignment): Likewise.
17763 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17764 (forward_propagate_into_cond): Likewise.
17765 (pass_forwprop::execute): Likewise.
17766 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17767 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17769 (move_computations_worker): Likewise.
17770 (execute_sm): Likewise.
17771 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17772 (remove_exits_and_undefined_stmts): Likewise.
17773 (remove_redundant_iv_tests): Likewise.
17774 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17775 (adjust_iv_update_pos): Likewise.
17776 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17777 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17778 (value_replacement): Likewise.
17779 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17780 * tree-ssa-pre.c (print_pre_expr): Likewise.
17781 (get_representative_for): Likewise.
17782 (create_expression_by_pieces): Likewise.
17783 (insert_into_preds_of_block): Likewise.
17784 (eliminate_insert): Likewise.
17785 (eliminate_dom_walker::before_dom_children): Likewise.
17786 (eliminate): Likewise.
17787 (remove_dead_inserted_code): Likewise.
17788 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17789 * tree-ssa-reassoc.c (get_rank): Likewise.
17790 (eliminate_duplicate_pair): Likewise.
17791 (eliminate_plus_minus_pair): Likewise.
17792 (eliminate_not_pairs): Likewise.
17793 (undistribute_ops_list): Likewise.
17794 (eliminate_redundant_comparison): Likewise.
17795 (update_range_test): Likewise.
17796 (optimize_range_tests_var_bound): Likewise.
17797 (optimize_vec_cond_expr): Likewise.
17798 (rewrite_expr_tree): Likewise.
17799 (rewrite_expr_tree_parallel): Likewise.
17800 (linearize_expr): Likewise.
17801 (break_up_subtract): Likewise.
17802 (linearize_expr_tree): Likewise.
17803 (attempt_builtin_powi): Likewise.
17804 (attempt_builtin_copysign): Likewise.
17805 (transform_stmt_to_copy): Likewise.
17806 (transform_stmt_to_multiply): Likewise.
17807 (dump_ops_vector): Likewise.
17808 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17809 (print_scc): Likewise.
17810 (set_ssa_val_to): Likewise.
17811 (visit_reference_op_store): Likewise.
17812 (visit_use): Likewise.
17813 (sccvn_dom_walker::before_dom_children): Likewise.
17814 (run_scc_vn): Likewise.
17815 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17817 (expr_hash_elt::print): Likewise.
17818 (const_and_copies::pop_to_marker): Likewise.
17819 (const_and_copies::record_const_or_copy_raw): Likewise.
17820 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17821 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17822 (dump_predicates): Likewise.
17823 (find_uninit_use): Likewise.
17824 (warn_uninitialized_phi): Likewise.
17825 (pass_late_warn_uninitialized::execute): Likewise.
17826 * tree-ssa.c (verify_vssa): Likewise.
17827 (verify_ssa): Likewise.
17828 (maybe_optimize_var): Likewise.
17829 * tree-vrp.c (dump_value_range): Likewise.
17830 (dump_all_value_ranges): Likewise.
17831 (dump_asserts_for): Likewise.
17832 (register_edge_assert_for_2): Likewise.
17833 (vrp_visit_cond_stmt): Likewise.
17834 (vrp_visit_switch_stmt): Likewise.
17835 (vrp_visit_stmt): Likewise.
17836 (vrp_visit_phi_node): Likewise.
17837 (simplify_cond_using_ranges_1): Likewise.
17838 (fold_predicate_in): Likewise.
17839 (evrp_dom_walker::before_dom_children): Likewise.
17840 (evrp_dom_walker::push_value_range): Likewise.
17841 (evrp_dom_walker::pop_value_range): Likewise.
17842 (execute_early_vrp): Likewise.
17844 2017-05-16 Richard Biener <rguenther@suse.de>
17846 * dwarf2out.c (loc_list_from_tree_1): Do not create
17847 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17849 2017-05-16 Richard Biener <rguenther@suse.de>
17851 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17853 (note_variable_value_in_expr): If we resolved the decl ref
17854 do not push to the stack.
17856 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17858 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17859 operations in fast-math mode.
17860 (vaddq_f16): Likewise.
17861 (vmul_f16): Likewise.
17862 (vmulq_f16): Likewise.
17863 (vsub_f16): Likewise.
17864 (vsubq_f16): Likewise.
17865 * config/arm/neon.md (add<mode>3): New.
17867 (fma:<VH:mode>3): New. Also remove outdated comment.
17870 2017-05-16 Martin Liska <mliska@suse.cz>
17874 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17875 (odr_types_equivalent_p): Likewise.
17877 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17879 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17881 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17884 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17885 non-interunit SSE move alternatives with '?'.
17886 (zero-extendsidi peephole2): New peephole to skip intermediate
17887 general register in SSE zero-extend sequence.
17889 2017-05-15 Jeff Law <law@redhat.com>
17891 * reorg.c (relax_delay_slots): Create a new variable to hold
17892 the temporary target rather than clobbering TARGET_LABEL.
17894 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17895 missing argument to extract_bit_field call.
17896 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17898 2017-05-15 Martin Liska <mliska@suse.cz>
17901 * gcc.c (process_command): Do not allow empty argument of -o option.
17903 2017-05-15 Renlin Li <renlin.li@arm.com>
17905 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17906 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17907 * config/aarch64/constraints.md (Usf): Add long call check.
17908 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17909 (call_value): Likewise.
17910 (sibcall): Likewise.
17911 (sibcall_value): Likewise.
17913 (call_value_insn): New.
17914 (sibcall_insn): Update rtx pattern.
17915 (sibcall_value_insn): Likewise.
17916 (call_internal): Remove.
17917 (call_value_internal): Likewise.
17918 (sibcall_internal): Likewise.
17919 (sibcall_value_internal): Likewise.
17920 (call_reg): Likewise.
17921 (call_symbol): Likewise.
17922 (call_value_reg): Likewise.
17923 (call_value_symbol): Likewise.
17925 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17928 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17930 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17932 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17933 compatible with CCGOCmode and with CCZmode.
17935 2017-05-14 Martin Sebor <msebor@redhat.com>
17937 PR middle-end/77671
17938 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17939 (gimple_fold_builtin_snprintf): Same.
17940 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17941 (gimple_fold_builtin_snprintf): Same.
17942 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17943 of character types.
17944 (is_call_safe): New function.
17945 (try_substitute_return_value): Call it.
17946 (try_simplify_call): New function.
17947 (pass_sprintf_length::handle_gimple_call): Call it.
17949 2017-05-14 Martin Sebor <msebor@redhat.com>
17951 PR middle-end/80669
17952 * builtins.c (expand_builtin_stpncpy): Simplify.
17954 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17956 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17957 * config/i386/i386.h
17958 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17959 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17960 (struct machine_function): Add new members call_ms2sysv,
17961 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17962 (struct machine_frame_state): New fields sp_realigned and
17963 sp_realigned_offset.
17964 * config/i386/i386.c
17965 (enum xlogue_stub): New enum.
17966 (enum xlogue_stub_sets): New enum.
17967 (class xlogue_layout): New class.
17968 (struct ix86_frame): New fields stack_realign_allocate_offset,
17969 stack_realign_offset and outlined_save_offset. Modify comments to
17970 detail stack layout when using out-of-line stubs.
17971 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17972 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17973 -mcall-ms2sysv-xlogues.
17974 (stub_managed_regs): New static variable.
17975 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17976 registers managed by out-of-line stub.
17977 (disable_call_ms2sysv_xlogues): New function.
17978 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17979 m->call_ms2sysv when appropriate and compute frame layout for
17981 (sp_valid_at, fp_valid_at): New inline functions.
17982 (choose_basereg): New function.
17983 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17985 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17986 Use align parameter of choose_baseaddr to generated aligned SSE movs
17988 (pro_epilogue_adjust_stack): Modify to track
17989 machine_frame_state::sp_realigned.
17990 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17991 (ix86_nsaved_sseregs): Likewise.
17992 (ix86_emit_save_regs): Likewise.
17993 (ix86_emit_save_regs_using_mov): Likewise.
17994 (ix86_emit_save_sse_regs_using_mov): Likewise.
17995 (get_scratch_register_on_entry): Likewise.
17996 (gen_frame_set): New function.
17997 (gen_frame_load): Likewise.
17998 (gen_frame_store): Likewise.
17999 (emit_outlined_ms2sysv_save): Likewise.
18000 (emit_outlined_ms2sysv_restore): Likewise.
18001 (ix86_expand_prologue): Modify stack re-alignment code and call
18002 emit_outlined_ms2sysv_save when appropriate.
18003 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18004 parameter rtx_insn *insn, which allows the function to be used to only
18005 generate the notes.
18006 (ix86_expand_epilogue): Modify validity checks of frame and stack
18007 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18008 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18009 * config/i386/predicates.md
18010 (save_multiple): New predicate.
18011 (restore_multiple): Likewise.
18012 * config/i386/sse.md
18013 (save_multiple<mode>): New pattern.
18014 (save_multiple_realign<mode>): Likewise.
18015 (restore_multiple<mode>): Likewise.
18016 (restore_multiple_and_return<mode>): Likewise.
18017 (restore_multiple_leave_return<mode>): Likewise.
18018 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18020 2017-05-14 Julia Koval <julia.koval@intel.com>
18022 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18023 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18024 (__builtin_ia32_xsetbv): New builtins.
18025 * config/i386/i386.c (ix86_expand_special_args_builtin):
18027 (ix86_expand_builtin): Special expand for new intrinsics.
18028 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18029 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18030 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18032 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18034 * cfganal.c (inverted_post_order_compute): Change argument type
18036 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18037 * df-core.c (rest_of_handle_df_initialize): Adjust.
18038 (rest_of_handle_df_finish): Likewise.
18039 (df_analyze_1): Likewise.
18040 (df_analyze): Likewise.
18041 (loop_inverted_post_order_compute): Change argument to be a vec *.
18042 (df_analyze_loop): Adjust.
18043 (df_get_n_blocks): Likewise.
18044 (df_get_postorder): Likewise.
18045 * df.h (struct df_d): Change field to be a vec.
18046 * lcm.c (compute_laterin): Adjust.
18047 (compute_available): Likewise.
18048 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18049 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18050 * tree-ssa-pre.c (compute_antic): Likewise.
18052 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18054 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18055 (depth_first_search::depth_first_search): Change structure init
18056 function to this constructor.
18057 (depth_first_search::add_bb): Rename function to this member.
18058 (depth_first_search::execute): Likewise.
18059 (flow_dfs_compute_reverse_finish): Adjust.
18061 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18063 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18064 (longest_simple_path): Likewise.
18065 * shrink-wrap.c (spread_components): Likewise.
18066 (disqualify_problematic_components): Likewise.
18067 (emit_common_heads_for_components): Likewise.
18068 (emit_common_tails_for_components): Likewise.
18069 (insert_prologue_epilogue_for_components): Likewise.
18071 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18073 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18076 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18078 * df-core.c (df_set_blocks): Start using auto_bitmap.
18079 (df_compact_blocks): Likewise.
18080 * df-problems.c (df_rd_confluence_n): Likewise.
18081 * df-scan.c (df_insn_rescan_all): Likewise.
18082 (df_process_deferred_rescans): Likewise.
18083 (df_update_entry_block_defs): Likewise.
18084 (df_update_exit_block_uses): Likewise.
18085 (df_entry_block_bitmap_verify): Likewise.
18086 (df_exit_block_bitmap_verify): Likewise.
18087 (df_scan_verify): Likewise.
18088 * lra-constraints.c (lra_constraints): Likewise.
18089 (undo_optional_reloads): Likewise.
18090 (lra_undo_inheritance): Likewise.
18091 * lra-remat.c (calculate_gen_cands): Likewise.
18092 (do_remat): Likewise.
18093 * lra-spills.c (assign_spill_hard_regs): Likewise.
18094 (spill_pseudos): Likewise.
18095 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18096 (bitmap_set_subtract_values): Likewise.
18098 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18100 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18101 management with auto_bitmap.
18102 (fix_inter_tick): Likewise.
18103 (fix_recovery_deps): Likewise.
18104 * ira.c (add_store_equivs): Likewise.
18105 (find_moveable_pseudos): Likewise.
18106 (split_live_ranges_for_shrink_wrap): Likewise.
18107 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18108 (rtx_reuse_manager::seen_def_p): Likewise.
18109 (rtx_reuse_manager::set_seen_def): Likewise.
18110 * print-rtl.h (class rtx_reuse_manager): Likewise.
18112 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18114 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18116 (migrate_btr_def): Likewise.
18117 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18118 * df-core.c (loop_post_order_compute): Likewise.
18119 (loop_inverted_post_order_compute): Likewise.
18120 * hsa-common.h: Likewise.
18121 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18122 * init-regs.c (initialize_uninitialized_regs): Likewise.
18123 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18124 (inline_small_functions): Likewise.
18125 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18126 * ira.c (combine_and_move_insns): Likewise.
18127 (build_insn_chain): Likewise.
18128 * loop-invariant.c (find_invariants): Likewise.
18129 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18130 * predict.c (tree_predict_by_opcode): Likewise.
18131 (predict_paths_leading_to): Likewise.
18132 (predict_paths_leading_to_edge): Likewise.
18133 (estimate_loops_at_level): Likewise.
18134 (estimate_loops): Likewise.
18135 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18136 (spread_components): Likewise.
18137 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18138 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18139 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18140 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18141 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18142 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18143 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18144 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18145 (mark_threaded_blocks): Likewise.
18146 (thread_through_all_blocks): Likewise.
18147 * tree-ssa.c (verify_ssa): Likewise.
18148 (execute_update_addresses_taken): Likewise.
18149 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18151 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18153 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18155 (post_order_compute): Likewise.
18156 (inverted_post_order_compute): Likewise.
18157 (pre_and_rev_post_order_compute_fn): Likewise.
18159 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18161 * genrecog.c (int_set::int_set): Explicitly construct our
18162 auto_vec base class.
18163 * vec.h (auto_vec::auto_vec): New constructor.
18165 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18167 * bitmap.h (class auto_bitmap): New constructor taking
18168 bitmap_obstack * argument.
18170 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18172 * bitmap.h (class auto_bitmap): Change type of m_bits to
18173 bitmap_head, and adjust ctor / dtor and member operators.
18175 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18177 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18178 when returned register mode doesn't match original mode.
18180 2017-05-12 Jeff Law <law@redhat.com>
18181 Jakub Jelinek <jakub@redhat.com>
18183 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18184 we look for cc setter after the compare-elim changes.
18185 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18186 within the vector to match what compare-elim now expects.
18187 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18188 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18190 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18191 after the compare-elim changes.
18192 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18193 the vector to match what compare-elim now expects.
18194 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18195 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18196 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18197 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18198 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18200 * config/visium/visium.c (single_set_and_flags): Fix where
18201 we look for cc setter after the compare-elim changes.
18202 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18203 with the vector to match what compare-elim now expects.
18204 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18205 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18206 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18207 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18208 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18209 (neg<mode>2_insn_set_overflow): Likewise.
18211 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18213 PR middle-end/79794
18214 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18215 maybe_expand_insn call, set ops[0].target. If still set after call,
18216 set alt_rtl. Add extra arg to recursive calls.
18217 (extract_bit_field): Add alt_rtl argument. Pass to
18219 * expmed.h (extract_bit_field): Fix prototype.
18220 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18221 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18222 to extract_bit_field_calls.
18223 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18224 Pass alt_rtl to extract_bit_field calls.
18225 * calls.c (store_unaligned_arguments_into_psuedos)
18226 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18227 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18229 * optabs.h (struct expand_operand): Add target bitfield.
18231 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18233 * compare-elim.c (try_eliminate_compare): Canonicalize
18234 operation with embedded compare to
18235 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18236 (set (reg) (operation)].
18238 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18240 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18243 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18244 cost of adding a carry flag for ADC instruction.
18245 [case MINUS]: Ignore the cost of subtracting a carry flag
18246 for SBB instruction.
18248 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18250 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18252 * config/rs6000/bmiintrin.h: New file.
18253 * config/rs6000/bmi2intrin.h: New file.
18254 * config/rs6000/x86intrin.h: New file.
18256 2017-05-12 Jeff Law <law@redhat.com>
18258 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18261 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18263 PR middle-end/80707
18264 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18266 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18268 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18269 early expansion of vector divide builtins.
18270 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18271 builtins identified as having unsigned arguments.
18273 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18275 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18276 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18277 expansion of vector logical operations (and, andc, or, xor,
18280 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18282 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18283 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18285 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18287 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18288 early GIMPLE expansion of vector multiplies.
18290 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18292 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18293 TARGET_HAVE_MOVT conditional.
18294 (movt splitter): Likewise.
18296 2017-05-12 Richard Biener <rguenther@suse.de>
18298 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18299 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18300 Fold all stmts not inplace.
18302 2017-05-12 Richard Biener <rguenther@suse.de>
18304 PR tree-optimization/80713
18305 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18306 inserted_exprs bit for not removed stmts.
18308 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18310 PR middle-end/69921
18311 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18312 parallelized" attribute for parallelized OpenACC kernels.
18313 * omp-offload.c (execute_oacc_device_lower): Use it.
18315 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18316 Set "oacc kernels" attribute.
18317 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18318 parameter. Adjust all users.
18319 (oacc_fn_attrib_kernels_p): Remove function.
18320 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18321 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18322 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18323 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18324 assert "oacc kernels" attribute is set.
18326 2017-05-11 Carl Love <cel@us.ibm.com>
18328 * config/rs6000/rs6000-c: Add support for built-in functions
18329 vector unsigned char vec_popcnt (vector signed char)
18330 vector unsigned char vec_popcnt (vector unsigned char)
18331 vector unsigned short vec_popcnt (vector signed short)
18332 vector unsigned short vec_popcnt (vector unsigned short)
18333 vector unsigned int vec_popcnt (vector signed int)
18334 vector unsigned int vec_popcnt (vector unsigned int)
18335 vector unsigned long long vec_popcnt (vector signed long long)
18336 vector unsigned long long vec_popcnt (vector unsigned long long)
18337 vector signed long long vec_slo (vector signed long long,
18338 vector signed char)
18339 vector signed long long vec_slo (vector signed long long,
18340 vector unsigned char)
18341 vector unsigned long long vec_slo (vector unsigned long long,
18342 vector signed char)
18343 vector unsigned long long vec_slo (vector unsigned long long,
18344 vector unsigned char)
18345 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18346 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18347 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18348 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18349 * doc/extend.texi: Update the built-in documentation file for the
18350 new built-in functions.
18352 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18354 * attribs.h (sorted_attr_string): Move machine independent
18355 functions for target clone support from the i386 port to common
18356 code. Rename ix86_function_versions to common_function_versions.
18357 Rename make_name to make_unique_name.
18358 (common_function_versions): Likewise.
18359 (make_unique_name): Likewise.
18360 (make_dispatcher_decl): Likewise.
18361 (is_function_default_version): Likewise.
18362 * attribs.c (attr_strcmp): Likewise.
18363 (sorted_attr_string): Likewise.
18364 (common_function_versions): Likewise.
18365 (make_unique_name): Likewise.
18366 (make_dispatcher_decl): Likewise.
18367 (is_function_default_version): Likewise.
18368 * config/i386/i386.c (attr_strcmp): Likewise.
18369 (sorted_attr_string): Likewise.
18370 (ix86_function_versions): Likewise.
18371 (make_name): Likewise.
18372 (make_dispatcher_decl): Likewise.
18373 (is_function_default_version): Likewise.
18374 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18376 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18379 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18380 Account for direct move costs for vec_construct of integer
18383 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18386 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18387 (UNSPEC_STX_ATOMIC): Ditto.
18388 (loaddi_via_sse): New insn.
18389 (storedi_via_sse): Ditto.
18390 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18391 Update corresponding peephole2 patterns.
18392 (atomic_storedi_fpu): Ditto.
18394 2017-05-11 Julia Koval <julia.koval@intel.com>
18396 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18397 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18399 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18400 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18401 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18403 2017-05-11 Nathan Sidwell <nathan@acm.org>
18405 * graphite-poly.c: Include dumpfile.h.
18407 * dumpfle.h (dump_function): Declare here ...
18408 * tree-dump.h (dump_function): ... not here.
18409 * dumpfile.c: #include tree-cfg.h.
18410 (dump_function): Move here from ...
18411 * tree-dump.c (dump_function): ... here.
18412 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18413 * graphite-poly.c: Don't include tree-dump.h.
18414 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18415 * print-tree.c: Likewise.
18416 * stor-layout.c: Likewise.
18417 * tree-nested.c: Likewise.
18419 * dumpfile.c (dump_start): Use TDF_FLAGS.
18420 (dump_enable_all): Fix TDF_KIND check thinko.
18422 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18424 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18425 array entries to represent two legal parameterizations of the
18426 overloaded __builtin_cmpb function, as represented by the
18427 P6_OV_BUILTIN_CMPB constant.
18428 (altivec_resolve_overloaded_builtin): Add special case handling
18429 for the __builtin_cmpb function, as represented by the
18430 P6_OV_BUILTIN_CMPB constant.
18431 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18432 (BU_P6_64BIT_2): New macro.
18433 (BU_P6_OVERLOAD_2): New macro
18434 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18435 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18436 (CMPB): Add overload support to represent both 32-bit and 64-bit
18437 compare-bytes function.
18438 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18439 support for TARGET_CMPB.
18440 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18441 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18442 documentation of the __builtin_cmpb overloaded built-in function.
18444 2017-05-11 Richard Biener <rguenther@suse.de>
18446 PR tree-optimization/80705
18447 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18448 bases are not vectorizable.
18450 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18452 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18453 when counting register pressure.
18455 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18457 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18458 (iv_ca_more_deps): Renamed to ...
18459 (iv_ca_compare_deps): ... this.
18460 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18462 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18464 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18466 (determine_group_iv_costs): ... here.
18467 (find_inv_vars_cb): Record inv var if it's not recorded before.
18469 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18471 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18472 (get_shiftadd_cost): Ditto.
18474 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18476 * tree-ssa-address.c: Include header file.
18477 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18479 (add_to_parts): Refactor.
18480 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18481 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18484 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18486 PR tree-optimization/53090
18487 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18489 (extract_cond_operands): Detect condition with IV on both sides
18490 and return COMP_IV_EXPR_2.
18491 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18492 (rewrite_use_compare): Simplify by removing call to function
18493 extract_cond_operands.
18495 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18497 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18498 (extract_cond_operands): Detect condition comparing against non-
18499 invariant bound and return appropriate enum value.
18500 (find_interesting_uses_cond): Update use of extract_cond_operands.
18501 Handle its return value accordingly.
18502 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18504 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18506 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18507 nonlinear iv_use computation in loop invariant sensitive way.
18509 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18511 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18512 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18514 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18516 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18517 (dump_cand): Support iv_cand.inv_exprs.
18518 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18520 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18523 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18525 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18527 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18528 as local function. Include necessary header files.
18529 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18531 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18533 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18535 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18537 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18538 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18539 RSHIFT_EXPR and BIT_NOT_EXPR.
18541 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18543 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18544 (adjust_setup_cost): New parameter supporting round up adjustment.
18545 (struct address_cost_data): Delete.
18546 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18547 (split_address_cost, ptr_difference_cost): Delete.
18548 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18549 (struct ainc_cost_data): New struct.
18550 (get_address_cost_ainc): New function.
18551 (get_address_cost, get_computation_cost): Reimplement.
18552 (determine_group_iv_cost_address): Record inv_expr for all uses of
18554 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18555 (iv_ca_has_deps): Reimplemented to ...
18556 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18558 (iv_ca_extend): Call iv_ca_more_deps.
18560 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18562 * tree-ssa-address.c (struct mem_address): Move to header file.
18563 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18564 * tree-ssa-address.h (struct mem_address): Move from C file.
18565 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18567 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18569 * tree-affine.h (aff_combination_type): New interface.
18570 (aff_combination_zero_p): Remove static.
18571 (aff_combination_const_p): New interface.
18572 (aff_combination_singleton_var_p): New interfaces.
18574 2017-05-11 Richard Biener <rguenther@suse.de>
18576 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18577 Skip unreachable blocks and destinations.
18578 (eliminate): Move stmt removal and fixup ...
18579 (fini_eliminate): ... here. Skip inserted exprs.
18580 (pass_pre::execute): Move fini_pre after fini_eliminate.
18581 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18582 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18583 PRE to get rid of dead code that has invalid SSA form and
18584 split critical edges again.
18586 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18588 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18590 2017-05-11 Richard Biener <rguenther@suse.de>
18592 * passes.c (execute_function_todo): Verify loops if they are
18593 said to be up-to-date.
18594 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18595 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18597 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18600 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18601 handle calling assemble_external ourself.
18604 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18605 modes with zero size. Enhance comment.
18607 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18609 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18610 built-ins for vec_xl and vec_xst with short and char pointer
18613 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18615 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18616 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18617 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18618 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18619 (_mm_maskz_min_round_ss): New intrinsics.
18620 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18621 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18622 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18623 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18624 (__builtin_ia32_minss_mask_round): New builtins.
18625 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18626 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18627 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18629 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18630 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18632 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18635 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18637 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18638 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18639 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18640 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18641 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18642 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18643 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18644 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18645 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18646 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18647 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18648 (__builtin_ia32_mulss_mask_round): New builtins.
18649 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18650 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18651 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18653 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18654 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18656 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18659 2017-05-10 Julia Koval <julia.koval@intel.com>
18661 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18662 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18663 (_mm256_setr_m128i): New intrinsics.
18665 2017-05-10 Julia Koval <julia.koval@intel.com>
18667 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18668 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18669 (_mm_maskz_rcp14_ss): New intrinsics.
18670 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18671 (__builtin_ia32_rcp14ss_mask): New builtins.
18672 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18674 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18676 PR tree-optimization/51513
18677 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18678 (assert_unreachable_fallthru_edge_p): Use it.
18679 (group_case_labels_stmt): Likewise.
18680 * tree-cfg.h: Prototype it.
18681 * stmt.c: Include cfghooks.h and tree-cfg.h.
18682 (emit_case_dispatch_table) <gap_label>: New local variable.
18683 Use it to fill dispatch table gaps.
18684 Test for default_label before updating probabilities.
18685 (expand_case) <default_label>: Remove unneeded initialization.
18686 Test for unreachable default case statement and remove its edge.
18687 Set default_label accordingly.
18688 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18690 2017-05-10 Carl Love <cel@us.ibm.com>
18692 * config/rs6000/rs6000-c: Add support for built-in functions
18693 vector signed char vec_neg (vector signed char)
18694 vector signed short int vec_neg (vector short int)
18695 vector signed int vec_neg (vector signed int)
18696 vector signed long long vec_neg (vector signed long long)
18697 vector float vec_neg (vector float)
18698 vector double vec_neg (vector double)
18699 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18701 * config/rs6000/altivec.h: Add define for vec_neg
18702 * doc/extend.texi: Update the built-in documentation for the
18703 new built-in functions.
18705 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18707 PR tree-optimization/77644
18708 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18710 2017-05-10 Nathan Sidwell <nathan@acm.org>
18712 * dumpfile.h (TDI_lang_all): New.
18713 (TDF_KIND): New. Renumber others
18714 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18716 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18718 (get_dump_file_name): Adjust suffix generation.
18719 (dump_enable_all): Use TDF_KIND.
18720 * doc/invoke.texi (-fdump-lang-all): Document.
18722 * dumpfile.h: Tabify.
18724 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18727 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18728 Move member access before delete.
18730 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18732 * tree-inline.c (expand_call_inline): Split block at stmt
18735 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18738 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18739 are now unused after splitting mov{sf,sd}_hardfloat.
18740 (f32_lr2): Likewise.
18741 (f32_lm): Likewise.
18742 (f32_lm2): Likewise.
18743 (f32_li): Likewise.
18744 (f32_li2): Likewise.
18745 (f32_lv): Likewise.
18746 (f32_sr): Likewise.
18747 (f32_sr2): Likewise.
18748 (f32_sm): Likewise.
18749 (f32_sm2): Likewise.
18750 (f32_si): Likewise.
18751 (f32_si2): Likewise.
18752 (f32_sv): Likewise.
18753 (f32_dm): Likewise.
18754 (f32_vsx): Likewise.
18755 (f32_av): Likewise.
18756 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18757 For movsf, order stores so the VSX stores occur before the GPR
18758 store which encourages the register allocator to use a traditional
18759 FPR instead of a GPR. For movsd, order the stores so that the GPR
18760 store comes before the VSX stores to allow the power6 to work.
18761 This is due to the power6 not having a 32-bit integer store
18762 instruction from a FPR.
18763 (movsf_hardfloat): Likewise.
18764 (movsd_hardfloat): Likewise.
18766 2017-05-09 Martin Sebor <msebor@redhat.com>
18768 PR translation/80280
18769 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18772 PR translation/80280
18773 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18774 data member added in r247778.
18775 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18777 2017-05-09 Nathan Sidwell <nathan@acm.org>
18779 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18781 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18784 2017-05-09 Marek Polacek <polacek@redhat.com>
18786 * doc/invoke.texi: Fix typo.
18788 2017-05-09 Richard Biener <rguenther@suse.de>
18790 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18791 (vrp_val_is_min): Likewise.
18792 (set_value_range_to_value): Likewise.
18793 (set_value_range_to_nonnegative): Likewise.
18794 (gimple_assign_nonzero_p): Likewise.
18795 (gimple_stmt_nonzero_p): Likewise.
18796 (vrp_int_const_binop): Likewise. Remove unreachable case.
18797 (adjust_range_with_scev): Adjust comments.
18798 (compare_range_with_value): Likewise.
18799 (extract_range_from_phi_node): Likewise.
18800 (test_for_singularity): Likewise.
18802 2017-05-09 Richard Biener <rguenther@suse.de>
18804 * tree-vrp.c (get_single_symbol): Add assert that we don't
18805 get overflowed constants as invariant part.
18806 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18807 checks. Use wi::cmp instead of recursing for integer constants.
18808 (compare_values): Just ignore whether we assumed undefined
18809 overflow instead of failing the compare.
18810 (extract_range_for_var_from_comparison_expr): Add comment before the
18811 TREE_NO_WARNING sets.
18812 (test_for_singularity): Likewise.
18813 (extract_range_from_comparison): Do not disable optimization
18814 when we assumed undefined overflow.
18815 (extract_range_basic): Remove init of unused var.
18817 2017-05-09 Richard Biener <rguenther@suse.de>
18819 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18820 (extract_range_from_multiplicative_op_1): Adjust.
18821 (extract_range_from_binary_expr_1): Use int_const_binop.
18823 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18826 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18827 rs6000_store_data_bypass_p in seven define_bypass directives and
18828 in several comments.
18829 * config/rs6000/rs6000-protos.h: Add prototype for
18830 rs6000_store_data_bypass_p function.
18831 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18832 function implements slightly different (rs6000-specific) semantics
18833 than store_data_bypass_p, returning false rather than aborting
18834 with assertion error when arguments do not satisfy the
18835 requirements of store data bypass.
18836 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18837 rs6000_store_data_bypass_p.
18839 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18841 * config/xtensa/xtensa-protos.h
18842 (xtensa_initial_elimination_offset): New declaration.
18843 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18844 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18845 macro definition, add case for FRAME_POINTER_REGNUM when
18846 FRAME_GROWS_DOWNWARD.
18847 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18848 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18849 xtensa_initial_elimination_offset.
18851 2017-05-08 Nathan Sidwell <nathan@acm.org>
18853 * doc/invoke.texi: Alphabetize -fdump options.
18855 2017-05-08 Martin Sebor <msebor@redhat.com>
18857 PR translation/80280
18858 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18860 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18862 * target.def (compute_frame_layout): New optional target hook.
18863 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18864 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18865 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18867 * reload1.c (verify_initial_elim_offsets): Likewise.
18868 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18869 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18870 (arm_get_frame_offsets): Split up into this ...
18871 (arm_compute_frame_layout): ... and this function.
18873 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18875 * config/aarch64/constraints.md (Usa): New constraint.
18876 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18878 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18880 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18881 with_multilib_list after it has been checked.
18883 2017-05-08 Richard Biener <rguenther@suse.de>
18885 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18886 (bitmap_set_subtract_values): Likewise.
18888 2017-05-08 Richard Biener <rguenther@suse.de>
18890 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18891 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18893 (gimple_stmt_nonzero_warnv_p): Rename to ...
18894 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18896 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18897 (extract_range_basic): Adjust, do not disable propagation on
18898 strict overflow sensitive simplification.
18899 (vrp_visit_cond_stmt): Likewise.
18901 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18903 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18904 body size unconditionally.
18906 2017-05-07 Jeff Law <law@redhat.com>
18909 2017-05-06 Jeff Law <law@redhat.com>
18910 PR tree-optimization/78496
18911 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18914 PR tree-optimization/78496
18915 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18916 (simplify_stmt_using_ranges): Call it.
18917 (vrp_dom_walker::before_dom_children): Extract equivalences
18918 from an ASSERT_EXPR with an equality comparison against a
18921 2017-05-06 Jeff Law <law@redhat.com>
18923 PR tree-optimization/78496
18924 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18927 PR tree-optimization/78496
18928 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18929 (simplify_stmt_using_ranges): Call it.
18930 (vrp_dom_walker::before_dom_children): Extract equivalences
18931 from an ASSERT_EXPR with an equality comparison against a
18934 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18936 * lra-constraints.c (lra_copy_reg_equiv): New function.
18937 (split_reg): Use it to copy equivalence information from the
18938 original register to the spill register.
18940 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18942 PR rtl-optimization/75964
18943 * simplify-rtx.c (simplify_const_relational_operation): Remove
18944 invalid handling of comparisons of integer ABS.
18946 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18948 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18949 initialize to zero.
18950 (init_regs): Remove declaration.
18951 (function_arg_advance_32): Initialize error_p as boolean variable.
18953 2017-05-05 Nathan Sidwell <nathan@acm.org>
18955 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18956 lines. Use for (;;).
18958 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18960 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18961 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18962 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18963 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18964 VF=2 that require versioning.
18966 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18968 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18971 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18973 * diagnostic.h (diagnostic_override_option_index): Convert from
18974 macro to inline function.
18976 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18978 * diagnostic.c (last_module_changed_p): New function.
18979 (set_last_module): New function.
18980 (diagnostic_report_current_module): Convert macro usage to
18981 the above functions.
18982 * diagnostic.h (diagnostic_context::last_module): Strengthen
18983 from const line_map * to const line_map_ordinary *.
18984 (diagnostic_last_module_changed): Delete macro.
18985 (diagnostic_set_last_module): Delete macro.
18987 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18989 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18990 with diagnostic_report_diagnostic.
18991 (diagnostic_n_impl_richloc): Likewise.
18992 * diagnostic.h (report_diagnostic): Delete macro.
18993 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18994 with diagnostic_report_diagnostic.
18995 * substring-locations.c (format_warning_va): Likewise.
18997 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18999 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19000 save/restor of format_spec. Move option-printing code to...
19001 (print_option_information): ...this new function, and
19002 reimplement by simply printing to the pretty_printer,
19003 rather than appending to the format string.
19005 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19007 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19008 handling logic into...
19009 (update_effective_level_from_pragmas): ...this new function.
19011 2017-05-04 Andrew Waterman <andrew@sifive.com>
19013 * config/riscv/riscv.opt (mstrict-align): New option.
19014 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19015 (SLOW_UNALIGNED_ACCESS): Define.
19016 (riscv_slow_unaligned_access): Declare.
19017 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19019 (riscv_slow_unaligned_access): New variable.
19020 (rocket_tune_info): Set slow_unaligned_access to true.
19021 (optimize_size_tune_info): Set slow_unaligned_access to false.
19022 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19023 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19024 (riscv_option_override): Set riscv_slow_unaligned_access.
19025 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19027 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19029 * config/riscv/riscv.md: Unify indentation.
19031 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19036 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19038 (extendsi<mode>2): Add support for doing sign extension via
19039 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19040 don't have ISA 3.0 instructions.
19041 (extendsi<mode>2 splitter): Likewise.
19042 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19043 generate the normal insns since SImode can now go in vector
19044 registers. Disallow the special UNSPECs needed for previous
19045 machines to hide SImode being used. Add new insns
19046 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19047 (fix_trunc<mode>si2_stfiwx): Likewise.
19048 (fix_trunc<mode>si2_internal): Likewise.
19049 (fixuns_trunc<mode>si2): Likewise.
19050 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19051 (fctiw<u>z_<mode>_smallint): Likewise.
19052 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19053 of floating point to 32-bit integer from doing a direct move to
19054 the GPR registers to do a store.
19055 (fctiwz_<mode>): Break long line.
19057 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19059 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19060 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19061 (addr_list, addr_offset_valid_p): New.
19062 (split_address_groups): Check offset validity with above function.
19063 (gt-tree-ssa-loop-ivopts.h): Include header file.
19065 2017-05-05 Nathan Sidwell <nathan@acm.org>
19067 * config.gcc (arm*-*-*): Add missing 'fi'.
19069 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19071 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19072 included in -fopt-info does not matter.
19073 * doc/optinfo.texi (-fopt-info): Fix description of default
19074 behavour. Explicitly say order of options included in -fopt-info
19077 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19079 * config.gcc: Allow combinations of aprofile and rmprofile values for
19080 --with-multilib-list.
19081 * config/arm/t-multilib: New file.
19082 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19083 variables. Remove setting of ISA and floating-point ABI in
19084 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19085 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19086 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19087 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19089 * config/arm/t-rmprofile: Likewise except for the matches changes.
19090 * doc/install.texi (--with-multilib-list): Document the combination of
19091 aprofile and rmprofile values and warn about pitfalls in doing that.
19093 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19095 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19096 (movdi_aarch64): Likewise.
19098 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19100 PR tree-optimization/80632
19101 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19103 (build_arrays): Initialize it for virtual phis.
19104 (fix_phi_nodes): Use it for virtual phis.
19106 PR tree-optimization/80558
19107 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19108 [x, y] op z into [x op, y op z] for op & or | if conditions
19111 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19112 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19115 * config/arm/arm.md (use_literal_pool): Remove.
19116 (64-bit immediate split): No longer takes cost into consideration
19117 if arm_disable_literal_pool is enabled.
19118 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19119 used when arm_disable_literal_pool is enabled.
19120 (arm_max_const_double_inline_cost): Remove use of
19121 arm_disable_literal_pool.
19122 (push_minipool_fix): Add assert.
19123 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19124 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19125 (no_literal_pool_sf_immediate): New.
19127 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19129 PR tree-optimization/80613
19130 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19131 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19133 2017-05-05 Richard Biener <rguenther@suse.de>
19135 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19137 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19139 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19140 of this flag from insn conditions due to removal from r247495.
19142 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19144 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19146 (arm_early_store_addr_dep_ptr): Likewise.
19147 * config/arm/aarch-common-protos.h
19148 (arm_early_load_addr_dep_ptr): Add prototype.
19149 (arm_early_store_addr_dep_ptr): Likewise.
19150 * config/arm/cortex-a53.md: Add new bypasses.
19152 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19154 * tree.c (next_type_uid): Change type to unsigned.
19155 (type_hash_canon): Decrement back next_type_uid if
19156 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19157 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19160 2017-05-04 Martin Sebor <msebor@redhat.com>
19162 * builtins.c: Fix a trivial typo in a comment.
19164 PR middle-end/79234
19165 * builtins.c (check_sizes): Adjust to handle reading past the end.
19166 Avoid printing excessive upper bound of ranges. Use %E to print
19167 tree nodes instead of converting them to %wu.
19168 (expand_builtin_memchr): New function.
19169 (compute_dest_size): Rename...
19170 (compute_objsize): ...to this.
19171 (expand_builtin_memcpy): Adjust.
19172 (expand_builtin_mempcpy): Adjust.
19173 (expand_builtin_strcat): Adjust.
19174 (expand_builtin_strcpy): Adjust.
19175 (check_strncat_sizes): Adjust.
19176 (expand_builtin_strncat): Adjust.
19177 (expand_builtin_strncpy): Adjust and simplify.
19178 (expand_builtin_memset): Adjust.
19179 (expand_builtin_bzero): Adjust.
19180 (expand_builtin_memcmp): Adjust.
19181 (expand_builtin): Handle memcmp.
19182 (maybe_emit_chk_warning): Check strncat just once.
19184 2017-05-04 Martin Sebor <msebor@redhat.com>
19186 PR preprocessor/79214
19187 PR middle-end/79222
19188 PR middle-end/79223
19189 * builtins.c (check_sizes): Add inlining context and issue
19190 warnings even when -Wno-system-headers is set.
19191 (check_strncat_sizes): Same.
19192 (expand_builtin_strncat): Same.
19193 (expand_builtin_memmove): New function.
19194 (expand_builtin_stpncpy): Same.
19195 (expand_builtin): Handle memmove and stpncpy.
19197 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19199 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19200 which is not used any more.
19202 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19204 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19206 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19208 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19209 (cortexa53_tunings): Likewise.
19210 (cortexa57_tunings): Likewise.
19211 (cortexa72_tunings): Likewise.
19212 (cortexa73_tunings): Likewise.
19214 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19216 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19217 Set loop alignment to 8.
19219 2017-05-04 Martin Sebor <msebor@redhat.com>
19221 PR translation/80280
19222 * builtins.c (expand_builtin_object_size): Add missing quoting to
19223 %D and like directives.
19224 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19225 (hsa_type_for_tree_type): Same.
19226 (verify_function_arguments): Same.
19227 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19228 * varasm.c (get_section): Same.
19231 2017-05-04 Martin Sebor <msebor@redhat.com>
19233 PR translation/80280
19234 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19236 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19238 * config/aarch64/aarch64.c (generic_addrcost_table):
19239 Change HI/TI mode setting.
19241 2017-05-04 Martin Jambor <mjambor@suse.cz>
19243 PR tree-optimization/80622
19244 * tree-sra.c (comes_initialized_p): New function.
19245 (build_accesses_from_assign): Only set write lazily when
19246 comes_initialized_p is false.
19247 (analyze_access_subtree): Use comes_initialized_p.
19248 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19249 instead of testing for PARM_DECL.
19251 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19253 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19254 constraint on operand 0 to allow more general addressing modes.
19255 Adjust output template.
19256 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19258 * config/aarch64/aarch64-protos.h
19259 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19260 * config/aarch64/constraints.md (Dp): New address constraint.
19261 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19264 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19266 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19267 update use of estimate_ipcp_clone_size_and_time.
19268 (estimate_local_effects): Update use of
19269 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19270 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19271 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19272 Return nonspecialized time.
19274 2017-05-04 Richard Biener <rguenther@suse.de>
19276 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19277 for the last VUSE which def dominates the PHI. Directly call
19279 (get_continuation_for_phi_1): Remove.
19281 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19283 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19284 to explain the use of truncating division. Cap the number of
19285 iterations to the maximum given by nb_iterations_upper_bound,
19288 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19290 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19291 * configure: Regenerate.
19292 * config.in: Regenerate.
19293 * config/i386/driver-mingw32.c: new file.
19294 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19295 * config.host: Link driver-mingw32.o on MinGW host.
19296 * doc/install.texi: Document new --enable-mingw-wildcard configure
19299 2017-05-04 Marek Polacek <polacek@redhat.com>
19301 PR tree-optimization/80612
19302 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19304 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19305 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19307 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19308 (movt splitter): Likewise.
19309 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19310 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19311 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19312 block for Thumb-1 with MOVT.
19313 (thumb2_legitimate_address_p): Move code block ...
19314 (can_avoid_literal_pool_for_label_p): ... into this new function.
19315 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19317 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19318 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19319 "M-profile targets with the MOVT instruction".
19321 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19323 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19324 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19325 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19327 2017-05-04 Martin Liska <mliska@suse.cz>
19329 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19330 variable cond_code.
19332 2017-05-04 Richard Biener <rguenther@suse.de>
19334 * tree.c (array_at_struct_end_p): Handle arrays at struct
19335 end with flexarrays more conservatively. Refactor and treat
19336 arrays of arrays or aggregates more strict. Fix
19337 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19338 * tree.c (array_at_struct_end_p): Adjust prototype.
19339 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19340 * gimple-fold.c (get_range_strlen): Likewise.
19341 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19343 2017-05-04 Richard Biener <rguenther@suse.de>
19345 PR tree-optimization/31130
19346 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19348 (supports_overflow_infinity): Likewise.
19349 (is_negative_overflow_infinity): Likewise.
19350 (is_positive_overflow_infinity): Likewise.
19351 (is_overflow_infinity): Likewise.
19352 (stmt_overflow_infinity): Likewise.
19353 (overflow_infinity_range_p): Likewise.
19354 (usable_range_p): Remove as always returning true.
19355 (make_overflow_infinity): Remove.
19356 (negative_overflow_infinity): Likewise.
19357 (positive_overflow_infinity): Likewise.
19358 (avoid_overflow_infinity): Likewise.
19359 (set_value_range): Adjust accordingly.
19360 (set_value_range_to_nonnegative): Likewise, remove now unused
19361 overflow_infinity arg.
19362 (vrp_operand_equal_p): Adjust.
19363 (update_value_range): Likewise.
19364 (range_int_cst_singleton_p): Likewise.
19365 (operand_less_p): Likewise.
19366 (compare_values_warnv): Likewise.
19367 (extract_range_for_var_from_comparison_expr): Likewise.
19368 (vrp_int_const_binop): Likewise.
19369 (zero_nonzero_bits_from_vr): Likewise.
19370 (extract_range_from_multiplicative_op_1): Likewise.
19371 (extract_range_from_binary_expr_1): Likewise.
19372 (extract_range_from_unary_expr): Likewise.
19373 (extract_range_from_comparison): Likewise.
19374 (extract_range_basic): Likewise.
19375 (adjust_range_with_scev): Likewise.
19376 (compare_ranges): Likewise.
19377 (compare_range_with_value): Likewise.
19378 (dump_value_range): Likewise.
19379 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19381 (simplify_cond_using_ranges): Adjust.
19383 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19385 * brig-builtins.def: Added a builtin for class_f64.
19386 * builtin-types.def: Added a builtin type needed by class_f64.
19388 2017-05-03 Jason Merrill <jason@redhat.com>
19390 * timevar.def: Add TV_CONSTEXPR.
19392 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19394 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19396 2017-05-03 Martin Jambor <mjambor@suse.cz>
19398 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19399 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19400 * ipa-cp.c (ipcp_propagate_stage): Do not call
19401 ipa_update_after_lto_read.
19402 * ipa-inline.c (ipa_inline): Likewise.
19404 2017-05-03 Martin Jambor <mjambor@suse.cz>
19406 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19407 tag. Added a default constructor and a destructor.
19408 (ipa_edge_args_sum_t): New class;
19409 (ipa_edge_args_sum): Declare.
19410 (ipa_edge_args_vector): Remove declaration.
19411 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19412 (ipa_free_edge_args_substructures): Remove declaration.
19413 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19414 (ipa_edge_args_info_available_for_edge_p): Likewise.
19415 * ipa-prop.c (ipa_edge_args_vector): Removed.
19416 (edge_removal_hook_holder): Likewise.
19417 (edge_duplication_hook_holder): Likewise.
19418 (ipa_edge_args_sum): New variable.
19419 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19420 ipa_edge_args_vector.
19421 (ipa_free_edge_args_substructures): Likewise.
19422 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19423 ipa_edge_args_vector.
19424 (ipa_edge_removal_hook): Turned into method
19425 ipa_edge_args_sum_t::remove.
19426 (ipa_edge_duplication_hook): Turned into method
19427 ipa_edge_args_sum_t::duplicate.
19428 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19429 registering edge hooks.
19430 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19431 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19432 ipa_edge_args_sum instead of ipa_edge_args_vector.
19433 * ipa-profile.c (ipa_profile): Likewise.
19435 2017-05-03 Martin Jambor <mjambor@suse.cz>
19437 * symbol-summary.h (function_summary): New method exists.
19438 (function_summary::symtab_removal): Deallocate through release.
19439 (call_summary): New class.
19440 (gt_ggc_mx): New overload.
19441 (gt_pch_nx): Likewise.
19442 (gt_pch_nx): Likewise.
19444 2017-05-03 Jeff Law <law@redhat.com>
19446 PR tree-optimization/78496
19447 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19448 from simplify_cond_using_ranges. Split off code to walk
19449 backwards through casts into ...
19450 (simplify_cond_using_ranges_2): New function.
19451 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19452 (execute_vrp): After identifying jump threads, call
19453 simplify_cond_using_ranges_2.
19455 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19458 * ipa-inline.h (inline_summary): Add ctor.
19459 (create_ggc): Do not use ggc_cleared_alloc.
19461 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19462 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19464 * gcc.c (handle_braces): Support escaping in switch matching
19466 * doc/invoke.texi (Spec Files): Document it.
19467 Remove superfluous @code markup in items.
19469 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19471 * diagnostic-show-locus.c (struct column_range): New struct.
19472 (get_affected_columns): New function.
19473 (get_printed_columns): New function.
19474 (struct correction): New struct.
19475 (correction::ensure_capacity): New function.
19476 (correction::ensure_terminated): New function.
19477 (struct line_corrections): New struct.
19478 (line_corrections::~line_corrections): New dtor.
19479 (line_corrections::add_hint): New function.
19480 (layout::print_trailing_fixits): Reimplement in terms of the new
19482 (selftest::test_overlapped_fixit_printing): New function.
19483 (selftest::diagnostic_show_locus_c_tests): Call it.
19485 2017-05-03 Nathan Sidwell <nathan@acm.org>
19487 Canonicalize canonical type hashing
19488 * tree.h (type_hash_canon_hash): Declare.
19489 * tree.c (type_hash_list, attribute_hash_list): Move into
19490 type_hash_canon_hash.
19491 (build_type_attribute_qual_variant): Break out hash code calc into
19492 type_hash_canon_hash.
19493 (type_hash_canon_hash): New. Generic type hash computation.
19494 (build_range_type_1, build_array_type_1, build_function_type,
19495 build_method_type_directly, build_offset_type, build_complex_type,
19496 make_vector_type): Call it.
19498 2017-05-03 Richard Biener <rguenther@suse.de>
19500 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19501 When all DRs have unknown misaligned do not always peel
19502 when there is a store but apply the same costing model as if
19503 there were only loads.
19505 2017-05-03 Richard Biener <rguenther@suse.de>
19508 PR tree-optimization/80492
19509 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19510 compare_base_decls returning dont-know properly.
19512 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19514 * config/arm/iterators.md (CCSI): New mode iterator.
19515 (arch): New mode attribute.
19516 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19517 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19518 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19519 code iterator for success result mode.
19520 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19521 the corresponding new insn generators.
19523 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19526 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19527 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19529 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19531 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19532 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19533 (DDR_A): Wrap DDR argument in brackets.
19534 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19535 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19536 (DDR_REVERSED_P): Likewise.
19538 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19540 PR tree-optimization/79472
19541 * tree-switch-conversion.c (struct switch_conv_info): Add
19542 contiguous_range and default_case_nonstandard fields.
19543 (collect_switch_conv_info): Compute contiguous_range and
19544 default_case_nonstandard fields, don't clear final_bb if
19545 contiguous_range and only the default case doesn't have the required
19547 (check_all_empty_except_final): Set default_case_nonstandard instead
19548 of failing if contiguous_range and the default case doesn't have empty
19550 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19551 and only the default case doesn't have the required constants. Skip
19553 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19554 if default_case_nonstandard.
19555 (build_constructors): Build constant 1 just once. Assert that default
19556 values aren't inserted in between cases if contiguous_range. Skip
19558 (build_arrays): Skip virtual phis.
19559 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19560 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19561 Handle virtual phis.
19562 (gen_inbound_check): Handle default_case_nonstandard case.
19563 (process_switch): Adjust check_final_bb caller. Call
19564 gather_default_values with the first non-default case instead of
19565 default case if default_case_nonstandard.
19567 2017-05-02 Nathan Sidwell <nathan@acm.org>
19569 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19570 check. Fix formatting.
19572 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19574 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19575 errors when comparing specialized and unspecialized times.
19577 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19579 * diagnostic-show-locus.c
19580 (layout::should_print_annotation_line_p): Make private.
19581 (layout::print_annotation_line): Make private.
19582 (layout::annotation_line_showed_range_p): Make private.
19583 (layout::show_ruler): Make private.
19584 (layout::print_source_line): Make private. Pass in line and
19585 line_width, rather than calling location_get_source_line. Drop
19587 (layout::print_leading_fixits): New method.
19588 (layout::print_any_fixits): Rename to...
19589 (layout::print_trailing_fixits): ...this, and make private.
19590 Don't print newline fixits.
19591 (diagnostic_show_locus): Move logic for printing one row into...
19592 (layout::print_line): ...this new function. Move the
19593 location_get_source_line call and error-handling from
19594 print_source_line to here. Call print_leading_fixits, and rename
19595 print_any_fixits to print_trailing_fixits.
19596 (selftest::test_fixit_insert_containing_newline): Update now that
19597 newlines are partially supported.
19598 (selftest::test_fixit_insert_containing_newline_2): New test.
19599 (selftest::test_fixit_replace_containing_newline): Update comments.
19600 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19601 * edit-context.c (class added_line): New class.
19602 (class edited_line): Describe newline handling in comment.
19603 (edited_line::actually_edited_p): New method.
19604 (edited_line::print_content): Delete redundant decl.
19605 (edited_line::m_predecessors): New field.
19606 (edited_file::print_content): Call edited_line::print_content.
19607 (edited_file::print_diff): Update to support newlines.
19608 (edited_file::print_diff_hunk): Likewise.
19609 (edited_file::print_run_of_changed_lines): New function.
19610 (edited_file::print_diff_line): Convert to...
19611 (print_diff_line): ...this.
19612 (edited_file::get_effective_line_count): New function.
19613 (edited_line::edited_line): Initialize new field m_predecessors.
19614 (edited_line::~edited_line): Clean up m_predecessors.
19615 (edited_line::apply_fixit): Handle newlines.
19616 (edited_line::get_effective_line_count): New function.
19617 (edited_line::print_content): New function.
19618 (edited_line::print_diff_lines): New function.
19619 (selftest::test_applying_fixits_insert_containing_newline): New
19621 (selftest::test_applying_fixits_replace_containing_newline): New
19623 (selftest::insert_line): New function.
19624 (selftest::test_applying_fixits_multiple_lines): Add example of
19626 (selftest::edit_context_c_tests): Call the new tests.
19628 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19630 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19631 parameter cand. Update dump information.
19632 (get_computation_cost): Update uses.
19634 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19636 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19637 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19638 (get_computation_at, rewrite_use_address): Update use of
19639 get_computation_aff.
19641 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19643 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19644 (get_computation): Delete.
19645 (get_computation_cost): Implement like get_computation_cost_at.
19646 Use get_computation_at.
19647 (get_computation_cost_at): Delete.
19648 (rewrite_use_nonlinear_expr): Use get_computation_at.
19649 (rewrite_use_compare, remove_unused_ivs): Ditto.
19651 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19653 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19655 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19657 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19658 (ivopts_global_cost_for_size): Rename parameter and update uses.
19659 (iv_ca_recount_cost): Update uses.
19660 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19661 candidates seperately in n_invs and n_cands.
19662 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19664 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19666 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19667 (find_inv_vars_cb): New.
19668 (find_depends): Renamed to ...
19669 (find_inv_vars): ... this.
19670 (add_candidate_1, force_var_cost): Call find_inv_vars.
19671 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19673 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19675 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19676 inv_vars. Add inv_exprs.
19677 (struct iv_cand): Rename depends_on to inv_vars.
19678 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19679 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19680 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19681 (dump_cand): Dump inv_vars.
19682 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19683 (record_invariant, find_depends, add_candidate_1): Ditto.
19684 (set_group_iv_cost, force_var_cost): Ditto.
19685 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19686 (get_computation_cost_at, get_computation_cost): Ditto.
19687 (determine_group_iv_cost_generic): Ditto.
19688 (determine_group_iv_cost_address): Ditto.
19689 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19690 (determine_group_iv_costs): Ditto.
19691 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19692 (iv_ca_set_remove_invariants): Renamed to ...
19693 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19694 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19695 (iv_ca_set_add_invariants): Renamed to ...
19696 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19697 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19698 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19699 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19700 (create_new_ivs): Remove useless dump.
19702 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19704 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19706 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19707 (iv_ca_set_no_cp, create_new_iv): Ditto.
19709 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19711 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19713 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19715 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19716 function tree_check2.
19718 2017-05-02 Martin Liska <mliska@suse.cz>
19720 * doc/gcov.texi: Add missing preposition.
19721 * gcov.c (function_info::function_info): Properly fill up
19722 all member variables.
19724 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19726 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19728 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19730 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19732 2017-05-02 Martin Liska <mliska@suse.cz>
19735 * lto-streamer-in.c (lto_read_tree_1): Remove
19736 LTO_STREAMER_DEBUG.
19737 * lto-streamer.c (struct tree_hash_entry): Likewise.
19738 (struct tree_entry_hasher): Likewise.
19739 (tree_entry_hasher::hash): Likewise.
19740 (tree_entry_hasher::equal): Likewise.
19741 (lto_streamer_init): Likewise.
19742 (lto_orig_address_map): Likewise.
19743 (lto_orig_address_get): Likewise.
19744 (lto_orig_address_remove): Likewise.
19745 * lto-streamer.h: Likewise.
19746 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19747 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19749 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19751 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19752 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19753 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19754 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19755 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19756 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19757 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19758 (mm_maskz_sub_ss): New intrinsics.
19759 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19760 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19761 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19762 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19763 (__builtin_ia32_subss_mask_round): New builtins.
19764 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19765 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19766 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19768 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19769 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19771 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19774 2017-05-02 Martin Jambor <mjambor@suse.cz>
19776 PR tree-optimization/78687
19777 * tree-sra.c (access): New field parent.
19778 (process_subtree_disqualification): New function.
19779 (disqualify_candidate): Call it.
19780 (build_accesses_from_assign): Reset write flag if creating an
19782 (build_access_subtree): Fill in parent field and also prpagate
19783 down grp_write flag.
19784 (create_artificial_child_access): New parameter set_grp_write, set
19785 grp_write to its value.
19786 (propagate_subaccesses_across_link): Also propagate grp_write flag
19788 (propagate_all_subaccesses): Push the closest parent back to work
19789 queue if add_access_to_work_queue returned true.
19791 2017-05-02 Richard Biener <rguenther@suse.de>
19793 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19794 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19795 -fstrict-overflow documentation.
19796 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19797 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19798 flag_strict_overflow.
19799 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19800 * lto-opts.c (lto_write_options): Do not stream it.
19801 * lto-wrapper.c (merge_and_complain): Do not handle it.
19802 * opts.c (default_options_table): Do not set -fstrict-overflow.
19803 (finish_options): Likewise do not clear it when sanitizing.
19804 * simplify-rtx.c (simplify_const_relational_operation): Do not
19805 test flag_strict_overflow.
19807 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19809 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19810 using enabled attribute.
19811 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19812 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19813 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19814 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19815 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19816 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19817 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19818 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19819 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19820 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19822 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19824 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19826 2017-05-02 Richard Biener <rguenther@suse.de>
19828 PR tree-optimization/80591
19830 2017-04-10 Richard Biener <rguenther@suse.de>
19832 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19835 2017-05-02 Richard Biener <rguenther@suse.de>
19837 PR tree-optimization/80549
19838 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19839 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19840 headers if they do not have a preheader.
19842 2017-05-02 Martin Liska <mliska@suse.cz>
19845 * common.opt: Fix typo.
19846 * doc/invoke.texi: Likewise.
19848 2017-05-01 Jan Beulich <jbeulich@suse.com>
19850 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19851 swapping, add (x,x,m,x,n) alternative.
19853 2017-05-01 Nathan Sidwell <nathan@acm.org>
19855 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19856 unnecessary unadjusted_alignment check.
19858 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19861 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19863 * gimplify.c (gimplify_cilk_detach): New function.
19864 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19865 * tree-core.h: Document EXPR_CILK_SPAWN.
19866 * tree.h (EXPR_CILK_SPAWN): Define.
19868 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19870 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19871 to use new fixit_hint representation, using the "replace" logic.
19872 (get_line_span_for_fixit_hint): Likewise.
19873 (layout::print_any_fixits): Likewise.
19874 (selftest::test_one_liner_many_fixits): Rename to...
19875 (selftest::test_one_liner_many_fixits_1): ...this, and update
19876 comment and expected output to reflect that the multiple fix-it
19877 hints are now consolidated into one insertion.
19878 (selftest::test_one_liner_many_fixits_2): New test.
19879 (selftest::test_diagnostic_show_locus_one_liner): Update for
19881 (selftest::test_fixit_consolidation): Update for fix-it API
19883 * diagnostic.c (print_parseable_fixits): Likewise.
19884 * edit-context.c (edited_line::m_line_events): Convert from
19885 auto_vec <line_event *> to auto_vec <line_event>.
19886 (class line_event): Convert from abstract base class to a concrete
19887 class, taking over the role of replace_event.
19888 (class insert_event): Delete.
19889 (class replace_event): Rename to class line_event. Convert to
19891 (edit_context::add_fixits): Reimplement.
19892 (edit_context::apply_insert): Delete.
19893 (edit_context::apply_replace): Rename to...
19894 (edit_context::apply_fixit): ...this. Convert to half-open range.
19895 (edited_file::apply_insert): Delete.
19896 (edited_file::apply_replace): Rename to...
19897 (edited_file::apply_fixit): ...this.
19898 (edited_line::~edited_line): Drop deletion of events.
19899 (edited_line::apply_insert): Delete.
19900 (edited_line::apply_replace): Rename to...
19901 (edited_line::apply_fixit): ...this. Convert to half-open range.
19902 Update for change to type of m_line_events.
19903 * edit-context.h (edit_context::apply_insert): Delete.
19904 (edit_context::apply_replace): Rename to...
19905 (edit_context::apply_fixit): ...this.
19907 2017-05-01 Martin Sebor <msebor@redhat.com>
19909 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19912 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19915 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19916 __get_cpuid_max returns 0.
19917 (__get_cpuid_count): Ditto.
19919 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19921 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19922 replacement expression is another instance of one of its arguments.
19924 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19927 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19928 check for stack push/pop autoinc.
19929 * config/i386/i386.c (ix86_agi_dependent): Return false
19930 if the only reason why modified_in_p returned true is that
19931 addr is SP based and set_insn is a push or pop.
19933 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19935 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19938 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19941 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19942 (account_size_time): Use two predicates - exec_pred and
19944 (evaluate_conditions_for_known_args): Compute both clause and
19946 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19947 (inline_summary_t::duplicate): Update.
19948 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19950 (compute_inline_parameters): Likewise.
19951 (estimate_edge_size_and_time): Update caluclation of time.
19952 (estimate_node_size_and_time): Compute both time and nonspecialized
19954 (estimate_ipcp_clone_size_and_time): Update.
19955 (inline_merge_summary): Update.
19956 (do_estimate_edge_time): Update.
19957 (do_estimate_edge_size): Update.
19958 (do_estimate_edge_hints): Update.
19959 (inline_read_section, inline_write_summary): Stream both new predicates.
19960 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19962 (compute_inlined_call_time): Cleanup.
19963 (big_speedup_p): Update.
19964 (edge_badness): Update.
19965 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19966 (size_time_entry): Replace predicate by exec_predicate and
19967 nonconst_predicate.
19968 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19969 (estimate_edge_time): Return also nonspec_time.
19970 (reset_edge_growth_cache): Update.
19972 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19974 PR rtl-optimization/80491
19975 * ifcvt.c (noce_process_if_block): When looking for x setter
19976 with missing else_bb, don't check only the insn right before
19977 cond_earliest, but look for the last insn that x is modified in
19978 within the same bb.
19980 PR rtl-optimization/80491
19981 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19983 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19985 PR tree-optimization/80487
19986 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19988 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19990 PR tree-optimization/79697
19991 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19992 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19993 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19995 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19996 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19998 2017-04-28 Martin Sebor <msebor@redhat.com>
20000 PR tree-optimization/80523
20001 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20002 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20004 (maybe_warn, format_directive, parse_directive): Use new functions.
20005 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20007 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20009 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20011 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20013 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20014 target_header_dir): Set correctly.
20015 * configure: Regenerated.
20016 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20017 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20018 instead of SYSTEM_HEADER_DIR.
20020 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20022 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20023 (estimate_local_effects): Likewise.
20024 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20025 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20026 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20027 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20028 do_estimate_edge_time, estimate_edge_time): Likewise.
20029 * ipa-inline-analysis.c (estimate_node_size_and_time,
20030 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20031 (estimate_time_after_inlining): Remove.
20033 2017-04-28 Martin Liska <mliska@suse.cz>
20035 * doc/gcov.texi: Enhance documentation of gcov.
20037 2017-04-28 Martin Liska <mliska@suse.cz>
20039 * doc/gcov.texi: Sort options in alphabetic order.
20040 * doc/gcov-dump.texi: Likewise.
20041 * doc/gcov-tool.texi: Likewise.
20042 * gcov.c (print_usage): Likewise.
20043 * gcov-dump.c (print_usage): Likewise.
20044 * gcov-tool.c (print_merge_usage_message): Likewise.
20045 (print_rewrite_usage_message): Likewise.
20046 (print_overlap_usage_message): Likewise.
20048 2017-04-28 Martin Liska <mliska@suse.cz>
20050 PR gcov-profile/53915
20051 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20053 2017-04-28 Martin Liska <mliska@suse.cz>
20055 PR gcov-profile/79891
20056 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20057 is marked by compiler as living on a line.
20058 (get_cycles_count): Remove usage of the union.
20059 (output_intermediate_file): Likewise.
20060 (find_source): Fix GNU coding style.
20061 (accumulate_line_counts): Remove old non-all block mode.
20062 (output_lines): Remove usage of the union.
20063 * profile.c (output_location): Include all BBs, even if
20064 belonging to a same line (and file) as a previous BB.
20066 2017-04-28 Martin Liska <mliska@suse.cz>
20068 * gcov.c (process_args): Handle new argument 'w'.
20069 (read_graph_file): Assign ID to BBs.
20070 (output_branch_count): Display BB # if verbose flag is set.
20071 (output_lines): Likewise for arcs.
20072 (print_usage): Add '--verbose' option help.
20073 * doc/gcov.texi: Document --verbose (-w) option.
20075 2017-04-28 Martin Liska <mliska@suse.cz>
20077 * gcov.c (struct block_location_info): New struct.
20078 (process_file): Fill up the new structure.
20079 (read_graph_file): Replace usage of encoding by the newly added
20081 (add_line_counts): Likewise.
20082 (accumulate_line_counts): Remove usage of the union.
20083 (function_info::function_info): New function.
20084 (function_info::~function_info): Likewise.
20085 (process_file): Call delete instead of release_function.
20086 (release_function): Release the function.
20087 (release_structures): Call delete instead of release_function.
20088 (solve_flow_graph): Replace usage of num_blocks.
20089 (find_exception_blocks): Likewise.
20090 (output_lines): Fix GNU coding style.
20092 2017-04-28 Martin Liska <mliska@suse.cz>
20095 * coverage.c (coverage_remove_note_file): New function.
20096 * coverage.h: Declare the function.
20097 * toplev.c (finalize): Clean if an error has been seen.
20099 2017-04-28 Martin Liska <mliska@suse.cz>
20101 PR gcov-profile/80031
20102 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20103 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20104 * gcov.c (read_graph_file): Read just number of blocks.
20105 * profile.c (branch_prob): Do not stream 0 flags per a basic
20108 2017-04-28 Martin Liska <mliska@suse.cz>
20110 * gcov-dump.c (tag_*): Add new argument to declarations.
20111 (dump_gcov_file): Likewise.
20112 (tag_blocks): Add and use new argument depth.
20113 (tag_arcs): Likewise.
20114 (tag_lines): Likewise.
20115 (tag_counters): Likewise.
20116 (tag_summary): Likewise.
20117 (dump_working_sets): Use depth to do a proper indentation.
20119 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20122 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20123 * symtab.c (symtab_node::debug_symtab): Move definition here.
20125 2017-04-28 Richard Biener <rguenther@suse.de>
20127 * lto-streamer.h (LTO_major_version): Bump to 7.
20129 2017-04-28 Richard Biener <rguenther@suse.de>
20131 * tree-vrp.c (assert_info): New struct.
20132 (add_assert_info): New helper.
20133 (register_edge_assert_for_2): Refactor to add asserts to a vector
20135 (register_edge_assert_for_1): Likewise.
20136 (register_edge_assert_for): Likewise.
20137 (finish_register_edge_assert_for): New helper actually registering
20138 asserts where live on edge.
20139 (find_conditional_asserts): Adjust.
20140 (find_switch_asserts): Likewise.
20141 (evrp_dom_walker::try_find_new_range): Generalize.
20142 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20144 2017-04-27 Marek Polacek <polacek@redhat.com>
20147 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20148 arg10 and arg11 to itype.
20150 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20152 * doc/extend.texi (Object Size Checking): Improve grammar.
20154 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20157 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20158 that the logic for permitting reciprocal estimates matches that
20161 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20164 * tree.c (type_cache_hasher::equal): Only compare
20165 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20166 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20167 non-aggregate element types.
20168 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20169 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20171 2017-04-27 Richard Biener <rguenther@suse.de>
20173 PR middle-end/80533
20174 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20175 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20176 keeping a reference to a trailing array.
20178 2017-04-27 Richard Biener <rguenther@suse.de>
20180 PR middle-end/80539
20181 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20182 being in loop-closed SSA form conservatively.
20183 (chrec_fold_multiply_poly_poly): Likewise.
20185 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20187 PR middle-end/79665
20188 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20189 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20191 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20194 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20195 (aarch64_function_arg_alignment): Return unsigned int again, but still
20196 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20197 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20198 Don't emit -Wpsabi note.
20199 (aarch64_function_arg_boundary): Likewise.
20200 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20203 2017-04-26 Nathan Sidwell <nathan@acm.org>
20205 * tree.h (crc32_unsigned_n): Declare.
20206 (crc32_unsigned, crc32_unsigned): Make inline.
20207 * tree.c (crc32_unsigned_bits): Replace with ...
20208 (crc32_unsigned_n): ... this.
20209 (crc32_unsigned, crc32_byte): Remove.
20210 (crc32_string): Remove unnecessary braces.
20212 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20214 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20215 * ipa-inline-analysis.c (MAX_TIME): Remove.
20216 (account_size_time): Use sreal for time.
20217 (dump_inline_summary): Update.
20218 (estimate_function_body_sizes): Update.
20219 (estimate_edge_size_and_time): Update.
20220 (estimate_calls_size_and_time): Update.
20221 (estimate_node_size_and_time): Update.
20222 (inline_merge_summary): Update.
20223 (inline_update_overall_summary): Update.
20224 (estimate_time_after_inlining): Update.
20225 (inline_read_section): Update.
20226 (inline_write_summary): Update.
20227 * ipa-inline.c (compute_uninlined_call_time): Update.
20228 (compute_inlined_call_time): Update.
20229 (recursive_inlining): Update.
20230 (inline_small_functions): Update.
20231 (dump_overall_stats): Update.
20232 * ipa-inline.h: Include sreal.h.
20233 (size_time_entry): Turn time to sreal.
20234 (inline_summary): Turn self_time nad time to sreal.
20236 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20238 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20240 (sreal::stream_out, sreal::stream_in): New.
20241 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20243 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20245 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20248 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20251 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20253 (dimode_scalar_chain::compute_convert_gain): Ditto.
20254 (dimode_scalar_chain::make_vector_copies): Ditto.
20255 (dimode_scalar_chain::convert_reg): Ditto.
20256 (dimode_scalar_chain::convert_insn): Ditto.
20257 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20258 (VI248_AVX512BW_1): New mode iterator.
20259 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20260 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20263 2017-04-25 Martin Sebor <msebor@redhat.com>
20265 PR tree-optimization/80497
20266 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20267 constants are representable in HOST_WIDE_INT.
20268 (parse_directive): Ditto.
20270 2017-04-25 Martin Sebor <msebor@redhat.com>
20273 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20274 (new_zero_array): Adjust signature.
20275 (dom_info::dom_init): Used unsigned rather that size_t.
20276 (dom_info::dom_info): Same.
20278 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20279 Jakub Jelinek <jakub@redhat.com>
20282 * config/arm/arm.c: Include gimple.h.
20283 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20284 returns negative, increment ncrn only if it returned positive.
20285 (arm_needs_doubleword_align): Return int instead of bool,
20286 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20287 members, but if there is any such non-FIELD_DECL
20288 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20289 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20290 returns negative, increment nregs only if it returned positive.
20291 (arm_setup_incoming_varargs): Likewise.
20292 (arm_function_arg_boundary): Emit -Wpsabi note if
20293 arm_needs_doubleword_align returns negative, return
20294 DOUBLEWORD_ALIGNMENT only if it returned positive.
20296 2017-04-25 Marek Polacek <polacek@redhat.com>
20299 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20300 first argument to type.
20302 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20305 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20306 type checks to test for compatibility instead of equality.
20308 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20309 Jakub Jelinek <jakub@redhat.com>
20312 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20314 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20315 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20316 the alignment computation, but return their maximum in warn_alignment.
20317 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20318 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20320 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20321 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20324 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20326 * config/arc/simdext.md (dmpyh): Fix typo.
20328 2017-04-25 Richard Biener <rguenther@suse.de>
20330 PR tree-optimization/80492
20331 * alias.c (compare_base_decls): Handle registers with asm
20332 specification conservatively.
20333 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20334 compare_base_decls returning dont-know properly.
20336 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20338 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20339 (legitimate_offset_address_p): New function.
20340 (arc_legitimate_address_p): Use above function.
20342 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20344 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20346 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20348 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20349 ACCH registers whenever they are available.
20351 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20353 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20354 double regs fix when not used.
20356 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20358 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20360 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20361 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20363 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20365 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20366 when emitting short ADD instructions.
20368 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20370 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20372 (cmpsi_cc_c_insn): Likewise.
20373 (cbranchsi4_scratch): Compute proper instruction length using
20374 compact_hreg_operand.
20375 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20377 2017-04-25 Richard Biener <rguenther@suse.de>
20379 PR middle-end/80509
20380 * passes.c (pass_manager::pass_manager): Initialize
20381 m_name_to_pass_map.
20383 2017-04-25 Richard Biener <rguenther@suse.de>
20385 PR tree-optimization/79201
20386 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20388 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20391 * config/s390/vector.md: Split MEM->GPR vector moves for
20392 non-s_operand addresses.
20394 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20397 * config/s390/predicates.md (reload_const_wide_int_operand): New
20399 * config/s390/s390.md ("movti"): Remove d/P alternative.
20400 ("movti_bigconst"): New pattern definition.
20402 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20405 * s390-protos.h (s390_expand_cs_hqi): Removed.
20406 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20407 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20408 modes as well as CCZ1mode and CCZmode.
20409 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20410 signature of s390_emit_compare_and_swap.
20411 (s390_expand_cs_hqi): Likewise, make static.
20412 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20413 compare-and-swap, in some cases.
20414 (s390_expand_cs): Wrapper function.
20415 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20417 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20418 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20419 patterns for small and large integers. Forbid symref memory operands.
20420 Move expander to s390.c. Require cc register.
20421 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20422 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20423 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20424 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20425 symref memory operands. Remove CC mode and call s390_match_ccmode
20427 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20429 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20431 * config/s390/s390.md (define_peephole2): New peephole to help
20432 combining the load-and-test pattern with volatile memory.
20434 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20436 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20437 with CCZmode for TARGET_Z196.
20439 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20441 PR rtl-optimization/80501
20442 * combine.c (make_compound_operation_int): Set subreg_code to SET
20443 even for AND with mask of the sign bit of mode.
20445 PR rtl-optimization/80500
20446 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20447 sum's initial value.
20449 2017-04-25 Julian Brown <julian@codesourcery.com>
20450 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20452 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20454 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20456 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20458 2017-04-25 Julian Brown <julian@codesourcery.com>
20459 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20461 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20462 (thunderx2t99_sha): New Reservation.
20464 2017-04-25 Julian Brown <julian@codesourcery.com>
20465 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20467 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20468 type for 1-element load.
20470 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20472 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20474 2017-04-24 Martin Jambor <mjambor@suse.cz>
20476 PR tree-optimization/80293
20477 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20478 char arrays not totally scalarizable if it is false.
20479 (analyze_all_variable_accesses): Pass correct value in the new
20480 parameter. Add a statistics counter.
20482 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20484 PR middle-end/79931
20485 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20487 2017-04-24 Richard Biener <rguenther@suse.de>
20489 PR tree-optimization/80494
20490 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20491 out for complex types.
20493 2017-04-24 Richard Biener <rguenther@suse.de>
20495 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20496 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20497 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20498 (DFS): Adjust and never fail.
20499 (sccvn_dom_walker::fail): Remove.
20500 (sccvn_dom_walker::before_dom_children): Adjust.
20501 (run_scc_vn): Likewise and never fail.
20502 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20503 (pass_fre::execute): Likewise.
20505 2017-04-24 Richard Biener <rguenther@suse.de>
20507 PR tree-optimization/79725
20508 * tree-ssa-sink.c (statement_sink_location): Return whether
20509 failure reason was zero uses. Move that check later.
20510 (sink_code_in_bb): Deal with zero uses by removing the stmt
20513 2017-04-24 Richard Biener <rguenther@suse.de>
20516 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20517 pointer-based references.
20519 2017-04-24 Richard Biener <rguenther@suse.de>
20522 * pass_manager.h (pass_manager::operator new): Remove.
20523 (pass_manager::operator delete): Likewise.
20524 * passes.c (pass_manager::operator new): Remove.
20525 (pass_manager::operator delete): Likewise.
20526 (pass_manager::pass_manager): Zero individual pass members.
20528 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20531 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20532 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20533 Check "XEXP (src, 1)" operand here.
20534 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20535 Check "XEXP (src, 1)" operand here.
20536 (dimode_scalar_chain::make_vector_copies): Detect count register
20537 of a shift instruction. Zero extend count register from QImode
20538 to DImode to satisfy vector shift pattern count operand predicate.
20539 Substitute vector shift count operand with a DImode copy.
20540 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20543 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20545 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20546 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20547 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20548 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20549 (UNSPEC_NOREX_MEM): Remove definition.
20551 2017-04-21 Richard Biener <rguenther@suse.de>
20553 PR tree-optimization/79547
20554 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20555 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20556 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20557 without any constraints.
20559 2017-04-21 Richard Biener <rguenther@suse.de>
20561 PR tree-optimization/78847
20562 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20564 2017-04-21 Richard Biener <rguenther@suse.de>
20566 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20567 (build_distinct_type_copy): Likewise.
20568 (build_variant_type_copy): Likewise.
20569 * tree.c (build_qualified_type): Pass down mem-stat info.
20570 (build_distinct_type_copy): Likewise.
20571 (build_variant_type_copy): Likewise.
20573 2017-04-21 Richard Biener <rguenther@suse.de>
20575 PR tree-optimization/80237
20576 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20578 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20579 for a simplified result.
20581 2016-04-21 Richard Biener <rguenther@suse.de>
20583 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20584 sth as strict as a simple_iv but a chrec without symbols and an
20585 operand defined in the loop we are peeling (and not some subloop).
20586 (propagate_constants_for_unrolling): Propagate all constants.
20588 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20591 * config/i386/i386.c (print_reg): Remove assert for disalowed
20592 regno values, call output_operand_lossage instead.
20594 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20597 * config/i386/constraints.md (Yc): New register constraint.
20598 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20599 Use Yc constraint for alternative 2 of operand 0. Remove
20600 preferred_for_speed attribute.
20602 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20604 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20605 lastprivate clauses in SIMT case.
20607 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20609 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20611 2017-04-20 Richard Biener <rguenther@suse.de>
20613 PR tree-optimization/57796
20614 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20615 as N scalar stores.
20616 (vect_model_load_cost): Cost gathers as N scalar loads.
20618 2017-04-20 Richard Biener <rguenther@suse.de>
20620 * ggc-page.c (ggc_allocated_p): Rename to ...
20621 (safe_lookup_page_table_entry): ... this and return the lookup
20623 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20625 2017-04-20 Richard Biener <rguenther@suse.de>
20627 PR tree-optimization/80453
20628 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20629 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20630 from the conditions.
20631 (vn_phi_eq): Pass them down.
20632 (vn_phi_lookup): Record them.
20633 (vn_phi_insert): Likewise.
20635 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20637 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20638 uninitialized variable warning to avoid buffer overrun.
20640 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20643 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20644 is suppressed for '{ 0 }' in C.
20646 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20648 * BASE-VER: Set to 8.0.0.
20650 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20652 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20653 priority .init_array and .fini_array section with SECTION_NOTYPE
20656 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20658 PR middle-end/80423
20659 * tree.h (build_array_type): Add typeless_storage default argument.
20660 * tree.c (type_cache_hasher::equal): Also compare
20661 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20662 (build_array_type): Add typeless_storage argument, set
20663 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20665 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20666 (build_array_type): Likewise. Add typeless_storage argument.
20668 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20669 Jakub Jelinek <jakub@redhat.com>
20671 PR tree-optimization/80426
20672 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20673 operation on symbolic operands, also compute the overflow for the
20674 invariant part when the operation degenerates into a negation.
20676 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20679 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20680 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20683 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20685 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20688 * config/avr/avr.c (tree.h): Include it.
20689 (cgraph.h): Include it.
20690 (avr_encode_section_info): Don't warn for uninitialized progmem
20691 variable if it's just an alias.
20693 2017-04-19 Richard Biener <rguenther@suse.de>
20696 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20697 when needed by AutoPGO.
20699 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20702 * doc/lto.texi: Remove an extra 'that'.
20704 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20706 PR rtl-optimization/80429
20707 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20708 are only used in debug insns.
20710 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20711 Vladimir Makarov <vmakarov@redhat.com>
20713 * config/sparc/predicates.md (input_operand): Add comment. Return
20714 true for any memory operand when LRA is in progress.
20715 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20717 2017-04-18 Jeff Law <law@redhat.com>
20720 * mips.md ({return,simple_return}_internal): Do not overwrite
20723 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20725 PR tree-optimization/80443
20726 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20727 instead of adding 1, subtract -1 and similarly instead of subtracting
20730 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20732 PR rtl-optimization/80357
20733 * haifa-sched.c (tmp_bitmap): New variable.
20734 (model_recompute): Handle duplicate use records.
20735 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20736 (free_global_sched_pressure_data): Free it.
20738 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20741 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20742 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20743 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20744 instead of SYSTEM_HEADER_DIR.
20746 2017-04-18 Jeff Law <law@redhat.com>
20748 PR middle-end/80422
20749 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20750 predecessors after walking up the insn chain.
20752 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20755 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20756 sizetype type into debug info.
20758 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20761 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20762 unneeded test for TARGET_UPPER_REGS_SF.
20763 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20765 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20768 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20769 instead of gsi_after_labels.
20771 2017-04-18 Jeff Law <law@redhat.com>
20773 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20777 2017-04-13 Jeff Law <law@redhat.com>
20778 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20779 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20781 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20784 * config/avr/avr.c (intl.h): Include it.
20785 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20787 2017-04-18 Martin Liska <mliska@suse.cz>
20789 PR gcov-profile/78783
20790 * gcov-tool.c (gcov_output_files): Validate that destination
20791 file is either removed by the tool or by a user.
20793 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20794 Guy Benyei <guybe@mellanox.com>
20796 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20797 block, and do not negate it, the stored id is already negative.
20799 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20801 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20803 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20806 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20807 masks of options that should be turned off if the VSX vector
20808 options are turned off.
20809 (OTHER_P8_VECTOR_MASKS): Likewise.
20810 (OTHER_VSX_VECTOR_MASKS): Likewise.
20811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20812 rs6000_disable_incompatible_switches to validate no type switches
20814 (rs6000_incompatible_switch): New function to disallow turning on
20815 other vector options if -mno-vsx, -mno-power8-vector, or
20816 -mno-power9-vector are specified.
20818 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20820 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20822 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20824 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20825 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20826 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20827 (ARG_POINTER_CFA_OFFSET): Likewise.
20829 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20831 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20832 conditions to take advantage of various optimizations.
20834 2017-04-13 Jeff Law <law@redhat.com>
20836 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20837 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20838 (zero_extendsidi2_dext): Likewise.
20840 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20843 * fold-const.c (fold_ternary_loc): Revert
20844 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20847 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20849 PR rtl-optimization/80343
20850 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20851 new scratch pseudo.
20853 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20856 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20857 to ubsan_encode_value.
20859 2017-04-13 Jeff Law <law@redhat.com>
20861 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20862 appearing in DEBUG_INSNs.
20864 2017-04-13 Martin Liska <mliska@suse.cz>
20866 PR gcov-profile/80413
20867 * gcov-io.c (gcov_write_string): Copy to buffer just when
20868 allocated size is greater than zero.
20870 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20873 * dwarf2out.c (decls_for_scope): Ignore declarations of
20874 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20876 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20879 * ipa-visibility.c (non_local_p): Fix typos.
20880 (localize_node): When localizing symbol in same comdat group,
20881 dissolve the group only when we know external symbols are going
20883 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20885 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20887 PR tree-optimization/79390
20888 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20889 order does not result in usable sequence, retry with reversed operand
20895 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20896 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20897 op0 instead of fold_convert_loc (loc, type, arg0).
20899 2017-04-12 Jeff Law <law@redhat.com>
20901 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20902 has a delay slot in the generated code.
20904 * config/cris/cris.md (cris_preferred_reload_class): Return
20905 GENNONACR_REGS rather than GENERAL_REGS.
20907 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20910 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20911 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20912 signedness of the result type.
20914 2017-04-12 Richard Biener <rguenther@suse.de>
20915 Jeff Law <law@redhat.com>
20917 PR tree-optimization/80359
20918 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20919 trim stores to TARGET_MEM_REFs.
20921 2017-04-12 Richard Biener <rguenther@suse.de>
20923 PR tree-optimization/79390
20924 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20925 threading case even more.
20927 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20930 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20931 for quad_address_p for TImode, instead of just not indexed_address.
20933 2017-04-12 Richard Biener <rguenther@suse.de>
20934 Bernd Edlinger <bernd.edlinger@hotmail.de>
20936 PR middle-end/79671
20937 * alias.c (component_uses_parent_alias_set_from): Handle
20938 TYPE_TYPELESS_STORAGE.
20939 (get_alias_set): Likewise.
20940 * tree-core.h (tree_type_common): Add typeless_storage flag.
20941 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20942 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20943 for types containing members with TYPE_TYPELESS_STORAGE.
20944 (place_field): Likewise.
20945 (layout_type): Likewise for ARRAY_TYPE.
20946 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20947 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20948 TYPE_TYPELESS_STORAGE.
20949 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20951 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20954 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20955 first argument to type.
20957 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20961 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20962 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20963 (rs6000_expand_binop_builtin): Likewise.
20964 (rs6000_expand_ternop_builtin): Likewise; also add missing
20965 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20966 vshasigma built-ins.
20967 * doc/extend.texi: Document that vec_xxpermdi's third argument
20968 must be a constant.
20970 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20972 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20973 Use shift_const cost parameter when calculating gain of STV shifts.
20975 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20977 PR rtl-optimization/70478
20978 * lra-constraints.c (process_alt_operands): Check memory for
20979 disfavoring memory insn operand.
20981 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20983 PR middle-end/80100
20984 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20985 left shift in unsigned HOST_WIDE_INT type.
20987 PR rtl-optimization/80385
20988 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20989 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20992 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20993 if they have any depend clauses.
20995 2017-04-11 Martin Liska <mliska@suse.cz>
20998 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20999 * ipa-split.c (split_function): Create a local comdat symbol
21000 if caller is in a comdat group.
21002 2017-04-11 Martin Liska <mliska@suse.cz>
21005 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21008 2017-04-11 Martin Sebor <msebor@redhat.com>
21010 PR middle-end/80364
21011 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21012 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21014 (directive::set_width, directive::set_precision, format_character):
21016 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21019 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21022 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21023 conflict, set target->arch_name instead of target->cpu_name.
21025 2017-04-11 Richard Biener <rguenther@suse.de>
21027 PR tree-optimization/80374
21028 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21029 build_zero_cst, remove fold_convertible_p check again.
21031 2017-04-11 Martin Liska <mliska@suse.cz>
21034 * ubsan.c (instrument_object_size): Do not instrument register
21037 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21040 * config/i386/i386-builtin-types.def
21041 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21042 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21043 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21044 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21045 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21046 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21047 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21048 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21049 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21050 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21051 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21052 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21053 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21054 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21055 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21056 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21057 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21058 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21059 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21060 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21061 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21062 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21063 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21064 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21065 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21066 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21067 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21068 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21069 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21070 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21071 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21072 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21073 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21074 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21075 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21076 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21077 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21078 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21079 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21080 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21081 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21082 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21083 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21084 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21085 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21086 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21088 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21089 flag to second_arg_count, handle 4 argument function type _COUNT
21090 aliases, handle second_arg_count on second argument rather than last.
21092 2017-04-10 Jeff Law <law@redhat.com>
21094 PR tree-optimization/80374
21095 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21096 record anything if we can not convert integer_zero_node to the
21099 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21102 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21103 Enhance special handling given to the TARGET_P9_MINMAX option in
21104 relation to certain other options.
21106 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21108 PR tree-optimization/80153
21109 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21110 remove POINTER_PLUS_EXPR's base part directly, rather than through
21113 2017-04-10 Richard Biener <rguenther@suse.de>
21114 Bin Cheng <bin.cheng@arm.com>
21116 PR tree-optimization/80153
21117 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21118 the first element of pointer type aff_tree. Build result expr in
21120 (add_elt_to_tree): Convert to type unconditionally. Remove other
21121 fold_convert calls.
21122 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21123 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21125 2017-04-10 Richard Biener <rguenther@suse.de>
21127 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21130 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21132 PR rtl-optimization/70478
21133 * lra-constraints.c (curr_small_class_check): New.
21134 (update_and_check_small_class_inputs): New.
21135 (process_alt_operands): Update curr_small_class_check. Disfavor
21136 alternative insn memory operands. Check available regs for small
21139 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21142 * config/mips/mips.opt (-mvirt): Update description.
21143 * doc/invoke.texi (-mvirt): Likewise.
21145 2017-04-10 Richard Biener <rguenther@suse.de>
21147 PR middle-end/80362
21148 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21149 looking for NEGATE_EXPR in -A / -B to A / B folding.
21151 2017-04-10 Martin Liska <mliska@suse.cz>
21153 PR gcov-profile/80224
21154 * gcov.c (print_usage): Fix usage string.
21155 (get_gcov_intermediate_filename): Remove.
21156 (output_gcov_file): Use both for normal and intermediate format.
21157 (generate_results): Do not initialize special file for
21158 intermediate format.
21160 2017-04-10 Richard Biener <rguenther@suse.de>
21162 PR tree-optimization/80304
21163 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21166 2017-04-10 Nathan Sidwell <nathan@acm.org>
21169 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21170 (rs6000_init_builtins): Use it.
21172 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21174 * config/arm/arm.md (<mrc>): Add mode to SET source.
21175 (<mrrc>): Likewise.
21177 2017-04-10 Richard Biener <rguenther@suse.de>
21179 PR middle-end/80344
21180 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21182 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21185 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21186 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21187 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21188 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21189 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21190 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21191 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21192 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21193 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21194 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21195 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21196 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21197 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21198 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21199 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21200 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21201 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21202 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21203 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21204 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21205 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21206 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21207 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21209 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21211 PR rtl-optimization/70478
21212 * lra-constraints.c: Reverse the last patch.
21214 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21216 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21217 Add comment for WCHAR_T.
21219 2017-04-08 Martin Liska <mliska@suse.cz>
21222 2017-04-07 Martin Liska <mliska@suse.cz>
21225 * ipa-split.c (split_function): Add function part to a same comdat
21228 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21231 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21233 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21235 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21236 (vec_pairing): ...to this.
21237 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21238 (rs6000_sched_init): Adjust for name change.
21239 (struct rs6000_sched_context): Likewise.
21240 (rs6000_init_sched_context): Likewise.
21241 (rs6000_set_sched_context): Likewise.
21243 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21249 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21251 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21252 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21253 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21255 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21257 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21259 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21261 PR rtl-optimization/70703
21262 * ira-color.c (update_conflict_hard_regno_costs): Use
21263 int64_t instead of HOST_WIDE_INT.
21265 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21267 PR rtl-optimization/70478
21268 * lra-constraints.c (process_alt_operands): Disfavor alternative
21269 insn memory operands.
21271 2017-04-07 Jeff Law <law@redhat.com>
21273 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21274 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21276 2017-04-07 Martin Liska <mliska@suse.cz>
21279 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21280 Show error message instead of an ICE.
21282 2017-04-07 Martin Liska <mliska@suse.cz>
21285 * ipa-split.c (split_function): Add function part to a same comdat
21288 2017-04-07 Richard Biener <rguenther@suse.de>
21290 PR middle-end/80341
21291 * tree.c (get_unwidened): Also handle ! for_type case for
21293 * convert.c (do_narrow): Split out from ...
21294 (convert_to_integer_1): ... here. Do not pass final truncation
21295 type to get_unwidened for TRUNC_DIV_EXPR.
21297 2017-04-07 Richard Biener <rguenther@suse.de>
21299 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21301 (aff_combination_const): Adjust.
21302 (aff_combination_scale): Likewise.
21303 (aff_combination_add_elt): Likewise.
21304 (aff_combination_add_cst): Likewise.
21305 (aff_combination_convert): Likewise.
21306 (add_elt_to_tree): Likewise. Remove unused argument.
21307 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21309 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21311 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21313 * config/arm/arm.c (arm_default_short_enums): Use
21314 ARM_DEFAULT_SHORT_ENUMS.
21315 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21317 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21320 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21321 members with redundant out-of-class redeclaration.
21323 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21326 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21327 * config/i386/i386.md (*zero_extendsidi2):
21328 Add (?*x,*x) and (?*v,*v) alternatives.
21330 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21333 * config/i386/i386.c (ix86_expand_builtin)
21334 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21335 mode from insn data. Convert operands to insn operand mode.
21336 Copy operands that don't satisfy insn predicate to a register.
21338 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21340 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21343 2017-04-06 Richard Biener <rguenther@suse.de>
21345 PR tree-optimization/80334
21346 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21347 preserve alignment of accesses.
21349 2017-04-06 Richard Biener <rguenther@suse.de>
21351 PR tree-optimization/80262
21352 * tree-sra.c (build_ref_for_offset): Preserve address-space
21354 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21355 Drop useless address-space information on MEM_REF offsets.
21357 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21359 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21361 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21363 PR rtl-optimization/70703
21364 * ira-color.c (update_conflict_hard_regno_costs): Use
21365 HOST_WIDE_INT instead of long.
21367 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21370 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21371 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21373 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21374 for x86_64 target. Handle -m3dnowa option.
21376 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21378 PR rtl-optimization/70703
21379 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21380 (update_conflict_hard_regno_costs): Use long instead of unsigned
21381 arithmetic for cost calculation.
21383 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21384 Bernd Edlinger <bernd.edlinger@hotmail.de>
21387 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21390 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21393 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21394 ptr_mode with Pmode throughout.
21395 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21396 into probe_stack_range and use DImode.
21398 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21401 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21402 call_eh_return is true.
21404 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21406 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21407 Initialize last_match_fntype_index.
21409 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21412 * tree-nvr.c: Include internal-fn.h.
21413 (pass_return_slot::execute): Ignore internal calls without
21416 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21417 Richard Biener <rguenther@suse.de>
21420 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21421 captures used multiple times, except for the last use.
21422 * generic-match-head.c: Include gimplify.h.
21424 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21426 PR tree-optimization/79390
21427 * target.h (struct noce_if_info): Declare.
21428 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21429 * target.def (noce_conversion_profitable_p): New target hook.
21430 * ifcvt.h (struct noce_if_info): New type, moved from ...
21431 * ifcvt.c (struct noce_if_info): ... here.
21432 (noce_conversion_profitable_p): Renamed to ...
21433 (default_noce_conversion_profitable_p): ... this. No longer
21435 (noce_try_store_flag_constants, noce_try_addcc,
21436 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21437 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21438 instead of noce_conversion_profitable_p.
21439 * config/i386/i386.c: Include ifcvt.h.
21440 (ix86_option_override_internal): Don't override
21441 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21442 (ix86_noce_conversion_profitable_p): New function.
21443 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21444 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21445 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21446 * doc/tm.texi: Regenerated.
21448 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21450 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21453 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21456 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21457 instructions for small multiply cores.
21459 2017-04-04 Jeff Law <law@redhat.com>
21461 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21463 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21464 that are not set by the loop over the elements.
21466 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21469 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21470 int mode, convert_modes it to mode as unsigned, otherwise use
21471 lowpart_subreg to mode rather than SImode.
21472 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21473 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21474 Use DImode instead of SImode for the shift count operand.
21475 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21478 2017-04-04 Richard Biener <rguenther@suse.de>
21480 PR middle-end/80281
21481 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21482 arithmetic done for the negate or the plus. Simplify.
21483 (A - (-B) -> A + B): Likewise.
21484 * fold-const.c (split_tree): Make sure to not negate pointers.
21486 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21488 PR rtl-optimization/60818
21489 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21490 a compare of comparisons with the thing compared if this results
21491 in a different machine mode.
21493 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21495 * alias.c (base_alias_check): Fix typo in comment.
21496 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21497 * cgraphunit.c (symbol_table::compile): Likewise.
21498 * collect2.c (maybe_run_lto_and_relink): Likewise.
21499 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21500 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21501 * config/avr/avr.c (avr_map_op_t): Likewise.
21502 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21503 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21504 * config/epiphany/epiphany.md (movcc): Likewise.
21505 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21506 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21508 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21509 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21510 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21511 * config/sh/sh.c (sh_rtx_costs): Likewise.
21512 * fold-const.c (fold_truth_andor): Likewise.
21513 * genautomata.c (collapse_flag): Likewise.
21514 * gengtype.h (struct type::u::s): Likewise.
21515 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21516 * input.c (FORMAT_AMOUNT): Likewise.
21517 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21518 (known_aggs_to_agg_replacement_list): Likewise.
21519 * ipa-inline-analysis.c: Likewise.
21520 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21521 * ipa-polymorphic-call.c
21522 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21523 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21524 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21526 * modulo-sched.c (apply_reg_moves): Likewise.
21527 * omp-expand.c (build_omp_regions_1): Likewise.
21528 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21529 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21530 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21531 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21532 * value-prof.c: Likewise.
21533 * var-tracking.c (val_reset): Likewise.
21535 2017-04-03 Richard Biener <rguenther@suse.de>
21537 PR tree-optimization/80275
21538 * fold-const.c (split_address_to_core_and_offset): Handle
21541 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21543 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21544 descriptors is at least equal to that of functions.
21546 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21548 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21550 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21553 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21554 (mov<IMOD4:mode>): New expander.
21555 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21557 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21559 PR rtl-optimization/79405
21560 * fwprop.c (propagations_left): New variable.
21561 (forward_propagate_into): Decrement it.
21562 (fwprop_init): Initialize it.
21563 (fw_prop): If the variable has reached zero, stop propagating.
21564 (fwprop_addr): Ditto.
21566 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21569 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21570 a FUNCTION_DECL, pass it as decl instead of origin to
21573 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21575 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21578 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21581 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21582 TARGET_VSX_SMALL_INTEGER.
21584 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21586 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21587 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21589 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21591 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21592 extraction from odd-numbered MSA register.
21594 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21596 PR middle-end/80173
21597 * expmed.c (store_bit_field_1): Don't attempt to create
21598 a word subreg out of hard registers wider than word if they
21599 have HARD_REGNO_NREGS of 1 for their mode.
21601 PR middle-end/80163
21602 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21603 conversions to integer types wider than word and pointer.
21606 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21607 (rtx_equal_for_cselib_p): Pass 0 to it.
21608 * cselib.c (cselib_hasher::equal): Likewise.
21609 (rtx_equal_for_cselib_1): Add depth argument. If depth
21610 is 128, don't look up VALUE locs and punt. Increment
21611 depth in recursive calls when walking VALUE locs.
21613 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21615 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21616 (make_gcov_file_name): Use the canonical path name for generating
21618 (read_line): Fix handling of files with ascii null bytes.
21620 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21622 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21623 to initialise a vector register instead
21624 of using a const_int.
21626 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21628 PR translation/80189
21629 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21630 diagnostic messages.
21632 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21635 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21636 (dfp_diex_<mode>): Update mode of operand 1.
21637 * doc/extend.texi (dxex, dxexq): Document change to return type.
21638 (diex, diexq): Document change to argument type.
21640 2017-03-30 Martin Jambor <mjambor@suse.cz>
21643 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21644 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21645 it reflects the signature changes performed at the callee side.
21646 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21647 to cgraph_build_function_type_skip_args.
21648 (build_function_decl_skip_args): Adjust call to the above function.
21650 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21653 * config/i386/sse.md
21654 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21655 register as dest whenever it is a MEM not rtx_equal_p to the
21656 corresponding dup operand, and when forcing into reg move the
21657 reg into the memory afterwards.
21658 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21659 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21660 for the force_reg mode.
21661 (avx512vl_vextractf128<mode>): Use register as dest either
21662 always when a MEM, or when it is a MEM not rtx_equal_p to the
21663 corresponding dup operand, or even not when it is a CONST_VECTOR
21664 depending on the mode and lo vs. hi.
21665 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21667 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21668 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21669 Likewise. Require that operands[2] is even.
21670 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21671 Remove extraneous parens. Require that operands[2] is a multiple
21673 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21674 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21675 disallow memory then.
21677 2017-03-30 Richard Biener <rguenther@suse.de>
21679 PR tree-optimization/77498
21680 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21681 to non-constants over backedges.
21683 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21685 PR rtl-optimization/80233
21686 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21687 as last_combined_insn. Do not test for BARRIER_P separately.
21689 2017-03-29 Andreas Schwab <schwab@suse.de>
21692 * calls.c (prepare_call_address): Convert funexp to Pmode before
21693 copying to temp reg.
21695 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21697 PR tree-optimization/80158
21698 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21699 Handle possible future case of more than one alternate
21701 (replace_rhs_if_not_dup): Likewise.
21702 (replace_one_candidate): Likewise.
21704 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21706 PR rtl-optimization/80193
21707 * ira.c (ira): Do not check allocation for LRA.
21709 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21711 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21712 (nvptx_output_simt_exit): Declare.
21713 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21714 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21715 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21716 (nvptx_declare_function_name): ...here. Emit declaration of local
21717 memory space buffer for omp_simt_enter insn.
21718 (nvptx_output_unisimt_switch): New.
21719 (nvptx_output_softstack_switch): New.
21720 (nvptx_output_simt_enter): New.
21721 (nvptx_output_simt_exit): New.
21722 * config/nvptx/nvptx.h (struct machine_function): New fields
21723 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21724 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21725 (UNSPECV_SIMT_EXIT): Ditto.
21726 (omp_simt_enter_insn): New insn.
21727 (omp_simt_enter): New expansion.
21728 (omp_simt_exit): New insn.
21729 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21731 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21732 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21733 (expand_GOMP_SIMT_EXIT): New.
21734 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21735 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21736 (GOMP_SIMT_EXIT): Ditto.
21737 * target-insns.def (omp_simt_enter): New insn.
21738 (omp_simt_exit): Ditto.
21739 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21741 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21742 (lower_rec_input_clauses): Likewise.
21743 (lower_lastprivate_clauses): Handle SIMT privatization.
21745 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21746 (ompdevlow_adjust_simt_enter): New.
21747 (find_simtpriv_var_op): New.
21748 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21749 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21751 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21752 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21753 (copy_decl_for_dup_finish): Ditto.
21755 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21757 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21760 * config/i386/i386.c (ix86_option_override_internal): Always
21761 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21763 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21765 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21767 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21769 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21770 mark new edge's irreducible flag accordign to it.
21771 (vect_do_peeling): Check loop preheader edge's irreducible flag
21772 and pass it to function slpeel_add_loop_guard.
21774 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21776 PR tree-optimization/80218
21777 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21778 Update block frequencies and counts.
21780 2017-03-28 Richard Biener <rguenther@suse.de>
21782 PR tree-optimization/78644
21783 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21784 of a simplification result we may not use it at all.
21786 2017-03-28 Richard Biener <rguenther@suse.de>
21789 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21790 without arguments, generate default definition of a SSA name.
21792 2017-03-28 Richard Biener <rguenther@suse.de>
21794 PR middle-end/80222
21795 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21796 TYPE_REF_CAN_ALIAS_ALL references.
21797 * fold-const.c (fold_indirect_ref_1): Likewise.
21799 2017-03-28 Martin Liska <mliska@suse.cz>
21802 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21803 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21805 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21806 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21808 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21809 (EXTRA_SPECS): Define.
21810 (SUBTARGET_EXTRA_SPECS): Likewise.
21811 (SUBTARGET_CPP_SPEC): Likewise.
21812 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21813 SUBTARGET_EXTRA_SPECS.
21814 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21816 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21818 * config/arc/simdext.md (vst64_insn): Update pattern.
21819 (vld32wh_insn): Likewise.
21820 (vld32wl_insn): Likewise.
21821 (vld64_insn): Likewise.
21822 (vld32_insn): Likewise.
21824 2017-03-28 Marek Polacek <polacek@redhat.com>
21827 * fold-const.c (fold_comparison): Use protected_set_expr_location
21828 instead of SET_EXPR_LOCATION.
21830 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21832 * tree.c (add_expr): Avoid name lookup warning.
21834 2017-03-27 Jeff Law <law@redhat.com>
21836 PR tree-optimization/80216
21837 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21838 function name. Limit recursion depth.
21839 (record_temporary_equivalences): Corresponding changes.
21841 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21843 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21846 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21849 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21851 * cfgcleanup.c (reg_note_cfa_p): New array.
21852 (insns_have_identical_cfa_notes): New function.
21853 (old_insns_match_p): Don't cross-jump in between /f
21854 and non-/f instructions. If both i1 and i2 are frame related,
21855 verify all CFA notes, their order and content.
21857 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21860 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21861 HImode and SImode with zero extend to DImode to one insn.
21862 (bswap<mode>2_extenddi): Likewise.
21863 (bswapsi2_extenddi): Likewise.
21864 (bswaphi2_extendsi): Likewise.
21865 (bswaphi2): Combine bswap HImode and SImode into one insn.
21866 Separate memory insns from swapping register.
21867 (bswapsi2): Likewise.
21868 (bswap<mode>2): Likewise.
21869 (bswaphi2_internal): Delete, no longer used.
21870 (bswapsi2_internal): Likewise.
21871 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21872 store, and gpr<-gpr swap insns.
21873 (bswap<mode>2_store): Likewise.
21874 (bswaphi2_reg): Register only splitter, combine with the splitter.
21875 (bswaphi2 splitter): Likewise.
21876 (bswapsi2_reg): Likewise.
21877 (bswapsi2 splitter): Likewise.
21878 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21879 the insns into load, store, and register/register insns.
21880 (bswapdi2_ldbrx): Likewise.
21881 (bswapdi2_load): Likewise.
21882 (bswapdi2_store): Likewise.
21883 (bswapdi2_reg): Likewise.
21885 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21887 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21888 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21890 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21893 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21896 special handling for target option conflicts between dform
21897 options (-mpower9-dform, -mpower9-dform-vector,
21898 -mpower9-dform-scalar) and -mno-direct-move.
21900 2017-03-27 Richard Biener <rguenther@suse.de>
21902 PR tree-optimization/80181
21903 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21905 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21907 * config/arc/predicates.md (move_double_src_operand): Replace the
21908 call to move_double_src_operand with a call to address_operand.
21910 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21912 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21913 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21914 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21916 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21918 * config/arc/predicates.md (long_immediate_loadstore_operand):
21919 Consider scaled addresses cases.
21921 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21923 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21924 restored when in interrupt.
21925 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21926 doesn't have delay slot.
21928 2017-03-27 Richard Biener <rguenther@suse.de>
21931 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21932 inlined thunk clones.
21934 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21937 * asan.c (instrument_derefs): Copy over last operand from
21938 original COMPONENT_REF to the new COMPONENT_REF with
21939 DECL_BIT_FIELD_REPRESENTATIVE.
21940 * ubsan.c (instrument_object_size): Likewise.
21942 2017-03-27 Richard Biener <rguenther@suse.de>
21944 PR tree-optimization/80170
21945 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21946 sure DR/SCEV didnt fold in constants we do not see when looking
21947 at the reference base alignment.
21949 2017-03-27 Richard Biener <rguenther@suse.de>
21951 PR middle-end/80171
21952 * gimple-fold.c (fold_ctor_reference): Properly guard against
21953 NULL return value from canonicalize_constructor_val.
21955 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21958 * config/i386/i386.c (ix86_expand_builtin)
21959 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21960 flags reg setting and flags reg using instructions.
21961 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21962 clobbering instructions to zero extend op2.
21964 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21966 * doc/install.texi (Configuration) <--with-aix-soname>:
21967 Update link to AIX ld.
21969 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21971 PR rtl-optimization/80160
21972 PR rtl-optimization/80159
21973 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21974 reg_alternate_class into account.
21976 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21979 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21980 to consider in curr_insn_transform.
21982 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21984 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21985 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21986 and emit_mode_inner.
21988 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21990 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21991 argument to the overloaded builtin variants. Use the new flag to
21992 deprecate certain builtin variants.
21993 * config/s390/s390-builtin-types.def: Add new builtin types.
21994 * config/s390/s390-builtins.h: Support new flags field for
21995 overloaded builtins.
21996 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21997 (s390_macro_to_expand): Enable vector float data type.
21998 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21999 builtins by incrementing the __VEC__ version number.
22000 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22002 (s390_resolve_overloaded_builtin): Emit error messages depending
22003 on the builtin flags.
22004 * config/s390/s390.c (s390_expand_builtin): Support additional
22005 flags argument. Change error message to match the messages
22006 emitted in s390-c.c.
22007 * config/s390/s390.md: New UNSPEC_* constants.
22008 (op_type): Add new instruction types.
22009 * config/s390/vecintrin.h: Add new builtins and test data class
22011 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22012 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22013 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22014 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22015 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22016 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22018 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22019 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22020 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22021 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22023 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22024 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22025 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22026 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22028 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22029 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22030 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22031 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22032 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22033 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22034 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22036 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22037 ("vec_scatter_element<V_HW_4:mode>_DI")
22038 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22039 ("vec_fpint<mode>", "vflls")
22040 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22041 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22042 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22043 ("*vec_cmphe<mode>_cc"): ... these.
22045 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22046 mode constant instead of magic value.
22048 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22050 * config/s390/s390.c (s390_expand_vec_compare): Support other
22051 vector floating point modes than just V2DF.
22052 (s390_expand_vcond): Likewise.
22053 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22054 (s390_cannot_change_mode_class): Prevent mode changes between TF
22055 and V1TF in vector registers.
22056 * config/s390/s390.md (DF, SF): New mode attributes.
22057 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22058 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22059 SFmode support for VRs.
22060 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22062 (VFT, VF_HW): New mode iterators.
22063 (vw, sdx): New mode attributes.
22064 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22065 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22066 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22067 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22068 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22069 also the new vector floating point modes. Renaming to ...
22071 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22072 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22073 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22074 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22075 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22076 ("vec_unordered<mode>"): ... these.
22078 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22079 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22080 ("*vec_extendv2df"): New insn definitions.
22082 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22084 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22085 ("mulditi3_2", "*muldi3_sign"): New patterns.
22086 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22087 rename the pattern definition.
22089 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22091 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22093 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22095 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22097 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22098 instruction if possible.
22099 * config/s390/vector.md (vec_halfnumelts): New mode
22101 ("*vec_vllezlf<mode>"): New pattern.
22103 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22105 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22106 ("popcountv4si2", "popcountv2di2"): Rename to ...
22107 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22108 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22110 ("popcount<mode>2_vxe"): New pattern.
22112 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22114 * common/config/s390/s390-common.c (processor_flags_table): Add
22116 * config.gcc: Add arch12.
22117 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22118 Default to arch12 for unknown CPU model numbers.
22119 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22120 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22121 PROCESSOR_max sanity check.
22122 * config/s390/s390-opts.h (enum processor_type): Add
22124 * config/s390/s390.c (processor_table): Add arch12.
22125 (s390_expand_builtin): Add check for B_VXE flag.
22126 (s390_issue_rate): Add PROCESSOR_ARCH12.
22127 (s390_get_sched_attrmask): Likewise.
22128 (s390_get_unit_mask): Likewise.
22129 (s390_sched_score): Enable z13 scheduling for arch12.
22130 (s390_sched_reorder): Likewise.
22131 (s390_sched_variable_issue): Likewise.
22132 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22134 (s390_tune_attr): Use z13 scheduling also for arch12.
22135 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22136 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22137 (TARGET_VXE_P): New macros.
22138 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22139 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22140 * config/s390/s390.opt: Add arch12 as processor_type.
22142 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22144 * config/s390/s390.md
22145 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22146 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22147 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22149 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22150 Rename expanders to ...
22152 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22153 ("fixuns_truncdddi2_emu"): ... these.
22155 ("fixuns_trunc<mode>si2_emu"): New expander.
22157 ("*fixuns_truncdfdi2_z13"): Rename to ...
22158 ("*fixuns_truncdfdi2_vx"): ... this.
22160 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22162 * config/s390/2964.md: Remove the single element vector compare
22163 instructions which are no longer used.
22164 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22166 (s390_canonicalize_comparison): Remove handling of DFmode
22168 (s390_expand_vec_compare_scalar): Remove function.
22169 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22170 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22172 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22176 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22178 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22179 will anyway by matched by mov<mode>_64dfp.
22181 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22183 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22184 vlef/vstef. Add missing operand to vleif.
22186 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22188 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22189 pair for all vector types with 64 bit elements.
22190 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22191 * config/s390/vector.md (V_HW_64): ... here.
22192 (V_128_NOSINGLE): New mode iterator.
22193 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22194 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22195 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22196 ("*vec_load_pairv2di"): Change to ...
22197 ("*vec_load_pair<mode>"): ... this one.
22199 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22201 * config/s390/constraints.md: Add comments.
22202 (jKK): Reject element sizes > 8 bytes.
22203 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22205 * config/s390/s390.md: Add the s_operand checks formerly in
22206 s390_split_ok_p to various splitters where they are still
22208 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22209 for 128 bit vectors. Plus two splitters.
22211 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22213 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22216 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22219 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22220 error if the boundary argument is not constant.
22222 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22224 PR rtl-optimization/80112
22225 * loop-doloop.c (doloop_condition_get): Don't check condition
22226 if cmp isn't SET with IF_THEN_ELSE src.
22228 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22230 PR tree-optimization/80158
22231 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22232 replacing a candidate statement, also replace it for the
22233 candidate's alternate interpretation.
22234 (replace_rhs_if_not_dup): Likewise.
22235 (replace_one_candidate): Likewise.
22237 2017-03-24 Richard Biener <rguenther@suse.de>
22239 PR tree-optimization/80167
22240 * graphite-isl-ast-to-gimple.c
22241 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22243 (translate_isl_ast_to_gimple::get_rename): Likewise.
22245 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22247 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22248 handling of certain combinations of target options, including the
22249 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22250 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22252 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22255 * config/arm/arm.md (*load_multiple): Add reload_completed to
22256 matching condition.
22258 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22259 Richard Biener <rguenth@suse.de>
22261 PR tree-optimization/79908
22262 PR tree-optimization/80136
22263 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22264 been cast away, gimplify_and_add suffices.
22266 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22268 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22270 2017-03-23 Richard Biener <rguenther@suse.de>
22272 PR tree-optimization/80032
22273 * gimplify.c (gimple_push_cleanup): Forced unconditional
22274 cleanups still have to go to the conditional_cleanups
22277 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22279 PR tree-optimization/80072
22280 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22282 (next_operand_entry_id): Change type to unsigned int.
22283 (sort_by_operand_rank): Make sure to return the right return value
22284 even if unsigned fields are bigger than INT_MAX.
22285 (struct oecount): Change cnt and id type to unsigned int.
22286 (oecount_hasher::equal): Formatting fix.
22287 (oecount_cmp): Make sure to return the right return value
22288 even if unsigned fields are bigger than INT_MAX.
22289 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22292 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22293 TREE_READONLY on result if writing it more than once.
22296 * doc/invoke.texi (-fsanitize=thread): Document that with
22297 -fnon-call-exceptions atomics are not able to throw
22301 * tsan.c: Include tree-eh.h.
22302 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22303 maybe_clean_or_replace_eh_stmt where needed.
22304 (instrument_memory_accesses): Add cfg_changed argument.
22305 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22306 if it returned true.
22307 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22309 PR rtl-optimization/63191
22310 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22311 wrapper function, moved the whole old content into ...
22312 (ix86_delegitimize_address_1): ... this. New inline function.
22313 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22314 true as last argument instead of ix86_delegitimize_address.
22316 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22318 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22319 cortexa57_branch_cost.
22321 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22323 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22325 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22328 * doc/md.texi (Constraints): Document wA constraint.
22329 * config/rs6000/constraints.md (wA): New.
22330 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22331 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22332 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22333 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22335 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22338 * gimplify.c (is_oacc_declared): New function.
22339 (oacc_default_clause): Use it to set default flags for acc declared
22340 variables inside parallel regions.
22341 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22342 declared variables.
22343 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22344 declare attribute to any decl as necessary.
22346 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22349 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22350 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22351 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22352 (arm_arch_lpae): This.
22353 * config/arm/arm.c (arm_arch7ve): Rename into ...
22354 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22355 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22358 2017-03-22 Martin Liska <mliska@suse.cz>
22361 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22362 error message instead of an ICE.
22364 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22366 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22368 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22370 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22372 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22375 2017-03-21 Martin Sebor <msebor@redhat.com>
22377 * doc/extend.texi: Use "cannot" instead of "can't."
22378 * doc/hostconfig.texi: Same.
22379 * doc/install.texi: Same.
22380 * doc/invoke.texi: Same.
22381 * doc/loop.texi: Same.
22382 * doc/md.texi: Same.
22383 * doc/objc.texi: Same.
22384 * doc/rtl.texi: Same.
22385 * doc/tm.texi: Same.
22386 * doc/tm.texi.in: Same.
22387 * doc/trouble.texi: Same.
22389 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22392 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22393 (collect_checksum_attributes): Set it.
22394 (die_checksum_ordered): Use it.
22396 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22398 PR tree-optimization/79908
22399 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22400 change: For a VA_ARG whose LHS has been cast away, use
22401 force_gimple_operand to construct the side effects.
22403 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22405 PR translation/80001
22406 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22407 more amenable to translation.
22408 (oacc_loop_auto_partitions): Likewise.
22410 2017-03-21 Marek Polacek <polacek@redhat.com>
22411 Martin Sebor <msebor@redhat.com>
22413 PR tree-optimization/80109
22414 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22415 on INTEGRAL_TYPE_P.
22417 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22418 Segher Boessenkool <segher@kernel.crashing.org>
22421 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22422 check reg_used_between_p between insn and one of succ or succ2
22423 depending on if succ is artificial insn not inserted into insn
22426 2017-03-21 Martin Liska <mliska@suse.cz>
22428 PR gcov-profile/80081
22429 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22430 * doc/gcc.texi: Include gcov-dump stuff.
22431 * doc/gcov-dump.texi: New file.
22433 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22435 PR rtl-optimization/79150
22436 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22437 conditional jump, if the jump is the last insn of the loop.
22439 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22440 Richard Biener <rguenth@suse.de>
22442 PR tree-optimization/79908
22443 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22444 been cast away, use force_gimple_operand to construct the side
22447 2017-03-21 Martin Liska <mliska@suse.cz>
22449 PR libfortran/79956
22450 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22453 2017-03-21 Brad Spengler <spender@grsecurity.net>
22456 * plugin.c (htab_hash_plugin): New function.
22457 (add_new_plugin): Use it and adjust.
22458 (parse_plugin_arg_opt): Adjust.
22459 (init_one_plugin): Likewise.
22461 2017-03-21 Richard Biener <rguenther@suse.de>
22463 PR tree-optimization/80032
22464 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22465 if set force the cleanup to happen unconditionally.
22466 (gimplify_target_expr): Push inserted clobbers with force_uncond
22467 to avoid them being removed by control-dependent DCE.
22469 2017-03-21 Richard Biener <rguenther@suse.de>
22471 PR tree-optimization/80122
22472 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22473 va_arg_pack_len when the inlined call stmt requires pack
22475 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22477 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22480 * tsan.c (instrument_builtin_call): If the memory model argument
22481 is not a constant, assume it is valid.
22484 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22487 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22489 PR rtl-optimization/79910
22490 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22491 if its dest is used by an insn before I2 (other than the combined
22492 insns themselves, which are properly handled already).
22494 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22497 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22499 * combine.c (record_used_regs): New static function.
22500 (try_combine): Handle situations where there is an additional
22501 instruction between I2 and I3 which needs to have a LOG_LINK
22505 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22507 * combine.c (try_combine): Delete redundant i1 test. Call
22508 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22510 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22513 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22514 alternatives 13/14.
22516 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22518 PR tree-optimization/80054
22519 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22520 the optimization if a PHI or any of its arguments is not dominated
22521 by the candidate's basis. Use gphi* rather than gimple* as
22523 (replace_profitable_candidates): Clean up a gimple* variable that
22524 should be a gphi* variable.
22526 2017-03-20 Martin Sebor <msebor@redhat.com>
22529 * doc/extend.texi (attribute constructor): Document present limitation.
22531 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22534 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22535 __POWER9_VECTOR__ #ifdef control, change template definition to
22536 use Power9-specific built-in function.
22537 (vec_any_eq): Likewise.
22538 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22539 to control outcomes from this test.
22540 (vector_ae_<mode>p): For VEC_F modes, likewise.
22542 2017-03-20 Ian Lance Taylor <iant@google.com>
22544 * config/i386/i386.c (ix86_function_regparm): Save an extra
22545 register for -fsplit-stack with DECL_STATIC_CHAIN.
22547 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22550 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22551 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22553 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22555 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22557 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22560 2017-03-20 Marek Polacek <polacek@redhat.com>
22563 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22565 2017-03-20 Richard Biener <rguenther@suse.de>
22567 PR tree-optimization/80113
22568 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22569 allocate extra SSA name for PHI def.
22570 (add_close_phis_to_outer_loops): Likewise.
22571 (add_close_phis_to_merge_points): Likewise.
22572 (copy_loop_close_phi_args): Likewise.
22573 (copy_cond_phi_nodes): Likewise.
22575 2017-03-20 Martin Liska <mliska@suse.cz>
22577 PR middle-end/79753
22578 * tree-chkp.c (chkp_build_returned_bound): Do not build
22579 returned bounds for a LHS that's not a BOUNDED_P type.
22581 2017-03-20 Martin Liska <mliska@suse.cz>
22585 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22586 COMPLEX_CST and VECTOR_CST.
22588 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22591 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22592 target operand. A new splitter adds the clobber statement in case
22593 the target operand is dead anyway.
22595 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22597 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22598 to age-old versions of binutils and glibc.
22600 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22602 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22604 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22606 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22608 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22610 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22611 requirement for binutils 2.13.
22613 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22615 * combine.c (try_combine): Delete redundant i1 test. Call
22616 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22618 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22620 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22621 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22623 <riscv64-*-elf>: Re-arrange section
22624 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22625 <riscv32-*-linux>: Likewise.
22626 <riscv64-*-elf>: Likewise
22627 <riscv64-*-linux>: Likewise.
22629 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22632 * aarch64.opt(verbose-cost-dump): Fix typo.
22634 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22637 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22638 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22640 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22642 * reload.c (find_reloads): When reloading a nonoffsettable address,
22643 use RELOAD_OTHER for it and its address reloads.
22645 PR rtl-optimization/79910
22646 * combine.c (record_used_regs): New static function.
22647 (try_combine): Handle situations where there is an additional
22648 instruction between I2 and I3 which needs to have a LOG_LINK
22651 2017-03-17 Jeff Law <law@redhat.com>
22653 PR tree-optimization/71437
22654 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22655 conditional in the hash table first.
22656 (vrp_dom_walker::before_dom_children): Extract condition from
22657 ASSERT_EXPR. Record condition, its inverion and any implied
22658 conditions as well.
22660 2017-03-17 Marek Polacek <polacek@redhat.com>
22661 Markus Trippelsdorf <markus@trippelsdorf.de>
22663 PR tree-optimization/80079
22664 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22667 2017-03-17 Richard Biener <rguenther@suse.de>
22669 PR middle-end/80075
22670 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22671 Properly verify the LHS before the RHS possibly claims to be
22673 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22676 2017-03-17 Martin Jambor <mjambor@suse.cz>
22678 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22679 (List of -O2 options): Likewise.
22680 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22683 2017-03-17 Tom de Vries <tom@codesourcery.com>
22685 * gcov-dump.c (print_usage): Print bug_report_url.
22687 2017-03-17 Richard Biener <rguenther@suse.de>
22689 PR middle-end/80050
22690 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22691 (parser::peek): Likewise.
22693 2017-03-17 Richard Biener <rguenther@suse.de>
22695 PR tree-optimization/80048
22696 * sese.c (free_sese_info): Properly release rename_map and
22697 copied_bb_map elements.
22699 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22701 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22702 Add linked-list forward and backlinks. Insert on
22703 construction, remove on destruction.
22704 (class pass_store_merging): Add m_stores_head field.
22705 (pass_store_merging::terminate_and_process_all_chains):
22706 Iterate over m_stores_head list.
22707 (pass_store_merging::terminate_all_aliasing_chains):
22709 (pass_store_merging::execute): Check for debug stmts first.
22710 Push new chains onto the m_stores_head stack.
22712 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22715 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22716 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22717 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22719 2017-03-16 Jeff Law <law@redhat.com>
22721 PR tree-optimization/71437
22722 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22723 member function. Implementation moved into after_dom_children
22724 member function and into the threader's thread_outgoing_edges
22726 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22727 some code into new thread_outgoing_edges.
22728 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22729 definition. Simplify marker handling (do it here). Assume we always
22730 have the available expression and the const/copies tables.
22731 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22733 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22734 * tree-vrp.c (equiv_stack): No longer file scoped.
22735 (vrp_dom_walker): New class.
22736 (vrp_dom_walker::before_dom_children): New member function.
22737 (vrp_dom_walker::after_dom_children): Likewise.
22738 (identify_jump_threads): Setup domwalker. Use it rather than
22739 walking edges in a random order by hand. Simplify setup/finalization.
22740 (finalize_jump_threads): Remove.
22741 (vrp_finalize): Do not call identify_jump_threads here.
22742 (execute_vrp): Do it here instead and call thread_through_all_blocks
22745 PR tree-optimization/71437
22746 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22748 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22750 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22751 (dom_opt_dom_walker::thread_across_edge): Remove
22752 handle_dominating_asserts argument. All callers changed.
22753 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22754 changes. Remove calls to lhs_of_dominating_assert. Other
22755 uses of handle_dominating_asserts turn into unconditional code
22756 (simplify_control_stmt_condition_1): Likewise.
22757 (simplify_control_stmt_condition): Likewise.
22758 (thread_through_normal_block, thread_across_edge): Likewise.
22759 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22760 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22761 object if it is not an SSA_NAME.
22762 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22763 before calling into the VRP specific simplifiers.
22764 (identify_jump_threads): Remove handle_dominating_asserts
22767 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22770 * tree-diagnostic.c (default_tree_printer): No longer static.
22771 * tree-diagnostic.h (default_tree_printer): New prototype.
22773 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22775 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22776 Change ins into fmov.
22778 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22780 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22781 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22782 Use h_con constraint for operand 1.
22783 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22784 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22786 2017-03-15 Jeff Law <law@redhat.com>
22788 PR tree-optimization/71437
22789 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22790 (record_temporary_equivalences): Use it.
22792 PR tree-optimization/71437
22793 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22794 tree-ssa-scopedtables.
22795 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22796 (record_conditions, record_cond, vuse_eq): Likewise.
22797 (record_edge_info): Adjust to API tweak of record_conditions.
22798 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22799 (record_temporary_equivalences, optimize_stmt): Likewise.
22800 (eliminate_redundant_computations): Likewise.
22801 (record_equivalences_from_stmt): Likewise.
22802 * tree-ssa-scopedtables.c: Include options.h and params.h.
22803 (vuse_eq): New function, moved from tree-ssa-dom.c
22804 (build_and_record_new_cond): Likewise.
22805 (record_conditions): Likewise. Accept vector of conditions rather
22806 than edge_equivalence structure for first argument.
22807 for the first argument.
22808 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22809 from tree-ssa-dom.c.
22810 (avail_exprs_stack::record_cond): Likewise.
22811 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22812 from tree-ssa-dom.c.
22813 (avail_exprs_stack): Add new member functions lookup_avail_expr
22815 (record_conditions): Declare.
22817 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22820 * lra-constraints.c (process_alt_operands): Increase reject for
22821 reloading an input/output operand.
22823 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22826 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22827 insns to convert from signed/unsigned char/short to IEEE 128-bit
22829 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22831 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22834 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22835 subreg of inner mode for values already in registers.
22837 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22839 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22840 iteration reg is used after the loop.
22842 2017-03-14 Martin Sebor <msebor@redhat.com>
22844 PR tree-optimization/79800
22845 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22846 precision in negative-positive range.
22847 (format_floating): Call non-const overload with adjusted precision.
22849 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22852 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22855 2017-03-14 Martin Sebor <msebor@redhat.com>
22857 PR middle-end/80020
22858 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22859 * builtins.def (aligned_alloc): Use it.
22862 * Makefile.in (GTFILES): Add calls.c.
22863 * calls.c: Include "gt-calls.h".
22865 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22867 PR rtl-optimization/79728
22868 * regs.h (struct target_regs): New field
22869 x_contains_allocatable_regs_of_mode.
22870 (contains_allocatable_regs_of_mode): New macro.
22871 * reginfo.c (init_reg_sets_1): Initialize it, and change
22872 contains_reg_of_mode so it includes global regs as well.
22873 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22874 rather than contains_regs_of_mode.
22876 2017-03-14 Martin Liska <mliska@suse.cz>
22878 * doc/invoke.texi: Document options that can't be combined with
22879 -fcheck-pointer-bounds.
22881 2017-03-14 Martin Liska <mliska@suse.cz>
22883 PR middle-end/79831
22884 * doc/invoke.texi (-Wchkp): Document the option.
22886 2017-03-14 Martin Liska <mliska@suse.cz>
22888 * Makefile.in: Install gcov-dump.
22890 2017-03-14 Martin Liska <mliska@suse.cz>
22892 * multiple_target.c (expand_target_clones): Bail out for
22893 an invalid attribute.
22895 2017-03-14 Richard Biener <rguenther@suse.de>
22897 * alias.c (struct alias_set_entry): Pack properly.
22898 * cfgloop.h (struct loop): Likewise.
22899 * cse.c (struct set): Likewise.
22900 * ipa-utils.c (struct searchc_env): Likewise.
22901 * loop-invariant.c (struct invariant): Likewise.
22902 * lra-remat.c (struct cand): Likewise.
22903 * recog.c (struct change_t): Likewise.
22904 * rtl.h (struct address_info): Likewise.
22905 * symbol-summary.h (function_summary): Likewise.
22906 * tree-loop-distribution.c (struct partition): Likewise.
22907 * tree-object-size.c (struct object_size_info): Likewise.
22908 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22909 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22910 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22911 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22912 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22913 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22914 (struct _stmt_vec_info): Likewise.
22916 2017-03-14 Martin Liska <mliska@suse.cz>
22919 * multiple_target.c (create_dispatcher_calls): Check that
22920 a target can create a function dispatcher.
22922 2017-03-14 Martin Liska <mliska@suse.cz>
22925 * multiple_target.c (expand_target_clones): Drop local.local
22926 flag for default implementation.
22928 2017-03-14 Richard Biener <rguenther@suse.de>
22930 PR tree-optimization/80030
22931 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22933 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22935 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22936 gcc_fallthrough() instead of __attribute__((fallthrough));
22938 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22940 * doc/gcc.texi: Remove "up" link to (DIR).
22941 * doc/gccint.texi: Ditto.
22943 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22945 * doc/install.texi (Specific) <avr>: Remove reference to
22948 2017-03-13 Jeff Law <law@redhat.com>
22950 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22951 attribute rather than comments.
22953 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22954 match_scratch operand is highest.
22956 2017-03-13 Martin Liska <mliska@suse.cz>
22958 PR middle-end/78339
22959 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22960 is a CHKP clone, use original declaration.
22962 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22964 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22965 (arc_conditional_register_usage): Use a different allocation order
22966 when optimizing for size.
22967 * common/config/arc/arc-common.c (arc_option_optimization_table):
22968 Section anchors default on when optimizing for size.
22970 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22972 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22974 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22976 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22977 * config/arc/arc.md (cpu_facility): Add cd variant.
22978 (*movqi_insn): Add code density variant.
22979 (*movhi_insn): Likewise.
22980 (*movqi_insn): Likewise.
22981 (*addsi3_mixed): Likewise.
22982 (subsi3_insn): Likewise.
22984 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22986 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22988 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22990 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22991 expressions with MINUS and UNARY ops.
22993 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22996 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22998 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22999 between vec_select and vector argument.
23000 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23001 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23002 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23003 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23004 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23005 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23007 2017-03-13 Richard Biener <rguenther@suse.de>
23010 * params.def (vect-max-peeling-for-alignment): Fix typo.
23012 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23014 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23015 issue that only occurred with binutils below 2.18.
23017 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23019 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23020 refer to binutils 2.11/2.12 minimum.
23022 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23024 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23025 ftp.kernel.org and simplify binutils requirement.
23027 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23029 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23031 (Optimize Options): Ditto. Also remove redundancy.
23033 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23035 PR translation/79848
23036 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23038 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23039 to G_ to avoid double translation.
23041 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23043 PR translation/79923
23044 * auto-profile.c (get_combined_location): Convert leading
23045 character of diagnostics to lower case and remove trailing period.
23046 (read_profile): Likewise for various diagnostics.
23047 * config/arm/arm.c (arm_option_override): Remove trailing period
23048 from various diagnostics.
23049 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23050 (msp430_expand_delay_cycles): Likewise.
23052 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23055 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23056 full command-line argument, rather than just "str".
23057 (aarch64_validate_march): Likewise.
23058 (aarch64_validate_mtune): Likewise.
23060 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23062 PR rtl-optimization/78911
23063 * lra-assigns.c (must_not_spill_p): New function.
23064 (spill_for): Use it.
23066 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23068 PR tree-optimization/79981
23069 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23070 ATOMIC_COMPARE_EXCHANGE ifn result.
23071 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23072 IFN_ATOMIC_COMPARE_EXCHANGE.
23074 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23077 * opts.c (parse_sanitizer_options): Add missing question mark to
23078 "did you mean" message.
23080 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23082 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23084 (VMULEUH_UNS): Likewise.
23085 (VMULOUB_UNS): Likewise.
23086 (VMULOUH_UNS): Likewise.
23087 * config/rs6000/rs6000.c (builtin_function_type): Remove
23088 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23090 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23093 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23094 x with result of extra_parsing_for_operand_code_0.
23095 (function_reader::extra_parsing_for_operand_code_0): Convert
23096 return type from void to rtx, returning x. When reading
23097 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23098 larger size containing struct block_symbol.
23100 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23102 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23103 -mfloat128-hardware without -m64.
23105 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23108 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23109 entries to the case statement that marks unsigned arguments to
23110 overloaded functions.
23112 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23114 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23115 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23117 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23120 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23121 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23123 2017-03-10 Martin Liska <mliska@suse.cz>
23127 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23128 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23131 2017-03-10 Olivier Hainque <hainque@adacore.com>
23133 * tree-switch-conversion (array_value_type): Start by resetting
23134 candidate type to it's main variant.
23136 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23138 PR rtl-optimization/79909
23139 * combine.c (try_combine): Use simplify_replace_rtx on individual
23140 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23141 of the whole CALL_INSN_FUNCTION_USAGE.
23143 PR tree-optimization/79972
23144 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23145 get_range_info on SSA_NAMEs. Formatting fixes.
23147 2017-03-10 Richard Biener <rguenther@suse.de>
23148 Jakub Jelinek <jakub@redhat.com>
23150 PR tree-optimization/77975
23151 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23152 edge to be constant.
23153 (get_val_for): For constant x return it. Formatting fix.
23154 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23155 would use the same bases as the current one.
23157 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23159 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23160 instead of vec_select for V1TImode.
23161 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23163 (VSX_LE_128): Add V1TI to this mode iterator.
23164 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23165 (*vsx_le_perm_store_<mode>): Likewise.
23166 (pre-reload splitter for VSX stores): Likewise.
23167 (post-reload splitter for VSX stores): Likewise.
23168 (*vsx_xxpermdi2_le_<mode>): Likewise.
23169 (*vsx_lxvd2x2_le_<mode>): Likewise.
23170 (*vsx_stxvd2x2_le_<mode>): Likewise.
23172 2017-03-09 Michael Eager <eager@eagercon.com>
23174 Correct failures with --enable-checking=yes,rtl.
23176 * config/microblaze/microblaze.c (microblaze_expand_shift):
23177 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23178 test for const0_rtx.
23179 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23180 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23182 2017-03-09 Richard Biener <rguenther@suse.de>
23184 PR tree-optimization/79977
23185 * graphite-scop-detection.c (scop_detection::merge_sese):
23186 Handle the case of extra exits to blocks dominating the entry.
23188 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23190 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23193 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23195 PR rtl-optimization/79949
23196 * lra-constraints.c (process_alt_operands): Check memory when
23197 trying to predict a cycle. Print about the overall increase.
23199 2017-03-09 Richard Biener <rguenther@suse.de>
23201 PR middle-end/79971
23202 * gimple-expr.c (useless_type_conversion_p): Preserve
23203 TYPE_SATURATING for fixed-point types.
23205 2017-03-09 Richard Biener <rguenther@suse.de>
23208 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23209 alignment of BLKmode params.
23211 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23214 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23215 (VALL_NO_V2Q): Likewise.
23217 * config/aarch64/aarch64-simd.md
23218 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23220 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23221 VALL_NO_V2Q mode iterator.
23222 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23224 2017-03-09 Martin Liska <mliska@suse.cz>
23226 PR tree-optimization/79631
23227 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23228 tree_int_cst_sign_bit just for INTEGER constants.
23230 2017-03-09 Martin Liska <mliska@suse.cz>
23234 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23237 2017-03-09 Marek Polacek <polacek@redhat.com>
23240 * tree.c (inchash::add_expr): Handle TREE_VEC.
23242 2017-03-09 Martin Liska <mliska@suse.cz>
23245 (chkp_narrow_size_and_offset): New function.
23246 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23247 (void chkp_parse_bit_field_ref): New function.
23248 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23249 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23251 2017-03-09 Martin Liska <mliska@suse.cz>
23254 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23255 (chkp_find_bounds_1): Remove gcc_unreachable.
23257 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23260 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23261 BUILT_IN_SYNC*, determine the access type from the size suffix and
23262 always build a MEM_REF with that type. Handle forgotten
23263 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23266 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23267 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23268 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23269 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23270 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23271 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23272 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23273 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23274 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23275 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23276 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23277 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23278 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23279 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23280 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23281 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23282 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23283 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23284 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23285 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23286 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23287 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23288 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23289 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23290 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23291 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23292 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23293 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23294 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23295 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23296 definitions outside of __OPTIMIZE__ guarded section.
23299 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23300 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23301 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23302 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23305 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23307 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23308 ("vfenez<mode>"): Add missing constraints.
23310 2017-03-08 Martin Sebor <msebor@redhat.com>
23313 * config/nds32/nds32.c (nds32_option_override):
23314 Fix misspelled diagnostic.
23316 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23319 * gimplify.c (gimplify_omp_for): Replace index var in outer
23320 taskloop statement with an artificial variable and add
23321 OMP_CLAUSE_PRIVATE clause for it.
23323 2017-03-08 Richard Biener <rguenther@suse.de>
23325 PR tree-optimization/79955
23326 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23327 for accesses that are completely outside of the variable.
23329 2017-03-08 Andrew Haley <aph@redhat.com>
23331 PR tree-optimization/79943
23332 * tree-ssa-loop-split.c (compute_new_first_bound): When
23333 calculating the new upper bound, (END-BEG) should be added, not
23336 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23338 * config/avr/avr.md (setmemhi): Make sure match_dup
23339 operand number comes before match_scratch.
23341 2017-03-08 Richard Biener <rguenther@suse.de>
23343 PR tree-optimization/79920
23344 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23345 with ncopies == 1 to ...
23346 (vect_transform_slp_perm_load): ... here. Properly compute
23347 all element loads by iterating VF times over the group. Do
23348 not handle ncopies (computed in a broken way) in
23349 vect_create_mask_and_perm.
23351 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23354 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23355 is a uniform vector, use uniform_vector_p return value instead of
23356 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23358 2017-03-07 Marek Polacek <polacek@redhat.com>
23360 PR middle-end/79809
23361 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23362 (alloca_call_type): Likewise.
23364 2017-03-07 Martin Liska <mliska@suse.cz>
23366 * gcov.c (process_args): Put comment to correct location.
23368 2017-03-07 Martin Liska <mliska@suse.cz>
23370 PR middle-end/68270
23371 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23372 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23373 (chkp_narrow_bounds_for_field): Likewise.
23374 (chkp_parse_array_and_component_ref): Pass one more argument to
23377 2017-03-07 Richard Biener <rguenther@suse.de>
23379 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23382 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23384 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23385 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23387 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23390 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23391 to end of description.
23392 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23394 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23396 PR rtl-optimization/79901
23397 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23399 (*avx512f_<code><mode>3<mask_name>): ... this.
23400 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23401 iterator instead of VI8_AVX2_AVX512BW.
23403 PR rtl-optimization/79901
23404 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23405 min/max expander, expand it using expand_vec_cond_expr.
23408 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23411 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23414 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23415 to void * for PCH reasons.
23416 * dwarf2out.c (output_loc_operands, output_die): Cast
23417 v.val_vec.array to unsigned char *.
23419 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23422 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23425 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23427 PR rtl-optimization/79571
23428 * lra-constraints.c (process_alt_operands): Calculate static
23429 reject and subtract it from overall when only addresses will be
23432 2017-03-06 Julia Koval <julia.koval@intel.com>
23435 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23436 incoming stack boundary to 128 for 64-bit targets.
23438 2017-03-06 Richard Biener <rguenther@suse.de>
23440 PR tree-optimization/79894
23441 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23442 to NULL after folding it.
23444 2017-03-06 Richard Biener <rguenther@suse.de>
23446 PR tree-optimization/79824
23447 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23448 check disabling peeling for gaps.
23450 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23452 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23453 attributes): Document gettimeofday.
23455 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23457 * config/s390/s390.c (s390_option_override_internal): Set
23458 PARAM_MIN_VECT_LOOP_BOUND
23460 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23462 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23463 * config/s390/s390.md: Likewise.
23465 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23468 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23469 (<avx2_avx512>_perm<mode>): Rename to ...
23470 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23472 (<avx512>_perm<mode>_mask): Rename to ...
23473 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23475 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23476 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23477 instead of VI8F_256_512.
23478 (avx512f_perm<mode>): New define_expand.
23479 (avx512f_perm<mode>_mask): Likewise.
23480 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23481 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23483 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23485 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23486 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23488 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23490 2017-03-06 Martin Liska <mliska@suse.cz>
23493 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23494 when having a SSA NAME w/o VAR_DECL assigned to it.
23496 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23498 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23499 msa_dpsub_<su>_d): Fix MODE for vec_select.
23501 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23503 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23505 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23507 2017-03-06 Richard Biener <rguenther@suse.de>
23509 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23510 * plugin.c (register_plugin_info): Likewise.
23511 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23513 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23515 * config/i386/sse.md (sse_storehps, sse_storelps,
23516 avx_<castmode><avxsizesuffix>_<castmode>,
23517 avx512f_<castmode><avxsizesuffix>_<castmode>,
23518 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23519 in condition that at least one operand is not a MEM.
23521 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23523 PR middle-end/79805
23524 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23525 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23527 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23528 gimple_call_nothrow_p flag based on whether original builtin can throw.
23529 If it can, emit following stmts on the fallthrough edge.
23530 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23531 don't create new bb if inserting just debug stmts on the edge, try to
23532 insert them on the fallthru bb or just reset debug stmts.
23534 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23537 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23538 restore recog_data (including the operand rtxes inside it) around
23539 the call to get_insn_template.
23541 2017-03-03 Martin Sebor <msebor@redhat.com>
23543 PR tree-optimization/79699
23544 * context.c (context::~context): Free MPFR caches to avoid
23545 a memory leak on program exit.
23547 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23549 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23550 Use wide_int::ulow () instead of .elt (0).
23552 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23554 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23555 (*pushxf): Limit oF constraint to 32bit targets and add oC
23556 constraint for 64bit targets.
23557 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23558 (*pushdf): Change rmF constraint to rmC.
23560 2017-03-03 Martin Liska <mliska@suse.cz>
23562 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23563 Remove unused variable.
23565 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23568 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23569 is a memory operand, increase num_memory.
23570 (ix86_expand_args_builtin): Likewise.
23572 2017-03-03 Jan Hubicka <jh@suse.cz>
23575 * ipa-devirt.c (maybe_record_node): Properly handle
23576 __cxa_pure_virtual visibility.
23578 2017-03-03 Martin Liska <mliska@suse.cz>
23580 PR tree-optimization/79803
23581 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23583 (pass_loop_prefetch::execute): Disabled optimization if an
23584 assumption about L1 cache size is not met.
23586 2017-03-03 Martin Liska <mliska@suse.cz>
23588 PR rtl-optimization/79574
23589 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23590 (hash_scan_set): Likewise.
23591 (dump_hash_table): Likewise.
23592 (hoist_code): Likewise.
23594 2017-03-03 Richard Biener <rguenther@suse.de>
23596 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23597 in place of uhigh ().
23598 (fixed_convert_from_real): Likewise.
23600 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23603 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23605 2017-03-03 Richard Biener <rguenther@suse.de>
23607 PR middle-end/79818
23608 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23609 TYPE_OVERFLOW_UNDEFINED check.
23611 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23613 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23615 (vector_ae_<mode>_p): Likewise.
23616 (vector_nez_<mode>_p): Likewise.
23617 (vector_ne_v2di_p): Likewise.
23618 (vector_ae_v2di_p): Likewise.
23619 (vector_ne_<mode>_p): Likewise.
23620 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23622 (vsx_tsqrt<mode>2_fe): Likewise.
23624 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23627 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23629 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23631 PR rtl-optimization/79780
23632 * cprop.c (one_cprop_pass): When second and further conditional trap
23633 in a single basic block is turned into an unconditional trap, turn it
23634 into a deleted note to avoid RTL verification failures.
23636 2017-03-02 Richard Biener <rguenther@suse.de>
23638 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23640 2017-03-02 Richard Biener <rguenther@suse.de>
23642 PR tree-optimization/79345
23644 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23645 param and abort the walk, returning -1 if it is hit.
23646 (walk_aliased_vdefs): Take a limit param and pass it on.
23647 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23648 defaulting to 0 and return a signed int.
23649 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23650 (check_defs): New helper.
23651 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23652 about uninitialized memory.
23653 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23654 bogus uninitialized warning.
23655 (fixed_convert_from_real): Likewise.
23657 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23659 PR tree-optimization/66768
23660 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23661 iv_use if base object can't be determined.
23663 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23665 PR tree-optimization/79345
23666 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23667 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23668 (get_pattern_stats): Initialize it.
23669 * genemit.c (gen_expand): Verify match_scratch numbers come after
23670 match_operand/match_dup numbers.
23671 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23672 match_scratch numbers.
23673 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23675 * config/s390/s390.md (trunctdsd2): Likewise.
23677 2017-03-02 Richard Biener <rguenther@suse.de>
23679 * wide-int.h (wide_int_storage::operator=): Implement in terms
23682 2017-03-02 Richard Biener <rguenther@suse.de>
23684 PR tree-optimization/79777
23685 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23686 the to insert expression to sth existing.
23688 2017-03-01 Martin Sebor <msebor@redhat.com>
23690 PR middle-end/79692
23691 * gimple-ssa-sprintf.c
23692 (directive::known_width_and_precision): New function.
23693 (format_integer): Use it.
23694 (get_mpfr_format_length): Consider the full range of precision
23695 when computing %g output with the # flag. Set the likely byte
23696 count to 3 rather than 1 when precision is indeterminate.
23697 (format_floating): Correct the lower bound of precision.
23699 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23701 * doc/invoke.texi: Document default code model for 64-bit Linux.
23703 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23706 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23707 udiv rather than div since input pattern is unsigned.
23709 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23711 * config/i386/i386.c (print_reg): Warn for values of
23712 unsupported size in integer register.
23714 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23717 * config/rs6000/predicates.md (current_file_function_operand): Do
23718 not allow self calls to be local if the function is replaceable.
23720 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23723 * config/rs6000/altivec.h (vec_ctz and others): Change the
23724 preprocessor macro that controls conditional compilation from
23725 _ARCH_PWR9 to __POWER9_VECTOR__.
23726 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23727 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23728 control (instead of _ARCH_PWR9 control) so that template
23729 definition uses power9-specific function.
23730 (vec_any_eq): Likewise.
23731 (vec_all_ne): Change macro definition to use a power9-specific
23732 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23733 _ARCH_PWR9 control).
23734 (vec_any_eq) Likewise.
23735 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23736 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23737 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23738 support for xvcmpnedp instruction.
23739 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23740 macro expansion so that Power9 implementation of vec_all_ne does
23741 not use the AltiVec predicate framework.
23742 (VCMPNEH_P): Likewise.
23743 (VCMPNEW_P): Likewise.
23744 (VCMPNED_P): Likewise.
23745 (VCMPNEFP_P): Likewise.
23746 (VCMPNEDP_P): Likewise.
23747 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23748 implementation of vec_any_eq to not use AltiVec predicate
23750 (VCMPAEH_P): Likewise.
23751 (VCMPAEW_P): Likewise.
23752 (VCMPAED_P): Likewise.
23753 (VCMPAEFP_P): Likewise.
23754 (VCMPAEDP_P): Likewise.
23755 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23756 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23757 not use the AltiVec predicate framework.
23758 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23759 of vec_any_eq to not use AltiVec predicate framework.
23760 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23761 support for predefined __POWER9_VECTOR__ macro to indicate that
23762 Power9 instruction selection is enabled.
23763 (altivec_overloaded_builtins): Remove extraneous
23764 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23765 function argument types RS6000_BTI_bool_V16QI and
23766 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23767 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23768 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23769 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23770 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23771 Power9 for implementations of vec_cmpne. Change the signature for
23772 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23773 (representing vec_all_ne) to remove the previously described first
23774 argument of type RS6000_BTI_INTSI, as this was an artifact of
23775 reliance on the AltiVec predicate framework, which is no longer
23776 used in the implementation of these functions. Add
23777 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23778 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23779 since, unlike the AltiVec predicate framework implementation, we
23780 do not share function descriptors between vec_alle and vec_anyeq.
23781 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23782 set of modes that receive special treatment even when
23783 TARGET_P9_VECTOR is true. The special treatment emits code that
23784 does not depend on Power9 instructions.
23785 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23786 define_expand to not rely on AltiVec predicate framework.
23787 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23789 (vector_ne_v2di_p): Change this define_expand to not rely on
23790 AltiVec predicate framework.
23791 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23793 (vector_ne_<mode>_p): Change this define_expand to not rely on
23794 AltiVec predicate framework.
23795 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23797 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23798 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23799 define_insn pattern.
23800 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23801 define_insn pattern because the xvcmpne<VSs>. instruction is not
23803 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23804 instruction is not supported.
23806 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23808 * config/nvptx/nvptx.c: Include intl.h.
23810 2017-03-01 Martin Jambor <mjambor@suse.cz>
23813 * ipa-prop.h (ipa_bits): Removed field known.
23814 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23815 to pointers. Adjusted their comments to warn about their sharing.
23816 (ipcp_transformation_summary): Change bits to a vector of pointers.
23817 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23818 (ipa_get_ipa_bits_for_value): Declare.
23819 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23820 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23821 (ipa_bits_hash_table): Likewise.
23822 (ipa_vr_ggc_hash_traits): Likewise.
23823 (ipa_vr_hash_table): Likewise.
23824 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23825 being pointers and vr_known being removed.
23826 (ipa_set_jf_unknown): Likewise.
23827 (ipa_get_ipa_bits_for_value): New function.
23828 (ipa_set_jfunc_bits): Likewise.
23829 (ipa_get_value_range): New overloaded functions.
23830 (ipa_set_jfunc_vr): Likewise.
23831 (ipa_compute_jump_functions_for_edge): Use the above functions to
23832 construct bits and vr parts of jump functions.
23833 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23834 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23836 (ipcp_grow_transformations_if_necessary): Also allocate
23837 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23839 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23840 them. Fix too long lines.
23841 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23842 vr_known being removed.
23843 (ipa_read_jump_function): Use new setter functions to construct bits
23844 and vr parts of jump functions or set them to NULL.
23845 (write_ipcp_transformation_info): Adjust for bits being pointers.
23846 (read_ipcp_transformation_info): Likewise.
23847 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23849 Include gt-ipa-prop.h.
23850 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23852 (ipcp_store_bits_results): Likewise.
23853 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23854 Do not write to existing jump functions but use a temporary instead.
23856 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23859 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23860 attempt to use its first operand as BIT_FIELD_REF base.
23862 2017-03-01 Richard Biener <rguenther@suse.de>
23864 PR middle-end/79721
23865 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23866 interpolating formula in wrapping arithmetic.
23867 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23869 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23871 PR tree-optimization/79734
23872 * tree-vect-generic.c (expand_vector_condition): Optimize
23873 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23874 Handle VEC_COND_EXPR where comparison has different inner width from
23875 type's inner width.
23877 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23879 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23880 markup, and similar issues. Remove @opindex entries for things
23881 that aren't options. Add missing -mmpy-option entries.
23883 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23885 PR tree-optimization/79737
23886 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23887 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23888 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23889 instead of byte_size. Formatting fix.
23890 (shift_bytes_in_array_right): Formatting fix.
23892 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23895 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23896 condition on optimize for the leaf function test.
23898 2017-02-28 Martin Liska <mliska@suse.cz>
23901 * read-rtl-function.c (function_reader::handle_unknown_directive):
23902 Bail out when one uses -flto.
23904 2017-02-28 Martin Liska <mliska@suse.cz>
23906 * common.opt: Replace space with tabular for options of <number>
23908 * config/i386/i386.opt: Show <number> value for
23909 -mlarge-data-threshold.
23910 * opts.c (print_filtered_help): Do not display number in hexadecimal
23913 2017-02-28 Martin Liska <mliska@suse.cz>
23915 * common.opt: Fix --help=option -Q for options which are of
23918 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23920 * config/i386/i386.c (print_reg): Error out for values
23921 of 8-bit size in invalid integer register.
23923 2017-02-28 Martin Sebor <msebor@redhat.com>
23925 PR tree-optimization/79691
23926 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23928 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23931 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23932 gcc_unreachable with output_operand_lossage.
23934 2017-02-28 Richard Biener <rguenther@suse.de>
23936 PR tree-optimization/79740
23937 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23939 (visit_nary_op): Insert the nary into the hashtable if we
23940 pattern-matched sth.
23941 * tree-ssa-pre.c (eliminate_insert): Robustify.
23943 2017-02-28 Richard Biener <rguenther@suse.de>
23945 PR middle-end/79731
23946 * fold-const.c (decode_field_reference): Reject out-of-bound
23949 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23951 * config/i386/i386.c: Include intl.h.
23952 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23953 instead of just cond ? "..." : "...".
23954 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23955 * coverage.c (read_counts_file): Likewise.
23956 * omp-offload.c: Include intl.h.
23957 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23958 of just cond ? "..." : "...".
23959 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23960 of just cond ? "..." : "...".
23962 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23965 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23967 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23968 'tune for' CPU name.
23969 * config/arm/arm-cpu-data.h: Regenerated.
23971 2017-02-28 Richard Biener <rguenther@suse.de>
23973 PR tree-optimization/79732
23974 * tree-inline.c (expand_call_inline): Do not shadow var.
23976 2017-02-28 Richard Biener <rguenther@suse.de>
23978 PR tree-optimization/79723
23979 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23980 address-space properly.
23982 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23984 * doc/optinfo.texi (Optimization groups): Fix option used for
23986 * doc/invoke.texi (-fopt-info): Document "omp".
23987 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23988 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23989 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23990 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23991 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23993 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23995 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23996 OPTGROUP_OMP with "omp".
23998 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24001 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24002 for arithmetic shift of unsigned V2DI.
24004 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24006 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24007 arc/linux.h headers.
24008 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24009 (LINK_SPEC): Likewise.
24010 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24011 (EXTRA_SPECS): Likewise.
24012 (STARTFILE_SPEC): Likewise.
24013 (ENDFILE_SPEC): Likewise.
24014 (LIB_SPEC): Likewise.
24015 (TARGET_SDATA_DEFAULT): Likewise.
24016 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24017 (MULTILIB_DEFAULTS): Likewise.
24018 (DWARF2_UNWIND_INFO): Likewise.
24019 * config/arc/big.h: New file.
24020 * config/arc/elf.h: Likewise.
24021 * config/arc/linux.h: Likewise.
24022 * config/arc/t-uClibc: Remove.
24024 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24026 PR tree-optimization/77536
24027 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24028 (tree_transform_and_unroll_loop): Use above function to compute the
24029 estimated niter of unrolled loop and use it when scaling profile.
24030 Also use count info rather than frequency if it's non-zero.
24031 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24032 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24033 (vect_transform_loop): Call above function.
24035 2017-02-27 Richard Biener <rguenther@suse.de>
24037 PR tree-optimization/45397
24038 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24039 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24040 (visit_nary_op): Add pattern matching for CSEing sign-changed
24041 or truncated operations with wider ones.
24043 2017-02-27 Richard Biener <rguenther@suse.de>
24045 PR tree-optimization/79690
24046 * tree-vect-stmts.c (vectorizable_store): Use vector type
24047 built from the DR with address-space.
24049 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24051 * doc/invoke.texi (Optimize Options): Refine the description
24052 of asan-use-after-return.
24054 2017-02-25 Alan Modra <amodra@gmail.com>
24056 PR rtl-optimization/79584
24057 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24058 base, not ad->base_term, the reg within base. Remove assertion
24059 that ad->base == ad->base_term. Replace gen_int_mode using
24060 bogus mode with const0_rtx.
24062 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24064 PR middle-end/79396
24065 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24066 FMA_EXPR like tcc_binary or tcc_unary.
24068 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24071 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24073 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24074 (output_loc_operands): Handle DW_OP_call_ref and
24075 DW_OP_GNU_variable_value.
24076 (struct variable_value_struct): New type.
24077 (struct variable_value_hasher): Likewise.
24078 (variable_value_hash): New variable.
24079 (string_types): Remove.
24080 (copy_loc_descr): New function.
24081 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24082 (prepend_loc_descr_to_each): New function.
24083 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24084 instead of add_loc_descr_to_each if the first argument is single
24085 location list and the second has multiple.
24086 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24087 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24088 when looking for variable value which doesn't have other location info.
24089 (loc_list_from_tree): Formatting fix.
24090 (gen_array_type_die): Simplify DW_AT_string_length handling.
24091 (adjust_string_types): Remove.
24092 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24093 string_types. Call resolve_variable_values.
24094 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24095 (resolve_addr_in_expr): Likewise. Add A argument.
24096 (copy_deref_exprloc): Remove deref argument. Adjust for the
24097 original expression being DW_OP_GNU_variable_value with optionally
24098 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24099 optionally after it.
24100 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24101 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24102 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24103 (variable_value_hasher::hash, variable_value_hasher::equal): New
24105 (resolve_variable_value_in_expr, resolve_variable_value,
24106 resolve_variable_values, note_variable_value_in_expr,
24107 note_variable_value): New functions.
24108 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24111 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24114 * opts.h (handle_generated_option): Add GENERATED_P argument.
24115 * opts-common.c (handle_option): Adjust function comment.
24116 (handle_generated_option): Add GENERATED_P argument, pass it to
24118 (control_warning_option): Pass false to handle_generated_option
24120 * opts.c (maybe_default_option): Pass true to handle_generated_option
24122 * optc-gen.awk: Likewise.
24124 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24126 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24127 a REG, look at the REG it is a SUBREG of.
24128 (splitter for cmpeqsi_t): Ditto.
24130 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24132 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24133 the special USEs with the pattern of the insn, not the insn itself.
24135 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24138 * doc/invoke.texi: Document -mload-store-pairs.
24140 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24141 Sandra Loosemore <sandra@codesourcery.com>
24143 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24144 argument isn't a CONST_INT.
24145 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24146 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24147 (nios2_validate_compare): Bypass alternate compare logic if *op2
24148 is not a CONST_INT.
24149 (ldstwm_operation_p): Return false if first_base is not a REG or
24150 if first_offset is not a CONST_INT.
24152 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24154 * config/cris/cris.md: Use correct operand in a define_peephole2.
24156 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24158 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24160 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24162 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24163 this_insn if it is an INSN or JUMP_INSN.
24164 (force_offsettable): Look at base, not at addr.
24165 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24166 on things that aren't necessarily CONST_INTs.
24168 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24170 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24171 -mfpmath=sse is the default also for x86-32 targets with SSE2
24172 instruction set when @option{-ffast-math} is enabled
24174 2017-02-24 Jeff Law <law@redhat.com>
24176 PR rtl-optimizatoin/79286
24177 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24180 2017-02-24 Richard Biener <rguenther@suse.de>
24182 PR tree-optimization/79389
24183 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24186 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24188 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24189 function comment to reflect reality.
24190 (loop_exits_before_overflow): Fix typo in function description.
24192 2017-02-24 Richard Biener <rguenther@suse.de>
24194 PR tree-optimization/79389
24195 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24196 properly that a threading opportunity exists. Detect conditional
24197 copy/constant propagation opportunities.
24199 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24201 * config/visium/visium.md (type): Add trap.
24202 (b): New mode attribute.
24203 (*btst): Rename into...
24204 (*btst<mode>): ...this and adjust.
24205 (*cbranchsi4_btst_insn): Rename into...
24206 (*cbranch<mode>4_btst_insn): ...this and adjust.
24207 (trap): New define_insn.
24209 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24211 PR tree-optimization/79389
24212 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24213 (noce_reversed_cond_code): New function.
24214 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24215 reversed_comparison_code. Formatting fix.
24216 (noce_try_store_flag): Test rev_cond != NULL in addition to
24217 reversed_comparison_code.
24218 (noce_try_store_flag_constants): Likewise.
24219 (noce_try_store_flag_mask): Likewise.
24220 (noce_try_addcc): Use rev_cond if non-NULL instead of
24221 reversed_comparison_code.
24222 (noce_try_cmove_arith): Likewise. Formatting fixes.
24223 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24224 (noce_find_if_block): Initialize rev_cond.
24225 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24226 instead of false as last argument never attempt to reverse it
24229 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24231 PR tree-optimization/79663
24232 * tree-predcom.c (combine_chains): Process refs in reverse order
24233 only for ZERO length chains, and add explaining comment.
24235 2017-02-23 Jeff Law <law@redhat.com>
24237 PR tree-optimization/79578
24238 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24239 in call to operand_equal_p.
24241 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24244 * config/i386/cpuid.h: Fix another undefined behavior.
24246 2017-02-23 Richard Biener <rguenther@suse.de>
24248 PR tree-optimization/79683
24249 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24250 vector types for data-refs.
24252 2017-02-23 Martin Liska <mliska@suse.cz>
24254 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24256 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24258 PR middle-end/79665
24259 * internal-fn.c (get_range_pos_neg): Moved to ...
24260 * tree.c (get_range_pos_neg): ... here. No longer static.
24261 * tree.h (get_range_pos_neg): New prototype.
24262 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24263 are known to be in between 0 and signed maximum inclusive, try to
24264 expand both unsigned and signed divmod and use the cheaper one from
24267 2017-02-22 Jeff Law <law@redhat.com>
24269 PR tree-optimization/79578
24270 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24271 to compare base operands.
24273 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24276 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24277 gpc_reg_operand instead of fpr_reg_operand.
24279 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24281 * config/mips/mips.c (mips_return_in_memory): Force FP
24282 vector types to be returned in memory for o32 ABI.
24284 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24286 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24287 instead of DW_TAG_member for static data member declarations and don't
24288 set no_linkage_name for static inline data members.
24289 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24292 2017-02-22 Martin Liska <mliska@suse.cz>
24294 * doc/invoke.texi: Replace inequality signs with square brackets
24297 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24299 PR tree-optimization/68644
24300 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24302 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24305 * lra-constraints.c (simplify_operand_subreg): Handle
24306 WORD_REGISTER_OPERATIONS targets.
24308 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24311 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24312 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24313 elimination by swapping fld*.
24315 2017-02-22 Richard Biener <rguenther@suse.de>
24317 PR tree-optimization/79673
24318 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24319 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24320 irrelevant address-space qualifiers and avoiding a
24321 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24323 2017-02-22 Richard Biener <rguenther@suse.de>
24325 PR tree-optimization/79666
24326 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24327 to not symbolically negate if that may introduce undefined
24330 2017-02-22 Martin Liska <mliska@suse.cz>
24333 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24334 * data-streamer-out.c (streamer_write_gcov_count_stream):
24336 * value-prof.c (stream_out_histogram_value): Make assert more
24337 precise based on type of counter.
24339 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24342 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24343 Use nonimmediate_operand instead of memory_operand for operand 1.
24344 (float-extend standard_x87sse_constant_load splitter): Ditto.
24346 2017-02-21 Jeff Law <law@redhat.com>
24348 PR tree-optimization/79621
24349 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24350 blocks with edges to themselves.
24352 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24355 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24356 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24357 Use gimple_call_builtin_p.
24360 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24361 on temporarily removed DEBUG_INSNs.
24363 PR tree-optimization/79649
24364 * tree-loop-distribution.c (classify_partition): Give up on
24365 non-generic address space loads/stores.
24367 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24369 * doc/loop.texi (Loop manipulation): Remove nonexistent
24370 tree_ssa_loop_version from the documentation.
24371 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24373 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24376 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24377 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24378 * config/rs6000/rs6000.c: Include except.h.
24379 (rs6000_expand_split_stack_prologue): Call
24380 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24382 2017-02-21 Martin Jambor <mjambor@suse.cz>
24385 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24386 have been analyzed.
24388 2017-02-21 Martin Jambor <mjambor@suse.cz>
24390 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24391 for backward compatibility only.
24392 * doc/invoke.texi (Option Summary): Remove all references to
24393 -fipa-cp-alignment.
24395 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24399 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24401 * lra-constraints.c (curr_insn_transform): Handle
24402 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24404 2017-02-21 Martin Liska <mliska@suse.cz>
24406 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24408 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24410 PR translation/79638
24411 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24413 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24416 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24417 (arm_function_ok_for_sibcall): Return false for an indirect call by
24418 descriptor if all the argument registers are used.
24419 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24420 alignment of the function.
24422 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24424 PR tree-optimization/61441
24425 * simplify-rtx.c (simplify_const_unary_operation): For
24426 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24427 the sNaN unmodified.
24429 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24431 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24432 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24433 instead of SYSTEM_HEADER_DIR.
24435 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24436 Martin Liška <mliska@suse.cz>
24438 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24439 Fix typos and grammar, use active voice, and clarify.
24441 2017-02-20 Marek Polacek <polacek@redhat.com>
24443 PR middle-end/79537
24444 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24447 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24449 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24452 * config/i386/i386.c (ix86_expand_builtin): Handle
24453 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24454 ix86_builtins_isa[fcode].isa as a requirement of those
24455 flags and any other flag in the bitmask.
24456 (ix86_init_mmx_sse_builtins): Use 0 instead of
24457 ~OPTION_MASK_ISA_64BIT as mask.
24458 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24459 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24460 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24461 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24463 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24466 * lra-constraints.c (split_reg): Check requested split mode
24467 is supported by the register.
24469 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24471 * lra-constraints.c (simplify_operand_subreg): Remove early
24474 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24477 * lra-constraints.c (curr_insn_transform): Tighten condition
24478 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24480 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24483 * lra-constraints.c (curr_insn_transform): Handle
24484 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24486 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24489 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24491 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24493 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24496 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24499 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24501 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24504 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24506 * final.c (last_columnnum, override_columnnum): New variables.
24507 (final_start_function): Set last_columnnum, pass it to begin_prologue
24508 hook and pass 0 to dwarf2out_begin_prologue.
24509 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24510 to source_line debug hook.
24511 (notice_source_line): Compute last_columnnum and for debug_column_info
24512 return true on column changes.
24513 * debug.h (struct gcc_debug_hooks): Add column argument to
24514 source_line and begin_prologue hooks.
24515 (debug_nothing_int_charstar_int_bool): Remove prototype.
24516 (debug_nothing_int_int_charstar,
24517 debug_nothing_int_int_charstar_int_bool): New prototypes.
24518 (dwarf2out_begin_prologue): Add column argument.
24519 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24520 begin_prologue hooks.
24521 (debug_nothing_int_charstar_int_bool): Remove.
24522 (debug_nothing_int_int_charstar,
24523 debug_nothing_int_int_charstar_int_bool): New functions.
24524 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24525 through to dwarf2out_source_line.
24526 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24527 (dwarf2out_source_line): Add column argument, emit it if requested.
24528 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24530 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24531 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24532 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24533 through to dwarf2out_begin_prologue.
24534 (vmsdbgout_source_line): Add column argument, pass it through to
24535 dwarf2out_source_line.
24536 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24537 dbxout_source_line caller.
24538 (dbxout_source_line): Add column argument.
24540 * common.opt (gno-column-info, gcolumn-info): New options.
24541 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24542 (check_die): Also test for multiple DW_AT_decl_column attributes.
24543 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24544 DW_AT_decl_column if requested.
24545 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24547 (gen_variable_die): Likewise.
24548 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24549 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24552 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24553 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24554 (ix86_handle_option): Handle OPT_m3dnowa.
24555 * doc/invoke.texi (-m3dnowa): Document.
24556 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24557 -m3dnowa instead of -m3dnow -march=athlon.
24560 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24561 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24563 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24566 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24567 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24568 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24569 generator for vsx_xxpermdi_<mode>_be.
24570 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24571 force big-endian semantics.
24572 (vsx_xxpermdi_<mode>_be): New define_expand with same
24573 implementation as previous version of vsx_xxpermdi_<mode>.
24575 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24577 PR tree-optimization/79327
24578 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24579 variable, its initialization and use.
24581 2017-02-17 Julia Koval <julia.koval@intel.com>
24583 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24584 (OPTION_MASK_ISA_PKU_UNSET): New.
24585 (ix86_handle_option): Handle -mrdpid.
24586 * config/i386/cpuid.h (bit_RDPID): New.
24587 * config/i386/driver-i386.c (host_detect_local_cpu):
24588 Detect RDPID feature.
24589 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24590 * config/i386/i386-c.c (ix86_target_macros_internal):
24592 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24593 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24594 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24595 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24596 * config/i386/i386.md (define_insn "rdpid"): New.
24597 * config/i386/i386.opt Add -mrdpid.
24598 * config/i386/immintrin.h (_rdpid_u32): New.
24600 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24602 PR rtl-optimization/79541
24603 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24604 instead of transforming it into USE.
24606 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24608 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24609 If HONOR_SNANS (SFmode) force the input to a register.
24610 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24611 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24612 an frsp or similar insn.
24614 2017-02-17 Martin Liska <mliska@suse.cz>
24616 PR rtl-optimization/79577
24617 * params.def (selsched-max-sched-times): Increase minimum to 1.
24619 2017-02-17 Martin Liska <mliska@suse.cz>
24621 PR rtl-optimization/79574
24622 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24624 2017-02-17 Martin Liska <mliska@suse.cz>
24626 PR tree-optimization/79529
24627 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24628 ssa_defined_default_def_p to handle cases which are implicitly
24630 * tree-ssa.c (ssa_defined_default_def_p): New function.
24631 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24632 which are implicitly defined.
24633 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24635 2017-02-17 Richard Biener <rguenther@suse.de>
24637 PR middle-end/79576
24638 * params.def (max-ssa-name-query-depth): Limit to 10.
24640 2017-02-17 Richard Biener <rguenther@suse.de>
24642 PR tree-optimization/79552
24643 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24646 2017-02-17 Richard Biener <rguenther@suse.de>
24649 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24651 2017-02-17 Marek Polacek <polacek@redhat.com>
24653 PR middle-end/79536
24654 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24655 (fold_negate_expr): New wrapper.
24657 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24659 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24660 Correct terminology and de-emphasize pre-standard behavior.
24662 2017-02-16 Alan Modra <amodra@gmail.com>
24664 PR rtl-optimization/79286
24665 * ira.c (def_dominates_uses): New function.
24666 (update_equiv_regs): Don't create an equivalence for insns that
24667 may trap where the register def does not dominate the use.
24669 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24671 PR rtl-optimization/78127
24672 * lra.c (lra): Call lra_eliminate before finish the loop after
24675 2017-02-16 Richard Biener <rguenther@suse.de>
24677 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24679 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24680 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24681 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24682 (isl_val_int_from_wi): New function.
24683 (extract_affine_gmp): Rename to ...
24684 (extract_affine_wi): ... this, take a widest_int.
24685 (extract_affine_int): Just wrap extract_affine_wi.
24686 (add_param_constraints): Use isl_val_int_from_wi.
24687 (add_loop_constraints): Likewise, and extract_affine_wi.
24689 2017-02-15 Jeff Law <law@redhat.com>
24691 PR middle-end/79521
24692 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24693 ira_init_register_move_cost_if_necessary.
24695 2017-02-15 Martin Sebor <msebor@redhat.com>
24697 PR middle-end/32003
24698 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24699 removed in a prior commit.
24701 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24703 PR tree-optimization/79347
24704 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24705 counters during peeling.
24707 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24709 * Makefile.in (site.exp): Remove "set ISLVER".
24711 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24714 * real.c (real_from_integer): Call real_convert even for decimal.
24716 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24719 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24721 2017-02-14 Andrew Pinski <apinski@cavium.com>
24723 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24724 cores and change the partno/implementer to be correct.
24725 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24726 the 'B" as the implementer.
24727 * config/aarch64/aarch64-tune.md: Regenerate.
24729 2017-02-14 Carl Love <cel@us.ibm.com>
24731 * config/rs6000/rs6000.c: Add case statement entry to make the
24732 xvcvuxdsp built-in argument unsigned.
24733 * config/rs6000/vsx.md: Fix the source and return operand types so they
24734 match the instruction definitions from the ISA document. Fix typo
24735 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24738 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24741 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24742 member early_clobber_alts.
24743 * lra-lives.c (reg_early_clobber_p): New.
24744 (process_bb_lives): Use it.
24745 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24746 (debug_operand_data): Initialize early_clobber_alts.
24747 (setup_operand_alternative): Set up early_clobber_alts.
24748 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24749 alternatives to new_insn_reg.
24750 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24752 (lra_update_insn_regno_info): Pass the new arg.
24754 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24756 PR middle-end/79505
24757 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24758 (new_oacc_loop_raw): Don't clear already cleared fields.
24761 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24762 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24763 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24765 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24768 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24770 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24773 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24774 the extra instruction to the right place to store 128-bit constant
24777 2017-02-14 Martin Sebor <msebor@redhat.com>
24779 PR middle-end/79448
24780 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24781 warning for strings of unknown length.
24783 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24785 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24787 2017-02-14 Jeff Law <law@redhat.com>
24790 * ira-costs.c (scan_one_insn): Initialize register move costs
24791 for pseudos seen in USE/CLOBBER insns.
24793 PR tree-optimization/79095
24794 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24795 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24796 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24797 if the operands are known to be not equal, then the resulting range
24799 (intersect_ranges): If the new range is ~[0,0] and the old range is
24800 wide, then prefer ~[0,0].
24801 * tree-vrp.c (overflow_comparison_p_1): New function.
24802 (overflow_comparison_p): New function.
24803 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24804 if NAME is used in an overflow test.
24805 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24806 overflow check that can be expressed as an equality test, then adjust
24807 ops to be that equality test.
24809 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24811 * config/s390/s390-builtin-types.def: Remove flags argument.
24812 * config/s390/s390.c (s390_init_builtins): Likewise.
24814 2017-02-14 Martin Liska <mliska@suse.cz>
24816 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24817 vector. Fix trailing white spaces.
24819 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24821 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24824 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24826 PR rtl-optimization/68664
24827 * config/arm/arm.c (arm_sched_can_speculate_insn):
24828 New function. Declare prototype.
24829 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24831 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24833 PR rtl-optimization/68664
24834 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24836 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24838 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24840 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24841 max skip bytes for function, loop and jump.
24843 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24845 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24846 ABS_EXPR for gimple dump.
24848 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24851 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24853 PR tree-optimization/79408
24854 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24855 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24856 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24857 also if rhs1 is INTEGER_CST.
24859 2017-02-14 Richard Biener <rguenther@suse.de>
24861 PR middle-end/79432
24862 * tree-into-ssa.c (insert_phi_nodes): When the function can
24863 have abnormal edges rewrite SSA names with broken use-def
24864 dominance out of SSA and register them for PHI insertion.
24866 2017-02-13 Martin Sebor <msebor@redhat.com>
24868 PR middle-end/79496
24869 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24870 clearing info.nowrite flag when snprintf size argument is a range.
24872 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24874 * cprop.c (cprop_jump): Add missing space in string literal.
24875 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24876 (get_constraint_for_component_ref): Likewise.
24877 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24878 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24879 * lra-constraints.c (process_alt_operands): Likewise.
24880 * ipa-inline.c (inline_small_functions): Likewise.
24881 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24882 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24883 * trans-mem.c (diagnose_tm_1_op): Likewise.
24884 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24885 (grid_parallel_clauses_gridifiable): Likewise.
24887 * config/nvptx/mkoffload.c (process): Add space in between
24890 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24891 "MOD4_SSE_REGS" and "ALL_REGS".
24893 * spellcheck.c (test_data): Add , in between "foo" and "food".
24895 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24898 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24899 boundary crossing check and subsequent code generation agree.
24901 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24903 * config/aarch64/aarch64.c (has_memory_op): Delete.
24904 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24907 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24909 PR rtl-optimization/79388
24910 PR rtl-optimization/79450
24911 * combine.c (distribute_notes): When removing TEM_INSN for which
24912 corresponding dest has last value recorded, invalidate that last
24915 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24917 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24918 of explicit '@'. Add missing assembly comment marker on branch costs
24921 2017-02-13 Nathan Sidwell <nathan@acm.org>
24923 * gengtype-lex.l (<in_struct>): Add '/'.
24925 2017-02-13 Martin Liska <mliska@suse.cz>
24928 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24930 2017-02-13 Richard Biener <rguenther@suse.de>
24932 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24934 * configure: Re-generate.
24935 * config.in: Likewise.
24936 * graphite-dependences.c: Simplify as if
24937 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24938 * graphite-isl-ast-to-gimple.c: Likewise.
24939 * graphite-optimize-isl.c: Likewise.
24940 * graphite-poly.c: Likewise.
24941 * graphite-sese-to-poly.c: Likewise.
24942 * graphite.h: Likewise.
24943 * toplev.c: Include isl/version.h and use isl_version () for
24944 printing the ISL version.
24945 * doc/install.texi: Update ISL requirement.
24947 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24949 * doc/standards.texi (Standards): Update reference to
24952 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24954 * doc/extend.texi (Named Address Spaces): sourceware.org now
24956 * doc/install.texi (Binaries): Ditto.
24959 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24961 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24962 standard terminology "stringize"/"stringizing" throughout.
24963 * doc/cppinternals.texi: Likewise.
24965 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24967 * doc/extend.texi: Fix some spelling mistakes and typos.
24968 * doc/invoke.texi: Likewise.
24970 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24973 * params.def (inline-min-speedup) Change from 10 to 8.
24975 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24977 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24980 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24983 * ipa-inline-analysis.c (get_minimal_bb): New function.
24984 (record_modified): Use it.
24985 (remap_edge_change_prob): Handle also ancestor functions.
24987 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24989 * doc/contrib.texi (Contributors): Remove broken link into
24990 the Mauve CVS repository.
24992 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24994 PR middle-end/79454
24995 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24996 result computation whenever lhs doesn't have vector mode, not
24997 just when it has BLKmode.
24999 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25001 * doc/makefile.texi (profiledbootstrap): Refer to the
25002 installation instructions only in textual form.
25004 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25007 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25009 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25011 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25012 (Specific): Update mingw-w64 reference.
25014 (Specific): Remove broken link to Renesas RX processor.
25016 2017-02-10 Richard Biener <rguenther@suse.de>
25018 * toplev.c (process_options): Do not mention obsolete graphite
25019 options when printing sorry message about missing graphite support.
25020 Mention -floop-nest-optimize.
25022 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25024 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25025 (vtst_p16): Likewise.
25026 (vtstq_p8): Likewise.
25027 (vtstq_p16): Likewise.
25029 (vtstq_p64): Likewise.
25030 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25031 (vset_lane_p64): New.
25032 (vsetq_lane_p64): New.
25034 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25036 PR tree-optimization/79411
25037 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25038 stmt operands are SSA_NAMEs used in abnormal phis.
25039 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25042 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25045 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25048 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25050 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25052 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25054 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25058 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25059 not arbitrary TREE_CONSTANT.
25062 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25063 "omp declare target link" attribute unless is_global_var.
25064 * omp-offload.c (find_link_var_op): Likewise.
25066 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25067 Chung-Lin Tang <cltang@codesourcery.com>
25069 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25071 (gimplify_adjust_omp_clauses): Don't delete TILE.
25072 (gimplify_omp_for): Deal with TILE.
25073 * internal-fn.c (expand_GOACC_TILE): New function.
25074 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25076 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25077 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25079 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25080 avoid DIV for outermost collapse var.
25081 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25082 Remove out of date comments, fix whitespace.
25083 * omp-general.c (omp_extract_for_data): Deal with tiling.
25084 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25085 adjust OLF_DIM_BASE value.
25086 (struct omp_for_data): Add tiling field.
25087 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25088 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25089 for auto loops. Remove default auto determining, moved to
25090 oacc_loop_fixed_partitions.
25091 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25092 stmts, add e_mask field.
25093 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25094 (oacc_thread_numbers): Use oacc_dim_call.
25095 (oacc_xform_tile): New.
25096 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25097 (finish_oacc_loop): Adjust for ifns vector.
25098 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25099 add case for GOACC_TILE fns.
25100 (oacc_loop_xform_loop): Delete.
25101 (oacc_loop_process): Iterate over call list directly, and add
25102 handling for GOACC_TILE fns.
25103 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25105 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25106 vector partitioning to outer loops. Assign 2 partitions to loops
25107 when available. Add TILE handling.
25108 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25109 (execite_oacc_device_lower): Process GOACC_TILE fns,
25110 ignore unknown specs.
25111 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25112 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25113 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25115 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25117 * configure.ac (ACX_BUGURL): Update.
25118 * configure: Regenerate.
25120 2017-02-09 Richard Biener <rguenther@suse.de>
25122 PR tree-optimization/69823
25123 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25124 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25126 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25128 * config/arc/arc-c.def: Add __NPS400__ definition.
25129 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25130 (TARGET_NPS400): Define.
25132 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25134 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25136 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25137 pointer, arch_info.
25138 (arc_cpu_types): Fill the arch_info field with a pointer into the
25139 arc_arch_types table.
25140 (arc_selected_cpu): Declare.
25141 * config/arc/arc.c (arc_selected_cpu): Make global.
25142 (arc_selected_arch): Delete.
25143 (arc_base_cpu): Delete.
25144 (arc_override_options): Remove references to deleted variables,
25145 update access to arch information.
25146 (ARC_OPT): Update access to arch information.
25147 (ARC_OPTX): Likewise.
25148 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25149 (TARGET_ARC600): Update access to arch information.
25150 (TARGET_ARC601): Likewise.
25151 (TARGET_ARC700): Likewise.
25152 (TARGET_EM): Likewise.
25153 (TARGET_HS): Likewise.
25154 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25157 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25160 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25161 condition/operands for integer GE/LE/GEU/LEU operations.
25163 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25165 PR translation/79397
25166 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25169 2017-02-08 Martin Jambor <mjambor@suse.cz>
25172 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25173 whether allocation happened.
25174 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25175 nothing was allocated.
25177 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25179 PR tree-optimization/79408
25180 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25181 constant, but SSA_NAME with a known integer range, use the minimum
25182 of that range instead of op1 to determine if modulo can be replaced
25183 with its first operand.
25185 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25187 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25189 2017-02-08 Richard Biener <rguenther@suse.de>
25191 PR tree-optimization/71824
25192 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25193 Check all loops contained in the merged region.
25195 2017-02-07 Andrew Pinski <apinski@cavium.com>
25197 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25199 2017-02-07 Andrew Pinski <apinski@cavium.com>
25201 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25202 (thunderxt88): Likewise.
25203 (thunderxt81): Disable LSE and change v8.1 to v8.
25204 (thunderxt83): Likewise.
25206 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25207 Richard Biener <rguenther@suse.de>
25209 PR middle-end/79399
25210 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25211 type from int to size_t.
25212 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25214 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25216 PR rtl-optimization/79386
25217 * cprop.c (bypass_conditional_jumps): Initialize
25218 bypass_last_basic_block already before splitting bbs after
25219 unconditional traps...
25220 (bypass_conditional_jumps): ... rather than here.
25223 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25224 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25225 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25226 fix -masm=intel patterns.
25228 2017-02-07 Richard Biener <rguenther@suse.de>
25230 PR tree-optimization/79256
25231 PR middle-end/79278
25232 * builtins.c (get_object_alignment_2): Use min_align_of_type
25233 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25234 and ADJUST_FIELD_ALIGN.
25236 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25238 * doc/tm.texi: Regenerate.
25239 * stor-layout.c (layout_decl): Adjust.
25240 (update_alignment_for_field): Likewise.
25241 (place_field): Likewise.
25242 (min_align_of_type): Likewise.
25243 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25244 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25245 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25246 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25247 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25248 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25249 * config/i386/i386.c (x86_field_alignment): Likewise.
25250 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25251 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25252 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25253 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25254 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25255 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25259 2017-01-30 Richard Biener <rguenther@suse.de>
25261 PR tree-optimization/79256
25262 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25263 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25266 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25268 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25269 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25270 builtins to SImode and emit a zero-extend, if necessary.
25272 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25274 * docs/invoke.texi (RISC-V Options): Alphabetize.
25276 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25278 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25281 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25283 * config/riscv/riscv.c: New file.
25284 * common/config/riscv/riscv-common.c: Likewise.
25285 * config.gcc: Likewise.
25286 * config/riscv/constraints.md: Likewise.
25287 * config/riscv/elf.h: Likewise.
25288 * config/riscv/generic.md: Likewise.
25289 * config/riscv/linux.h: Likewise.
25290 * config/riscv/multilib-generator: Likewise.
25291 * config/riscv/peephole.md: Likewise.
25292 * config/riscv/pic.md: Likewise.
25293 * config/riscv/predicates.md: Likewise.
25294 * config/riscv/riscv-builtins.c: Likewise.
25295 * config/riscv/riscv-c.c: Likewise.
25296 * config/riscv/riscv-ftypes.def: Likewise.
25297 * config/riscv/riscv-modes.def: Likewise.
25298 * config/riscv/riscv-opts.h: Likewise.
25299 * config/riscv/riscv-protos.h: Likewise.
25300 * config/riscv/riscv.h: Likewise.
25301 * config/riscv/riscv.md: Likewise.
25302 * config/riscv/riscv.opt: Likewise.
25303 * config/riscv/sync.md: Likewise.
25304 * config/riscv/t-elf-multilib: Likewise.
25305 * config/riscv/t-linux: Likewise.
25306 * config/riscv/t-linux-multilib: Likewise.
25307 * config/riscv/t-riscv: Likewise.
25308 * configure.ac: Likewise.
25309 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25310 Waterman as RISC-V maintainers.
25311 * doc/install.texi: Add RISC-V entries.
25312 * doc/invoke.texi: Add RISC-V options section.
25313 * doc/md.texi: Add RISC-V constraints section.
25314 * configure: Regenerated.
25316 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25319 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25320 false values to be constant vectors with all 0 or all 1 bits set.
25321 (vcondu<mode><mode>): Likewise.
25322 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25324 (fpmask_comparison_operator): Update comment.
25325 (vecint_comparison_operator): New predicate.
25326 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25327 vector conditionals when the true and false values are constant
25328 vectors with all 0 bits or all 1 bits set.
25330 2017-02-06 Martin Sebor <msebor@redhat.com>
25332 PR tree-optimization/79376
25333 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25335 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25337 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25338 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25339 to simplify split condition.
25341 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25343 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25344 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25347 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25349 PR rtl-optimization/68664
25350 * target.def (can_speculate_insn): New hook.
25351 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25352 * doc/tm.texi: Regenerate.
25353 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25354 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25355 (rs6000_sched_can_speculate_insn): New function.
25357 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25359 PR tree-optimization/79284
25360 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25361 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25362 vectorizable_mask_load_store, vectorizable_operation,
25363 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25364 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25365 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25366 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25367 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25368 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25369 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25370 is_gimple_assign (stmt). Replace another such test with
25371 is_gimple_assign (stmt).
25373 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25376 * config/avr/avr.c (rtl-iter.h): Include it.
25377 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25378 (avr_legitimate_combined_insn): ...and implementation.
25380 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25382 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25383 * config/s390/s390.c (s390_const_operand_ok)
25384 (s390_canonicalize_comparison, s390_extract_part)
25385 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25386 (s390_contiguous_bitmask_p, s390_rtx_costs)
25387 (legitimize_pic_address): Likewise.
25388 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25389 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25390 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25391 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25392 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25394 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25396 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25397 REGNO($0) == REGNO($1).
25399 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25401 * config/s390/linux.h(SIZE_TYPE): Add comment.
25403 2017-02-06 Julian Brown <julian@codesourcery.com>
25404 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25405 Virendra Pathak <virendra.pathak@broadcom.com>
25407 * config/aarch64/aarch64-cores.def: Change the scheduler
25409 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25410 * config/aarch64/thunderx2t99.md: New file.
25412 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25414 * doc/standards.texi (Go Language): Update link to language
25417 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25419 * tree-eh.c (lower_resx): Sanitize profile.
25420 (cleanup_empty_eh_move_lp): Likewise.
25422 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25425 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25427 * cfgloopmanip.h (loop_version): Update prototype.
25428 * modulo-sched.c (sms_schedule): Update call of loop_version.
25429 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25430 * tree-parloops.c (gen_parallel_loop): Likewise.
25431 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25432 * tree-ssa-loop-split.c (split_loop): Likewise.
25433 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25434 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25436 2017-02-05 Martin Liska <mliska@suse.cz>
25439 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25441 (print_operand_address): Initialize a struct to zero.
25443 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25445 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25446 garbage collector only in textual form.
25448 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25450 * doc/extend.texi (x86 specific memory model extensions for
25451 transactional memory): Simplify a phrase.
25453 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25456 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25457 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25458 (atomic_storedi_1): Likewise.
25460 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25462 PR tree-optimization/79338
25463 * tree-parloops.c (gather_scalar_reductions): Don't call
25464 vect_analyze_loop_form for loop->inner before destroying loop's
25467 2017-02-03 Martin Sebor <msebor@redhat.com>
25469 PR tree-optimization/79327
25470 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25471 when precision has resulted in leading zeros.
25472 (format_integer): Adjust the likely counter to assume an unknown
25473 argument that may be zero is non-zero.
25475 2017-02-03 Jason Merrill <jason@redhat.com>
25478 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25479 avoid copying non-taken branch.
25481 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25483 PR tree-optimization/79340
25484 * tree-vect-loop.c (vectorizable_reduction): Release
25485 vec_defs elements after safe_splicing them into other vectors.
25488 PR tree-optimization/79327
25489 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25490 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25492 (format_integer): Use wide_int_to_tree instead of build_int_cst
25493 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25494 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25495 of shortest and longest sequence.
25497 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25499 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25500 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25502 2017-02-03 Walter Lee <walt@tilera.com>
25505 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25506 after initial stackframe link reg save.
25507 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25509 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25512 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25513 wu for stxssp alternative.
25515 2017-02-03 Martin Sebor <msebor@redhat.com>
25517 PR tree-optimization/79352
25518 * gimple-fold.c (get_range_strlen): Add argument.
25519 (get_range_strlen): Change return type to bool.
25520 (get_maxval_strlen): Pass in a dummy argument.
25521 * gimple-fold.h (get_range_strlen): Change return type to bool.
25522 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25523 * tree.h (array_at_struct_end_p): Add argument.
25524 * tree.c (array_at_struct_end_p): Handle it.
25526 2017-02-03 Martin Liska <mliska@suse.cz>
25529 * multiple_target.c (create_dispatcher_calls): Redirect edge
25530 from a caller of a dispatcher.
25531 (expand_target_clones): Make the clones local.
25532 (ipa_target_clone): Do both target clones and resolvers.
25533 (ipa_dispatcher_calls): Remove the pass.
25534 (pass_dispatcher_calls::gate): Likewise.
25535 (make_pass_dispatcher_calls): Likewise.
25536 * passes.def (pass_target_clone): Put as very first IPA early
25539 2017-02-03 Martin Liska <mliska@suse.cz>
25541 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25542 in case of a function with ifunc attribute.
25544 2017-02-03 Martin Liska <mliska@suse.cz>
25546 * cgraph.c (cgraph_node::dump): Dump function version info.
25547 * symtab.c (symtab_node::dump_base): Add missing new line.
25549 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25551 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25552 (ifcombine_ifandif): Use it.
25554 2017-02-03 Martin Liska <mliska@suse.cz>
25556 * doc/invoke.texi: Document default value for
25557 use-after-scope-direct-emission-threshold.
25559 2017-02-03 Martin Liska <mliska@suse.cz>
25561 PR tree-optimization/79339
25562 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25563 (format_floating): Likewise.
25565 2017-02-03 Martin Liska <mliska@suse.cz>
25568 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25570 (ipa_node_params_t::remove): Likewise.
25571 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25572 initialization from removed ipa_node_params_t::insert.
25573 (ipa_node_params::~ipa_node_params): Move from removed
25574 ipa_node_params_t::release.
25575 * symbol-summary.h (symbol_summary::m_released): New member.
25576 Do not release a summary twice. Do not allow to call finalizer
25577 for types of a summary that live in GGC memory.
25579 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25581 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25584 2017-02-02 Martin Sebor <msebor@redhat.com>
25586 PR middle-end/79275
25587 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25588 (format_string): Tighten up the range of output for non-constant
25589 strings and correct the expected range for wide non-constant strings.
25591 2017-02-02 Martin Sebor <msebor@redhat.com>
25593 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25595 PR middle-end/32003
25596 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25598 (-fdump-tree-@var): Add to index and document how to come up
25599 with pass-specific option and dump file names.
25600 (-fdump-passes): Clarify where to look for output.
25602 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25604 PR middle-end/77445
25605 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25606 statistics of the analyzed path; allow threading for speed when
25607 any of BBs along the path are optimized for speed.
25609 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25611 PR middle-end/78468
25612 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25613 settings of the virtual registers.
25616 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25618 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25619 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25622 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25624 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25625 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25627 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25629 * config/s390/s390.md: Add missing comments with the expanded
25631 * config/s390/vector.md: Likewise.
25632 * config/s390/vx-builtins.md: Likewise.
25634 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25637 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25638 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25639 conditions on a single line.
25641 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25643 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25644 __S390_VX__ to __VX__.
25646 2017-02-01 Andrew Pinski <apinski@cavium.com>
25648 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25649 stmt_info to record_stmt_cost.
25650 (vect_get_known_peeling_cost): Pass stmt_info if known to
25652 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25653 cpu_vector_cost field into
25654 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25655 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25656 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25657 splitting of scalar_stmt_cost and vec_stmt_cost.
25658 (thunderx_vector_cost): Likewise.
25659 (cortexa57_vector_cost): LIkewise.
25660 (exynosm1_vector_cost): Likewise.
25661 (xgene1_vector_cost): Likewise.
25662 (thunderx2t99_vector_cost): Improve after the splitting of the two
25664 (aarch64_builtin_vectorization_cost): Update for the splitting of
25665 scalar_stmt_cost and vec_stmt_cost.
25667 2017-02-01 Torvald Riegel <triegel@redhat.com>
25668 Richard Henderson <rth@redhat.com>
25670 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25671 conditional on existance of a fast atomic load.
25672 * optabs-query.c (can_atomic_load_p): New function.
25673 * optabs-query.h (can_atomic_load_p): Declare it.
25674 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25675 no fast atomic load is available for the particular size of access.
25676 (expand_atomic_compare_and_swap): Likewise.
25677 (expand_atomic_load): Likewise.
25678 (expand_atomic_store): Likewise.
25679 (expand_atomic_fetch_op): Likewise.
25680 * testsuite/lib/target-supports.exp
25681 (check_effective_target_sync_int_128): Remove x86 because it provides
25682 no fast atomic load.
25683 (check_effective_target_sync_int_128_runtime): Likewise.
25685 2017-02-01 Richard Biener <rguenther@suse.de>
25687 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25688 (graphite_transform_loops): Provide opt-info for optimized nests.
25689 * tree-parloop.c (parallelize_loops): Provide opt-info for
25690 parallelized loops.
25692 2017-02-01 Richard Biener <rguenther@suse.de>
25694 PR middle-end/79315
25695 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25696 was not set before.
25698 2017-02-01 Richard Biener <rguenther@suse.de>
25700 PR tree-optimization/71824
25701 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25702 Verify the loops are valid in the merged SESE region.
25703 (scop_detection::can_represent_loop_1): Check analyzing the
25704 evolution of the number of iterations in the region succeeds.
25706 2017-01-31 Ian Lance Taylor <iant@golang.org>
25708 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25709 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25711 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25713 PR preprocessor/79210
25714 * input.c (get_substring_ranges_for_loc): Replace line_width
25715 assertion with error-handling.
25717 2017-01-31 Richard Biener <rguenther@suse.de>
25719 PR tree-optimization/77318
25720 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25721 (create_pw_aff_from_tree): Take loop parameter.
25722 (add_condition_to_pbb): Pass loop of the condition to
25723 create_pw_aff_from_tree.
25725 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25727 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25728 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25730 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25734 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25736 (convert_int_to_float128): Likewise.
25737 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25738 (convert_int_to_float128): Likewise.
25739 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25740 (UNSPEC_IEEE128_CONVERT): Likewise.
25741 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25742 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25743 Use local variables for IBM extended format.
25744 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25745 (fix_trunc<mode>si2_fprs): Likewise.
25746 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25747 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25748 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25749 to know that we can now have integers of all sizes in vector
25751 (fix<uns>_<mode>di2_hw): Likewise.
25752 (float<uns>_<mode>si2_hw): Likewise.
25753 (fix_<mode>si2_hw): Likewise.
25754 (fixuns_<mode>si2_hw): Likewise.
25755 (float<uns>_<mode>di2_hw): Likewise.
25756 (float_<mode>di2_hw): Likewise.
25757 (float_<mode>si2_hw): Likewise.
25758 (floatuns_<mode>di2_hw): Likewise.
25759 (floatuns_<mode>si2_hw): Likewise.
25760 (xscvqp<su>wz_<mode>): Delete, no longer used.
25761 (xscvqp<su>dz_<mode>): Likewise.
25762 (xscv<su>dqp_<mode>): Likewise.
25763 (ieee128_mfvsrd_64bit): Likewise.
25764 (ieee128_mfvsrd_32bit): Likewise.
25765 (ieee128_mfvsrwz): Likewise.
25766 (ieee128_mtvsrw): Likewise.
25767 (ieee128_mtvsrd_64bit): Likewise.
25768 (ieee128_mtvsrd_32bit): Likewise.
25770 2017-01-31 Martin Liska <mliska@suse.cz>
25773 * ipa-prop.c (ipa_free_all_node_params): Call release method
25774 instead of ~sumbol_summary to not to trigger double times
25777 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25779 PR tree-optimization/71691
25780 * bitmap.h (class auto_bitmap): New.
25781 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25782 is_maybe_undefined instead of ssa_undefined_value_p.
25784 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25786 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25787 __S390_ARCH_LEVEL__ to __ARCH__.
25789 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25791 PR tree-optimization/79267
25792 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25793 if should_remove_lhs_p is true.
25795 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25798 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25799 (add_alignment_attribute): New.
25800 (base_type_die): Add alignment attribute.
25801 (subrange_type_die): Likewise.
25802 (modified_type_die): Likewise.
25803 (gen_array_type_die): Likewise.
25804 (gen_descr_array_type_die: Likewise.
25805 (gen_enumeration_type_die): Likewise.
25806 (gen_subprogram_die): Likewise.
25807 (gen_variable_die): Likewise.
25808 (gen_field_die): Likewise.
25809 (gen_ptr_to_mbr_type_die): Likewise.
25810 (gen_struct_or_union_type_die): Likewise.
25811 (gen_subroutine_type_die): Likewise.
25812 (gen_typedef_die): Likewise.
25813 (base_type_cmp): Compare alignment attribute.
25815 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25818 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25819 (setb_unsigned) New pattern for setb with CCUNS.
25820 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25821 subfc./subfe sequence to avoid overflow problems. Generate a
25822 shorter sequence with cmpld/setb for power9.
25823 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25824 for generating subfc. instruction.
25825 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25826 now uses this instruction.
25828 2017-01-30 Ian Lance Taylor <iant@google.com>
25831 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25832 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25834 2017-01-30 Martin Sebor <msebor@redhat.com>
25836 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25837 Move constant to the right of a relational operator.
25838 (get_mpfr_format_length, format_character, format_string): Ditto.
25839 (should_warn_p, maybe_warn): Same.
25841 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25843 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25846 * asan.c (get_translation_unit_decl): Remove function.
25847 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25849 2017-01-30 Martin Liska <mliska@suse.cz>
25851 PR gcov-profile/79259
25852 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25853 -fprofile-generate.
25855 2017-01-30 Martin Liska <mliska@suse.cz>
25858 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25859 Initialize variables with NULL value.
25861 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25864 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25866 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25868 2017-01-30 Richard Biener <rguenther@suse.de>
25870 PR tree-optimization/79276
25871 * tree-vrp.c (process_assert_insertions): Properly adjust common
25872 when removing a duplicate.
25874 * gcc.dg/torture/pr79276.c: New testcase.
25876 2017-01-30 Richard Biener <rguenther@suse.de>
25878 PR tree-optimization/79256
25879 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25880 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25882 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25884 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25887 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25888 ("*r<noxa>sbg_<mode>_sll_bitmask")
25889 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25890 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25891 Use contiguous_bitmask_nowrap_operand.
25893 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25896 * config/rs6000/altivec.h (vec_xl): Revise #define.
25897 (vec_xst): Likewise.
25899 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25901 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25903 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25905 PR rtl-optimization/79194
25906 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25907 traps before call to bypass_conditional_jumps.
25909 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25911 PR tree-optimization/71374
25912 * lra-constraints.c (check_conflict_input_operands): New.
25913 (match_reload): Use it.
25915 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25918 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25919 account to calculate conflict_set.
25921 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25923 PR rtl-optimization/78559
25924 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25925 other_insn in combine.
25927 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25929 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25930 uint16_type_node for BT_UINT16.
25932 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25934 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25935 "RTL Tests" to menu.
25936 (GIMPLE Tests): New node.
25937 (RTL Tests): New node.
25939 2017-01-27 Richard Biener <rguenther@suse.de>
25941 PR tree-optimization/79245
25942 * tree-loop-distribution.c (distribute_loop): Apply cost
25943 modeling also to detected patterns.
25945 2017-01-27 Richard Biener <rguenther@suse.de>
25947 PR tree-optimization/71433
25948 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25949 (compare_assert_loc): New function.
25950 (process_assert_insertions): Sort and optimize assert locations
25951 to remove duplicates and push down identical assertions on
25952 edges to their destination block.
25954 2017-01-27 Richard Biener <rguenther@suse.de>
25956 PR tree-optimization/79244
25957 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25958 out SSA names even if abnormal.
25960 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25962 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25963 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25964 instead of MPFR_RNDN.
25966 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25969 * arm.c (arm_option_override): Don't call build_target_option_node
25970 until after doing all option overrides.
25971 (arm_valid_target_attribute_tree): Likewise.
25973 2017-01-27 Martin Liska <mliska@suse.cz>
25975 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25976 for {cd}tors and C++ {cd}tors.
25978 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25980 * config/s390/s390.md ("*setmem_long_and")
25981 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25983 2017-01-26 Martin Sebor <msebor@redhat.com>
25985 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25988 2017-01-26 Martin Sebor <msebor@redhat.com>
25990 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25991 HAVE_DFmode before using XFmode or DFmode.
25992 (parse_directive): Avoid using the z length modifier to avoid
25993 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25995 PR middle-end/78703
25996 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25997 to accept adjustment as an array.
25998 (get_int_range): New function.
25999 (struct directive): Make width and prec arrays.
26000 (directive::set_width, directive::set_precision): Call get_int_range.
26001 (format_integer, format_floating): Handle width and precision ranges.
26002 (format_string, parse_directive): Same.
26004 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26007 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26008 comdat_type_p set, just clone them, but keep the children in the
26012 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26013 which have direct callers with -fvar-tracking-assignments enabled
26015 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26016 inside of type units.
26018 2017-01-26 Martin Sebor <msebor@redhat.com>
26020 PR middle-end/78703
26021 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26023 (struct format_result): Replace number_chars, number_chars_min,
26024 and number_chars_max with a single member of struct result_range.
26026 (format_result::operator+=): Adjust.
26027 (struct fmtresult): Remove bounded. Handle likely and unlikely
26029 (fmtresult::adjust_for_width_or_precision): New function.
26030 (fmtresult:type_max_digits): New function.
26031 (bytes_remaining): Handle likely and unlikely counters.
26032 (min_bytes_remaining): Remove.
26033 (format_percent): Simplify.
26034 (format_integer, format_floating): Set likely and unlikely counters.
26035 (get_string_length, format_character, format_string): Same.
26036 (format_plain, should_warn_p): New function.
26037 (maybe_warn): Call should_warn_p. Update diagnostic messages
26038 and handle those for all directives, including plain strings.
26039 (format_directive): Handle likely and unlikely counters.
26040 Remove unnecessary quoting from diagnostics. Add an informational
26042 (add_bytes): Remove.
26043 (pass_sprintf_length::compute_format_length): Simplify.
26044 (try_substitute_return_value): Handle likely and unlikely counters.
26046 2017-01-26 Carl Love <cel@us.ibm.com>
26048 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26049 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26051 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26054 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26055 endianess for subregs into account.
26056 * lra-constraints.c (lra_constraints): Do risky transformations
26057 always on the first iteration.
26058 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26059 last_call_used_reg_set.
26060 (process_bb_lives): Define and use last_call_used_reg_set.
26061 * lra.c (lra): Always continue after lra_constraints on the first
26064 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26066 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26068 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26070 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26072 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26073 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26074 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26075 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26076 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26077 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26078 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26079 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26080 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26082 2017-01-26 Marek Polacek <polacek@redhat.com>
26085 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26086 for the third operand.
26088 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26090 PR middle-end/79236
26091 * omp-low.c (struct omp_context): Add simt_stmt field.
26092 (scan_omp_for): Return omp_context *.
26093 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26094 context to the _simt_ SIMD stmt.
26095 (lower_omp_for): For combined SIMD with sibling _simt_
26096 SIMD, make sure to use the same decls in _looptemp_
26097 clauses as in the sibling.
26099 2017-01-26 David Sherwood <david.sherwood@arm.com>
26101 PR middle-end/79212
26102 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26105 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26108 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26109 emit fld b; fld a; if possible.
26111 * brig-builtins.def: Update copyright years.
26112 * config/arm/arm_acle_builtins.def: Update copyright years.
26114 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26117 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26118 constraint instead of o for the stxsd instruction.
26120 2017-01-25 Carl Love <cel@us.ibm.com>
26122 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26123 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26125 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26127 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26129 2017-01-25 Richard Biener <rguenther@suse.de>
26131 PR tree-optimization/69264
26132 * target.def (vector_alignment_reachable): Improve documentation.
26133 * doc/tm.texi: Regenerate.
26134 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26136 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26137 earlier changes with respect to TYPE_USER_ALIGN.
26138 (vector_alignment_reachable_p): Likewise. Improve dumping.
26140 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26143 * config/arm/arm.md (xordi3): Force constant operand into a register
26146 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26148 * doc/invoke.texi (-fstore-merging): Correct default optimization
26149 levels at which it is enabled.
26150 (-O): Move -fstore-merging from list to...
26153 2017-01-25 Richard Biener <rguenther@suse.de>
26156 * omp-expand.c: Include debug.h.
26157 (expand_omp_taskreg): Make sure to generate early debug before
26158 outlining anything from a function.
26159 (expand_omp_target): Likewise.
26160 (grid_expand_target_grid_body): Likewise.
26162 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26165 * asan.c (get_translation_unit_decl): New function.
26166 (asan_add_global): Extract modules file name from globals
26167 TRANSLATION_UNIT_DECL name.
26169 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26172 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26173 for long calls with APCS frame and VFP.
26175 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26177 * cfg.c (original_copy_tables_initialized_p): New function.
26178 * cfg.h (original_copy_tables_initialized_p): New decl.
26179 * cfgrtl.c (relink_block_chain): Guard the call to
26180 free_original_copy_tables with a call to
26181 original_copy_tables_initialized_p.
26182 * cgraph.h (symtab_node::native_rtl_p): New decl.
26183 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26184 (symtab_node::needed_p): Don't assert for early assembly output
26185 for __RTL functions.
26186 (cgraph_node::finalize_function): Set "force_output" for __RTL
26188 (cgraph_node::analyze): Bail out early for __RTL functions.
26189 (analyze_functions): Update assertion to support __RTL functions.
26190 (cgraph_node::expand): Bail out early for __RTL functions.
26191 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26193 * function.h (struct function): Update comment for field
26195 * gimple-expr.c: Include "tree-pass.h".
26196 (gimple_has_body_p): Return false for __RTL functions.
26197 * Makefile.in (OBJS): Add run-rtl-passes.o.
26198 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26200 (gcc::pass_manager::get_clean_slate): New accessor.
26201 * passes.c: Include "insn-addr.h".
26202 (should_skip_pass_p): Add logging. Update logic for running
26203 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26204 property-provider override so it is only done for gimple passes.
26206 (skip_pass): New function.
26207 (execute_one_pass): Call skip_pass when skipping passes.
26208 * read-md.c (md_reader::read_char): Support filtering
26209 the input to a subset of line numbers.
26210 (md_reader::md_reader): Initialize fields
26211 m_first_line and m_last_line.
26212 (md_reader::read_file_fragment): New function.
26213 * read-md.h (md_reader::read_file_fragment): New decl.
26214 (md_reader::m_first_line): New field.
26215 (md_reader::m_last_line): New field.
26216 * read-rtl-function.c (function_reader::create_function): Only
26217 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26218 curr_properties. Set DECL_INITIAL to a dummy block.
26219 (read_rtl_function_body_from_file_range): New function.
26220 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26222 * run-rtl-passes.c: New file.
26223 * run-rtl-passes.h: New file.
26225 2017-01-24 Jeff Law <law@redhat.com>
26227 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26230 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26232 PR tree-optimization/79159
26233 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26234 (record_nonwrapping_iv): Improve boundary using above function if no
26235 value range information.
26237 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26238 Martin Jambor <mjambor@suse.cz>
26240 * brig-builtins.def: New file.
26241 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26242 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26243 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26244 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26245 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26246 * builtin-types.def (BT_INT8): New.
26247 (BT_INT16): Likewise.
26248 (BT_UINT8): Likewise.
26249 (BT_UINT16): Likewise.
26250 (BT_FN_ULONG): Likewise.
26251 (BT_FN_UINT_INT): Likewise.
26252 (BT_FN_UINT_ULONG): Likewise.
26253 (BT_FN_UINT_LONG): Likewise.
26254 (BT_FN_UINT_PTR): Likewise.
26255 (BT_FN_ULONG_PTR): Likewise.
26256 (BT_FN_INT8_FLOAT): Likewise.
26257 (BT_FN_INT16_FLOAT): Likewise.
26258 (BT_FN_UINT32_FLOAT): Likewise.
26259 (BT_FN_UINT16_FLOAT): Likewise.
26260 (BT_FN_UINT8_FLOAT): Likewise.
26261 (BT_FN_UINT64_FLOAT): Likewise.
26262 (BT_FN_UINT16_UINT32): Likewise.
26263 (BT_FN_UINT32_UINT16): Likewise.
26264 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26265 (BT_FN_INT_PTR_INT): Likewise.
26266 (BT_FN_UINT_PTR_UINT): Likewise.
26267 (BT_FN_LONG_PTR_LONG): Likewise.
26268 (BT_FN_ULONG_PTR_ULONG): Likewise.
26269 (BT_FN_VOID_UINT64_UINT64): Likewise.
26270 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26271 (BT_FN_INT8_INT8_INT8): Likewise.
26272 (BT_FN_INT16_INT16_INT16): Likewise.
26273 (BT_FN_INT_INT_INT): Likewise.
26274 (BT_FN_UINT_FLOAT_UINT): Likewise.
26275 (BT_FN_FLOAT_UINT_UINT): Likewise.
26276 (BT_FN_ULONG_UINT_UINT): Likewise.
26277 (BT_FN_ULONG_UINT_PTR): Likewise.
26278 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26279 (BT_FN_UINT_UINT_UINT): Likewise.
26280 (BT_FN_VOID_UINT_PTR): Likewise.
26281 (BT_FN_UINT_UINT_PTR: Likewise.
26282 (BT_FN_UINT32_UINT64_PTR): Likewise.
26283 (BT_FN_INT_INT_UINT_UINT): Likewise.
26284 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26285 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26286 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26287 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26288 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26289 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26290 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26291 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26292 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26293 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26294 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26295 * doc/frontends.texi: List BRIG FE.
26296 * doc/install.texi (Testing): Add BRIG tesring requirements.
26297 * doc/invoke.texi (Overall Options): Mention BRIG.
26298 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26300 2017-01-24 Richard Biener <rguenther@suse.de>
26302 PR translation/79208
26303 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26305 2017-01-24 Martin Jambor <mjambor@suse.cz>
26308 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26309 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26310 and known_contexts.
26312 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26314 PR middle-end/79123
26315 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26316 casts from signed to unsigned really don't have a range.
26318 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26320 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26321 GMP_RNDx for compatiblity.
26323 2017-01-24 Martin Liska <mliska@suse.cz>
26326 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26327 that would prevent us to call alloca with -1 as argument.
26329 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26331 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26332 Avoid -Wformat-security warning.
26334 2017-01-23 Andrew Pinski <apinski@cavium.com>
26336 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26339 2017-01-23 Martin Sebor <msebor@redhat.com>
26341 PR middle-end/78703
26342 * gimple-ssa-sprintf.c (warn_level): New global.
26343 (format_integer): Use it here and throughout the rest of the file.
26344 Use the same switch to compute sign as base.
26345 (maybe_warn): New function.
26346 (format_directive): Factor out warnings into maybe_warn.
26347 Add debugging output. Use warn_level.
26348 (add_bytes): Use warn_level.
26349 (pass_sprintf_length::compute_format_length): Add debugging output.
26350 (try_substitute_return_value): Same.
26351 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26353 PR middle-end/78703
26354 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26355 (struct fmtresult, format_integer, format_floating): Adjust.
26356 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26357 (get_string_length, format_string,format_directive): Same.
26358 (pass_sprintf_length::compute_format_length): Same.
26359 (try_substitute_return_value): Simplify slightly.
26361 PR middle-end/78703
26362 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26363 (fmtresult::operator+=): Outlined.
26364 (struct fmtresult): Add ctors.
26365 (struct conversion_spec): Rename...
26366 (struct directive): ...to this. Add and remove data members.
26367 (directive::set_width, directive::set_precision): New functions.
26368 (format_percent): Use fmtresult ctor.
26369 (get_width_and_precision): Remove.
26370 (format_integer): Make naming changes. Avoid computing width and
26372 (format_floating): Same. Adjust indentation.
26373 (format_character, format_none): New functions.
26374 (format_string): Moved character handling to format_character.
26375 (format_directive): Remove arguments, change return type.
26376 (parse_directive): New function.
26377 (pass_sprintf_length::compute_format_length): Move directive
26378 parsing to parse_directive.
26380 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26382 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26383 (assign_assembler_name_if_needed): ... this.
26384 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26385 (assign_assembler_name_if_needed): ... this.
26386 (free_lang_data_in_cgraph): Adjust callers.
26387 * cgraphunit.c (cgraph_node::analyze): Likewise.
26388 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26391 2017-01-23 Richard Biener <rguenther@suse.de>
26393 PR tree-optimization/79088
26394 PR tree-optimization/79188
26395 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26396 resetting loop bounds after last path deletion. Reset loop
26397 bounds of the target loop, make code match the comments.
26398 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26399 Make sure loops need no fixups.
26401 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26403 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26404 exponent support with double type for first argument.
26405 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26406 type returned by __builtin_vec_extract_sig,
26407 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26408 functions from "vector int" to "vector unsigned int" or from
26409 "vector long long int" to "vector unsigned long long int".
26410 Changed type returned by __builtin_vec_extract_exp,
26411 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26412 functions from "vector int" to "vector unsigned int" or from
26413 "vector long long int" to "vector unsigned long long int".
26414 Changed return type of __builtin_vec_test_data_class,
26415 __builtin_vec_test_data_class_sp, and
26416 __builtin_vec_test_data_class_dp from "vector int" to
26417 "vector bool int" or from "vector long long int" to "vector bool
26418 long long int" and changed second argument type from "unsigned
26419 int" to "int". Added new overloaded function forms "vector float
26420 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26421 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26422 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26423 double, vector unsigned long long int)" and "vector double
26424 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26425 long int)". Changed return type of
26426 __builtin_scalar_test_data_class and
26427 __builtin_scalar_test_data_class_sp and
26428 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26429 int" and changed second argument from "unsigned int" to "int".
26430 Changed type returned by __builtin_scalar_test_neg,
26431 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26432 from "int" to "bool int". Added new overloaded function form
26433 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26434 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26435 exponent double-precision with floating point first argument.
26436 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26437 documentation of scalar_test_data_class, scalar_test_neg,
26438 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26439 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26440 vec_test_data_class built-in functions to reflect refinements in
26441 their type signatures.
26443 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26445 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26447 (aarch64_elf_asm_destructor): Likewise.
26449 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26451 PR rtl-optimization/78634
26452 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26453 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26454 * ifcvt.c (noce_try_cmove): Add missing cost check.
26456 PR rtl-optimization/71724
26457 * combine.c (if_then_else_cond): Look for situations where it is
26458 beneficial to undo the work of one of the recursive calls.
26460 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26462 PR tree-optimization/70754
26463 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26464 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26465 combined stmt before it if not NULL.
26466 (combine_chains): Process refs reversely and compute dominance point
26469 2017-01-23 Martin Liska <mliska@suse.cz>
26471 PR tree-optimization/79196
26472 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26473 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26475 (strlen_optimize_stmt): Call the renamed function.
26477 2017-01-23 Michael Matz <matz@suse.de>
26479 PR tree-optimization/78384
26480 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26482 2017-01-23 Richard Biener <rguenther@suse.de>
26484 PR tree-optimization/79186
26485 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26486 both incoming edges before moving an assert.
26488 2017-01-23 Martin Jambor <mjambor@suse.cz>
26490 * ipa-prop.c (load_from_param_1): Removed.
26491 (load_from_unmodified_param): Bits from load_from_param_1 put back
26493 (load_from_param): Removed.
26494 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26495 with stmt. Reverted back to use of load_from_unmodified_param.
26497 2017-01-23 Martin Jambor <mjambor@suse.cz>
26500 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26501 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26502 field a pointer to garbage collected vector, mark lattices and
26503 ipcp_orig_node with GTY((skip)).
26504 (ipa_get_param_count): Adjust to descriptors being a pointer.
26505 (ipa_get_param): Likewise.
26506 (ipa_get_type): Likewise.
26507 (ipa_get_param_move_cost): Likewise.
26508 (ipa_set_param_used): Likewise.
26509 (ipa_get_controlled_uses): Likewise.
26510 (ipa_set_controlled_uses): Likewise.
26511 (ipa_is_param_used): Likewise.
26512 (ipa_node_params_t): Move into garbage collector. New methods insert
26514 (ipa_node_params_sum): Annotate wth GTY(()).
26515 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26517 (ipa_load_from_parm_agg): Adjust declaration.
26518 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26519 * ipa-profile.c (ipa_profile): Likewise.
26520 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26521 (ipa_populate_param_decls): Make descriptors parameter garbage
26523 (ipa_dump_param): Adjust to descriptors being a pointer.
26524 (ipa_alloc_node_params): Likewise.
26525 (ipa_initialize_node_params): Likewise.
26526 (load_from_param_1): Make descriptors parameter garbage collected.
26527 (load_from_unmodified_param): Likewise.
26528 (load_from_param): Likewise.
26529 (ipa_load_from_parm_agg): Likewise.
26530 (ipa_node_params::~ipa_node_params): Removed.
26531 (ipa_free_all_node_params): Remove call to delete operator.
26532 (ipa_node_params_t::insert): New.
26533 (ipa_node_params_t::remove): Likewise.
26534 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26535 copy known_csts and known_contexts vectors.
26536 (ipa_read_node_info): Adjust to descriptors being a pointer.
26537 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26539 (ipcp_transform_function): Make descriptors variable garbage
26542 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26544 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26545 * config/i386/avx512dqintrin.h: Ditto.
26546 * config/i386/avx512fintrin.h: Ditto.
26547 * config/i386/i386.c: Handle new builtins.
26548 * config/i386/i386-builtin.def: Add new builtins.
26549 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26550 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26552 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26553 Martin Liska <mliska@suse.cz>
26555 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26556 * asan.c (asan_expand_poison_ifn): Support stores and use
26557 appropriate ASAN report function.
26558 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26559 * internal-fn.def (ASAN_POISON_USE): Declare.
26560 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26561 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26562 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26563 ASAN_POISON calls w/o LHS.
26564 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26565 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26566 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26567 * gimplify.c (asan_poison_variables): Add attribute
26568 use_after_scope_memory to variables that really needs to live
26570 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26571 having the attribute.
26573 2017-01-23 Martin Liska <mliska@suse.cz>
26575 * asan.c (create_asan_shadow_var): New function.
26576 (asan_expand_poison_ifn): Likewise.
26577 * asan.h (asan_expand_poison_ifn): New declaration.
26578 * internal-fn.c (expand_ASAN_POISON): Likewise.
26579 * internal-fn.def (ASAN_POISON): New builtin.
26580 * sanopt.c (pass_sanopt::execute): Expand
26581 asan_expand_poison_ifn.
26582 * tree-inline.c (copy_decl_for_dup_finish): Make function
26584 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26585 * tree-ssa.c (is_asan_mark_p): New function.
26586 (execute_update_addresses_taken): Rewrite local variables
26587 (identified just by use-after-scope as addressable) into SSA.
26589 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26591 * doc/install.texi (Specific): opensource.apple.com uses https
26592 now. Remove trailing slash.
26594 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26596 * README.Portability: Remove note on an Irix compatibility issue.
26598 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26600 * gcov.c (INCLUDE_ALGORITHM): Define.
26601 (INCLUDE_VECTOR): Define.
26602 No longer include <vector> and <algorithm> directly.
26604 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26606 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26608 * doc/invoke.texi (Code Gen Options): Ditto.
26610 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26613 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26615 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26617 rtl-optimization/79125
26618 * cprop.c (local_cprop_pass): Handle cases where we make an
26619 unconditional trap.
26621 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26625 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26626 read from, for big endian.
26628 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26630 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26631 register pauth builtins for LP64 only.
26633 2017-01-20 Marek Polacek <polacek@redhat.com>
26636 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26639 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26641 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26643 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26644 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26645 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26647 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26650 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26651 in big-endian mode when they are not a single duplicated value.
26653 2017-01-20 Richard Biener <rguenther@suse.de>
26655 * BASE-VER: Bump to 7.0.1.
26657 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26659 * omp-low.c (omplow_simd_context): New struct. Use it...
26660 (lower_rec_simd_input_clauses): ...here and...
26661 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26662 references to idx, lane, max_vf, is_simt.
26664 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26666 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26669 2017-01-20 Martin Jambor <mjambor@suse.cz>
26671 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26672 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26674 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26675 (GTFILES): Rename hsa.c to hsa-common.c.
26676 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26677 * hsa-dump.c: Likewise.
26678 * hsa-gen.c: Likewise.
26679 * hsa-regalloc.c: Likewise.
26680 * ipa-hsa.c: Likewise.
26681 * omp-expand.c: Likewise.
26682 * omp-low.c: Likewise.
26683 * toplev.c: Likewise.
26685 2017-01-20 Marek Polacek <polacek@redhat.com>
26688 * doc/invoke.texi: Document -Wduplicated-branches.
26689 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26690 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26691 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26692 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26693 return 0 only when not OEP_LEXICOGRAPHIC.
26694 (fold_build_cleanup_point_expr): Use the expression
26695 location when building CLEANUP_POINT_EXPR.
26696 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26697 * tree.c (add_expr): Handle error_mark_node.
26699 2017-01-20 Martin Liska <mliska@suse.cz>
26702 * tree-profile.c (init_ic_make_global_vars): Do not call
26704 (gimple_init_gcov_profiler): Likewise.
26706 2017-01-20 Martin Liska <mliska@suse.cz>
26709 * cgraph.h (maybe_create_reference): Remove argument and
26711 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26713 * ipa-cp.c (create_specialized_node): Likewise.
26714 * symtab.c (symtab_node::maybe_create_reference): Handle
26715 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26717 2017-01-20 Martin Liska <mliska@suse.cz>
26719 * read-rtl-function.c (function_reader::create_function): Use
26720 build_decl instread of build_decl_stat.
26722 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26724 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26725 * config/i386/avx512dqintrin.h: Ditto.
26726 * config/i386/avx512fintrin.h: Ditto.
26727 * config/i386/i386-builtin-types.def: Add new types.
26728 * config/i386/i386.c: Handle new types.
26729 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26730 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26731 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26732 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26733 (__builtin_ia32_kshiftridi): New.
26734 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26736 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26740 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26741 define to rs6000_init_stack_protect_guard.
26742 (rs6000_init_stack_protect_guard): New function.
26744 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26745 Yunqiang Su <yunqiang.su@imgtec.com>
26747 * config.gcc (supported_defaults): Add madd4.
26748 (with_madd4): Add validation.
26749 (all_defaults): Add madd4.
26750 * config/mips/mips.opt (mmadd4): New option.
26751 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26753 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26755 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26756 (ISA_HAS_FUSED_MADD4): Likewise.
26757 * doc/invoke.texi (-mmadd4): Document the new option.
26758 * doc/install.texi (--with-madd4): Document the new option.
26760 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26762 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26763 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26764 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26765 (aarch64_init_pauth_hint_builtins): New.
26766 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26767 (aarch64_expand_builtin): Expand new builtins.
26769 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26771 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26772 * combine-stack-adj.c (no_unhandled_cfa): Handle
26773 REG_CFA_TOGGLE_RA_MANGLE.
26774 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26775 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26776 info for return address signing.
26777 (aarch64_expand_epilogue): Likewise.
26779 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26781 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26782 * config/aarch64/aarch64-protos.h
26783 (aarch64_return_address_signing_enabled): New declaration.
26784 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26786 (aarch64_expand_prologue): Sign return address before it's pushed onto
26788 (aarch64_expand_epilogue): Authenticate return address fetched from
26790 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26791 (aarch64_attributes): New function attributes for "sign-return-address".
26792 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26793 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26794 ("*do_return"): Generate combined instructions according to key index.
26795 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26796 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26798 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26799 * config/aarch64/aarch64.opt (msign-return-address=): New.
26800 * doc/extend.texi (AArch64 Function Attributes): Documents
26801 "sign-return-address=".
26802 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26804 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26806 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26807 overall option summary.
26809 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26811 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26812 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26813 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26814 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26816 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26818 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26819 -mpower9-minmax by default for -mcpu=power9.
26820 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26821 128-bit floating point.
26823 2017-01-20 Alan Modra <amodra@gmail.com>
26825 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26826 optimizing for size.
26828 2017-01-20 Alan Modra <amodra@gmail.com>
26831 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26832 for strcmp and strncmp from corresponding builtin decl.
26834 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26836 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26837 instead of i386/rtems-64.h.
26838 * config/i386/rtems-64.h: Remove.
26840 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26844 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26846 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26848 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26850 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26851 Change int to HOST_WIDE_INT.
26852 * config/aarch64/aarch64-protos.h
26853 (aarch64_simd_gen_const_vector_dup): Likewise.
26854 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26856 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26858 * langhooks-def.h (lhd_type_for_size): New decl.
26859 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26860 * langhooks.c (lhd_type_for_size): New function, taken from
26863 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26865 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26866 define_bypass for CR latency.
26867 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26868 (power9-alu2): Add define_bypass for CR latency.
26870 (power9-mul): Update insn latency.
26871 (power9-mul-compare): Update insn latency, bypass latency and remove
26874 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26876 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26878 * config/aarch64/aarch64.md
26879 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26880 aarch64_nopcrelative_literal_loads.
26881 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26883 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26885 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26886 TARGET_LOONGSON_3A.
26887 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26889 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26892 * config.gcc (supported_defaults): Add lxc1-sxc1.
26893 (with_lxc1_sxc1): Add validation.
26894 (all_defaults): Add lxc1-sxc1.
26895 * config/mips/mips.opt (mlxc1-sxc1): New option.
26896 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26898 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26899 __mips_no_lxc1_sxc1.
26900 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26901 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26902 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26904 2017-01-19 Richard Biener <rguenther@suse.de>
26906 PR tree-optimization/72488
26907 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26908 sure to restore SSA info.
26909 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26911 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26913 PR rtl-optimization/79121
26914 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26915 of the inner type when shifting an extended value.
26917 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26920 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26921 interposable aliases.
26923 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26926 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26927 Use the evmergelohi instruction.
26928 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26929 (mov_sitf_e500_subreg8_2_be): Likewise.
26930 (mov_sitf_e500_subreg12_2_le): Likewise.
26931 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26932 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26933 (mov_sitf_e500_subreg8_2_le): Likewise.
26934 (mov_sitf_e500_subreg12_2_be): Likewise.
26936 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26938 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26939 attribute from vecsimple to vecperm.
26940 (altivec_vbpermq2): Likewise.
26942 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26945 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26947 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26948 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26949 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26950 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26951 case where N arg is SIZE_MAX.
26952 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26953 (cmpstrsi): Add pattern.
26955 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26957 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26958 __builtin_vec_revb builtins.
26959 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26960 built-in functions to support generation of the ISA 3.0 XXBR<x>
26961 vector byte reverse instructions.
26962 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26963 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26964 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26965 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26966 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26967 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26968 (P9V_BUILTIN_VEC_REVB): Likewise.
26969 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26970 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26971 (p9_xxbrq_v16qi): Likewise.
26972 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26973 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26974 (p9_xxbrh_v8hi): Likewise.
26975 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26976 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26977 vec_revb built-in functions.
26979 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26981 PR rtl-optimization/78952
26982 * config/i386/i386.md (any_extract): New code iterator.
26983 (*insvqi_2): Use any_extract for source operand.
26984 (*insvqi_3): Use any_shiftrt for source operand.
26986 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26988 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26990 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26992 2017-01-18 Matthias Klose <doko@ubuntu.com>
26994 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26996 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26998 * config/rs6000/altivec.h (vec_bperm): Change #define.
26999 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27000 (altivec_vbpermq2): New define_insn.
27001 (altivec_vbpermd): Likewise.
27002 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27003 function interface.
27004 (VBPERMD): Likewise.
27005 (VBPERM): New polymorphic function interface.
27006 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27007 Add entries for P9V_BUILTIN_VEC_VBPERM.
27008 * doc/extend.texi: Add interfaces for vec_bperm.
27010 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27012 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27013 first letter of error messages.
27014 (s390_resolve_overloaded_builtin): Likewise.
27015 * config/s390/s390.c (s390_expand_builtin): Likewise.
27016 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27017 (s390_valid_target_attribute_inner_p): Likewise.
27018 * config/s390/s390.md ("tabort"): Likewise.
27020 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27022 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27023 (ISA_AVOID_DIV_HILO): New macro.
27024 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27025 (ISA_HAS_DDIV): Likewise.
27027 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27029 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27031 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27033 * doc/invoke.texi (fabi-version): Spelling fix.
27035 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27038 * doc/invoke.texi (fabi-version): Mention mangling fix for
27041 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27044 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27046 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27049 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27050 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27052 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27054 (stack_protect_test): Ditto.
27055 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27056 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27058 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27059 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27060 -mstack-protector-guard-offset=.
27061 (RS/6000 and PowerPC Options): Ditto.
27063 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27065 * config/i386/i386.h (MASK_CLASS_P): New define.
27066 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27067 there are no registers from different register sets also when
27068 mask registers are used. Update function comment.
27069 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27070 to (*k/*r) and (*k/*km) alternatives.
27072 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27074 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27075 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27076 (EH_RETURN_HANDLER_RTX): New define.
27077 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27078 Force frame pointer in EH return functions.
27079 (aarch64_expand_epilogue): Add barrier for eh_return.
27080 (aarch64_final_eh_return_addr): Remove.
27081 (aarch64_eh_return_handler_rtx): New function.
27082 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27084 (aarch64_eh_return_handler_rtx): New prototype.
27086 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27088 * config/rs6000/altivec.h (vec_rlmi): New #define.
27089 (vec_vrlnm): Likewise.
27090 (vec_rlnm): Likewise.
27091 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27092 (UNSPEC_VRLNM): Likewise.
27093 (VIlong): New mode iterator.
27094 (altivec_vrl<VI_char>mi): New define_insn.
27095 (altivec_vrl<VI_char>nm): Likewise.
27096 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27098 (VRLDNM): Likewise.
27099 (RLNM): New polymorphic function entry.
27100 (VRLWMI): New monomorphic function entry.
27101 (VRLDMI): Likewise.
27102 (RLMI): New polymorphic function entry.
27103 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27104 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27105 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27108 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27111 * dwarf2out.c (field_byte_offset): Restore the
27112 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27113 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27114 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27117 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27120 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27122 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27125 * dwarf2out.c (add_data_member_location_attribute): For constant
27126 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27127 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27128 DW_AT_byte_size attributes.
27130 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27132 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27133 after forcing to constant memory when the code model is medium.
27135 2017-01-17 Julia Koval <julia.koval@intel.com>
27138 * config/i386/avx512fintrin.h
27139 (_mm512_i32gather_ps): Change __addr type to void const*.
27140 (_mm512_mask_i32gather_ps): Ditto.
27141 (_mm512_i32gather_pd): Ditto.
27142 (_mm512_mask_i32gather_pd): Ditto.
27143 (_mm512_i64gather_ps): Ditto.
27144 (_mm512_mask_i64gather_ps): Ditto.
27145 (_mm512_i64gather_pd): Ditto.
27146 (_mm512_mask_i64gather_pd): Ditto.
27147 (_mm512_i32gather_epi32): Ditto.
27148 (_mm512_mask_i32gather_epi32): Ditto.
27149 (_mm512_i32gather_epi64): Ditto.
27150 (_mm512_mask_i32gather_epi64): Ditto.
27151 (_mm512_i64gather_epi32): Ditto.
27152 (_mm512_mask_i64gather_epi32): Ditto.
27153 (_mm512_i64gather_epi64): Ditto.
27154 (_mm512_mask_i64gather_epi64): Ditto.
27155 (_mm512_i32scatter_ps): Change __addr type to void*.
27156 (_mm512_mask_i32scatter_ps): Ditto.
27157 (_mm512_i32scatter_pd): Ditto.
27158 (_mm512_mask_i32scatter_pd): Ditto.
27159 (_mm512_i64scatter_ps): Ditto.
27160 (_mm512_mask_i64scatter_ps): Ditto.
27161 (_mm512_i64scatter_pd): Ditto.
27162 (_mm512_mask_i64scatter_pd): Ditto.
27163 (_mm512_i32scatter_epi32): Ditto.
27164 (_mm512_mask_i32scatter_epi32): Ditto.
27165 (_mm512_i32scatter_epi64): Ditto.
27166 (_mm512_mask_i32scatter_epi64): Ditto.
27167 (_mm512_i64scatter_epi32): Ditto.
27168 (_mm512_mask_i64scatter_epi32): Ditto.
27169 (_mm512_i64scatter_epi64): Ditto.
27170 (_mm512_mask_i64scatter_epi64): Ditto.
27171 * config/i386/avx512pfintrin.h
27172 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27173 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27174 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27175 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27176 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27177 (_mm512_prefetch_i32scatter_ps): Ditto.
27178 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27179 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27180 (_mm512_prefetch_i64scatter_pd): Ditto.
27181 (_mm512_prefetch_i64scatter_ps): Ditto.
27182 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27183 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27184 * config/i386/avx512vlintrin.h
27185 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27186 (_mm_mmask_i32gather_ps): Ditto.
27187 (_mm256_mmask_i32gather_pd): Ditto.
27188 (_mm_mmask_i32gather_pd): Ditto.
27189 (_mm256_mmask_i64gather_ps): Ditto.
27190 (_mm_mmask_i64gather_ps): Ditto.
27191 (_mm256_mmask_i64gather_pd): Ditto.
27192 (_mm_mmask_i64gather_pd): Ditto.
27193 (_mm256_mmask_i32gather_epi32): Ditto.
27194 (_mm_mmask_i32gather_epi32): Ditto.
27195 (_mm256_mmask_i32gather_epi64): Ditto.
27196 (_mm_mmask_i32gather_epi64): Ditto.
27197 (_mm256_mmask_i64gather_epi32): Ditto.
27198 (_mm_mmask_i64gather_epi32): Ditto.
27199 (_mm256_mmask_i64gather_epi64): Ditto.
27200 (_mm_mmask_i64gather_epi64): Ditto.
27201 (_mm256_i32scatter_ps): Change __addr type to void*.
27202 (_mm256_mask_i32scatter_ps): Ditto.
27203 (_mm_i32scatter_ps): Ditto.
27204 (_mm_mask_i32scatter_ps): Ditto.
27205 (_mm256_i32scatter_pd): Ditto.
27206 (_mm256_mask_i32scatter_pd): Ditto.
27207 (_mm_i32scatter_pd): Ditto.
27208 (_mm_mask_i32scatter_pd): Ditto.
27209 (_mm256_i64scatter_ps): Ditto.
27210 (_mm256_mask_i64scatter_ps): Ditto.
27211 (_mm_i64scatter_ps): Ditto.
27212 (_mm_mask_i64scatter_ps): Ditto.
27213 (_mm256_i64scatter_pd): Ditto.
27214 (_mm256_mask_i64scatter_pd): Ditto.
27215 (_mm_i64scatter_pd): Ditto.
27216 (_mm_mask_i64scatter_pd): Ditto.
27217 (_mm256_i32scatter_epi32): Ditto.
27218 (_mm256_mask_i32scatter_epi32): Ditto.
27219 (_mm_i32scatter_epi32): Ditto.
27220 (_mm_mask_i32scatter_epi32): Ditto.
27221 (_mm256_i32scatter_epi64): Ditto.
27222 (_mm256_mask_i32scatter_epi64): Ditto.
27223 (_mm_i32scatter_epi64): Ditto.
27224 (_mm_mask_i32scatter_epi64): Ditto.
27225 (_mm256_i64scatter_epi32): Ditto.
27226 (_mm256_mask_i64scatter_epi32): Ditto.
27227 (_mm_i64scatter_epi32): Ditto.
27228 (_mm_mask_i64scatter_epi32): Ditto.
27229 (_mm256_i64scatter_epi64): Ditto.
27230 (_mm256_mask_i64scatter_epi64): Ditto.
27231 (_mm_i64scatter_epi64): Ditto.
27232 (_mm_mask_i64scatter_epi64): Ditto.
27233 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27234 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27235 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27236 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27237 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27238 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27239 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27240 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27241 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27242 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27243 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27244 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27245 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27246 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27247 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27248 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27249 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27250 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27251 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27252 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27253 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27254 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27255 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27256 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27257 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27258 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27259 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27260 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27261 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27262 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27263 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27264 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27265 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27266 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27267 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27268 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27269 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27270 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27271 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27272 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27273 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27274 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27275 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27276 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27277 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27278 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27279 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27280 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27281 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27282 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27283 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27284 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27285 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27286 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27287 definitions accordingly.
27289 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27290 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27293 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27296 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27299 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27300 hard regs for inner regno.
27302 2017-01-17 Martin Liska <mliska@suse.cz>
27305 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27306 assumption and add comment.
27308 2017-01-17 Nathan Sidwell <nathan@acm.org>
27310 * ipa-visibility.c (localize_node): New function, broken out of ...
27311 (function_and_variable_visibility): ... here. Call it.
27313 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27315 PR middle-end/77445
27316 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27317 correctly set frequency of oudgoing edge.
27318 (duplicate_thread_path): Fix profile updating.
27320 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27323 * configure.ac: Add GCC_BASE_VER.
27324 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27325 version from BASE-VER file.
27326 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27327 (gcc.o): Depend on $(BASEVER).
27328 * common.opt (dumpfullversion): New option.
27329 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27330 * doc/invoke.texi: Document -dumpfullversion.
27331 * doc/install.texi: Document --with-gcc-major-version-only.
27332 * configure: Regenerated.
27334 2017-01-17 Richard Biener <rguenther@suse.de>
27336 PR tree-optimization/71433
27337 * tree-vrp.c (register_new_assert_for): Merge same asserts
27338 on all incoming edges.
27339 (process_assert_insertions_for): Handle insertions at the
27342 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27344 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27345 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27347 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27350 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27353 2017-01-17 Alan Modra <amodra@gmail.com>
27356 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27357 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27358 symbolic stack limit when pic.
27360 2017-01-16 Martin Sebor <msebor@redhat.com>
27362 PR tree-optimization/78608
27363 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27365 2017-01-16 Jeff Law <law@redhat.com>
27368 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27369 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27370 for several include directories that may be relative to sysroot.
27371 * config/i386/x-mingw32 (gplus_includedir): Define.
27372 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27373 (native_system_includedir): Likewise.
27374 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27375 override if TARGET_SYSTEM_ROOT is defined.
27376 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27378 PR tree-optimization/79090
27379 PR tree-optimization/33562
27380 PR tree-optimization/61912
27381 PR tree-optimization/77485
27382 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27383 and computed trims into the dump file.
27385 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27387 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27389 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27392 * gimplify.c (gimplify_init_constructor): If want_value and
27393 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27397 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27398 sequence. Formatting fixes.
27399 (doloop_optimize): Formatting fixes.
27402 * gcc.c (debug_level_greater_than_spec_func): New function.
27403 (static_spec_functions): Add debug-level-gt spec function.
27404 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27406 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27407 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27408 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27409 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27410 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27411 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27413 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27415 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27416 QImode fixups to general and mask registers only.
27418 2017-01-16 Carl Love <cel@us.ibm.com>
27420 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27421 for built-in functions
27422 vector signed char vec_nabs (vector signed char)
27423 vector signed short vec_nabs (vector signed short)
27424 vector signed int vec_nabs (vector signed int)
27425 vector signed long long vec_nabs (vector signed long long)
27426 vector float vec_nabs (vector float)
27427 vector double vec_nabs (vector double)
27428 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27430 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27431 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27432 * doc/extend.texi: Update the documentation file for the new built-in
27435 2017-01-16 Martin Sebor <msebor@redhat.com>
27437 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27440 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27442 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27443 UNSPEC_VSX__XXSPLTD to require special splat handling.
27445 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27448 * system.h: Poison strndup.
27450 2017-01-16 Alan Modra <amodra@gmail.com>
27453 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27456 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27458 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27460 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27462 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27463 call recog here. Assert that INSN_CODE (insn) is non-negative.
27465 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27468 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27470 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27471 in the currently scheduled RTL fragment.
27473 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27475 PR rtl-optimization/78751
27476 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27479 2017-01-14 Jeff Law <law@redhat.com>
27481 PR tree-optimization/79090
27482 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27483 variable length stores.
27484 (compute_trims): Delete dead assignment to *trim_tail.
27485 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27488 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27490 PR rtl-optimization/78626
27491 PR rtl-optimization/78727
27492 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27493 of a block, and split such blocks after everything else is finished.
27495 2017-01-14 Alan Modra <amodra@gmail.com>
27498 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27499 target legitimate_combined_insn.
27500 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27501 (rs6000_legitimate_combined_insn): New function.
27502 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27504 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27505 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27506 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27508 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27510 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27512 2017-01-13 Jeff Law <law@redhat.com>
27514 PR tree-optimization/33562
27515 PR tree-optimization/61912
27516 PR tree-optimization/77485
27517 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27519 (delete_dead_assignment): Likewise.
27520 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27521 statement to delete_dead_call and delete_dead_assignment.
27523 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27526 * substring-locations.c (format_warning_va): Strengthen case 1 so
27527 that both endpoints of the substring must be within the format
27528 range for just the substring to be printed.
27530 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27532 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27533 * config/i386/i386.c (ix86_target_string): Add missing options
27534 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27535 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27536 flags_other and ix86_target_other to flags2_other. Display unknown
27538 (ix86_valid_target_attribute_inner_p): Add missing options and
27539 reorder options by implied ISAs, as in ix86_target_string.
27541 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27543 * hash-table.h (hash_table::too_empty_p): New function.
27544 (hash_table::expand): Use it.
27545 (hash_table::traverse): Likewise.
27546 (hash_table::empty_slot): Use sizeof (value_type) instead of
27547 sizeof (PTR) to convert bytes to elements. Shrink the table
27548 if the current size is excessive for the current number of
27551 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27553 * ira-costs.c (record_reg_classes): Break from the inner loop
27554 early once alt_fail is known to be true. Update outer loop
27555 handling accordingly.
27557 2017-01-13 Jeff Law <law@redhat.com>
27559 * tree-ssa-dse.c (decrement_count): New function.
27560 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27561 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27562 when we know the partially dead statement is a mem* function.
27564 PR tree-optimization/61912
27565 PR tree-optimization/77485
27566 * tree-ssa-dse.c: Include expr.h.
27567 (maybe_trim_constructor_store): New function.
27568 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27570 PR tree-optimization/33562
27571 PR tree-optimization/61912
27572 PR tree-optimization/77485
27573 * doc/invoke.texi: Document new dse-max-object-size param.
27574 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27575 * tree-ssa-dse.c: Include params.h.
27576 (dse_store_status): New enum.
27577 (initialize_ao_ref_for_dse): New, partially extracted from
27579 (valid_ao_ref_for_dse, normalize_ref): New.
27580 (setup_live_bytes_from_ref, compute_trims): Likewise.
27581 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27582 (maybe_trim_partially_dead_store): Likewise.
27583 (maybe_trim_complex_store): Likewise.
27584 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27585 Track what bytes live from the original store. Return tri-state
27586 for dead, partially dead or live.
27587 (dse_dom_walker): Add constructor, destructor and new private members.
27588 (delete_dead_call, delete_dead_assignment): New extracted from
27590 (dse_optimize_stmt): Make a member of dse_dom_walker.
27591 Use initialize_ao_ref_for_dse.
27593 PR tree-optimization/33562
27594 PR tree-optimization/61912
27595 PR tree-optimization/77485
27596 * sbitmap.h (bitmap_count_bits): Prototype.
27597 (bitmap_clear_range, bitmap_set_range): Likewise.
27598 * sbitmap.c (bitmap_clear_range): New function.
27599 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27601 2017-01-13 Martin Liska <mliska@suse.cz>
27604 * function.c (set_cfun): Add new argument force.
27605 * function.h (set_cfun): Likewise.
27606 * ipa-inline-transform.c (inline_call): Use the function when
27607 strict alising from is dropped for function we inline to.
27609 2017-01-13 Richard Biener <rguenther@suse.de>
27611 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27612 for dumping GIMPLE INTEGER_CSTs.
27614 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27616 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27617 to 201112L since C++17.
27619 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27622 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27623 if -fsanitize=kernel-address is present.
27625 2017-01-13 Richard Biener <rguenther@suse.de>
27627 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27628 as _Literal ( type ) number in case usual suffixes do not
27629 preserve all information.
27631 2017-01-13 Richard Biener <rguenther@suse.de>
27633 PR tree-optimization/77283
27634 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27635 and ssa-iterators.h.
27636 (is_feasible_trace): Implement a cost model based on joiner
27639 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27642 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27643 char or short to __float128/_Float128 directly.
27645 2017-01-12 Martin Sebor <msebor@redhat.com>
27647 to -Wformat-overflow.
27648 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27649 (min_bytes_remaining): Same.
27650 (get_string_length): Same.
27651 (format_string): Same.
27652 (format_directive): Same.
27654 (pass_sprintf_length::handle_gimple_call): Same.
27656 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27658 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27659 info.nowrite calls with no lhs that can't throw. Return bool
27660 whether gsi_remove has been called or not.
27661 (pass_sprintf_length::handle_gimple_call): Return bool whether
27662 try_substitute_return_value called gsi_remove. Formatting fix.
27663 (pass_sprintf_length::execute): Don't use gsi_remove if
27664 handle_gimple_call returned true.
27667 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27668 be removed due to side-effects, don't remove following barrier nor
27669 turn the successor edge into fallthru edge.
27671 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27674 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27675 element-reversing loads and stores as not swappable.
27677 2017-01-12 Nathan Sidwell <nathan@acm.org>
27678 Nicolai Stange <nicstange@gmail.com>
27680 * combine.c (try_combine): Don't ignore result of overlap checking
27681 loop. Combine overlap & asm check into single loop.
27683 2017-01-12 Richard Biener <rguenther@suse.de>
27685 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27686 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27688 2017-01-12 Richard Biener <rguenther@suse.de>
27690 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27691 and TS_TARGET_OPTION directly derive from TS_BASE.
27692 * tree-core.h (tree_optimization_option): Derive from tree_base.
27693 (tree_target_option): Likewise.
27695 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27697 * config/i386/i386.c (memory_address_length): Increase len
27698 only when rip_relative_addr_p returns false.
27700 2017-01-11 Julia Koval <julia.koval@intel.com>
27702 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27703 (OPTION_MASK_ISA_SGX_SET): New.
27704 (ix86_handle_option): Handle OPT_msgx.
27705 * config.gcc: Added sgxintrin.h.
27706 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27707 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27708 * config/i386/i386.c (ix86_target_string): Add -msgx.
27710 (ix86_option_override_internal): Handle new options.
27711 (ix86_valid_target_attribute_inner_p): Add sgx.
27712 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27713 * config/i386/i386.opt: Add msgx.
27714 * config/i386/sgxintrin.h: New file.
27715 * config/i386/x86intrin.h: Add sgxintrin.h.
27717 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27720 * fold-const.c (maybe_nonzero_address): Return 1 for function
27722 (tree_single_nonzero_warnv_p): Don't handle function local objects
27726 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27729 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27732 * opts.c: Include "spellcheck.h"
27733 (struct string_fragment): New struct.
27734 (struct edit_distance_traits<const string_fragment &>): New
27736 (get_closest_sanitizer_option): New function.
27737 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27739 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27741 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27743 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27744 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27745 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27746 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27747 for initial die_offset if dwarf_split_debug_info.
27748 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27749 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27751 (output_skeleton_debug_sections): Formatting fix. Use
27752 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27753 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27755 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27757 * config/arm/cortex-a53.md: Add bypasses for
27758 cortex_a53_r2f_cvt.
27759 (cortex_a53_r2f): Only use for transfers.
27760 (cortex_a53_f2r): Likewise.
27761 (cortex_a53_r2f_cvt): Add reservation for conversions.
27762 (cortex_a53_f2r_cvt): Likewise.
27764 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27766 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27767 to all inlined functions, change static to extern.
27769 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27772 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27774 (arm_assemble_integer): Likewise.
27776 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27778 * config.gcc: Use new awk script to check CPU, FPU and architecture
27779 parameters for --with-... options.
27780 * config/arm/parsecpu.awk: New file
27781 * config/arm/arm-cpus.in: New file.
27782 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27784 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27786 * config/arm/t-arm: Update dependency rules.
27787 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27788 of processing .def files.
27789 * config/arm/genopt.sh: Deleted.
27790 * config/arm/gentune.sh: Deleted.
27791 * config/arm/arm-cores.def: Deleted.
27792 * config/arm/arm-arches.def: Deleted.
27793 * config/arm/arm-fpus.def: Deleted.
27794 * config/arm/arm-tune.md: Regenerated.
27795 * config/arm/arm-tables.opt: Regenerated.
27796 * config/arm/arm-cpu.h: New generated file.
27797 * config/arm/arm-cpu-data.h: New generated file.
27798 * config/arm/arm-cpu-cdata.h: New generated file.
27800 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27803 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27805 (input_varpool_node): Unpack dynamically_initialized bit.
27807 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27809 PR rtl-optimization/79032
27810 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27811 the alignment of the adjusted memory reference against that of MODE,
27812 instead of the alignment of the original memory reference.
27814 2017-01-11 Martin Jambor <mjambor@suse.cz>
27816 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27818 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27819 decorated functions.
27821 2017-01-11 Richard Biener <rguenther@suse.de>
27823 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27824 set range/nonnull info for PHI results. Do not set it on
27825 stmts marked for removal.
27827 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27829 * expr.c (store_field): In the bitfield case, fetch the return value
27830 from the registers before applying a single big-endian adjustment.
27831 Always do a final load for a BLKmode value not larger than a word.
27833 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27836 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27837 that we correctly handle column numbers greater than
27838 LINE_MAP_MAX_COLUMN_NUMBER.
27840 2017-01-10 Martin Sebor <msebor@redhat.com>
27842 PR middle-end/78245
27843 * gimple-ssa-sprintf.c (get_destination_size): Call
27844 {init,fini}object_sizes.
27845 * tree-object-size.c (addr_object_size): Adjust.
27846 (pass_through_call): Adjust.
27847 (pass_object_sizes::execute): Adjust.
27848 * tree-object-size.h (fini_object_sizes): Declare.
27850 2017-01-10 Martin Sebor <msebor@redhat.com>
27852 PR tree-optimization/78775
27853 * builtins.c (get_size_range): Move...
27854 * calls.c: ...to here.
27855 (alloc_max_size): Accept zero argument.
27856 (operand_signed_p): Remove.
27857 (maybe_warn_alloc_args_overflow): Call get_size_range.
27858 * calls.h (get_size_range): Declare.
27860 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27862 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27863 from TI's devices.csv file as of September 2016.
27864 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27866 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27868 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27869 * doc/invoke.texi: Likewise.
27870 * doc/md.texi: Likewise.
27871 * doc/objc.texi: Likewise.
27873 2017-01-10 Joshua Conner <joshconner@google.com>
27875 * config/arm/fuchsia-elf.h: New file.
27876 * config/fuchsia.h: New file.
27877 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27878 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27880 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27882 2016-01-10 Richard Biener <rguenther@suse.de>
27884 PR tree-optimization/79034
27885 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27886 Propagate out degenerate PHIs in the joiner.
27888 2017-01-10 Martin Liska <mliska@suse.cz>
27890 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27891 (sort_congruence_classes_by_decl_uid): Likewise.
27892 (sort_congruence_class_groups_by_decl_uid): Likewise.
27893 (sem_item_optimizer::merge_classes): Sort class, groups in these
27894 classes and members in the groups by DECL_UID of declarations.
27895 This would make merge operations stable.
27897 2017-01-10 Martin Liska <mliska@suse.cz>
27899 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27900 usage of m_classes_vec.
27901 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27902 (sem_item_optimizer::get_group_by_hash): Likewise.
27903 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27904 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27905 (sem_item_optimizer::verify_classes): Likewise.
27906 (sem_item_optimizer::process_cong_reduction): Likewise.
27907 (sem_item_optimizer::dump_cong_classes): Likewise.
27908 (sem_item_optimizer::merge_classes): Likewise.
27909 * ipa-icf.h (congruence_class_hash): Rename from
27910 congruence_class_group_hash. Remove declaration of m_classes_vec.
27912 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27914 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27915 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27916 * config.gcc: Add avx512vpopcntdqintrin.h.
27917 * config/i386/avx512vpopcntdqintrin.h: New.
27918 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27919 * config/i386/i386-builtin-types.def: Add new types.
27920 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27921 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27922 __builtin_ia32_vpopcountq_v8di_mask): New.
27923 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27924 __AVX512VPOPCNTDQ__.
27925 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27926 (PTA_AVX512VPOPCNTDQ): Define.
27927 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27928 TARGET_AVX512VPOPCNTDQ_P): Define.
27929 * config/i386/i386.opt: Add mavx512vpopcntdq.
27930 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27931 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27933 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27935 PR middle-end/77484
27936 * predict.def (PRED_CALL): Set to 67.
27938 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27940 * expr.c (store_field): In the bitfield case, if the value comes from
27941 a function call and is of an aggregate type returned in registers, do
27942 not modify the field mode; extract the value in all cases if the mode
27943 is BLKmode and the size is not larger than a word.
27945 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27948 * config/i386/cpuid.h: Fix undefined behavior.
27950 2017-01-04 Jeff Law <law@redhat.com>
27952 PR tree-optimization/79007
27953 PR tree-optimization/67955
27954 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27955 conservative for pt.null when flag_non_call_exceptions is on.
27957 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27959 PR translation/79019
27960 PR translation/79020
27961 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27962 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27963 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27965 * config/avr/avr.opt (maccumulate-args): Likewise.
27966 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27967 * common.opt (freport-bug): Likewise.
27968 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27969 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27970 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27971 translatable string.
27972 * config/i386/i386.c (function_value_32): Likewise.
27973 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27974 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27976 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27977 * common/config/msp430/msp430-common.c (msp430_handle_option):
27979 * symtab.c (symtab_node::verify_base): Likewise.
27980 * opts.c (set_debug_level): Likewise.
27981 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27982 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27983 missing whitespace to translatable strings.
27984 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27985 * config/sh/superh.h: Likewise.
27986 * config/i386/xopintrin.h: Likewise.
27987 * config/i386/znver1.md: Likewise.
27988 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27989 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27990 * double-int.h (struct double_int): Likewise.
27991 * double-int.c (div_and_round_double): Likewise.
27992 * wide-int.cc: Likewise.
27993 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27994 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27995 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27996 (crossjumps_occurred): ... this.
27997 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28000 PR tree-optimization/78899
28001 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28002 returning bool return struct loop *, NULL for failure and the new
28004 (versionable_outer_loop_p): Don't version outer loop if it has
28005 dont_vectorized bit set.
28006 (tree_if_conversion): When versioning outer loop, ensure
28007 tree_if_conversion is performed also on the inner loop of the
28008 non-vectorizable outer loop copy.
28009 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28010 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28011 prevent vectorization of it.
28012 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28013 the outer loop vectorization of the non-scalar version is attempted
28014 before vectorization of the inner loop in scalar version. If
28015 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28016 vectorization of its inner loop.
28017 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28018 has 2 inner loops, rename also on edges from bb whose single pred
28019 is outer_loop->header. Fix typo in function comment.
28021 2017-01-09 Martin Sebor <msebor@redhat.com>
28024 * asan.c (asan_emit_stack_protection): Increase local buffer size
28025 to avoid snprintf truncation warning.
28027 2017-01-09 Andrew Pinski <apinski@cavium.com>
28029 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28030 to reference thunderx2t99 for the tuning structure
28031 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28033 (thunderx2t99_extra_costs): This.
28034 * config/aarch64/aarch64-tune.md: Regenerate.
28035 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28036 (vulcan_addrcost_table): This.
28037 (vulcan_regmove_cost): Rename to ...
28038 (thunderx2t99_regmove_cost): This.
28039 (vulcan_vector_cost): Rename to ...
28040 (thunderx2t99_vector_cost): this.
28041 (vulcan_branch_cost): Rename to ...
28042 (thunderx2t99_branch_cost): This.
28043 (vulcan_tunings): Rename to ...
28044 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28045 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28047 2017-01-09 Martin Jambor <mjambor@suse.cz>
28051 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28052 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28053 (propagate_vr_accross_jump_function): Use the above function for all
28054 value range computations for pass-through jump functions and type
28055 converasion from explicit value range values.
28056 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28057 parameters from TYPE_ARG_TYPES.
28058 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28059 (ipa_write_node_info): Stream type of the actual argument.
28060 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28062 2017-01-09 Martin Liska <mliska@suse.cz>
28065 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28066 (lookup_compiler): Do not show error message with have_E.
28068 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28070 PR tree-optimization/78938
28071 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28072 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28073 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28074 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28077 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28079 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28082 2017-01-09 Richard Biener <rguenther@suse.de>
28084 PR tree-optimization/78997
28085 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28086 name condition properly.
28088 2017-01-09 Richard Biener <rguenther@suse.de>
28091 * dwarf2out.c (is_cxx): New overload with context.
28092 (is_naming_typedef_decl): Use it.
28094 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28096 * invoke.texi (Option Summary): Correct spacing in option lists
28097 and add line breaks to fix over-long lines.
28099 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28101 PR middle-end/17660
28103 * extend.texi (Common Variable Attributes): Add xref to GCC
28104 Internals manual to explain mode attribute keywords.
28106 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28109 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28110 and Preprocessor Options.
28111 (Options for Linking): Document -pthread here....
28112 (RS/6000 and PowerPC Options): ...not here.
28113 (Solaris 2 Options): ...or here.
28114 * doc/cppopts.texi: Document -pthread.
28116 2017-01-08 Martin Sebor <msebor@redhat.com>
28118 PR middle-end/77708
28119 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28120 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28121 New member functions.
28122 (format_directive): Used them.
28124 (pass_sprintf_length::handle_gimple_call): Same.
28125 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28126 to avoid truncation for any argument.
28127 (extract_affine_mul): Same.
28128 * tree.c (get_file_function_name): Same.
28130 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28132 PR middle-end/77484
28133 * predict.def (PRED_INDIR_CALL): Set to 86.
28135 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28137 PR preprocessor/54124
28138 * doc/cppopts.texi: Reformat -d subtable to list the full name
28139 of the options. Add cross-reference to the docs for the general
28140 compiler -d options.
28141 * doc/invoke.texi (Developer Options): Add cross-reference to the
28142 preprocessor-specific -d option documentation.
28144 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28146 PR preprocessor/13498
28147 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28148 redudant material, and reflect new command-line options.
28149 (System Headers): Likewise.
28151 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28153 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28154 -isystem, and -idirafter. Copy-edit.
28155 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28156 default for -ftrack-macro-expansion. Delete obsolete and
28157 badly-formatted implementation details about -fdebug-cpp output.
28158 * doc/cppwarnopts.texi: Copy-edit.
28160 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28163 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28164 that the transition from a max line width >= 1<<10 to narrower
28165 lines works correctly.
28167 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28169 * doc/options.texi (PerFunction): New.
28170 * opt-functions.awk (switch_flags): Map both Optimization and
28171 PerFunction to CL_OPTIMIZATION.
28172 * opth-gen.awk: Test for PerFunction flag along with
28174 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28175 it only when the latter is present. Skip those that don't in
28176 the hash function generator.
28177 * common.opt (fvar-tracking): Mark as PerFunction instead of
28179 (fvar-tracking-assignments): Likewise.
28180 (fvar-tracking-assignments-toggle): Likewise.
28181 (fvar-tracking-uninit): Likewise.
28183 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28185 PR translation/79018
28186 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28189 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28192 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28193 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28194 TARGET_LONG_JUMP_TABLE_OFFSETS.
28195 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28196 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28197 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28198 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28199 * config/m68k/m68k.md (tablejump expander): Likewise.
28200 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28201 TARGET_LONG_JUMP_TABLE_OFFSETS.
28202 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28203 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28205 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28206 David Holsgrove <david.holsgrove@xilinx.com>
28208 * common/config/microblaze/microblaze-common.c
28209 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28210 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28212 * config/microblaze/microblaze.c (microblaze_must_save_register)
28213 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28215 (microblaze_eh_return): New function.
28216 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28217 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28218 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28219 * config/microblaze/microblaze.md (eh_return): New pattern.
28221 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28223 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28224 GCC_DIAGNOSTIC_STRINGIFY): Define.
28226 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28228 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28230 * config/arm/arm.md (<mcrr>): New.
28232 * config/arm/arm.c (arm_arch5te): New.
28233 (arm_option_override): Set arm_arch5te.
28234 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28236 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28237 (arm_mcrr_qualifiers): ... this. New.
28238 (MRRC_QUALIFIERS): Define to...
28239 (arm_mrrc_qualifiers): ... this. New.
28240 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28242 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28243 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28244 (MRRCI, mrrc, MRRC): New.
28245 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28246 VUNSPEC_MRRC2): New.
28248 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28250 * config/arm/arm.md (<mcr>): New.
28252 * config/arm/arm.c (arm_coproc_builtin_available): Add
28253 support for mcr, mrc, mcr2 and mrc2.
28254 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28255 (arm_mcr_qualifiers): ... this. New.
28256 (MRC_QUALIFIERS): Define to ...
28257 (arm_mrc_qualifiers): ... this. New.
28258 (MCR_QUALIFIERS): Define to ...
28259 (arm_mcr_qualifiers): ... this. New.
28260 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28262 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28263 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28264 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28265 VUNSPEC_MRC2): New.
28267 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28269 * config/arm/arm.md (*ldc): New.
28273 * config/arm/arm.c (arm_coproc_builtin_available): Add
28274 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28275 (arm_coproc_ldc_stc_legitimate_address): New.
28276 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28277 'qualifier_const_pointer'.
28278 (LDC_QUALIFIERS): Define to...
28279 (arm_ldc_qualifiers): ... this. New.
28280 (STC_QUALIFIERS): Define to...
28281 (arm_stc_qualifiers): ... this. New.
28282 * config/arm/arm-protos.h
28283 (arm_coproc_ldc_stc_legitimate_address): New.
28284 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28285 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28286 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28287 stc2, stcl, stc2l): New.
28288 * config/arm/constraints.md (Uz): New.
28289 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28290 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28291 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28292 VUNSPEC_STC2L): New.
28294 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28296 * config/arm/arm.md (<cdp>): New.
28297 * config/arm/arm.c (neon_const_bounds): Rename this ...
28298 (arm_const_bounds): ... this.
28299 (arm_coproc_builtin_available): New.
28300 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28301 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28302 (CDP_QUALIFIERS): Define to...
28303 (arm_cdp_qualifiers): ... this. New.
28305 (arm_expand_builtin_args): Add case for 6 arguments.
28306 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28307 (arm_const_bounds): ... this.
28308 (arm_coproc_builtin_available): New.
28309 * config/arm/arm_acle.h (__arm_cdp): New.
28311 * config/arm/arm_acle_builtins.def (cdp): New.
28313 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28314 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28315 'arm_const_bounds'.
28316 * config/arm/types.md (coproc): New.
28317 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28318 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28319 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28320 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28322 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28324 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28325 (UBINOP_QUALIFIERS): New.
28327 (acle_builtin_data): New. Change comment.
28328 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28329 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28330 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28331 arm_acle_builtins.def.
28332 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28333 (arm_init_acle_builtins): New.
28334 (CRC32_BUILTIN): Remove.
28335 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28336 crc32cb, crc32ch and crc32cw.
28337 (arm_init_crc32_builtins): Remove.
28338 (arm_init_builtins): Use arm_init_acle_builtins rather
28339 than arm_init_crc32_builtins.
28340 (arm_expand_acle_builtin): New.
28341 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28342 * config/arm/arm_acle_builtins.def: New.
28344 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28346 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28347 (arm_builtin_datum): ... this.
28348 (arm_init_neon_builtin): Rename to ...
28349 (arm_init_builtin): ... this. Add a new parameters PREFIX
28350 and USE_SIG_IN_NAME.
28351 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28352 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28353 'arm_builtin_datum'.
28354 (arm_init_vfp_builtins): Likewise.
28355 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28356 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28357 (arm_expand_neon_args): Rename to ...
28358 (arm_expand_builtin_args): ... this. Rename builtin_arg
28359 enum values and differentiate between ARG_BUILTIN_MEMORY
28360 and ARG_BUILTIN_NEON_MEMORY.
28361 (arm_expand_neon_builtin_1): Rename to ...
28362 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28363 values, arm_expand_builtin_args and add bool parameter NEON.
28364 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28365 (arm_expand_vfp_builtin): Likewise.
28366 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28368 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28370 PR middle-end/77484
28371 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28372 * predict.c (tree_estimate_probability_bb): Reverse direction of
28373 polymorphic call predictor.
28375 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28377 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28378 (determine_pass_name_match): ...this new function and...
28379 (should_skip_pass_p): ...this new function.
28381 2017-01-06 Nathan Sidwell <nathan@acm.org>
28383 * ipa-visibility.c (function_and_variable_visibility): Reformat
28384 comments and long lines. Remove extrneous if.
28385 * symtab.c (symtab_node::make_decl_local): Fix code format.
28386 (symtab_node::set_section_for_node): Fix comment typo.
28388 2017-01-06 Martin Liska <mliska@suse.cz>
28391 * lra-constraints.c: Rename invariant to lra_invariant.
28392 * predict.c (set_even_probabilities): Initialize e to NULL.
28394 2017-01-05 Martin Sebor <msebor@redhat.com>
28396 PR tree-optimization/78910
28397 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28398 (format_integer): Correct off-by-one error in the handling
28399 of precision with negative numbers in signed conversions..
28401 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28403 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28405 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28407 PR tree-optimization/71016
28408 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28409 factor_out_conditional_conversion. Formatting fix.
28410 (factor_out_conditional_conversion): Add cond_stmt argument.
28411 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28412 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28415 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28417 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28418 read-rtl-function.o, and selftest-rtl.o.
28419 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28420 (selftest::aarch64_test_loading_full_dump): New function.
28421 (selftest::aarch64_run_selftests): New function.
28422 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28423 selftest::aarch64_run_selftests.
28424 * config/i386/i386.c
28425 (selftest::ix86_test_loading_dump_fragment_1): New function.
28426 (selftest::ix86_test_loading_call_insn): New function.
28427 (selftest::ix86_test_loading_full_dump): New function.
28428 (selftest::ix86_test_loading_unspec): New function.
28429 (selftest::ix86_run_selftests): Call the new functions.
28430 * emit-rtl.c (maybe_set_max_label_num): New function.
28431 * emit-rtl.h (maybe_set_max_label_num): New decl.
28432 * function.c (instantiate_decls): Guard call to
28433 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28434 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28436 * gensupport.c (gen_reader::gen_reader): Pass "false"
28437 for new "compact" param of rtx_reader.
28438 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28439 rather than an empty string for NULL strings.
28440 * read-md.c: Potentially include config.h rather than bconfig.h.
28441 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28442 (have_error): New global, copied from errors.c.
28443 (md_reader::read_name): Rename to...
28444 (md_reader::read_name_1): ...this, adding "out_loc" param,
28445 and converting "missing name or number" to returning false, rather
28447 (md_reader::read_name): Reimplement in terms of read_name_1.
28448 (md_reader::read_name_or_nil): New function.
28449 (md_reader::read_string): Handle "(nil)" by returning NULL.
28450 (md_reader::md_reader): Add new param "compact".
28451 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28452 (md_reader::read_file): New method.
28453 * read-md.h (md_reader::md_reader): Add new param "compact".
28454 (md_reader::read_file): New method.
28455 (md_reader::is_compact): New accessor.
28456 (md_reader::read_name): Convert return type from void to file_location.
28457 (md_reader::read_name_or_nil): New decl.
28458 (md_reader::read_name_1): New decl.
28459 (md_reader::m_compact): New field.
28460 (noop_reader::noop_reader): Pass "false" for new "compact" param
28462 (rtx_reader::rtx_reader): Add new "compact" param.
28463 (rtx_reader::read_rtx_operand): Make virtual and convert return
28464 type from void to rtx.
28465 (rtx_reader::read_until): New decl.
28466 (rtx_reader::handle_any_trailing_information): New virtual function.
28467 (rtx_reader::postprocess): New virtual function.
28468 (rtx_reader::finalize_string): New virtual function.
28469 (rtx_reader::m_in_call_function_usage): New field.
28470 (rtx_reader::m_reuse_rtx_by_id): New field.
28471 * read-rtl-function.c: New file.
28472 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28473 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28474 (selftest::verify_three_block_rtl_cfg): New decl.
28475 * read-rtl-function.h: New file.
28476 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28477 For host, include function.h, memmodel.h, and emit-rtl.h.
28478 (one_time_initialization): New function.
28479 (struct compact_insn_name): New struct.
28480 (compact_insn_names): New array.
28481 (find_code): Handle insn codes in compact dumps.
28482 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28483 (bind_subst_iter_and_attr): Likewise.
28484 (add_condition_to_string): Likewise.
28485 (add_condition_to_rtx): Likewise.
28486 (apply_attribute_uses): Likewise.
28487 (add_current_iterators): Likewise.
28488 (apply_iterators): Likewise.
28489 (initialize_iterators): Guard usage of apply_subst_iterator with
28490 #ifdef GENERATOR_FILE.
28491 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28492 (md_reader::read_mapping): Likewise.
28493 (add_define_attr_for_define_subst): Likewise.
28494 (add_define_subst_attr): Likewise.
28495 (read_subst_mapping): Likewise.
28496 (check_code_iterator): Likewise.
28497 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28499 (one_time_initialization): New function.
28500 (rtx_reader::read_until): New method.
28501 (read_flags): New function.
28502 (parse_reg_note_name): New function.
28503 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28504 Handle reuse_rtx ids.
28505 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28506 Add parsing support for RTL dumps, mirroring the special-cases in
28507 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28508 values, and calling handle_any_trailing_information.
28509 (rtx_reader::read_rtx_operand): Convert return type from void
28510 to rtx, returning return_rtx. Handle case 'e'. Call
28511 finalize_string on XSTR and XTMPL fields.
28512 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28513 "(nil)" values were omitted. Call the postprocess vfunc on the
28515 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28516 class ctor. Initialize m_in_call_function_usage. Call
28517 one_time_initialization.
28518 * rtl-tests.c (selftest::test_uncond_jump): Call
28519 set_new_first_and_last_insn.
28520 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28521 * selftest-rtl.c: New file.
28522 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28523 (selftest::get_insn_by_uid): New decl.
28524 * selftest-run-tests.c (selftest::run_tests): Call
28525 read_rtl_function_c_tests.
28526 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28527 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28530 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28532 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28533 operands in a special way. Assert that pos+len <= mode precision.
28535 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28537 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28538 3 argument Alias with unlimited for the negative form.
28539 (fno-vect-cost-model): Removed.
28541 2017-01-05 Martin Liska <mliska@suse.cz>
28543 * hsa-gen.c (gen_hsa_divmod): New function.
28544 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28546 2017-01-05 Martin Liska <mliska@suse.cz>
28549 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28552 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28554 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28555 small constant length operands.
28557 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28559 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28560 between loop iterations.
28562 2017-01-05 Martin Liska <mliska@suse.cz>
28565 * gimplify.c (gimplify_decl_expr): Compare to
28566 asan_poisoned_variables instread of checking flags.
28567 (gimplify_target_expr): Likewise.
28568 (gimplify_expr): Likewise.
28569 (gimplify_function_tree): Conditionally initialize
28570 asan_poisoned_variables.
28572 2017-01-04 Jeff Law <law@redhat.com>
28574 PR tree-optimizatin/78812
28575 * rtl.h (contains_mem_rtx_p): Prototype.
28576 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28577 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28578 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28579 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28580 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28582 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28584 * input.c (assert_char_at_range): Default-initialize actual_range.
28586 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28588 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28589 to match the caller.
28591 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28593 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28594 insns after final jump in test to emit dummy move.
28596 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28598 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28599 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28601 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28603 * multiple_target.c (create_dispatcher_calls): Init e_next.
28604 * tree-ssa-loop-split.c (split_loop): Init border.
28605 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28608 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28613 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28614 (altivec_register_operand): Do not return true if the operand
28615 contains a SUBREG mixing SImode and SFmode.
28616 (vsx_register_operand): Likewise.
28617 (vsx_reg_sfsubreg_ok): New predicate.
28618 (vfloat_operand): Do not return true if the operand contains a
28619 SUBREG mixing SImode and SFmode.
28620 (vint_operand): Likewise.
28621 (vlogical_operand): Likewise.
28622 (gpc_reg_operand): Likewise.
28623 (int_reg_operand): Likewise.
28624 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28625 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28626 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28628 (rs6000_emit_move_si_sf_subreg): New helper function.
28629 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28630 fixup SUBREGs involving SImode and SFmode.
28631 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28632 numbers for the new peephole2 optimization.
28633 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28634 the GLIBC math library that do AND/IOR/XOR operations on single
28635 precision floating point.
28636 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28637 target macros to say whether we need to avoid SUBREGs mixing
28639 (TARGET_ALLOW_SF_SUBREG): Likewise.
28640 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28641 (UNSPEC_SI_FROM_SF): Likewise.
28642 (iorxor): Change spacing.
28643 (and_ior_xor): New iterator for AND, IOR, and XOR.
28644 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28645 (movdi_from_sf_zero_ext): Likewise.
28646 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28647 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28648 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28649 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28650 (fms<mode>4): Likewise.
28651 (fnma<mode>4): Likewise.
28652 (fnms<mode>4): Likewise.
28653 (nfma<mode>4): Likewise.
28654 (nfms<mode>4): Likewise.
28656 2017-01-04 Marek Polacek <polacek@redhat.com>
28659 * doc/invoke.texi: Document -Wpointer-compare.
28661 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28663 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28666 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28667 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28670 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28673 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28674 documentation of the powerpc_popcntb_ok attribute.
28675 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28676 code to issue warning messages if a requested CPU configuration is
28677 not supported by the binary (assembler and loader) toolchain.
28678 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28679 made to define a built-in function that has been disabled.
28680 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28681 made to define a built-in function that has been disabled.
28682 (altivec_init_builtins): Add comment explaining why definition
28683 of the DST built-in functions is not preceded by an assertion
28684 check. Add assertions to prevent ICE if attempts are made to
28685 define an altivec predicate or an abs* built-in function that has
28687 (htm_init_builtins): Add comment explaining why definition of the
28688 htm built-in functions is not preceded by an assertion check.
28690 2017-01-04 Jeff Law <law@redhat.com>
28692 PR tree-optimizatin/67955
28693 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28694 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28695 the points-to solution does not include pt_null. Use DECL_PT_UID
28698 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28700 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28701 Use gen_int_mode instead of gen_lopwart for const_int operands.
28703 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28705 PR tree-optimization/71563
28706 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28707 out of range value - has low bits known to be zero.
28709 2017-01-04 Alan Modra <amodra@gmail.com>
28711 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28712 * configure: Regenerate.
28713 * config.in: Regenerate.
28715 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28718 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28719 a substring of the message, but strcmp with the whole message. Ifdef
28720 ENABLE_NLS, translate the message first using dgettext.
28722 2017-01-03 Jeff Law <law@redhat.com>
28724 PR tree-optimizatin/78856
28725 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28726 (mark_threaded_blocks): Remove code to truncate thread paths that
28727 cross multiple loop headers. Instead invalidate the cached loop
28728 iteration information and handle case of a thread path walking
28729 into an irreducible region.
28731 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28734 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28735 assertions. Add support for doing the signbit if the IEEE 128-bit
28736 floating point value is in a GPR.
28737 * config/rs6000/rs6000.md (Fsignbit): Delete.
28738 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28739 Update the length attribute if the value is in a GPR.
28740 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28741 the sign or zero extension instruction, since the value is always 0/1.
28742 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28745 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28746 extracting SImode to a GPR register so that we can generate a
28747 store, limit the vector to be in a traditional Altivec register
28748 for the vextuwrx instruction.
28750 2017-01-03 Ian Lance Taylor <iant@google.com>
28752 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28754 2017-01-03 Martin Sebor <msebor@redhat.com>
28756 PR tree-optimization/78696
28757 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28758 precision. Use MPFR for %f for greater fidelity. Correct handling
28760 (pass_sprintf_length::compute_format_length): Set width and precision
28761 specified by asrerisk to void_node for vararg functions.
28762 (try_substitute_return_value): Adjust dump output.
28764 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28766 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28768 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28770 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28771 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28772 -mlra/-mno-lra was passed to the compiler.
28774 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28776 PR rtl-optimization/65618
28777 * emit-rtl.c (try_split): Move initialization of "before" and
28778 "after" to just before the call to emit_insn_after_setloc.
28780 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28782 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28784 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28786 * dwarf2out.c (gen_enumeration_type_die): When
28787 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28789 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28791 PR tree-optimization/78965
28792 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28793 Change first argument from const call_info & to call_info &. For %n
28794 set info.nowrite to false.
28796 PR middle-end/78901
28797 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28798 possibly throwing calls.
28800 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28801 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28802 and fns handling, rather than in a separate case SSA_NAME.
28804 2017-01-02 Jeff Law <law@redhat.com>
28806 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28807 fixes for first_period and second_period variables.
28809 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28812 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28813 (*insvqi_1): New insn pattern.
28814 (*insvqi_1_mem_rex64): Ditto.
28815 (*insvqi_2): Ditto.
28816 (*insvqi_3): Rename from *insvqi.
28818 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28820 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28822 * doc/cfg.texi (Edges): Remove reference to Java.
28823 (Maintaining the CFG): Ditto.
28825 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28827 PR middle-end/77674
28828 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28829 transparent aliases.
28831 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28833 PR middle-end/77484
28834 * predict.def (PRED_CALL): Update hitrate.
28835 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28836 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28837 into direct/indirect/polymorphic variants.
28839 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28841 Update copyright years.
28843 * gcc.c (process_command): Update copyright notice dates.
28844 * gcov-dump.c (print_version): Ditto.
28845 * gcov.c (print_version): Ditto.
28846 * gcov-tool.c (print_version): Ditto.
28847 * gengtype.c (create_file): Ditto.
28848 * doc/cpp.texi: Bump @copying's copyright year.
28849 * doc/cppinternals.texi: Ditto.
28850 * doc/gcc.texi: Ditto.
28851 * doc/gccint.texi: Ditto.
28852 * doc/gcov.texi: Ditto.
28853 * doc/install.texi: Ditto.
28854 * doc/invoke.texi: Ditto.
28856 Copyright (C) 2017 Free Software Foundation, Inc.
28858 Copying and distribution of this file, with or without modification,
28859 are permitted in any medium without royalty provided the copyright
28860 notice and this notice are preserved.