1 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
4 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6 pentium4_cost, nocona_cost): Set reassociation width to 1.
7 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8 width to 2 for fp operations and 1 otherwise.
9 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11 (atom_cost): Set reassociation width to 2.
12 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
13 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
14 (core_cost): Set fp reassociation width to 4 and vector to 2.
15 (ix86_reassociation_width): Rewrite using cost table; special case
16 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
17 and TARGET_AVX128_OPTIMAL.
18 * i386.h (processor_costs): Add
19 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
20 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
21 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
22 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
23 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
24 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
26 2017-10-05 Nathan Sidwell <nathan@acm.org>
28 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
30 2017-10-05 Tamar Christina <tamar.christina@arm.com>
32 * config/arm/arm.c (arm_test_fpu_data): New.
33 (arm_run_selftests): Call arm_test_fpu_data.
35 2017-10-04 Nathan Sidwell <nathan@acm.org>
37 * toplev.c (toplev::main): Remove excess parens on pretty_printer
39 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
41 2017-10-04 Sudakshina Das <sudi.das@arm.com>
43 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
44 check type for aarch64_simd_valid_immediate.
45 (aarch64_output_simd_mov_immediate): Update prototype.
46 (aarch64_simd_valid_immediate): Update prototype.
47 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
48 support for ORR-immediate.
49 (and<mode>3): modified pattern to add support for BIC-immediate.
50 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
51 now checks for valid immediate for BIC and ORR based on new enum
53 (aarch64_output_simd_mov_immediate): Function now used to output
54 BIC/ORR imm as well based on new enum argument.
55 * config/aarch64/constraints.md (Do): New vector immediate constraint.
57 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
58 (aarch64_reg_or_bic_imm): Likewise.
60 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
62 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
63 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
65 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
69 PR rtl-optimization/82396
70 * haifa-sched.c (autopref_multipass_init): Simplify
72 (autopref_rank_data): Simplify sort order.
73 * sched-int.h (autopref_multipass_data_): Remove
74 multi_mem_insn_p, min_offset and max_offset.
76 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
78 * doc/sourcebuild.texi: Document vect_peeling_profitable.
80 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
82 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
85 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
87 * doc/sourcebuild.texi: Document vect_long_mult.
89 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
91 PR tree-optimization/82413
92 * fold-const.c (build_range_check): Use widest_int when comparing
93 the maximum ETYPE value with HIGH.
95 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
97 PR rtl-optimization/82396
98 * haifa-sched.c (autopref_multipass_init): Simplify
100 (autopref_rank_data): Simplify sort order.
101 * sched-int.h (autopref_multipass_data_): Remove
102 multi_mem_insn_p, min_offset and max_offset.
104 2017-10-04 Jakub Jelinek <jakub@redhat.com>
106 PR tree-optimization/82381
107 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
108 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
111 PR tree-optimization/82374
112 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
113 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
114 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
115 current_function_decl to the new decl.
117 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
119 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
120 helper macro for IEEE float128 hardware built-in functions.
121 (SQRTF128_ODD): Add built-in functions with the round-to-odd
123 (TRUNCF128_ODD): Likewise.
124 (ADDF128_ODD): Likewise.
125 (SUBF128_ODD): Likewise.
126 (MULF128_ODD): Likewise.
127 (DIVF128_ODD): Likewise.
128 (FMAF128_ODD): Likewise.
129 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
130 UNSPEC_TRUNC_ROUND_TO_ODD.
131 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
132 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
133 floating point round to odd instructions.
134 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
135 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
136 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
137 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
138 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
139 (trunc<mode>sf2_hw): Change the truncate with round to odd
140 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
141 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
142 to odd hardware instructions.
143 (sub<mode>3_odd): Likewise.
144 (mul<mode>3_odd): Likewise.
145 (div<mode>3_odd): Likewise.
146 (sqrt<mode>2_odd): Likewise.
147 (fma<mode>4_odd): Likewise.
148 (fms<mode>4_odd): Likewise.
149 (nfma<mode>4_odd): Likewise.
150 (nfms<mode>4_odd): Likewise.
151 (trunc<mode>df2_odd): Change the truncate with round to odd
152 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
154 * doc/extend.texi (PowerPC built-in functions): Update documentation
155 for existing IEEE float128-bit built-in functions. Add built-in
156 functions that generate the IEEE 128-bit floating point round to
159 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
161 PR rtl-optimization/77729
162 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
163 to (X&(C1&~C2))|C2 transformations.
165 2017-10-03 Martin Jambor <mjambor@suse.cz>
167 PR tree-optimization/82363
168 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
169 mismatch, mark lacc written regardless of racc.
171 2017-10-03 Jakub Jelinek <jakub@redhat.com>
173 PR tree-optimization/82381
174 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
175 stmt_to_insert nor wheather SSA_NAMEs are default defs.
176 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
177 fallthrough into reassoc_stmt_dominates_stmt_p.
180 * combine.c (combine_instructions): Don't combine in unreachable
183 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
186 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
187 function to not use the have_cpu variable. Do not set cpu_index,
188 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
190 (rs6000_valid_attribute_p): Remove duplicate initializations of
191 old_optimize and func_optimize.
192 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
193 (rs6000_activate_target_options): Make global.
194 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
197 2017-10-02 Jakub Jelinek <jakub@redhat.com>
199 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
200 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
201 Set *poffset to 0 and *psize and *pmax_size to -1 if
202 *poffset + *psize overflows in HOST_WIDE_INT.
204 PR tree-optimization/82387
205 PR tree-optimization/82388
206 PR tree-optimization/82389
207 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
208 instead of live_bytes non-NULL.
210 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
213 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
216 2017-10-02 Richard Biener <rguenther@suse.de>
218 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
219 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
220 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
222 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
224 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
225 requested precision matches the type's.
226 * calls.c (alloc_max_size): Calculate the new candidate size as
227 a widest_int and use wi::to_widest when comparing it with the
228 current candidate size.
229 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
230 zero rather than integer_zero_node.
231 * match.pd: Check for a no-op conversion before using wi::add
232 rather than after. Use tree_to_uhwi when summing small shift
233 counts into an unsigned int.
235 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
236 Alan Hayward <alan.hayward@arm.com>
237 David Sherwood <david.sherwood@arm.com>
240 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
241 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
242 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
245 2017-10-02 Richard Biener <rguenther@suse.de>
247 PR tree-optimization/82355
248 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
249 a mapping for the enclosing loop but avoid generating one for
251 (copy_bb_and_scalar_dependences): Remove premature codegen
252 error on PHIs in blocks duplicated into multiple places.
253 * graphite-scop-detection.c
254 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
255 in the region use it as loop and nest to analyze the DR in.
256 (try_generate_gimple_bb): Likewise.
257 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
258 (add_loop_constraints): For blocks in a loop not in the region
259 create a dimension with a single iteration.
260 * sese.h (gbb_loop_at_index): Remove assert.
262 2017-10-01 Kevin Buettner <kevinb@redhat.com>
264 * omp-expand.c (adjust_context_scope): New function.
265 (expand_parallel_call): Call adjust_context_scope.
267 2017-10-01 Jeff Law <law@redhat.com>
269 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
270 dom_opt_dom_walker class with direct access to private members.
271 Add comments. Call test_for_singularity.
272 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
273 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
274 m_dummy_cond anymore.
275 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
277 (pass_dominator:execute): Build the dummy_cond here and pass it
278 to the dom_opt_dom_walker ctor.
279 (test_for_singularity): New function.
281 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
282 Maya Rashish <coypu@sdf.org>
284 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
285 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
286 (alpha*-*-netbsd*) Use nbsd_tm_file.
287 (arm*-*-netbsdelf*) Likewise.
288 (i[34567]86-*-netbsdelf*) Likewise.
289 (x86_64-*-netbsd*) Likewise.
290 (mips*-*-netbsd*) Likewise.
291 (powerpc-*-netbsd*) Likewise.
292 (sh*-*-netbsd*) Likewise.
293 (sparc-*-netbsdelf*) Likewise.
294 (sparc64-*-netbsd*) Likewise.
295 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
297 (vax-*-netbsdelf*) Likewise.
298 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
299 (UINT_FAST8_TYPE) Likewise.
300 (INT_FAST16_TYPE) Check CHAR_FAST16.
301 (UINT_FAST16_TYPE) Likewise.
303 2017-09-30 Jakub Jelinek <jakub@redhat.com>
306 * config/i386/i386.md
307 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
308 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
309 *divmodsi4_zext_2): New define_insn_and_split.
310 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
311 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
312 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
313 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
314 New define_insn_and_split.
315 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
316 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
317 operands[1] having DImode when mode is SImode.
319 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
320 always SImode for DIV and MOD in REG_EQUAL notes.
322 2017-09-29 Yury Gribov <tetra2005@gmail.com>
325 * match.pd: Fix handling of NaNs in pattern.
327 2017-09-29 Jeff Law <law@redhat.com>
329 * sbitmap.c (bitmap_bit_in_range_p): New function.
330 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
331 * tree-ssa-dse.c (live_bytes_read): New function.
332 (dse_classify_store): Ignore reads of dead bytes.
334 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
335 typos and whitespace errors.
336 * config/i386/predicates.md (address_no_seg_operand): Likewise.
337 * config/s390/s390.c (s390_emit_prologue): Likewise.
339 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
342 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
343 with a symbol for LRA.
345 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
347 PR rtl-optimization/82338
348 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
350 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
352 * genmodes.c (calc_wider_mode): Suppress qsort macro.
353 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
354 (qsort_chk): Declare.
355 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
356 (qsort_chk): New function.
358 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
360 PR tree-optimization/82337
361 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
362 phi definition if the PHI result appears in an abnormal PHI.
363 (find_basis_for_base_expr): Don't record a basis if the LHS of the
364 basis appears in an abnormal PHI.
366 2017-09-29 Richard Biener <rguenther@suse.de>
368 * graphite-isl-ast-to-gimple.c
369 (translate_isl_ast_to_gimple::set_codegen_error): New function.
370 (binary_op_to_tree): Use it.
371 (get_rename_from_scev): Likewise.
372 (copy_loop_phi_nodes): Likewise.
373 (copy_bb_and_scalar_dependences): Likewise.
374 (translate_pending_phi_nodes): Likewise.
376 2017-09-29 Jakub Jelinek <jakub@redhat.com>
379 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
380 for movabsq $(i32 << shift), r64.
382 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
384 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
385 index when encoding %esp as %rsp to avoid 0x67 prefix.
387 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
389 * config/i386/i386.md (*movsf_internal, *movdf_internal):
390 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
392 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
394 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
395 Extensions with more than 16 double VFP registers.
396 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
397 to_clear_mask and all code related to it. Replace the remaining
398 entry by a sbitmap and adapt code accordingly.
400 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
402 * brig-builtins.def: Change pure attributes to const.
404 2017-09-28 Joseph Myers <joseph@codesourcery.com>
406 * config.gcc (default_gnu_indirect_function): Default to yes for
407 sparc*-*-linux* with glibc.
409 2017-09-28 Joseph Myers <joseph@codesourcery.com>
411 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
412 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
413 when creating .init_array and .fini_array sections with priority
416 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
419 * config/aarch64/aarch64.c
420 (aarch64_builtin_support_vector_misalignment): Always return false
421 when misalignment is unknown.
423 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
425 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
426 this function to return false if the definition used by the swap
427 instruction is artificial, or if the memory address from which the
428 constant value is loaded is not represented by a base address held
429 in a register or if the base address register is a frame or stack
430 pointer. Additionally, return false if the base address of the
431 loaded constant is a SYMBOL_REF but is not considered to be a
433 (replace_swapped_load_constant): New function.
434 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
435 loaded constant vector with a load of a swapped constant vector.
437 2017-09-27 Carl Love <cel@us.ibm.com>
439 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
440 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
441 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
444 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
446 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
447 first, always call autopref_rank_data otherwise.
449 2017-09-27 Richard Biener <rguenther@suse.de>
451 * graphite-scop-detection.c (find_scop_parameters): Move
452 loop bound handling ...
453 (gather_bbs::before_dom_children): ... here, avoiding the need
454 to build scop_info->loop_nest.
455 (record_loop_in_sese): Remove.
456 * sese.h (sese_info_t::loop_nest): Remove.
457 * sese.c (new_sese_info): Do not allocate loop_nest.
458 (free_sese_info): Do not free loop_nest.
460 2017-09-27 Jakub Jelinek <jakub@redhat.com>
463 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
464 lhs from calls if the lhs has addressable type.
466 2017-09-27 Richard Biener <rguenther@suse.de>
468 * graphite.h (scop::max_alias_set): New member.
469 * graphite-scop-detection.c: Remove references to non-existing
471 (build_alias_sets): Record the maximum alias set used for drs.
472 (build_scops): Support zero as unlimited for
473 --param graphite-max-arrays-per-scop.
474 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
476 (build_poly_sr_1): ... here. Compute alias set based on the
477 maximum alias set used for drs rather than
478 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
480 2017-09-27 Richard Biener <rguenther@suse.de>
482 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
483 --param loop-block-tile-size=0 to disable tiling.
485 2017-09-27 Richard Biener <rguenther@suse.de>
487 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
488 (graphite-max-nb-scop-params): Document special value zero.
489 * domwalk.h (dom_walker::STOP): New symbolical constant.
490 (dom_walker::dom_walker): Add optional parameter for bb to
492 (dom_walker::~dom_walker): Declare.
493 (dom_walker::before_dom_children): Document STOP return value.
494 (dom_walker::m_user_bb_to_rpo): New member.
495 (dom_walker::m_bb_to_rpo): Likewise.
496 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
497 mapping here if not provided by the user.
498 (dom_walker::~dom_walker): Free bb to RPO mapping if not
499 provided by the user.
500 (dom_walker::STOP): Define.
501 (dom_walker::walk): Do not compute bb to RPO mapping here.
502 Support STOP return value from before_dom_children to stop
504 * graphite-optimize-isl.c (optimize_isl): If the schedule
505 is the same still generate code if -fgraphite-identity
506 or -floop-parallelize-all are given.
507 * graphite-scop-detection.c: Include cfganal.h.
508 (gather_bbs::gather_bbs): Get and pass through bb to RPO
510 (gather_bbs::before_dom_children): Return STOP for BBs
512 (build_scops): Compute bb to RPO mapping and pass it to
513 the domwalk. Treat --param graphite-max-nb-scop-params=0
514 as not limiting the number of params.
515 * graphite.c (graphite_initialize): Remove limit on the
516 number of basic-blocks in a function.
517 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
518 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
521 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
523 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
524 Adjust code to eliminate needing to do the shift right 32-bits
525 operation after XSCVDPSPN.
527 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
529 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
530 ((X / Y) != 0 -> X >= Y): Likewise.
532 2017-09-26 Carl Love <cel@us.ibm.com>
534 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
535 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
536 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
537 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
538 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
539 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
540 definitions and overloading.
541 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
542 statement for P9V_BUILTIN_XST_LEN_R.
543 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
544 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
545 define_expand and define_insn for the instructions and builtins.
546 * doc/extend.texi: Update the built-in documentation file for the new
548 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
549 define_insn for the instructions
551 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
554 * gcc/config/netbsd-protos.h: New file.
555 * gcc/config/netbsd.c: New file.
556 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
557 * gcc/config/t-netbsd: New file.
558 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
559 (tmake_file) Add t-netbsd.
560 (extra_objs) Add netbsd.o.
562 2017-09-26 Janus Weil <janus@gcc.gnu.org>
566 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
568 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
570 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
571 sign extension from a vector register to a GPR by doing a 32-bit
572 direct move and then an EXTSW.
573 (extendsi<mode>2 splitter): Likewise.
574 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
575 right or vector extract after doing XSCVDPSPN. Use
576 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
578 (movdi_from_sf_zero_ext): Likewise.
579 (reload_gpr_from_vsxsf): Likewise.
580 (p8_mfvsrd_4_disf): Delete, no longer used.
581 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
582 then needing to move the SFmode to a GPR to use the XSCVDPSP
583 instruction instead of FRSP and XSCVDPSPN.
584 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
585 it is adjacent to the other XSCVSPDP insns.
586 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
587 SFmode to be in traditional Altivec registers.
588 (vsx_xscvdpspn): Eliminate useless alternative constraint.
589 (vsx_xscvspdpn): Likewise.
590 (vsx_xscvspdpn_scalar): Likewise.
592 2017-09-26 Martin Jambor <mjambor@suse.cz>
594 * tree-sra.c (compare_access_positions): Put integral types first,
595 stabilize sorting of integral types, remove conditions putting
596 non-full-precision integers last.
597 (sort_and_splice_var_accesses): Disable scalarization if a
598 non-integert would be represented by a non-full-precision integer.
600 2017-09-26 Joseph Myers <joseph@codesourcery.com>
602 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
603 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
604 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
605 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
606 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
607 conditionals inside the function instead of around it. Call
608 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
609 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
611 2017-09-26 Richard Biener <rguenther@suse.de>
613 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
615 (scop_detection::build_scop_breadth): ... this. Removed.
616 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
617 (scop_detection::harmful_stmt_in_bb): Likewise.
618 (scop_detection::graphite_can_represent_stmt): Likewise.
619 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
620 (scop_detection::can_represent_loop): Remove recursion, fold in ...
621 (scop_detection::can_represent_loop_1): ... this. Removed.
622 (scop_detection::harmful_loop_in_region): Simplify after inlining
623 the above and remove more quadraticness.
624 (build_scops): Adjust.
625 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
628 2017-09-26 Jakub Jelinek <jakub@redhat.com>
631 * config/i386/i386.c (ix86_print_operand_address_as): Only test
632 REGNO (base) == SP_REG if base is a REG.
635 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
636 if it is different SSA_NAME.
637 (optimize_range_tests_cmp_bitwise): New function.
638 (optimize_range_tests): Call it.
640 2017-09-26 Richard Biener <rguenther@suse.de>
642 PR tree-optimization/82321
643 * graphite.c (canonicalize_loop_closed_ssa): Properly check
644 for the def being inside the loop.
646 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
648 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
650 * config/s390/s390-builtins.def: Fix constraint on op4.
652 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
654 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
655 independent expanders.
656 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
657 ("vec_ordered", "vec_unordered"): New expanders.
659 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
661 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
664 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
666 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
667 vec_unpacks_lo_v16qi.
668 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
670 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
672 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
673 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
674 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
676 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
678 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
680 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
682 ("*vec_slb<mode>"): New insn pattern.
683 ("vec_shr_<mode>"): New expander.
684 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
685 and force the shift count operand to V16QImode.
686 ("vec_srb<mode>"): Set shift count mode to V16QI.
688 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
690 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
691 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
692 ("vec_widen_smult_hi_<mode>"): New expander definitions.
694 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
697 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
699 2017-09-26 Richard Biener <rguenther@suse.de>
701 PR tree-optimization/82320
702 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
705 2017-09-25 Jeff Law <law@redhat.com>
707 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
708 prototype for new argument.
709 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
710 mostly extracted from rs6000_emit_allocate_stack.
711 (rs6000_emit_probe_stack_range_stack_clash): New function.
712 (rs6000_emit_allocate_stack): Call
713 rs6000_emit_probe_stack_range_stack_clash as needed.
714 (rs6000_emit_probe_stack_range): Add additional argument
715 to call to gen_probe_stack_range{si,di}.
716 (output_probe_stack_range): New.
717 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
718 (output_probe_stack_range_stack_clash): New.
719 (rs6000_emit_prologue): Emit notes into dump file as requested.
720 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
721 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
722 Add additional operand and pass it to output_probe_stack_range.
724 2017-09-25 Bin Cheng <bin.cheng@arm.com>
726 PR tree-optimization/82163
727 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
728 (checking_verify_loop_closed_ssa): New parameter.
729 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
730 (check_loop_closed_ssa_stmt): Delete.
731 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
732 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
733 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
736 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
738 * brig-builtins.def: Treat HSAIL barrier builtins as
739 setjmp/longjump style functions.
741 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
743 * target.def (constant_alignment): New hook.
744 * defaults.h (CONSTANT_ALIGNMENT): Delete.
745 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
746 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
747 * doc/tm.texi: Regenerate.
748 * targhooks.h (default_constant_alignment): Declare.
749 (constant_alignment_word_strings): Likewise.
750 * targhooks.c (default_constant_alignment): New function.
751 (constant_alignment_word_strings): Likewise.
752 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
753 instead of CONSTANT_ALIGNMENT.
754 * varasm.c (align_variable, get_variable_align, build_constant_desc)
755 (force_const_mem): Likewise.
756 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
757 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
758 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
759 (TARGET_CONSTANT_ALIGNMENT): Redefine.
760 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
762 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
763 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
764 constant_alignment_word_strings.
765 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
766 (CONSTANT_ALIGNMENT): Likewise.
767 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
768 (arm_constant_alignment): New function.
769 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
770 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
771 constant_alignment_word_strings.
772 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
773 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
774 constant_alignment_word_strings.
775 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
776 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
777 (cris_constant_alignment): New function.
778 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
779 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
780 (epiphany_constant_alignment): New function.
781 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
782 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
783 constant_alignment_word_strings.
784 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
785 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
786 constant_alignment_word_strings.
787 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
788 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
789 constant_alignment_word_strings.
790 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
791 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
792 * config/i386/i386.c (ix86_constant_alignment): Make static.
793 Use the same interface as the target hook.
794 (TARGET_CONSTANT_ALIGNMENT): Redefine.
795 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
796 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
797 constant_alignment_word_strings.
798 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
799 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
800 (TARGET_CONSTANT_ALIGNMENT): Redefine.
801 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
802 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
803 constant_alignment_word_strings.
804 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
805 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
806 constant_alignment_word_strings.
807 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
808 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
809 constant_alignment_word_strings.
810 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
811 * config/microblaze/microblaze.c (microblaze_constant_alignment):
813 (TARGET_CONSTANT_ALIGNMENT): Redefine.
814 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
815 * config/mips/mips.c (mips_constant_alignment): New function.
816 (TARGET_CONSTANT_ALIGNMENT): Redefine.
817 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
818 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
819 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
820 (mmix_constant_alignment): Make static. Use the same interface
822 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
823 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
824 constant_alignment_word_strings.
825 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
826 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
827 constant_alignment_word_strings.
828 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
829 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
830 constant_alignment_word_strings.
831 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
832 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
833 (rs6000_constant_alignment): New function.
834 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
835 * config/riscv/riscv.c (riscv_constant_alignment): New function.
836 (TARGET_CONSTANT_ALIGNMENT): Redefine.
837 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
838 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
839 (rs6000_constant_alignment): New function.
840 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
841 * config/s390/s390.c (s390_constant_alignment): New function.
842 (TARGET_CONSTANT_ALIGNMENT): Redefine.
843 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
844 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
845 constant_alignment_word_strings.
846 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
847 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
848 (sparc_constant_alignment): New function.
849 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
850 * config/spu/spu.c (spu_constant_alignment): New function.
851 (TARGET_CONSTANT_ALIGNMENT): Redefine.
852 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
853 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
854 constant_alignment_word_strings.
855 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
856 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
857 constant_alignment_word_strings.
858 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
859 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
860 constant_alignment_word_strings.
861 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
862 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
863 (visium_constant_alignment): New function.
864 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
865 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
866 (xtensa_constant_alignment): New function.
867 * system.h (CONSTANT_ALIGNMENT): Poison.
869 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
872 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
873 (rs6000_builtin_valid_without_lhs): New helper function.
874 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
875 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
877 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
879 * target.h (vec_perm_indices): Use unsigned short rather than
881 (auto_vec_perm_indices): Likewise.
882 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
883 Use unsigned int rather than unsigned char.
884 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
886 2017-09-25 Richard Biener <rguenther@suse.de>
888 * cfgloop.h (sort_sibling_loops): Declare.
889 * cfgloop.c (sort_sibling_loops_cmp): New helper.
890 (sort_sibling_loops): New function sorting the sibling loop list
892 * graphite.c (graphite_transform_loops): Sort sibling loops.
894 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
896 * target.def (vec_perm_const_ok): Change sel parameter to
898 * optabs-query.c (can_vec_perm_p): Update accordingly.
899 * doc/tm.texi: Regenerate.
900 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
901 auto_vec_perm_indices and remove separate nelt field.
902 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
903 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
904 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
905 (aarch64_expand_vec_perm_const): Update accordingly.
906 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
908 * config/arm/arm.c (expand_vec_perm_d): Change perm to
909 auto_vec_perm_indices and remove separate nelt field.
910 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
911 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
912 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
914 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
916 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
917 sel to vec_perm_indices.
918 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
919 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
920 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
922 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
925 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
928 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
929 on the FUNCTION_DECL function context if it has a DIE that is a
932 2017-09-25 Richard Biener <rguenther@suse.de>
934 PR tree-optimization/82285
935 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
938 2017-09-25 Tom de Vries <tom@codesourcery.com>
942 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
945 2017-09-25 Richard Biener <rguenther@suse.de>
947 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
948 ISL errors other than isl_error_quota happen. Dump if the
949 schedule is the same.
950 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
951 errors instead of aborting inside ISL.
953 2017-09-25 Iain Sandoe <iain@codesourcery.com>
956 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
957 of libgcc_eh for m64.
958 * config/i386/darwin64.h: Likewise.
960 2017-09-25 Richard Biener <rguenther@suse.de>
963 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
964 attribute for incomplete types nor twice for complete ones.
966 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
969 * config/i386/i386.c (ix86_print_operand_address_as): Encode
970 %esp as %rsp to avoid 0x67 prefix if there is no index or base
973 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
976 * config/i386/i386.opt (mprefer-avx256): Use
977 ix86_target_flags variable.
978 * config/i386/i386.c (ix86_target_string): Move
979 -mprefer-avx256 to flag2_opts.
981 2017-09-22 Jakub Jelinek <jakub@redhat.com>
984 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
985 and x != -1 | y != -1 into (x & y) != -1.
987 2017-09-22 Steve Ellcey <sellcey@cavium.com>
989 * config.gcc: Add new case statement to set
990 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
991 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
992 s390x-*-linux* case statements. Added aarch64 to the list of
993 supported architectures.
995 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
997 PR tree-optimization/82289
998 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
999 STMT_VINFO_RELEVANT_P.
1001 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1002 Alan Hayward <alan.hayward@arm.com>
1003 David Sherwood <david.sherwood@arm.com>
1005 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1006 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1007 (extract_range_from_binary_expr_1): Don't call
1008 extract_range_from_multiplicative_op_1 if !range_int_cst_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 * target.def (preferred_vector_alignment): New hook.
1015 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1017 * doc/tm.texi: Regenerate.
1018 * targhooks.h (default_preferred_vector_alignment): Declare.
1019 * targhooks.c (default_preferred_vector_alignment): New function.
1020 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1022 (DR_TARGET_ALIGNMENT): New macro.
1023 (aligned_access_p): Update commentary.
1024 (vect_known_alignment_in_bytes): New function.
1025 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1027 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1028 Calculate the misalignment based on the target alignment rather than
1030 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1031 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1032 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1033 the target alignment, rather than masking the element misalignment
1034 with the number of elements in a vector. Also use the target
1035 alignment when calculating the maximum number of peels.
1036 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1037 instead of TYPE_ALIGN_UNIT.
1038 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1039 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1040 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1041 (vect_create_data_ref_ptr): Likewise.
1042 (vect_setup_realignment): Realign by ANDing with
1043 -DR_TARGET_MISALIGNMENT.
1044 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1045 the number of peels based on DR_TARGET_ALIGNMENT.
1046 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1047 with the guaranteed alignment boundary when deciding whether
1049 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1050 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1051 (ensure_base_align): Remove stmt_info parameter. Get the
1052 target base alignment from DR_TARGET_ALIGNMENT.
1053 (vectorizable_store): Update call accordingly. Interpret
1054 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1056 (vectorizable_load): Likewise.
1058 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1059 Alan Hayward <alan.hayward@arm.com>
1060 David Sherwood <david.sherwood@arm.com>
1062 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1063 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1064 (vect_enhance_data_refs_alignment): Likewise.
1066 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1068 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1069 error. Only quit immediately if parsing is complete.
1070 (BEGIN): Initialize fatal_err and parse_done.
1071 (begin fpu, end fpu): Check number of arguments.
1072 (begin arch, end arch): Likewise.
1073 (begin cpu, end cpu): Likewise.
1074 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1075 (optalias): Likewise.
1077 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1079 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1080 * config/arm/arm-isa.h: Delete. Move definitions to ...
1081 * arm-cpus.in: ... here. Use new feature and fgroup values.
1082 * config/arm/arm.c (arm_option_override): Use lower case for feature
1084 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1085 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1086 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1088 (print_isa_bits_for): New function.
1089 (gen_isa): New function.
1090 (gen_comm_data): Use print_isa_bits_for.
1091 (define feature): New keyword.
1092 (define fgroup): New keyword.
1093 * config/arm/t-arm (TM_H): Remove.
1094 (GTM_H): Add arm-isa.h.
1095 (arm-isa.h): Add rule to generate file.
1096 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1097 case for feature bit names.
1099 2017-09-22 Richard Biener <rguenther@suse.de>
1101 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1103 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1104 print of no dependency loops ...
1105 * graphite.c (graphite_transform_loops): ... here.
1106 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1108 (same_close_phi_node, remove_duplicate_close_phi,
1109 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1110 (canonicalize_loop_closed_ssa): ... here and simplify.
1111 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1112 (optimize_isl): Use dump_printf_loc to tell when we stopped
1113 optimizing because of an ISL timeout.
1115 2017-09-22 Richard Biener <rguenther@suse.de>
1117 PR tree-optimization/82291
1118 * tree-if-conv.c (predicate_mem_writes): Make sure to
1119 remove writes in blocks predicated with false.
1121 2017-09-22 Richard Biener <rguenther@suse.de>
1123 * sese.c: Include cfganal.h.
1124 (if_region_set_false_region): Remove.
1125 (create_if_region_on_edge): Likewise.
1126 (move_sese_in_condition): Re-implement without destroying
1129 2017-09-22 Richard Biener <rguenther@suse.de>
1131 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1132 Verify both BBs contain loop PHI nodes before dispatching to
1134 (graphite_regenerate_ast_isl): Do not recompute dominators,
1135 do not verify three times. Restructure for clarity.
1136 * graphite-scop-detection.c (same_close_phi_node,
1137 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1138 defined_in_loop_p, canonicalize_loop_closed_ssa,
1139 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1140 checking and SSA rewrite, move to ...
1141 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1142 (graphite_initialize): Do not pass in ctx, do not reset the
1143 SCEV cache, compute only dominators.
1144 (graphite_transform_loops): Allocate ISL ctx after
1145 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1146 Maintain post-dominators only around build_scops.
1147 * sese.c (if_region_set_false_region): Make static. Free
1148 and recompute dominators.
1149 (move_sese_in_condition): Assert we don't get called with
1150 post-dominators computed.
1151 * sese.h (if_region_set_false_region): Remove.
1153 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1155 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1156 mode attribute for TARGET_AVX512VL.
1158 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1160 * config/i386/i386.opt (mprefer-avx256): New option.
1161 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1163 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1164 for TARGET_PREFER_AVX256.
1165 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1167 2017-09-21 Jeff Law <law@redhat.com>
1169 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1170 Fix dump output if the only stack space is for pushed registers.
1172 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1174 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1177 2017-09-21 Martin Sebor <msebor@redhat.com>
1180 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1181 code (in C++) or code that triggers warnings.
1183 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1185 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1187 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1189 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1190 * sched-rgn.c: Ditto.
1191 * sel-sched-ir.c: Ditto.
1193 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1195 * toplev.h (set_random_seed): Adjust return type.
1196 * toplev.c (init_local_tick): Move eager initialization of random_seed
1197 to get_random_seed. Adjust comment.
1198 (init_random_seed): Inline to get_random_seed, delete.
1199 (get_random_seed): Initialize random_seed lazily.
1200 (set_random_seed): Do not return previous value.
1201 (print_switch_value): Do not call get_random_seed.
1203 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1205 * cgraph.c (delete_function_version): New, broken out from...
1206 (cgraph_node::delete_function_version): ...here. Rename to
1207 cgraph_node::delete_function_version_by_decl. Update all uses.
1208 (cgraph_node::remove): Call delete_function_version.
1210 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1213 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1214 VAR_DECLs to which addressable non-volatile parameters are mapped
1215 and for id->retvar after the return value assignment. Clear
1216 id->retval and id->retbnd after inlining.
1218 2017-09-21 Richard Biener <rguenther@suse.de>
1220 PR tree-optimization/82276
1221 PR tree-optimization/82244
1222 * tree-vrp.c (build_assert_expr_for): Set
1223 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1225 (remove_range_assertions): Revert earlier change.
1227 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1230 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1232 2017-09-21 Richard Biener <rguenther@suse.de>
1234 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1235 Restore valid IL after code generation errors.
1236 * graphite.c (graphite_transform_loops): Diagnose code
1237 generation issues as MSG_MISSED_OPTIMIZATION and continue
1238 with processing SCOPs.
1240 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1241 Alan Hayward <alan.hayward@arm.com>
1242 David Sherwood <david.sherwood@arm.com>
1244 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1245 rather than choosing between plus_constant and gen_rtx_<CODE>.
1246 * expr.c (emit_push_insn): Likewise.
1247 (expand_expr_real_2): Likewise.
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 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1255 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1256 Alan Hayward <alan.hayward@arm.com>
1257 David Sherwood <david.sherwood@arm.com>
1259 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1260 calling tree_to_uhwi.
1262 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1263 Alan Hayward <alan.hayward@arm.com>
1264 David Sherwood <david.sherwood@arm.com>
1266 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1267 INTEGER_CST rather than a negative test for ADDR_EXPR.
1269 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1270 Alan Hayward <alan.hayward@arm.com>
1271 David Sherwood <david.sherwood@arm.com>
1273 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1274 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1276 2017-09-21 Richard Biener <rguenther@suse.de>
1278 PR tree-optimization/71351
1279 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1280 graphite_create_new_loop_guard): Remove, fold remaining parts
1282 (translate_isl_ast_node_for): ... here and simplify.
1284 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1287 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1288 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1289 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1290 alternative always use QI mode, for -Os imov (=R,R) alternative
1291 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1294 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1295 Jeff Law <law@redhat.com>
1297 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1298 (allocate_stack_space): New function, partially extracted from
1300 (s390_emit_prologue): Track offset to most recent stack probe.
1301 Code to allocate space moved into allocate_stack_space.
1302 Dump actions when no stack is allocated.
1303 (s390_prologue_plus_offset): New function.
1304 (s390_emit_stack_probe): Likewise.
1306 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1308 * common.opt (Wa, Wl, Wp, g, gz=): Add
1310 (gno-column-info): Remove.
1311 (gcolumn-info): Drop RejectNegative.
1313 (gno-record-gcc-switches): Remove.
1314 (grecord-gcc-switches): Drop RejectNegative.
1315 (gno-split-dwarf): Remove.
1316 (gsplit-dwarf): Drop RejectNegative.
1317 (gno-strict-dwarf): Remove.
1318 (gstrict-dwarf): Drop RejectNegative.
1319 * config/darwin.opt (gfull, gused): Add RejectNegative.
1320 * dwarf2out.c (gen_producer_string): Drop
1321 gno-record-gcc-switches handler.
1322 * optc-gen.awk: Add g to prefixes with negative forms.
1323 * opts-common.c (remapping_prefix_p): New.
1324 (find_opt): Check it.
1325 (generate_canonical_option): Test g prefix.
1326 (option_map): Add -gno- mapping.
1327 (add_misspelling_candidates): Check remapping_prefix_p.
1329 2017-09-20 Jeff Law <law@redhat.com>
1331 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1332 thinko in stack clash protection support.
1334 * explow.c (compute_stack_clash_protection_loop_data): Use
1335 CONST_INT_P instead of explicit test. Verify object is a
1336 CONST_INT_P before looking at INTVAL.
1337 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1338 instead of explicit test.
1340 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1343 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1344 address instead of to r1 and r11.
1346 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1348 * config.gcc: Support "knm".
1349 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1350 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1352 * config/i386/i386.c (m_KNM): Define.
1353 (processor_target_table): Add "knm".
1355 (ix86_option_override_internal): Add "knm".
1356 (ix86_issue_rate): Add PROCESSOR_KNM.
1357 (ix86_adjust_cost): Ditto.
1358 (ia32_multipass_dfa_lookahead): Ditto.
1359 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1360 (fold_builtin_cpu): Add M_INTEL_KNM.
1361 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1362 (processor_type): Add PROCESSOR_KNM.
1363 * config/i386/x86-tune.def: Add m_KNM.
1364 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1366 2017-09-20 Richard Biener <rguenther@suse.de>
1368 PR tree-optimization/80213
1369 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1370 are allowed in empty BBs as well.
1371 (canonicalize_loop_closed_ssa): Also look for other complex
1373 (scop_detection::get_sese): Include the loop-closed PHI block
1375 (scop_detection::merge_sese): Remove code adding extra blocks.
1376 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1377 (build_scops): Assert the final returned scop is invalid.
1379 2017-09-20 Richard Biener <rguenther@suse.de>
1381 PR tree-optimization/82264
1382 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1384 (vn_phi_lookup): Likewise.
1385 (vn_phi_insert): Likewise.
1387 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1389 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1390 that fits into uhwi or shwi, add DW_AT_const_value regardless
1391 of early_dwarf without going through RTL, using add_AT_unsigned
1394 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1395 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1396 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1397 (DEBUG_MACRO_SECTION): Likewise.
1398 (DEBUG_LTO_MACRO_SECTION): Likewise.
1399 (DEBUG_STR_DWO_SECTION): Likewise.
1400 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1401 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1402 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1403 (DEBUG_LTO_LINE_STR_SECTION): Define.
1404 (init_sections_and_labels): Initialize debug_line_str_section
1405 variable. Initialize debug_loc_section for -gdwarf-5 to
1406 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1408 2017-09-20 Richard Biener <rguenther@suse.de>
1410 * graphite-sese-to-poly.c (extract_affine): Properly handle
1411 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1413 2017-09-20 Richard Biener <rguenther@suse.de>
1415 PR tree-optimization/81373
1416 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1417 Force SESE live-out defs to be handled even if they are
1420 2017-09-19 Jeff Law <law@redhat.com>
1422 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1423 nothing for stack adjustments with REG_STACK_CHECK.
1424 * sched-deps.c (parse_add_or_inc): Reject insns with
1425 REG_STACK_CHECK from dependency breaking.
1426 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1427 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1428 * reg-notes.def (STACK_CHECK): New note.
1430 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1431 (ix86_expand_prologue): Dump stack clash info as needed.
1432 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1434 * function.c (dump_stack_clash_frame_info): New function.
1435 * function.h (dump_stack_clash_frame_info): Prototype.
1436 (enum stack_clash_probes): New enum.
1438 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1439 flag_stack_clash_protection.
1440 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1441 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1442 (arm_frame_pointer_required): Likewise.
1443 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1444 (ia64_expand_prologue): Likewise.
1445 * config/mips/mips.c (mips_expand_prologue): Likewise.
1446 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1447 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1448 (sparc_flat_expand_prologue): Likewise.
1449 * config/spu/spu.c (spu_expand_prologue): Likewise.
1451 * explow.c: Include "params.h".
1452 (anti_adjust_stack_and_probe_stack_clash): New function.
1453 (get_stack_check_protect): Likewise.
1454 (compute_stack_clash_protection_loop_data): Likewise.
1455 (emit_stack_clash_protection_loop_start): Likewise.
1456 (emit_stack_clash_protection_loop_end): Likewise.
1457 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1458 Use anti_adjust_stack_and_probe_stack_clash.
1459 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1460 (emit_stack_clash_protection_loop_start): Likewise.
1461 (emit_stack_clash_protection_loop_end): Likewise.
1462 * rtl.h (get_stack_check_protect): Prototype.
1463 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1464 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1465 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1467 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1469 * doc/tm.texi: Rebuilt.
1470 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1471 get_stack_check_protect.
1472 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1473 * config/arm/arm.c (arm_expand_prologue): Likewise.
1474 (arm_frame_pointer_required): Likewise.
1475 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1476 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1477 * config/mips/mips.c (mips_expand_prologue): Likewise.
1478 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1479 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1480 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1481 (sparc_flat_expand_prologue): Likewise.
1483 * common.opt (-fstack-clash-protection): New option.
1484 * flag-types.h (enum stack_check_type): Note difference between
1485 -fstack-check= and -fstack-clash-protection.
1486 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1487 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1488 * toplev.c (process_options): Issue warnings/errors for cases
1489 not handled with -fstack-clash-protection.
1490 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1491 (-fstack-check): Note additional problem with -fstack-check=generic.
1492 Note that -fstack-check is primarily for Ada and refer users
1493 to -fstack-clash-protection for stack-clash-protection.
1494 Document new params for stack clash protection.
1496 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1498 * config/i386/i386.c (ix86_split_long_move): Do not handle
1499 address used for LEA in a special way.
1501 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1503 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1505 2017-09-19 Martin Sebor <msebor@redhat.com>
1508 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1509 of incompatible types.
1511 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1514 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1515 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1516 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1518 2017-09-19 Richard Biener <rguenther@suse.de>
1520 PR tree-optimization/82244
1521 * tree-vrp.c (remove_range_assertions): Do not propagate
1522 a constant to abnormals but replace the assert with a copy.
1524 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1526 PR rtl-optimization/57878
1527 PR rtl-optimization/68988
1528 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1529 avoidance test involving non_reload_pseudos. Move frequency test
1530 below the general fragmentation avoidance test.
1532 2017-09-19 Richard Biener <rguenther@suse.de>
1534 PR tree-optimization/69728
1535 * graphite-sese-to-poly.c (schedule_error): New global.
1536 (add_loop_schedule): Handle empty domain by failing the
1538 (build_original_schedule): Handle schedule_error.
1540 2017-09-19 Richard Biener <rguenther@suse.de>
1542 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1543 Do not iterate to sibling loops but only to siblings of inner
1546 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1549 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1550 that the registers are different.
1552 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1554 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1555 to processor_model and "amdfam17h" to arch_names_table.
1556 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1558 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1561 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1563 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1564 Alan Hayward <alan.hayward@arm.com>
1565 David Sherwood <david.sherwood@arm.com>
1567 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1569 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1571 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1572 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1573 vect_schedule_slp_instance.
1574 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1575 Update call to vect_slp_analyze_node_operations. Simplify return
1577 (vect_slp_analyze_bb_1): Update call accordingly.
1578 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1579 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1580 (vect_schedule_slp): Update call accordingly.
1582 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1583 Alan Hayward <alan.hayward@arm.com>
1584 David Sherwood <david.sherwood@arm.com>
1586 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1587 with types that aren't in fact scalar.
1589 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1591 * tree-vect-slp.c (vect_record_max_nunits): New function,
1593 (vect_build_slp_tree_1): ...here.
1594 (vect_build_slp_tree_2): Call it for phis too.
1596 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1598 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1599 to vect_get_vec_def_for_operand when getting the mask operand.
1601 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1602 Alan Hayward <alan.hayward@arm.com>
1603 David Sherwood <david.sherwood@arm.com>
1605 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1608 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1609 Alan Hayward <alan.hayward@arm.com>
1610 David Sherwood <david.sherwood@arm.com>
1612 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1613 calculation for vector booleans.
1615 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1616 Alan Hayward <alan.hayward@arm.com>
1617 David Sherwood <david.sherwood@arm.com>
1619 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1621 (vect_transform_stmt): ...here.
1622 (vect_analyze_stmt): Use it instead of calling
1623 vectorizable_live_operation directly.
1625 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1627 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1628 non-SIMT targets in acc vector loops.
1630 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1632 * configure.ac: Add arc and check if assembler supports gdwarf2.
1633 * configure: Regenerate.
1635 2017-09-18 Richard Biener <rguenther@suse.de>
1637 PR tree-optimization/82220
1638 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1639 epilogue niters from the min_profitable_iters compute.
1641 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1644 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1645 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1646 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1648 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1651 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1652 switching to a new text section.
1654 2017-09-18 Richard Biener <rguenther@suse.de>
1656 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1658 (build_alias_set): Reject aliases with no access function.
1660 2017-09-18 Richard Biener <rguenther@suse.de>
1662 PR tree-optimization/79622
1663 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1665 (build_cross_bb_scalars_use): Likewise.
1667 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1669 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1671 2017-09-18 Alan Modra <amodra@gmail.com>
1674 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1675 stack_pointer_rtx for count 0. Update comments. Break up
1676 large rtl expression.
1678 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1680 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1681 Increase to 20 bytes.
1682 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1683 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1684 or avx version of the stub.
1686 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1689 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1690 compute the minimum stack alignment. Also update preferred stack
1691 boundary for leaf functions.
1693 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1695 PR tree-optimization/82228
1696 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1699 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1701 * common/config/nds32/nds32-common.c
1702 (nds32_option_optimization_table): Refine formatting.
1703 (nds32_option_optimization_table): Use -fsched-pressure and
1704 -fomit-frame-pointer for specific optimization level.
1706 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1708 * config/nds32/nds32.c: Refine formatting and comments.
1709 * config/nds32/nds32.h: Likewise.
1710 * config/nds32/nds32.md: Likewise.
1711 * config/nds32/nds32-cost.c: Likewise.
1712 * config/nds32/nds32-isr.c: Likewise.
1713 * config/nds32/nds32-md-auxiliary.c: Likewise.
1714 * config/nds32/nds32-multiple.md: Likewise.
1715 * config/nds32/nds32-predicates.c: Likewise.
1717 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1718 Jakub Jelinek <jakub@redhat.com>
1720 Add support for -std=c++2a.
1721 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1723 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1725 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1728 * doc/extend.texi (Common Function Attributes): Add
1729 references to ARM, AArch64, and S/390 specific attributes.
1730 (Function Specific Option Pragmas): Add AArch64 and S/390
1731 to list of back ends that support the target pragma.
1733 2017-09-15 Nathan Sidwell <nathan@acm.org>
1735 * doc/standards.texi: Fix C++17 description. Update URLs for
1738 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1740 * common.opt (Wcast-align=strict): New warning option.
1741 * doc/invoke.texi: Document -Wcast-align=strict.
1743 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1745 * cgraph.h (cgraph_thunk_info): Add comments.
1746 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1747 assert for VIRTUAL_* arguments stricter.
1749 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1751 PR tree-optimization/71026
1752 * match.pd: Move RDIV patterns from fold-const.c
1753 * fold-const.c (distribute_real_division): Removed.
1754 (fold_binary_loc): Remove calls to distribute_real_divison.
1756 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1758 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1759 c++1z and gnu++1z as deprecated. Change other references to
1760 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1761 Change -Wc++1z-compat to -Wc++17-compat.
1762 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1763 * dwarf2out.c (highest_c_language): Handle C++17.
1764 (gen_compile_unit_die): Likewise.
1766 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1768 PR rtl-optimization/82192
1769 * combine.c (make_extraction): Don't look through non-paradoxical
1770 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1773 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1774 Alan Hayward <alan.hayward@arm.com>
1775 David Sherwood <david.sherwood@arm.com>
1777 * target.def (function_arg_offset): New hook.
1778 * targhooks.h (default_function_arg_offset): Declare.
1779 * targhooks.c (default_function_arg_offset): New function.
1780 * function.c (locate_and_pad_parm): Use
1781 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1782 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1783 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1784 * doc/tm.texi: Regenerate.
1785 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1786 * config/spu/spu.c (spu_function_arg_offset): New function.
1787 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1788 * system.h (FUNCTION_ARG_OFFSET): Poison.
1790 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayard <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * target.def (truly_noop_truncation): New hook.
1795 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1796 than TRULY_NOOP_TRUNCATION.
1797 * hooks.h (hook_bool_uint_uint_true): Declare.
1798 * hooks.c (hook_bool_uint_uint_true): New function.
1799 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1800 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1801 * doc/tm.texi: Regenerate.
1802 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1803 rather than TRULY_NOOP_TRUNCATION in comments.
1804 (simplify_comparison): Likewise.
1805 (record_truncated_value): Likewise.
1806 * expmed.c (extract_bit_field_1): Likewise.
1807 (extract_split_bit_field): Likewise.
1808 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1809 instead of TRULY_NOOP_TRUNCATION.
1810 * function.c (assign_parm_setup_block): Likewise.
1811 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1812 * rtlhooks.c: Include target.h.
1813 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1814 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1815 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1816 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1817 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1818 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1819 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1820 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1821 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1822 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1823 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1824 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1825 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1826 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1827 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1828 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1829 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1830 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1831 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1832 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1833 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1834 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1835 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1836 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1837 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1838 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1839 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1840 rather than TRULY_NOOP_TRUNCATION in comments.
1841 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1842 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1843 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1844 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1845 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1846 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1847 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1848 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1849 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1850 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1851 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1852 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1853 rather than TRULY_NOOP_TRUNCATION in comments.
1854 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1855 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1856 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1857 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1858 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1859 TRULY_NOOP_TRUNCATION condition.
1860 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1861 (TRULY_NOOP_TRUNCATION): Delete.
1862 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1863 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1864 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1865 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1866 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1867 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1868 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1869 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1870 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1871 rather than TRULY_NOOP_TRUNCATION in comments.
1872 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1873 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1874 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1875 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1876 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1877 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1879 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1882 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1883 (*cmp_ior): Likewise.
1884 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1885 (*ior_scc_scc_cmp): Likewise.
1886 (*and_scc_scc): Likewise.
1887 (*and_scc_scc_cmp): Likewise.
1889 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1890 Alan Hayard <alan.hayward@arm.com>
1891 David Sherwood <david.sherwood@arm.com>
1893 * target.def (can_change_mode_class): New hook.
1894 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1895 (hard_regno_nregs): Likewise.
1896 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1897 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1898 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1899 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1900 (LOAD_EXTEND_OP): Update accordingly.
1901 * doc/tm.texi: Regenerate.
1902 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1903 CANNOT_CHANGE_MODE_CLASS.
1904 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1905 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1906 * combine.c (simplify_set): Update accordingly.
1907 * emit-rtl.c (validate_subreg): Likewise.
1908 * recog.c (general_operand): Likewise.
1909 * regcprop.c (mode_change_ok): Likewise.
1910 * reload1.c (choose_reload_regs): Likewise.
1911 (inherit_piecemeal_p): Likewise.
1912 * rtlanal.c (simplify_subreg_regno): Likewise.
1913 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1914 instead of CANNOT_CHANGE_MODE_CLASS.
1915 (reload_cse_simplify_operands): Likewise.
1916 * reload.c (push_reload): Use targetm.can_change_mode_class
1917 instead of CANNOT_CHANGE_MODE_CLASS.
1918 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1919 REG_CANNOT_CHANGE_MODE_P.
1920 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1921 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1922 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1923 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1924 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1925 (arm_can_change_mode_class): New function.
1926 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1927 than CANNOT_CHANGE_MODE_CLASS in comments.
1928 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1929 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1930 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1931 (ix86_can_change_mode_class): ...this new function, inverting the
1932 sense of the return value.
1933 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1934 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1935 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1936 (ia64_can_change_mode_class): New function.
1937 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1938 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1939 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1940 (m32c_can_change_mode_class): ...this new function, inverting the
1941 sense of the return value.
1942 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1943 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1944 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1945 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1946 (mips_can_change_mode_class): ...this new function, inverting the
1947 sense of the return value.
1948 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1949 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1950 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1951 (msp430_can_change_mode_class): New function.
1952 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1953 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1954 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1955 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1956 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1957 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1958 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1959 (pa_cannot_change_mode_class): Replace with...
1960 (pa_can_change_mode_class): ...this new function, inverting the
1961 sense of the return value.
1962 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1963 than CANNOT_CHANGE_MODE_CLASS in comments.
1964 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1965 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1966 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1967 (pdp11_cannot_change_mode_class): Replace with...
1968 (pdp11_can_change_mode_class): ...this new function, inverting the
1969 sense of the return value.
1970 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1971 * config/powerpcspe/powerpcspe-protos.h
1972 (rs6000_cannot_change_mode_class_ptr): Delete.
1973 * config/powerpcspe/powerpcspe.c
1974 (rs6000_cannot_change_mode_class_ptr): Delete.
1975 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1976 (rs6000_option_override_internal): Assign to
1977 targetm.can_change_mode_class instead of
1978 rs6000_cannot_change_mode_class_ptr.
1979 (rs6000_cannot_change_mode_class): Replace with...
1980 (rs6000_can_change_mode_class): ...this new function, inverting the
1981 sense of the return value.
1982 (rs6000_debug_cannot_change_mode_class): Replace with...
1983 (rs6000_debug_can_change_mode_class): ...this new function.
1984 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1985 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1986 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1987 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1988 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1990 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1991 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1992 (rs6000_option_override_internal): Assign to
1993 targetm.can_change_mode_class instead of
1994 rs6000_cannot_change_mode_class_ptr.
1995 (rs6000_cannot_change_mode_class): Replace with...
1996 (rs6000_can_change_mode_class): ...this new function, inverting the
1997 sense of the return value.
1998 (rs6000_debug_cannot_change_mode_class): Replace with...
1999 (rs6000_debug_can_change_mode_class): ...this new function.
2000 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2001 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2002 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2003 (s390_can_change_mode_class): ...this new function, inverting the
2004 sense of the return value.
2005 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2006 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2007 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2008 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2009 (sh_cannot_change_mode_class): Replace with...
2010 (sh_can_change_mode_class): ...this new function, inverting the
2011 sense of the return value.
2012 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2013 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2014 (sparc_can_change_mode_class): New function.
2015 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2016 * config/spu/spu.c (spu_can_change_mode_class): New function.
2017 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2018 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2019 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2020 (visium_can_change_mode_class): New function.
2021 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2023 2017-09-15 Richard Biener <rguenther@suse.de>
2025 PR tree-optimization/82217
2026 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2027 but not undefined case.
2029 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2032 * postreload.c (reload_cse_simplify_operands): Skip
2033 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2035 2017-09-15 Richard Biener <rguenther@suse.de>
2037 PR tree-optimization/68823
2038 * graphite-scop-detection.c (build_alias_set): If we have a
2039 possible dependence check whether we can handle them by just
2040 looking at the DRs DR_ACCESS_FNs.
2041 (build_scops): If build_alias_set fails, fail the SCOP.
2043 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2045 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2046 to support float128 built-in functions that require the ISA 3.0
2048 (BU_FLOAT128_3_HW): Likewise.
2049 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2051 (FMAF128): Likewise.
2053 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2054 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2055 floating point instructions.
2056 (rs6000_invalid_builtin): Likewise.
2057 (rs6000_builtin_mask_names): Likewise.
2058 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2059 (RS6000_BTM_FLOAT128_HW): Likewise.
2060 (RS6000_BTM_COMMON): Likewise.
2061 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2063 * doc/extend.texi (RS/6000 built-in functions): Document the
2064 IEEE 128-bit floating point square root and fused multiply-add
2067 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2069 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2070 reg (r2) isn't in the set of registers defined in the prologue.
2072 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2073 Alan Hayward <alan.hayward@arm.com>
2074 David Sherwood <david.sherwood@arm.com>
2076 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2077 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2078 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2079 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2080 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2082 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2083 max_vectorization_factor.
2084 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2086 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2087 Alan Hayward <alan.hayward@arm.com>
2088 David Sherwood <david.sherwood@arm.com>
2090 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2091 (vect_worthwhile_without_simd_p): Declare.
2092 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2093 (vectorizable_reduction): Use it.
2094 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2095 (vectorizable_operation): Likewise.
2097 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2098 Alan Hayward <alan.hayward@arm.com>
2099 David Sherwood <david.sherwood@arm.com>
2101 * tree-vectorizer.h (vect_get_num_copies): New function.
2102 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2103 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2104 (vectorizable_induction): Likewise.
2105 (vectorizable_live_operation): Likewise.
2106 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2107 (vectorizable_bswap): Likewise.
2108 (vectorizable_call): Likewise.
2109 (vectorizable_conversion): Likewise.
2110 (vectorizable_assignment): Likewise.
2111 (vectorizable_shift): Likewise.
2112 (vectorizable_operation): Likewise.
2113 (vectorizable_store): Likewise.
2114 (vectorizable_load): Likewise.
2115 (vectorizable_condition): Likewise.
2116 (vectorizable_comparison): Likewise.
2117 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2119 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2120 Alan Hayward <alan.hayward@arm.com>
2121 David Sherwood <david.sherwood@arm.com>
2123 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2124 of vect_init_vector.
2126 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2127 Alan Hayward <alan.hayward@arm.com>
2128 David Sherwood <david.sherwood@arm.com>
2130 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2131 an inline wrapper that provides a location.
2132 (gimple_build_vector): Likewise.
2133 * gimple-fold.c (gimple_build_vector_from_val): New function.
2134 (gimple_build_vector): Likewise.
2135 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2136 functions to build the initial value. Always return a gimple value.
2137 (get_initial_defs_for_reduction): Likewise. Only compute
2139 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2140 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2141 (vectorizable_induction): Use gimple_build_vector rather than
2144 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2145 Alan Hayward <alan.hayward@arm.com>
2146 David Sherwood <david.sherwood@arm.com>
2148 * target.h (vec_perm_indices): New typedef.
2149 (auto_vec_perm_indices): Likewise.
2150 * optabs-query.h: Include target.h
2151 (can_vec_perm_p): Take a vec_perm_indices *.
2152 * optabs-query.c (can_vec_perm_p): Likewise.
2153 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2154 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2155 * tree-vect-generic.c (lower_vec_perm): Likewise.
2156 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2157 (vect_grouped_load_supported): Likewise.
2158 (vect_shift_permute_load_chain): Likewise.
2159 (vect_permute_store_chain): Use auto_vec_perm_indices.
2160 (vect_permute_load_chain): Likewise.
2161 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2162 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2163 Update uses of can_vec_perm_p.
2164 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2165 mode with a number of elements. Take a vec_perm_indices *.
2166 (vect_create_epilog_for_reduction): Update accordingly.
2167 Use auto_vec_perm_indices.
2168 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2169 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2170 (vect_transform_slp_perm_load): Likewise.
2171 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2172 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2173 (vect_gen_perm_mask_checked): Likewise.
2174 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2175 (vect_gen_perm_mask_checked): Likewise.
2176 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2177 (vectorizable_store): Likewise.
2178 (vectorizable_load): Likewise.
2179 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2180 (vectorizable_bswap): Likewise.
2182 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2183 Alan Hayward <alan.hayward@arm.com>
2184 David Sherwood <david.sherwood@arm.com>
2186 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2187 * tree.c (build_vector): Likewise.
2188 (build_vector_from_ctor): Update accordingly.
2189 (build_vector_from_val): Likewise.
2190 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2191 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2192 * tree-vect-generic.c (add_rshift): Likewise.
2193 (expand_vector_divmod): Likewise.
2194 (optimize_vector_constructor): Likewise.
2195 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2196 (vect_transform_slp_perm_load): Likewise.
2197 (vect_schedule_slp_instance): Likewise.
2198 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2199 (vectorizable_call): Likewise.
2200 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2201 * expmed.c (make_tree): Likewise.
2202 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2203 a vector passed to build_vector.
2204 (fold_convert_const): Likewise.
2205 (exact_inverse): Likewise.
2206 (fold_ternary_loc): Likewise.
2207 (fold_relational_const): Likewise.
2208 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2209 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2210 (const_unop): Likewise. Store the reduction accumulator in a
2211 variable rather than an array.
2212 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2213 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2214 the new vector, rather than constructing it after the input arrays.
2215 (native_interpret_vector): Use auto_vec<tree> when building
2216 a vector passed to build_vector. Add elements in order.
2217 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2218 auto_vec<tree> when building a vector passed to build_vector.
2219 (vect_create_epilog_for_reduction): Likewise.
2220 (vectorizable_induction): Likewise.
2221 (get_initial_def_for_reduction): Likewise. Fix indentation of
2223 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2225 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2226 passed to build_vector.
2228 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2229 Alan Hayward <alan.hayward@arm.com>
2230 David Sherwood <david.sherwood@arm.com>
2232 * tree-core.h (tree_base::u): Add an "nelts" field.
2233 (tree_vector): Use VECTOR_CST_NELTS as the length.
2234 * tree.c (tree_size): Likewise.
2235 (make_vector): Initialize VECTOR_CST_NELTS.
2236 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2237 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2238 TYPE_VECTOR_SUBPARTS.
2239 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2240 as the number of units, setting it from VECTOR_CST_NELTS.
2241 (const_vector_from_tree): Likewise.
2242 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2243 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2244 (fold_negate_expr_1): Likewise.
2245 (fold_convert_const): Likewise.
2246 (const_binop): Likewise. Differentiate the number of output and
2248 (const_unop): Likewise.
2249 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2250 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2251 in cases that did the opposite.
2253 2017-09-14 Richard Biener <rguenther@suse.de>
2255 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2258 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2260 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2262 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2265 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2266 if and where to split a bb, except for splitting before debug insn
2267 sequences followed by non-label real insn. Delete debug insns
2268 in between basic blocks.
2270 * combine.c (make_compound_operation_int): Formatting fixes.
2272 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2273 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2274 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2275 * config/sol2.h (LINK_EH_SPEC): Likewise.
2276 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2277 * config/s390/linux.h (LINK_SPEC): Likewise.
2278 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2279 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2280 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2281 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2282 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2283 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2284 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2285 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2287 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2289 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2291 (ENDFILE_LINUX_SPEC): Likewise.
2292 (LINK_EH_SPEC): Likewise.
2293 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2294 (LINK_OS_LINUX_SPEC32): Likewise.
2295 (LINK_OS_LINUX_SPEC64): Likewise.
2296 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2297 (LINK_OS_LINUX_SPEC): Likewise.
2299 2017-09-13 Martin Liska <mliska@suse.cz>
2302 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2303 CASE_HIGH is NULL_TREE.
2305 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2306 Alan Hayward <alan.hayward@arm.com>
2307 David Sherwood <david.sherwood@arm.com>
2309 * target.def (secondary_memory_needed): New hook.
2310 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2311 instead of SECONDARY_MEMORY_NEEDED.
2312 (secondary_memory_needed_mode): Likewise.
2313 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2314 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2315 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2316 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2317 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2318 * doc/tm.texi: Regenerate.
2319 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2320 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2321 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2322 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2323 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2324 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2325 mode argument first and change the reg_class arguments to reg_class_t.
2326 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2327 Make static. Update the call to inline_secondary_memory_needed.
2328 (ix86_register_move_cost): Update the call to
2329 inline_secondary_memory_needed.
2330 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2331 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2333 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2334 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2336 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2337 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2338 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2339 and match hook interface. Add comment from mips.h.
2340 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2341 * config/mmix/mmix.md (truncdfsf2): Refer to
2342 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2344 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2345 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2346 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2347 (pa_secondary_memory_needed): New function.
2348 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2349 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2350 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2351 (pdp11_secondary_memory_needed): Make static and match hook interface.
2352 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2353 * config/powerpcspe/powerpcspe-protos.h
2354 (rs6000_secondary_memory_needed_ptr): Delete.
2355 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2357 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2358 (rs6000_option_override_internal): Assign to
2359 targetm.secondary_memory_needed rather than
2360 rs6000_secondary_memory_needed_ptr.
2361 (rs6000_secondary_memory_needed): Match hook interface.
2362 (rs6000_debug_secondary_memory_needed): Likewise.
2363 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2364 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2365 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2366 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2367 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2368 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2370 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2371 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2372 (rs6000_option_override_internal): Assign to
2373 targetm.secondary_memory_needed rather than
2374 rs6000_secondary_memory_needed_ptr.
2375 (rs6000_secondary_memory_needed): Match hook interface.
2376 (rs6000_debug_secondary_memory_needed): Likewise.
2377 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2378 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2379 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2380 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2381 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2382 (sparc_secondary_memory_needed): New function.
2383 * lra-constraints.c (check_and_process_move): Refer to
2384 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2386 (curr_insn_transform): Likewise.
2387 (process_alt_operands): Use targetm.secondary_memory_needed
2388 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2389 (check_secondary_memory_needed_p): Likewise.
2390 (choose_split_class): Likewise.
2391 * reload.c: Unconditionally include code that was previously
2392 conditional on SECONDARY_MEMORY_NEEDED.
2393 (push_secondary_reload): Use targetm.secondary_memory_needed
2394 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2395 (push_reload): Likewise.
2396 * reload1.c: Unconditionally include code that was previously
2397 conditional on SECONDARY_MEMORY_NEEDED.
2398 (choose_reload_regs): Use targetm.secondary_memory_needed
2399 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2400 (gen_reload): Likewise.
2401 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2403 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2404 Alan Hayward <alan.hayward@arm.com>
2405 David Sherwood <david.sherwood@arm.com>
2407 * target.def (secondary_memory_needed_mode): New hook:
2408 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2409 * targhooks.h (default_secondary_memory_needed_mode): New function.
2410 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2411 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2412 * doc/tm.texi: Regenerate.
2413 * lra-constraints.c (check_and_process_move): Use
2414 targetm.secondary_memory_needed_mode instead of
2415 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2416 (curr_insn_transform): Likewise.
2417 * reload.c (get_secondary_mem): Likewise.
2418 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2419 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2421 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2422 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2423 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2424 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2425 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2427 * config/powerpcspe/powerpcspe-protos.h
2428 (rs6000_secondary_memory_needed_mode): Delete.
2429 * config/powerpcspe/powerpcspe.c
2430 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2431 (rs6000_secondary_memory_needed_mode): Make static.
2432 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2433 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2435 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2437 (rs6000_secondary_memory_needed_mode): Make static.
2438 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2439 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2440 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2441 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2442 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2444 (sparc_secondary_memory_needed_mode): New function.
2445 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2447 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2449 * config/aarch64/constraints.md (Umq): New constraint.
2450 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2452 (mov<mode>): Update condition.
2454 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2456 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2457 when bitposition is the same.
2459 2017-09-13 Richard Biener <rguenther@suse.de>
2461 * dwarf2out.c (output_die_symbol): Remove.
2462 (output_die): Do not output a DIEs symbol.
2464 2017-09-13 Richard Biener <rguenther@suse.de>
2467 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2468 default-def to avoid breaking iterator update with the weird
2469 interaction with cgraph_update_edges_for_call_stmt_node.
2471 2017-09-13 Richard Biener <rguenther@suse.de>
2473 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2474 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2475 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2476 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2478 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2480 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2481 Disable pc relative literal load irrespective of
2482 TARGET_FIX_ERR_A53_84341 for default.
2484 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2486 * config/sparc/sparc.c (output_return): Output the source location of
2487 the insn in the delay slot, if any.
2488 (output_sibcall): Likewise.
2490 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2493 * common.opt (-static-pie): New alias.
2494 (shared): Negate static-pie.
2495 (-no-pie): Update help text.
2497 (static-pie): New option.
2498 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2499 -static-pie support.
2500 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2501 (LINK_EH_SPEC): Likewise.
2502 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2503 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2504 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2505 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2506 (init_gcc_specs): Likewise.
2507 (init_spec): Likewise.
2508 (display_help): Update help message for -pie.
2509 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2512 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2514 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2515 (movdi_aarch64): Likewise.
2516 (movti_aarch64): Likewise.
2518 2017-09-12 Simon Wright <simon@pushface.org>
2521 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2522 calculation of the minor version, always output as 0.
2524 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2527 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2528 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2529 on it early, rather than manual conversion late. For
2530 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2531 instead of performing manual conversion.
2533 2017-09-12 Carl Love <cel@us.ibm.com>
2535 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2536 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2538 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2539 VMULOSW): Add definitions.
2540 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2541 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2542 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2543 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2544 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2546 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2548 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2550 (movti_aarch64): Likewise.
2551 (movdf_aarch64): Likewise.
2552 (movtf_aarch64): Likewise.
2553 (load_pairdi): Likewise.
2554 (store_pairdi): Likewise.
2555 (load_pairdf): Likewise.
2556 (store_pairdf): Likewise.
2557 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2558 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2559 (ldr_got_small_<mode>): Likewise.
2560 (ldr_got_small_28k_<mode>): Likewise.
2561 (ldr_got_tiny): Likewise.
2562 * config/aarch64/iterators.md (ldst_sz): New.
2563 (ldpstp_sz): Likewise.
2564 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2566 (thunderx_load): Split load_8 to load_16.
2567 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2569 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2570 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2571 (xgene1_store_pair): Split store_8 to store_16.
2572 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2573 (falkor_st_0_st_sd): Split store_8 to store_16.
2575 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2577 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2578 and store1/2/3/4 to store_4/8/12/16.
2579 * config/aarch64/aarch64.md: Update for rename.
2580 * config/arm/arm.md: Likewise.: Likewise.
2581 * config/arm/arm.c: Likewise.
2582 * config/arm/thumb1.md: Likewise.
2583 * config/arm/thumb2.md: Likewise.
2584 * config/arm/vfp.md: Likewise.
2585 * config/arm/arm-generic.md: Likewise.
2586 * config/arm/arm1020e.md: Likewise.
2587 * config/arm/arm1026ejs.md: Likewise.
2588 * config/arm/arm1136jfs.md: Likewise.
2589 * config/arm/arm926ejs.md: Likewise.
2590 * config/arm/cortex-a15.md: Likewise.
2591 * config/arm/cortex-a17.md: Likewise.
2592 * config/arm/cortex-a5.md: Likewise.
2593 * config/arm/cortex-a53.md: Likewise.
2594 * config/arm/cortex-a57.md: Likewise.
2595 * config/arm/cortex-a7.md: Likewise.
2596 * config/arm/cortex-a8.md: Likewise.
2597 * config/arm/cortex-a9.md: Likewise.
2598 * config/arm/cortex-m4.md: Likewise.
2599 * config/arm/cortex-m7.md: Likewise.
2600 * config/arm/cortex-r4.md: Likewise.
2601 * config/arm/exynos-m1.md: Likewise.
2602 * config/arm/fa526.md: Likewise.
2603 * config/arm/fa606te.md: Likewise.
2604 * config/arm/fa626te.md: Likewise.
2605 * config/arm/fa726te.md: Likewise.
2606 * config/arm/fmp626.md: Likewise.
2607 * config/arm/iwmmxt.md: Likewise.
2608 * config/arm/ldmstm.md: Likewise.
2609 * config/arm/marvell-pj4.md: Likewise.
2610 * config/arm/xgene1.md: Likewise.
2611 * config/aarch64/thunderx.md: Likewise.
2612 * config/aarch64/thunderx2t99.md: Likewise.
2613 * config/aarch64/falkor.md: Likewise.
2615 2017-09-12 Martin Liska <mliska@suse.cz>
2617 * attribs.c (private_lookup_attribute): New function.
2618 * attribs.h (private_lookup_attribute): Declared here.
2619 (lookup_attribute): Called from this place.
2621 2017-09-12 Richard Biener <rguenther@suse.de>
2623 PR tree-optimization/82157
2624 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2625 stmts with side-effects.
2627 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2628 Alan Hayward <alan.hayward@arm.com>
2629 David Sherwood <david.sherwood@arm.com>
2631 * target.def (hard_regno_nregs): New hook.
2632 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2633 * targhooks.h (default_hard_regno_nregs): Declare.
2634 * targhooks.c (default_hard_regno_nregs): New function.
2635 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2636 (TARGET_HARD_REGNO_NREGS): ...this hook.
2637 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2638 (CLASS_MAX_NREGS): Likewise.
2639 * doc/tm.texi: Regenerate.
2640 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2641 instead of HARD_REGNO_NREGS.
2642 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2643 HARD_REGNO_NREGS in the comment.
2644 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2645 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2646 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2647 Return an unsigned int.
2648 (TARGET_HARD_REGNO_NREGS): Redefine.
2649 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2650 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2651 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2652 (arc_hard_regno_nregs): New function.
2653 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2654 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2655 (arm_hard_regno_nregs): New function.
2656 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2657 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2658 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2659 (TARGET_HARD_REGNO_NREGS): Redefine.
2660 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2661 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2662 (HARD_REGNO_NREGS): Delete.
2663 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2664 (cr16_hard_regno_nregs): New function.
2665 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2666 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2667 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2668 (cris_hard_regno_nregs): New function.
2669 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2670 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2671 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2672 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2673 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2674 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2675 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2676 (frv_hard_regno_nregs): Make static. Take and return an
2678 (frv_class_max_nregs): Remove outdated copy of documentation.
2679 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2680 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2681 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2682 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2683 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2684 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2685 (TARGET_HARD_REGNO_NREGS): Redefine.
2686 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2687 (CLASS_MAX_NREGS): Update comment.
2688 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2689 (ia64_hard_regno_nregs): New function.
2690 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2691 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2692 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2693 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2694 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2696 (m32c_hard_regno_nregs): Likewise. Make static.
2697 (TARGET_HARD_REGNO_NREGS): Redefine.
2698 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2699 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2700 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2701 (m68k_hard_regno_nregs): New function.
2702 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2703 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2704 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2705 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2706 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2707 Take and return an unsigned int.
2708 (TARGET_HARD_REGNO_NREGS): Redefine.
2709 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2710 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2711 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2712 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2713 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2714 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2715 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2716 (msp430_hard_regno_nregs): Make static. Take and return an
2718 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2719 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2720 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2721 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2722 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2723 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2724 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2725 (TARGET_HARD_REGNO_NREGS): Redefine.
2726 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2727 (PA_HARD_REGNO_NREGS): ...this.
2728 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2729 (PA_HARD_REGNO_NREGS): ...this.
2730 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2731 (pa_hard_regno_nregs): New function.
2732 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2733 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2734 (pdp11_hard_regno_nregs): New function.
2735 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2736 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2737 (rs6000_hard_regno_nregs_hook): New function.
2738 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2739 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2740 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2741 Take and return an unsigned int. Move earlier in file.
2742 (TARGET_HARD_REGNO_NREGS): Redefine.
2743 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2744 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2745 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2746 (rl78_hard_regno_nregs): Make static. Take and return an
2748 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2749 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2750 (rs6000_hard_regno_nregs_hook): New function.
2751 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2752 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2753 (TARGET_HARD_REGNO_NREGS): Redefine.
2754 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2755 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2756 instead of HARD_REGNO_NREGS.
2757 (s390_hard_regno_nregs): New function.
2758 (s390_hard_regno_mode_ok): Add comment from s390.h.
2759 (TARGET_HARD_REGNO_NREGS): Redefine.
2760 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2761 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2762 (sh_hard_regno_nregs): New function.
2763 (sh_pass_in_reg_p): Use it.
2764 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2765 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2766 (sparc_hard_regno_nregs): New function.
2767 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2768 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2769 (spu_function_arg_advance): Use it, supplying a valid register number.
2770 (TARGET_HARD_REGNO_NREGS): Redefine.
2771 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2772 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2773 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2774 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2775 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2776 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2777 (CLASS_MAX_NREGS): Remove copy of old documentation.
2778 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2779 (visium_hard_regno_nregs): New function.
2780 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2781 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2782 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2783 xtensa_hard_regno_nregs): New function.
2784 * system.h (HARD_REGNO_NREGS): Poison.
2786 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2788 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2789 hard_regno_nregs instead of HARD_REGNO_NREGS.
2790 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2791 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2792 (c6x_expand_epilogue): Likewise.
2793 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2794 (frv_read_iacc_argument): Likewise.
2795 * config/sh/sh.c: Include regs.h.
2796 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2797 (regs_used): Likewise.
2798 (output_stack_adjust): Likewise.
2799 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2800 * expmed.c: Include regs.h.
2801 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2802 * ree.c: Include regs.h.
2803 (combine_reaching_defs): Use hard_regno_nregs instead of
2805 (add_removable_extension): Likewise.
2807 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2809 * regs.h (hard_regno_nregs): Turn into a function.
2810 (end_hard_regno): Update accordingly.
2811 * caller-save.c (setup_save_areas): Likewise.
2812 (save_call_clobbered_regs): Likewise.
2813 (replace_reg_with_saved_mem): Likewise.
2814 (insert_restore): Likewise.
2815 (insert_save): Likewise.
2816 * combine.c (can_change_dest_mode): Likewise.
2817 (move_deaths): Likewise.
2818 (distribute_notes): Likewise.
2819 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2820 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2821 (rs6000_split_multireg_move): Likewise.
2822 (rs6000_register_move_cost): Likewise.
2823 (rs6000_memory_move_cost): Likewise.
2824 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2825 (rs6000_split_multireg_move): Likewise.
2826 (rs6000_register_move_cost): Likewise.
2827 (rs6000_memory_move_cost): Likewise.
2828 * cselib.c (cselib_reset_table): Likewise.
2829 (cselib_lookup_1): Likewise.
2830 * emit-rtl.c (set_mode_and_regno): Likewise.
2831 * function.c (aggregate_value_p): Likewise.
2832 * ira-color.c (setup_profitable_hard_regs): Likewise.
2833 (check_hard_reg_p): Likewise.
2834 (calculate_saved_nregs): Likewise.
2835 (assign_hard_reg): Likewise.
2836 (improve_allocation): Likewise.
2837 (calculate_spill_cost): Likewise.
2838 * ira-emit.c (modify_move_list): Likewise.
2839 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2840 (ira_hard_reg_in_set_p): Likewise.
2841 * ira.c (setup_reg_mode_hard_regset): Likewise.
2842 (clarify_prohibited_class_mode_regs): Likewise.
2843 (check_allocation): Likewise.
2844 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2845 (lra_setup_reg_renumber): Likewise.
2846 (setup_try_hard_regno_pseudos): Likewise.
2847 (spill_for): Likewise.
2848 (assign_hard_regno): Likewise.
2849 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2850 * lra-constraints.c (in_class_p): Likewise.
2851 (lra_constraint_offset): Likewise.
2852 (simplify_operand_subreg): Likewise.
2853 (lra_constraints): Likewise.
2854 (split_reg): Likewise.
2855 (split_if_necessary): Likewise.
2856 (invariant_p): Likewise.
2857 (inherit_in_ebb): Likewise.
2858 * lra-lives.c (process_bb_lives): Likewise.
2859 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2860 (get_hard_regs): Likewise.
2861 (do_remat): Likewise.
2862 * lra-spills.c (assign_spill_hard_regs): Likewise.
2863 * mode-switching.c (create_pre_exit): Likewise.
2864 * postreload.c (reload_combine_recognize_pattern): Likewise.
2865 * recog.c (peep2_find_free_register): Likewise.
2866 * regcprop.c (kill_value_regno): Likewise.
2867 (set_value_regno): Likewise.
2868 (copy_value): Likewise.
2869 (maybe_mode_change): Likewise.
2870 (find_oldest_value_reg): Likewise.
2871 (copyprop_hardreg_forward_1): Likewise.
2872 * regrename.c (check_new_reg_p): Likewise.
2873 (regrename_do_replace): Likewise.
2874 * reload.c (push_reload): Likewise.
2875 (combine_reloads): Likewise.
2876 (find_dummy_reload): Likewise.
2877 (operands_match_p): Likewise.
2878 (find_reloads): Likewise.
2879 (find_equiv_reg): Likewise.
2880 (reload_adjust_reg_for_mode): Likewise.
2881 * reload1.c (count_pseudo): Likewise.
2882 (count_spilled_pseudo): Likewise.
2883 (find_reg): Likewise.
2884 (clear_reload_reg_in_use): Likewise.
2885 (free_for_value_p): Likewise.
2886 (allocate_reload_reg): Likewise.
2887 (choose_reload_regs): Likewise.
2888 (reload_adjust_reg_for_temp): Likewise.
2889 (emit_reload_insns): Likewise.
2890 (delete_output_reload): Likewise.
2891 * rtlanal.c (subreg_get_info): Likewise.
2892 * sched-deps.c (sched_analyze_reg): Likewise.
2893 * sel-sched.c (init_regs_for_mode): Likewise.
2894 (mark_unavailable_hard_regs): Likewise.
2895 (choose_best_reg_1): Likewise.
2896 (verify_target_availability): Likewise.
2897 * valtrack.c (dead_debug_insert_temp): Likewise.
2898 * var-tracking.c (track_loc_p): Likewise.
2899 (emit_note_insn_var_location): Likewise.
2900 * varasm.c (make_decl_rtl): Likewise.
2901 * reginfo.c (choose_hard_reg_mode): Likewise.
2902 (init_reg_modes_target): Refer directly to
2903 this_target_regs->x_hard_regno_nregs.
2905 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2907 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2908 instead of hard_regno_nregs.
2910 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2912 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2913 end_hard_regno instead of hard_regno_nregs.
2914 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2915 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2916 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2917 * ira-color.c (improve_allocation): Likewise.
2918 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2919 * lra-lives.c (mark_regno_live): Likewise.
2920 (mark_regno_dead): Likewise.
2921 * lra-remat.c (operand_to_remat): Likewise.
2922 * lra.c (collect_non_operand_hard_regs): Likewise.
2923 * postreload.c (reload_combine_note_store): Likewise.
2924 (move2add_valid_value_p): Likewise.
2925 * reload.c (regno_clobbered_p): Likewise.
2927 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2929 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2931 * config/v850/v850.c (v850_reorg): Likewise.
2932 * reload.c (refers_to_regno_for_reload_p): Likewise.
2933 (find_equiv_reg): Likewise.
2934 * reload1.c (reload_reg_reaches_end_p): Likewise.
2936 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2938 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2940 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2941 * config/arm/arm.c (output_move_neon): Likewise.
2942 (arm_attr_length_move_neon): Likewise.
2943 (neon_split_vcombine): Likewise.
2944 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2945 (c6x_mark_reg_written): Likewise.
2946 (c6x_dwarf_register_span): Likewise.
2947 * config/i386/i386.c (ix86_save_reg): Likewise.
2948 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2949 (rws_access_reg): Likewise.
2950 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2951 * mode-switching.c (create_pre_exit): Likewise.
2952 * ree.c (combine_reaching_defs): Likewise.
2953 (add_removable_extension): Likewise.
2954 * regcprop.c (find_oldest_value_reg): Likewise.
2955 (copyprop_hardreg_forward_1): Likewise.
2956 * reload.c (reload_inner_reg_of_subreg): Likewise.
2957 (push_reload): Likewise.
2958 (combine_reloads): Likewise.
2959 (find_dummy_reload): Likewise.
2960 (reload_adjust_reg_for_mode): Likewise.
2961 * reload1.c (find_reload_regs): Likewise.
2962 (forget_old_reloads_1): Likewise.
2963 (reload_reg_free_for_value_p): Likewise.
2964 (reload_adjust_reg_for_temp): Likewise.
2965 (emit_reload_insns): Likewise.
2966 (delete_output_reload): Likewise.
2967 * sel-sched.c (choose_best_reg_1): Likewise.
2968 (choose_best_pseudo_reg): Likewise.
2970 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2971 Alan Hayward <alan.hayward@arm.com>
2972 David Sherwood <david.sherwood@arm.com>
2974 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2975 * target.def (slow_unaligned_access): New hook.
2976 * targhooks.h (default_slow_unaligned_access): Declare.
2977 * targhooks.c (default_slow_unaligned_access): New function.
2978 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2979 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2980 * doc/tm.texi: Regenerate.
2981 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2982 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2983 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2985 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2986 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2988 (rs6000_slow_unaligned_access): New function.
2989 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2990 (expand_block_compare): Likewise.
2991 (expand_strn_compare): Likewise.
2992 (rs6000_rtx_costs): Likewise.
2993 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2994 (riscv_slow_unaligned_access): Likewise.
2995 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2996 (riscv_slow_unaligned_access_p): ...this and make static.
2997 (riscv_option_override): Update accordingly.
2998 (riscv_slow_unaligned_access): New function.
2999 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3000 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3001 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3002 (rs6000_slow_unaligned_access): New function.
3003 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3004 (rs6000_rtx_costs): Likewise.
3005 * config/rs6000/rs6000-string.c (expand_block_compare)
3006 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3007 of SLOW_UNALIGNED_ACCESS.
3008 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3009 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3010 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3011 of SLOW_UNALIGNED_ACCESS.
3012 * expmed.c (simple_mem_bitfield_p): Likewise.
3013 * expr.c (alignment_for_piecewise_move): Likewise.
3014 (emit_group_load_1): Likewise.
3015 (emit_group_store): Likewise.
3016 (copy_blkmode_from_reg): Likewise.
3017 (emit_push_insn): Likewise.
3018 (expand_assignment): Likewise.
3019 (store_field): Likewise.
3020 (expand_expr_real_1): Likewise.
3021 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3022 * lra-constraints.c (simplify_operand_subreg): Likewise.
3023 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3024 * gimple-ssa-store-merging.c: Likewise in block comment at start
3026 * tree-ssa-strlen.c: Include target.h.
3027 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3028 of SLOW_UNALIGNED_ACCESS.
3029 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3031 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3033 PR rtl-optimization/82185
3034 * expmed.c (emit_store_flag_int): Only test tem if it has been
3037 2017-09-12 Richard Biener <rguenther@suse.de>
3040 * match.pd ((FTYPE) N CMP CST): Fix typo.
3042 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3044 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3046 (mips_near_type_p): Add 'short_call' attribute as a synonym
3048 * doc/extend.texi (short_call): Document new function attribute.
3050 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3053 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3054 assertion check that in the condition.
3055 (get_atomic_generic_size): Likewise. Before testing if parameter
3056 has pointer type, if it has array type, call for C++
3057 default_conversion to perform array-to-pointer conversion.
3059 2017-09-12 Richard Biener <rguenther@suse.de>
3061 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3062 for operations we cannot scalarize.
3064 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3066 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3067 vectors heap vectors. Clean up comments.
3068 Make visited_bbs a reference.
3069 (profitable_jump_thread_path): Make GC
3070 vectors heap vectors. Clean up comments.
3072 (convert_and_register_jump_thread_path): Make GC vectors heap
3074 (check_subpath_and_update_thread_path): Same. Clean up comments.
3075 Make visited_bbs a reference.
3076 (handle_phi): Abstract common code to to
3077 register_jump_thread_path_if_profitable.
3078 Rename VAR_BB to DEF_BB.
3080 Make GC vectors heap vectors.
3081 Make visited_bbs a reference.
3082 (handle_assignment): Same.
3083 (register_jump_thread_path_if_profitable): New.
3084 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3086 Make GC vectors heap vectors. Clean up comments.
3087 Make visited_bbs a reference.
3088 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3089 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3092 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3095 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3096 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3098 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3100 Revert r251800 and r251799.
3102 2017-09-11 Martin Jambor <mjambor@suse.cz>
3105 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3106 arguments in advance.
3107 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3108 use it to find predecessor edges.
3109 (naive_outof_ssa): Collect vector of predecessors.
3111 2017-09-08 Jason Merrill <jason@redhat.com>
3113 PR c++/70029 - ICE with ref-qualifier and -flto
3114 * langhooks.h (struct lang_hooks_for_types): Add
3115 copy_lang_qualifiers.
3116 * attribs.c (build_type_attribute_qual_variant): Use it.
3117 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3119 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3120 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3122 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3125 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3126 (*mulsi3_sp64): New instruction.
3127 (mulsi3): New expander.
3129 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3131 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3133 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3135 * sancov.c: Include memmodel.h.
3137 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3140 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3143 2017-09-07 Carl Love <cel@us.ibm.com>
3145 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3146 the sldi instruction.
3148 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3150 * sancov.c: Include tm_p.h.
3152 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3155 * output.h (switch_to_other_text_partition): New declaration.
3156 * varasm.c (switch_to_other_text_partition): New function.
3157 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3158 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3159 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3160 to the other text partition before emitting LCL label and switch back
3161 after emitting the word after it.
3163 2017-09-07 Richard Biener <rguenther@suse.de>
3165 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3166 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3167 critical edges here, after loop init.
3168 (pass_data_pre): Remove PROP_no_crit_edges flags.
3169 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3170 for valueization of call args to avoid leaking VN_TOP.
3171 (visit_use): Assert we do not visit default defs.
3172 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3173 Use error_mark_node to more easily detect leaking VN_TOP.
3174 All default-defs are varying, not VN_TOP. Mark them visited.
3175 (run_scc_vn): Make code match comment.
3177 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3179 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3180 OPTION_MASK_FLOAT128_KEYWORD.
3181 (POWERPC_MASKS): Likewise.
3182 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3183 support for the -mfloat128-type option, and make -mfloat128
3184 default on PowerPC Linux systems. Define or undefine
3185 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3186 Define __float128 to be __ieee128 if IEEE 128-bit support is
3187 enabled, or undefine it.
3188 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3189 Delete defining __FLOAT128_TYPE__.
3190 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3191 -mfloat128-type option and make -mfloat128 default on PowerPC
3193 (TARGET_FLOAT128_TYPE): Likewise.
3194 (-mfloat128-type): Likewise.
3195 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3196 Delete the -mfloat128-type option and make -mfloat128 default on
3197 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3198 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3199 128-bit floating point is enabled. Change tests from using
3200 -mfloat128-type to -mfloat128.
3201 (rs6000_mangle_type): Use the correct mangling for the __float128
3202 type even if normal long double is restricted to 64-bits.
3203 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3205 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3206 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3207 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3209 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3210 documentation for -mfloat128.
3212 2017-09-06 Olivier Hainque <hainque@adacore.com>
3214 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3216 2017-09-06 Wish Wu <wishwu007@gmail.com>
3217 Jakub Jelinek <jakub@redhat.com>
3219 * asan.c (initialize_sanitizer_builtins): Add
3220 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3221 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3222 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3223 BT_FN_VOID_UINT64_PTR variables.
3224 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3225 (BT_FN_VOID_UINT16_UINT16): Likewise.
3226 (BT_FN_VOID_UINT32_UINT32): Likewise.
3227 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3228 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3229 (BT_FN_VOID_UINT64_PTR): Likewise.
3230 * common.opt (flag_sanitize_coverage): New variable.
3231 (fsanitize-coverage=trace-pc): Remove.
3232 (fsanitize-coverage=): Add.
3233 * flag-types.h (enum sanitize_coverage_code): New enum.
3234 * fold-const.c (fold_range_test): Disable non-short-circuit
3235 optimization if flag_sanitize_coverage.
3236 (fold_truth_andor): Likewise.
3237 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3238 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3239 (coverage_sanitizer_opts): New array.
3240 (get_closest_sanitizer_option): Add OPTS argument, handle also
3241 OPT_fsanitize_coverage_.
3242 (parse_sanitizer_options): Adjusted to also handle
3243 OPT_fsanitize_coverage_.
3244 (common_handle_option): Add OPT_fsanitize_coverage_.
3245 * sancov.c (instrument_comparison, instrument_switch): New function.
3246 (sancov_pass): Add trace-cmp support.
3247 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3248 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3249 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3250 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3251 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3252 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3253 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3254 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3255 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3256 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3258 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3260 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3261 error. Only quit immediately if parsing is complete.
3262 (BEGIN): Initialize fatal_err and parse_done.
3263 (begin fpu, end fpu): Check number of arguments.
3264 (begin arch, end arch): Likewise.
3265 (begin cpu, end cpu): Likewise.
3266 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3267 (optalias): Likewise.
3269 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3271 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3272 * config/arm/arm-isa.h: Delete. Move definitions to ...
3273 * arm-cpus.in: ... here. Use new feature and fgroup values.
3274 * config/arm/arm.c (arm_option_override): Use lower case for feature
3276 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3277 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3278 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3280 (print_isa_bits_for): New function.
3281 (gen_isa): New function.
3282 (gen_comm_data): Use print_isa_bits_for.
3283 (define feature): New keyword.
3284 (define fgroup): New keyword.
3285 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3286 (arm-isa.h): Add rule to generate file.
3287 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3288 case for feature bit names.
3290 2017-09-06 Richard Biener <rguenther@suse.de>
3292 * tree-ssa-pre.c (NECESSARY): Remove.
3293 (create_expression_by_pieces): Do not touch pass-local flags.
3294 (insert_into_preds_of_block): Likewise.
3295 (do_pre_regular_insertion): Likewise.
3296 (eliminate_insert): Likewise.
3297 (eliminate_dom_walker::before_dom_children): Likewise.
3298 (fini_eliminate): Do not look at inserted_exprs.
3299 (mark_operand_necessary): Remove.
3300 (remove_dead_inserted_code): Replace with simple work-list
3301 algorithm based on inserted_exprs and SSA uses.
3302 (pass_pre::execute): Re-order fini_eliminate and
3303 remove_dead_inserted_code.
3305 2017-09-06 Olivier Hainque <hainque@adacore.com>
3307 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3308 for VxWorks 7. Adjust surrounding comments.
3310 2017-09-06 Richard Biener <rguenther@suse.de>
3312 * gimple-ssa-strength-reduction.c
3313 (find_candidates_dom_walker::before_dom_children): Also allow
3316 2017-09-06 Richard Biener <rguenther@suse.de>
3318 PR tree-optimization/82108
3319 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3320 for gap in the non-permutation SLP case.
3322 2017-09-06 Martin Jambor <mjambor@suse.cz>
3324 PR tree-optimization/82078
3325 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3326 add_access_to_work_queue...
3327 (build_accesses_from_assign): ...here.
3328 (propagate_all_subaccesses): Make sure racc is the group
3329 representative, if there is one.
3331 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3334 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3337 2017-09-06 Richard Biener <rguenther@suse.de>
3339 * gimple-ssa-strength-reduction.c
3340 (find_candidates_dom_walker::before_doom_children): Use a
3341 type and not a mode check.
3343 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3346 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3348 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3350 2017-09-05 Jeff Law <law@redhat.com>
3352 PR tree-optimization/64910
3353 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3354 cases where we have 3 or more operands.
3356 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3359 * omp-low.c (lower_omp_for): Recompute tree invariant if
3360 gimple_omp_for_initial/final is ADDR_EXPR.
3363 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3364 into gimple val before gimplification fo the COND_EXPR.
3366 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3368 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3369 REGION_COPY argument.
3370 (thread_through_all_blocks): Remove unused argument to
3371 duplicate_thread_path.
3373 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3374 Alan Hayward <alan.hayward@arm.com>
3375 David Sherwood <david.sherwood@arm.com>
3377 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3378 Take a scalar_mode rather than a machine_mode.
3379 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3380 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3381 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3382 (aarch64_gen_adjusted_ldpstp): Likewise.
3383 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3385 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3386 Alan Hayward <alan.hayward@arm.com>
3387 David Sherwood <david.sherwood@arm.com>
3389 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3390 Take a scalar_int_mode instead of a machine_mode.
3391 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3392 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3393 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3394 (aarch64_simd_attr_length_rglist): Delete.
3395 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3396 a scalar_int_mode instead of a machine_mode.
3397 (aarch64_add_offset): Likewise.
3398 (aarch64_internal_mov_immediate): Likewise
3399 (aarch64_add_constant_internal): Likewise.
3400 (aarch64_add_constant): Likewise.
3401 (aarch64_movw_imm): Likewise.
3402 (aarch64_rtx_arith_op_extract_p): Likewise.
3403 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3404 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3405 Remove assert that the mode isn't a vector.
3406 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3407 (aarch64_expand_mov_immediate): Update calls after above changes.
3408 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3409 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3410 (aarch64_move_imm): Likewise.
3411 (aarch64_can_const_movi_rtx_p): Likewise.
3412 (aarch64_strip_extend): Likewise.
3413 (aarch64_extr_rtx_p): Likewise.
3414 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3415 a CONST_INT when the mode parameter is VOIDmode.
3416 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3418 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3420 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3421 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3422 (bitwise_type_for_mode): Update accordingly.
3424 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3426 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3427 * stor-layout.c (mode_for_size_tree): Likewise.
3428 (mode_for_array): Update accordingly.
3429 (layout_decl): Likewise.
3430 (compute_record_mode): Likewise. Only set the mode once.
3432 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3434 * target.def (get_mask_mode): Change return type to opt_mode.
3436 * doc/tm.texi: Regenerate.
3437 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3438 * targhooks.c (default_get_mask_mode): Likewise.
3439 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3440 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3441 targetm.get_mask_mode.
3442 * tree.c (build_truth_vector_type): Likewise.
3444 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3446 * machmode.h (mode_for_vector): Return an opt_mode.
3447 * stor-layout.c (mode_for_vector): Likewise.
3448 (mode_for_int_vector): Update accordingly.
3449 (layout_type): Likewise.
3450 * config/i386/i386.c (emit_memmov): Likewise.
3451 (ix86_expand_set_or_movmem): Likewise.
3452 (ix86_expand_vector_init): Likewise.
3453 (ix86_get_mask_mode): Likewise.
3454 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3456 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3457 * expmed.c (extract_bit_field_1): Likewise.
3458 * expr.c (expand_expr_real_2): Likewise.
3459 * optabs-query.c (can_vec_perm_p): Likewise.
3460 (can_vec_mask_load_store_p): Likewise.
3461 * optabs.c (expand_vec_perm): Likewise.
3462 * targhooks.c (default_get_mask_mode): Likewise.
3463 * tree-vect-stmts.c (vectorizable_store): Likewise.
3464 (vectorizable_load): Likewise.
3465 (get_vectype_for_scalar_type_and_size): Likewise.
3467 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3469 * machmode.h (mode_for_int_vector): New function.
3470 * stor-layout.c (mode_for_int_vector): Likewise.
3471 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3472 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3473 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3474 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3475 (s390_expand_vcond): Likewise.
3477 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3479 * machmode.h (opt_machine_mode): New type.
3480 (opt_mode<T>): Allow construction from anything that can be
3482 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3483 (mode_for_size): Return an opt_machine_mode.
3484 * stor-layout.c (mode_for_size): Likewise.
3485 (mode_for_size_tree): Update call accordingly.
3486 (bitwise_mode_for_mode): Likewise.
3487 (make_fract_type): Likewise.
3488 (make_accum_type): Likewise.
3489 * caller-save.c (replace_reg_with_saved_mem): Update call
3491 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3492 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3493 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3494 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3495 * expmed.c (extract_bit_field_1): Likewise.
3496 * reload.c (get_secondary_mem): Likewise.
3497 * varasm.c (assemble_integer): Likewise.
3498 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3501 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3503 * machmode.h (decimal_float_mode_for_size): New function.
3504 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3505 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3506 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3507 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3508 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3509 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3511 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3513 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3514 (get_builtin_sync_mode): Likewise.
3515 (expand_ifn_atomic_compare_exchange): Likewise.
3516 (expand_builtin_atomic_clear): Likewise.
3517 (expand_builtin_atomic_test_and_set): Likewise.
3518 (fold_builtin_atomic_always_lock_free): Likewise.
3519 * calls.c (compute_argument_addresses): Likewise.
3520 (emit_library_call_value_1): Likewise.
3521 (store_one_arg): Likewise.
3522 * combine.c (combine_instructions): Likewise.
3523 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3524 * config/arm/arm.c (arm_function_value): Likewise.
3525 (aapcs_allocate_return_reg): Likewise.
3526 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3527 * config/i386/i386.c (construct_container): Likewise.
3528 (ix86_gimplify_va_arg): Likewise.
3529 (ix86_expand_sse_cmp): Likewise.
3530 (emit_memmov): Likewise.
3531 (emit_memset): Likewise.
3532 (expand_small_movmem_or_setmem): Likewise.
3533 (ix86_expand_pextr): Likewise.
3534 (ix86_expand_pinsr): Likewise.
3535 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3536 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3538 * config/mips/mips.c (mips_function_value_1) Likewise.
3539 (mips_block_move_straight): Likewise.
3540 (mips_expand_ins_as_unaligned_store): Likewise.
3541 * config/powerpcspe/powerpcspe.c
3542 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3543 (rs6000_darwin64_record_arg_flush): Likewise.
3544 * config/rs6000/rs6000.c
3545 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3546 (rs6000_darwin64_record_arg_flush): Likewise.
3547 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3548 (sparc_function_value_1): Likewise.
3549 * config/spu/spu.c (adjust_operand): Likewise.
3550 (spu_emit_branch_or_set): Likewise.
3551 (arith_immediate_p): Likewise.
3552 * emit-rtl.c (gen_lowpart_common): Likewise.
3553 * expr.c (expand_expr_real_1): Likewise.
3554 * function.c (assign_parm_setup_block): Likewise.
3555 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3556 * reload1.c (alter_reg): Likewise.
3557 * stor-layout.c (mode_for_vector): Likewise.
3558 (layout_type): Likewise.
3560 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3562 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3563 (spu_convert_move): Likewise.
3564 * lower-subreg.c (resolve_simple_move): Likewise.
3566 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3569 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3570 define_insn to a define_expand.
3571 (altivec_vsum2sws_direct): New define_insn.
3572 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3574 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3576 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3579 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3584 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3585 realign stack if stack alignment needed is less than incoming
3588 2017-09-05 Marek Polacek <polacek@redhat.com>
3591 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3594 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3596 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3598 2017-09-05 Richard Biener <rguenther@suse.de>
3600 PR tree-optimization/82084
3601 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3603 2017-09-05 Richard Biener <rguenther@suse.de>
3605 PR tree-optimization/82102
3606 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3608 2017-09-05 Martin Liska <mliska@suse.cz>
3610 PR tree-optimization/82032
3611 * tree-cfg.c (generate_range_test): New function.
3612 * tree-cfg.h (generate_range_test): Declared here.
3613 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3614 (cleanup_control_expr_graph): Use it.
3615 * tree-switch-conversion.c (try_switch_expansion): Remove
3617 (emit_case_nodes): Use generate_range_test.
3619 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3622 * config/i386/i386.md (*<btsc><mode>_mask): Add
3623 TARGET_USE_BT to insn constraint.
3624 (*btr<mode>_mask): Ditto.
3626 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3628 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3629 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3631 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3634 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3635 TARGET_NEON and TARGET_IWMMXT.
3636 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3637 TARGET_NEON and TARGET_IWMMXT.
3638 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3640 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3642 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3643 (ix86_rewrite_tls_address): Ditto.
3644 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3645 (ix86_rewrite_tls_address_1): Ditto.
3646 (ix86_rewrite_tls_address): Ditto.
3647 * config/i386/predicates.md (tls_address_pattern): New predicate.
3648 * config/i386/i386.md (TLS address splitter): New splitter.
3650 2017-09-04 Richard Biener <rguenther@suse.de>
3652 PR tree-optimization/82084
3653 * fold-const.h (can_native_encode_string_p): Declare.
3654 * fold-const.c (can_native_encode_string_p): Factor out from ...
3655 (native_encode_string): ... here.
3656 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3657 vectorizing stores from constants we later cannot handle.
3659 2017-09-04 Marek Polacek <polacek@redhat.com>
3662 * doc/invoke.texi: Update -Wtautological-compare documentation.
3664 2017-09-04 Jeff Law <law@redhat.com>
3666 PR tree-optimization/64910
3667 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3668 swap the first and last operand if the last is a constant.
3670 2017-09-04 Marek Polacek <polacek@redhat.com>
3673 * convert.c (do_narrow): When sanitizing signed integer overflows,
3674 bail out for signed types.
3675 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3677 2017-09-04 Richard Biener <rguenther@suse.de>
3679 PR tree-optimization/82060
3680 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3681 Move devirtualization after stmt folding and before EH/AB/noreturn
3682 cleanup to get the stmt refs canonicalized. Use a bool instead
3683 of gimple_modified_p since that doesn't work for NOPs. Schedule
3684 NOPs generated by folding for removal.
3686 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3687 Alan Hayward <alan.hayward@arm.com>
3688 David Sherwood <david.sherwood@arm.com>
3690 * coretypes.h (pad_direction): New enum.
3691 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3692 (FUNCTION_ARG_PADDING): Likewise.
3693 * target.def (function_arg_padding): New hook.
3694 * targhooks.h (default_function_arg_padding): Declare.
3695 * targhooks.c (default_function_arg_padding): New function.
3696 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3697 (TARGET_FUNCTION_ARG_PADDING): ...this.
3698 * doc/tm.texi: Regenerate.
3699 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3700 instead of direction.
3701 (compute_argument_addresses): Likewise.
3702 (load_register_parameters): Likewise.
3703 (emit_library_call_value_1): Likewise.
3704 (store_one_arg): Use targetm.calls.function_arg_padding instead
3705 of FUNCTION_ARG_PADDING.
3706 (must_pass_in_stack_var_size_or_pad): Likewise.
3707 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3708 (emit_group_store): Likewise.
3709 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3710 instead of FUNCTION_ARG_PADDING.
3711 (emit_push_insn): Likewise, and propagate enum change throughout
3713 * function.h (direction): Delete.
3714 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3716 * function.c (assign_parm_find_stack_rtl): Likewise.
3717 (assign_parm_setup_block_p): Likewise.
3718 (assign_parm_setup_block): Likewise.
3719 (gimplify_parameters): Likewise.
3720 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3721 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3723 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3724 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3725 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3726 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3727 (aarch64_function_arg_padding): ...this new function.
3728 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3729 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3730 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3731 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3732 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3733 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3734 (arm_pad_arg_upward): Replace with...
3735 (arm_function_arg_padding): ...this new function.
3736 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3738 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3739 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3740 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3741 (ia64_hpux_function_arg_padding): Replace with...
3742 (ia64_function_arg_padding): ...this new function. Use pad_direction
3743 instead of direction. Check for TARGET_HPUX.
3744 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3745 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3746 (iq2000_function_arg_padding): New function.
3747 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3748 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3749 (mips_function_arg_padding): ...this new function.
3750 (mips_pad_reg_upward): Update accordingly.
3751 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3752 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3753 targetm.calls.function_arg_padding.
3754 (FUNCTION_ARG_PADDING): Delete.
3755 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3756 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3757 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3758 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3759 (nios2_block_reg_padding): Return pad_direction instead of direction.
3760 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3761 instead of direction.
3762 (nios2_function_arg_padding): Likewise. Make static.
3763 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3764 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3765 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3766 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3767 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3768 (pa_function_arg_padding): Make static. Return pad_direction instead
3770 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3771 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3772 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3773 instead of direction. Use targetm.calls.function_arg_padding.
3774 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3775 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3776 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3777 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3778 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3780 (function_arg_padding): Rename to...
3781 (rs6000_function_arg_padding): ...this. Make static. Return
3782 pad_direction instead of direction.
3783 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3784 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3785 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3786 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3787 instead of direction. Use targetm.calls.function_arg_padding.
3788 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3789 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3790 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3791 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3792 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3793 (function_arg_padding): Rename to...
3794 (rs6000_function_arg_padding): ...this. Make static. Return
3795 pad_direction instead of direction.
3796 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3797 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3798 * config/s390/s390.c (s390_function_arg_padding): New function.
3799 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3800 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3801 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3802 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3803 (function_arg_padding): Rename to...
3804 (sparc_function_arg_padding): ...this. Make static. Return
3805 pad_direction instead of direction.
3806 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3807 * config/spu/spu.c (spu_function_arg_padding): New function.
3808 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3809 * system.h (FUNCTION_ARG_PADDING): Poison.
3811 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3812 Alan Hayward <alan.hayward@arm.com>
3813 David Sherwood <david.sherwood@arm.com>
3815 * target.def (modes_tieable_p): New hook.
3816 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3817 (TARGET_MODES_TIEABLE_P): ...this.
3818 * doc/tm.texi.in: Regenerate.
3819 * hooks.h (hook_bool_mode_mode_true): Declare.
3820 * hooks.c (hook_bool_mode_mode_true): New function.
3821 * combine.c (subst): Use targetm.modes_tieable_p instead of
3823 * dse.c (find_shift_sequence): Likewise.
3824 * expmed.c (extract_low_bits): Likewise.
3825 * lower-subreg.c: Include target.h.
3826 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3828 * rtlanal.c (rtx_cost): Likewise.
3829 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3830 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3831 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3832 (TARGET_MODES_TIEABLE_P): Redefine.
3833 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3834 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3835 (TARGET_MODES_TIEABLE_P): Redefine.
3836 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3837 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3838 (arc_modes_tieable_p): New function.
3839 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3840 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3841 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3842 (arm_modes_tieable_p): Make static.
3843 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3844 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3845 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3846 (TARGET_MODES_TIEABLE_P): Redefine.
3847 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3848 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3849 (TARGET_MODES_TIEABLE_P): Redefine.
3850 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3851 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3852 (cr16_modes_tieable_p): New function.
3853 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3854 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3855 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3856 (TRULY_NOOP_TRUNCATION): Update comment.
3857 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3858 (TRULY_NOOP_TRUNCATION): Update comment.
3859 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3860 (frv_modes_tieable_p): New function.
3861 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3862 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3863 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3864 (TARGET_MODES_TIEABLE_P): Redefine.
3865 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3866 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3867 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3868 (TARGET_MODES_TIEABLE_P): Redefine.
3869 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3870 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3871 (ia64_modes_tieable_p): New function.
3872 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3873 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3874 (iq2000_modes_tieable_p): New function.
3875 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3876 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3877 (lm32_modes_tieable_p): New function.
3878 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3879 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3880 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3881 (TARGET_MODES_TIEABLE_P): Redefine.
3882 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3883 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3884 (m32r_modes_tieable_p): New function.
3885 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3886 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3887 (m68k_modes_tieable_p): New function.
3888 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3889 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3890 (mcore_modes_tieable_p): New function.
3891 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3892 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3894 (TARGET_MODES_TIEABLE_P): Redefine.
3895 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3896 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3897 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3898 (TARGET_MODES_TIEABLE_P): Redefine.
3899 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3900 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3901 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3902 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3903 (mn10300_modes_tieable_p): ...this and make static.
3904 (TARGET_MODES_TIEABLE_P): Redefine.
3905 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3906 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3907 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3908 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3909 (msp430_modes_tieable_p): Make static.
3910 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3911 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3912 (TARGET_MODES_TIEABLE_P): Redefine.
3913 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3914 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3915 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3916 (TARGET_MODES_TIEABLE_P): Redefine.
3917 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3918 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3919 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3920 (TARGET_MODES_TIEABLE_P): Redefine.
3921 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3922 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3923 (pdp11_modes_tieable_p): New function.
3924 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3925 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3926 (rs6000_modes_tieable_p): New function.
3927 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3928 * config/powerpcspe/powerpcspe.md: Update comment.
3929 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3930 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3931 (TARGET_MODES_TIEABLE_P): Redefine.
3932 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3933 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3934 (rl78_modes_tieable_p): New function.
3935 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3936 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3937 (rs6000_modes_tieable_p): New function.
3938 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3939 * config/rs6000/rs6000.md: Update comment.
3940 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3941 * config/rx/rx.c (rx_modes_tieable_p): New function.
3942 (TARGET_MODES_TIEABLE_P): Redefine.
3943 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3944 * config/s390/s390.c (s390_modes_tieable_p): New function.
3945 (TARGET_MODES_TIEABLE_P): Redefine.
3946 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3947 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3948 (sh_modes_tieable_p): New function.
3949 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3950 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3951 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3952 (sparc_modes_tieable_p): Make static.
3953 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3954 * config/spu/spu.c (spu_modes_tieable_p): New function.
3955 (TARGET_MODES_TIEABLE_P): Redefine.
3956 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3957 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3958 (TARGET_MODES_TIEABLE_P): Redefine.
3959 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3960 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3961 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3962 * config/v850/v850.c (v850_modes_tieable_p): New function.
3963 (TARGET_MODES_TIEABLE_P): Redefine.
3964 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3965 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3966 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3967 (visium_modes_tieable_p): New function.
3968 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3969 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3970 (xtensa_modes_tieable_p): New function.
3971 * system.h (MODES_TIEABLE_P): Poison.
3973 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3974 Alan Hayward <alan.hayward@arm.com>
3975 David Sherwood <david.sherwood@arm.com>
3977 * target.def (hard_regno_mode_ok): New hook.
3978 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3979 (TARGET_HARD_REGNO_MODE_OK): ...this.
3980 * doc/tm.texi.in: Regenerate.
3981 * hooks.h (hook_bool_uint_mode_true): Declare.
3982 * hooks.c (hook_bool_uint_mode_true): New function.
3983 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3985 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3986 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3987 instead of HARD_REGNO_MODE_OK.
3988 * caller-save.c: Include target.h.
3989 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3991 * combine.c (can_combine_p): Likewise.
3992 (combinable_i3pat): Likewise.
3993 (can_change_dest_mode): Likewise.
3994 * expr.c (init_expr_target): Likewise.
3995 (convert_move): Likewise.
3996 (convert_modes): Likewise.
3997 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3998 (setup_prohibited_mode_move_regs): Likewise.
3999 * ira.h (target_ira): Likewise.
4000 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4001 * lra-constraints.c (process_alt_operands): Likewise.
4002 (split_reg): Likewise.
4003 * recog.c (peep2_find_free_register): Likewise.
4004 * ree.c (combine_reaching_defs): Likewise.
4005 * regcprop.c (maybe_mode_change): Likewise.
4006 * reginfo.c (init_reg_sets_1): Likewise.
4007 (choose_hard_reg_mode): Likewise.
4008 (simplifiable_subregs): Likewise.
4009 * regrename.c (check_new_reg_p): Likewise.
4010 * reload.c (find_valid_class): Likewise.
4011 (find_valid_class_1): Likewise.
4012 (reload_inner_reg_of_subreg): Likewise.
4013 (push_reload): Likewise.
4014 (combine_reloads): Likewise.
4015 (find_dummy_reload): Likewise.
4016 (find_reloads): Likewise.
4017 * reload1.c (find_reg): Likewise.
4018 (set_reload_reg): Likewise.
4019 (allocate_reload_reg): Likewise.
4020 (choose_reload_regs): Likewise.
4021 (reload_adjust_reg_for_temp): Likewise.
4022 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4023 (simplify_subreg_regno): Likewise.
4024 * sel-sched.c (init_regs_for_mode): Likewise.
4025 * varasm.c (make_decl_rtl): Likewise.
4026 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4027 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4029 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4030 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4031 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4032 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4033 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4034 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4035 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4036 (arc_mode_class): Delete.
4037 (HARD_REGNO_MODE_OK): Delete.
4038 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4039 (arc_hard_regno_mode_ok): Rename old array to...
4040 (arc_hard_regno_mode_ok_modes): ...this.
4041 (arc_conditional_register_usage): Update accordingly.
4042 (arc_mode_class): Make static.
4043 (arc_hard_regno_mode_ok): New function.
4044 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4045 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4046 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4047 (arm_hard_regno_mode_ok): Make static.
4048 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4050 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4051 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4052 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4054 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4055 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4056 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4057 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4058 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4059 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4060 * config/bfin/predicates.md (valid_reg_operand): Use
4061 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4062 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4063 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4064 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4065 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4066 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4067 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4068 (cr16_hard_regno_mode_ok): Make static and return a bool.
4069 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4070 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4071 (cris_hard_regno_mode_ok): New function.
4072 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4073 (epiphany_mode_class): Delete.
4074 (HARD_REGNO_MODE_OK): Delete.
4075 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4076 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4077 (hard_regno_mode_ok): Rename to...
4078 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4079 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4080 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4082 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4083 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4084 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4085 (frv_hard_regno_mode_ok): Make static and return a bool.
4086 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4088 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4089 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4090 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4091 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4093 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4094 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4095 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4096 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4098 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4099 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4100 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4101 (ia64_hard_regno_mode_ok): New function.
4102 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4103 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4104 (iq2000_hard_regno_mode_ok): New function.
4105 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4106 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4107 (lm32_hard_regno_mode_ok): New function.
4108 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4109 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4110 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4111 instead of HARD_REGNO_MODE_OK.
4112 (m32c_hard_regno_ok): Rename to...
4113 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4114 (m32c_cannot_change_mode_class): Update accordingly.
4115 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4116 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4117 (m32r_mode_class): Delete.
4118 (HARD_REGNO_MODE_OK): Delete.
4119 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4120 (m32r_hard_regno_mode_ok): Rename to...
4121 (m32r_hard_regno_modes): ...this.
4122 (m32r_mode_class): Make static.
4123 (m32r_hard_regno_mode_ok): New function.
4124 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4125 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4126 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4127 (m68k_hard_regno_mode_ok): Make static.
4128 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4129 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4130 (mcore_hard_regno_mode_ok): New function.
4131 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4132 (HARD_REGNO_MODE_OK): Delete.
4133 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4135 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4136 (microblaze_hard_regno_mode_ok): New function.
4137 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4138 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4139 (mips_hard_regno_mode_ok): Delete.
4140 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4141 (mips_hard_regno_mode_ok_p): ...this and make static.
4142 (mips_hard_regno_mode_ok_p): Rename to...
4143 (mips_hard_regno_mode_ok_uncached): ...this.
4144 (mips_hard_regno_mode_ok): New function.
4145 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4146 of HARD_REGNO_MODE_OK.
4147 (mips_option_override): Update after above name changes.
4148 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4149 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4150 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4151 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4152 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4154 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4155 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4156 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4157 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4158 (msp430_hard_regno_mode_ok): Make static and return a bool.
4159 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4160 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4161 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4163 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4164 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4165 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4166 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4167 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4168 (PA_HARD_REGNO_MODE_OK): ...this
4169 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4170 (PA_HARD_REGNO_MODE_OK): ...this.
4171 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4172 (pa_hard_regno_mode_ok): New function.
4173 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4174 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4175 (pdp11_hard_regno_mode_ok): New function.
4176 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4177 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4179 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4181 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4182 (rs6000_hard_regno_mode_ok): Rename to...
4183 (rs6000_hard_regno_mode_ok_uncached): ...this.
4184 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4185 (rs6000_hard_regno_mode_ok): New function.
4186 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4187 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4188 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4189 (riscv_hard_regno_mode_ok): ...this and make static.
4190 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4191 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4192 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4193 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4194 (rl78_hard_regno_mode_ok): Make static and return bool.
4195 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4196 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4198 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4200 (rs6000_hard_regno_mode_ok): Rename to...
4201 (rs6000_hard_regno_mode_ok_uncached): ...this.
4202 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4203 (rs6000_hard_regno_mode_ok): New function.
4204 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4205 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4206 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4207 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4208 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4209 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4210 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4211 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4212 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4213 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4214 (sh_hard_regno_mode_ok): Make static.
4215 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4216 instead of HARD_REGNO_MODE_OK.
4217 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4218 (sparc_mode_class): Delete.
4219 (HARD_REGNO_MODE_OK): Delete.
4220 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4221 (hard_regno_mode_classes): Make static.
4222 (sparc_mode_class): Likewise.
4223 (sparc_hard_regno_mode_ok): New function.
4224 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4225 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4226 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4228 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4229 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4230 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4231 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4232 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4233 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4234 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4235 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4236 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4237 (visium_hard_regno_mode_ok): New function.
4238 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4239 instead of HARD_REGNO_MODE_OK.
4240 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4241 (HARD_REGNO_MODE_OK): Delete.
4242 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4243 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4244 (xtensa_option_override): Update accordingly.
4245 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4246 (xtensa_hard_regno_mode_ok): New function.
4247 * system.h (HARD_REGNO_MODE_OK): Poison.
4249 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4250 Alan Hayward <alan.hayward@arm.com>
4251 David Sherwood <david.sherwood@arm.com>
4253 * target.def (hard_regno_call_part_clobbered): New hook.
4254 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4255 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4256 * doc/tm.texi: Regenerate.
4257 * hooks.h (hook_bool_uint_mode_false): Declare.
4258 * hooks.c (hook_bool_uint_mode_false): New function.
4259 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4260 * cselib.c (cselib_process_insn): Use
4261 targetm.hard_regno_call_part_clobbered instead of
4262 HARD_REGNO_CALL_PART_CLOBBERED.
4263 * ira-conflicts.c (ira_build_conflicts): Likewise.
4264 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4265 * lra-constraints.c (need_for_call_save_p): Likewise.
4266 * lra-lives.c: Include target.h.
4267 (check_pseudos_live_through_calls): Use
4268 targetm.hard_regno_call_part_clobbered instead of
4269 HARD_REGNO_CALL_PART_CLOBBERED.
4270 * regcprop.c: Include target.h.
4271 (copyprop_hardreg_forward_1): Use
4272 targetm.hard_regno_call_part_clobbered instead of
4273 HARD_REGNO_CALL_PART_CLOBBERED.
4274 * reginfo.c (choose_hard_reg_mode): Likewise.
4275 * regrename.c (check_new_reg_p): Likewise.
4276 * reload.c (find_equiv_reg): Likewise.
4277 * reload1.c (emit_reload_insns): Likewise.
4278 * sched-deps.c (deps_analyze_insn): Likewise.
4279 * sel-sched.c (init_regs_for_mode): Likewise.
4280 (mark_unavailable_hard_regs): Likewise.
4281 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4282 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4283 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4285 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4286 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4287 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4289 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4291 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4292 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4293 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4295 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4296 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4297 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4299 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4300 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4302 * config/powerpcspe/powerpcspe.c
4303 (rs6000_hard_regno_call_part_clobbered): New function.
4304 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4305 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4306 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4308 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4309 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4310 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4312 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4313 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4314 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4316 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4317 Alan Hayward <alan.hayward@arm.com>
4318 David Sherwood <david.sherwood@arm.com>
4320 * rtl.h (subreg_memory_offset): Declare.
4321 * emit-rtl.c (subreg_memory_offset): New function.
4322 * expmed.c (store_bit_field_1): Use it.
4323 * expr.c (undefined_operand_subword_p): Likewise.
4324 * simplify-rtx.c (simplify_subreg): Likewise.
4326 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4328 PR rtl-optimization/57448
4331 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4332 using atomic_load pattern.
4333 (expand_atomic_store): Likewise.
4335 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4338 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4339 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4340 handling. Use replace_call_with_value with NULL instead of
4341 gsi_replace, unlink_stmt_vdef and release_defs.
4343 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4346 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4348 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4351 * rtl.h (emit_library_call_value_1): Declare.
4352 (emit_library_call): Replace declaration with a series of overloads.
4353 Remove the parameter count argument.
4354 (emit_library_call_value): Likewise.
4355 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4356 with an "rtx_mode_t *".
4357 (emit_library_call_value): Delete.
4358 (emit_library_call): Likewise.
4359 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4360 (asan_emit_allocas_unpoison): Likewise.
4361 * builtins.c (expand_builtin_powi): Likewise.
4362 (expand_asan_emit_allocas_unpoison): Likewise.
4363 * cfgexpand.c (expand_main_function): Likewise.
4364 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4365 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4366 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4367 * config/arm/arm.c (arm_trampoline_init): Likewise.
4368 (arm_call_tls_get_addr): Likewise.
4369 (arm_expand_divmod_libfunc): Likewise.
4370 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4371 (smulsi3_highpart): Likewise.
4372 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4373 (c6x_expand_compare): Likewise.
4374 (c6x_expand_movmem): Likewise.
4375 * config/frv/frv.c (frv_trampoline_init): Likewise.
4376 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4377 (ix86_expand_divmod_libfunc): Likewise.
4378 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4379 (ia64_expand_compare): Likewise.
4380 (ia64_profile_hook): Likewise.
4381 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4382 (nonlocal_goto): Likewise.
4383 (restore_stack_nonlocal): Likewise.
4384 * config/m32r/m32r.c (block_move_call): Likewise.
4385 (m32r_trampoline_init): Likewise.
4386 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4387 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4388 (m68k_call_m68k_read_tp): Likewise.
4389 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4390 (microblaze_expand_divide): Likewise.
4391 * config/mips/mips.h (mips_args): Likewise.
4392 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4393 (MIPS_ICACHE_SYNC): Likewise.
4394 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4395 (nios2_trampoline_init): Likewise.
4396 * config/pa/pa.c (hppa_tls_call): Likewise.
4397 (pa_trampoline_init): Likewise.
4398 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4399 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4400 (expand_strn_compare): Likewise.
4401 (rs6000_generate_compare): Likewise.
4402 (rs6000_expand_float128_convert): Likewise.
4403 (output_profile_hook): Likewise.
4404 (rs6000_trampoline_init): Likewise.
4405 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4406 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4407 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4408 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4409 (rs6000_generate_compare): Likewise.
4410 (rs6000_expand_float128_convert): Likewise.
4411 (output_profile_hook): Likewise.
4412 (rs6000_trampoline_init): Likewise.
4413 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4414 * config/sh/sh.c (sh_trampoline_init): Likewise.
4415 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4416 (sparc_emit_float_lib_cmp): Likewise.
4417 (sparc32_initialize_trampoline): Likewise.
4418 (sparc64_initialize_trampoline): Likewise.
4419 (sparc_profile_hook): Likewise.
4420 * config/spu/spu.c (ea_load_store): Likewise.
4421 * config/spu/spu.md (floatunssidf2): Likewise.
4422 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4423 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4424 * config/visium/visium.c (expand_block_move_4): Likewise.
4425 (expand_block_move_2): Likewise.
4426 (expand_block_move_1): Likewise.
4427 (expand_block_set_4): Likewise.
4428 (expand_block_set_2): Likewise.
4429 (expand_block_set_1): Likewise.
4430 (visium_trampoline_init): Likewise.
4431 (visium_profile_hook): Likewise.
4432 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4433 (xtensa_setup_frame_addresses): Likewise.
4434 (xtensa_trampoline_init): Likewise.
4435 * except.c (sjlj_emit_function_enter): Likewise.
4436 (sjlj_emit_function_exit): Likewise.
4437 * explow.c (allocate_dynamic_stack_space): Likewise.
4438 (probe_stack_range): Likewise.
4439 * expr.c (convert_mode_scalar): Likewise.
4440 * optabs.c (expand_binop): Likewise.
4441 (expand_twoval_binop_libfunc): Likewise.
4442 (expand_unop): Likewise.
4443 (prepare_cmp_insn): Likewise.
4444 (prepare_float_lib_cmp): Likewise.
4445 (expand_float): Likewise.
4446 (expand_fix): Likewise.
4447 (expand_fixed_convert): Likewise.
4448 (maybe_emit_sync_lock_test_and_set): Likewise.
4449 (expand_atomic_compare_and_swap): Likewise.
4450 (expand_mem_thread_fence): Likewise.
4451 (expand_atomic_fetch_op): Likewise.
4453 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4455 * doc/generic.texi (OpenACC): Adjust URL.
4456 * doc/invoke.texi (C Dialect Options): Ditto.
4458 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4460 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4461 predicate for operand 1. Add (m,<S>) constraint.
4462 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4463 Prevent memory operand 1 with register operand 2.
4465 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4467 PR rtl-optimization/82024
4468 * combine.c (try_combine): If the combination result is a PARALLEL,
4469 and we only need to retain the SET in there that would be placed
4470 at I2, check that we can place that at I3 instead, before doing so.
4472 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4475 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4477 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4478 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4481 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4482 Jeff Law <law@redhat.com>
4484 * varasm.c (bss_initializer_p): Do not put constants into .bss
4485 (categorize_decl_for_section): Handle bss_initializer_p returning
4486 false when DECL_INITIAL is NULL.
4488 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4491 * config/s390/s390.c (s390_can_inline_p): New function.
4493 2017-09-01 Jeff Law <law@redhat.com>
4495 PR tree-optimization/82052
4496 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4497 Always initialize the returned slot after a hash table miss
4498 when INSERT is true.
4500 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4502 * config/s390/s390.md (mem_signal_fence): Remove.
4503 * doc/md.texi (mem_signal_fence): Remove.
4504 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4506 * target-insns.def (mem_signal_fence): Remove.
4508 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4511 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4514 * asan.c (create_odr_indicator): Strip name encoding from assembler
4515 name before appending it after __odr_asan_.
4517 2017-09-01 Martin Liska <mliska@suse.cz>
4519 PR tree-optimization/82059
4520 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4521 frequency only when an edge is redirected.
4523 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4525 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4526 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4527 (arc_conditional_register_usage): Remove ARC600 lp_count
4529 (arc_file_start): Emit Tag_ARC_CPU_variation.
4530 (arc_can_use_doloop_p): New conditions to use ZOLs.
4531 (hwloop_fail): New function.
4532 (hwloop_optimize): Likewise.
4533 (hwloop_pattern_reg): Likewise.
4534 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4535 (arc_reorg_loops): New function, calls reorg_loops.
4536 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4537 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4539 (arc_loop_hazard): Remove function, functionality moved into
4541 (arc_hazard): Remove arc_loop_hazard call.
4542 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4543 into hwloop_optimize.
4544 (arc_label_align): Remove ZOL handling.
4545 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4546 * config/arc/arc.md (doloop_begin): Remove pattern.
4547 (doloop_begin_i): Likewise.
4548 (doloop_end_i): Likewise.
4549 (doloop_fallback): Likewise.
4550 (doloop_fallback_m): Likewise.
4551 (doloop_end): Reimplement expand.
4552 (arc_lp): New pattern for LP instruction.
4553 (loop_end): New pattern.
4554 (loop_fail): Likewise.
4555 (decrement_and_branch_until_zero): Likewise.
4556 * config/arc/arc.opt (mlpc-width): New option.
4557 * doc/invoke.texi (mlpc-width): Document option.
4559 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4561 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4562 (arc_ccfsm_advance): Fix checking for delay slots.
4563 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4565 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4567 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4569 (movhi_insn): Update store instruction constraint which are saving
4570 6-bit short immediates.
4571 (movsi_insn): Consider also short scaled load operations.
4572 (zero_extendhisi2_i): Use Usd constraint instead of T.
4573 (extendhisi2_i): Add q constraint.
4574 (arc_clzsi2): Add type and length attributes.
4575 (arc_ctzsi2): Likewise.
4576 * config/arc/constraints.md (Usc): Update constraint, the
4577 assembler can parse two relocations for a single instruction.
4579 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4581 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4582 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4584 2017-08-31 Olivier Hainque <hainque@adacore.com>
4586 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4587 match as powerpc-wrs-vxworks*.
4589 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4591 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4592 register constraint for by-element operand.
4593 (aarch64_mls_elt_merge<mode>): Likewise.
4595 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4597 * config/arc/arc.c (arc_can_follow_jump): Check for short
4600 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4602 * config.gcc: Use g.opt for arc.
4603 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4604 functionality moved to ...
4605 (legitimate_scaled_address_p): New function, ...here.
4606 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4607 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4608 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4610 (arc_override_options): Handle G option.
4611 (arc_output_pic_addr_const): Correct function definition.
4612 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4613 (arc_decl_anon_ns_mem_p): Delete.
4614 (arc_in_small_data_p): Overhaul this function to take into
4615 consideration the value given via G option.
4616 (arc_rewrite_small_data_1): Renamed and corrected old
4617 arc_rewrite_small_data function.
4618 (arc_rewrite_small_data): New function.
4619 (small_data_pattern): Don't use pic_offset_table_rtx.
4620 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4621 * config/arc/simdext.md (movmisalignv2hi): Use
4622 prepare_move_operands function.
4624 (movmisalign*): Likewise.
4625 * doc/invoke.texi (ARC options): Document -G option.
4627 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4629 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4631 * config/arc/arc.c (arc_print_operand): Output scalled address for
4632 sdata whenever is possible.
4633 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4634 load/stores are available.
4635 (compact_sda_memory_operand): Check for the alignment required by
4636 code density instructions.
4637 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4639 * config/arc/constraints.md (Usd): Update constraint.
4640 (Us0): New constraint.
4641 (Usc): Update constraint.
4643 2017-08-31 Richard Biener <rguenther@suse.de>
4646 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4649 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4651 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4652 Resize type_signature.
4654 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4655 Alan Hayward <alan.hayward@arm.com>
4656 David Sherwood <david.sherwood@arm.com>
4658 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4659 subregs whose inner modes can be stored in GPRs.
4660 (aarch64_classify_index): Likewise.
4662 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4663 Alan Hayward <alan.hayward@arm.com>
4664 David Sherwood <david.sherwood@arm.com>
4666 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4667 (V_INT_EQUIV): ...this.
4668 (v_cmp_result): Rename to...
4669 (v_int_equiv): ...this.
4670 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4671 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4672 (copysign<mode>3): Likewise.
4673 (aarch64_simd_bsl<mode>_internal): Likewise.
4674 (aarch64_simd_bsl<mode>): Likewise.
4675 (vec_cmp<mode><mode>): Likewise.
4676 (vcond<mode><mode>): Likewise.
4677 (vcond<v_cmp_mixed><mode>): Likewise.
4678 (vcondu<mode><v_cmp_mixed>): Likewise.
4679 (aarch64_cm<optab><mode>): Likewise.
4680 (aarch64_cmtst<mode>): Likewise.
4681 (aarch64_fac<optab><mode>): Likewise.
4682 (vec_perm_const<mode>): Likewise.
4683 (vcond_mask_<mode><v_cmp_result>): Rename to...
4684 (vcond_mask_<mode><v_int_equiv>): ...this.
4685 (vec_cmp<mode><v_cmp_result>): Rename to...
4686 (vec_cmp<mode><v_int_equiv>): ...this.
4688 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4689 Alan Hayward <alan.hayward@arm.com>
4690 David Sherwood <david.sherwood@arm.com>
4692 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4694 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4695 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4696 (UNSPEC_LD4_DREG): New unspecs.
4697 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4698 (aarch64_ld2<mode>_dreg_be): Replace with...
4699 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4701 (aarch64_ld3<mode>_dreg_le)
4702 (aarch64_ld3<mode>_dreg_be): Replace with...
4703 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4705 (aarch64_ld4<mode>_dreg_le)
4706 (aarch64_ld4<mode>_dreg_be): Replace with...
4707 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4710 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4712 PR tree-optimization/81987
4713 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4714 insert an initializer in a location not dominated by the stride
4717 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4719 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4720 on the entire header of the finally block in the fallthru case.
4722 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4724 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4726 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4728 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4729 rs6000_emit_move_from_cr and call renamed function.
4730 (rs6000_emit_prologue): Call renamed functions.
4731 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4732 movesi_from_cr, remove volatile CRs.
4734 2017-08-30 Jon Beniston <jon@beniston.com>
4735 Richard Biener <rguenther@suse.de>
4737 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4738 of VECTOR_MODE_P check.
4739 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4740 element vector types.
4742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4744 * df.h (df_read_modify_subreg_p): Remove in favor of...
4745 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4746 const_rtx instead of an rtx.
4747 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4748 * df-problems.c (df_word_lr_mark_ref): Likewise.
4749 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4750 (mark_pseudo_reg_dead): Likewise.
4751 (mark_ref_dead): Likewise.
4752 * reginfo.c (init_subregs_of_mode): Likewise.
4753 * sched-deps.c (sched_analyze_1): Likewise.
4754 * df-scan.c (df_def_record_1): Likewise.
4755 (df_uses_record): Likewise.
4756 (df_read_modify_subreg_p): Remove in favor of...
4757 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4758 const_rtx instead of an rtx.
4760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4761 Alan Hayward <alan.hayward@arm.com>
4762 David Sherwood <david.sherwood@arm.com>
4764 * rtl.h (partial_subreg_p): New function.
4765 * caller-save.c (save_call_clobbered_regs): Use it.
4766 * calls.c (expand_call): Likewise.
4767 * combine.c (combinable_i3pat): Likewise.
4768 (simplify_set): Likewise.
4769 (make_extraction): Likewise.
4770 (make_compound_operation_int): Likewise.
4771 (gen_lowpart_or_truncate): Likewise.
4772 (force_to_mode): Likewise.
4773 (make_field_assignment): Likewise.
4774 (reg_truncated_to_mode): Likewise.
4775 (record_truncated_value): Likewise.
4776 (move_deaths): Likewise.
4777 * cse.c (record_jump_cond): Likewise.
4778 (cse_insn): Likewise.
4779 * cselib.c (cselib_lookup_1): Likewise.
4780 * expmed.c (extract_bit_field_using_extv): Likewise.
4781 * function.c (assign_parm_setup_reg): Likewise.
4782 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4783 * ira-build.c (create_insn_allocnos): Likewise.
4784 * lra-coalesce.c (merge_pseudos): Likewise.
4785 * lra-constraints.c (match_reload): Likewise.
4786 (simplify_operand_subreg): Likewise.
4787 (curr_insn_transform): Likewise.
4788 * lra-lives.c (process_bb_lives): Likewise.
4789 * lra.c (new_insn_reg): Likewise.
4790 (lra_substitute_pseudo): Likewise.
4791 * regcprop.c (mode_change_ok): Likewise.
4792 (maybe_mode_change): Likewise.
4793 (copyprop_hardreg_forward_1): Likewise.
4794 * reload.c (push_reload): Likewise.
4795 (find_reloads): Likewise.
4796 (find_reloads_subreg_address): Likewise.
4797 * reload1.c (alter_reg): Likewise.
4798 (eliminate_regs_1): Likewise.
4799 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4801 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4803 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4804 back to if statements, including unpack.
4806 2017-08-30 Martin Liska <mliska@suse.cz>
4809 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4811 (func_checker::compare_asm_inputs_outputs): ... this function.
4812 (func_checker::compare_gimple_asm): Use the function to compare
4813 also ASM constrains.
4814 * ipa-icf-gimple.h: Rename the function.
4816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4817 Alan Hayward <alan.hayward@arm.com>
4818 David Sherwood <david.sherwood@arm.com>
4820 * coretypes.h (complex_mode): New type.
4821 * gdbhooks.py (build_pretty_printer): Handle it.
4822 * machmode.h (complex_mode): New class.
4823 (complex_mode::includes_p): New function.
4824 (is_complex_int_mode): Likewise.
4825 (is_complex_float_mode): Likewise.
4826 * genmodes.c (get_mode_class): Handle complex mode classes.
4827 * function.c (expand_function_end): Use is_complex_int_mode.
4829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4830 Alan Hayward <alan.hayward@arm.com>
4831 David Sherwood <david.sherwood@arm.com>
4833 * coretypes.h (scalar_mode_pod): New typedef.
4834 * gdbhooks.py (build_pretty_printer): Handle it.
4835 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4836 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4837 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4838 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4842 Alan Hayward <alan.hayward@arm.com>
4843 David Sherwood <david.sherwood@arm.com>
4845 * machmode.h (mode_for_vector): Take a scalar_mode instead
4847 * stor-layout.c (mode_for_vector): Likewise.
4848 * explow.c (promote_mode): Use as_a <scalar_mode>.
4849 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4852 Alan Hayward <alan.hayward@arm.com>
4853 David Sherwood <david.sherwood@arm.com>
4855 * target.def (preferred_simd_mode): Take a scalar_mode
4856 instead of a machine_mode.
4857 * targhooks.h (default_preferred_simd_mode): Likewise.
4858 * targhooks.c (default_preferred_simd_mode): Likewise.
4859 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4860 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4861 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4862 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4863 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4864 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4865 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4866 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4868 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4869 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4870 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4871 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4872 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4873 * doc/tm.texi: Regenerate.
4874 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4878 Alan Hayward <alan.hayward@arm.com>
4879 David Sherwood <david.sherwood@arm.com>
4881 * target.def (scalar_mode_supported_p): Take a scalar_mode
4882 instead of a machine_mode.
4883 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4884 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4885 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4886 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4887 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4888 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4889 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4890 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4891 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4892 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4893 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4894 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4895 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4896 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4898 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4899 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4900 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4901 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4902 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4904 * doc/tm.texi: Regenerate.
4906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4907 Alan Hayward <alan.hayward@arm.com>
4908 David Sherwood <david.sherwood@arm.com>
4910 * coretypes.h (opt_scalar_mode): New typedef.
4911 * gdbhooks.py (build_pretty_printers): Handle it.
4912 * machmode.h (mode_iterator::get_2xwider): Add overload for
4914 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4916 * expr.c (convert_mode_scalar): Likewise.
4917 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4918 * optabs.c (expand_float): Likewise.
4919 (expand_fix): Likewise.
4920 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4923 Alan Hayward <alan.hayward@arm.com>
4924 David Sherwood <david.sherwood@arm.com>
4926 * optabs.c (expand_float): Explicitly check for scalars before
4927 using a branching expansion.
4928 (expand_fix): Likewise.
4930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4931 Alan Hayward <alan.hayward@arm.com>
4932 David Sherwood <david.sherwood@arm.com>
4934 * expr.c (convert_mode): Split scalar handling out into...
4935 (convert_mode_scalar): ...this new function. Treat the modes
4938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4939 Alan Hayward <alan.hayward@arm.com>
4940 David Sherwood <david.sherwood@arm.com>
4942 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4944 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4947 Alan Hayward <alan.hayward@arm.com>
4948 David Sherwood <david.sherwood@arm.com>
4950 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4951 rather than a machine_mode.
4952 (fixed_from_string): Likewise.
4953 (fixed_convert): Likewise.
4954 (fixed_convert_from_int): Likewise.
4955 (fixed_convert_from_real): Likewise.
4956 (real_convert_from_fixed): Likewise.
4957 * fixed-value.c (fixed_from_double_int): Likewise.
4958 (fixed_from_string): Likewise.
4959 (fixed_convert): Likewise.
4960 (fixed_convert_from_int): Likewise.
4961 (fixed_convert_from_real): Likewise.
4962 (real_convert_from_fixed): Likewise.
4963 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4966 Alan Hayward <alan.hayward@arm.com>
4967 David Sherwood <david.sherwood@arm.com>
4969 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4970 of separate mode class checks. Do not allow vector modes here.
4971 (immed_wide_int_const): Use as_a <scalar_mode>.
4972 * explow.c (trunc_int_for_mode): Likewise.
4973 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4974 (wi::shwi): Likewise.
4975 (wi::min_value): Likewise.
4976 (wi::max_value): Likewise.
4977 * dwarf2out.c (loc_descriptor): Likewise.
4978 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4982 Alan Hayward <alan.hayward@arm.com>
4983 David Sherwood <david.sherwood@arm.com>
4985 * tree.h (SCALAR_TYPE_MODE): New macro.
4986 * expr.c (expand_expr_addr_expr_1): Use it.
4987 (expand_expr_real_2): Likewise.
4988 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4989 (fold_convert_const_fixed_from_int): Likewise.
4990 (fold_convert_const_fixed_from_real): Likewise.
4991 (native_encode_fixed): Likewise
4992 (native_encode_complex): Likewise
4993 (native_encode_vector): Likewise.
4994 (native_interpret_fixed): Likewise.
4995 (native_interpret_real): Likewise.
4996 (native_interpret_complex): Likewise.
4997 (native_interpret_vector): Likewise.
4998 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4999 (simd_clone_adjust_argument_types): Likewise.
5000 (simd_clone_init_simd_arrays): Likewise.
5001 (simd_clone_adjust): Likewise.
5002 * stor-layout.c (layout_type): Likewise.
5003 * tree.c (build_minus_one_cst): Likewise.
5004 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5005 * tree-inline.c (estimate_move_cost): Likewise.
5006 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5007 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5008 (vectorizable_reduction): Likewise.
5009 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5010 (vect_recog_mixed_size_cond_pattern): Likewise.
5011 (check_bool_pattern): Likewise.
5012 (adjust_bool_pattern): Likewise.
5013 (search_type_for_mask_1): Likewise.
5014 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5015 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5016 (vectorizable_load): Likewise.
5017 (vectorizable_store): Likewise.
5018 * ubsan.c (ubsan_encode_value): Likewise.
5019 * varasm.c (output_constant): Likewise.
5021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5022 Alan Hayward <alan.hayward@arm.com>
5023 David Sherwood <david.sherwood@arm.com>
5025 * coretypes.h (scalar_mode): New class.
5026 * machmode.h (scalar_mode): Likewise.
5027 (scalar_mode::includes_p): New function.
5028 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5029 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5030 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5031 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5032 * expmed.c (store_bit_field_1): Likewise.
5033 (extract_bit_field_1): Likewise.
5034 * expr.c (write_complex_part): Likewise.
5035 (read_complex_part): Likewise.
5036 (emit_move_complex_push): Likewise.
5037 (expand_expr_real_2): Likewise.
5038 * function.c (assign_parm_setup_reg): Likewise.
5039 (assign_parms_unsplit_complex): Likewise.
5040 * optabs.c (expand_binop): Likewise.
5041 * rtlanal.c (subreg_get_info): Likewise.
5042 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5043 * varasm.c (output_constant_pool_2): Likewise.
5045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5046 Alan Hayward <alan.hayward@arm.com>
5047 David Sherwood <david.sherwood@arm.com>
5049 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5051 (expmed_mult_highpart_optab): Likewise.
5052 (expmed_mult_highpart): 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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5059 instead of a machine_mode.
5060 (builtin_memset_read_str): Likewise.
5061 * builtins.c (c_readstr): Likewise.
5062 (builtin_memcpy_read_str): Likewise.
5063 (builtin_strncpy_read_str): Likewise.
5064 (builtin_memset_read_str): Likewise.
5065 (builtin_memset_gen_str): Likewise.
5066 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5067 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5068 instead of a machine_mode.
5069 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5071 (make_extraction): Likewise.
5072 (try_widen_shift_mode): Take and return scalar_int_modes instead
5074 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5075 a scalar_int_mode instead of a machine_mode.
5076 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5077 (avr_addr_space_pointer_mode): Likewise.
5078 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5079 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5080 (msp430_unwind_word_mode): Likewise.
5081 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5082 (spu_addr_space_pointer_mode): Likewise.
5083 (spu_addr_space_address_mode): Likewise.
5084 (spu_libgcc_cmp_return_mode): Likewise.
5085 (spu_libgcc_shift_count_mode): Likewise.
5086 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5087 (rl78_addr_space_pointer_mode): Likewise.
5088 (fl78_unwind_word_mode): Likewise.
5089 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5091 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5092 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5093 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5094 (mips_valid_pointer_mode): Likewise.
5095 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5096 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5097 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5099 (ft32_addr_space_address_mode): Likewise.
5100 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5101 scalar_int_mode instead of a machine_mode.
5102 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5104 (m32c_addr_space_address_mode): Likewise.
5105 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5106 (rs6000_eh_return_filter_mode): Likewise.
5107 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5108 (rs6000_eh_return_filter_mode): Likewise.
5109 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5110 (s390_libgcc_shift_count_mode): Likewise.
5111 (s390_unwind_word_mode): Likewise.
5112 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5114 * target.def (mode_rep_extended): Likewise.
5115 (valid_pointer_mode): Likewise.
5116 (addr_space.valid_pointer_mode): Likewise.
5117 (eh_return_filter_mode): Return a scalar_int_mode rather than
5119 (libgcc_cmp_return_mode): Likewise.
5120 (libgcc_shift_count_mode): Likewise.
5121 (unwind_word_mode): Likewise.
5122 (addr_space.pointer_mode): Likewise.
5123 (addr_space.address_mode): Likewise.
5124 * doc/tm.texi: Regenerate.
5125 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5127 (do_jump): Use scalar_int_mode for local variables.
5128 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5129 rather than a machine_mode.
5130 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5131 (scompare_loc_descriptor_wide): Likewise.
5132 (scompare_loc_descriptor_narrow): Likewise.
5133 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5135 * except.c (sjlj_emit_dispatch_table): Likewise.
5136 (expand_builtin_eh_copy_values): Likewise.
5137 * explow.c (convert_memory_address_addr_space_1): Likewise.
5138 Take a scalar_int_mode rather than a machine_mode.
5139 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5140 than a machine_mode.
5141 (memory_address_addr_space): Use scalar_int_mode for local variables.
5142 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5143 rather than a machine_mode.
5144 * expmed.c (mask_rtx): Likewise.
5145 (init_expmed_one_conv): Likewise.
5146 (expand_mult_highpart_adjust): Likewise.
5147 (extract_high_half): Likewise.
5148 (expmed_mult_highpart_optab): Likewise.
5149 (expmed_mult_highpart): Likewise.
5150 (expand_smod_pow2): Likewise.
5151 (expand_sdiv_pow2): Likewise.
5152 (emit_store_flag_int): Likewise.
5153 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5155 (extract_low_bits): Likewise.
5156 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5158 * expr.c (pieces_addr::adjust): Likewise.
5159 (can_store_by_pieces): Likewise.
5160 (store_by_pieces): Likewise.
5161 (clear_by_pieces_1): Likewise.
5162 (expand_expr_addr_expr_1): Likewise.
5163 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5164 (expand_expr_real_1): Likewise.
5165 (try_casesi): Likewise.
5166 * final.c (shorten_branches): Likewise.
5167 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5168 type of "mode" to machine_mode.
5169 * internal-fn.c (expand_arith_overflow_result_store): Take a
5170 scalar_int_mode rather than a machine_mode.
5171 (expand_mul_overflow): Use scalar_int_mode for local variables.
5172 * loop-doloop.c (doloop_modify): Likewise.
5173 (doloop_optimize): Likewise.
5174 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5175 than a machine_mode.
5176 (expand_doubleword_shift_condmove): Likewise.
5177 (expand_doubleword_shift): Likewise.
5178 (expand_doubleword_clz): Likewise.
5179 (expand_doubleword_popcount): Likewise.
5180 (expand_doubleword_parity): Likewise.
5181 (expand_absneg_bit): Use scalar_int_mode for local variables.
5182 (prepare_float_lib_cmp): Likewise.
5183 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5184 rather than a machine_mode.
5185 (convert_memory_address_addr_space): Likewise.
5186 (get_mode_bounds): Likewise.
5187 (get_address_mode): Return a scalar_int_mode rather than a
5189 * rtlanal.c (get_address_mode): Likewise.
5190 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5191 than a machine_mode.
5192 * targhooks.c (default_mode_rep_extended): Likewise.
5193 (default_valid_pointer_mode): Likewise.
5194 (default_addr_space_valid_pointer_mode): Likewise.
5195 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5196 than a machine_mode.
5197 (default_libgcc_cmp_return_mode): Likewise.
5198 (default_libgcc_shift_count_mode): Likewise.
5199 (default_unwind_word_mode): Likewise.
5200 (default_addr_space_pointer_mode): Likewise.
5201 (default_addr_space_address_mode): Likewise.
5202 * targhooks.h (default_eh_return_filter_mode): Likewise.
5203 (default_libgcc_cmp_return_mode): Likewise.
5204 (default_libgcc_shift_count_mode): Likewise.
5205 (default_unwind_word_mode): Likewise.
5206 (default_addr_space_pointer_mode): Likewise.
5207 (default_addr_space_address_mode): Likewise.
5208 (default_mode_rep_extended): Take a scalar_int_mode rather than
5210 (default_valid_pointer_mode): Likewise.
5211 (default_addr_space_valid_pointer_mode): Likewise.
5212 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5214 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5215 rather than a machine_mode.
5216 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5217 for local variables.
5218 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5219 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5220 than a machine_mode.
5222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5223 Alan Hayward <alan.hayward@arm.com>
5224 David Sherwood <david.sherwood@arm.com>
5226 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5227 the mode argument to scalar_int_mode.
5228 (do_jump_by_parts_zero_rtx): Likewise.
5229 (do_jump_by_parts_equality_rtx): Likewise.
5230 (do_jump_by_parts_greater): Take a mode argument.
5231 (do_jump_by_parts_equality): Likewise.
5232 (do_jump_1): Update calls accordingly.
5234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5235 Alan Hayward <alan.hayward@arm.com>
5236 David Sherwood <david.sherwood@arm.com>
5238 * is-a.h (safe_dyn_cast): New function.
5239 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5240 (jump_table_for_label): Likewise.
5241 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5242 instead of an rtx_insn *.
5243 (shorten_branches): Use dyn_cast instead of LABEL_P and
5244 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5245 rtx_jump_table_data::get_data_mode.
5246 (final_scan_insn): Likewise.
5248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5249 Alan Hayward <alan.hayward@arm.com>
5250 David Sherwood <david.sherwood@arm.com>
5252 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5253 trying to combine a full-register integer set with a subreg
5256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5257 Alan Hayward <alan.hayward@arm.com>
5258 David Sherwood <david.sherwood@arm.com>
5260 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5261 that is always either address_mode or pointer_mode.
5263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5264 Alan Hayward <alan.hayward@arm.com>
5265 David Sherwood <david.sherwood@arm.com>
5267 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5268 when the two are known to be equal.
5270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5271 Alan Hayward <alan.hayward@arm.com>
5272 David Sherwood <david.sherwood@arm.com>
5274 * simplify-rtx.c (simplify_const_unary_operation): Use
5275 is_a <scalar_int_mode> instead of checking for a nonzero
5276 precision. Forcibly convert op_mode to a scalar_int_mode
5277 in that case. More clearly differentiate the operand and
5278 result modes and use the former when deciding what the value
5279 of a count-bits operation should be. Use is_int_mode instead
5280 of checking for a MODE_INT. Remove redundant check for whether
5281 this mode has a zero precision.
5283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5284 Alan Hayward <alan.hayward@arm.com>
5285 David Sherwood <david.sherwood@arm.com>
5287 * optabs.c (widen_leading): Change the type of the mode argument
5288 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5289 (widen_bswap): Likewise.
5290 (expand_parity): Likewise.
5291 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5292 (expand_ffs): Likewise.
5293 (epand_unop): Check for scalar integer modes before calling the
5296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5297 Alan Hayward <alan.hayward@arm.com>
5298 David Sherwood <david.sherwood@arm.com>
5300 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5302 (expand_expr_real_1): Update call accordingly.
5304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5305 Alan Hayward <alan.hayward@arm.com>
5306 David Sherwood <david.sherwood@arm.com>
5308 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5309 value_mode arguments. Use scalar_int_mode internally.
5310 (store_bit_field_1): Rename the new integer mode from imode
5311 to op0_mode and use it instead of GET_MODE (op0). Update calls
5312 to store_split_bit_field, store_bit_field_using_insv and
5313 store_fixed_bit_field.
5314 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5315 Use scalar_int_mode internally. Use a bit count rather than a mode
5316 when calculating the largest bit size for get_best_mode.
5317 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5318 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5319 Remove assertion that OP0 has a scalar integer mode.
5320 (store_split_bit_field): Add op0_mode and value_mode arguments.
5321 Update calls to extract_fixed_bit_field.
5322 (extract_bit_field_using_extv): Add an op0_mode argument.
5323 Use scalar_int_mode internally.
5324 (extract_bit_field_1): Rename the new integer mode from imode to
5325 op0_mode and use it instead of GET_MODE (op0). Update calls to
5326 extract_split_bit_field, extract_bit_field_using_extv and
5327 extract_fixed_bit_field.
5328 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5329 to extract_split_bit_field and extract_fixed_bit_field_1.
5330 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5331 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5333 (extract_split_bit_field): Add an op0_mode argument. Update call
5334 to extract_fixed_bit_field.
5336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5337 Alan Hayward <alan.hayward@arm.com>
5338 David Sherwood <david.sherwood@arm.com>
5340 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5341 * explow.c (hard_function_value): Likewise.
5342 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5343 convert_to_mode call outside the loop.
5344 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5345 for the mode iterator. Require the mode specified by max_pieces
5347 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5349 (copy_blkmode_to_reg): Likewise.
5350 (set_storage_via_setmem): Likewise.
5351 * optabs.c (prepare_cmp_insn): Likewise.
5352 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5353 * stor-layout.c (finish_bitfield_representative): Likewise.
5355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5356 Alan Hayward <alan.hayward@arm.com>
5357 David Sherwood <david.sherwood@arm.com>
5359 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5360 * expr.c (convert_move): Use them.
5361 (convert_modes): Likewise.
5362 (store_expr_with_bounds): 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 (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5369 parameter for the mode of "x". Remove the "known_x", "known_mode"
5370 and "known_ret" arguments. Change the type of the mode argument
5372 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5373 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5374 (reg_num_sign_bit_copies_for_combine): Likewise.
5375 * rtlanal.c (nonzero_bits1): Likewise.
5376 (num_sign_bit_copies1): Likewise.
5377 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5378 (reg_num_sign_bit_copies_general): Likewise.
5379 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5380 (reg_nonzero_bits_general): Likewise.
5382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5383 Alan Hayward <alan.hayward@arm.com>
5384 David Sherwood <david.sherwood@arm.com>
5386 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5387 than in subroutines. Return 1 for non-integer modes.
5388 (cached_num_sign_bit_copies): Change the type of the mode parameter
5390 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5391 classes. Handle CONST_INT_P first and then check whether X also
5392 has a scalar integer mode. Check the same thing for inner registers
5393 of a SUBREG and for values that are being extended or truncated.
5395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5396 Alan Hayward <alan.hayward@arm.com>
5397 David Sherwood <david.sherwood@arm.com>
5399 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5400 in subroutines. Return the mode mask for non-integer modes.
5401 (cached_nonzero_bits): Change the type of the mode parameter
5403 (nonzero_bits1): Likewise. Remove early exit for other mode
5404 classes. Handle CONST_INT_P first and then check whether X
5405 also has a scalar integer mode.
5407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5408 Alan Hayward <alan.hayward@arm.com>
5409 David Sherwood <david.sherwood@arm.com>
5411 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5412 Return a scalar_int_mode and assert that the size is greater than
5414 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5415 (op_by_pieces_d::op_by_pieces_d): Likewise.
5416 (op_by_pieces_d::run): Likewise.
5417 (can_store_by_pieces): Likewise.
5419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5420 Alan Hayward <alan.hayward@arm.com>
5421 David Sherwood <david.sherwood@arm.com>
5423 * combine.c (extract_left_shift): Add a mode argument and update
5425 (make_compound_operation_int): Change the type of the mode parameter
5426 to scalar_int_mode and update the call to extract_left_shift.
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 (simplify_and_const_int): Change the type of the mode
5433 parameter to scalar_int_mode.
5434 (simplify_and_const_int_1): Likewise. Update recursive call.
5436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5437 Alan Hayward <alan.hayward@arm.com>
5438 David Sherwood <david.sherwood@arm.com>
5440 * combine.c (simplify_compare_const): Check that the mode is a
5441 scalar_int_mode (rather than VOIDmode) before testing its
5443 (simplify_comparison): Move COMPARISON_P handling out of the
5444 loop and restrict the latter part of the loop to scalar_int_modes.
5445 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5446 and when considering SUBREG_REGs. Use is_int_mode instead of
5447 checking GET_MODE_CLASS against MODE_INT.
5449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5450 Alan Hayward <alan.hayward@arm.com>
5451 David Sherwood <david.sherwood@arm.com>
5453 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5454 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5455 shift_unit_mode and for modes involved in scalar shifts.
5457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5458 Alan Hayward <alan.hayward@arm.com>
5459 David Sherwood <david.sherwood@arm.com>
5461 * combine.c (force_int_to_mode): New function, split out from...
5462 (force_to_mode): ...here. Keep xmode up-to-date and use it
5463 instead of GET_MODE (x).
5465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5466 Alan Hayward <alan.hayward@arm.com>
5467 David Sherwood <david.sherwood@arm.com>
5469 * optabs-query.h (extraction_insn::struct_mode): Change type to
5470 opt_scalar_int_mode and update comment.
5471 (extraction_insn::field_mode): Change type to scalar_int_mode.
5472 (extraction_insn::pos_mode): Likewise.
5473 * combine.c (make_extraction): Update accordingly.
5474 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5475 (get_optab_extraction_insn): Likewise.
5476 * recog.c (simplify_while_replacing): Likewise.
5477 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5478 parameter to opt_scalar_int_mode.
5480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5481 Alan Hayward <alan.hayward@arm.com>
5482 David Sherwood <david.sherwood@arm.com>
5484 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5485 to a scalar_int_mode instead of a machine_mode.
5486 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5487 (get_best_mode): Return a boolean and use a pointer argument to store
5488 the selected mode. Replace the limit mode parameter with a bit limit.
5489 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5490 for the values returned by bit_field_mode_iterator::next_mode.
5491 (store_bit_field): Update call to get_best_mode.
5492 (store_fixed_bit_field): Likewise.
5493 (extract_fixed_bit_field): Likewise.
5494 * expr.c (optimize_bitfield_assignment_op): Likewise.
5495 * fold-const.c (optimize_bit_field_compare): Likewise.
5496 (fold_truth_andor_1): Likewise.
5497 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5498 Update for new type of m_mode.
5499 (get_best_mode): As above.
5501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5502 Alan Hayward <alan.hayward@arm.com>
5503 David Sherwood <david.sherwood@arm.com>
5505 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5506 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5507 (store_bit_field): Check is_a <scalar_int_mode> before calling
5508 strict_volatile_bitfield_p.
5509 (extract_bit_field): Likewise.
5511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5512 Alan Hayward <alan.hayward@arm.com>
5513 David Sherwood <david.sherwood@arm.com>
5515 * target.def (cstore_mode): Return a scalar_int_mode.
5516 * doc/tm.texi: Regenerate.
5517 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5518 * targhooks.h (default_cstore_mode): Likewise.
5519 * targhooks.c (default_cstore_mode): Likewise, using a forced
5521 * expmed.c (emit_cstore): Expect the target of the cstore to be
5524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5525 Alan Hayward <alan.hayward@arm.com>
5526 David Sherwood <david.sherwood@arm.com>
5528 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5530 (niter_desc): Likewise mode.
5531 (iv_analyze): Add a mode parameter.
5533 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5534 and change its type to scalar_int_mode.
5535 * loop-iv.c: Update commentary at head of file.
5536 (iv_constant): Pass the mode paraeter before the rtx it describes
5537 and change its type to scalar_int_mode. Remove VOIDmode handling.
5538 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5539 (iv_extend): Likewise.
5540 (shorten_into_mode): Likewise.
5541 (iv_add): Use scalar_int_mode.
5542 (iv_mult): Likewise.
5543 (iv_shift): Likewise.
5544 (canonicalize_iv_subregs): Likewise.
5545 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5546 it describes and change its mode to scalar_int_mode. Also change
5547 the type of the returned inner_mode to scalar_int_mode.
5548 (get_biv_step): Likewise, turning outer_mode from a pointer
5549 into a direct parameter. Update call to get_biv_step_1.
5550 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5551 iv_constant and get_biv_step.
5552 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5553 and change its type to scalar_int_mode. Don't initialise iv->mode
5554 to VOIDmode and remove later checks for its still being VOIDmode.
5555 Update calls to iv_analyze_op and iv_analyze_expr. Check
5556 is_a <scalar_int_mode> when changing the mode under consideration.
5557 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5558 Update call to iv_analyze_expr.
5559 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5560 inner register is not also a scalar_int_mode. Update call to
5562 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5563 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5564 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5565 separate mode class checks. Update calls to iv_analyze. Remove
5566 fix-up of VOIDmodes after iv_analyze_biv.
5567 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5568 don't have a scalar_int_mode. Update call to biv_p.
5570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5571 Alan Hayward <alan.hayward@arm.com>
5572 David Sherwood <david.sherwood@arm.com>
5574 * cfgexpand.c (convert_debug_memory_address): Use
5575 as_a <scalar_int_mode>.
5576 * combine.c (expand_compound_operation): Likewise.
5577 (make_extraction): Likewise.
5578 (change_zero_ext): Likewise.
5579 (simplify_comparison): Likewise.
5580 * cse.c (cse_insn): Likewise.
5581 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5582 (mem_loc_descriptor): Likewise.
5583 (loc_descriptor): Likewise.
5584 * expmed.c (init_expmed_one_mode): Likewise.
5585 (synth_mult): Likewise.
5586 (emit_store_flag_1): Likewise.
5587 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5588 of a comparison with size.
5589 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5590 (reduce_to_bit_field_precision): Likewise.
5591 * function.c (expand_function_end): Likewise.
5592 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5593 * loop-doloop.c (doloop_modify): Likewise.
5594 * optabs.c (expand_binop): Likewise.
5595 (expand_unop): Likewise.
5596 (expand_copysign_absneg): Likewise.
5597 (prepare_cmp_insn): Likewise.
5598 (maybe_legitimize_operand): Likewise.
5599 * recog.c (const_scalar_int_operand): Likewise.
5600 * rtlanal.c (get_address_mode): Likewise.
5601 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5602 (simplify_cond_clz_ctz): Likewise.
5603 * tree-nested.c (get_nl_goto_field): Likewise.
5604 * tree.c (build_vector_type_for_mode): Likewise.
5605 * var-tracking.c (use_narrower_mode): Likewise.
5607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5608 Alan Hayward <alan.hayward@arm.com>
5609 David Sherwood <david.sherwood@arm.com>
5611 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5612 * builtins.c (expand_builtin_signbit): Use it.
5613 * cfgexpand.c (expand_debug_expr): Likewise.
5614 * dojump.c (do_jump): Likewise.
5615 (do_compare_and_jump): Likewise.
5616 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5617 * expmed.c (make_tree): Likewise.
5618 * expr.c (expand_expr_real_2): Likewise.
5619 (expand_expr_real_1): Likewise.
5620 (try_casesi): Likewise.
5621 * fold-const-call.c (fold_const_call_ss): Likewise.
5622 * fold-const.c (unextend): Likewise.
5623 (extract_muldiv_1): Likewise.
5624 (fold_single_bit_test): Likewise.
5625 (native_encode_int): Likewise.
5626 (native_encode_string): Likewise.
5627 (native_interpret_int): Likewise.
5628 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5629 * internal-fn.c (expand_addsub_overflow): Likewise.
5630 (expand_neg_overflow): Likewise.
5631 (expand_mul_overflow): Likewise.
5632 (expand_arith_overflow): Likewise.
5633 * match.pd: Likewise.
5634 * stor-layout.c (layout_type): Likewise.
5635 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5636 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5637 * tree-ssanames.c (get_range_info): Likewise.
5638 * tree-switch-conversion.c (array_value_type) Likewise.
5639 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5640 (vect_recog_divmod_pattern): Likewise.
5641 (vect_recog_mixed_size_cond_pattern): Likewise.
5642 * tree-vrp.c (extract_range_basic): Likewise.
5643 (simplify_float_conversion_using_ranges): Likewise.
5644 * tree.c (int_fits_type_p): Likewise.
5645 * ubsan.c (instrument_bool_enum_load): Likewise.
5646 * varasm.c (mergeable_string_section): Likewise.
5647 (narrowing_initializer_constant_valid_p): Likewise.
5648 (output_constant): Likewise.
5650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5651 Alan Hayward <alan.hayward@arm.com>
5652 David Sherwood <david.sherwood@arm.com>
5654 * machmode.h (NARROWEST_INT_MODE): New macro.
5655 * expr.c (alignment_for_piecewise_move): Use it instead of
5656 GET_CLASS_NARROWEST_MODE (MODE_INT).
5657 (push_block): Likewise.
5658 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5660 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5663 Alan Hayward <alan.hayward@arm.com>
5664 David Sherwood <david.sherwood@arm.com>
5666 * postreload.c (move2add_valid_value_p): Change the type of the
5667 mode parameter to scalar_int_mode.
5668 (move2add_use_add2_insn): Add a mode parameter and use it instead
5670 (move2add_use_add3_insn): Likewise.
5671 (reload_cse_move2add): Update accordingly.
5673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5674 Alan Hayward <alan.hayward@arm.com>
5675 David Sherwood <david.sherwood@arm.com>
5677 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5679 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5680 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5687 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5688 (popcount_loc_descriptor): Likewise.
5689 (bswap_loc_descriptor): Likewise.
5690 (rotate_loc_descriptor): Likewise.
5691 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5692 calling the functions above.
5694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5695 Alan Hayward <alan.hayward@arm.com>
5696 David Sherwood <david.sherwood@arm.com>
5698 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5700 (try_combine): Likewise.
5701 (simplify_if_then_else): Likewise.
5702 * cse.c (cse_insn): Likewise.
5703 * dwarf2out.c (mem_loc_descriptor): Likewise.
5704 * emit-rtl.c (gen_lowpart_common): Likewise.
5705 * simplify-rtx.c (simplify_truncation): Likewise.
5706 (simplify_binary_operation_1): Likewise.
5707 (simplify_const_relational_operation): Likewise.
5708 (simplify_ternary_operation): Likewise.
5709 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5712 Alan Hayward <alan.hayward@arm.com>
5713 David Sherwood <david.sherwood@arm.com>
5715 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5716 * reload.c (push_reload): Likewise.
5717 (find_reloads): Likewise.
5719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5720 Alan Hayward <alan.hayward@arm.com>
5721 David Sherwood <david.sherwood@arm.com>
5723 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5724 (make_compound_operation_int): Likewise.
5725 (change_zero_ext): Likewise.
5726 * expr.c (convert_move): Likewise.
5727 (convert_modes): Likewise.
5728 * fwprop.c (forward_propagate_subreg): Likewise.
5729 * loop-iv.c (get_biv_step_1): Likewise.
5730 * optabs.c (widen_operand): Likewise.
5731 * postreload.c (move2add_valid_value_p): Likewise.
5732 * recog.c (simplify_while_replacing): Likewise.
5733 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5734 (simplify_binary_operation_1): Likewise. Remove redundant
5735 mode equality check.
5737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5738 Alan Hayward <alan.hayward@arm.com>
5739 David Sherwood <david.sherwood@arm.com>
5741 * combine.c (combine_simplify_rtx): Add checks for
5742 is_a <scalar_int_mode>.
5743 (simplify_if_then_else): Likewise.
5744 (make_field_assignment): Likewise.
5745 (simplify_comparison): Likewise.
5746 * ifcvt.c (noce_try_bitop): Likewise.
5747 * loop-invariant.c (canonicalize_address_mult): Likewise.
5748 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5751 Alan Hayward <alan.hayward@arm.com>
5752 David Sherwood <david.sherwood@arm.com>
5754 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5755 is_a <scalar_int_mode> instead of != BLKmode.
5757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5758 Alan Hayward <alan.hayward@arm.com>
5759 David Sherwood <david.sherwood@arm.com>
5761 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5762 instead of != VOIDmode.
5763 * combine.c (if_then_else_cond): Likewise.
5764 (change_zero_ext): Likewise.
5765 * dwarf2out.c (mem_loc_descriptor): Likewise.
5766 (loc_descriptor): Likewise.
5767 * rtlanal.c (canonicalize_condition): Likewise.
5768 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5771 Alan Hayward <alan.hayward@arm.com>
5772 David Sherwood <david.sherwood@arm.com>
5774 * simplify-rtx.c (simplify_binary_operation_1): Use
5775 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5778 Alan Hayward <alan.hayward@arm.com>
5779 David Sherwood <david.sherwood@arm.com>
5781 * wide-int.h (int_traits<unsigned char>) New class.
5782 (int_traits<unsigned short>) Likewise.
5783 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5784 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5786 * combine.c (set_nonzero_bits_and_sign_copies): Use
5787 is_a <scalar_int_mode>.
5788 (find_split_point): Likewise.
5789 (combine_simplify_rtx): Likewise.
5790 (simplify_logical): Likewise.
5791 (expand_compound_operation): Likewise.
5792 (expand_field_assignment): Likewise.
5793 (make_compound_operation): Likewise.
5794 (extended_count): Likewise.
5795 (change_zero_ext): Likewise.
5796 (simplify_comparison): Likewise.
5797 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5798 (ucompare_loc_descriptor): Likewise.
5799 (minmax_loc_descriptor): Likewise.
5800 (mem_loc_descriptor): Likewise.
5801 (loc_descriptor): Likewise.
5802 * expmed.c (init_expmed_one_mode): Likewise.
5803 * lra-constraints.c (lra_constraint_offset): Likewise.
5804 * optabs.c (prepare_libcall_arg): Likewise.
5805 * postreload.c (move2add_note_store): Likewise.
5806 * reload.c (operands_match_p): Likewise.
5807 * rtl.h (load_extend_op): Likewise.
5808 * rtlhooks.c (gen_lowpart_general): Likewise.
5809 * simplify-rtx.c (simplify_truncation): Likewise.
5810 (simplify_unary_operation_1): Likewise.
5811 (simplify_binary_operation_1): Likewise.
5812 (simplify_const_binary_operation): Likewise.
5813 (simplify_const_relational_operation): Likewise.
5814 (simplify_subreg): Likewise.
5815 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5816 * var-tracking.c (adjust_mems): Likewise.
5817 (prepare_call_arguments): Likewise.
5819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5820 Alan Hayward <alan.hayward@arm.com>
5821 David Sherwood <david.sherwood@arm.com>
5823 * machmode.h (is_int_mode): New fuction.
5824 * combine.c (find_split_point): Use it.
5825 (combine_simplify_rtx): Likewise.
5826 (simplify_if_then_else): Likewise.
5827 (simplify_set): Likewise.
5828 (simplify_shift_const_1): Likewise.
5829 (simplify_comparison): Likewise.
5830 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5831 * cse.c (notreg_cost): Likewise.
5832 (cse_insn): Likewise.
5833 * cselib.c (cselib_lookup_1): Likewise.
5834 * dojump.c (do_jump_1): Likewise.
5835 (do_compare_rtx_and_jump): Likewise.
5836 * dse.c (get_call_args): Likewise.
5837 * dwarf2out.c (rtl_for_decl_init): Likewise.
5838 (native_encode_initializer): Likewise.
5839 * expmed.c (emit_store_flag_1): Likewise.
5840 (emit_store_flag): Likewise.
5841 * expr.c (convert_modes): Likewise.
5842 (store_field): Likewise.
5843 (expand_expr_real_1): Likewise.
5844 * fold-const.c (fold_read_from_constant_string): Likewise.
5845 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5846 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5847 * optabs.c (expand_binop): Likewise.
5848 (expand_unop): Likewise.
5849 (expand_abs_nojump): Likewise.
5850 (expand_one_cmpl_abs_nojump): Likewise.
5851 * simplify-rtx.c (mode_signbit_p): Likewise.
5852 (val_signbit_p): Likewise.
5853 (val_signbit_known_set_p): Likewise.
5854 (val_signbit_known_clear_p): Likewise.
5855 (simplify_relational_operation_1): Likewise.
5856 * tree.c (vector_type_mode): Likewise.
5858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5859 Alan Hayward <alan.hayward@arm.com>
5860 David Sherwood <david.sherwood@arm.com>
5862 * machmode.h (smallest_mode_for_size): Fix formatting.
5863 (smallest_int_mode_for_size): New function.
5864 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5865 instead of smallest_mode_for_size.
5866 * combine.c (make_extraction): Likewise.
5867 * config/arc/arc.c (arc_expand_movmem): Likewise.
5868 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5869 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5870 * config/s390/s390.c (s390_expand_insv): Likewise.
5871 * config/sparc/sparc.c (assign_int_registers): Likewise.
5872 * config/spu/spu.c (spu_function_value): Likewise.
5873 (spu_function_arg): Likewise.
5874 * coverage.c (get_gcov_type): Likewise.
5875 (get_gcov_unsigned_t): Likewise.
5876 * dse.c (find_shift_sequence): Likewise.
5877 * expmed.c (store_bit_field_1): Likewise.
5878 * expr.c (convert_move): Likewise.
5879 (store_field): Likewise.
5880 * internal-fn.c (expand_arith_overflow): Likewise.
5881 * optabs-query.c (get_best_extraction_insn): Likewise.
5882 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5883 * stor-layout.c (layout_type): Likewise.
5884 (initialize_sizetypes): Likewise.
5885 * targhooks.c (default_get_mask_mode): Likewise.
5886 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5889 Alan Hayward <alan.hayward@arm.com>
5890 David Sherwood <david.sherwood@arm.com>
5892 * machmode.h (opt_mode::else_blk): New function.
5893 (int_mode_for_mode): Declare.
5894 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5895 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5897 * cfgexpand.c (expand_debug_expr): Likewise.
5898 * combine.c (gen_lowpart_or_truncate): Likewise.
5899 (gen_lowpart_for_combine): Likewise.
5900 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5901 * config/avr/avr.c (avr_to_int_mode): Likewise.
5902 (avr_out_plus_1): Likewise.
5903 (avr_out_plus): Likewise.
5904 (avr_out_round): Likewise.
5905 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5906 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5907 (s390_expand_vcond): Likewise.
5908 * config/spu/spu.c (spu_split_immediate): Likewise.
5909 (spu_expand_mov): Likewise.
5910 * dse.c (get_stored_val): Likewise.
5911 * expmed.c (store_bit_field_1): Likewise.
5912 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5914 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5915 (extract_low_bits): Likewise.
5916 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5917 handling rather than repeating the check.
5918 (emit_group_store): Likewise.
5919 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5920 * optabs.c (expand_absneg_bit): Likewise.
5921 (expand_copysign_absneg): Likewise.
5922 (expand_copysign_bit): Likewise.
5923 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5924 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5925 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5926 * var-tracking.c (prepare_call_arguments): Likewise.
5927 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5928 int_mode_for_mode instead of mode_for_size.
5929 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5932 Alan Hayward <alan.hayward@arm.com>
5933 David Sherwood <david.sherwood@arm.com>
5935 * machmode.h (int_mode_for_size): New function.
5936 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5937 instead of mode_for_size.
5938 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5940 * combine.c (expand_field_assignment): Use int_mode_for_size
5941 instead of mode_for_size.
5942 (make_extraction): Likewise.
5943 (simplify_shift_const_1): Likewise.
5944 (simplify_comparison): Likewise.
5945 * dojump.c (do_jump): Likewise.
5946 * dwarf2out.c (mem_loc_descriptor): Likewise.
5947 * emit-rtl.c (init_derived_machine_modes): Likewise.
5948 * expmed.c (flip_storage_order): Likewise.
5949 (convert_extracted_bit_field): Likewise.
5950 * expr.c (copy_blkmode_from_reg): Likewise.
5951 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5952 * internal-fn.c (expand_mul_overflow): Likewise.
5953 * lower-subreg.c (simple_move): Likewise.
5954 * optabs-libfuncs.c (init_optabs): Likewise.
5955 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5956 * tree.c (vector_type_mode): Likewise.
5957 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5958 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5959 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5960 * tree-vect-stmts.c (vectorizable_load): Likewise.
5961 (vectorizable_store): Likewise.
5963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5964 Alan Hayward <alan.hayward@arm.com>
5965 David Sherwood <david.sherwood@arm.com>
5967 * coretypes.h (pod_mode): New type.
5968 (scalar_int_mode_pod): New typedef.
5969 * machmode.h (pod_mode): New class.
5970 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5971 * genmodes.c (emit_mode_int_n): Update accordingly.
5972 * lower-subreg.h (target_lower_subreg): Change type to
5973 scalar_int_mode_pod.
5974 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5975 scalar_int_mode_pod.
5977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5978 Alan Hayward <alan.hayward@arm.com>
5979 David Sherwood <david.sherwood@arm.com>
5981 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5982 machine_mode to scalar_int_mode.
5983 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5984 (rs6000_option_override_internal): Remove cast to int.
5985 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5986 machine_mode to scalar_int_mode.
5987 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5988 (rs6000_option_override_internal): Remove cast to int.
5989 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5990 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5992 * config/s390/s390.c (s390_expand_builtin): Likewise.
5993 * coretypes.h (scalar_int_mode): New type.
5994 (opt_scalar_int_mode): New typedef.
5995 * machmode.h (scalar_int_mode): New class.
5996 (scalar_int_mode::includes_p): New function.
5997 (byte_mode): Change type to scalar_int_mode.
5998 (word_mode): Likewise.
5999 (ptr_mode): Likewise.
6000 * emit-rtl.c (byte_mode): Likewise.
6001 (word_mode): Likewise.
6002 (ptr_mode): Likewise.
6003 (init_derived_machine_modes): Update accordingly.
6004 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6005 and MODE_PARTIAL_INT.
6006 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6007 opt_scalar_int_mode.
6009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6010 Alan Hayward <alan.hayward@arm.com>
6011 David Sherwood <david.sherwood@arm.com>
6013 * target.def (libgcc_floating_mode_supported_p): Take a
6015 * doc/tm.texi: Regenerate.
6016 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6018 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6019 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6023 Alan Hayward <alan.hayward@arm.com>
6024 David Sherwood <david.sherwood@arm.com>
6026 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6027 * doc/tm.texi: Regenerate.
6028 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6029 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6030 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6031 * targhooks.h (default_floatn_mode): Likewise.
6032 * targhooks.c (default_floatn_mode): Likewise.
6033 * tree.c (build_common_tree_nodes): Update accordingly.
6035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6036 Alan Hayward <alan.hayward@arm.com>
6037 David Sherwood <david.sherwood@arm.com>
6039 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6040 (mode_iterator::iterate_p): Likewise.
6041 (mode_iterator::get_wider): Likewise.
6042 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
6049 * machmode.h (float_mode_for_size): New function.
6050 * emit-rtl.c (double_mode): Delete.
6051 (init_emit_once): Use float_mode_for_size.
6052 * stor-layout.c (layout_type): Likewise.
6053 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6056 Alan Hayward <alan.hayward@arm.com>
6057 David Sherwood <david.sherwood@arm.com>
6059 * output.h (assemble_real): Take a scalar_float_mode.
6060 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6061 * config/arm/arm.md (consttable_4): Likewise.
6062 (consttable_8): Likewise.
6063 (consttable_16): Likewise.
6064 * config/mips/mips.md (consttable_float): Likewise.
6065 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6066 * varasm.c (assemble_real): Take a scalar_float_mode.
6067 (output_constant_pool_2): Update accordingly.
6068 (output_constant): Likewise.
6070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6071 Alan Hayward <alan.hayward@arm.com>
6072 David Sherwood <david.sherwood@arm.com>
6074 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6075 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6076 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6077 (native_encode_real): Likewise.
6078 (native_interpret_real): Likewise.
6079 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6080 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6083 Alan Hayward <alan.hayward@arm.com>
6084 David Sherwood <david.sherwood@arm.com>
6086 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6087 <scalar_float_mode>. Simplify.
6088 (gen_extend_conv_libfunc): Likewise.
6090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6091 Alan Hayward <alan.hayward@arm.com>
6092 David Sherwood <david.sherwood@arm.com>
6094 * coretypes.h (scalar_float_mode): New type.
6095 * machmode.h (mode_traits::from_int): Use machine_mode if
6096 USE_ENUM_MODES is defined.
6097 (is_a): New function.
6099 (dyn_cast): Likewise.
6100 (scalar_float_mode): New class.
6101 (scalar_float_mode::includes_p): New function.
6102 (is_float_mode): Likewise.
6103 * gdbhooks.py (MachineModePrinter): New class.
6104 (build_pretty_printer): Use it for scalar_float_mode.
6105 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6106 (format_helper::format_helper): Turn into a template.
6107 * genmodes.c (get_mode_class): New function.
6108 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6109 or machine_mode if none.
6110 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6111 as_a <scalar_float_mode>.
6112 * dwarf2out.c (mem_loc_descriptor): Likewise.
6113 (insert_float): Likewise.
6114 (add_const_value_attribute): Likewise.
6115 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6116 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6117 (expand_unop): Update accordingly.
6118 (expand_abs_nojump): Likewise.
6119 (expand_copysign_absneg): Take a scalar_float_mode.
6120 (expand_copysign_bit): Likewise.
6121 (expand_copysign): Update accordingly.
6123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6124 Alan Hayward <alan.hayward@arm.com>
6125 David Sherwood <david.sherwood@arm.com>
6127 * coretypes.h (opt_mode): New class.
6128 * machmode.h (opt_mode): Likewise.
6129 (opt_mode::else_void): New function.
6130 (opt_mode::require): Likewise.
6131 (opt_mode::exists): Likewise.
6132 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6133 (GET_MODE_2XWIDER_MODE): Likewise.
6134 (mode_iterator::get_wider): Update accordingly.
6135 (mode_iterator::get_2xwider): Likewise.
6136 (mode_iterator::get_known_wider): Likewise, turning into a template.
6137 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6138 forcing a wider mode to exist.
6139 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6140 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6141 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6142 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6143 * lower-subreg.c (init_lower_subreg): Likewise.
6144 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6145 on the final iteration.
6146 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6147 a wider mode exists before asking for a move pattern.
6148 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6149 forcing a wider mode to exist.
6150 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6151 returning false if no such mode exists.
6152 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6153 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6154 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6155 Avoid checking for a MODE_INT if we already know the mode is not a
6157 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6158 forcing a wider mode to exist.
6159 (expmed_mult_highpart_optab): Likewise.
6160 (expmed_mult_highpart): Likewise.
6161 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6163 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6164 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6165 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6166 * internal-fn.c (expand_mul_overflow): Update use of
6167 GET_MODE_2XWIDER_MODE.
6168 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6169 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6170 GET_MODE_WIDER_MODE.
6171 (convert_plusminus_to_widen): Likewise.
6172 * tree-switch-conversion.c (array_value_type): Likewise.
6173 * var-tracking.c (emit_note_insn_var_location): Likewise.
6174 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6175 Return false inside rather than outside the loop if no wider mode
6177 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6178 and GET_MODE_2XWIDER_MODE
6179 (can_compare_p): Use else_void.
6180 * gdbhooks.py (OptMachineModePrinter): New class.
6181 (build_pretty_printer): Use it for opt_mode.
6183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6184 Alan Hayward <alan.hayward@arm.com>
6185 David Sherwood <david.sherwood@arm.com>
6187 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6188 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6191 Alan Hayward <alan.hayward@arm.com>
6192 David Sherwood <david.sherwood@arm.com>
6194 * machmode.h (mode_traits): New structure.
6195 (get_narrowest_mode): New function.
6196 (mode_iterator::start): Likewise.
6197 (mode_iterator::iterate_p): Likewise.
6198 (mode_iterator::get_wider): Likewise.
6199 (mode_iterator::get_known_wider): Likewise.
6200 (mode_iterator::get_2xwider): Likewise.
6201 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6202 (FOR_EACH_MODE): Likewise.
6203 (FOR_EACH_MODE_FROM): Likewise.
6204 (FOR_EACH_MODE_UNTIL): Likewise.
6205 (FOR_EACH_WIDER_MODE): Likewise.
6206 (FOR_EACH_2XWIDER_MODE): Likewise.
6207 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6208 * combine.c (simplify_comparison): Likewise
6209 * config/i386/i386.c (type_natural_mode): Likewise.
6210 * cse.c (cse_insn): Likewise.
6211 * dse.c (find_shift_sequence): Likewise.
6212 * emit-rtl.c (init_derived_machine_modes): Likewise.
6213 (init_emit_once): Likewise.
6214 * explow.c (hard_function_value): Likewise.
6215 * expmed.c (extract_fixed_bit_field_1): Likewise.
6216 (extract_bit_field_1): Likewise.
6217 (expand_divmod): Likewise.
6218 (emit_store_flag_1): Likewise.
6219 * expr.c (init_expr_target): Likewise.
6220 (convert_move): Likewise.
6221 (alignment_for_piecewise_move): Likewise.
6222 (widest_int_mode_for_size): Likewise.
6223 (emit_block_move_via_movmem): Likewise.
6224 (copy_blkmode_to_reg): Likewise.
6225 (set_storage_via_setmem): Likewise.
6226 (compress_float_constant): Likewise.
6227 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6228 * optabs-query.c (get_best_extraction_insn): Likewise.
6229 * optabs.c (expand_binop): Likewise.
6230 (expand_twoval_unop): Likewise.
6231 (expand_twoval_binop): Likewise.
6232 (widen_leading): Likewise.
6233 (widen_bswap): Likewise.
6234 (expand_parity): Likewise.
6235 (expand_unop): Likewise.
6236 (prepare_cmp_insn): Likewise.
6237 (prepare_float_lib_cmp): Likewise.
6238 (expand_float): Likewise.
6239 (expand_fix): Likewise.
6240 (expand_sfix_optab): Likewise.
6241 * postreload.c (move2add_use_add2_insn): Likewise.
6242 * reg-stack.c (reg_to_stack): Likewise.
6243 * reginfo.c (choose_hard_reg_mode): Likewise.
6244 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6245 * stor-layout.c (mode_for_size): Likewise.
6246 (smallest_mode_for_size): Likewise.
6247 (mode_for_vector): Likewise.
6248 (finish_bitfield_representative): Likewise.
6249 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6250 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6251 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6252 * var-tracking.c (prepare_call_arguments): Likewise.
6254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6255 Alan Hayward <alan.hayward@arm.com>
6256 David Sherwood <david.sherwood@arm.com>
6258 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6259 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6260 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6261 * machmode.h (mode_size): Move earlier in file.
6262 (mode_precision): Likewise.
6263 (mode_inner): Likewise.
6264 (mode_nunits): Likewise.
6265 (mode_unit_size): Likewise.
6266 (unit_unit_precision): Likewise.
6267 (mode_wider): Likewise.
6268 (mode_2xwider): Likewise.
6269 (machine_mode): New class.
6270 (mode_to_bytes): New function.
6271 (mode_to_bits): Likewise.
6272 (mode_to_precision): Likewise.
6273 (mode_to_inner): Likewise.
6274 (mode_to_unit_size): Likewise.
6275 (mode_to_unit_precision): Likewise.
6276 (mode_to_nunits): Likewise.
6277 (GET_MODE_SIZE): Use mode_to_bytes.
6278 (GET_MODE_BITSIZE): Use mode_to_bits.
6279 (GET_MODE_PRECISION): Use mode_to_precision.
6280 (GET_MODE_INNER): Use mode_to_inner.
6281 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6282 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6283 (GET_MODE_NUNITS): Use mode_to_nunits.
6284 * system.h (ALWAYS_INLINE): New macro.
6285 * config/powerpcspe/powerpcspe-c.c
6286 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6287 int for arg1_mode and arg2_mode.
6289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6290 Alan Hayward <alan.hayward@arm.com>
6291 David Sherwood <david.sherwood@arm.com>
6293 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6294 Prefix mode names with E_ in case statements.
6295 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6296 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6297 (aarch64_split_simd_move): Likewise.
6298 (aarch64_gen_storewb_pair): Likewise.
6299 (aarch64_gen_loadwb_pair): Likewise.
6300 (aarch64_gen_store_pair): Likewise.
6301 (aarch64_gen_load_pair): Likewise.
6302 (aarch64_get_condition_code_1): Likewise.
6303 (aarch64_constant_pool_reload_icode): Likewise.
6304 (get_rsqrte_type): Likewise.
6305 (get_rsqrts_type): Likewise.
6306 (get_recpe_type): Likewise.
6307 (get_recps_type): Likewise.
6308 (aarch64_gimplify_va_arg_expr): Likewise.
6309 (aarch64_simd_container_mode): Likewise.
6310 (aarch64_emit_load_exclusive): Likewise.
6311 (aarch64_emit_store_exclusive): Likewise.
6312 (aarch64_expand_compare_and_swap): Likewise.
6313 (aarch64_gen_atomic_cas): Likewise.
6314 (aarch64_emit_bic): Likewise.
6315 (aarch64_emit_atomic_swap): Likewise.
6316 (aarch64_emit_atomic_load_op): Likewise.
6317 (aarch64_evpc_trn): Likewise.
6318 (aarch64_evpc_uzp): Likewise.
6319 (aarch64_evpc_zip): Likewise.
6320 (aarch64_evpc_ext): Likewise.
6321 (aarch64_evpc_rev): Likewise.
6322 (aarch64_evpc_dup): Likewise.
6323 (aarch64_gen_ccmp_first): Likewise.
6324 (aarch64_gen_ccmp_next): Likewise.
6325 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6326 (alpha_emit_xfloating_libcall): Likewise.
6327 (emit_insxl): Likewise.
6328 (alpha_arg_type): Likewise.
6329 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6330 (arc_preferred_simd_mode): Likewise.
6331 (arc_secondary_reload): Likewise.
6332 (get_arc_condition_code): Likewise.
6333 (arc_print_operand): Likewise.
6334 (arc_legitimate_constant_p): Likewise.
6335 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6336 * config/arc/arc.md (casesi_load): Likewise.
6337 (casesi_compact_jump): Likewise.
6338 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6339 (cc_use_register): Likewise.
6340 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6341 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6342 (arm_init_iwmmxt_builtins): Likewise.
6343 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6344 (neon_expand_vector_init): Likewise.
6345 (arm_attr_length_move_neon): Likewise.
6346 (maybe_get_arm_condition_code): Likewise.
6347 (arm_emit_vector_const): Likewise.
6348 (arm_preferred_simd_mode): Likewise.
6349 (arm_output_iwmmxt_tinsr): Likewise.
6350 (thumb1_output_casesi): Likewise.
6351 (thumb2_output_casesi): Likewise.
6352 (arm_emit_load_exclusive): Likewise.
6353 (arm_emit_store_exclusive): Likewise.
6354 (arm_expand_compare_and_swap): Likewise.
6355 (arm_evpc_neon_vuzp): Likewise.
6356 (arm_evpc_neon_vzip): Likewise.
6357 (arm_evpc_neon_vrev): Likewise.
6358 (arm_evpc_neon_vtrn): Likewise.
6359 (arm_evpc_neon_vext): Likewise.
6360 (arm_validize_comparison): Likewise.
6361 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6362 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6363 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6364 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6365 (c6x_preferred_simd_mode): Likewise.
6366 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6367 (epiphany_rtx_costs): Likewise.
6368 * config/epiphany/predicates.md (proper_comparison_operator):
6370 * config/frv/frv.c (condexec_memory_operand): Likewise.
6371 (frv_emit_move): Likewise.
6372 (output_move_single): Likewise.
6373 (output_condmove_single): Likewise.
6374 (frv_hard_regno_mode_ok): Likewise.
6375 (frv_matching_accg_mode): Likewise.
6376 * config/h8300/h8300.c (split_adds_subs): Likewise.
6377 (h8300_rtx_costs): Likewise.
6378 (h8300_print_operand): Likewise.
6379 (compute_mov_length): Likewise.
6380 (output_logical_op): Likewise.
6381 (compute_logical_op_length): Likewise.
6382 (compute_logical_op_cc): Likewise.
6383 (h8300_shift_needs_scratch_p): Likewise.
6384 (output_a_shift): Likewise.
6385 (compute_a_shift_length): Likewise.
6386 (compute_a_shift_cc): Likewise.
6387 (expand_a_rotate): Likewise.
6388 (output_a_rotate): Likewise.
6389 * config/i386/i386.c (classify_argument): Likewise.
6390 (function_arg_advance_32): Likewise.
6391 (function_arg_32): Likewise.
6392 (function_arg_64): Likewise.
6393 (function_value_64): Likewise.
6394 (ix86_gimplify_va_arg): Likewise.
6395 (ix86_legitimate_constant_p): Likewise.
6396 (put_condition_code): Likewise.
6397 (split_double_mode): Likewise.
6398 (ix86_avx256_split_vector_move_misalign): Likewise.
6399 (ix86_expand_vector_logical_operator): Likewise.
6400 (ix86_split_idivmod): Likewise.
6401 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6402 (ix86_build_const_vector): Likewise.
6403 (ix86_build_signbit_mask): Likewise.
6404 (ix86_match_ccmode): Likewise.
6405 (ix86_cc_modes_compatible): Likewise.
6406 (ix86_expand_branch): Likewise.
6407 (ix86_expand_sse_cmp): Likewise.
6408 (ix86_expand_sse_movcc): Likewise.
6409 (ix86_expand_int_sse_cmp): Likewise.
6410 (ix86_expand_vec_perm_vpermi2): Likewise.
6411 (ix86_expand_vec_perm): Likewise.
6412 (ix86_expand_sse_unpack): Likewise.
6413 (ix86_expand_int_addcc): Likewise.
6414 (ix86_split_to_parts): Likewise.
6415 (ix86_vectorize_builtin_gather): Likewise.
6416 (ix86_vectorize_builtin_scatter): Likewise.
6417 (avx_vpermilp_parallel): Likewise.
6418 (inline_memory_move_cost): Likewise.
6419 (ix86_tieable_integer_mode_p): Likewise.
6420 (x86_maybe_negate_const_int): Likewise.
6421 (ix86_expand_vector_init_duplicate): Likewise.
6422 (ix86_expand_vector_init_one_nonzero): Likewise.
6423 (ix86_expand_vector_init_one_var): Likewise.
6424 (ix86_expand_vector_init_concat): Likewise.
6425 (ix86_expand_vector_init_interleave): Likewise.
6426 (ix86_expand_vector_init_general): Likewise.
6427 (ix86_expand_vector_set): Likewise.
6428 (ix86_expand_vector_extract): Likewise.
6429 (emit_reduc_half): Likewise.
6430 (ix86_emit_i387_round): Likewise.
6431 (ix86_mangle_type): Likewise.
6432 (ix86_expand_round_sse4): Likewise.
6433 (expand_vec_perm_blend): Likewise.
6434 (canonicalize_vector_int_perm): Likewise.
6435 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6436 (expand_vec_perm_1): Likewise.
6437 (expand_vec_perm_interleave3): Likewise.
6438 (expand_vec_perm_even_odd_pack): Likewise.
6439 (expand_vec_perm_even_odd_1): Likewise.
6440 (expand_vec_perm_broadcast_1): Likewise.
6441 (ix86_vectorize_vec_perm_const_ok): Likewise.
6442 (ix86_expand_vecop_qihi): Likewise.
6443 (ix86_expand_mul_widen_hilo): Likewise.
6444 (ix86_expand_sse2_abs): Likewise.
6445 (ix86_expand_pextr): Likewise.
6446 (ix86_expand_pinsr): Likewise.
6447 (ix86_preferred_simd_mode): Likewise.
6448 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6449 * config/i386/sse.md (*andnot<mode>3): Likewise.
6450 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6451 (*<code><mode>3): Likewise.
6452 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6453 (ia64_expand_atomic_op): Likewise.
6454 (ia64_arg_type): Likewise.
6455 (ia64_mode_to_int): Likewise.
6456 (ia64_scalar_mode_supported_p): Likewise.
6457 (ia64_vector_mode_supported_p): Likewise.
6458 (expand_vec_perm_broadcast): Likewise.
6459 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6460 (iq2000_function_arg_advance): Likewise.
6461 (iq2000_function_arg): Likewise.
6462 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6463 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6464 (m68k_libcall_value): Likewise.
6465 (m68k_function_value): Likewise.
6466 (sched_attr_op_type): Likewise.
6467 * config/mcore/mcore.c (mcore_output_move): Likewise.
6468 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6470 (microblaze_function_arg): Likewise.
6471 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6472 (mips_print_operand): Likewise.
6473 (mips_mode_ok_for_mov_fmt_p): Likewise.
6474 (mips_vector_mode_supported_p): Likewise.
6475 (mips_preferred_simd_mode): Likewise.
6476 (mips_expand_vpc_loongson_even_odd): Likewise.
6477 (mips_expand_vec_unpack): Likewise.
6478 (mips_expand_vi_broadcast): Likewise.
6479 (mips_expand_vector_init): Likewise.
6480 (mips_expand_vec_reduc): Likewise.
6481 (mips_expand_msa_cmp): Likewise.
6482 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6483 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6484 (cc_flags_for_mode): Likewise.
6485 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6486 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6487 (nds32_output_casesi_pc_relative): Likewise.
6488 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6489 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6490 (nvptx_gen_unpack): Likewise.
6491 (nvptx_gen_pack): Likewise.
6492 (nvptx_gen_shuffle): Likewise.
6493 (nvptx_gen_wcast): Likewise.
6494 (nvptx_preferred_simd_mode): Likewise.
6495 * config/pa/pa.c (pa_secondary_reload): Likewise.
6496 * config/pa/predicates.md (base14_operand): Likewise.
6497 * config/powerpcspe/powerpcspe-c.c
6498 (altivec_resolve_overloaded_builtin): Likewise.
6499 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6501 (rs6000_preferred_simd_mode): Likewise.
6502 (output_vec_const_move): Likewise.
6503 (rs6000_expand_vector_extract): Likewise.
6504 (rs6000_split_vec_extract_var): Likewise.
6505 (reg_offset_addressing_ok_p): Likewise.
6506 (rs6000_legitimate_offset_address_p): Likewise.
6507 (rs6000_legitimize_address): Likewise.
6508 (rs6000_emit_set_const): Likewise.
6509 (rs6000_const_vec): Likewise.
6510 (rs6000_emit_move): Likewise.
6511 (spe_build_register_parallel): Likewise.
6512 (rs6000_darwin64_record_arg_recurse): Likewise.
6513 (swap_selector_for_mode): Likewise.
6514 (spe_init_builtins): Likewise.
6515 (paired_init_builtins): Likewise.
6516 (altivec_init_builtins): Likewise.
6517 (do_load_for_compare): Likewise.
6518 (rs6000_generate_compare): Likewise.
6519 (rs6000_expand_float128_convert): Likewise.
6520 (emit_load_locked): Likewise.
6521 (emit_store_conditional): Likewise.
6522 (rs6000_output_function_epilogue): Likewise.
6523 (rs6000_handle_altivec_attribute): Likewise.
6524 (rs6000_function_value): Likewise.
6525 (emit_fusion_gpr_load): Likewise.
6526 (emit_fusion_p9_load): Likewise.
6527 (emit_fusion_p9_store): Likewise.
6528 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6529 (fusion_gpr_mem_load): Likewise.
6530 (fusion_addis_mem_combo_load): Likewise.
6531 (fusion_addis_mem_combo_store): Likewise.
6532 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6533 (fusion_gpr_mem_load): Likewise.
6534 (fusion_addis_mem_combo_load): Likewise.
6535 (fusion_addis_mem_combo_store): Likewise.
6536 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6538 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6539 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6540 (rs6000_preferred_simd_mode): Likewise.
6541 (output_vec_const_move): Likewise.
6542 (rs6000_expand_vector_extract): Likewise.
6543 (rs6000_split_vec_extract_var): Likewise.
6544 (reg_offset_addressing_ok_p): Likewise.
6545 (rs6000_legitimate_offset_address_p): Likewise.
6546 (rs6000_legitimize_address): Likewise.
6547 (rs6000_emit_set_const): Likewise.
6548 (rs6000_const_vec): Likewise.
6549 (rs6000_emit_move): Likewise.
6550 (rs6000_darwin64_record_arg_recurse): Likewise.
6551 (swap_selector_for_mode): Likewise.
6552 (paired_init_builtins): Likewise.
6553 (altivec_init_builtins): Likewise.
6554 (rs6000_expand_float128_convert): Likewise.
6555 (emit_load_locked): Likewise.
6556 (emit_store_conditional): Likewise.
6557 (rs6000_output_function_epilogue): Likewise.
6558 (rs6000_handle_altivec_attribute): Likewise.
6559 (rs6000_function_value): Likewise.
6560 (emit_fusion_gpr_load): Likewise.
6561 (emit_fusion_p9_load): Likewise.
6562 (emit_fusion_p9_store): Likewise.
6563 * config/rx/rx.c (rx_gen_move_template): Likewise.
6564 (flags_from_mode): Likewise.
6565 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6566 (s390_slb_comparison): Likewise.
6567 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6568 (s390_vector_mode_supported_p): Likewise.
6569 (s390_cc_modes_compatible): Likewise.
6570 (s390_match_ccmode_set): Likewise.
6571 (s390_canonicalize_comparison): Likewise.
6572 (s390_emit_compare_and_swap): Likewise.
6573 (s390_branch_condition_mask): Likewise.
6574 (s390_rtx_costs): Likewise.
6575 (s390_secondary_reload): Likewise.
6576 (__SECONDARY_RELOAD_CASE): Likewise.
6577 (s390_expand_cs): Likewise.
6578 (s390_preferred_simd_mode): Likewise.
6579 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6580 * config/sh/sh.c (sh_print_operand): Likewise.
6581 (dump_table): Likewise.
6582 (sh_secondary_reload): Likewise.
6583 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6584 * config/sh/sh.md (casesi_worker_1): Likewise.
6585 (casesi_worker_2): Likewise.
6586 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6587 (fcc_comparison_operator): Likewise.
6588 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6589 (emit_soft_tfmode_cvt): Likewise.
6590 (sparc_preferred_simd_mode): Likewise.
6591 (output_cbranch): Likewise.
6592 (sparc_print_operand): Likewise.
6593 (sparc_expand_vec_perm_bmask): Likewise.
6594 (vector_init_bshuffle): Likewise.
6595 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6596 (spu_vector_mode_supported_p): Likewise.
6597 (spu_expand_insv): Likewise.
6598 (spu_emit_branch_or_set): Likewise.
6599 (spu_handle_vector_attribute): Likewise.
6600 (spu_builtin_splats): Likewise.
6601 (spu_builtin_extract): Likewise.
6602 (spu_builtin_promote): Likewise.
6603 (spu_expand_sign_extend): Likewise.
6604 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6605 (tilegx_simd_int): Likewise.
6606 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6607 (tilepro_simd_int): Likewise.
6608 * config/v850/v850.c (const_double_split): Likewise.
6609 (v850_print_operand): Likewise.
6610 (ep_memory_offset): Likewise.
6611 * config/vax/vax.c (vax_rtx_costs): Likewise.
6612 (vax_output_int_move): Likewise.
6613 (vax_output_int_add): Likewise.
6614 (vax_output_int_subtract): Likewise.
6615 * config/visium/predicates.md (visium_branch_operator): Likewise.
6616 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6617 (visium_print_operand_address): Likewise.
6618 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6619 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6620 (xtensa_expand_conditional_branch): Likewise.
6621 (xtensa_copy_incoming_a7): Likewise.
6622 (xtensa_output_literal): Likewise.
6623 * dfp.c (decimal_real_maxval): Likewise.
6624 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6627 Alan Hayward <alan.hayward@arm.com>
6628 David Sherwood <david.sherwood@arm.com>
6630 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6631 (mode_nunits_inline): Likewise.
6632 (mode_inner_inline): Likewise.
6633 (mode_unit_size_inline): Likewise.
6634 (mode_unit_precision_inline): Likewise.
6635 (emit_insn_modes_h): Likewise. Also emit a #define of the
6637 (emit_mode_wider): Add an E_ prefix to mode names.
6638 (emit_mode_complex): Likewise.
6639 (emit_mode_inner): Likewise.
6640 (emit_mode_adjustments): Likewise.
6641 (emit_mode_int_n): Likewise.
6642 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6643 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6644 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6645 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6646 (CRC32_BUILTIN, ENTRY): Likewise.
6647 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6648 (aarch64_pop_regs): Likewise.
6649 (aarch64_process_components): Likewise.
6650 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6651 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6652 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6653 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6654 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6655 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6657 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6658 (pcmpistr): Likewise.
6659 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6660 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6661 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6663 * config/rl78/rl78.c (mduc_regs): Likewise.
6664 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6665 (htm_expand_builtin): Likewise.
6666 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6667 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6668 * config/xtensa/xtensa.c (print_operand): Likewise.
6669 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6670 (NUM_MODE_VECTOR_INT): Likewise.
6671 * genoutput.c (null_operand): Likewise.
6672 (output_operand_data): Likewise.
6673 * genrecog.c (print_parameter_value): Likewise.
6674 * lra.c (debug_operand_data): Likewise.
6676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6678 * dwarf2out.c (scompare_loc_descriptor_wide)
6679 (scompare_loc_descriptor_narrow): New functions, split out from...
6680 (scompare_loc_descriptor): ...here.
6681 * expmed.c (emit_store_flag_int): New function, split out from...
6682 (emit_store_flag): ...here.
6684 2017-08-30 Richard Biener <rguenther@suse.de>
6686 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6687 (dwarf2out_early_finish): Move setting of AT_pubnames from
6688 early debug output to early finish.
6690 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6692 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6693 and -mdata-region to the assembler.
6695 2017-08-30 Richard Biener <rguenther@suse.de>
6697 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6699 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6700 (dwarf2out_early_global_decl): Only generate a DIE for the
6701 abstract origin if it doesn't already exist or is a declaration DIE.
6702 (resolve_addr): Do not add the linkage name twice when
6703 generating a stub DIE for the DW_TAG_GNU_call_site target.
6705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6707 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6708 Use machine_mode rather than int for arg1_mode.
6710 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6713 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6714 that the second argument of the built-in functions to unpack
6715 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6716 switch statement instead a lot of if statements.
6717 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6718 Allow 64-bit values to be in Altivec registers as well as
6719 traditional floating point registers.
6720 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6722 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6724 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6725 MAX_REGS_PER_ADDRESS == 1.
6727 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6729 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6730 * config/i386/i386.c (ix86_option_override_internal): Simplify
6731 setting of opts->x_flag_entry.
6733 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6734 Jakub Jelinek <jakub@redhat.com>
6735 Richard Biener <rguenther@suse.de>
6737 PR tree-optimization/81503
6738 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6739 folded constant fits in the target type; reorder tests for clarity.
6741 2017-08-29 Martin Liska <mliska@suse.cz>
6743 * passes.def: Include pass_lower_switch.
6744 * stmt.c (dump_case_nodes): Remove and move to
6745 tree-switch-conversion.
6746 (case_values_threshold): Likewise.
6747 (expand_switch_as_decision_tree_p): Likewise.
6748 (emit_case_decision_tree): Likewise.
6749 (expand_case): Likewise.
6750 (balance_case_nodes): Likewise.
6751 (node_has_low_bound): Likewise.
6752 (node_has_high_bound): Likewise.
6753 (node_is_bounded): Likewise.
6754 (emit_case_nodes): Likewise.
6755 (struct simple_case_node): New struct.
6756 (add_case_node): Remove.
6757 (emit_case_dispatch_table): Use vector instead of case_list.
6758 (reset_out_edges_aux): Remove.
6759 (compute_cases_per_edge): Likewise.
6760 (expand_case): Build list of simple_case_node.
6761 (expand_sjlj_dispatch_table): Use it.
6762 * tree-switch-conversion.c (struct case_node): Moved from
6763 stmt.c and adjusted.
6764 (emit_case_nodes): Likewise.
6765 (node_has_low_bound): Likewise.
6766 (node_has_high_bound): Likewise.
6767 (node_is_bounded): Likewise.
6768 (case_values_threshold): Likewise.
6769 (reset_out_edges_aux): Likewise.
6770 (compute_cases_per_edge): Likewise.
6771 (add_case_node): Likewise.
6772 (dump_case_nodes): Likewise.
6773 (balance_case_nodes): Likewise.
6774 (expand_switch_as_decision_tree_p): Likewise.
6775 (emit_jump): Likewise.
6776 (emit_case_decision_tree): Likewise.
6777 (try_switch_expansion): Likewise.
6778 (do_jump_if_equal): Likewise.
6779 (emit_cmp_and_jump_insns): Likewise.
6780 (fix_phi_operands_for_edge): New function.
6781 (record_phi_operand_mapping): Likewise.
6782 (class pass_lower_switch): New pass.
6783 (pass_lower_switch::execute): New function.
6784 (make_pass_lower_switch): Likewise.
6785 (conditional_probability):
6786 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6787 * tree-pass.h: Add make_pass_lower_switch.
6789 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6792 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6795 2017-08-29 Richard Biener <rguenther@suse.de>
6797 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6798 we do not add a DW_AT_inline attribute twice.
6799 (gen_subprogram_die): Remove code setting DW_AT_inline on
6800 DECL_ABSTRACT_P nodes.
6802 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6804 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6805 calls to internal functions.
6806 (gimplify_modify_expr): Likewise.
6807 * tree-call-cdce.c (use_internal_fn): Likewise.
6808 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6809 (convert_to_divmod): Set the nothrow flag.
6810 * tree-if-conv.c (predicate_mem_writes): Likewise.
6811 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6812 (vectorizable_call): Likewise.
6813 (vectorizable_store): Likewise.
6814 (vectorizable_load): Likewise.
6815 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6816 (vect_recog_mask_conversion_pattern): Likewise.
6818 2017-08-29 Martin Liska <mliska@suse.cz>
6821 * gcc.c (driver_handle_option): Add new argument.
6822 * opts-common.c (handle_option): Pass
6823 target_option_override_hook.
6824 * opts-global.c (lang_handle_option): Add new option.
6825 (set_default_handlers): Add new argument.
6826 (decode_options): Likewise.
6827 * opts.c (target_handle_option): Likewise.
6828 (common_handle_option): Call target_option_override_hook.
6829 * opts.h (struct cl_option_handler_func): Add hook for
6830 target option override.
6831 (struct cl_option_handlers): Likewise.
6832 (set_default_handlers): Add new argument.
6833 (decode_options): Likewise.
6834 (common_handle_option): Likewise.
6835 (target_handle_option): Likewise.
6836 * toplev.c (toplev::main): Pass targetm.target_option.override
6839 2017-08-29 Richard Biener <rguenther@suse.de>
6840 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6842 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6843 life to the active subtree.
6845 2017-08-28 Jeff Law <law@redhat.com>
6847 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6848 derive_equivalences.
6849 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6851 (edge_info::derive_equivalences): New private member function
6853 * tree-ssa-dom.c (class edge_info): Changed from a struct
6854 to a class. Add ctor/dtor, methods and data members.
6855 (edge_info::edge_info): Renamed from allocate_edge_info.
6856 Initialize additional members.
6857 (edge_info::~edge_info): New.
6858 (free_dom_edge_info): Delete the edge info.
6859 (record_edge_info): Use new class & associated member functions.
6860 Tighten forms for testing for edge equivalences.
6861 (record_temporary_equivalences): Iterate over the simple
6862 equivalences rather than assuming there's only one per edge.
6863 (cprop_into_successor_phis): Iterate over the simple
6864 equivalences rather than assuming there's only one per edge.
6865 (optimize_stmt): Use operand_equal_p rather than pointer
6866 equality for mini-DSE code.
6868 2017-08-28 Nathan Sidwell <nathan@acm.org>
6870 * gcc.c (execute): Fold SIGPIPE handling into switch
6871 statement. Adjust internal error message.
6873 2017-08-28 Richard Biener <rguenther@suse.de>
6876 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6877 Do nothing for removed DIEs.
6879 2017-08-28 Richard Biener <rguenther@suse.de>
6881 PR tree-optimization/81977
6882 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6885 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6888 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6889 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6890 using targetm.gen_mem_thread_fence.
6892 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6895 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6896 predicate to register_operand. Reorder operands.
6897 (*btr<mode>): Ditto.
6898 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6899 (*btr<mode>_mask): Ditto.
6901 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6903 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6904 * config/rs6000/xmmintrin.h: New file.
6905 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6907 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6910 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6911 parameter and_insn and return it.
6912 (recombine_lvx_pattern): Insert a copy to ensure availability of
6913 the base register of the copied masking operation at the point of
6914 the instruction replacement.
6915 (recombine_stvx_pattern): Likewise.
6917 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6919 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6920 undocumented switches.
6921 (-mpower9-dform-vector): Likewise.
6922 (-mpower9-dform): Likewise.
6923 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6924 comments to delete references to -mpower9-dform* switches.
6925 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6926 Delete reference to -mpower9-dform* switches, test for
6927 -mpower9-vector instead.
6928 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6929 (OTHER_P9_VECTOR_MASKS): Likewise.
6930 (POWERPC_MASKS): Likewise.
6931 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6932 tests against -mpower9-dform* to -mpower9-vector. Delete code
6933 that checked for -mpower9-dform* consistancy with other options.
6934 Add test for -mpower9-misc to enable other power9 switches.
6935 (rs6000_init_hard_regno_mode_ok): Likewise.
6936 (rs6000_option_override_internal): Likewise.
6937 (rs6000_emit_prologue): Likewise.
6938 (rs6000_emit_epilogue): Likewise.
6939 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6940 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6941 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6943 (emit_fusion_p9_store): Likewise.
6944 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6945 resetting these macros if the assembler does not support ISA 3.0
6947 (TARGET_P9_DFORM_VECTOR): Likewise.
6948 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6949 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6951 2017-08-25 Alan Modra <amodra@gmail.com>
6953 PR rtl-optimization/81747
6954 * cse.c (cse_extended_basic_block): Don't attempt to record
6955 equivalences for degenerate conditional branches that branch
6956 to their fall-through.
6958 2017-08-24 Martin Sebor <msebor@redhat.com>
6961 * gimple-fold.c (size_must_be_zero_p): New function.
6962 (gimple_fold_builtin_memory_op): Call it.
6964 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6966 * config/rs6000/mm_malloc.h: New file.
6968 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6970 PR tree-optimization/81913
6971 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6972 analysis when either IVs in condition can wrap.
6974 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6976 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6977 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6979 2017-08-24 Richard Biener <rguenther@suse.de>
6982 * targhooks.c (default_target_can_inline_p): Properly
6983 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6984 is present and always compare.
6985 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6986 infer -mfpmath=sse from TARGET_SSE_P.
6987 (ix86_can_inline_p): Properly use target_option_default_node when
6988 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6990 2017-08-24 Richard Biener <rguenther@suse.de>
6993 * dwarf2out.c (output_die): Handle flag_generate_offload like
6995 (output_comp_unit): Likewise.
6996 (gen_array_type_die): Likewise.
6997 (dwarf2out_early_finish): Likewise.
6998 (note_variable_value_in_expr): Likewise.
6999 (dwarf2out_finish): Likewise. Adjust assert.
7000 * cgraphunit.c (symbol_table::compile): Move setting of
7001 flag_generate_offload earlier ...
7002 (symbol_table::finalize_compilation_unit): ... here, before
7003 early debug finalization.
7005 2017-08-24 Richard Biener <rguenther@suse.de>
7007 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7008 and ipa-fnsummary.h.
7009 (ix86_can_inline_p): When ix86_fpmath flags do not match
7010 check whether the callee uses FP math at all.
7012 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7015 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7016 instead of TYPE_PRECISION.
7018 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7019 Alan Hayward <alan.hayward@arm.com>
7020 David Sherwood <david.sherwood@arm.com>
7022 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7024 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7026 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7028 (ix86_frame::stack_realign_allocate): New field.
7029 (struct machine_frame_state): Modify comments.
7030 (machine_frame_state::sp_realigned_fp_end): New field.
7031 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7033 (sp_valid_at): Add assertion to assure no attempt to access invalid
7034 offset of a realigned stack.
7035 (fp_valid_at): Likewise.
7036 (choose_baseaddr): Modify comments.
7037 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7038 ix86_expand_prologue.
7039 (ix86_expand_prologue): Modify stack realignment and allocation.
7040 (ix86_expand_epilogue): Modify comments.
7041 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7042 avx2_runtime, avx512f, and avx512f_runtime.
7044 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7046 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7047 (mstackrealign): Do not init to -1.
7048 * config/i386/i386.c (ix86_option_override_internal):
7049 Check opts_set, not opts when setting default value of
7050 opts->x_ix86_force_align_arg_pointer.
7052 2017-08-23 Richard Biener <rguenther@suse.de>
7054 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7055 lang_hooks.decl_printable_name.
7056 * print-rtl-function.c (print_rtx_function): Likewise.
7057 * tree-pretty-print.c (dump_function_header): Likewise.
7059 2017-08-23 Richard Biener <rguenther@suse.de>
7062 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7063 -g0 at compile-time.
7065 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7068 * doc/sourcebuild.texi (Other hardware attributes):
7071 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7074 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7075 Fix single-use check.
7077 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7079 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7081 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7083 * doc/install.texi: Modify to add more details on running selected
7086 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7088 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7089 is combined with -mabi=ms.
7090 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7093 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7095 PR tree-optimization/81488
7096 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7097 and cached_basis fields.
7098 (MAX_SPREAD): New constant.
7099 (alloc_cand_and_find_basis): Initialize new fields.
7100 (clear_visited): New function.
7101 (create_phi_basis_1): Rename from create_phi_basis, set visited
7102 and cached_basis fields.
7103 (create_phi_basis): New wrapper function.
7104 (phi_add_costs_1): Rename from phi_add_costs, add spread
7105 parameter, set visited field, short-circuit when limits reached.
7106 (phi_add_costs): New wrapper function.
7107 (record_phi_increments_1): Rename from record_phi_increments, set
7109 (record_phi_increments): New wrapper function.
7110 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7111 (phi_incr_cost): New wrapper function.
7112 (all_phi_incrs_profitable_1): Rename from
7113 all_phi_incrs_profitable, set visited field.
7114 (all_phi_incrs_profitable): New wrapper function.
7116 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7117 Alan Hayward <alan.hayward@arm.com>
7118 David Sherwood <david.sherwood@arm.com>
7120 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7121 that takes the outer and inner modes.
7122 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7123 comparison as the canonical test for a paradoxical subreg.
7124 * combine.c (simplify_set): Use paradoxical_subreg_p.
7125 (make_extraction): Likewise.
7126 (force_to_mode): Likewise.
7127 (rtx_equal_for_field_assignment_p): Likewise.
7128 (gen_lowpart_for_combine): Likewise.
7129 (simplify_comparison): Likewise.
7130 * cse.c (equiv_constant): Likewise.
7131 * expmed.c (store_bit_field_1): Likewise.
7132 * final.c (alter_subreg): Likewise.
7133 * fwprop.c (propagate_rtx): Likewise.
7134 (forward_propagate_subreg): Likewise.
7135 * ira-conflicts.c (ira_build_conflicts): Likewise.
7136 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7137 * lra-constraints.c (curr_insn_transform): Likewise.
7138 (split_reg): Likewise.
7139 * lra-eliminations.c (move_plus_up): Likewise.
7140 (lra_eliminate_regs_1): Likewise.
7141 * recog.c (general_operand): Likewise.
7142 * ree.c (combine_reaching_defs): Likewise.
7143 * reload.c (push_reload): Likewise.
7144 (find_reloads): Likewise.
7145 * reload1.c (elimination_effects): Likewise.
7146 (compute_reload_subreg_offset): Likewise.
7147 (choose_reload_regs): Likewise.
7148 * rtlanal.c (subreg_lsb_1): Likewise.
7149 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7150 (simplify_subreg): Likewise.
7151 * var-tracking.c (track_loc_p): Likewise.
7152 * emit-rtl.c (byte_lowpart_offset): Likewise.
7153 (paradoxical_subreg_p): Delete out-of-line definition.
7155 2017-08-22 Jeff Law <law@redhat.com>
7157 PR tree-optimization/81741
7158 PR tree-optimization/71947
7159 * tree-ssa-dom.c: Include tree-inline.h.
7160 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7161 equivalences if one is more expensive to compute than the other.
7162 * tree-ssa-scopedtables.h (class const_or_copies): Make
7163 record_const_or_copy_raw method private.
7164 (class avail_exprs_stack): New method simplify_binary_operation.
7165 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7166 avail_exprs_stack::simplify_binary_operation as needed.
7167 (avail_exprs_stack::simplify_binary_operation): New function.
7169 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7171 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7172 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7173 (DOT_SYMBOLS): Likewise.
7174 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7175 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7176 (RS6000_ABI_NAME): Likewise.
7177 (TARGET_CMODEL): Likewise.
7178 (TOC_SECTION_ASM_OP): Likewise.
7179 (SET_CMODEL): New macro.
7180 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7182 2017-08-22 Richard Biener <rguenther@suse.de>
7184 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7185 to allow for free-lang-data replacements similar to verify_type_variant.
7187 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7188 Alan Hayward <alan.hayward@arm.com>
7189 David Sherwood <david.sherwood@arm.com>
7191 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7192 VOIDmode for the LABEL_REF.
7194 2017-08-22 Richard Biener <rguenther@suse.de>
7196 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7198 2017-08-22 Richard Biener <rguenther@suse.de>
7200 * common.opt (feliminate-dwarf2-dups): Ignore.
7201 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7202 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7203 same_die_p_wrap, compute_section_prefix,
7204 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7205 (comdat_symbol_id, comdat_symbol_number): Likewise.
7206 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7208 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7209 (output_die): Mark unreachable path unreachable.
7210 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7211 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7212 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7213 (dwarf2out_early_finish): Likewise.
7215 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7217 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7219 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7222 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7223 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7224 (avr_attribute_table) <io, io_low, address>: Initialize
7225 .decl_required with true.
7227 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7229 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7230 undocumented debugging options.
7231 (-mvsx-scalar-double): Likewise.
7232 (-mallow-df-permute): Likewise.
7233 (-mvectorize-builtins): Likewise.
7234 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7235 (rs6000_builtin_vectorized_function): Likewise.
7236 (rs6000_builtin_md_vectorized_function): Likewise.
7237 (rs6000_opt_vars): Likewise.
7239 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7242 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7243 (*btrq_imm): Rename from *btrq.
7244 (*btcq_imm): Rename from *btcq.
7245 (btsc): New code attribute.
7246 (*<btsc><mode>): New insn pattern.
7247 (*btr<mode>): Ditto.
7248 (*<btsc><mode>_mask): New insn_and_split pattern.
7249 (*btr<mode>_mask): Ditto.
7251 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7252 Alan Hayward <alan.hayward@arm.com>
7253 David Sherwood <david.sherwood@arm.com>
7255 * function.c (pad_below): Simplify padding calculation.
7257 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7258 Alan Hayward <alan.hayward@arm.com>
7259 David Sherwood <david.sherwood@arm.com>
7261 * target.def (function_prologue): Remove frame size argument.
7262 (function_epilogue): Likewise.
7263 * doc/tm.texi: Regenerate.
7264 * final.c (final_start_function): Update call to function_prologue.
7265 (final_end_function): Update call to function_epilogue.
7266 (default_function_pro_epilogue): Remove frame size argument.
7267 * output.h (default_function_pro_epilogue): Likewise.
7268 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7269 (arm_output_function_prologue): Likewise.
7270 * config/frv/frv.c (frv_function_prologue): Likewise.
7271 (frv_function_epilogue): Likewise.
7272 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7273 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7274 (ia64_output_function_epilogue): Likewise.
7275 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7276 (m32r_output_function_epilogue): Likewise.
7277 * config/microblaze/microblaze.c (microblaze_function_prologue)
7278 (microblaze_function_epilogue): Likewise.
7279 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7280 (mips_output_function_epilogue): Likewise.
7281 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7282 (mmix_target_asm_function_epilogue): Likewise.
7283 * config/msp430/msp430.c (msp430_start_function): Likewise.
7284 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7285 (nds32_asm_function_epilogue): Likewise.
7286 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7287 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7288 (pa_output_function_epilogue): Likewise.
7289 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7290 (rs6000_output_function_epilogue): Likewise.
7291 * config/rl78/rl78.c (rl78_start_function): Likewise.
7292 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7293 (rs6000_output_function_epilogue): Likewise.
7294 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7295 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7296 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7297 (sparc_asm_function_epilogue): Likewise.
7299 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7301 * tree.h (type_has_mode_precision_p): New function.
7302 * convert.c (convert_to_integer_1): Use it.
7303 * expr.c (expand_expr_real_2): Likewise.
7304 (expand_expr_real_1): Likewise.
7305 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7306 * match.pd: Likewise.
7307 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7308 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7309 * tree-tailcall.c (process_assignment): Likewise.
7310 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7311 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7312 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7313 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7314 (vectorizable_assignment): Likewise.
7315 (vectorizable_shift): Likewise.
7316 (vectorizable_operation): Likewise.
7317 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7319 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7321 * match.pd: Add pow (C, x) simplification.
7323 2017-08-21 Richard Biener <rguenther@suse.de>
7325 PR tree-optimization/81900
7326 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7327 for blocks with abnormal predecessors.
7328 (compute_antic): Do not set visited flag prematurely.
7330 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7333 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7335 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7337 * stor-layout.h (vector_type_mode): Move to...
7338 * tree.h (vector_type_mode): ...here.
7339 * stor-layout.c (vector_type_mode): Move to...
7340 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7342 2017-08-21 Richard Biener <rguenther@suse.de>
7344 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7345 register_external_die hooks.
7346 (debug_false_tree_charstarstar_uhwistar): Declare.
7347 (debug_nothing_tree_charstar_uhwi): Likewise.
7348 * debug.c (do_nothing_debug_hooks): Adjust.
7349 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7350 (debug_nothing_tree_charstar_uhwi): Likewise.
7351 * dbxout.c (dbx_debug_hooks): Adjust.
7352 (xcoff_debug_hooks): Likewise.
7353 * sdbout.c (sdb_debug_hooks): Likewise.
7354 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7355 * dwarf2out.c (macinfo_label_base): New global.
7356 (dwarf2out_register_external_die): New function for the
7357 register_external_die hook.
7358 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7359 (dwarf2_debug_hooks): Use them.
7360 (dwarf2_lineno_debug_hooks): Adjust.
7361 (struct die_struct): Add with_offset flag.
7362 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7363 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7364 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7365 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7366 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7367 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7368 defining section names for the early LTO debug variants.
7369 (reset_indirect_string): New helper.
7370 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7371 (print_dw_val): Add support for offsetted symbol references.
7372 (get_ultimate_context): Split out from is_cxx.
7373 (is_cxx): Use get_ultimate_context.
7374 (is_fortran): Add decl overload.
7375 (compute_comp_unit_symbol): Split out worker from
7376 compute_section_prefix.
7377 (compute_section_prefix): Call compute_comp_unit_symbol and
7378 set comdat_type_p here.
7379 (output_die): Skip DIE symbol output for the LTO added one.
7380 Handle DIE symbol references with offset.
7381 (output_comp_unit): Guard section name mangling properly.
7382 For LTO debug sections emit a symbol at the section beginning
7383 which we use to refer to its DIEs.
7384 (add_abstract_origin_attribute): For DIEs registered via
7385 dwarf2out_register_external_die directly refer to the early
7386 DIE rather than indirectly through the shadow one we created.
7387 Remove obsolete call to dwarf2out_abstract_function for
7388 non-function/block origins.
7389 (gen_array_type_die): When generating early LTO debug do
7390 not emit DW_AT_string_length.
7391 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7392 late when in LTO. As suggested place a gcc_unreachable for
7393 the DECL_ABSTRACT_P case.
7394 (gen_subprogram_die): Avoid another specification DIE
7395 for early built declarations/definitions for the late LTO case.
7396 (gen_variable_die): Add type references for late duplicated VLA dies
7398 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7399 we have the abstract instance already.
7400 (process_scope_var): Adjust decl DIE contexts in LTO which
7401 first puts them in limbo.
7402 (gen_decl_die): Do not generate type DIEs late apart from
7403 types for VLAs or for decls we do not yet have a DIE. Do not
7404 call dwarf2out_abstract_function late.
7405 (dwarf2out_early_global_decl): Make sure to create DIEs
7406 for abstract instances of a decl first.
7407 (dwarf2out_late_global_decl): Adjust comment.
7408 (output_macinfo_op): With multiple macro sections use
7409 macinfo_label_base to distinguish labels.
7410 (output_macinfo): Likewise. Update macinfo_label_base.
7411 Pass in the line info label.
7412 (note_variable_value_in_expr): When generating LTO resolve
7413 all variable values here by generating DIEs as needed.
7414 (init_sections_and_labels): Add early LTO debug flag parameter
7415 and generate different sections and names if set. Add generation
7416 counter for the labels so we can have multiple of them.
7417 (reset_dies): Helper to allow DIEs to be output multiple times.
7418 (dwarf2out_finish): When outputting DIEs to the fat part of an
7419 LTO object first reset DIEs.
7420 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7421 (modified_type_die): Check for decl_ultimate_origin being self
7423 (gen_type_die_with_usage): Likewise.
7424 (gen_typedef_die): Allow decl_ultimate_origin being self.
7425 (set_decl_abstract_flags): Remove.
7426 (set_block_abstract_flags): Likewise.
7427 (dwarf2out_abstract_function): Treat the early generated DIEs
7428 as the abstract copy and only add DW_AT_inline and
7429 DW_AT_artificial here and call set_decl_origin_self.
7430 If the DIE has an abstract origin don't do anything.
7431 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7432 if we have none yet (Go fails to build one, PR78628).
7433 (variably_modified_type_p): Prevent endless recursion for Ada
7434 cyclic pointer types.
7435 * lto-streamer-in.c: Include debug.h.
7436 (dref_queue): New global.
7437 (lto_read_tree_1): Stream in DIE references.
7438 (lto_input_tree): Register DIE references.
7439 (input_function): Stream DECL_DEBUG_ARGS.
7440 * lto-streamer-out.c: Include debug.h.
7441 (lto_write_tree_1): Output DIE references.
7442 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7443 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7444 (output_function): Stream DECL_DEBUG_ARGS.
7445 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7446 Stream DECL_ABSTRACT_ORIGIN.
7447 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7448 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7449 DECL_CONTEXT for file-scope decls.
7450 * lto-streamer.h (struct dref_entry): Declare.
7451 (dref_queue): Likewise.
7452 * cfgexpand.c (pass_expand::execute): Do not call the
7453 outlining_inline_function hook here.
7454 * lto-wrapper.c (debug_obj): New global.
7455 (tool_cleanup): Unlink it if required.
7456 (debug_objcopy): New function.
7457 (run_gcc): Handle early debug sections in the IL files by
7458 extracting them to separate files, partially linkin them and
7459 feeding the result back as result to the linker.
7460 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7461 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7462 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7463 sections into a separate segment.
7464 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7466 (darwin_asm_dwarf_section): Likewise.
7467 (darwin_asm_output_dwarf_offset): Likewise.
7468 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7470 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7471 Alan Hayward <alan.hayward@arm.com>
7472 David Sherwood <david.sherwood@arm.com>
7474 * read-md.h (md_reader::record_potential_iterator_use): Replace
7475 pointer argument with an rtx and an index.
7476 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7477 (apply_mode_iterator): Likewise.
7478 (apply_code_iterator): Likewise.
7479 (apply_int_iterator): Likewise.
7480 (apply_subst_iterator): Likewise.
7481 (record_iterator_use): Likewise.
7482 (record_attribute_use): Likewise.
7483 (md_reader::record_potential_iterator_use): Likewise. Update calls
7484 to record_iterator_use and apply_iterator.
7485 (iterator_use): Replace ptr with x and index.
7486 (attribute_use): Likewise.
7487 (apply_attribute_uses): Update calls to apply_iterator.
7488 (apply_iterators): Likewise. Update initialization of iterator_use.
7489 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7490 and record_potential_iterator_use.
7491 (rtx_reader::read_rtx_operand): Likewise.
7493 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7494 Alan Hayward <alan.hayward@arm.com>
7495 David Sherwood <david.sherwood@arm.com>
7497 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7500 2017-08-21 Richard Biener <rguenther@suse.de>
7503 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7504 at struct end conservatively when comparing common bases.
7506 2017-08-21 Richard Biener <rguenther@suse.de>
7508 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7509 (mem_ref_in_stmt): Remove.
7510 (determine_max_movement): Use ref index to get at the reference.
7511 (invariantness_dom_walker::before_dom_children): Deal with
7512 lim data already initialized.
7513 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7515 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7517 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7518 (TARGET_ISA_ROUND): Ditto.
7519 (TARGET_ROUND): Ditto.
7520 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7521 * config/i386/i386.md: Ditto.
7522 * config/i386/sse.md: Ditto.
7523 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7524 with OPTION_MASK_ISA_SSE4_1.
7526 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7529 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7530 __builtin_ia32_lzcnt_u16.
7532 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7535 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7536 (rs6000_set_current_function): Rewrite function to use it.
7538 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7541 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7542 and TYPE_WARN_IF_NOT_ALIGN.
7543 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7544 (handle_warn_if_not_align): New.
7545 (place_union_field): Call handle_warn_if_not_align.
7546 (place_field): Call handle_warn_if_not_align.
7547 Copy TYPE_WARN_IF_NOT_ALIGN.
7548 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7549 (layout_type): Likewise.
7550 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7552 (tree_decl_common): Add warn_if_not_align.
7553 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7554 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7555 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7556 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7557 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7558 * doc/extend.texi: Document warn_if_not_aligned attribute.
7559 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7561 2017-08-17 Martin Liska <mliska@suse.cz>
7564 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7565 (get_data_dependence): Use it as pointer type.
7566 (distribute_loop): Likewise.
7568 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7570 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7571 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7572 (p8_vmrgow_<mode>_direct): New define_insn.
7573 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7574 handle endianness for vmrgew and vmrgow permute patterns.
7576 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7578 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7579 * config/rs6000/rs6000-cpus.def: Remove comment.
7580 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7581 (POWERPC_MASKS): Likewise.
7582 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7583 use of TARGET_VSX_TIMODE.
7584 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7585 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7586 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7587 (rs6000_option_override_internal): Remove dead code.
7588 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7589 (rs6000_legitimize_reload_address): Likewise.
7590 (rs6000_legitimate_address_p): Likewise.
7591 (rs6000_opt_masks): Delete "vsx-timode".
7592 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7593 from function comment.
7594 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7595 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7596 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7598 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7599 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7600 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7603 (define_peephole2): Likewise.
7605 2017-08-17 Martin Sebor <msebor@redhat.com>
7608 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7609 past the end of an array.
7610 (test_pp_format): Add test cases.
7612 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7614 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7615 missing ECF_NOTHROW flags.
7617 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7620 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7621 operands residing in integer registers.
7622 (*vsx_le_perm_load_<mode>): Likewise.
7623 (*vsx_le_perm_store_<mode>): Likewise.
7624 (define_peephole2): Add peepholes to optimize the above.
7626 2017-08-17 Marek Polacek <polacek@redhat.com>
7629 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7630 to mimic what shorten_compare did. Change the return type to bool.
7631 (fold_cond_expr_with_comparison): Update call to
7632 operand_equal_for_comparison_p.
7633 (fold_ternary_loc): Likewise.
7635 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7637 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7639 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7641 2017-08-17 Richard Biener <rguenther@suse.de>
7643 * tree-ssa-structalias.c (solve_graph): When propagating
7644 to successors update the graphs succ edges and avoid duplicate work.
7646 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7649 * config/i386/i386.c (ix86_option_override_internal): Save target
7650 specific options after ix86_stack_protector_guard_reg was changed.
7652 2017-08-17 Richard Biener <rguenther@suse.de>
7654 PR tree-optimization/81827
7655 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7656 (new_var_info): Initialize it conservatively.
7657 (get_call_vi): Mark register vars.
7658 (new_scalar_tmp_constraint_exp): Likewise.
7659 (handle_rhs_call): Likewise.
7660 (handle_const_call): Likewise.
7661 (create_function_info_for): Likewise.
7662 (solve_constraints): Sort varinfos to separate register from
7663 non-register vars to pack points-to solution bitmaps during
7666 2017-08-17 Marek Polacek <polacek@redhat.com>
7668 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7670 2017-08-17 Richard Biener <rguenther@suse.de>
7672 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7673 to true when overflow is undefined and we saturated the result.
7675 2017-08-17 Alan Modra <amodra@gmail.com>
7678 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7679 Don't use store multiple if only one reg needs saving.
7680 (interesting_frame_related_regno): New function.
7681 (rs6000_frame_related): Don't emit frame info for regs that
7683 (rs6000_emit_epilogue): Likewise.
7685 2017-08-16 Nathan Sidwell <nathan@acm.org>
7687 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7688 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7689 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7690 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7691 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7692 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7693 (verify_type): Adjust for TYPE_BINFO move.
7694 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7695 process TYPE_BINFO directly.
7696 (hash_tree): Likewise.
7697 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7699 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7702 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7704 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7706 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7709 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7710 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7711 Add ix86_binary_operator_ok to insn constraint.
7712 (*iordi_1_bts): Ditto.
7713 (*xordi_1_btc): Ditto.
7714 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7715 Update corresponding peephole2 pattern.
7719 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7721 PR tree-optimization/81832
7722 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7723 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7725 2017-08-16 Marek Polacek <polacek@redhat.com>
7728 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7729 regarding size_zero_node.
7731 2017-08-16 Martin Liska <mliska@suse.cz>
7734 * config.gcc: Respect previously set extra_objs in case
7737 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7739 PR tree-optimization/81835
7740 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7741 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7742 not depend on the phi.
7744 2017-08-16 Alan Modra <amodra@gmail.com>
7746 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7749 2017-08-16 Alan Modra <amodra@gmail.com>
7751 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7752 (save_reg_p): ..into this. Update all callers.
7753 (first_reg_to_save): Simplify.
7755 2017-08-16 Alan Modra <amodra@gmail.com>
7757 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7760 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7764 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7765 constant count if that count is less than 32.
7767 2017-08-15 Nathan Sidwell <nathan@acm.org>
7769 * gcc.c (execute): Emit friendlier message if inferior is killed
7770 by an external cause.
7772 2017-08-15 Richard Biener <rguenther@suse.de>
7774 PR tree-optimization/81790
7775 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7776 CONSTRUCTORs from simplifying and VN.
7778 2017-08-14 Martin Sebor <msebor@redhat.com>
7780 * builtin-attrs.def: Add comments.
7782 2017-08-14 Martin Sebor <msebor@redhat.com>
7785 * doc/extend.texi (attribute nonstring): Document new attribute.
7787 2017-08-14 Martin Sebor <msebor@redhat.com>
7790 * tree-diagnostic.c (default_tree_printer): Handle %G.
7791 * gimple-pretty-print.h (percent_G_format): Declare new function.
7792 * gimple-pretty-print.c (percent_G_format): Define.
7793 * tree-pretty-print.c (percent_K_format): Add argument.
7795 2017-08-14 Martin Sebor <msebor@redhat.com>
7797 PR translation/79998
7798 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7799 Remove a stray space.
7801 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7804 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7805 (*iordi_1_bts): Ditto.
7806 (*xordi_1_btc): Ditto.
7808 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7811 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7812 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7814 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7815 quoted strings, and make more translator-friendly.
7816 (darwin_rs6000_override_options): Likewise.
7817 (rs6000_option_override_internal): Likewise.
7818 (rs6000_return_in_memory): Fix overlong line.
7819 (init_cmulative_args): Use quoted strings, and make more
7820 translator-friendly.
7821 (rs6000_pass_by_reference): Fix overlong line.
7822 (def_builtin): Use quoted strings.
7823 (altivec_expand_predicate_builtin): Use quoted strings, and make
7824 more translator-friendly.
7825 (htm_expand_builtin): Use quoted strings.
7826 (cpu_expand_builtin): Use quoted strings, and make more
7827 translator-friendly.
7828 (altivec_expand_builtin): Likewise.
7829 (paired_expand_predicate_builtin): Likewise.
7830 (rs6000_invalid_builtin): Likewise.
7831 (builtin_function_type): Use quoted strings.
7832 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7833 more translator-friendly.
7834 (rs6000_trampoline_init): Likewise.
7835 (rs6000_handle_altivec_attribute): Likewise.
7836 (rs6000_inner_target_options): Use quoted strings.
7837 (rs6000_disable_incompatible_switches): Likewise.
7838 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7839 strings, and make more translator-friendly.
7840 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7842 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7844 PR tree-optimization/81799
7845 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7846 cond_expr to simple gimple operand.
7848 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7851 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7853 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7857 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7858 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7859 TARGET_GASISR_PROLOGUES.
7860 * config/avr/avr.c (avr_option_override): Same.
7861 (avr_pass_pre_proep::execute): Same.
7863 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7866 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7867 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7869 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7871 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7872 define_insn_and_split. Split to a memory load from 0 in
7873 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7874 using PTR mode iterator.
7875 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7876 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7877 (*add_tp_<mode>): Redefine as define_insn_and_split.
7878 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7879 address space. Merge with *add_tp_x32 using PTR mode iterator.
7880 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7881 Split to an add with a memory load from 0 in
7882 DEFAULT_TLS_SEG_REG address space.
7884 2017-08-12 Andrew Pinski <apinski@cavium.com>
7886 * config/aarch64/aarch64-option-extensions.def (rdma):
7887 Fix feature string to what Linux prints out in /proc/cpuinfo.
7889 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7892 * dwarf2out.c (modified_type_die): For C typedef types that have
7893 an ultimate origin, process the ultimate origin instead of the
7895 (gen_typedef_die): Assert that input DECLs have no ultimate
7897 (gen_type_die_with_usage): For typedef variants that have an
7898 ultimate origin, just call gen_decl_die on the original DECL.
7899 (process_scope_var): Avoid creating DIEs for local typedefs and
7900 concrete static variables.
7902 2017-08-12 Alan Modra <amodra@gmail.com>
7906 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7907 match gnu-user.h startfile.
7908 (ENDFILE_LINUX_SPEC): Similarly.
7910 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7913 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7915 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7917 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7918 * config/aarch64/aarch64.md (mov<mode>): Change.
7919 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7920 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7921 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7923 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7925 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7926 for storage order barriers.
7928 2017-08-11 Martin Liska <mliska@suse.cz>
7931 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7932 variables of void type.
7934 2017-08-11 Martin Liska <mliska@suse.cz>
7936 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7937 TARGET_SUPPORTS_ALIASES.
7938 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7939 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7940 (optimize_weakref): Likewise.
7941 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7942 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7943 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7945 2017-08-11 Martin Liska <mliska@suse.cz>
7948 * config/i386/i386.c (make_resolver_func): Do complete
7949 refactoring of the function.
7951 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7954 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7955 * config/i386/i386.c (ix86_stack_protect_guard): Use
7956 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7957 * doc/invoke.texi (x86 Options): Document
7958 -mstack-protector-guard-symbol= option.
7960 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7962 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7963 * config/i386/i386.c (ix86_split_stack_guard): New function.
7964 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7965 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7966 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7967 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7968 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7969 (split_stack_space_check): Call ix86_split_stack_guard.
7971 2017-08-10 Martin Sebor <msebor@redhat.com>
7973 * print-tree.c (print_node): Print location using the established
7975 Replace spaces with colons.
7976 (debug_raw, debug): Ditto.
7978 2017-08-10 Martin Sebor <msebor@redhat.com>
7981 * pretty-print.c (pp_format): Correct the handling of %s precision.
7983 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7986 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7988 (ix86_finalize_stack_frame_flags): This. Also clear
7989 frame_pointer_needed if -fno-omit-frame-pointer is used without
7991 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7992 with ix86_finalize_stack_frame_flags.
7993 (ix86_expand_epilogue): Likewise.
7994 (ix86_expand_split_stack_prologue): Likewise.
7995 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7997 2017-08-10 Martin Liska <mliska@suse.cz>
8000 * c-attribs.c (handle_target_attribute):
8001 Report warning for an empty string argument of target attribute.
8003 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8006 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8008 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8009 or DECL_NONLOCAL labels.
8010 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8011 or DECL_NONLOCAL labels here.
8013 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8015 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8016 to indicate when early gimple folding has been disabled.
8017 (rs6000_gimple_fold_builtin): Add debug content.
8018 (rs6000_invalid_builtin): Fix whitespace.
8019 (rs6000_expand_builtin): Fix whitespace.
8020 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8022 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8025 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8026 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8028 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8030 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8031 (qdf24xx): Likewise.
8032 * config/aarch64/aarch64.md: Include falkor.md.
8033 * config/aarch64/falkor.md: New.
8035 2017-08-09 Marek Polacek <polacek@redhat.com>
8038 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8039 * diagnostic.c (emit_diagnostic): Add a comment.
8040 (emit_diagnostic_valist): New function.
8042 2017-08-09 Marek Polacek <polacek@redhat.com>
8045 * input.c (make_location): New overload.
8046 * input.h (make_location): Declare.
8048 2017-08-08 Alan Modra <amodra@gmail.com>
8049 H.J. Lu <hongjiu.lu@intel.com>
8052 * gcc.c (NO_PIE_SPEC): Delete.
8053 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8055 (LINK_PIE_SPEC): ..to here.
8056 (LINK_COMMAND_SPEC): Support -no-pie.
8057 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8058 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8060 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8061 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8062 (ENDFILE_CRTEND_SPEC): Similarly.
8064 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8067 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8068 (mstack-protector-guard-offset=): Ditto.
8069 * config/i386/i386.c (ix86_option_override): Handle
8070 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8072 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8073 ix86_stack_protect_guard_offset variables.
8074 (TARGET_STACK_PROTECT_GUARD): Always define.
8075 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8076 and -mstack-protector-guard-offset= options.
8078 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8080 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8081 boundary case for the last candidate.
8083 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8085 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8086 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8088 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8091 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8092 * config/aarch64/aarch64-builtins.c
8093 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8094 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8095 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8097 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8098 Andrew Pinski <pinskia@gmail.com>
8101 * internal-fn.def (XORSIGN): New.
8102 * optabs.def (xorsign_optab): New.
8103 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8104 (convert_expand_mult_copysign): New.
8105 (pass_optimize_widening_mul::execute): Call
8106 convert_expand_mult_copysign.
8108 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8110 PR tree-optimization/81354
8111 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8112 Insert on edges rather than explicitly creating landing pads.
8113 (analyze_candidates_and_replace): Commit edge inserts.
8115 2017-08-08 Richard Biener <rguenther@suse.de>
8118 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8119 (expand_simple_operations): Also look through ADDR_EXPRs with
8120 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8122 2017-08-08 Richard Biener <rguenther@suse.de>
8124 PR tree-optimization/81723
8125 * tree-vect-slp.c (struct bst_traits): New hash traits.
8126 (bst_fail): New global.
8127 (vect_build_slp_tree_2): New worker, split out from ...
8128 (vect_build_slp_tree): ... this now wrapping it with using
8129 bst_fail set to cache SLP tree build fails. Properly handle
8131 (vect_analyze_slp_instance): Allocate and free bst_fail.
8133 2017-08-08 Martin Liska <mliska@suse.cz>
8136 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8137 LABEL_DECLs that can be from a different function.
8139 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8141 PR tree-optimization/81744
8142 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8143 loop's number of iterations.
8145 2017-08-08 Martin Liska <mliska@suse.cz>
8147 * asan.c: Include header files.
8148 * attribs.c (build_decl_attribute_variant): New function moved
8150 (build_type_attribute_qual_variant): Likewise.
8151 (cmp_attrib_identifiers): Likewise.
8152 (simple_cst_list_equal): Likewise.
8153 (omp_declare_simd_clauses_equal): Likewise.
8154 (attribute_value_equal): Likewise.
8155 (comp_type_attributes): Likewise.
8156 (build_type_attribute_variant): Likewise.
8157 (lookup_ident_attribute): Likewise.
8158 (remove_attribute): Likewise.
8159 (merge_attributes): Likewise.
8160 (merge_type_attributes): Likewise.
8161 (merge_decl_attributes): Likewise.
8162 (merge_dllimport_decl_attributes): Likewise.
8163 (handle_dll_attribute): Likewise.
8164 (attribute_list_equal): Likewise.
8165 (attribute_list_contained): Likewise.
8166 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8167 (lookup_attribute_by_prefix): Likewise.
8168 * bb-reorder.c: Include header files.
8169 * builtins.c: Likewise.
8170 * calls.c: Likewise.
8171 * cfgexpand.c: Likewise.
8172 * cgraph.c: Likewise.
8173 * cgraphunit.c: Likewise.
8174 * convert.c: Likewise.
8175 * dwarf2out.c: Likewise.
8176 * final.c: Likewise.
8177 * fold-const.c: Likewise.
8178 * function.c: Likewise.
8179 * gimple-expr.c: Likewise.
8180 * gimple-fold.c: Likewise.
8181 * gimple-pretty-print.c: Likewise.
8182 * gimple.c: Likewise.
8183 * gimplify.c: Likewise.
8184 * hsa-common.c: Likewise.
8185 * hsa-gen.c: Likewise.
8186 * internal-fn.c: Likewise.
8187 * ipa-chkp.c: Likewise.
8188 * ipa-cp.c: Likewise.
8189 * ipa-devirt.c: Likewise.
8190 * ipa-fnsummary.c: Likewise.
8191 * ipa-inline.c: Likewise.
8192 * ipa-visibility.c: Likewise.
8194 * lto-cgraph.c: Likewise.
8195 * omp-expand.c: Likewise.
8196 * omp-general.c: Likewise.
8197 * omp-low.c: Likewise.
8198 * omp-offload.c: Likewise.
8199 * omp-simd-clone.c: Likewise.
8200 * opts-global.c: Likewise.
8201 * passes.c: Likewise.
8202 * predict.c: Likewise.
8203 * sancov.c: Likewise.
8204 * sanopt.c: Likewise.
8205 * symtab.c: Likewise.
8206 * toplev.c: Likewise.
8207 * trans-mem.c: Likewise.
8208 * tree-chkp.c: Likewise.
8209 * tree-eh.c: Likewise.
8210 * tree-into-ssa.c: Likewise.
8211 * tree-object-size.c: Likewise.
8212 * tree-parloops.c: Likewise.
8213 * tree-profile.c: Likewise.
8214 * tree-ssa-ccp.c: Likewise.
8215 * tree-ssa-live.c: Likewise.
8216 * tree-ssa-loop.c: Likewise.
8217 * tree-ssa-sccvn.c: Likewise.
8218 * tree-ssa-structalias.c: Likewise.
8219 * tree-ssa.c: Likewise.
8220 * tree-streamer-in.c: Likewise.
8221 * tree-vectorizer.c: Likewise.
8222 * tree-vrp.c: Likewise.
8224 * ubsan.c: Likewise.
8225 * varasm.c: Likewise.
8226 * varpool.c: Likewise.
8227 * tree.c: Remove functions moved to attribs.[ch].
8229 * config/aarch64/aarch64.c: Add attrs.h header file.
8230 * config/alpha/alpha.c: Likewise.
8231 * config/arc/arc.c: Likewise.
8232 * config/arm/arm.c: Likewise.
8233 * config/avr/avr.c: Likewise.
8234 * config/bfin/bfin.c: Likewise.
8235 * config/c6x/c6x.c: Likewise.
8236 * config/cr16/cr16.c: Likewise.
8237 * config/cris/cris.c: Likewise.
8238 * config/darwin.c: Likewise.
8239 * config/epiphany/epiphany.c: Likewise.
8240 * config/fr30/fr30.c: Likewise.
8241 * config/frv/frv.c: Likewise.
8242 * config/ft32/ft32.c: Likewise.
8243 * config/h8300/h8300.c: Likewise.
8244 * config/i386/winnt.c: Likewise.
8245 * config/ia64/ia64.c: Likewise.
8246 * config/iq2000/iq2000.c: Likewise.
8247 * config/lm32/lm32.c: Likewise.
8248 * config/m32c/m32c.c: Likewise.
8249 * config/m32r/m32r.c: Likewise.
8250 * config/m68k/m68k.c: Likewise.
8251 * config/mcore/mcore.c: Likewise.
8252 * config/microblaze/microblaze.c: Likewise.
8253 * config/mips/mips.c: Likewise.
8254 * config/mmix/mmix.c: Likewise.
8255 * config/mn10300/mn10300.c: Likewise.
8256 * config/moxie/moxie.c: Likewise.
8257 * config/msp430/msp430.c: Likewise.
8258 * config/nds32/nds32-isr.c: Likewise.
8259 * config/nds32/nds32.c: Likewise.
8260 * config/nios2/nios2.c: Likewise.
8261 * config/nvptx/nvptx.c: Likewise.
8262 * config/pa/pa.c: Likewise.
8263 * config/pdp11/pdp11.c: Likewise.
8264 * config/powerpcspe/powerpcspe.c: Likewise.
8265 * config/riscv/riscv.c: Likewise.
8266 * config/rl78/rl78.c: Likewise.
8267 * config/rx/rx.c: Likewise.
8268 * config/s390/s390.c: Likewise.
8269 * config/sh/sh.c: Likewise.
8270 * config/sol2.c: Likewise.
8271 * config/sparc/sparc.c: Likewise.
8272 * config/spu/spu.c: Likewise.
8273 * config/stormy16/stormy16.c: Likewise.
8274 * config/tilegx/tilegx.c: Likewise.
8275 * config/tilepro/tilepro.c: Likewise.
8276 * config/v850/v850.c: Likewise.
8277 * config/vax/vax.c: Likewise.
8278 * config/visium/visium.c: Likewise.
8279 * config/xtensa/xtensa.c: Likewise.
8281 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8284 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8285 constraints since the -mupper-regs-* switches have been
8287 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8288 into a vector from a double word element that was extracted from
8289 another vector, and eliminate extra XXPERMDI instructions.
8290 (vsx_concat_<mode>_2): Likewise.
8291 (vsx_concat_<mode>_3): Likewise.
8292 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8293 concat to allow optimizing inserts from previous extracts.
8295 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8297 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8298 memory reference to a SSP offset in TLS address space.
8299 (ix86_print_operand) <case '@'>: Remove.
8300 (ix86_print_operand_punct_valid_p): Remove '@' code.
8301 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8303 (stack_tls_protect_set_<mode>): Remove.
8304 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8305 (stack_tls_protect_test_<mode>): Remove.
8306 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8308 2017-08-07 Olivier Hainque <hainque@adacore.com>
8311 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8313 2017-08-07 Douglas Rupp <rupp@adacore.com>
8315 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8316 variable was referenced as multidir in command.
8318 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8321 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8324 2017-08-07 Martin Liska <mliska@suse.cz>
8326 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8327 * config/rl78/rl78.c: Add include of attribs.h.
8328 * config/sh/sh.c: Likewise.
8329 * config/v850/v850.c: Likewise.
8331 2017-08-07 Tom de Vries <tom@codesourcery.com>
8334 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8336 2017-08-07 Martin Liska <mliska@suse.cz>
8338 * config/mips/mips.c: Include attribs.h.
8340 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8343 * doc/invoke.texi: Document change in behvaior for -Ofast for
8346 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8348 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8350 (aarch64_pop_regs): Likewise.
8351 (aarch64_gen_load_pair): Likewise.
8352 (aarch64_save_callee_saves): Likewise.
8353 (aarch64_restore_callee_saves): Likewise.
8355 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8357 * config/i386/i386.c: Revert the last change.
8359 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8362 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8364 (ix86_finalize_stack_frame_flags): This. Also clear
8365 frame_pointer_needed if -fno-omit-frame-pointer is used without
8367 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8368 with ix86_finalize_stack_frame_flags.
8369 (ix86_expand_epilogue): Likewise.
8370 (ix86_expand_split_stack_prologue): Likewise.
8372 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8375 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8376 to P_AES for Westmere.
8378 2017-08-07 Jonathan Yong <10walls@gmail.com>
8380 * config/i386/mingw.opt (fset-stack-executable): Removed.
8381 * config/i386/cygming.opt (fset-stack-executable): Moved
8383 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8385 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8387 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8389 2017-08-07 Marek Polacek <polacek@redhat.com>
8392 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8394 2017-08-07 Martin Liska <mliska@suse.cz>
8396 * attribs.h (canonicalize_attr_name): New function.
8397 (cmp_attribs): Move from c-format.c and adjusted.
8398 (is_attribute_p): Moved from tree.h.
8399 * tree-inline.c: Add new includes.
8400 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8401 (private_is_attribute_p): Remove.
8402 (private_lookup_attribute): Likewise.
8403 (private_lookup_attribute_by_prefix): Simplify.
8404 (remove_attribute): Use is_attribute_p.
8405 * tree.h: Remove removed declarations.
8407 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8410 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8411 instead of computing it in the function. Formatting fix.
8412 (expand_case): Don't rely on default_edge being the first edge,
8413 clear it if removing it, pass default_edge to
8414 emit_case_dispatch_table.
8415 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8418 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8420 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8421 insn in the function, emit NOP after the insn.
8423 2017-08-06 Tom de Vries <tom@codesourcery.com>
8425 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8428 2017-08-06 Tom de Vries <tom@codesourcery.com>
8430 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8433 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8435 PR tree-optimization/57371
8436 * match.pd: New pattern.
8438 2017-08-04 Marek Polacek <polacek@redhat.com>
8441 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8442 perform the computation in offset_int.
8444 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8446 PR tree-optimization/81136
8447 * tree-vectorizer.h: Include tree-hash-traits.h.
8448 (vec_base_alignments): New typedef.
8449 (vec_info): Add a base_alignments field.
8450 (vect_record_base_alignments): Declare.
8451 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8453 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8454 (create_data_ref): Add an is_conditional_in_stmt argument.
8455 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8456 the is_conditional_in_stmt field.
8457 (data_ref_loc): Add an is_conditional_in_stmt field.
8458 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8459 (find_data_references_in_stmt): Update call to create_data_ref.
8460 (graphite_find_data_references_in_stmt): Likewise.
8461 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8462 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8463 (vect_record_base_alignment): New function.
8464 (vect_record_base_alignments): Likewise.
8465 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8466 for nested statements even if we fail to compute a misalignment.
8467 Use pooled base alignments for unconditional references.
8468 (vect_find_same_alignment_drs): Compare base addresses instead
8470 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8471 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8473 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8475 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8476 Add an explicit name for the enum. Use auto_vec for slp_instances
8478 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8480 (_bb_vec_info): Add a constructor and destructor.
8481 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8482 (destroy_loop_vec_info): Delete.
8483 (vect_destroy_datarefs): Likewise.
8484 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8485 (vec_info::vec_info): New function.
8486 (vec_info::~vec_info): Likewise.
8487 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8488 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8489 destroy_loop_vec_info.
8490 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8491 (_loop_vec_info::_loop_vec_info): ...this.
8492 (destroy_loop_vec_info): Replace with...
8493 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8494 the stmt_vec_infos. Leave handling of vec_info information to its
8495 destructor. Remove explicit vector releases.
8496 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8497 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8498 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8499 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8500 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8501 (destroy_bb_vec_info): Replace with...
8502 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8503 information to its destructor.
8504 (vect_slp_analyze_bb_1): Use new and delete instead of
8505 new_bb_vec_info and destroy_bb_vec_info.
8506 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8509 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8511 * tree-data-ref.h (subscript): Add access_fn field.
8512 (data_dependence_relation): Add could_be_independent_p.
8513 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8514 (same_access_functions): Move to tree-data-ref.c.
8515 * tree-data-ref.c (ref_contains_union_access_p): New function.
8516 (access_fn_component_p): Likewise.
8517 (access_fn_components_comparable_p): Likewise.
8518 (dr_analyze_indices): Add a reference to access_fn_component_p.
8519 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8521 (constant_access_functions): Likewise.
8522 (add_other_self_distances): Likewise.
8523 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8524 (initialize_data_dependence_relation): Use XCNEW and remove
8525 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8526 of access functions that have the same type. Allow the
8527 subsequence to end with different bases in some circumstances.
8528 Record the chosen access functions in SUB_ACCESS_FN.
8529 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8530 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8531 (subscript_dependence_tester_1): Likewise dra and drb.
8532 (build_classic_dist_vector): Update calls accordingly.
8533 (subscript_dependence_tester): Likewise.
8534 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8535 DDR_COULD_BE_INDEPENDENT_P.
8536 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8537 comp_alias_ddrs instead of may_alias_ddrs.
8538 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8540 (vect_analyze_data_ref_dependence): Use it if
8541 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8542 distance vectors if that fails.
8543 (dependence_distance_ge_vf): New function.
8544 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8545 LOOP_VINFO_MAY_ALIAS_DDRS.
8547 2017-08-04 Richard Biener <rguenther@suse.de>
8550 * fold-const.c (fold_binary_loc): Properly restrict
8551 minus_var0 && minus_var1 case when associating undefined overflow
8554 2017-08-04 Tom de Vries <tom@codesourcery.com>
8556 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8558 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8560 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8561 Don't start diagnostic messages with a capital letter.
8562 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8564 (rs6000_invalid_builtin): Likewise.
8565 (rs6000_trampoline_init): Likewise.
8567 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8570 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8571 after setting changeable df flags.
8573 2017-08-03 Richard Biener <rguenther@suse.de>
8575 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8576 up if the use is in USE - X.
8578 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8580 * toplev.c (dumpfile.h): New include.
8581 (internal_error_reentered): New static function. Use it...
8582 (internal_error_function): ...here to handle reentered internal_error.
8584 2017-08-03 Richard Biener <rguenther@suse.de>
8587 * fold-const.c (split_tree): Add minus_var and minus_con
8588 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8589 here but always use minus_*.
8590 (associate_trees): Assert we never associate with MINUS_EXPR
8591 and NULL first operand. Do not recurse for PLUS_EXPR operands
8592 when associating as MINUS_EXPR either.
8593 (fold_binary_loc): Track minus_var and minus_con.
8595 2017-08-03 Tom de Vries <tom@codesourcery.com>
8598 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8599 ACCEL_COMPILER, apply finish_options on
8600 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8602 2017-08-03 Tom de Vries <tom@codesourcery.com>
8605 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8606 function_entry_patch_area_size > 0.
8608 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8611 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8612 instead of 10??LU, perform unit multiplication in wide_int,
8613 don't change alloc_object_size_limit if the limit is larger
8616 PR tree-optimization/81655
8617 PR tree-optimization/81588
8618 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8619 the case when ranges[i].low and high are 1 for unsigned type with
8623 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8624 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8626 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8628 * tree-vrp.h: Add include guard.
8630 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8633 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8634 (ud2): New insn pattern.
8635 * config/i386/i386.c (ix86_expand_epilogue):
8636 For naked functions, generate ud2 instead of trap insn.
8638 2017-08-02 Marek Polacek <polacek@redhat.com>
8641 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8643 2017-08-02 Tom de Vries <tom@codesourcery.com>
8644 Cesar Philippidis <cesar@codesourcery.com>
8646 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8647 Add missing edge probabilities.
8649 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8651 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8654 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8657 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8658 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8659 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8660 prologue_seq sequences - if any.
8662 2017-08-02 Richard Biener <rguenther@suse.de>
8664 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8665 via vectors if supported, integer extracts via punning if supported
8666 or otherwise vector extracts.
8668 2017-08-02 Richard Biener <rguenther@suse.de>
8670 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8672 (bitmap_insert_into_set): ... this.
8674 2017-08-02 Richard Biener <rguenther@suse.de>
8676 PR tree-optimization/81633
8678 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8680 PR tree-optimization/71752
8681 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8683 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8685 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8686 (machine_function::call_ms2sysv_pad_out): Remove field.
8687 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8688 (ix86_compute_frame_layout): Likewise.
8690 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8693 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8694 attribute with interrupt attribute.
8696 2017-08-01 Andrew Pinski <apinski@cavium.com>
8698 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8699 BIT_INSERT_EXPR's operand 1
8700 to see if the types precision matches.
8702 2017-08-01 Martin Liska <mliska@suse.cz>
8705 * builtins.c (expand_builtin_memcpy_args): Remove.
8706 (expand_builtin_memcpy): Call newly added function
8707 expand_builtin_memory_copy_args.
8708 (expand_builtin_memcpy_with_bounds): Likewise.
8709 (expand_builtin_mempcpy): Remove last argument.
8710 (expand_builtin_mempcpy_with_bounds): Likewise.
8711 (expand_builtin_memory_copy_args): New function created from
8712 expand_builtin_mempcpy_args with small modifications.
8713 (expand_builtin_mempcpy_args): Remove.
8714 (expand_builtin_stpcpy): Remove unused argument.
8715 (expand_builtin): Likewise.
8716 (expand_builtin_with_bounds): Likewise.
8718 2017-08-01 Martin Liska <mliska@suse.cz>
8721 Make mempcpy more optimal (PR middle-end/70140).
8723 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8726 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8727 __builtin_vec_cmpne verify both arguments are compatible vectors
8728 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8729 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8730 move computation of aligned to after checking the argument types.
8734 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8735 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8738 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8739 Doug Rupp <rupp@adacore.com>
8740 Olivier Hainque <hainque@adacore.com>
8742 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8743 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8745 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8746 for TARGET_OS_CPP_BUILTIN.
8747 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8748 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8750 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8751 passing required abi options to the assembler for EABI configurations.
8752 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8753 of .text.hot and .text.unlikely sections for kernel modules when
8754 using ARM style exceptions.
8755 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8756 options. Add EXTRA_CC1_SPEC.
8757 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8759 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8761 (ARM_TARGET2_DWARF_FORMAT): Define.
8762 * config/arm/t-vxworks: Adjust multilib control to removal of the
8763 Diab command line options.
8765 2017-08-01 Martin Liska <mliska@suse.cz>
8767 PR gcov-profile/81561
8768 * gcov.c (unblock): Make unblocking safe as we need to preserve
8769 index correspondence of blocks and block_lists.
8771 2017-08-01 Richard Biener <rguenther@suse.de>
8773 PR tree-optimization/81181
8774 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8775 (compute_antic): ... end of iteration here.
8777 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8779 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8780 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8781 (ftree-slp-vectorize): Likewise.
8782 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8783 can no longer be set independent of flag_tree_loop_vectorize.
8784 * omp-general.c (emp_max_vf): Likewise.
8785 * opts.c (enable_fdo_optimizations): Remove references to
8786 flag_tree_vectorize, these are now implicit.
8787 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8788 and leave it for the options machinery.
8790 2017-08-01 Martin Liska <mliska@suse.cz>
8793 * builtins.c (expand_builtin_memcpy_args): Remove.
8794 (expand_builtin_memcpy): Call newly added function
8795 expand_builtin_memory_copy_args.
8796 (expand_builtin_memcpy_with_bounds): Likewise.
8797 (expand_builtin_mempcpy): Remove last argument.
8798 (expand_builtin_mempcpy_with_bounds): Likewise.
8799 (expand_builtin_memory_copy_args): New function created from
8800 expand_builtin_mempcpy_args with small modifications.
8801 (expand_builtin_mempcpy_args): Remove.
8802 (expand_builtin_stpcpy): Remove unused argument.
8803 (expand_builtin): Likewise.
8804 (expand_builtin_with_bounds): Likewise.
8806 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8809 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8810 print "ds:" only for immediates in generic address space.
8812 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8815 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8816 (ix86_function_ok_for_sibcall): Return false for naked functions.
8818 2017-08-01 Richard Biener <rguenther@suse.de>
8820 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8821 (compute_antic): Seed worklist with exit block predecessors.
8822 * cfganal.c (dfs_find_deadend): For a cycle return the source
8823 of the edge closing it.
8825 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8827 * config/aarch64/aarch64.c
8828 (aarch64_can_const_movi_rtx_p): Move 0 check.
8830 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8832 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8833 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8835 * match.pd: Ditto in address comparison pattern.
8837 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8839 PR tree-optimization/81627
8840 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8841 closed ssa form for store-store chain.
8843 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8845 PR tree-optimization/81620
8846 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8847 for store-store chain.
8849 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8851 PR tree-optimization/81588
8852 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8853 ranges[i].in_p, invert comparison code ccode. For >/>=,
8854 swap rhs1 and rhs2 and comparison code unconditionally,
8855 for </<= don't do that. Don't swap rhs1/rhs2 again if
8856 ranges[i].in_p, instead invert comparison code ccode if
8857 opcode or oe->rank is BIT_IOR_EXPR.
8860 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8861 a direct optab to conversion optab.
8862 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8863 with GET_MODE_INNER as last argument instead of optab_handler.
8864 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8865 vector extraction if possible and optab is available.
8866 * expr.c (store_constructor): Use convert_optab_handler instead
8867 of optab_handler. Use vector initialization from smaller
8868 vectors if possible and optab is available.
8869 * tree-vect-stmts.c (vectorizable_load): Likewise.
8870 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8872 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8873 of vec_init from half-sized vectors with the same element mode.
8874 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8875 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8876 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8877 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8878 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8879 after mode in gen_vec_extract* calls.
8880 (vec_extract<mode>): Renamed to ...
8881 (vec_extract<mode><ssescalarmodelower>): ... this.
8882 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8883 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8884 element mode after mode in gen_vec_init* calls.
8885 (VEC_INIT_HALF_MODE): New mode iterator.
8886 (vec_init<mode>): Renamed to ...
8887 (vec_init<mode><ssescalarmodelower>): ... this.
8888 (vec_init<mode><ssehalfvecmodelower>): New expander.
8889 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8890 (vec_extractv2sfsf): ... this.
8891 (vec_initv2sf): Renamed to ...
8892 (vec_initv2sfsf): ... this.
8893 (vec_extractv2si): Renamed to ...
8894 (vec_extractv2sisi): ... this.
8895 (vec_initv2si): Renamed to ...
8896 (vec_initv2sisi): ... this.
8897 (vec_extractv4hi): Renamed to ...
8898 (vec_extractv4hihi): ... this.
8899 (vec_initv4hi): Renamed to ...
8900 (vec_initv4hihi): ... this.
8901 (vec_extractv8qi): Renamed to ...
8902 (vec_extractv8qiqi): ... this.
8903 (vec_initv8qi): Renamed to ...
8904 (vec_initv8qiqi): ... this.
8905 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8906 (vec_init<mode>): Renamed to ...
8907 (vec_init<mode><VEC_base_l>): ... this.
8908 (vec_extract<mode>): Renamed to ...
8909 (vec_extract<mode><VEC_base_l>): ... this.
8910 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8911 (vec_initv2sfsf): ... this.
8912 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8913 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8914 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8915 element mode after mode in gen_vec_init* calls.
8916 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8917 (vec_init<mode><Vel>): ... this.
8918 (vec_extract<mode>): Renamed to ...
8919 (vec_extract<mode><Vel>): ... this.
8920 * config/aarch64/iterators.md (Vel): New mode attribute.
8921 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8922 Add element mode after mode in gen_vec_extract* calls.
8923 * config/s390/vector.md (non_vec_l): New mode attribute.
8924 (vec_extract<mode>): Renamed to ...
8925 (vec_extract<mode><non_vec_l>): ... this.
8926 (vec_init<mode>): Renamed to ...
8927 (vec_init<mode><non_vec_l>): ... this.
8928 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8929 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8931 * config/arm/iterators.md (V_elem_l): New mode attribute.
8932 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8933 (vec_extract<mode><V_elem_l>): ... this.
8934 (vec_extractv2di): Renamed to ...
8935 (vec_extractv2didi): ... this.
8936 (vec_init<mode>): Renamed to ...
8937 (vec_init<mode><V_elem_l>): ... this.
8938 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8939 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8940 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8941 Add element mode after gen_vec_extract* calls.
8942 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8943 (vec_init<mode><unitmode>): ... this.
8944 (vec_extract<mode>): Renamed to ...
8945 (vec_extract<mode><unitmode>): ... this.
8946 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8947 (vec_init<mode><unitmode>): ... this.
8948 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8949 (vec_initv2sfsf): ... this.
8950 (vec_extractv2sf): Renamed to ...
8951 (vec_extractv2sfsf): ... this.
8952 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8953 Add element mode after gen_vec_extract* calls.
8954 * config/mips/mips.md (unitmode): New mode iterator.
8955 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8956 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8957 * config/spu/spu.md (inner_l): New mode attribute.
8958 (vec_init<mode>): Renamed to ...
8959 (vec_init<mode><inner_l>): ... this.
8960 (vec_extract<mode>): Renamed to ...
8961 (vec_extract<mode><inner_l>): ... this.
8962 * config/sparc/sparc.md (veltmode): New mode iterator.
8963 (vec_init<VMALL:mode>): Renamed to ...
8964 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8965 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8966 (vec_initv2sisi): ... this.
8967 (vec_initv2sf): Renamed to ...
8968 (vec_initv2sfsf): ... this.
8969 (vec_extractv2sf): Renamed to ...
8970 (vec_extractv2sfsf): ... this.
8971 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8972 (vec_init<mode>): Renamed to ...
8973 (vec_init<mode><VEC_base_l>): ... this.
8974 (vec_extract<mode>): Renamed to ...
8975 (vec_extract<mode><VEC_base_l>): ... this.
8976 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8977 (vec_initv2sfsf): ... this.
8978 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8979 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8980 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8981 gen_vec_init* calls.
8983 2017-08-01 Richard Biener <rguenther@suse.de>
8985 PR tree-optimization/81297
8986 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8987 TREE_OVERFLOW from INTEGER_CSTs.
8989 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8991 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8993 2017-07-31 Carl Love <cel@us.ibm.com>
8995 * config/rs6000/rs6000-c: Add support for built-in functions
8996 vector signed char vec_xl_be (signed long long, signed char *);
8997 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8998 vector signed int vec_xl_be (signed long long, signed int *);
8999 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9000 vector signed long long vec_xl_be (signed long long, signed long long *);
9001 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9002 vector signed short vec_xl_be (signed long long, signed short *);
9003 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9004 vector double vec_xl_be (signed long long, double *);
9005 vector float vec_xl_be (signed long long, float *);
9006 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9007 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9008 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9010 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9011 (altivec_expand_builtin): Add switch statement to call
9012 altivec_expand_xl_be for each builtin.
9013 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9014 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9015 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9016 __builtin_vsx_le_be_v16qi.
9017 * doc/extend.texi: Update the built-in documentation file for the
9018 new built-in functions.
9020 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9023 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9025 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9027 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9029 * config.gcc: Add z14.
9030 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9031 CPU model numbers for z13s and z14.
9032 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9034 * config/s390/s390-opts.h (enum processor_type): Rename
9035 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9036 * config/s390/s390.c (processor_table): Add field for CPU name to
9037 be passed to Binutils.
9038 (s390_asm_output_machine_for_arch): Use the new field in
9039 processor_table for Binutils.
9040 (s390_expand_builtin): Replace arch12 with z14.
9041 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9042 (s390_get_sched_attrmask): Likewise.
9043 (s390_get_unit_mask): Likewise.
9044 * config/s390/s390.opt: Add z14 to processor_type enum.
9046 2017-07-31 Martin Jambor <mjambor@suse.cz>
9049 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9050 regardless of optimization level.
9052 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9053 Martin Liska <mliska@suse.cz>
9055 * predict.def: Remove old comment and adjust probability.
9056 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9059 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9062 * config/i386/i386.c (ix86_function_naked): New function.
9063 (ix86_can_use_return_insn_p): Return false for naked functions.
9064 (ix86_expand_prologue): Skip prologue for naked functions.
9065 (ix86_expand_epilogue): Skip epilogue for naked functions
9066 and emit trap instruction.
9067 (ix86_warn_func_return): New function.
9068 (ix86_attribute_table): Add "naked" attribute specification.
9069 (TARGET_WARN_FUNC_RETURN): Define.
9070 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9072 2017-07-31 Martin Liska <mliska@suse.cz>
9074 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9075 (dump_gimple_bb_header): Always dump BB info.
9076 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9078 2017-07-31 Martin Liska <mliska@suse.cz>
9081 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9082 also with current_function_decl non-null equality.
9084 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9087 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9088 change type to the element type, instead add eltype variable and
9089 use it where we are interested in the element type.
9091 PR tree-optimization/81603
9092 * ipa-polymorphic-call.c
9093 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9094 offset arithmetic in offset_int, bail out if the resulting bit offset
9095 doesn't fit into shwi.
9097 2017-07-31 Martin Liska <mliska@suse.cz>
9099 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9100 (gimplify_save_expr): Fix comment.
9102 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9105 * config/i386/i386.c (ix86_function_arg): Update arguments for
9107 (ix86_compute_frame_layout): Set the initial stack offset to
9108 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9109 INCOMING_FRAME_SP_OFFSET.
9110 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9111 stack before exception handler returns.
9112 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9113 the 'ERROR_CODE' for exception handler.
9115 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9117 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9118 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9119 (ASM_OUTPUT_REG_POP): Ditto.
9120 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9121 instead of asm_fprintf to output pure string.
9123 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9125 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9126 to imported_module_or_decl hook.
9127 (debug_nothing_tree_tree_tree_bool): Remove.
9128 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9129 * debug.c (do_nothing_debug_hooks): Use
9130 debug_nothing_tree_tree_tree_bool_bool instead of
9131 debug_nothing_tree_tree_tree_bool.
9132 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9133 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9134 * sdbout.c (sdb_debug_hooks): Likewise.
9135 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9136 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9138 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9139 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9140 attribute, don't add anything.
9142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9144 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9145 (fold_build2_stat_loc): Likewise.
9146 (fold_build3_stat_loc): Likewise.
9147 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9148 (fold_build1_loc): Remove macro.
9149 (fold_build2_loc): Likewise.
9150 (fold_build3_loc): Likewise.
9152 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9154 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9155 (gimple_build_debug_bind_source_stat): Likewise.
9156 * gimple.h (gimple_build_debug_bind): Remove macro.
9157 (gimple_build_debug_bind_source): Likewise.
9159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9161 * bitmap.c (bitmap_alloc): Adjust.
9162 (bitmap_gc_alloc): Likewise.
9163 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9167 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9168 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9169 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9170 (bitmap_gc_alloc_stat): Likewise.
9171 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9173 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9175 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9176 * rtl.h (shallow_copy_rtx): Remove macro.
9178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9180 * emit-rtl.c (gen_raw_REG): Adjust.
9181 * gengenrtl.c (gendef): Likewise.
9182 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9183 * rtl.h (rtx_alloc): Remove macro.
9185 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9187 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9188 (build_tree_list_stat): Likewise.
9189 * tree.h (build_tree_list): Remove macro.
9190 (build_tree_list_vec): Likewise.
9192 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9194 * tree.c (make_vector_stat): Remove _stat from name.
9195 (build_vector_stat): Likewise.
9196 * tree.h (make_vector_stat): Remove macro.
9197 (build_vector_stat): Likewise.
9199 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9201 * tree.h (build_var_debug_value): Remove prototype.
9203 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9205 * tree.c (tree_cons_stat): Remove _stat from name.
9206 * tree.h (tree_cons): Remove macro.
9208 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9210 * tree.c (build_vl_exp_stat): Remove _stat from name.
9211 * tree.h (build_vl_exp): Remove macro.
9213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9215 * tree.c (build_decl_stat): Remove _stat from name.
9216 * tree.h (build_decl): Remove macro.
9218 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9220 * gimple.c (gimple_build_with_ops_stat): Adjust.
9221 (gimple_alloc_stat): Remove _stat from name.
9222 * gimple.h (gimple_alloc): Remove macro.
9224 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9226 * tree.c (make_tree_vec_stat): Remove _stat from name.
9227 (grow_tree_vec_stat): Likewise.
9228 * tree.h (make_tree_vec_stat): Adjust prototype.
9229 (grow_tree_vec_stat): Likewise.
9230 (make_tree_vec): Remove macro.
9231 (grow_tree_vec): Likewise.
9233 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9235 * fold-const.c (fold_build1_stat_loc): Adjust.
9236 (fold_build2_stat_loc): Likewise.
9237 (fold_build3_stat_loc): Likewise.
9238 * tree.c (build0_stat): Remove _stat from name.
9239 (build1_stat): Likewise.
9240 (build2_stat): Likewise.
9241 (build3_stat): Likewise.
9242 (build4_stat): Likewise.
9243 (build5_stat): Likewise.
9244 * tree.h (build1_loc): Remove macro, and rename _stat function
9246 (build2_loc): Likewise.
9247 (build3_loc): Likewise.
9248 (build4_loc): Likewise.
9249 (build5_loc): Likewise.
9251 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9253 * tree.c (make_int_cst_stat): Remove _stat from name.
9254 * tree.h (make_int_cst_stat): Adjust prototype.
9255 (make_int_cst): Remove macro.
9257 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9259 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9260 * tree.h (make_tree_binfo_stat): Adjust prototype.
9261 (make_tree_binfo): Remove.
9263 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9265 * tree.c (copy_node_stat): Rename to copy_node.
9266 (build_distinct_type_copy): Adjust.
9267 * tree.h (copy_node_stat): Adjust prototype.
9268 (copy_node): Remove macro.
9270 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9272 * tree.c (make_node_stat): rename to make_node.
9273 (build_tree_list_stat): Adjust.
9274 (build0_stat): Likewise.
9275 (build2_stat): Likewise.
9276 (build3_stat): Likewise.
9277 (build4_stat): Likewise.
9278 (build5_stat): Likewise.
9279 (build_decl_stat): Likewise.
9280 * tree.h (make_node_stat): Adjust prototype.
9281 (make_node): remove macro.
9283 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9285 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9286 (PPC_FEATURE2_SCV): Likewise.
9287 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9289 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9291 * config/aarch64/aarch64.c
9292 (aarch64_internal_mov_immediate): Add new special pattern.
9293 * config/aarch64/aarch64.md (*movdi_aarch64):
9294 Add reg/32bit const mov case.
9296 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9297 Richard Sandiford <richard.sandiford@linaro.org>
9299 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9300 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9301 Add integer and movi cases.
9302 (movi-split-hf-df-sf split, fp16): New.
9303 (enabled): Added TARGET_FP_F16INST.
9304 * config/aarch64/iterators.md (GPF_HF): New.
9305 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9307 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9309 * config/aarch64/aarch64.c
9310 (aarch64_simd_container_mode): Add prototype.
9311 (aarch64_expand_mov_immediate): Add HI support.
9312 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9313 (aarch64_can_const_movi_rtx_p): New.
9314 (aarch64_preferred_reload_class):
9315 Remove restrictions of using FP registers for certain SIMD operations.
9316 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9317 (aarch64_valid_floating_const): Add integer move validation.
9318 (aarch64_simd_imm_scalar_p): Remove.
9319 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9320 (aarch64_legitimate_constant_p): Expand list of supported cases.
9321 * config/aarch64/aarch64-protos.h
9322 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9323 (aarch64_reinterpret_float_as_int): New.
9324 (aarch64_simd_imm_scalar_p): Remove.
9325 * config/aarch64/constraints.md (Uvi): New.
9326 (Dd): Split into Ds and new Dd.
9327 * config/aarch64/aarch64.md (*movsi_aarch64):
9329 (*movdi_aarch64): Add SIMD mov case.
9331 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9333 * tree-predcom.c: (struct chain): Handle store-store chain in which
9334 stores for elimination only store loop invariant values.
9335 (execute_pred_commoning_chain): Ditto.
9336 (prepare_initializers_chain_store_elim): Ditto.
9337 (prepare_finalizers): Ditto.
9338 (is_inv_store_elimination_chain): New function.
9339 (initialize_root_vars_store_elim_1): New function.
9341 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9343 * tree-predcom.c: Revise general description of the pass.
9344 (enum chain_type): New enum type for store elimination.
9345 (struct chain): New field supporting store elimination.
9346 (struct component): Ditto.
9347 (dump_chain): Dump store-stores chain.
9348 (release_chain): Release resources.
9349 (split_data_refs_to_components): Compute and create component
9350 contains only stores for elimination.
9351 (get_chain_last_ref_at): New function.
9352 (make_invariant_chain): Initialization.
9353 (make_rooted_chain): Specify chain type in parameter and record it.
9354 (add_looparound_copies): Skip for store-stores chain.
9355 (determine_roots_comp): Compute type of chain and pass it to
9357 (initialize_root_vars_store_elim_2): New function.
9358 (finalize_eliminated_stores): New function.
9359 (remove_stmt): Handle store for elimination.
9360 (execute_pred_commoning_chain): Execute predictive commoning on
9362 (determine_unroll_factor): Skip unroll for store-stores chain.
9363 (prepare_initializers_chain_store_elim): New function.
9364 (prepare_initializers_chain): Hanlde store-store chain.
9365 (prepare_finalizers_chain, prepare_finalizers): New function.
9366 (tree_predictive_commoning_loop): Return integer value indicating
9367 if loop is unrolled or lcssa form is corrupted.
9368 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9370 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9372 * tree-predcom.c (initialize_root): Delete.
9373 (execute_pred_commoning_chain): Initialize root vars and replace
9374 reference of non-combined chain directly, rather than call above
9377 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9379 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9380 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9382 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9384 * tree-predcom.c (struct chain): New field init_seq.
9385 (release_chain): Release init_seq.
9386 (prepare_initializers_chain): Record intialization stmts in above
9388 (insert_init_seqs): New function.
9389 (tree_predictive_commoning_loop): Call insert_init_seqs.
9391 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9393 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9395 2017-07-28 Richard Biener <rguenther@suse.de>
9397 * match.pd: Remove superfluous :c.
9398 * genmatch.c (simplify::id): Add member.
9399 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9401 (current_id): New global.
9402 (dt_node::parent): Move from ...
9403 (dt_operand::parent): ... here. Add for_id member.
9404 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9405 (decision_tree::find_node): Relax order requirement when
9406 merging DT_TRUE nodes to ones inbetween the current simplify
9407 and the one we try to merge with. Add diagnostic whenever
9408 we need to enforce pattern order by not merging.
9409 (decision_tree::insert): Set current_id.
9410 (decision_tree::print_node): Dump parent node and for_id.
9411 (parser::last_id): Add member.
9412 (parser::push_simplify): Assign unique id.
9413 (parser::parser): Initialize last_id.
9415 2017-07-28 Martin Liska <mliska@suse.cz>
9418 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9419 gimple_build_debug_bind.
9421 2017-07-28 Richard Biener <rguenther@suse.de>
9423 PR tree-optimization/81502
9424 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9426 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9428 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9429 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9430 for BIT_FIELD_REF args.
9431 * fold-const.c (make_bit_field_ref): Likewise.
9432 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9434 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9437 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9438 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9439 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9440 Or it into SANITIZER_UNDEFINED.
9441 * ubsan.c: Include gimple-fold.h and varasm.h.
9442 (ubsan_expand_ptr_ifn): New function.
9443 (instrument_pointer_overflow): New function.
9444 (maybe_instrument_pointer_overflow): New function.
9445 (instrument_object_size): Formatting fix.
9446 (pass_ubsan::execute): Call instrument_pointer_overflow
9447 and maybe_instrument_pointer_overflow.
9448 * internal-fn.c (expand_UBSAN_PTR): New function.
9449 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9450 * sanitizer.def (__ubsan_handle_pointer_overflow,
9451 __ubsan_handle_pointer_overflow_abort): New builtins.
9452 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9453 * internal-fn.def (UBSAN_PTR): New internal function.
9454 * opts.c (sanitizer_opts): Add pointer-overflow.
9455 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9456 * fold-const.c (build_range_check): Compute pointer range check in
9457 integral type if pointer arithmetics would be needed. Formatting
9460 2017-07-28 Martin Liska <mliska@suse.cz>
9463 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9464 parameters that are of a variable-length.
9466 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9468 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9470 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9471 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9472 (CRT_CALL_STATIC_FUNCTION): Likewise.
9473 (ASM_DEFAULT_SPEC): New define.
9474 (ASM_SPEC32): Likewise.
9475 (ASM_SPEC64): Likewise.
9476 (ASM_SPEC_COMMON): Likewise.
9477 (ASM_SPEC): Likewise.
9478 (INVALID_64BIT): Likewise.
9479 (LINK_OS_DEFAULT_SPEC): Likewise.
9480 (LINK_OS_SPEC32): Likewise.
9481 (LINK_OS_SPEC64): Likewise.
9482 (POWERPC_LINUX): Likewise.
9483 (PTRDIFF_TYPE): Likewise.
9484 (RESTORE_FP_PREFIX): Likewise.
9485 (RESTORE_FP_SUFFIX): Likewise.
9486 (SAVE_FP_PREFIX): Likewise.
9487 (SAVE_FP_SUFFIX): Likewise.
9488 (SIZE_TYPE): Likewise.
9489 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9490 (TARGET_64BIT): Likewise.
9491 (TARGET_64BIT): Likewise.
9492 (TARGET_AIX): Likewise.
9493 (WCHAR_TYPE_SIZE): Likewise.
9494 (WCHAR_TYPE): Undefine.
9495 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9496 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9497 (CPP_OS_RTEMS_SPEC): Delete.
9498 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9499 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9501 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9503 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9505 PR tree-optimization/81578
9506 * tree-parloops.c (build_new_reduction): Bail out if
9507 reduction_code isn't one of the standard OpenMP reductions.
9508 Move the details printing after that decision.
9510 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9512 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9513 related to reload_in_progress.
9514 (splat_input_operand): Likewise.
9515 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9517 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9519 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9520 (TARGET_INSTANTIATE_DECLS): Likewise.
9521 (legitimate_indexed_address_p): Delete reload_in_progress code.
9522 (rs6000_debug_legitimate_address_p): Likewise.
9523 (rs6000_eliminate_indexed_memrefs): Likewise.
9524 (rs6000_emit_le_vsx_store): Likewise.
9525 (rs6000_emit_move_si_sf_subreg): Likewise.
9526 (rs6000_emit_move): Likewise.
9527 (register_to_reg_type): Likewise.
9528 (rs6000_pre_atomic_barrier): Likewise.
9529 (rs6000_machopic_legitimize_pic_address): Likewise.
9530 (rs6000_allocate_stack_temp): Likewise.
9531 (rs6000_address_for_fpconvert): Likewise.
9532 (rs6000_address_for_altivec): Likewise.
9533 (rs6000_secondary_memory_needed_rtx): Delete function.
9534 (rs6000_check_sdmode): Likewise.
9535 (rs6000_alloc_sdmode_stack_slot): Likewise.
9536 (rs6000_instantiate_decls): Likewise.
9537 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9538 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9539 Delete reload_in_progress.
9540 (*vec_reload_and_plus_<mptrsize>): Likewise.
9541 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9542 (vsx_div_v2di): Likewise.
9543 (vsx_udiv_v2di): Likewise.
9545 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9547 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9548 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9549 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9550 (rs6000_debug_reg_global): Delete print of LRA status.
9551 (rs6000_option_override_internal): Delete dead LRA related code.
9552 (rs6000_lra_p): Delete function.
9553 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9555 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9557 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9558 * config/riscv/rtems.h: New file.
9560 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9561 Sudakshina Das <sudi.das@arm.com>
9563 * config/aarch64/aarch64.md
9564 (define_split for and<mode>3nr_compare): Move
9565 non aarch64_logical_operand to a register.
9566 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9567 register immediate operand to a register.
9568 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9570 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9573 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9575 2017-07-27 Richard Biener <rguenther@suse.de>
9577 PR tree-optimization/81573
9578 PR tree-optimization/81494
9579 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9580 multi defuse cycle case.
9582 2017-07-27 Richard Biener <rguenther@suse.de>
9584 PR tree-optimization/81571
9585 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9588 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9590 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9591 earlier and only if MASK_FPU is set. Adjust formatting.
9593 2017-07-27 Martin Liska <mliska@suse.cz>
9595 * opt-functions.awk: Add validation of value of Init.
9596 * optc-gen.awk: Pass new argument.
9598 2017-07-27 Martin Liska <mliska@suse.cz>
9600 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9601 Fix wrong condition.
9603 2017-07-27 Martin Liska <mliska@suse.cz>
9605 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9606 BBs and edges seen by autoFDO.
9608 2017-07-27 Richard Biener <rguenther@suse.de>
9610 PR tree-optimization/81502
9611 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9612 with incompatible but same sized type.
9613 (execute_update_addresses_taken): Likewise.
9615 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9617 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9618 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9620 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9623 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9624 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9625 Change s_operand to memory_operand.
9627 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9629 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9630 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9631 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9632 Emit instructions rather than returning an expression. Handle TFmode
9633 and KFmode by casting to TImode.
9634 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9635 (rs6000_emit_le_vsx_store): Likewise.
9636 * config/rs6000/vsx.md (VSX_TI): New iterator.
9637 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9638 (*vsx_le_undo_permute_<mode>): Likewise.
9639 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9640 emit the split sequence.
9641 (*vsx_le_perm_store_<mode>): Likewise.
9643 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9645 PR tree-optimization/81555
9646 PR tree-optimization/81556
9647 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9648 if true, force CHANGED for the recursive invocation.
9649 (reassociate_bb): Remember original length of ops array, pass
9650 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9652 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9653 attributes for noipa attribute. For naked attribute use
9654 lookup_attribute first before lookup_attribute_spec.
9655 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9657 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9658 for functions with noipa attribute.
9659 (cgraph_externally_visible_p): Return true for functions with noipa
9661 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9662 for functions with noipa attribute.
9663 * doc/extend.texi: Document noipa function attribute.
9664 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9665 also for functions with noipa attribute.
9666 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9668 2017-07-26 Andrew Pinski <apinski@cavium.com>
9670 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9671 vec_unalign_load_cost and vec_unalign_store_cost.
9673 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9675 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9676 -mvsx-small-integer option.
9677 (ISA_3_0_MASKS_IEEE): Likewise.
9678 (OTHER_VSX_VECTOR_MASKS): Likewise.
9679 (POWERPC_MASKS): Likewise.
9680 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9681 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9682 code, only testing for DImode being allowed in non-VSX floating
9684 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9685 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9687 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9688 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9689 TARGET_P8_VECTOR test.
9690 (rs6000_secondary_reload_simple_move): Likewise.
9691 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9692 since TARGET_P9_VECTOR was already tested.
9693 (rs6000_opt_masks): Remove -mvsx-small-integer.
9694 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9695 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9697 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9698 test for TARGET_VEXTRACTUB was used, and that uses
9700 (p9 extract splitter): Likewise.
9701 (vsx_extract_<mode>_di_p9): Likewise.
9702 (vsx_extract_<mode>_store_p9): Likewise.
9703 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9704 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9705 the elimination of TARGET_VSX_SMALL_INTEGER.
9706 (vsx_extract_<mode>_p8): Likewise.
9707 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9708 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9709 (vsx_set_<mode>_p9): Likewise.
9710 (vsx_set_v4sf_p9): Likewise.
9711 (vsx_set_v4sf_p9_zero): Likewise.
9712 (vsx_insert_extract_v4sf_p9): Likewise.
9713 (vsx_insert_extract_v4sf_p9_2): Likewise.
9714 * config/rs6000/rs6000.md (sign extend splitter): Change
9715 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9716 (floatsi<mode>2_lfiwax_mem): Likewise.
9717 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9718 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9719 since a test for TARGET_P9_VECTOR was used.
9720 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9721 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9722 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9723 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9724 TARGET_P8_VECTOR test.
9725 (fix_trunc<mode>si2_stfiwx): Likewise.
9726 (fix_trunc<mode>si2_internal): Likewise.
9727 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9728 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9730 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9731 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9732 TARGET_P8_VECTOR test.
9733 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9734 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9735 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9737 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9738 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9739 since a test for TARGET_P9_VECTOR was used.
9740 (splitter for loading small constants): Likewise.
9742 2017-07-26 Andrew Pinski <apinski@cavium.com>
9744 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9747 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9750 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9751 (fp_valid_at): Likewise.
9753 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9755 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9756 (qdf24xx_addrcost_table): Likewise.
9757 (cortexa57_tunings): Update to use generic_branch_cost.
9758 (cortexa72_tunings): Likewise.
9759 (cortexa73_tunings): Likewise.
9760 (qdf24xx_tunings): Likewise.
9762 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9764 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9765 (thunderx2t99_branch_cost): Likewise.
9766 (cortexa35_tunings): Update to use generic_branch_cost.
9767 (cortexa53_tunings): Likewise.
9768 (cortexa57_tunings): Likewise.
9769 (cortexa72_tunings): Likewise.
9770 (cortexa73_tunings): Likewise.
9771 (thunderx2t99_tunings): Likewise.
9773 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9775 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9776 (sparc_option_override): Honour MASK_FSMULD.
9777 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9778 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9779 * config/sparc/sparc.opt (mfsmuld): New option.
9780 * doc/invoke.texi (mfsmuld): Document option.
9782 2017-07-26 Marek Polacek <polacek@redhat.com>
9785 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9787 2017-07-26 Richard Biener <rguenther@suse.de>
9789 * gimple-match-head.c (do_valueize): Return OP if valueize
9791 (get_def): New helper to get at the def stmt of a SSA name
9793 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9794 do_valueize to get at the def stmt.
9795 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9797 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9800 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9802 2017-07-26 Martin Liska <mliska@suse.cz>
9805 * function.c (expand_function_start): Make expansion of
9806 nonlocal_goto_save_area after parm_birth_insn.
9808 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9810 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9811 from all CPU target flags enable members.
9813 2017-07-26 Richard Biener <rguenther@suse.de>
9815 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9816 (decision_tree::gen): Make 'type' const.
9817 (write_predicate): Likewise.
9819 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9821 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9822 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9823 (rs6000_option_override_internal): Likewise.
9824 (rs6000_expand_vector_set): Likewise.
9825 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9826 (TARGET_UPPER_REGS_SF): Likewise.
9827 (TARGET_UPPER_REGS_DI): Likewise.
9828 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9829 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9830 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9831 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9832 (Splitters for DI constants in Altivec registers): Likewise.
9833 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9834 (vsx_set_v4sf_p9): Likewise.
9835 (vsx_set_v4sf_p9_zero): Likewise.
9836 (vsx_insert_extract_v4sf_p9): Likewise.
9837 (vsx_insert_extract_v4sf_p9_2): Likewise.
9839 2017-07-25 Carl Love <cel@us.ibm.com>
9841 * doc/extend.texi: Update the built-in documentation file for the
9842 existing built-in functions
9843 vector signed char vec_cnttz (vector signed char);
9844 vector unsigned char vec_cnttz (vector unsigned char);
9845 vector signed short vec_cnttz (vector signed short);
9846 vector unsigned short vec_cnttz (vector unsigned short);
9847 vector signed int vec_cnttz (vector signed int);
9848 vector unsigned int vec_cnttz (vector unsigned int);
9849 vector signed long long vec_cnttz (vector signed long long);
9850 vector unsigned long long vec_cnttz (vector unsigned long long);
9852 2017-07-25 Andrew Pinski <apinski@cavium.com>
9854 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9855 accesses where the use is for the first operand of a BIT_INSERT.
9857 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9860 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9861 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9863 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9865 * config/i386/gstabs.h: Delete.
9866 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9868 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9870 * config/i386/i386.c (ix86_decompose_address): Do not check for
9871 register RTX when looking at index_reg or base_reg.
9872 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9874 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9876 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9877 to update EH info here.
9879 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9881 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9883 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9885 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9887 2017-07-25 Torsten Duwe <duwe@suse.de>
9889 * common.opt: Introduce -fpatchable-function-entry
9890 command line option, and its variables function_entry_patch_area_size
9891 and function_entry_patch_area_start.
9892 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9893 including a two-value parser.
9894 * target.def (print_patchable_function_entry): New target hook.
9895 * targhooks.h (default_print_patchable_function_entry): New function.
9896 * targhooks.c (default_print_patchable_function_entry): Likewise.
9897 * toplev.c (process_options): Switch off IPA-RA if
9898 patchable function entries are being generated.
9899 * varasm.c (assemble_start_function): Look at the
9900 patchable-function-entry command line switch and current
9901 function attributes and maybe generate NOP instructions by
9902 calling the print_patchable_function_entry hook.
9903 * doc/extend.texi: Document patchable_function_entry attribute.
9904 * doc/invoke.texi: Document -fpatchable_function_entry
9905 command line option.
9906 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9908 * doc/tm.texi: Re-generate.
9910 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9913 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9914 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9916 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9918 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9920 2017-07-25 Richard Biener <rguenther@suse.de>
9922 PR tree-optimization/81455
9923 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9924 not walk in cycles when looking for guards.
9926 2017-07-25 Richard Biener <rguenther@suse.de>
9928 PR tree-optimization/81529
9929 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9930 when optimizing backedge uses.
9932 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9934 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9936 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9937 to dl_section_ref. On AIX, append an expression to subtract
9938 the size of the section length to dl_section_ref.
9940 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9942 * configure.ac: If any of the config.* scripts fail, exit 1.
9943 * configure: Regenerate.
9945 2017-07-25 Richard Biener <rguenther@suse.de>
9948 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9949 of immediate uses, be more verbose on errors.
9951 2017-07-25 Richard Biener <rguenther@suse.de>
9953 PR tree-optimization/81510
9954 * tree-vect-loop.c (vect_is_simple_reduction): When the
9955 reduction stmt is not inside the loop bail out.
9957 2017-07-25 Richard Biener <rguenther@suse.de>
9959 PR tree-optimization/81303
9960 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9961 profitability check against LOOP_VINFO_NITERSM1.
9963 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9965 * domwalk.c (cmp_bb_postorder): Simplify.
9966 (sort_bbs_postorder): New function. Use it...
9967 (dom_walker::walk): ...here to optimize common cases.
9969 2017-07-25 Martin Liska <mliska@suse.cz>
9972 * ipa-visibility.c (function_and_variable_visibility): Make the
9973 redirection just on target that supports aliasing.
9974 Fix GNU coding style.
9976 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9979 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9981 * config/arm/rtems.h: Likewise.
9982 * config/bfin/rtems.h: Likewise.
9983 * config/i386/rtemself.h: Likewise.
9984 * config/lm32/rtems.h: Likewise.
9985 * config/m32c/rtems.h: Likewise.
9986 * config/m68k/rtemself.h: Likewise.
9987 * config/microblaze/rtems.h: Likewise.
9988 * config/mips/rtems.h: Likewise.
9989 * config/moxie/rtems.h: Likewise.
9990 * config/nios2/rtems.h: Likewise.
9991 * config/powerpcspe/rtems.h: Likewise.
9992 * config/rs6000/rtems.h: Likewise.
9993 * config/rtems.h: Likewise.
9994 * config/sh/rtems.h: Likewise.
9995 * config/sh/rtemself.h: Likewise.
9996 * config/sparc/rtemself.h: Likewise.
9998 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10001 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10002 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10003 * tree-ssa-structalias.c (alias_get_name): Same.
10005 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10008 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10009 instructions if no du chain is found.
10011 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10013 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10015 2017-07-25 Richard Biener <rguenther@suse.de>
10017 PR middle-end/81505
10018 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10021 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10023 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10024 upper-regs options.
10025 (ISA_2_7_MASKS_SERVER): Likewise.
10026 (ISA_3_0_MASKS_IEEE): Likewise.
10027 (OTHER_P8_VECTOR_MASKS): Likewise.
10028 (OTHER_VSX_VECTOR_MASKS): Likewise.
10029 (POWERPC_MASKS): Likewise.
10030 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10031 duplicate list of options.
10032 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10033 explicit -mupper-regs options.
10034 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10035 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10036 alias for -mupper-regs-df.
10037 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10038 (rs6000_init_hard_regno_mode_ok): Likewise.
10039 (rs6000_option_override_internal): Likewise.
10040 (rs6000_opt_masks): Likewise.
10041 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10042 options in terms of whether -mvsx or -mpower8-vector was used.
10043 (TARGET_UPPER_REGS_DI): Likewise.
10044 (TARGET_UPPER_REGS_SF): Likewise.
10045 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10046 -mupper-regs-* options.
10048 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10050 * passes.c (emergency_dump_function): Print some empty lines and a
10051 header before the RTL dump.
10053 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10055 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10057 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10060 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10061 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10063 2017-07-24 Carl Love <cel@us.ibm.com>
10065 * config/rs6000/rs6000-c.c: Add support for built-in functions
10066 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10067 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10068 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10069 vec_extract_fp_from_shortl): Add defines for the two builtins.
10070 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10071 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10073 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10074 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10075 * doc/extend.texi: Update the built-in documentation file for the
10076 new built-in function.
10078 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10081 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10083 * doc/generic.texi: Likewise.
10084 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10085 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10087 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10089 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10090 (aarch64_mls_elt_merge<mode>): Likewise.
10092 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10094 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10095 having __cxa_atexit.
10097 2017-07-23 Michael Collison <michael.collison@arm.com>
10099 * config/arm/arm.c (arm_option_override): Deprecate
10100 use of -mstructure-size-boundary.
10101 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10102 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10104 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10107 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10108 Reduce cost estimate for direct moves.
10110 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10113 * config/i386/i386.c (ix86_option_override_internal): Disable
10114 BMI, BMI2 and TBM instructions for -m16.
10116 2017-07-21 Carl Love <cel@us.ibm.com>
10118 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10119 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10120 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10121 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10122 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10123 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10124 VMULOSW): New enum "unspec" values.
10125 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10126 altivec_vmulosw): New patterns.
10127 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10128 VMULOSW): Add definitions.
10130 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10132 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10133 (qdf24xx): Likewise.
10134 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10135 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10136 (AARCH64_FL_V8_1): Renumber.
10137 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10138 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10139 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10140 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10141 rdma to feature modifiers list.
10143 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10145 PR middle-end/56727
10146 * ipa-visibility (function_and_variable_visibility): Convert
10147 recursive PLT call to direct call if appropriate.
10149 2017-07-21 Andrew Pinski <apinski@cavium.com>
10151 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10152 operand 1 to see if the types precision matches.
10153 * fold-const.c (operand_equal_p): Likewise.
10155 2017-07-21 Richard Biener <rguenther@suse.de>
10157 PR tree-optimization/81303
10158 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10159 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10160 (vect_peeling_hash_get_lowest_cost): Adjust.
10161 (vect_enhance_data_refs_alignment): Likewise. Use
10162 vect_get_peeling_costs_all_drs to compute the penalty for no
10163 peeling to match up costs.
10165 2017-07-21 Richard Biener <rguenther@suse.de>
10167 PR tree-optimization/81500
10168 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10169 we didn't identify a reduction path.
10171 2017-07-21 Tom de Vries <tom@codesourcery.com>
10172 Cesar Philippidis <cesar@codesourcery.com>
10174 PR gcov-profile/81442
10175 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10178 2017-07-21 Tom de Vries <tom@codesourcery.com>
10181 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10183 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10184 nvptx_override_options_after_change.
10186 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10188 * dwarf2out.c (output_file_names): Avoid double testing for
10189 dwarf_version >= 5.
10191 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10193 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10195 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10197 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10199 (try_crossjump_to_edge): Do not punt on partitioned functions.
10201 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10203 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10204 Put all BBs reachable only via paths crossing cold region to cold
10206 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10208 2016-07-21 Richard Biener <rguenther@suse.de>
10210 PR tree-optimization/81303
10211 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10212 into account prologue and epilogue iterations when raising
10213 min_profitable_iters to sth at least covering one vector iteration.
10215 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10217 * config/arm/arm.c (arm_test_cpu_arch_dat):
10220 2017-07-20 Nathan Sidwell <nathan@acm.org>
10222 Remove TYPE_METHODS.
10223 * tree.h (TYPE_METHODS): Delete.
10224 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10225 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10226 (dbxout_type_methods): Scan TYPE_FIELDS.
10227 (dbxout_type): Don't check TYPE_METHODS here.
10228 * function.c (use_register_for_decl): Always ignore register for
10229 class types when not optimizing.
10230 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10231 * tree.c (free_lang_data_in_type): Stitch out member functions and
10232 templates from TYPE_FIELDS.
10233 (build_distinct_type_copy, verify_type_variant,
10234 verify_type): Member fns are on TYPE_FIELDS.
10235 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10236 * tree-pretty-print.c (dump_generic_node): Likewise.
10238 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10241 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10242 V2TImode and V4TImode.
10243 (ix86_expand_vector_extract): Likewise.
10244 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10245 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10246 (ssescalarmode): Handle V4TImode and V2TImode.
10247 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10248 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10249 (VEXTRACTI128_MODE): New mode iterator.
10250 (splitter for *vec_extractv?ti first element): New.
10251 (VEC_INIT_MODE): New mode iterator.
10252 (vec_init<mode>): Consolidate 3 expanders into one using
10253 VEC_INIT_MODE mode iterator.
10255 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10257 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10258 non_spilled_static_chain_regno_p.
10260 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10262 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10264 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10266 * bb-reorder.c (connect_traces): Allow copying of blocks within
10269 2017-07-20 Richard Biener <rguenther@suse.de>
10271 * gimple.h (gimple_phi_result): Add gphi * overload.
10272 (gimple_phi_result_ptr): Likewise.
10273 (gimple_phi_arg): Likewise. Adjust index assert to only
10274 allow actual argument accesses rather than all slots available
10276 (gimple_phi_arg_def): Add gphi * overload.
10277 * tree-phinodes.c (make_phi_node): Initialize only actual
10279 (resize_phi_node): Clear memory not covered by old node,
10280 do not initialize excess argument slots.
10281 (reserve_phi_args_for_new_edge): Initialize new argument slot
10284 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10286 PR tree-optimization/81388
10288 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10290 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10291 by removing computation of may_be_zero.
10293 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10294 Tom de Vries <tom@codesourcery.com>
10296 PR middle-end/81030
10297 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10298 when gimple level profile disagrees with what RTL expander did.
10300 2017-07-20 Richard Biener <rguenther@suse.de>
10302 PR tree-optimization/61171
10303 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10304 (vect_analyze_stmt): Add slp instance parameter.
10305 (vectorizable_reduction): Likewise.
10306 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10307 (vect_is_simple_reduction): Deal with chains not detected
10308 as SLP reduction chain, specifically not properly associated
10309 chains containing a mix of plus/minus.
10310 (get_reduction_op): Remove.
10311 (get_initial_defs_for_reduction): Simplify, pass in whether
10312 this is a reduction chain, pass in the SLP node for the PHIs.
10313 (vect_create_epilog_for_reduction): Get the SLP instance as
10315 (vectorizable_reduction): Get the SLP instance as arg.
10316 During analysis remember the SLP node with the PHIs in the
10317 instance. Simplify getting at the vectorized reduction PHIs.
10318 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10319 through SLP instance.
10320 (vect_slp_analyze_operations): Likewise.
10321 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10322 (vect_transform_stmt): Likewise.
10324 2017-07-20 Tom de Vries <tom@codesourcery.com>
10326 PR tree-optimization/81489
10327 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10328 read of phi arg location to before loop that modifies phi.
10330 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10332 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10335 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10337 PR middle-end/81331
10338 * except.c (execute): Fix ordering issue.
10340 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10342 PR rtl-optimization/81423
10343 * combine.c (make_compound_operation_int): Don't try to optimize
10344 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10346 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10348 PR rtl-optimization/81423
10349 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10350 with a constant that is -1 in the truncated to mode.
10352 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10354 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10355 (determine_unlikely_bbs): ... here.
10356 * predict.h (propagate_unlikely_bbs_forward): Declare.
10357 * cfgexpand.c (pass_expand::execute): Use it.
10358 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10360 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10361 propagate_unlikely_bbs_forward.
10363 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10365 PR middle-end/81331
10366 * except.c (maybe_add_nop_after_section_switch): New function.
10369 2017-07-19 Tom de Vries <tom@codesourcery.com>
10371 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10373 2017-07-19 Tom de Vries <tom@codesourcery.com>
10375 * gimple.h (gimple_phi_arg): Make assert more strict.
10377 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10379 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10380 * config/rs6000/mmintrin.h: New file.
10381 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10383 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10385 PR tree-optimization/81346
10386 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10388 2017-07-19 Tom de Vries <tom@codesourcery.com>
10390 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10392 2017-07-19 Tom de Vries <tom@codesourcery.com>
10394 * config/nvptx/nvptx-modes.def: Add V2DImode.
10395 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10396 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10397 (nvptx_output_mov_insn): Handle lack of mov.b128.
10398 (nvptx_print_operand): Handle 'H' and 'L' codes.
10399 (nvptx_vector_mode_supported): Allow V2DImode.
10400 (nvptx_preferred_simd_mode): New function.
10401 (nvptx_data_alignment): New function.
10402 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10403 nvptx_preferred_simd_mode.
10404 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10406 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10408 2017-07-19 Tom de Vries <tom@codesourcery.com>
10410 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10411 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10412 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10413 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10414 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10415 (mov<VECIM>_insn): New define_insn.
10416 (define_expand "mov<VECIM>): New define_expand.
10418 2017-07-19 Tom de Vries <tom@codesourcery.com>
10420 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10422 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10424 PR tree-optimization/81346
10425 * fold-const.h (fold_div_compare, range_check_type): Declare.
10426 * fold-const.c (range_check_type): New function.
10427 (build_range_check): Use range_check_type.
10428 (fold_div_compare): No longer static, rewritten into
10429 a match.pd helper function.
10430 (fold_comparison): Don't call fold_div_compare here.
10431 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10432 as helper function.
10434 2017-07-19 Nathan Sidwell <nathan@acm.org>
10436 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10437 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10438 * tree.c (find_decls_types_r, verify_type): Use
10439 TYPE_{MIN,MAX}_VALUE_RAW.
10440 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10441 (hash_tree): Likewise.
10442 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10444 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10447 2017-07-18 Tom de Vries <tom@codesourcery.com>
10449 PR middle-end/81464
10450 * omp-expand.c (expand_omp_for_static_chunk): Handle
10451 equal-argument loop exit phi.
10453 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10456 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10457 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10458 operand 2 predicate.
10459 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10460 operand 2 predicate.
10461 (ror,rol -> rorx splitters): Use const_int_operand as
10462 operand 2 predicate.
10464 2017-06-18 Richard Biener <rguenther@suse.de>
10466 PR tree-optimization/81410
10467 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10468 the gap in the ! slp_perm SLP case after each group.
10470 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10472 PR middle-end/81463
10473 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10476 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10478 PR middle-end/81462
10479 * predict.c (set_even_probabilities): Cleanup; do not affect
10480 probabilities that are already known.
10481 (combine_predictions_for_bb): Call even when count is set.
10483 2017-07-18 Nathan Sidwell <nathan@acm.org>
10485 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10488 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10491 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10492 optimization for loop niter analysis.
10494 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10497 * config/avr/avr.c (avr_optimize_casesi): Don't use
10498 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10500 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10502 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10503 body_cost_vec from _vect_peel_extended_info.
10504 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10505 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10508 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10510 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10512 2017-07-18 Richard Biener <rguenther@suse.de>
10514 PR tree-optimization/80620
10515 PR tree-optimization/81403
10516 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10517 info when re-using a VN table entry.
10519 2017-07-18 Richard Biener <rguenther@suse.de>
10521 PR tree-optimization/81418
10522 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10523 vectype_in. Verify that with lane-reducing reduction operations
10524 we have a single def-use cycle.
10526 2017-07-17 Carl Love <cel@us.ibm.com>
10528 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10530 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10531 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10532 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10534 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10535 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10536 VMULOSW): New enum "unspec" values.
10537 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10538 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10539 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10540 altivec_vmulosw): New patterns.
10541 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10542 VMULOSW): Add definitions.
10543 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10545 * config/alpha/alpha.c: Include predict.h.
10547 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10549 * tree-vrp.c (compare_assert_loc): Fix comparison function
10550 to return predictable results.
10552 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10554 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10556 (subdi3): Likewise.
10557 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10558 * doc/invoke.texi (mexpand-adddi): Update text.
10560 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10562 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10563 that also clobbers the CC register. The old expand code is moved
10565 (*arc_clzsi2): ... here.
10566 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10567 the CC register. The old expand code is moved to ...
10568 (arc_ctzsi2): ... here.
10570 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10572 * config/arc/arc.opt (mindexed-loads): Use initial value
10573 TARGET_INDEXED_LOADS_DEFAULT.
10574 (mauto-modify-reg): Use initial value
10575 TARGET_AUTO_MODIFY_REG_DEFAULT.
10576 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10577 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10578 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10579 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10581 2017-07-17 Martin Liska <mliska@suse.cz>
10584 * opts.c (finish_options): Do not allow -fgnu-tm
10585 w/ -fsanitize={kernel-,}address. Say sorry.
10587 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10590 * tree-loop-distribution.c (classify_partition): Only assert on
10591 numer of iterations.
10592 (merge_dep_scc_partitions): Delete prameter. Update function call.
10593 (distribute_loop): Remove code handling loop with unknown niters.
10594 (pass_loop_distribution::execute): Skip loop with unknown niters.
10596 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10599 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10600 function sort_partitions_by_post_order.
10602 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10604 PR tree-optimization/81374
10605 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10606 the max index of basic blocks, rather than number of basic blocks.
10608 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10610 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10612 (arc_legitimate_pic_operand_p): Likewise.
10613 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10615 (arc_needs_pcl_p): Likewise.
10616 (arc_legitimate_pc_offset_p): Likewise.
10617 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10618 function is also used in constrains.md.
10619 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10620 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10621 PLUS. Only return true/false in known cases, otherwise assert.
10622 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10623 is already called in arc_legitimate_constant_p.
10624 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10626 (LEGITIMATE_PIC_OPERAND_P): Use
10627 arc_raw_symbolic_reference_mentioned_p function.
10628 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10633 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10634 Andrew Burgess <andrew.burgess@embecosm.com>
10636 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10637 (arc_return_address_register): New function.
10638 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10639 (arc_handle_fndecl_attribute): Add naked attribute.
10640 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10641 (TARGET_WARN_FUNC_RETURN): Likewise.
10642 (arc_allocate_stack_slots_for_args): New function.
10643 (arc_warn_func_return): Likewise.
10644 (machine_function): Change type fn_type.
10645 (arc_compute_function_type): Consider new naked function type,
10646 change function return type.
10647 (arc_must_save_register): Adapt to handle new
10648 arc_compute_function_type's return type.
10649 (arc_expand_prologue): Likewise.
10650 (arc_expand_epilogue): Likewise.
10651 (arc_return_address_regs): Delete.
10652 (arc_return_address_register): New function.
10653 (arc_epilogue_uses): Use above function.
10654 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10655 (arc_function_type): Change encoding, add naked type.
10656 (ARC_INTERRUPT_P): Change to handle the new encoding.
10657 (ARC_FAST_INTERRUPT_P): Likewise.
10658 (ARC_NORMAL_P): Define.
10659 (ARC_NAKED_P): Likewise.
10660 (arc_compute_function_type): Delete prototype.
10661 * config/arc/arc.md (in_ret_delay_slot): Use
10662 arc_return_address_register function.
10663 (simple_return): Likewise.
10664 (p_return_i): Likewise.
10666 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10668 PR tree-optimization/81428
10669 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10670 can't be built for those types.
10672 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10674 Remove stuff dead since r239246.
10676 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10677 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10678 (avr_inform_devices): Remove dead stuff.
10680 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10682 * config/arm/arm_neon.h: Fix softp typo.
10684 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10686 PR tree-optimization/81365
10687 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10688 aggregate moves onto bb predecessor edges, make sure there are no
10689 loads that could alias the lhs in between the start of bb and the
10692 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10695 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10696 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10697 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10699 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10701 PR tree-optimization/81396
10702 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10703 (init_symbolic_number): Initialize it to 1.
10704 (perform_symbolic_merge): Add n_ops from both operands into the new
10706 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10707 without base_addr as useless if they need more than one operation.
10708 (bswap_replace): Handle !bswap case for NULL base_addr.
10710 2017-07-17 Tom de Vries <tom@codesourcery.com>
10713 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10716 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10718 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10719 conditional builtin define __FIX_LEON3FT_B2BST.
10721 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10723 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10724 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10727 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10729 PR rtl-optimization/81424
10730 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10731 to remove potential trapping from operands if -fnon-call-exceptions.
10733 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10735 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10736 profile_proability for scalling.
10737 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10739 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10741 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10743 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10745 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10746 fixpoint arithmetics.
10748 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10750 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10751 fixpoint arithmetics.
10753 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10755 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10756 fixpoint arithmetics.
10758 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10760 * profile-count.h (profile_probability::from_reg_br_prob_note,
10761 profile_probability::to_reg_br_prob_note): New functions.
10762 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10763 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10764 * predict.c (probability_reliable_p): Update.
10765 (edge_probability_reliable_p): Update.
10766 (br_prob_note_reliable_p): Update.
10767 (invert_br_probabilities): Update.
10768 (add_reg_br_prob_note): New function.
10769 (combine_predictions_for_insn): Update.
10770 * asan.c (asan_clear_shadow): Update.
10771 * cfgbuild.c (compute_outgoing_frequencies): Update.
10772 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10773 (update_br_prob_note): Update.
10774 (rtl_verify_edges): Update.
10775 (purge_dead_edges): Update.
10776 (fixup_reorder_chain): Update.
10777 * emit-rtl.c (try_split): Update.
10778 * ifcvt.c (cond_exec_process_insns): Update.
10779 (cond_exec_process_if_block): Update.
10780 (dead_or_predicable): Update.
10781 * internal-fn.c (expand_addsub_overflow): Update.
10782 (expand_neg_overflow): Update.
10783 (expand_mul_overflow): Update.
10784 * loop-doloop.c (doloop_modify): Update.
10785 * loop-unroll.c (compare_and_jump_seq): Update.
10786 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10787 * predict.h: Update.
10788 * reorg.c (mostly_true_jump): Update.
10790 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10791 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10792 * config/arc/arc.c: (emit_unlikely_jump): Update.
10793 * config/arm/arm.c: (emit_unlikely_jump): Update.
10794 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10795 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10796 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10797 (ix86_print_operand): Update.
10798 (ix86_split_fp_branch): Update.
10799 (predict_jump): Update.
10800 * config/ia64/ia64.c (ia64_print_operand): Update.
10801 * config/mmix/mmix.c (mmix_print_operand): Update.
10802 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10803 (rs6000_expand_split_stack_prologue): Update.
10804 * config/rs6000/rs6000.c: Update.
10805 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10806 (s390_expand_vec_movstr): Update.
10807 (s390_expand_cs_tdsi): Update.
10808 (s390_expand_split_stack_prologue): Update.
10809 * config/sh/sh.c (sh_print_operand): Update.
10810 (expand_cbranchsi4): Update.
10811 (expand_cbranchdi4): Update.
10812 * config/sparc/sparc.c (output_v9branch): Update.
10813 * config/spu/spu.c (get_branch_target): Update.
10814 (ea_load_store_inline): Update.
10815 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10816 * config/tilepro/tilepro.c: Update.
10818 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10820 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10821 (gimplify_save_expr): Likewise.
10823 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10825 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10827 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10829 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10831 * timevar.def (TV_IPA_FNSUMMARY): Define.
10833 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10835 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10836 to back store errata sensitive sequence from being generated.
10837 (sqrtdf2_fix): Likewise.
10839 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10841 * tree-ssa-threadupdate.c (compute_path_counts,
10842 update_joiner_offpath_counts): Use profile_probability.
10844 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10847 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10849 * config/arm/arm-c.c (arm_cpu_builtins): Define
10850 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10852 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10854 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10855 array entries to represent __ieee128 versions of the
10856 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10857 scalar_extract_sig, and scalar_insert_exp built-in functions.
10858 (altivec_resolve_overloaded_builtin): Add special case handling
10859 for the __builtin_scalar_insert_exp function, as represented by
10860 the P9V_BUILTIN_VEC_VSIEDP constant.
10861 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10862 exponent support for __ieee128 argument.
10863 (VSESQP): Add scalar extract signature support for __ieee128
10865 (VSTDCNQP): Add scalar test negative support for __ieee128
10867 (VSIEQP): Add scalar insert exponent support for __int128 argument
10868 with __ieee128 result.
10869 (VSIEQPF): Add scalar insert exponent support for __ieee128
10870 argument with __ieee128 result.
10871 (VSTDCQP): Add scalar test data class support for __ieee128
10873 (VSTDCNQP): Add overload support for scalar test negative with
10874 __ieee128 argument.
10875 (VSTDCQP): Add overload support for scalar test data class
10876 __ieee128 argument.
10877 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10878 UNSPEC_VSX_SXSIGDP.
10879 (UNSPEC_VSX_SIEXPQP): New constant.
10880 (xsxexpqp): New insn for VSX scalar extract exponent quad
10882 (xsxsigqp): New insn for VSX scalar extract significand quad
10884 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10885 precision with floating point argument.
10886 (xststdcqp): New expand for VSX scalar test data class quad
10888 (xststdcnegqp): New expand for VSX scalar test negative quad
10890 (xststdcqp): New insn to match expansions for VSX scalar test data
10891 class quad precision and VSX scalar test negative quad precision.
10892 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10893 special case operand checking to enforce that second operand of
10894 VSX scalar test data class with quad precision argument is a 7-bit
10896 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10897 prototypes and descriptions of __ieee128 versions of
10898 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10899 scalar_test_data_class, and scalar_test_neg built-in functions.
10901 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10903 PR tree-optimization/81162
10904 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10905 replace a negate with an add.
10907 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10909 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10911 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10913 * config/arm/arm-c.c (arm_cpu_builtins): Define
10914 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10916 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10918 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10919 (armv8-r): Set ARM Cortex-R52 as default CPU.
10920 * config/arm/arm-tables.opt: Regenerate.
10921 * config/arm/arm-tune.md: Regenerate.
10922 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10924 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10925 extension for -mcpu=cortex-r52.
10927 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10929 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10930 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10931 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10932 (fp-armv8): Define it as FP_ARMv8 only.
10933 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10934 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10936 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10937 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10938 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10939 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10940 than TARGET_FPU_ARMV8.
10941 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10942 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10943 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10945 * config/arm/neon.md (neon_vrint): Likewise.
10946 (neon_vcvt): Likewise.
10947 (neon_<fmaxmin_op><mode>): Likewise.
10948 (<fmaxmin><mode>3): Likewise.
10949 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10950 * config/arm/predicates.md (arm_cond_move_operator): Check against
10951 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10953 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10955 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10956 to top of function.
10958 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10960 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10961 loop in comment with memset.
10963 2017-07-14 Martin Liska <mliska@suse.cz>
10965 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10966 * dwarf2out.c (is_java): Remove the function.
10967 (output_pubname): Remove usage of the function.
10968 (lower_bound_default): Remove usage of DW_LANG_Java.
10969 (gen_compile_unit_die): Likewise.
10970 * gcc.c: Remove compiler defaults for .java and .zip files.
10971 * gimple-expr.c (remove_suffix): Change as there's no longer
10972 extension than 4-letter one.
10973 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10974 (gimplify_save_expr): Likewise.
10975 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10976 as it's possible even for other languages than Java.
10977 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10978 * lto-opts.c (lto_write_options): Remove reference to Java.
10979 * opts.c (strip_off_ending): Update file extension handling.
10980 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10981 * tree-eh.c (lower_resx): Likewise.
10982 * tree.c (free_lang_data_in_type): Remove dead code.
10983 (find_decls_types_r): Likewise.
10984 (build_common_builtin_nodes): Remove Java from a comment.
10985 (verify_type): Remove dead code.
10986 * varasm.c (assemble_external): Remove Java from a comment.
10988 2017-07-14 Martin Liska <mliska@suse.cz>
10990 * opts.c (finish_options): Add quotes.
10991 (common_handle_option): Likewise.
10993 2017-07-14 Martin Liska <mliska@suse.cz>
10995 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10996 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10997 Remove N_SO_PASCAL.
10998 * dwarf2out.c (lower_bound_default): Do not handle
11000 (gen_compile_unit_die): Likewise.
11001 * gcc.c: Remove default extension binding for GNU Pascal.
11002 * stmt.c: Remove Pascal language from a comment.
11003 * xcoffout.c: Likewise.
11005 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11008 * diagnostic-show-locus.c (fixit_cmp): New function.
11009 (layout::layout): Sort m_fixit_hints.
11010 (column_range::column_range): Assert that the values are valid.
11011 (struct char_span): New struct.
11012 (correction::overwrite): New method.
11013 (struct source_line): New struct.
11014 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11015 calls in terms of classes source_line and char_span, and
11016 correction::overwrite.
11017 (selftest::test_overlapped_fixit_printing_2): New function.
11018 (selftest::diagnostic_show_locus_c_tests): Call it.
11020 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11022 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11023 early if there is no lhs.
11025 2017-07-13 Martin Liska <mliska@suse.cz>
11027 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11028 (gen_reference_type_die): Likewise.
11029 * stor-layout.c: Remove Pascal-related comment.
11031 2017-07-13 Martin Liska <mliska@suse.cz>
11033 * opts.c (finish_options): Add quotes to error messages.
11034 (parse_sanitizer_options): Likewise.
11036 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11038 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11040 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11042 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11044 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11046 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11048 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11050 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11053 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11054 provides the hardware capability bits, define the macro
11055 __BUILTIN_CPU_SUPPORTS__.
11056 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11057 if GLIBC does not provide the hardware capability bits. Add a
11058 gcc_unreachable call if the built-in cpu function is neither
11059 __builtin_cpu_is nor __builtin_cpu_supports.
11060 (rs6000_get_function_versions_dispatcher): Change the warning
11061 that an old GLIBC is used which does not export the capability
11062 bits to be an error.
11063 * doc/extend.texi (target_clones attribute): Document the
11064 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11065 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11066 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11067 the macros defined by GCC if the newer GLIBC is available.
11069 2017-07-12 Jeff Law <law@redhat.com>
11071 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11072 remaining includes slightly.
11073 * config/riscv/riscv-builtins.c: Include profile-count.h.
11075 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11078 * config/avr/avr.c (avr_set_current_function): In diagnostic
11079 messages: Quote keywords and (parts of) identifiers.
11080 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11083 2017-07-12 Carl Love <cel@us.ibm.com>
11085 * config/rs6000/rs6000-c.c: Add support for built-in functions
11086 vector bool char vec_revb (vector bool char);
11087 vector bool short vec_revb (vector short char);
11088 vector bool int vec_revb (vector bool int);
11089 vector bool long long vec_revb (vector bool long long);
11090 * doc/extend.texi: Update the built-in documentation file for the
11091 new built-in functions.
11093 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11095 * config/s390/s390.md: Remove movcc splitter.
11097 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11099 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11100 load/store on condition.
11102 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11105 * config/avr/avr.c (avr_encode_section_info)
11106 [progmem && !TREE_READONLY]: Error if progmem object needs
11109 2017-07-11 Michael Collison <michael.collison@arm.com>
11111 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11114 2017-07-11 Carl Love <cel@us.ibm.com>
11116 * config/rs6000/rs6000-c.c: Add support for builtins
11117 vector unsigned int vec_parity_lsbb (vector signed int);
11118 vector unsigned int vec_parity_lsbb (vector unsigned int);
11119 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11120 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11121 vector unsigned long long vec_parity_lsbb (vector signed long long);
11122 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11123 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11124 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11125 * doc/extend.texi: Update the built-in documentation file for the
11126 new built-in functions.
11128 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11130 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11131 (layout::m_primary_loc): New field.
11132 (layout::layout): Initialize new field. Move location filtering
11133 logic from here to...
11134 (layout::maybe_add_location_range): ...this new method. Add
11135 support for filtering to just the lines already specified by other
11137 (layout::will_show_line_p): New method.
11138 (gcc_rich_location::add_location_if_nearby): New method.
11139 (selftest::test_add_location_if_nearby): New test function.
11140 (selftest::diagnostic_show_locus_c_tests): Call it.
11141 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11144 2017-07-11 Tom de Vries <tom@codesourcery.com>
11146 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11147 (bb_first_real_insn): New function.
11148 (nvptx_single): Add extra initialization of broadcasted condition
11151 2017-07-11 Nathan Sidwell <nathan@acm.org>
11153 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11155 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11157 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11158 Binutils doc as TEXI will mess them up.
11159 * doc/invoke.texi (AVR Options): Same here.
11161 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11163 * config/sparc/sparc.opt (mfix-ut700): New option.
11164 (mfix-gr712rc): Likewise.
11165 (sparc_fix_b2bst): New variable.
11166 * doc/invoke.texi (SPARC options): Document them.
11167 (ARM options): Fix warnings.
11168 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11169 instructions to prevent sequences that can trigger the store-store
11170 errata for certain LEON3FT processors.
11171 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11172 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11173 * config/sparc/sparc.md (fix_b2bst): New attribute.
11174 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11176 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11179 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11181 (*rsqrtsf2_sse): Ditto.
11183 (div<mode>3): Macroize insn from divdf3 and divsf3
11184 using MODEF mode iterator.
11186 2017-07-10 Martin Sebor <msebor@redhat.com>
11188 PR tree-optimization/80397
11189 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11190 instead of testing for equality to INTEGER_TYPE.
11192 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11194 * config.gcc: Remove uclibc from arc target spec.
11196 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11198 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11200 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11203 * lto-wrapper.c (remove_option): New function.
11204 (merge_and_complain): Merge PIC/PIE options more realistically.
11206 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11208 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11212 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11213 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11214 * config.in: Regenerate.
11215 * configure: Regenerate.
11216 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11217 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11218 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11219 (TARGET_GASISR_PROLOGUES): ...target mask.
11220 * common/config/avr/avr-common.c
11221 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11222 Set -mgas-isr-prologues.
11223 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11224 INSERT_PASS_BEFORE for it.
11225 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11226 * config/avr/avr.c (avr_option_override)
11227 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11228 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11229 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11230 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11231 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11233 (make_avr_pass_pre_proep): New function.
11234 (emit_push_sfr) <treg>: Add argument to function and use it
11235 instead of TMP_REG.
11236 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11237 and set machine->gasisr.yes.
11238 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11239 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11240 __gcc_isr.n_pushed to .L__stack_usage.
11241 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11242 (avr_asm_final_postscan_insn): ...this new static function.
11243 * config/avr/avr.h (machine_function)
11244 <is_no_gccisr, use_L__stack_usage>: New fields.
11245 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11246 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11247 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11248 (gasisr, *gasisr): New expander and insn.
11249 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11250 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11251 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11253 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11255 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11258 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11260 Move jump-tables out of .text again.
11263 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11264 (ASM_OUTPUT_ADDR_VEC): New function.
11265 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11266 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11267 INSN_ADDRESSes as asm comment.
11268 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11269 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11270 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11271 * config/avr/avr.md (*tablejump): Adjust comment.
11272 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11273 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11275 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11276 (avr_output_addr_vec): New proto.
11277 (avr_log_t) <insn_addresses>: New field.
11279 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11282 * config/i386/i386.c (ix86_function_arg_advance): Set
11283 outgoing_args_on_stack to true if there are outgoing arguments
11285 (ix86_function_arg): Likewise.
11286 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11287 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11288 * config/i386/i386.h (machine_function): Add
11289 outgoing_args_on_stack.
11291 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11293 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11294 supporting pthreds.
11295 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11297 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11299 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11300 (REAL_H): Remove $(MACHMODE_H).
11301 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11303 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11304 $(MACHMODE_H) and double-int.h.
11305 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11307 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11310 2017-07-07 Andrew Pinski <apinski@cavium.com>
11312 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11313 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11315 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11317 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11318 Add warning if GCC was not configured to link against a GLIBC that
11319 exports the hardware capability bits.
11320 (make_resolver_func): Make resolver function private and not a
11321 COMDAT function. Create the name with clone_function_name instead
11322 of make_unique_name.
11325 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11326 correct operand in doing the split.
11328 2017-07-07 Carl Love <cel@us.ibm.com>
11330 * config/rs6000/rs6000-c: Add support for built-in function
11331 vector unsigned short vec_pack_to_short_fp32 (vector float,
11333 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11334 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11335 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11336 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11337 (convert_4f32_8i16): Add define_expand.
11338 * doc/extend.texi: Update the built-in documentation file for the
11339 new built-in function.
11341 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11343 * config/sparc/m8.md: New file.
11344 * config/sparc/sparc.md: Include m8.md.
11346 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11348 * config/sparc/sparc.opt: New option -mvis4b.
11349 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11350 (sparc_option_override): Handle VIS4B.
11351 (enum sparc_builtins): Define
11352 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11353 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11354 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11355 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11356 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11357 (check_constant_argument): New function.
11358 (sparc_vis_init_builtins): Define builtins
11359 __builtin_vis_dictunpack{8,16,32},
11360 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11361 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11362 __builtin_vis_fpcmpde{8,16,32}shl and
11363 __builtin_vis_fpcmpur{8,16,32}shl.
11364 (sparc_expand_builtin): Check that the constant operands to
11365 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11366 constant and in range.
11367 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11369 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11370 (SPARC_IMM5_P): Likewise.
11371 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11372 (enabled): Handle vis4b.
11373 (UNSPEC_DICTUNPACK): New unspec.
11374 (UNSPEC_FPCMPSHL): Likewise.
11375 (UNSPEC_FPUCMPSHL): Likewise.
11376 (UNSPEC_FPCMPDESHL): Likewise.
11377 (UNSPEC_FPCMPURSHL): Likewise.
11378 (cpu_feature): New CPU feature `vis4b'.
11379 (dictunpack{8,16,32}): New insns.
11380 (FPCSMODE): New mode iterator.
11381 (fpcscond): New code iterator.
11382 (fpcsucond): Likewise.
11383 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11384 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11385 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11386 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11387 * config/sparc/constraints.md: Define constraints `q' for unsigned
11388 2-bit integer constants and `t' for unsigned 5-bit integer
11390 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11392 (imm5_operand_dictunpack16): Likewise.
11393 (imm5_operand_dictunpack32): Likewise.
11394 (imm2_operand): Likewise.
11395 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11396 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11397 ditunpack* and fpcmp*shl builtins.
11399 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11401 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11402 * config.in: Add HAVE_AS_SPARC6 define.
11403 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11405 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11407 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11408 (CPP_CPU_SPEC): Handle m8.
11409 (ASM_CPU_SPEC): Likewise.
11410 * config/sparc/sparc-opts.h (enum processor_type): Add
11412 * config/sparc/sparc.c (m8_costs): New struct.
11413 (sparc_option_override): Handle TARGET_CPU_m8.
11414 (sparc32_initialize_trampoline): Likewise.
11415 (sparc64_initialize_trampoline): Likewise.
11416 (sparc_issue_rate): Likewise.
11417 (sparc_register_move_cost): Likewise.
11418 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11419 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11420 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11421 (CPP_CPU_SPEC): Handle M8.
11422 (ASM_CPU_SPEC): Likewise.
11423 (AS_M8_FLAG): Define.
11424 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11425 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11426 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11428 * configure: Regenerate.
11429 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11432 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11434 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11436 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11437 ("*movdi_insn_sp32"): Do not set v3pipe.
11438 ("*movsi_insn"): Likewise.
11439 ("*movdi_insn_sp64"): Likewise.
11440 ("*movsf_insn"): Likewise.
11441 ("*movdf_insn_sp32"): Likewise.
11442 ("*movdf_insn_sp64"): Likewise.
11443 ("*zero_extendsidi2_insn_sp64"): Likewise.
11444 ("*sign_extendsidi2_insn"): Likewise.
11445 ("*mov<VM32:mode>_insn"): Likewise.
11446 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11447 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11448 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11449 ("<vlop:code><VL:mode>3"): Likewise.
11450 ("*not_<vlop:code><VL:mode>3"): Likewise.
11451 ("*nand<VL:mode>_vis"): Likewise.
11452 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11453 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11454 ("one_cmpl<VL:mode>2"): Likewise.
11455 ("faligndata<VM64:mode>_vis"): Likewise.
11456 ("alignaddrsi_vis"): Likewise.
11457 ("alignaddrdi_vis"): Likweise.
11458 ("alignaddrlsi_vis"): Likewise.
11459 ("alignaddrldi_vis"): Likewise.
11460 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11461 ("bmaskdi_vis"): Likewise.
11462 ("bmasksi_vis"): Likewise.
11463 ("bshuffle<VM64:mode>_vis"): Likewise.
11464 ("cmask8<P:mode>_vis"): Likewise.
11465 ("cmask16<P:mode>_vis"): Likewise.
11466 ("cmask32<P:mode>_vis"): Likewise.
11467 ("pdistn<P:mode>_vis"): Likewise.
11468 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11470 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11472 * config/sparc/sparc.md ("subtype"): New insn attribute.
11473 ("*wrgsr_sp64"): Set insn subtype.
11474 ("*rdgsr_sp64"): Likewise.
11475 ("alignaddrsi_vis"): Likewise.
11476 ("alignaddrdi_vis"): Likewise.
11477 ("alignaddrlsi_vis"): Likewise.
11478 ("alignaddrldi_vis"): Likewise.
11479 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11480 ("fexpand_vis"): Likewise.
11481 ("fpmerge_vis"): Likewise.
11482 ("faligndata<VM64:mode>_vis"): Likewise.
11483 ("bshuffle<VM64:mode>_vis"): Likewise.
11484 ("cmask8<P:mode>_vis"): Likewise.
11485 ("cmask16<P:mode>_vis"): Likewise.
11486 ("cmask32<P:mode>_vis"): Likewise.
11487 ("fchksm16_vis"): Likewise.
11488 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11489 ("fmean16_vis"): Likewise.
11490 ("fp<plusminus_insn>64_vis"): Likewise.
11491 ("<plusminus_insn>v8qi3"): Likewise.
11492 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11493 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11494 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11495 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11496 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11497 ("*movqi_insn"): Likewise.
11498 ("*movhi_insn"): Likewise.
11499 ("*movsi_insn"): Likewise.
11500 ("movsi_pic_gotdata_op"): Likewise.
11501 ("*movdi_insn_sp32"): Likewise.
11502 ("*movdi_insn_sp64"): Likewise.
11503 ("movdi_pic_gotdata_op"): Likewise.
11504 ("*movsf_insn"): Likewise.
11505 ("*movdf_insn_sp32"): Likewise.
11506 ("*movdf_insn_sp64"): Likewise.
11507 ("*zero_extendhisi2_insn"): Likewise.
11508 ("*zero_extendqihi2_insn"): Likewise.
11509 ("*zero_extendqisi2_insn"): Likewise.
11510 ("*zero_extendqidi2_insn"): Likewise.
11511 ("*zero_extendhidi2_insn"): Likewise.
11512 ("*zero_extendsidi2_insn_sp64"): Likewise.
11513 ("ldfsr"): Likewise.
11514 ("prefetch_64"): Likewise.
11515 ("prefetch_32"): Likewise.
11516 ("tie_ld32"): Likewise.
11517 ("tie_ld64"): Likewise.
11518 ("*tldo_ldub_sp32"): Likewise.
11519 ("*tldo_ldub1_sp32"): Likewise.
11520 ("*tldo_ldub2_sp32"): Likewise.
11521 ("*tldo_ldub_sp64"): Likewise.
11522 ("*tldo_ldub1_sp64"): Likewise.
11523 ("*tldo_ldub2_sp64"): Likewise.
11524 ("*tldo_ldub3_sp64"): Likewise.
11525 ("*tldo_lduh_sp32"): Likewise.
11526 ("*tldo_lduh1_sp32"): Likewise.
11527 ("*tldo_lduh_sp64"): Likewise.
11528 ("*tldo_lduh1_sp64"): Likewise.
11529 ("*tldo_lduh2_sp64"): Likewise.
11530 ("*tldo_lduw_sp32"): Likewise.
11531 ("*tldo_lduw_sp64"): Likewise.
11532 ("*tldo_lduw1_sp64"): Likewise.
11533 ("*tldo_ldx_sp64"): Likewise.
11534 ("*mov<VM32:mode>_insn"): Likewise.
11535 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11536 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11538 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11540 * config/sparc/sparc.md ("type"): New insn type viscmp.
11541 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11543 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11544 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11545 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11546 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11548 ("n7_vis_logical_11cycle"): Likewise.
11549 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11550 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11551 * config/sparc/niagara.md ("niag_vis"): Likewise.
11552 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11553 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11555 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11557 * config/sparc/sparc.md: New instruction type `bmask'.
11558 (bmaskdi_vis): Use the `bmask' type.
11559 (bmasksi_vis): Likewise.
11560 * config/sparc/ultra3.md (us3_array): Likewise.
11561 * config/sparc/niagara7.md (n7_array): Likewise.
11562 * config/sparc/niagara4.md (n4_array): Likewise.
11563 * config/sparc/niagara2.md (niag2_vis): Likewise.
11564 (niag3_vis): Likewise.
11565 * config/sparc/niagara.md (niag_vis): Likewise.
11567 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11569 * ipa-comdats.c: Remove optimize check from gate.
11570 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11571 for functions not optimized.
11572 (ipa_fn_summary_read): Skip optimize check.
11573 (ipa_fn_summary_write): Likewise.
11574 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11576 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11578 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11579 (check_callers): Give up on caller which is not optimized.
11580 (inline_small_functions): Likewise.
11581 (ipa_inline): Do not give up when not optimizing.
11582 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11583 away unoptimizes cdtors.
11584 (whole_program_function_and_variable_visibility): Do
11585 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11586 * ipa.c (process_references): Do not check optimize.
11587 (symbol_table::remove_unreachable_nodes): Update optimize check.
11588 (set_writeonly_bit): Update optimize check.
11589 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11590 (pass_ipa_single_use::gate): Remove.
11592 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11594 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11595 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11596 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11597 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11598 permute_load, permute_store, adjust_extract, adjust_splat,
11599 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11600 replace_swap_with_copy, dump_swap_insn_table,
11601 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11602 recombine_lvx_pattern, recombine_stvx_pattern,
11603 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11604 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11605 to file rs6000-p8swap.c.
11606 * config/rs6000/rs6000-p8swap.c: New file.
11607 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11608 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11609 and rs6000*-*-* targets.
11611 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11613 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11615 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11617 * lto-wrapper.c (merge_and_complain): Do not merge
11618 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11619 fsigned_zeros, ftrapping_math, fwrapv.
11620 (append_compiler_options): Do not track these options.
11621 (append_linker_options): Likewie
11623 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11625 * cgraphunit.c (cgraph_node::finalize_function): When
11626 !flag_toplevel_reorde set no_reorder flag.
11627 (varpool_node::finalize_decl): Likewise.
11628 (symbol_table::compile): Drop no toplevel reorder path.
11630 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11632 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11633 edges; zero probability is not better than uninitialized.
11635 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11637 * asan.h (asan_sanitize_allocas_p): Declare.
11638 * asan.c (asan_sanitize_allocas_p): New function.
11639 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11640 (handle_builtin_alloca): Likewise.
11641 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11642 if !asan_sanitize_allocas_p.
11643 * params.def (asan-instrument-allocas): Add new option.
11644 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11645 * opts.c (common_handle_option): Disable allocas sanitization for
11648 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11650 * asan.c: Include gimple-fold.h.
11651 (get_last_alloca_addr): New function.
11652 (handle_builtin_stackrestore): Likewise.
11653 (handle_builtin_alloca): Likewise.
11654 (asan_emit_allocas_unpoison): Likewise.
11655 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11656 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11657 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11658 (instrument_builtin_call): Pass gimple iterator to
11659 get_mem_refs_of_builtin_call.
11660 (last_alloca_addr): New global.
11661 * asan.h (asan_emit_allocas_unpoison): Declare.
11662 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11663 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11664 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11665 if function calls alloca.
11666 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11667 * gimple-fold.h (replace_call_with_value): Declare.
11668 * internal-fn.c: Include asan.h.
11669 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11670 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11672 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11674 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11675 (C_SELFTEST_FLAGS): New.
11676 (CPP_SELFTEST_FLAGS): New.
11677 (SELFTEST_DEPS): New, from deps of s-selftest.
11678 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11679 (CPP_SELFTEST_DEPS): New.
11680 (selftest): Add dependency on s-selftest-c++.
11681 (s-selftest): Rename to...
11682 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11683 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11684 than SELFTEST_FLAGS.
11685 (selftest-gdb): Rename to...
11686 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11688 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11689 (selftest-valgrind): Rename to...
11690 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11692 (selftest-valgrind): Reintroduce as an alias for
11693 selftest-c-valgrind.
11694 (s-selftest-c++): New.
11695 (selftest-c++-gdb): New.
11696 (selftest-c++-valgrind): New.
11698 2017-07-06 Olivier Hainque <hainque@adacore.com>
11700 * gcc.c (process_command): When deciding if undefined variables
11701 should be ignored when processing specs, accept "gcc -v" as well.
11703 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11705 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11706 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11708 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11710 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11711 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11712 * config/arm/arm-tables.opt: Regenerate.
11713 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11715 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11717 2017-07-06 Carl Love <cel@us.ibm.com>
11719 * ChangeLog: Clean up from mid air collision
11721 2017-07-06 Carl Love <cel@us.ibm.com>
11723 * config/rs6000/rs6000-c.c: Add support for built-in functions
11724 vector signed int vec_subc (vector signed int, vector signed int);
11725 vector signed __int128 vec_subc (vector signed __int128,
11726 vector signed __int128);
11727 vector unsigned __int128 vec_subc (vector unsigned __int128,
11728 vector unsigned __int128);
11729 vector signed int vec_sube (vector signed int, vector signed int,
11730 vector signed int);
11731 vector unsigned int vec_sube (vector unsigned int,
11732 vector unsigned int,
11733 vector unsigned int);
11734 vector signed __int128 vec_sube (vector signed __int128,
11735 vector signed __int128,
11736 vector signed__int128);
11737 vector unsigned __int128 vec_sube (vector unsigned __int128,
11738 vector unsigned __int128,
11739 vector unsigned __int128);
11740 vector signed int vec_subec (vector signed int, vector signed int,
11741 vector signed int);
11742 vector unsigned int vec_subec (vector unsigned int,
11743 vector unsigned int,
11744 vector unsigned int);
11745 vector signed __int128 vec_subec (vector signed __int128,
11746 vector signed __int128,
11747 vector signed__int128);
11748 vector unsigned __int128 vec_subec (vector unsigned __int128,
11749 vector unsigned __int128,
11750 vector unsigned __int128);
11751 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11752 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11753 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11754 BU_ALTIVEC_OVERLOAD_X definitions.
11755 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11756 * doc/extend.texi: Update the built-in documentation file for the new
11757 built-in functions.
11759 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11762 * diagnostic-show-locus.c (layout::layout): Use start and finish
11763 spelling location for the start and finish of each range.
11764 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11765 Add unused aspect param.
11766 * input.c (expand_location_1): Add "aspect" param, and use it
11767 to access the correct part of the location.
11768 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11770 (expand_location_to_spelling_point): Likewise.
11771 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11772 param, and pass it to expand_location_1.
11774 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11776 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11777 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11778 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11779 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11780 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11781 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11782 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11783 _mm_maskz_getmant_ss): New intrinsics.
11784 (__builtin_ia32_getexpss128_mask): Changed to ...
11785 __builtin_ia32_getexpss128_round ... this.
11786 (__builtin_ia32_getexpsd128_mask): Changed to ...
11787 __builtin_ia32_getexpsd128_round ... this.
11788 * config/i386/i386-builtin-types.def
11789 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11790 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11791 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11792 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11793 __builtin_ia32_getmantss_mask_round): New builtins.
11794 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11795 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11796 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11797 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11798 * config/i386/sse.md
11799 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11800 avx512f_sgetexp<mode><mask_scalar_name>
11801 <round_saeonly_scalar_name> ... this.
11802 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11803 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11804 vgetexp<ssescalarmodesuffix>
11805 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11806 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11807 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11808 avx512f_vgetmant<mode><mask_scalar_name>
11809 <round_saeonly_scalar_name> ... this.
11810 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11811 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11812 vgetmant<ssescalarmodesuffix>
11813 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11814 %0<mask_scalar_operand4>, %1, %2
11815 <round_saeonly_scalar_mask_op4>, %3} ... this.
11816 * config/i386/subst.md (mask_scalar_operand4,
11817 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11818 round_saeonly_scalar_nimm_predicate): New subst attributes.
11820 2017-07-06 Julia Koval <julia.koval@intel.com>
11822 * config/i386/i386.c (ix86_erase_embedded_rounding):
11823 Remove code for old rounding pattern.
11825 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11827 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11829 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11831 * doc/sourcebuild.texi (Test Directives, Variants of
11832 dg-require-support): Add documentation for dg-require-stack-check.
11834 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11836 * config/i386/subst.md (mask_scalar, round_scalar,
11837 round_saeonly_scalar): New meta-templates.
11838 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11839 round_scalar_mask_operand3, round_scalar_mask_op3,
11840 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11841 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11842 round_saeonly_scalar_constraint,
11843 round_saeonly_scalar_prefix): New subst attribute.
11844 * config/i386/sse.md
11845 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11846 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11847 <round_scalar_name> ... this.
11848 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11849 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11850 <round_scalar_name> ... this.
11851 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11852 <sse>_vm<code><mode>3<mask_scalar_name>
11853 <round_saeonly_scalar_name> ... this.
11854 (v<plusminus_mnemonic><ssescalarmodesuffix>
11855 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11856 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11857 v<plusminus_mnemonic><ssescalarmodesuffix>
11858 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11859 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11860 (v<multdiv_mnemonic><ssescalarmodesuffix>
11861 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11862 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11863 v<multdiv_mnemonic><ssescalarmodesuffix>
11864 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11865 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11866 (v<maxmin_float><ssescalarmodesuffix>
11867 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11868 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11869 v<maxmin_float><ssescalarmodesuffix>
11870 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11871 %0<mask_scalar_operand3>, %1, %<iptr>2
11872 <round_saeonly_scalar_mask_op3>} ... this.
11874 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11876 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11877 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11879 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11880 Alan Hayward <alan.hayward@arm.com>
11881 David Sherwood <david.sherwood@arm.com>
11883 * combine.c (simplify_if_then_else): Remove "enum" before
11885 * compare-elim.c (can_eliminate_compare): Likewise.
11886 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11888 (aarch64_lookup_simd_builtin_type): Likewise.
11889 (aarch64_simd_builtin_type): Likewise.
11890 (aarch64_init_simd_builtin_types): Likewise.
11891 (aarch64_simd_expand_args): Likewise.
11892 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11894 (aarch64_reverse_mask): Likewise.
11895 (aarch64_simd_emit_reg_reg_move): Likewise.
11896 (aarch64_gen_adjusted_ldpstp): Likewise.
11897 (aarch64_ccmp_mode_to_code): Likewise.
11898 (aarch64_operands_ok_for_ldpstp): Likewise.
11899 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11900 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11902 (aarch64_min_divisions_for_recip_mul): Likewise.
11903 (aarch64_reassociation_width): Likewise.
11904 (aarch64_get_condition_code_1): Likewise.
11905 (aarch64_simd_emit_reg_reg_move): Likewise.
11906 (aarch64_simd_attr_length_rglist): Likewise.
11907 (aarch64_reverse_mask): Likewise.
11908 (aarch64_operands_ok_for_ldpstp): Likewise.
11909 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11910 (aarch64_gen_adjusted_ldpstp): Likewise.
11911 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11913 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11914 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11915 (arm_lookup_simd_builtin_type): Likewise.
11916 (arm_simd_builtin_type): Likewise.
11917 (arm_init_simd_builtin_types): Likewise.
11918 (arm_expand_builtin_args): Likewise.
11919 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11920 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11921 (ft32_setup_incoming_varargs): Likewise.
11922 (ft32_function_arg): Likewise.
11923 (ft32_function_arg_advance): Likewise.
11924 (ft32_pass_by_reference): Likewise.
11925 (ft32_arg_partial_bytes): Likewise.
11926 (ft32_valid_pointer_mode): Likewise.
11927 (ft32_addr_space_pointer_mode): Likewise.
11928 (ft32_addr_space_legitimate_address_p): Likewise.
11929 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11931 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11932 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11933 (iamcu_alignment): Likewise.
11934 (canonicalize_vector_int_perm): Likewise.
11935 (ix86_noce_conversion_profitable_p): Likewise.
11936 (ix86_mpx_bound_mode): Likewise.
11937 (ix86_operands_ok_for_move_multiple): Likewise.
11938 * config/microblaze/microblaze-protos.h
11939 (microblaze_expand_conditional_branch_reg): Likewise.
11940 * config/microblaze/microblaze.c
11941 (microblaze_expand_conditional_branch_reg): Likewise.
11942 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11944 (rs6000_reassociation_width): Likewise.
11945 (rs6000_invalid_binary_op): Likewise.
11946 (fusion_p9_p): Likewise.
11947 (emit_fusion_p9_load): Likewise.
11948 (emit_fusion_p9_store): Likewise.
11949 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11951 (riscv_hard_regno_mode_ok_p): Likewise.
11952 (riscv_address_insns): Likewise.
11953 (riscv_split_symbol): Likewise.
11954 (riscv_legitimize_move): Likewise.
11955 (riscv_function_value): Likewise.
11956 (riscv_hard_regno_nregs): Likewise.
11957 (riscv_expand_builtin): Likewise.
11958 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11959 (riscv_build_integer): Likewise.
11960 (riscv_split_integer): Likewise.
11961 (riscv_legitimate_constant_p): Likewise.
11962 (riscv_cannot_force_const_mem): Likewise.
11963 (riscv_regno_mode_ok_for_base_p): Likewise.
11964 (riscv_valid_base_register_p): Likewise.
11965 (riscv_valid_offset_p): Likewise.
11966 (riscv_valid_lo_sum_p): Likewise.
11967 (riscv_classify_address): Likewise.
11968 (riscv_legitimate_address_p): Likewise.
11969 (riscv_address_insns): Likewise.
11970 (riscv_load_store_insns): Likewise.
11971 (riscv_force_binary): Likewise.
11972 (riscv_split_symbol): Likewise.
11973 (riscv_force_address): Likewise.
11974 (riscv_legitimize_address): Likewise.
11975 (riscv_move_integer): Likewise.
11976 (riscv_legitimize_const_move): Likewise.
11977 (riscv_legitimize_move): Likewise.
11978 (riscv_address_cost): Likewise.
11979 (riscv_subword): Likewise.
11980 (riscv_output_move): Likewise.
11981 (riscv_canonicalize_int_order_test): Likewise.
11982 (riscv_emit_int_order_test): Likewise.
11983 (riscv_function_arg_boundary): Likewise.
11984 (riscv_pass_mode_in_fpr_p): Likewise.
11985 (riscv_pass_fpr_single): Likewise.
11986 (riscv_pass_fpr_pair): Likewise.
11987 (riscv_get_arg_info): Likewise.
11988 (riscv_function_arg): Likewise.
11989 (riscv_function_arg_advance): Likewise.
11990 (riscv_arg_partial_bytes): Likewise.
11991 (riscv_function_value): Likewise.
11992 (riscv_pass_by_reference): Likewise.
11993 (riscv_setup_incoming_varargs): Likewise.
11994 (riscv_print_operand): Likewise.
11995 (riscv_elf_select_rtx_section): Likewise.
11996 (riscv_save_restore_reg): Likewise.
11997 (riscv_for_each_saved_reg): Likewise.
11998 (riscv_register_move_cost): Likewise.
11999 (riscv_hard_regno_mode_ok_p): Likewise.
12000 (riscv_hard_regno_nregs): Likewise.
12001 (riscv_class_max_nregs): Likewise.
12002 (riscv_memory_move_cost): Likewise.
12003 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12004 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12005 (rl78_addr_space_address_mode): Likewise.
12006 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12008 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12009 (rs6000_reassociation_width): Likewise.
12010 (rs6000_invalid_binary_op): Likewise.
12011 (fusion_p9_p): Likewise.
12012 (emit_fusion_p9_load): Likewise.
12013 (emit_fusion_p9_store): Likewise.
12014 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12015 (ok_for_simple_move_operands): Likewise.
12016 (ok_for_simple_move_strict_operands): Likewise.
12017 (ok_for_simple_arith_logic_operands): Likewise.
12018 (visium_legitimize_reload_address): Likewise.
12019 (visium_select_cc_mode): Likewise.
12020 (output_cbranch): Likewise.
12021 (visium_split_double_move): Likewise.
12022 (visium_expand_copysign): Likewise.
12023 (visium_expand_int_cstore): Likewise.
12024 (visium_expand_fp_cstore): Likewise.
12025 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12026 (visium_function_arg): Likewise.
12027 (visium_function_arg_advance): Likewise.
12028 (visium_libcall_value): Likewise.
12029 (visium_setup_incoming_varargs): Likewise.
12030 (visium_legitimate_constant_p): Likewise.
12031 (visium_legitimate_address_p): Likewise.
12032 (visium_legitimize_address): Likewise.
12033 (visium_secondary_reload): Likewise.
12034 (visium_register_move_cost): Likewise.
12035 (visium_memory_move_cost): Likewise.
12036 (prepare_move_operands): Likewise.
12037 (ok_for_simple_move_operands): Likewise.
12038 (ok_for_simple_move_strict_operands): Likewise.
12039 (ok_for_simple_arith_logic_operands): Likewise.
12040 (visium_function_value_1): Likewise.
12041 (rtx_ok_for_offset_p): Likewise.
12042 (visium_legitimize_reload_address): Likewise.
12043 (visium_split_double_move): Likewise.
12044 (visium_expand_copysign): Likewise.
12045 (visium_expand_int_cstore): Likewise.
12046 (visium_expand_fp_cstore): Likewise.
12047 (visium_split_cstore): Likewise.
12048 (visium_select_cc_mode): Likewise.
12049 (visium_split_cbranch): Likewise.
12050 (output_cbranch): Likewise.
12051 (visium_print_operand_address): Likewise.
12052 * expmed.c (flip_storage_order): Likewise.
12053 * expmed.h (emit_cstore): Likewise.
12054 (flip_storage_order): Likewise.
12055 * genrecog.c (validate_pattern): Likewise.
12056 * hsa-gen.c (gen_hsa_addr): Likewise.
12057 * internal-fn.c (expand_arith_overflow): Likewise.
12058 * ira-color.c (allocno_copy_cost_saving): Likewise.
12059 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12060 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12061 (process_invariant_for_inheritance): Likewise.
12062 * lra-eliminations.c (move_plus_up): Likewise.
12063 * omp-low.c (lower_oacc_reductions): Likewise.
12064 * simplify-rtx.c (simplify_subreg): Likewise.
12065 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12066 (TARGET_CHKP_BOUND_MODE): Likewise..
12067 * targhooks.c (default_chkp_bound_mode): Likewise.
12068 (default_setup_incoming_vararg_bounds): Likewise.
12069 * targhooks.h (default_chkp_bound_mode): Likewise.
12070 (default_setup_incoming_vararg_bounds): Likewise.
12071 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12072 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12073 (have_whole_vector_shift): Likewise.
12074 * tree-vect-stmts.c (vectorizable_load): Likewise.
12075 * doc/tm.texi: Regenerate.
12077 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12079 Graceful degrade if Binutils PR21472 is not available.
12082 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12083 .rodata in flash test fails.
12084 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12085 * confgure: Regenerate.
12086 * config.in: Regenerate.
12087 * config/avr/avr.c (avr_asm_named_section)
12088 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12089 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12090 (avr_asm_init_sections): Same.
12092 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12094 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12095 (fma<VH:mode>4_intrinsic): Likewise.
12096 (*fmsub<VCVTF:mode>4): Likewise.
12097 (*fmsub<VH:mode>4_intrinsic): Likewise.
12099 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12102 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12103 Don't depend on "optimize > 0".
12104 (out_movhi_r_mr, out_movqi_mr_r): Same.
12105 (out_movhi_mr_r, out_movqi_r_mr): Same.
12106 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12107 io_address_operand on "optimize > 0".
12109 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12111 * tree-loop-distribution.c: Add general explanantion on the pass.
12112 (generate_loops_for_partition): Mark distributed loop.
12113 (pg_add_dependence_edges): New parameter. Handle alias data
12114 dependence specially and record it in the parameter if asked.
12115 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12116 (init_partition_graph_vertices, add_partition_graph_edge): New.
12117 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12118 (free_partition_graph_vdata, build_partition_graph): New.
12119 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12120 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12121 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12122 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12123 (version_for_distribution_p, finalize_partitions): New.
12124 (distribute_loop): Handle alias data dependence specially. Factor
12125 out loop fusion code as functions and call these functions.
12127 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12129 * tree-loop-distribution.c (classify_partition): New parameter and
12130 better handle reduction statement.
12131 (rdg_build_partitions): Revise comment.
12132 (distribute_loop): Compute statements in all partitions and pass it
12133 to classify_partition.
12135 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12137 * tree-loop-distribution.c (enum partition_type): New.
12138 (struct partition): New field type.
12139 (partition_merge_into): Add parameter. Update partition type.
12140 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12141 (build_rdg_partition_for_vertex): Compute partition type.
12142 (rdg_build_partitions): Dump partition type.
12143 (distribute_loop): Update calls to partition_merge_into.
12145 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12147 * tree-loop-distribution.c (struct ddr_hasher): New.
12148 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12150 (classify_partition): Call get_data_dependence.
12151 (pg_add_dependence_edges): Ditto.
12152 (distribute_loop): Release data dependence hash table.
12154 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12156 * tree-loop-distribution.c (ref_base_address): Delete.
12157 (similar_memory_accesses): Rename ...
12158 (share_memory_accesses): ... to this. Check if partitions access
12159 the same memory reference.
12160 (distribute_loop): Call share_memory_accesses.
12162 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12164 * tree-loop-distribution.c (struct partition): New field recording
12165 its data reference.
12166 (partition_alloc, partition_free): Init and release data refs.
12167 (partition_merge_into): Merge data refs.
12168 (build_rdg_partition_for_vertex): Collect data refs for partition.
12169 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12171 (distribute_loop): Remve data refs from vertice data of partition
12174 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12176 * tree-loop-distribution.c (params.h): Include header file.
12177 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12178 (datarefs_vec): New global var.
12179 (create_rdg_vertices): Use datarefs_vec directly.
12180 (free_rdg): Don't free data references.
12181 (build_rdg): Update use. Don't free data references.
12182 (distribute_loop): Compute global variable for data references.
12183 Bail out if there are too many data references.
12185 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12187 * tree-loop-distribution.c (loop_nest): New global var.
12188 (build_rdg): Use loop directly, rather than loop nest.
12189 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12191 (distribute_loop): Compute global variable loop nest. Update use.
12193 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12195 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12196 (partition_merge_into): New parameter. Dump reason for fusion.
12197 (distribute_loop): Update use of partition_merge_into.
12199 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12201 * tree-loop-distribution.c (bb_top_order_index): New.
12202 (bb_top_order_index_size, bb_top_order_cmp): New.
12203 (stmts_from_loop): Use topological order.
12204 (pass_loop_distribution::execute): Compute and release topological
12205 order for basic blocks.
12207 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12209 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12212 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12214 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12215 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12216 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12217 * internal-fn.def (LOOP_DIST_ALIAS): New.
12218 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12219 (fold_loop_internal_call): ... this.
12220 (vect_loop_dist_alias_call): New function.
12221 (set_uid_loop_bbs): Call fold_loop_internal_call.
12222 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12225 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12228 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12229 Require dead FLAGS_REG at the beginning of a peephole.
12231 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12234 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12235 arguments in the call to __builtin_ia32_sbb_u32.
12236 (_subborrow_u64): Swap _X and _Y arguments in the call to
12237 __builtin_ia32_sbb_u64.
12239 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12242 * tree-vrp.c (compare_assert_loc): Turn into a function template
12243 with stable template parameter. Only test if a->e is NULL,
12244 !a->e == !b->e has been verified already. Use e == NULL or
12245 e != NULL instead of e or ! e tests. If stable is true, don't use
12246 iterative_hash_expr, on the other side allow a or b or both NULL
12247 and sort the NULLs last.
12248 (process_assert_insertions): Sort using compare_assert_loc<false>
12249 instead of compare_assert_loc, later sort using
12250 compare_assert_loc<true> before calling process_assert_insertions_for
12251 in a loop. Use break instead of continue once seen NULL pointer.
12253 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12255 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12256 Cortex-R7 and Cortex-R8 processors.
12258 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12260 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12261 uninitialized while src is not.
12263 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12265 * common/config/arm/arm-common.c: Adjust include path for
12267 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12268 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12269 (arm-cpu-data.h): Likewise.
12270 (arm-cpu-cdata.h): Likewise.
12271 * config/arm/arm-cpu.h: Delete.
12272 * config/arm/arm-cpu-cdata.h: Delete.
12273 * config/arm/arm-cpu-data.h: Delete.
12275 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12277 * config/arm/arm-cpus.in (cortex-a55): New.
12278 (cortex-a75): Likewise.
12279 (cortex-a75.cortex-a55): Likewise.
12280 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12282 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12283 * config/arm/arm-cpu-cdata.h: Regenerate.
12284 * config/arm/arm-cpu-data.h: Regenerate.
12285 * config/arm/arm-cpu.h: Regenerate.
12286 * config/arm/arm-tables.opt: Regenerate.
12287 * config/arm/arm-tune.md: Regenerate.
12289 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12291 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12293 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12295 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12298 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12300 PR tree-optimization/81292
12301 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12302 full_string_p, also call adjust_related_strinfos if the adjustment
12303 is simple, otherwise invalidate related strinfos.
12305 2017-07-04 Martin Liska <mliska@suse.cz>
12308 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12309 newly created variable as DECL_IGNORED_P.
12311 2017-07-04 Martin Liska <mliska@suse.cz>
12314 * ipa-inline.c (inline_small_functions):
12315 Use xstrdup_for_dump.
12317 2017-07-04 Tom de Vries <tom@codesourcery.com>
12319 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12321 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12324 * config/darwin.c (darwin_function_switched_text_sections):
12327 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12329 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12331 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12333 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12335 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12337 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12338 min_profitable_iters, and th as inclusive lower bounds.
12339 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12340 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12341 for min_profitable_iters and min_profitable_estimate.
12342 (vect_transform_loop): Treat th as an inclusive lower bound.
12343 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12345 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12348 * config/darwin.c (darwin_function_switched_text_sections):
12349 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12350 in two pieces, and suppress the use of buf.
12352 2017-07-03 Nathan Sidwell <nathan@acm.org>
12354 * hash-table.h (hash_table_mod1): Fix indentation.
12356 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12358 PR middle-end/81290
12359 * predict.c (force_edge_cold): Be more careful about propagation
12361 * profile-count.h (profile_probability::guessed,
12362 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12364 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12366 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12368 * doc/invoke.texi (rcpc architecture extension): Document it.
12370 2017-07-03 Richard Biener <rguenther@suse.de>
12372 PR tree-optimization/60510
12373 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12374 the scalar reduction PHI and use it.
12375 (vectorizable_reduction): Properly guard the single_defuse_cycle
12376 path for non-SLP reduction chains where we cannot use it.
12377 Rework reduc_def/index and vector type deduction. Rework
12378 vector operand gathering during reduction op code-gen.
12379 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12380 chains dissolve the chain and leave it to non-SLP reduction
12383 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12385 * tree-data-ref.h (dr_alignment): Declare.
12386 * tree-data-ref.c (dr_alignment): New function.
12387 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12388 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12390 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12392 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12394 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12395 and base_misalignment fields.
12396 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12397 * tree-data-ref.c: Include builtins.h.
12398 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12399 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12400 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12401 * tree-vect-data-refs.c: Include tree-cfg.h.
12402 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12403 fields instead of calculating an alignment here.
12404 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12405 innermost_loop_behavior fields.
12407 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12409 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12411 (DR_STEP_ALIGNMENT): New macro.
12412 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12413 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12414 (create_data_ref): Print it.
12415 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12416 to tell whether the step preserves vector (mis)alignment.
12417 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12418 Move the check for an integer step and generalise to all INTEGER_CST.
12419 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12420 Print the outer step alignment.
12422 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12424 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12425 with offset_alignment.
12426 (DR_ALIGNED_TO): Delete.
12427 (DR_OFFSET_ALIGNMENT): New macro.
12428 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12429 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12430 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12431 (create_data_ref): Likewise.
12432 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12433 (vect_analyze_data_refs): Likewise.
12434 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12435 creating dummy innermost behavior.
12437 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12439 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12440 with a "innermost_loop_behavior *" and refeence tree.
12441 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12442 (create_data_ref): Update call accordingly.
12443 * tree-predcom.c (find_looparound_phi): Likewise.
12445 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12447 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12448 fields with dr_wrt_vec_loop.
12449 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12450 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12451 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12452 (vect_dr_behavior): New function.
12453 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12454 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12455 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12456 track whether the step preserves the misalignment.
12457 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12458 Use vect_dr_behavior.
12459 (vect_setup_realignment): Update call accordingly.
12460 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12461 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12462 call to vect_create_addr_base_for_vector_ref.
12463 (vect_create_cond_for_align_checks): Likewise.
12464 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12465 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12466 (vect_recog_mask_conversion_pattern): Likewise.
12467 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12468 (new_stmt_vec_info): Remove redundant zeroing.
12470 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12472 * common/config/arm/arm-common.c (arm_be8_option): New function.
12473 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12474 (ISA_ARMv6): Add isa_bit_be8.
12475 * config/arm/arm.h (arm_be8_option): Add prototype.
12476 (BE8_SPEC_FUNCTION): New define.
12477 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12478 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12479 (mlittle-endian): Similarly.
12480 (mbe8, mbe32): New options.
12481 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12482 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12484 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12486 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12488 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12490 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12491 (cleanup_tree_cfg_bb): Use it.
12492 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12494 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12496 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12499 * loop-doloop.c (add_test): Update profile.
12501 2017-07-03 Martin Liska <mliska@suse.cz>
12504 * sanopt.c (rewrite_usage_of_param): New function.
12505 (sanitize_rewrite_addressable_params): Likewise.
12506 (pass_sanopt::execute): Call rewrite_usage_of_param.
12508 2017-07-03 Richard Biener <rguenther@suse.de>
12510 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12511 back to using VIEW_CONVERT_EXPR.
12513 2017-07-03 Martin Liska <mliska@suse.cz>
12516 * doc/extend.texi: Document when a resolver function is
12517 generated for target_clones.
12519 2017-07-03 Martin Liska <mliska@suse.cz>
12521 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12522 and shadow memory of auto variables which are subject of
12523 use-after-scope sanitization.
12524 (asan_expand_mark_ifn): Add do set only when is_poison.
12526 2016-07-03 Richard Biener <rguenther@suse.de>
12528 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12530 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12531 (vectorizable_reduction): Perform reduction PHI creation when
12532 visiting a reduction PHI and adjust and simplify code generation
12533 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12534 (vect_transform_loop): Visit reduction PHIs.
12535 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12536 defs into the SLP tree.
12537 (vect_build_slp_tree): Reduction defs terminate the recursion.
12538 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12540 (vect_get_vec_defs_for_stmt_copy): Export.
12541 (vect_get_vec_defs): Likewise.
12542 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12544 (vect_get_vec_defs_for_stmt_copy): Declare.
12545 (vect_get_vec_defs): Likewise.
12547 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12549 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12550 parameter with a "loop" parameter and use it instead of the
12551 loop containing DR_STMT. Don't check simple_iv when doing
12552 BB analysis. Describe the two analysis modes in the comment.
12554 2017-07-03 Tom de Vries <tom@codesourcery.com>
12556 PR tree-optimization/69468
12557 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12558 (find_same_succ_bb): Handle ignore_edge_flags.
12560 2017-07-03 Tom de Vries <tom@codesourcery.com>
12562 PR tree-optimization/81192
12563 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12565 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12567 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12569 2017-07-03 Tom de Vries <tom@codesourcery.com>
12571 PR tree-optimization/81192
12572 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12573 BB_SAME_SUCC (bb) == NULL.
12575 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12577 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12580 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12582 * dumpfile.c: Include profile-count.h
12583 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12585 (insert_cond_bb): Update profile.
12586 * tree-cfg.h (insert_cond_bb): Update prototype.
12587 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12588 * tree-dump.c: Do not include tree-cfg.
12590 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12592 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12594 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12596 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12599 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12601 * tree-complex.c (expand_complex_div_wide): update profile.
12603 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12604 Alan Hayward <alan.hayward@arm.com>
12605 David Sherwood <david.sherwood@arm.com>
12607 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12608 (CORETYPES_H): New define.
12609 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12610 (insn-modes-inline.h, s-modes-inline-h): New rules.
12611 (generated_files): Add insn-modes-inline.h.
12612 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12613 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12614 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12615 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12616 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12617 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12618 (build/gencodes.o, build/genconditions.o): Likewise.
12619 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12620 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12621 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12622 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12623 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12624 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12625 * coretypes.h: Include everything up to real.h for generators.
12626 Include insn-modes.h first. Include wide-int-print.h after
12627 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12628 * machmode.h: Don't include insn-modes.h here.
12629 * function-tests.c: Remove includes of signop.h, machmode.h,
12630 double-int.h and wide-int.h.
12632 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12634 * optc-save-gen.awk: Likewise.
12635 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12636 * godump.c: Remove include of wide-int-print.h.
12637 * pretty-print.h: Likewise.
12638 * wide-int-print.cc: Likewise.
12639 * wide-int.cc: Likewise.
12640 * hash-map-tests.c: Remove include of signop.h.
12641 * hash-set-tests.c: Likewise.
12642 * rtl-tests.c: Likewise.
12643 * mkconfig.sh: Remove include of machmode.h.
12644 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12646 (emit_insn_modes_inline_h): ...this new function. Emit the code
12647 into an insn-modes-inline.h header file, adding appropriate
12648 include guards and end comments.
12649 (emit_insn_modes_c_header): Remove include of machmode.h.
12650 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12652 (main): Handle -i flag and call emit_insn_modes_inline_h when
12655 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12657 * tree-ssa-strlen.c (strinfo): Rename the length field to
12658 nonzero_chars. Add a full_string_p field.
12659 (compare_nonzero_chars, zero_length_string_p): New functions.
12660 (get_addr_stridx): Add an offset_out parameter.
12661 Use compare_nonzero_chars.
12662 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12663 (new_strinfo): Update after above changes to strinfo.
12664 (set_endptr_and_length): Set full_string_p.
12665 (get_string_length): Update after above changes to strinfo.
12666 (unshare_strinfo): Update call to new_strinfo.
12667 (maybe_invalidate): Likewise.
12668 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12669 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12670 as a uhwi instead of an shwi. Update after above changes to
12671 strinfo and new_strinfo.
12672 (zero_length_string): Assert that chainsi contains full strings.
12673 Use zero_length_string_p. Update call to new_strinfo.
12674 (adjust_related_strinfos): Update after above changes to strinfo.
12675 Copy full_string_p from origsi.
12676 (adjust_last_stmt): Use zero_length_string_p.
12677 (handle_builtin_strlen): Update after above changes to strinfo and
12678 new_strinfo. Install the lhs as the string length if the previous
12679 entry didn't describe a full string.
12680 (handle_builtin_strchr): Update after above changes to strinfo
12682 (handle_builtin_strcpy): Likewise.
12683 (handle_builtin_strcat): Likewise.
12684 (handle_builtin_malloc): Likewise.
12685 (handle_pointer_plus): Likewise.
12686 (handle_builtin_memcpy): Likewise. Track nonzero characters
12687 that aren't necessarily followed by a nul terminator.
12688 (handle_char_store): Likewise.
12690 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12692 PR tree-optimization/80769
12693 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12694 for malloc and calloc. Document the new invariant that all related
12695 strinfos have delayed lengths or none do.
12696 (verify_related_strinfos): Move earlier in file.
12697 (set_endptr_and_length): New function, split out from...
12698 (get_string_length): ...here. Also set the lengths of related
12700 (zero_length_string): Assert that chainsi has known (rather than
12702 (adjust_related_strinfos): Likewise.
12704 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12706 PR tree-optimization/81136
12707 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12708 assert that two references with the same misalignment have the same
12709 compile-time misalignment if those compile-time misalignments
12712 2017-07-01 Andi Kleen <ak@linux.intel.com>
12714 * print-tree.c (print_node): Print all attributes.
12716 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12718 * cfg.c (scale_bbs_frequencies): New function.
12719 * cfg.h (scale_bbs_frequencies): Declare it.
12720 * cfgloopanal.c (single_likely_exit): Cleanup.
12721 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12723 (scale_loop_profile): Likewise.
12724 (loop_version): Likewise.
12725 (create_empty_loop_on_edge): Update.
12726 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12727 scale_loop_frequencies, scale_loop_profile, loopify,
12728 loop_version): Update prototypes.
12729 * modulo-sched.c (sms_schedule): Update.
12730 * predict.c (unlikely_executed_edge_p): Also check probability.
12731 (probably_never_executed_edge_p): Fix typo.
12732 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12733 * tree-parloops.c (gen_parallel_loop): Update.
12734 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12735 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12736 * tree-ssa-loop-split.c (split_loop): Update.
12737 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12738 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12739 (vect_loop_versioning): Update.
12740 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12742 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12744 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12746 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12748 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12749 to keep profile consistent.
12751 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12753 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12754 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12755 * profile-count.h (max_safe_multiplier): Make unsigned.
12756 (profile_count::guessed_zero): New.
12758 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12760 * bb-reorder.c (fix_up_crossing_landing_pad,
12761 fix_crossing_conditional_branches): Use make_single_succ_edge
12762 to keep profile consistent.
12764 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12766 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12769 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12772 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12773 the right scopes, make sure cond_jump isn't preserved between multiple
12774 iterations. Search for fallthru edge whenever there are 3+ edges and
12775 use find_fallthru_edge for it.
12777 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12779 Patch by Alexander Monakov <amonakov@ispras.ru>
12780 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12781 probabilities consistently.
12783 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12785 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12786 emit_cmp_and_jump_insns.
12788 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12791 * tree-inline.c (expand_call_inline): Combine profile statuses.
12793 2017-06-30 Andrew Pinski <apinski@cavium.com>
12795 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12796 fold_stmt returned true.
12798 2017-06-30 Nathan Sidwell <nathan@acm.org>
12800 * ggc.h (empty_string): Delete.
12801 * cfgexpand.c (expand_asm_stmt): Use plain "".
12802 * optabs.c (expand_asm_memory_barrier): Likewise.
12803 * stringpool.c (empty_string): Delete.
12804 (digit_vector, digit_string): Delete.
12805 (ggc_alloc_string): Use plain "", don't optimize single digit
12806 strings. Use ggc_alloc_atomic.
12808 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12810 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12811 comparison set and one other set, use the cost of the non-comparison
12814 2017-06-30 Nathan Sidwell <nathan@acm.org>
12816 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12819 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12821 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12822 loops. Remove now unneeded calls to gimple_switch_set_label() that
12823 just set removed labels to NULL_TREE.
12825 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12827 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12828 (set_range_info): ...here. Only call set_range_info_raw if domain
12830 (set_nonzero_bits): Call set_range_info_raw.
12831 * tree-ssanames.h (set_range_info_raw): New.
12833 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12836 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12837 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12838 of nonimmediate_operand and <store_mask_constraint> instead of m
12839 for the input operand. For V8FI iterator, always split if input
12840 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12841 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12842 <store_mask_predicate> instead of register_operand and
12843 <store_mask_constraint> instead of v for the input operand. Make
12844 sure both operands aren't MEMs for if not <mask_applied>.
12846 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12848 * lto-wrapper.c (copy_file) Close both file descriptors before
12851 2017-06-30 Martin Liska <mliska@suse.cz>
12854 * multiple_target.c (create_dispatcher_calls): Make ifunc
12855 also for function that don't have calls or are not referenced.
12857 2017-06-30 Richard Biener <rguenther@suse.de>
12859 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12860 analyze the first scalar stmt. Move vector type computation
12861 for the BB case here from ...
12862 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12863 live operation processing in the SLP case properly.
12865 2017-06-30 Richard Biener <rguenther@suse.de>
12867 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12869 2017-06-30 Martin Liska <mliska@suse.cz>
12872 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12873 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12875 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12877 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12880 2017-06-30 Richard Biener <rguenther@suse.de>
12882 PR tree-optimization/81249
12883 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12884 condition reduction result to original scalar type.
12886 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12888 * profile-count.h (enum profile_quality): Fix typos and whitespace
12891 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12893 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12894 type for branch probabilities.
12896 2017-06-29 Julian Brown <julian@codesourcery.com>
12897 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12899 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12900 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12901 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12902 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12904 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12906 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12907 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12908 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12909 CC usage from generic code to here.
12910 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12911 CC usage into the target macros.
12913 2017-06-29 Maya Rashish <coypu@sdf.org>
12915 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12918 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12920 * arm/arm-builtins.c: Include profile-count.h
12921 * except.c (sjlj_emit_function_enter): Use
12922 profile_probability::unlikely.
12924 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12926 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12927 and tocrel_offset be pointer args rather than implicitly using
12929 (legitimate_constant_pool_address_p, rs6000_emit_move,
12930 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12931 tocrel_offset and use in toc_relative_expr_p call.
12932 (print_operand, print_operand_address): Use static tocrel_base_oac
12933 and tocrel_offset_oac.
12934 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12937 2017-06-29 Maya Rashish <coypu@sdf.org>
12939 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12941 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12943 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12944 objects, take into account only the alignment of 'op0' and 'mode1' if
12947 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12949 * ccmp.c (ccmp_tree_comparison_p): New function.
12950 (ccmp_candidate_p): Update to use above function.
12951 (get_compare_parts): New function.
12952 (expand_ccmp_next): Update to use new functions.
12953 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12955 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12956 take mode as argument.
12957 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12958 * expr.c (expand_expr_real_1): Pass mode as argument.
12960 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12962 * combine.c (combine_instructions): Print insns to dump_file, together
12965 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12967 * asan.c (asan_emit_stack_protection): Update.
12968 (create_cond_insert_point): Update.
12969 * auto-profile.c (afdo_propagate_circuit): Update.
12970 * basic-block.h (struct edge_def): Turn probability to
12971 profile_probability.
12972 (EDGE_FREQUENCY): Update.
12973 * bb-reorder.c (find_traces_1_round): Update.
12974 (better_edge_p): Update.
12975 (sanitize_hot_paths): Update.
12976 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12977 (make_single_succ_edge): Update.
12978 (check_bb_profile): Update.
12979 (dump_edge_info): Update.
12980 (update_bb_profile_for_threading): Update.
12981 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12982 probabilitycount to 0.
12983 * cfgbuild.c (compute_outgoing_frequencies): Update.
12984 * cfgcleanup.c (try_forward_edges): Update.
12985 (outgoing_edges_match): Update.
12986 (try_crossjump_to_edge): Update.
12987 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12988 (expand_gimple_tailcall): Update.
12989 (construct_init_block): Use make_single_succ_edge.
12990 (construct_exit_block): Use make_single_succ_edge.
12991 * cfghooks.c (verify_flow_info): Update.
12992 (redirect_edge_succ_nodup): Update.
12993 (split_edge): Update.
12994 (account_profile_record): Update.
12995 * cfgloopanal.c (single_likely_exit): Update.
12996 * cfgloopmanip.c (scale_loop_profile): Update.
12997 (set_zero_probability): Remove.
12998 (duplicate_loop_to_header_edge): Update.
12999 * cfgloopmanip.h (loop_version): Update prototype.
13000 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13001 (force_nonfallthru_and_redirect): Update.
13002 (update_br_prob_note): Update.
13003 (rtl_verify_edges): Update.
13004 (purge_dead_edges): Update.
13005 (rtl_lv_add_condition_to_bb): Update.
13006 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13007 * cgraphunit.c (init_lowered_empty_function): Update.
13008 (cgraph_node::expand_thunk): Update.
13009 * cilk-common.c: Include profile-count.h
13010 * dojump.c (inv): Remove.
13011 (jumpifnot): Update.
13012 (jumpifnot_1): Update.
13013 (do_jump_1): Update.
13015 (do_jump_by_parts_greater_rtx): Update.
13016 (do_compare_rtx_and_jump): Update.
13017 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13018 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13019 * dwarf2cfi.c: Include profile-count.h
13020 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13021 (sjlj_emit_dispatch_table): Likewise.
13022 * explow.c: Include profile-count.h
13023 * expmed.c (emit_store_flag_force): Update.
13024 (do_cmp_and_jump): Update.
13025 * expr.c (compare_by_pieces_d::generate): Update.
13026 (compare_by_pieces_d::finish_mode): Update.
13027 (emit_block_move_via_loop): Update.
13028 (store_expr_with_bounds): Update.
13029 (store_constructor): Update.
13030 (expand_expr_real_2): Update.
13031 (expand_expr_real_1): Update.
13032 * expr.h (try_casesi, try_tablejump): Update prototypes.
13033 * gimple-pretty-print.c (dump_probability): Update.
13034 (dump_profile): New.
13035 (dump_gimple_label): Update.
13036 (dump_gimple_bb_header): Update.
13037 * graph.c (draw_cfg_node_succ_edges): Update.
13038 * hsa-gen.c (convert_switch_statements): Update.
13039 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13040 (find_if_case_1): Update.
13041 (find_if_case_2): Update.
13042 * internal-fn.c (expand_arith_overflow_result_store): Update.
13043 (expand_addsub_overflow): Update.
13044 (expand_neg_overflow): Update.
13045 (expand_mul_overflow): Update.
13046 (expand_vector_ubsan_overflow): Update.
13047 * ipa-cp.c (good_cloning_opportunity_p): Update.
13048 * ipa-split.c (split_function): Use make_single_succ_edge.
13049 * ipa-utils.c (ipa_merge_profiles): Update.
13050 * loop-doloop.c (add_test): Update.
13051 (doloop_modify): Update.
13052 * loop-unroll.c (compare_and_jump_seq): Update.
13053 (unroll_loop_runtime_iterations): Update.
13054 * lra-constraints.c (lra_inheritance): Update.
13055 * lto-streamer-in.c (input_cfg): Update.
13056 * lto-streamer-out.c (output_cfg): Update.
13057 * mcf.c (adjust_cfg_counts): Update.
13058 * modulo-sched.c (sms_schedule): Update.
13059 * omp-expand.c (expand_omp_for_init_counts): Update.
13060 (extract_omp_for_update_vars): Update.
13061 (expand_omp_ordered_sink): Update.
13062 (expand_omp_for_ordered_loops): Update.
13063 (expand_omp_for_generic): Update.
13064 (expand_omp_for_static_nochunk): Update.
13065 (expand_omp_for_static_chunk): Update.
13066 (expand_cilk_for): Update.
13067 (expand_omp_simd): Update.
13068 (expand_omp_taskloop_for_outer): Update.
13069 (expand_omp_taskloop_for_inner): Update.
13070 * omp-simd-clone.c (simd_clone_adjust): Update.
13071 * optabs.c (expand_doubleword_shift): Update.
13072 (expand_abs): Update.
13073 (emit_cmp_and_jump_insn_1): Update.
13074 (expand_compare_and_swap_loop): Update.
13075 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13076 * predict.c (predictable_edge_p): Update.
13077 (edge_probability_reliable_p): Update.
13078 (set_even_probabilities): Update.
13079 (combine_predictions_for_insn): Update.
13080 (combine_predictions_for_bb): Update.
13081 (propagate_freq): Update.
13082 (estimate_bb_frequencies): Update.
13083 (force_edge_cold): Update.
13084 * profile-count.c (profile_count::dump): Add missing space into dump.
13085 (profile_count::debug): Add newline.
13086 (profile_count::differs_from_p): Explicitly convert to unsigned.
13087 (profile_count::stream_in): Update.
13088 (profile_probability::dump): New member function.
13089 (profile_probability::debug): New member function.
13090 (profile_probability::differs_from_p): New member function.
13091 (profile_probability::differs_lot_from_p): New member function.
13092 (profile_probability::stream_in): New member function.
13093 (profile_probability::stream_out): New member function.
13094 * profile-count.h (profile_count_quality): Rename to ...
13095 (profile_quality): ... this one.
13096 (profile_probability): New.
13097 (profile_count): Update.
13098 * profile.c (compute_branch_probabilities): Update.
13099 * recog.c (peep2_attempt): Update.
13100 * sched-ebb.c (schedule_ebbs): Update.
13101 * sched-rgn.c (find_single_block_region): Update.
13102 (compute_dom_prob_ps): Update.
13103 (schedule_region): Update.
13104 * sel-sched-ir.c (compute_succs_info): Update.
13105 * stmt.c (struct case_node): Update.
13106 (do_jump_if_equal): Update.
13107 (get_outgoing_edge_probs): Update.
13108 (conditional_probability): Update.
13109 (emit_case_dispatch_table): Update.
13110 (expand_case): Update.
13111 (expand_sjlj_dispatch_table): Update.
13112 (emit_case_nodes): Update.
13113 * targhooks.c: Update.
13114 * tracer.c (better_p): Update.
13115 (find_best_successor): Update.
13116 * trans-mem.c (expand_transaction): Update.
13117 * tree-call-cdce.c: Update.
13118 * tree-cfg.c (gimple_split_edge): Upate.
13119 (move_sese_region_to_fn): Upate.
13120 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13121 * tree-eh.c (lower_resx): Upate.
13122 (cleanup_empty_eh_move_lp): Upate.
13123 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13124 * tree-inline.c (copy_edges_for_bb): Update.
13125 (copy_cfg_body): Update.
13126 * tree-parloops.c (gen_parallel_loop): Update.
13127 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13128 (gimple_gen_time_profiler): Update.
13129 * tree-ssa-dce.c (remove_dead_stmt): Update.
13130 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13131 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13132 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13133 (unloop_loops): Update.
13134 (try_peel_loop): Update.
13135 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13136 * tree-ssa-loop-split.c (connect_loops): Update.
13137 (split_loop): Update.
13138 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13139 (hoist_guard): Update.
13140 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13141 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13142 (value_replacement): Update.
13143 * tree-ssa-reassoc.c (branch_fixup): Update.
13144 * tree-ssa-tail-merge.c (replace_block_by): Update.
13145 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13146 (create_edge_and_update_destination_phis): Update.
13147 (compute_path_counts): Update.
13148 (recompute_probabilities): Update.
13149 (update_joiner_offpath_counts): Update.
13150 (freqs_to_counts_path): Update.
13151 (duplicate_thread_path): Update.
13152 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13153 (struct switch_conv_info): Update.
13154 (gen_inbound_check): Update.
13155 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13156 (vect_do_peeling): Update.
13157 (vect_loop_versioning): Update.
13158 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13159 (optimize_mask_stores): Update.
13160 * ubsan.c (ubsan_expand_null_ifn): Update.
13161 * value-prof.c (gimple_divmod_fixed_value): Update.
13162 (gimple_divmod_fixed_value_transform): Update.
13163 (gimple_mod_pow2): Update.
13164 (gimple_mod_pow2_value_transform): Update.
13165 (gimple_mod_subtract): Update.
13166 (gimple_mod_subtract_transform): Update.
13167 (gimple_ic): Update.
13168 (gimple_stringop_fixed_value): Update.
13169 (gimple_stringops_transform): Update.
13170 * value-prof.h: Update.
13172 2017-06-29 Carl Love <cel@us.ibm.com>
13174 * config/rs6000/rs6000-c.c: Add support for built-in functions
13175 vector signed int vec_signed (vector float);
13176 vector signed long long vec_signed (vector double);
13177 vector signed int vec_signed2 (vector double, vector double);
13178 vector signed int vec_signede (vector double);
13179 vector signed int vec_signedo (vector double);
13180 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13181 instruction generator.
13182 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13183 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13184 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13186 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13187 vunsignede_v2df): Add define_expands.
13188 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13189 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13190 VEC_UNSIGNEDO): Add definitions.
13191 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13192 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13193 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13194 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13195 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13196 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13197 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13198 vec_unsignede, vec_unsignedo): Add builtin defines.
13199 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13201 * doc/extend.texi: Update the built-in documentation file for the
13202 new built-in functions.
13204 2017-06-29 Richard Biener <rguenther@suse.de>
13206 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13207 reduction chains to LOOP_VINFO_REDUCTIONS.
13208 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13209 SLP reductions after processing reduction chains.
13211 2017-06-29 Nathan Sidwell <nathan@acm.org>
13213 * builtins.c (fold_builtin_FUNCTION): Use
13214 lang_hooks.decl_printable_name.
13216 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13218 PR middle-end/81194
13219 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13220 with only one label.
13221 * stmt.c (expand_case): Assert NCASES is greater than one.
13223 2017-06-29 Richard Biener <rguenther@suse.de>
13225 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13227 (group_case_labels): Likewise.
13228 (find_taken_edge): Push sanity checking on val to workers...
13229 (find_taken_edge_cond_expr): ... here
13230 (find_taken_edge_switch_expr): ... and here, handle cases
13231 with just a default label.
13232 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13233 (group_case_labels): Likewise.
13234 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13235 group_case_labels does anything cleanup the CFG again.
13237 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13239 PR tree-optimization/81196
13240 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13241 exit condition comparing two IVs.
13243 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13245 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13246 profile to the dummy entry at the end of the list of architectures.
13247 * config/arm/arm-cpu-cdata.h: Regenerated.
13249 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13250 Michael Collison <michael.collison@arm.com>
13253 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13255 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13256 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13257 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13258 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13259 of shift when the shift amount is masked with constant equal to
13260 the size of the mode.
13261 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13264 2017-06-29 Martin Liska <mliska@suse.cz>
13266 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13268 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13271 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13274 (edge_growth_cache_entry::edge_growth_cache_entry): New
13276 (reset_edge_growth_cache): Update to use constructor.
13278 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13280 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13281 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13282 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13284 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13286 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13287 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13289 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13291 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13292 (*-linux-uclibc*): Add t-uclibc tmake_file.
13293 * config/t-musl: New.
13294 * config/t-uclibc: New.
13296 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13298 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13300 (gen_comm_data): Emit architectural setting of arch_prof.
13301 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13303 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13304 (armv8-m.base, armv8-m.main): Likewise.
13305 * arm-protos.h (arm_build_target): Add profile field.
13306 (arch_option): Likewise.
13307 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13309 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13310 arm_active_target.profile.
13312 2017-06-28 Richard Biener <rguenther@suse.de>
13314 PR middle-end/81227
13315 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13316 TYPE_OVERFLOW_WRAPS.
13317 * match.pd (negate_expr_p): Likewise.
13318 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13319 fold_build2, not fold_binary.
13321 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13323 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13324 Convert memory address to Pmode.
13325 (aarch64_print_operand): Assert MEM operands are always Pmode.
13327 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13330 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13331 Remove redundant if.
13332 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13333 * config/arm/aarch-common-protos.h
13334 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13335 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13337 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13340 * multiple_target.c (create_dispatcher_calls): Set the default
13341 clone to be static, not public.
13343 2017-06-28 Richard Biener <rguenther@suse.de>
13345 * tree-vect-loop.c (vectorizable_reduction): Move special
13346 cond reduction IV var creation ...
13347 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13348 parameter. Use STMT_VINFO_VECTYPE.
13349 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13352 2017-06-28 Martin Liska <mliska@suse.cz>
13355 * ipa-visibility.c (non_local_p): Handle visibility.
13357 2017-06-28 Martin Liska <mliska@suse.cz>
13360 * common.opt: Add IntegerRange to various options.
13361 * opt-functions.awk (integer_range_info): New function.
13362 * optc-gen.awk: Add integer_range_info to cl_options struct.
13363 * opts-common.c (decode_cmdline_option): Handle
13364 CL_ERR_INT_RANGE_ARG.
13365 (cmdline_handle_error): Likewise.
13366 * opts.c (print_filtered_help): Show valid interval in
13367 when --help is provided.
13368 * opts.h (struct cl_option): Add range_min and range_max fields.
13369 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13371 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13373 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13374 (x * C EQ/NE y * C): New transformation.
13376 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13378 * genmultilib (combination_space): Accept '+' in option names.
13380 2017-06-28 Martin Liska <mliska@suse.cz>
13383 * asan.c (instrument_derefs): Bail out inner references
13384 that are hard register variables.
13386 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13389 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13390 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13392 2017-06-28 Richard Biener <rguenther@suse.de>
13394 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13395 (vect_get_slp_defs): Adjust.
13396 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13398 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13400 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13401 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13402 (vectorizable_reduction): Adjust.
13403 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13405 (vect_get_slp_defs): Likewise.
13406 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13407 (vectorizable_bswap): Adjust.
13408 (vectorizable_call): Likewise.
13409 (vectorizable_conversion): Likewise.
13410 (vectorizable_assignment): Likewise.
13411 (vectorizable_shift): Likewise.
13412 (vectorizable_operation): Likewise.
13413 (vectorizable_store): Likewise.
13414 (vectorizable_condition): Likewise.
13415 (vectorizable_comparison): Likewise.
13417 2017-06-28 Michael Collison <michael.collison@arm.com>
13420 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13422 (arm_gen_movmemqi): Removed unused variable 'i'.
13423 Convert 'for' loop into 'while' loop.
13424 (arm_expand_prologue): Remove last unnecessary set of insn.
13425 (thumb_pop): Remove unused variable 'pushed_words'.
13426 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13428 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13430 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13431 * config/s390/s390-protos.h: Add prototype of
13432 s390_rel_address_ok_p.
13433 * config/s390/s390.c (s390_got_symbol): New function.
13434 (s390_rel_address_ok_p): New function.
13435 (legitimize_pic_address): Use s390_rel_address_ok_p.
13436 (s390_load_got): Use s390_got_symbol.
13437 (s390_option_override): Issue error if
13438 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13439 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13441 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13443 2017-06-27 Andrew Pinski <apinski@cavium.com>
13445 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13446 (X * copysign (1.0, X)): New pattern.
13447 (X * copysign (1.0, -X)): New pattern.
13448 (copysign (-1.0, CST)): New pattern.
13450 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13452 * genmultilib (combination_space): Remove variable.
13453 Validate reuse rules against regular expression for any sequence
13454 of multilib options in any order.
13456 2017-06-27 Michael Collison <michael.collison@arm.com>
13458 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13459 call aarch64_split_simd_combine.
13460 * (aarch64_combine_internal<mode>): Delete pattern.
13461 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13462 Allow register and subreg operands.
13464 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13466 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13467 specific need, just fallback on defaults.
13468 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13470 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13471 Olivier Hainque <hainque@adacore.com>
13473 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13475 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13476 targets. Pick a default if no particular attempt applied.
13477 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13480 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13482 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13483 (x86_64-wrs-vxworks7): Likewise.
13485 2017-06-27 Marek Polacek <polacek@redhat.com>
13488 * ubsan.c (instrument_null): Check get_base_address's result for null.
13490 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13492 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13494 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13496 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13497 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13498 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13499 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13500 New function types.
13501 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13502 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13503 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13504 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13505 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13506 BUILT_IN_FEUPDATEENV): New builtins.
13507 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13508 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13509 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13510 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13512 (builtin_structptr_types): Adjust size.
13513 * tree.c (builtin_structptr_types): Add four entries.
13515 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13516 Olivier Hainque <hainque@adacore.com>
13518 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13519 (TLS_SYM): New local macro, forcing reference to __tls__ on
13520 link command lines for VxWorks 7 RTPs, triggering initialization
13522 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13523 OS features TLS support, true for RTPs on VxWorks 7.
13524 * config/vxworks.c (vxworks_override_options): Setup emutls
13527 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13529 * predict.c (test_prediction_value_range): Use -1U instead of -1
13530 to avoid narrowing conversion warning.
13531 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13532 to avoid narrowing conversion warning.
13533 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13535 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13537 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13539 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13540 64bit configurations.
13541 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13542 (SIZE_TYPE): Likewise.
13543 * config/vxworks.c (vxworks_emutls_var_fields): Use
13544 long_unsigned_type_node instead of unsigned_type_node as the offset
13545 field type, which is "pointer" mode in emutls.c.
13547 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13550 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13552 PR middle-end/81207
13553 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13554 gimple_vuse copying separately from gimple_vdef copying.
13556 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13558 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13561 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13562 Olivier Hainque <hainque@adacore.com>
13564 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13565 for all vxworks7 targets.
13566 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13567 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13568 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13569 variations for VX6/VX7 and 32/64bits later on in ...
13570 (VXWORKS_LIB_SPEC): Leverage new macros.
13571 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13572 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13574 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13576 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13577 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13579 2017-06-26 Carl Love <cel@us.ibm.com>
13581 * config/rs6000/rs6000-c.c: Add support for built-in functions
13582 vector bool char vec_reve (vector bool char);
13583 vector signed char vec_reve (vector signed char);
13584 vector unsigned char vec_reve (vector unsigned char);
13585 vector bool int vec_reve (vector bool int);
13586 vector signed int vec_reve (vector signed int);
13587 vector unsigned int vec_reve (vector unsigned int);
13588 vector bool long long vec_reve (vector bool long long);
13589 vector signed long long vec_reve (vector signed long long);
13590 vector unsigned long long vec_reve (vector unsigned long long);
13591 vector bool short vec_reve (vector bool short);
13592 vector signed short vec_reve (vector signed short);
13593 vector double vec_reve (vector double);
13594 vector float vec_reve (vector float);
13595 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13596 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13597 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13598 (altivec_vreve): New pattern.
13599 * config/rs6000/altivec.h (vec_reve): New define.
13600 * doc/extend.texi (vec_rev): Update the built-in documentation file
13601 for the new built-in functions.
13603 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13605 PR tree-optimization/71815
13606 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13608 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13610 (slsr_process_phi): Likewise.
13611 (replace_uncond_cands_and_profitable_phis): Don't replace a
13612 multiply candidate with a stride of 1 (copy or cast).
13613 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13615 (lowest_cost_path): Likewise.
13616 (total_savings): Likewise.
13618 2017-06-26 Richard Biener <rguenther@suse.de>
13621 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13622 Use def_builtin_pure for all gather builtins.
13624 2017-06-26 Richard Biener <rguenther@suse.de>
13626 PR tree-optimization/81203
13627 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13628 non-dominating BBs.
13630 2017-06-26 Marek Polacek <polacek@redhat.com>
13633 * doc/invoke.texi: Document -Wmultistatement-macros.
13635 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13637 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13638 arm_neon_ok_no_float_abi effective target.
13640 2017-06-26 Richard Biener <rguenther@suse.de>
13642 PR tree-optimization/80928
13643 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13644 (copy_bbs): Set BB_DUPLICATED flag early.
13645 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13647 (execute_on_shrinking_pred): Likewise.
13648 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13649 BB_DUPLICATED blocks.
13650 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13651 iterate over all PHIs considering removal of *gsi.
13653 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13655 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13658 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13660 * config/rs6000/rs6000-string.c: (expand_block_clear,
13661 do_load_for_compare, select_block_compare_mode,
13662 compute_current_alignment, expand_block_compare,
13663 expand_strncmp_align_check, expand_strn_compare,
13664 expand_block_move, rs6000_output_load_multiple)
13665 Move functions related to string/block move/compare
13666 to a separate file.
13667 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13668 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13669 for this function which is now used in two files.
13670 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13671 * config.gcc: Add rs6000-string.o to extra_objs for
13672 targets powerpc*-*-* and rs6000*-*-*.
13674 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13677 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13678 32-bit, since indexed is not valid for DImode.
13679 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13680 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13681 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13682 (define_peephole2 for Altivec d-form store): Likewise.
13685 * multiple_target.c (create_dispatcher_calls): Only create the
13686 dispatcher call if the function is the default clone of a
13687 versioned function.
13689 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13691 PR middle-end/80902
13692 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13693 a call, force the call to not be a tail call.
13695 2017-06-23 Jeff Law <law@redhat.com>
13697 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13700 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13702 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13703 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13704 handling for early expansion of vector shifts (sl,sr,sra,rl).
13705 (builtin_function_type): Add vector shift right instructions
13706 to the unsigned argument list.
13708 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13710 rtl-optimizatoin/79286
13711 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13712 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13713 trap. PIC register plus a const unspec without offset can never trap.
13715 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13717 * tree.h (builtin_structptr_type): New type.
13718 (builtin_structptr_types): Declare new array.
13719 * tree.c (builtin_structptr_types): New array.
13720 (free_lang_data, build_common_tree_nodes): Use it.
13722 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13725 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13728 2017-06-22 Matt Turner <mattst88@gmail.com>
13730 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13731 Lake models to skylake case. Assume skylake for unknown
13732 models with clflushopt.
13734 2017-06-22 Jeff Law <law@redhat.com>
13736 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13737 frame sizes that do not satisfy aarch64_uimm12_shift.
13739 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13741 * profile-count.h (apply_probability,
13742 apply_scale, probability_in): Fix checks for zero.
13744 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13746 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13747 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13749 2016-06-22 Richard Biener <rguenther@suse.de>
13751 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13752 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13753 REDUC_MAX_EXPR support.
13754 (vectorizable_reduction): Likewise.
13755 (vect_create_epilog_for_reduction): Likewise.
13757 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13759 * match.pd (A / (1 << B) -> A >> B): New.
13760 * generic-match-head.c: Include optabs-tree.h.
13761 * gimple-match-head.c: Likewise.
13762 * optabs-tree.h (target_supports_op_p): New.
13763 * optabs-tree.c (target_supports_op_p): New.
13765 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13767 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13768 $gcc_cv_ld --help output.
13769 (gcc_cv_ld_demangle): Likewise.
13770 (gcc_cv_ld_eh_frame_hdr): Likewise.
13771 (gcc_cv_ld_pie): Likewise.
13772 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13773 (gcc_cv_ld_buildid): Likewise.
13774 (gcc_cv_ld_sysroot): Likewise.
13775 (ld_bndplt_support): Likewise.
13776 (ld_pushpopstate_support): Likewise.
13777 * configure: Regenerate.
13778 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13780 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13783 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13784 operands indexes to avoid gap between operands and match_dups.
13786 2017-06-21 Andrew Pinski <apinski@cavium.com>
13788 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13789 Increment Arith_shift and Arith_shift_reg by 1.
13790 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13792 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13793 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13794 (aarch64_strip_extend): Add new argument and test for it.
13795 (aarch64_cheap_mult_shift_p): New function.
13796 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13797 add a cost if it is true.
13798 Update calls to aarch64_strip_extend.
13799 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13801 2017-06-21 Andrew Pinski <apinski@cavium.com>
13803 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13805 (thunderxt88): Likewise.
13806 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13807 (thunderx_prefetch_tune): New variable.
13808 (thunderx2t99_prefetch_tune): Update for the correct values.
13809 (thunderxt88_tunings): New variable.
13810 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13811 generic_prefetch_tune.
13812 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13814 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13816 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13817 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13818 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13819 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13820 (aarch64_atomic_cas<mode>, GPI): Likewise.
13822 2017-06-21 Martin Liska <mliska@suse.cz>
13824 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13825 statements on cold and hot labels.
13826 * predict.c (tree_estimate_probability_bb): Remove the
13827 prediction from this place.
13829 2017-06-21 Martin Liska <mliska@suse.cz>
13831 PR tree-optimization/79489
13832 * gimplify.c (maybe_add_early_return_predict_stmt): New
13834 (gimplify_return_expr): Call the function.
13835 * predict.c (tree_estimate_probability_bb): Remove handling
13837 * predict.def: Update comment about early return predictor.
13838 * gimple-predict.h (is_gimple_predict): New function.
13839 * predict.def: Change default value of early return to 66.
13840 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13842 * passes.def: Put pass_strip_predict_hints to the beginning of
13845 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13847 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13848 FUNCTION_DECL declarations.
13849 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13851 (dwaf2out_decl): Likewise.
13852 * godump.c (go_early_global_decl): Skip call to the real debug hook
13853 for FUNCTION_DECL declarations.
13854 * passes.c (rest_of_decl_compilation): Skip call to the
13855 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13856 -fdump-go-spec is passed.
13858 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13860 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13861 Reorder for compactness.
13862 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13863 (def_builtin_pure, def_builtin_pure2): New functions.
13864 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13866 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13868 * match.pd (nop_convert): New predicate.
13869 ((A +- CST1) +- CST2): Allow some NOP conversions.
13871 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13874 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13875 with ctors/dtors if GOVD_SHARED is set.
13877 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13879 * config/aarch64/aarch64.md (movti_aarch64):
13880 Emit mov rather than orr.
13881 (movtf_aarch64): Likewise.
13882 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13883 Emit mov rather than orr.
13885 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13887 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13888 Swap alternatives, make integer dup more expensive.
13890 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13892 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13893 Return true for non-tls symbols.
13895 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13897 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13898 (cortex-a75): Likewise.
13899 (cortex-a75.cortex-a55): Likewise.
13900 * config/aarch64/aarch64-tune.md: Regenerate.
13901 * doc/invoke.texi (-mtune): Document new values for -mtune.
13903 2017-06-21 Tom de Vries <tom@codesourcery.com>
13905 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13906 stack_size feature.
13907 (Effective-Target Keywords, Other attributes): Suggest using
13908 dg-add-options stack_size feature to get stack limit in stack_size
13909 effective target documentation.
13911 2017-06-21 Julian Brown <julian@codesourcery.com>
13912 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13914 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13915 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13916 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13918 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13919 attribute type list for neon_multiply.
13920 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13921 attribute type list for neon_multiply.
13922 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13923 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13924 attribute type list for neon_multiply.
13925 * config/arm/types.md (crypto_pmull): Add.
13926 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13927 attribute type list.
13929 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13931 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13934 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13936 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13937 to make sure not to dereference a NULL cost_classes_ptr pointer.
13939 2017-06-20 Carl Love <cel@us.ibm.com>
13941 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13942 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13943 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13944 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13945 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13946 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13947 VMULOSW): New enum "unspec" values.
13948 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13949 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13950 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13951 altivec_vmulosw): New patterns.
13952 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13953 VMULOSW): Add definitions.
13955 2017-06-20 Julia Koval <julia.koval@intel.com>
13957 * config/i386/i386.c: Fix rounding expand for new pattern.
13958 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13960 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13962 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13963 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13965 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13967 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13970 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13972 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13973 architecture, then by implementer ID.
13974 * config/aarch64/aarch64-tune.md: Regenerate.
13976 2017-06-20 Richard Biener <rguenther@suse.de>
13978 PR middle-end/81097
13979 * fold-const.c (split_tree): Fold to type before negating.
13981 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13983 * diagnostic-show-locus.c
13984 (selftest::test_fixit_deletion_affecting_newline): New function.
13985 (selftest::diagnostic_show_locus_c_tests): Call it.
13987 2017-06-20 Andreas Schwab <schwab@suse.de>
13990 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13993 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13995 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13996 __ARM_FEATURE_COPROC according to support.
13998 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14000 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14001 Rewritten to avoid overflow for > 32-bit pointers.
14004 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14005 by removing enum keyword.
14006 (ubsan_type_descriptor): Likewise. Formatting fix.
14009 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14010 splitter): Require TARGET_SSE2 in the condition.
14012 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14015 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14016 for doing vector set of SFmode on ISA 3.0.
14017 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14018 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14020 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14021 SFmode value into a V4SF variable that was extracted from another
14022 V4SF variable without converting the element to double precision
14023 and back to single precision vector format.
14024 (vsx_insert_extract_v4sf_p9_2): Likewise.
14026 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14028 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14029 in UWHI to avoid undefined overflow.
14032 * ubsan.h (enum ubsan_encode_value_phase): New.
14033 (ubsan_encode_value): Change second argument to
14034 enum ubsan_encode_value_phase with default value of
14035 UBSAN_ENCODE_VALUE_GENERIC.
14036 * ubsan.c (ubsan_encode_value): Change second argument to
14037 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14038 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14039 create_tmp_var_raw instead of create_tmp_var and use a
14041 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14042 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14043 ubsan_encode_value callers.
14046 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14047 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14048 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14050 2017-06-19 Richard Biener <rguenther@suse.de>
14052 PR middle-end/81118
14053 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14054 estimates if we changed anything.
14056 2017-06-19 Richard Biener <rguenther@suse.de>
14058 PR tree-optimization/80887
14059 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14060 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14061 simplified lookups, then reset mprts_hook.
14062 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14064 (try_to_simplify): Likewise.
14066 2017-06-19 Martin Liska <mliska@suse.cz>
14069 * gimplify.c (gimplify_switch_expr):
14070 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14072 2017-06-19 Martin Liska <mliska@suse.cz>
14074 * doc/install.texi: Document that PGO runs in 4 stages.
14076 2017-06-19 Martin Liska <mliska@suse.cz>
14079 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14080 to dispatcher function name.
14081 * multiple_target.c (replace_function_decl): New function.
14082 (create_dispatcher_calls): Redirect both edges and references.
14084 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14086 * profile-count.c (profile_count::dump): Dump quality.
14087 (profile_count::differs_from_p): Update for unsigned val.
14088 * profile-count.h (profile_count_quality): New enum.
14089 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14091 2017-06-19 Richard Biener <rguenther@suse.de>
14093 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14094 struct function as arg.
14095 (estimate_numbers_of_iterations): Export overload with loop arg.
14096 (free_numbers_of_iterations_estimates_loop): Use an overload of
14097 free_numbers_of_iterations_estimates instead.
14098 * tree-cfg.c (remove_bb): Adjust.
14099 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14100 * tree-parloops.c (gen_parallel_loop): Likewise.
14101 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14103 (tree_unroll_loops_completely): Likewise.
14104 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14105 Use an overload instead and export.
14106 (estimated_loop_iterations): Adjust.
14107 (max_loop_iterations): Likewise.
14108 (likely_max_loop_iterations): Likewise.
14109 (estimate_numbers_of_iterations): Take struct function as arg
14111 (loop_exits_before_overflow): Adjust.
14112 (free_numbers_of_iterations_estimates_loop): Use an overload.
14113 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14114 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14116 2017-06-19 Richard Biener <rguenther@suse.de>
14119 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14120 RANGE_EXPR conservatively.
14122 2017-06-16 Carl Love <cel@us.ibm.com>
14124 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14125 definitions for vec_float, vec_float2, vec_floato,
14126 vec_floate built-ins.
14127 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14128 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14130 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14131 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14132 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14133 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14134 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14135 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14136 vec_floato): Add builtin defines.
14137 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14138 Update the built-in documentation file for the new built-in
14141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14143 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14144 (mthumb): Mark as the negative of -marm.
14146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14148 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14150 (ARM Options, -mtune): Document that this accepts the same
14151 extension options as -mcpu.
14152 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14154 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14156 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14157 permitted extensions.
14159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14161 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14162 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14163 (armv8-m.main): Add option +nodsp.
14164 * config/arm/arm-cpu-cdata.h: Regenerated.
14166 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14168 * config/arm/t-fuchsia: New file.
14169 * config.gcc (arm*-*-fuchsia*): Use it.
14171 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14173 * config/arm/t-symbian: Rewrite for new option infrastructure.
14175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14177 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14178 (MULTILIB_REQUIRED): Likewise.
14180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14182 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14183 (MULTILIB_RESUE): Likewise.
14184 (MULTILIB_MATCHES): Likewise.
14185 (MULTLIB_REQUIRED): Likewise.
14187 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14189 * config/arm/t-rtems: Rewrite for new option framework.
14191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14193 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14194 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14195 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14196 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14197 * config/arm/t-multilib: ... here.
14198 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14199 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14200 armv7-a and armv8*-a when A-profile libraries have not been built.
14201 * config/arm/t-rmprofile: Rewrite.
14203 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14205 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14206 with a backslash. Remove the backslash after substituting unescaped
14208 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14210 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14212 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14213 the driver to pass through the default setting of -mfloat-abi.
14214 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14216 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14217 (all_feat_combs): New rule.
14218 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14220 * config/arm/t-aprofile: Rewrite.
14222 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14224 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14225 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14226 fpu is not specified by the user/command-line.
14227 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14228 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14229 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14230 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14231 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14232 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14234 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14236 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14237 * config/arm/t-arm-elf: Rewritten.
14239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14241 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14242 have some floating-point instructions.
14243 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14244 (TARGET_MAYBE_HARD_FLOAT): New macro.
14245 * config/arm/arm-builtins.c (arm_init_builtins): Use
14246 TARGET_MAYBE_HARD_FLOAT.
14247 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14249 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14251 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14252 (configargs.h): Include it.
14253 (arm_print_hint_for_fpu_option): New function.
14254 (arm_parse_fpu_option): New function.
14255 (candidate_extension): New class.
14256 (arm_canon_for_multilib): New function.
14257 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14258 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14259 (ARCH_CANONICAL_SPECS): New macro.
14260 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14262 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14264 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14265 are set after handling multilib fragments. Set target_cpu_default2
14268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14270 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14272 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14273 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14275 (check_arch): Likewise.
14276 * config/arm/arm.c (arm_configure_build_target): Handle
14277 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14278 options in the default.
14280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14282 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14283 when an option is an alias of another.
14284 * config/arm/parsecpu.awk (optalias): New parser token.
14285 (gen_comm_data): Mark non-alias options as such. Emit entries
14286 for extension aliases.
14287 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14288 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14289 (armv6kz, armv6zk, armv6t2): Likewise.
14290 (armv7): Make vfpv3-d16 an alias.
14291 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14293 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14294 Sort in canonical order.
14295 (armv8-a): Sort in canonical order.
14296 (armv8.1-a, armv8.2-a): Likewise.
14297 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14299 (cortex-a9): Sort in canonical order.
14300 * config/arm/arm.c (selftests.h): Include it.
14301 (arm_test_cpu_arch_data): New function.
14302 (arm_run_self_tests): New function.
14303 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14304 (targetm): Move declaration to the end of the file.
14305 * arm-cpu-cdata.h: Regenerated.
14307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14309 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14310 call to target_mode_check describing the type of option passed.
14311 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14312 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14313 arm_parse_cpu_option_name to match parameters against list of
14315 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14316 arm_arch_core_flags data structure.
14317 * config/arm/arm-cpu_cdata.h: Regenerated.
14319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14321 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14323 (arm_print_hint_for_cpu_option): Likewise.
14324 (arm_print_hint_for_arch_option): Likewise.
14325 (arm_parse_cpu_option_name): Likewise.
14326 (arm_parse_arch_option_name): Likewise.
14327 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14328 of entries in the all_fpus list.
14329 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14330 (arm_parse_cpu_option_name): Declare.
14331 (arm_parse_arch_option_name): Declare.
14332 (arm_parse_option_features): Declare.
14333 (arm_intialize_isa): Declare.
14334 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14336 (gen_comm_data): ... here. Make definitions non-static.
14337 * config/arm/arm-cpu-data.h: Regenerated.
14338 * config/arm/arm-cpu-cdata.h: Regenerated.
14340 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14342 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14343 (cpu_arch_extension): New structure.
14344 (cpu_arch_option, arch_option, cpu_option): New structures.
14345 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14346 architecture types.
14347 (gen_data): Generate new format data tables.
14348 * config/arm/arm.c (cpu_tune): New structure.
14349 (cpu_option, processors): Delete.
14350 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14351 (arm_print_hint_for_cpu_option): ... this and ...
14352 (arm_print_hint_for_arch_option): ... this.
14353 (arm_parse_arch_cpu_name): Delete. Replace with ...
14354 (arm_parse_cpu_option_name): ... this and ...
14355 (arm_parse_arch_option_name): ... this.
14356 (arm_unrecognized_feature): Change type of target parameter to
14358 (arm_parse_arch_cpu_features): Delete. Replace with ...
14359 (arm_parse_option_features): ... this.
14360 (arm_configure_build_target): Rework to use new configuration data
14362 (arm_print_tune_info): Rework for new configuration data tables.
14363 * config/arm/arm-cpu-data.h: Regenerated.
14364 * config/arm/arm-cpu.h: Regenerated.
14366 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14368 * Makefile.in (OBJS): Move sbitmap.o from here ...
14369 (OBJS-libcommon): ... to here.
14371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14373 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14374 (ISA_ALL_CRYPTO): New macro.
14375 (ISA_ALL_SIMD): New macro
14376 (ISA_ALL_FP): New macro.
14377 * config/arm/arm.c (fpu_bitlist): Update initializer.
14378 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14380 (arm9e): Add fpu. Add option for nofp
14381 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14382 (arm926ej-s, arm1026ej-s): Likewise.
14383 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14384 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14385 neon-fp16, neon-vfpv4, nofp and nosimd.
14386 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14387 (cortex-a8): Add fpu. Add option for nofp.
14388 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14389 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14390 (cortex-r4f): Add fpu.
14391 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14392 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14394 (cortex-r8): Likewise.
14395 (cortex-m4): Add fpu. Add option for nofp.
14396 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14397 (cortex-a17.cortex-a7): Likewise.
14398 (cortex-a32): Add fpu. Add options for crypto and nofp.
14399 (cortex-a35, cortex-a53): Likewise.
14400 (cortex-a57): Add fpu. Add option for crypto.
14401 (cortex-a72, cortex-a73): Likewise.
14402 (exynos-m1): Likewise.
14403 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14404 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14405 (cortex-m33): Add fpu. Add option for nofp.
14406 * config/arm/arm-cpu-cdata.h: Regenerated
14407 * config/arm/arm-cpu-data.h: Regenerated.
14409 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14411 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14412 (armv5te, armv5tej): Likewise.
14413 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14414 (armv7): Add options fp and vfpv3-d16.
14415 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14416 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14418 (armv7ve): Likewise.
14419 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14420 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14421 (armv8-a): Add nocrypto option.
14422 (armv8.1-a, armv8.2-a): Likewise.
14423 (armv8-m.main): add options fp, fp.dp and nofp.
14425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14427 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14429 (armv8-a+crc): Delete.
14430 (armv8.1-a): Add options simd, crypto and nofp.
14431 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14432 (armv8.2-a+fp16): Delete.
14433 (armv8-m.main): Add option dsp.
14434 (armv8-m.main+dsp): Delete.
14435 (cortex-a8): Add fpu. Add nofp option.
14436 (cortex-a9): Add fpu. Add nofp and nosimd options.
14437 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14438 link to main cpu and architecture data structures.
14439 (gen_comm_data): Only put isa attributes from the main architecture
14441 (option): New statement for architecture and CPU entries.
14442 * arm.c (struct cpu_option): New structure.
14443 (struct processors): Add entry for options.
14444 (arm_unrecognized_feature): New function.
14445 (arm_parse_arch_cpu_name): Ignore any characters after the first
14447 (arm_parse_arch_cpu_feature): New function.
14448 (arm_configure_build_target): Separate out any CPU and architecture
14449 features and parse separately. Don't error out if -mfpu=auto is
14450 used with only an architecture string.
14451 (arm_print_asm_arch_directives): New function.
14452 (arm_file_start): Call it.
14453 * config/arm/arm-cpu-cdata.h: Regenerated.
14454 * config/arm/arm-cpu-data.h: Likewise.
14455 * config/arm/arm-tables.opt: Likewise.
14457 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14459 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14460 assembler when it is not -mfpu=auto.
14462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14464 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14465 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14466 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14467 (ASM_CPU_SPEC): Rewrite.
14468 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14469 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14470 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14471 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14472 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14473 copied string is NUL-terminated. Also strip any characters prefixed
14475 (arm_rewrite_selected_arch): New function.
14476 (arm_rewrite_march): New function.
14478 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14480 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14481 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14482 (march, mcpu, mtune): Convert to string-based options.
14483 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14484 (arm_parse_arch_cpu_name): New function.
14485 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14486 identify selected architecture or CPU.
14487 (arm_option_save): New function.
14488 (TARGET_OPTION_SAVE): Redefine.
14489 (arm_option_restore): Restore string options.
14490 (arm_option_print): Print string options.
14492 2017-06-16 Martin Sebor <msebor@redhat.com>
14494 PR tree-optimization/80933
14495 PR tree-optimization/80934
14496 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14497 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14498 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14499 (gimple_fold_builtin): Call them.
14501 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14503 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14504 as unlikely; update profile.
14506 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14508 * predict.c (force_edge_cold): Handle declaring edges impossible
14511 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14513 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14515 (try_unroll_loop_completely): Fix reporting.
14517 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14519 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14521 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14524 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14525 if given a non-constant argument for an intrinsic which requires a
14528 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14530 * profile.c (compare_freqs): New function.
14531 (branch_prob): Sort edge list.
14532 (find_spanning_tree): Assume that the list is priority sorted.
14534 2017-06-16 Richard Biener <rguenther@suse.de>
14536 PR tree-optimization/81090
14537 * passes.def (pass_record_bounds): Remove.
14538 * tree-pass.h (make_pass_record_bounds): Likewise.
14539 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14540 make_pass_record_bounds): Likewise.
14541 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14542 not free niter estimates at the beginning but at the end.
14543 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14545 2017-06-16 Richard Biener <rguenther@suse.de>
14547 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14548 initializer to workaround ICE in host GCC 4.8.
14550 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14552 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14554 (clone_inlined_nodes): Update.
14556 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14558 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14559 prefetch settings, and enable prefetching by default at -O3.
14561 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14563 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14564 Set flag_prefetch_loop_arrays according to tuning data.
14566 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14568 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14569 New tune structure.
14570 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14571 [Unrelated to main purpose of the patch] Place the pointer field last
14572 to enable type checking errors when tune structure are wrongly merged.
14573 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14574 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14575 (thunderx2t99_prefetch_tune): New tune constants.
14576 (tune_params *_tunings): Update all tunings (no functional change).
14577 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14578 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14579 from tunings structures.
14581 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14584 * ubsan.c (instrument_null): Add T argument, use it instead
14585 of computing it based on IS_LHS.
14586 (instrument_object_size): Likewise.
14587 (pass_ubsan::execute): Adjust instrument_null and
14588 instrument_object_size callers to pass gimple_get_lhs or
14589 gimple_assign_rhs1 result to it. Use instrument_null instead of
14590 calling get_base_address and instrument_mem_ref. Handle
14591 aggregate call arguments for object-size sanitization.
14593 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14595 PR tree-optimization/81089
14596 * tree-vrp.c (is_masked_range_test): Validate operands of
14599 2017-06-15 Martin Sebor <msebor@redhat.com>
14602 * dumpfile.c (dump_register): Avoid calling memset to initialize
14603 a class with a default ctor.
14604 * gcc.c (struct compiler): Remove const qualification.
14605 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14606 * hash-table.h: Ditto.
14607 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14609 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14610 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14612 * params.h (struct param_info): Make struct members non-const.
14613 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14614 with default initialization.
14615 * vec.h (vec_copy_construct, vec_default_construct): New helper
14617 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14618 with vec_copy_construct.
14619 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14620 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14621 * doc/invoke.texi (-Wclass-memaccess): Document.
14623 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14625 * emit-rtl.h (is_leaf): Update comment about local
14626 register allocator.
14628 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14631 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14632 for a variable to have a section before checking if the section has a
14634 Set section to.persistent if persistent attribute is set.
14635 Warn if .persistent attribute is used on an automatic variable.
14637 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14639 PR rtl-optimization/80474
14640 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14642 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14644 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14647 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14649 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14650 (casesi): Emit bounds checking as RTL.
14651 (casesi_internal_mips16_<mode>): Remove bounds checking.
14653 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14655 * config/xtensa/xtensa.c (xtensa_option_override): Append
14656 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14657 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14658 xtensa_doloop_hooks): Define unconditionally.
14659 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14661 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14662 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14663 for it in xtensa_option_override.
14664 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14665 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14667 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14669 * doc/cppopts.texi: Document '-' special value to -MF.
14671 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14673 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14674 (cortex_a53_fconst): Likewise.
14675 (cortex_a53_fpmul): Likewise.
14676 (cortex_a53_f_load_64): Likewise.
14677 (cortex_a53_f_load_many): Likewise.
14678 (cortex_a53_advsimd_alu): Likewise.
14679 (cortex_a53_advsimd_alu_q): Likewise.
14680 (cortex_a53_advsimd_mul): Likewise.
14681 (cortex_a53_advsimd_mul_q): Likewise.
14682 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14683 Add missing fmul, r2f_cvt and fconst cases.
14685 2017-06-14 Richard Biener <rguenther@suse.de>
14687 PR middle-end/81088
14688 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14690 (fold_binary_loc): When associating do not treat pre-existing
14691 TREE_OVERFLOW on literal constants as a reason to allow
14692 TREE_OVERFLOW on associated literal constants.
14694 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14696 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14697 (MASK_FEATURES): New macro.
14698 * config/sparc/sparc.c (sparc_option_override): Remove the special
14699 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14701 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14703 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14704 a division of 0 if non-call exceptions are enabled.
14706 2017-06-14 Andrew Pinski <apinski@cavium.com>
14707 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14710 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14711 Improve vector initialization code gen for only variable case.
14713 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14715 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14717 2017-06-14 Richard Biener <rguenther@suse.de>
14719 PR tree-optimization/81083
14720 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14723 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14725 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14726 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14727 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14728 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14729 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14730 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14732 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14734 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14735 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14737 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14739 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14741 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14743 * config/rs6000/t-rtems: Don't handle SPE.
14745 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14747 * config/rs6000/t-linux: Don't handle SPE.
14749 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14751 * config/rs6000/eabispe.h: Delete file.
14753 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14755 * config/rs6000/t-spe: Delete file.
14757 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14759 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14760 (rs6000_legitimate_offset_address_p): Return false for anything in
14761 V2SImode or V2SFmode.
14763 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14765 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14766 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14767 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14769 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14770 (rs6000_legitimate_offset_address_p): Ditto.
14771 (rs6000_emit_move): Ditto.
14772 (rs6000_init_builtins): Remove V4HI_type_node.
14774 2017-06-13 Martin Liska <mliska@suse.cz>
14777 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14778 (gate_asan): Likewise.
14779 * asan.h (asan_no_sanitize_address_p): Remove the function.
14780 (sanitize_flags_p): New function.
14781 * builtins.def: Fix coding style.
14782 * common.opt: Use renamed enum value.
14783 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14784 * doc/extend.texi: Document no_sanitize attribute.
14785 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14786 to SANITIZE_UNDEFINED_NONDEFAULT.
14787 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14788 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14789 Use sanitize_flags_p.
14790 * gimplify.c (gimplify_function_tree): Likewise.
14791 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14792 * opts.c (parse_no_sanitize_attribute): New function.
14793 (common_handle_option): Use renamed enum value.
14794 * opts.h (parse_no_sanitize_attribute): Declare.
14795 * tree.c (sanitize_flags_p): New function.
14796 * tree.h: Declared here.
14797 * tsan.c: Use sanitize_flags_p.
14798 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14799 (instrument_mem_ref): Likewise.
14800 (instrument_bool_enum_load): Likewise.
14801 (do_ubsan_in_current_function): Remove the function.
14802 (pass_ubsan::execute): Use sanitize_flags_p.
14803 * ubsan.h: Remove do_ubsan_in_current_function
14804 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14805 (dump_function_to_file): Use it here.
14807 2017-06-13 Martin Jambor <mjambor@suse.cz>
14809 PR tree-optimization/80803
14810 PR tree-optimization/81063
14811 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14812 (propagate_subaccesses_across_link): Enqueue subtree whenever
14813 necessary instead of relying on the caller.
14815 2017-06-13 Martin Jambor <mjambor@suse.cz>
14817 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14818 that have a first_link.
14819 (sort_and_splice_var_accesses): Do not check first_link before
14821 (subtree_mark_written_and_enqueue): Likewise.
14822 (propagate_all_subaccesses): Likewise and do not stop at first
14823 parent with a first_link.
14825 2017-06-13 Martin Jambor <mjambor@suse.cz>
14827 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14830 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14832 * match.pd: New pattern.
14834 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14836 * tree-vrp.c (is_masked_range_test): New function.
14837 (register_edge_assert_for): Determine ranges for
14840 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14842 PR tree-optimization/67328
14843 * fold-const.c (maskable_range_p): New function.
14844 (build_range_check): Generate bittests if possible.
14846 2017-06-13 Martin Liska <mliska@suse.cz>
14848 * gimple-pretty-print.c (dump_probability): Add new argument.
14849 (dump_edge_probability): Dump both probability and count.
14850 (dump_gimple_label): Likewise.
14851 (dump_gimple_bb_header): Likewise.
14853 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14856 * config/avr/avr-devices.c: Fix indentation.
14857 * config/avr/gen-avr-mmcu-specs.c: Dito.
14859 2017-06-13 Richard Biener <rguenther@suse.de>
14861 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14862 instead get vector type from stmt_info.
14863 (vectorizable_reduction): Adjust. Remove dead code.
14865 2017-06-13 Richard Biener <rguenther@suse.de>
14867 PR middle-end/81065
14868 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14869 case of C * (x * C2 + C3).
14870 (fold_addr_of_array_ref_difference): Properly fold index difference.
14872 2017-06-12 David S. Miller <davem@davemloft.net>
14875 * config/sparc/sparc.md (return expander): Emit frame blockage if
14876 function uses alloca.
14878 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14880 * combine.c (make_field_assignment): Check len rather than the mode
14881 precision when calling force_to_mode.
14883 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14885 Support multilibs and devices that see flash in RAM address range.
14888 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14889 (avr_mcu_t) <flash_pm_offset>: New field.
14890 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14891 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14892 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14893 (AVR_TINY_PM_OFFSET): Remove macro.
14894 * config/avr/avr.opt (-mshort-calls): New option.
14895 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14896 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14897 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14898 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14899 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14900 instead of avr_arch->have_jmp_call.
14901 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14902 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14903 avr_arch->flash_pm_offset to define.
14904 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14905 new field flash_pm_offset. Add entry for avrxmega3.
14906 (avr_texinfo): Add entry for avrxmega3.
14907 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14908 attiny212, attiny214,
14909 attiny412, attiny414, attiny416, attiny417,
14910 attiny814, attiny816, attiny817,
14911 attiny1614, attiny1616, attiny1617,
14912 attiny3214, attiny3216, attiny3217.
14913 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14914 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14915 (avr_print_operand_address) [AVR_TINY]: Same.
14916 (avr_asm_init_sections) <readonly_data_section>: Only patch
14917 callback if avr_arch->flash_pm_offset = 0.
14918 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14919 for rodata if avr_arch->flash_pm_offset != 0.
14920 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14921 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14922 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14923 (m_options): Append opt_rcall.
14924 (m_dirnames): Append dir_rcall.
14925 * config/avr/t-multilib: Regenerate.
14927 * configure.ac [target=avr]: Check whether avrxmega3 default
14928 linker description file works as needed.
14929 * configure: Regenerate.
14930 * doc/avr-mmcu.texi: Regenerate.
14931 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14932 <__AVR_ARCH__>: Document avrxmega3 and 103.
14933 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14934 <__AVR_SHORT_CALLS__>: Document it.
14935 <__AVR_PM_BASE_ADDRESS__>: Document it.
14936 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14937 (AVR Variable Attributes) <progmem>: Document this is
14938 not needed for avrxmega3.
14939 (AVR Named Address Spaces) <__flash>: Dito.
14941 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14943 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14945 2017-06-12 Doug Rupp <rupp@adacore.com>
14947 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14948 Append vxworks-stdint.h to the tm_file list.
14949 * config/vxworks-stdint.h: New file.
14951 2017-06-12 Martin Liska <mliska@suse.cz>
14953 PR tree-optimization/81041
14954 * tree-profile.c (gimple_gen_ic_func_profiler):
14955 Create an extra BB in profile-generate
14956 (gimple_gen_time_profiler): Likewise.
14958 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14960 PR tree-optimization/81003
14961 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14962 (update_range_test): Use it instead of force_gimple_operand_gsi.
14964 2017-06-12 Richard Biener <rguenther@suse.de>
14966 PR tree-optimization/81053
14967 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14968 with backedge value not defined in loop. Simplify def stmt
14971 2017-06-11 Tom de Vries <tom@codesourcery.com>
14974 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14976 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14977 nvptx_cannot_force_const_mem.
14979 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14981 * opts.c (finish_options): Move test for flag_split_stack after
14982 it has been initialized.
14984 2017-06-11 Jason Merrill <jason@redhat.com>
14986 * tree.h (id_equal): New.
14987 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14988 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14989 instead of strcmp of IDENTIFIER_POINTER.
14991 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14993 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14994 (mark_all_inlined_calls_cdtor): Fix formating.
14995 (inline_transform): Rescale profile before inlining.
14997 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14999 * cgraph.h (cgraph_edge::clone): Update prototype.
15000 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15001 (cgraph_node::create_clone): Update.
15002 (cgraph_node::create_version_clone): Update.
15003 * tree-inline.c (copy_bb): Update.
15004 (expand_call_inline): Update.
15006 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15008 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15009 factored out from ...
15010 (rs6000_emit_prologue): ... here.
15012 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15014 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15015 factored out from ...
15016 (rs6000_emit_prologue): ... here.
15018 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15020 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15022 (handle_missing_profiles): Fix computation of tp_first_run.
15023 (counts_to_freqs): Do not touch freqs when count is 0.
15025 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15027 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15030 2017-06-10 Tom de Vries <tom@codesourcery.com>
15032 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15033 attributes): Document signal effective target.
15035 2017-06-10 Tom de Vries <tom@codesourcery.com>
15037 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15038 Document effective target stack_size.
15040 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15042 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15043 to the edit_context if they can be auto-applied.
15045 2017-06-9 Ian Lance Taylor <iant@golang.org>
15047 * opts.c (finish_options): If -fsplit-stack, disable implicit
15048 -forder-blocks-and-partition.
15049 * doc/invoke.texi (Optimize Options): Document that when using
15050 -fsplit-stack -forder-blocks-and-partition is not implicitly
15053 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15055 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15056 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15057 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15058 * builtins.def (abort, trap, unreachable): Declare cold.
15059 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15060 * tree-core.h (ECF_COLD): New.
15061 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15062 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15064 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15066 * predict.c (unlikely_executed_stmt_p): Cleanup.
15068 2017-06-09 Richard Biener <rguenther@suse.de>
15070 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15071 model if the ref is always written to.
15073 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15075 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15077 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15079 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15082 2017-06-09 Tom de Vries <tom@codesourcery.com>
15085 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15086 "target cannot support label values" when encountering LABEL_REF.
15088 2017-06-09 Martin Liska <mliska@suse.cz>
15090 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15091 (gimple_gen_ic_func_profiler): Emit direct comparison
15092 of __gcov_indirect_call_callee with NULL.
15093 (gimple_gen_time_profiler): Change probability from
15094 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15096 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15098 * profile.c (edge_gcov_counts): Turn to pointer.
15099 (compute_branch_probabilities, compute_branch_probabilities): Update.
15100 (branch_prob): Do not clear edge_gcov_count.
15101 * profile.h (edge_gcov_counts): Turn to pointer.
15102 (edge_gcov_count): Update.
15104 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15106 * gimple.h (gimple_check_failed): Mark cold.
15108 2017-06-09 Richard Biener <rguenther@suse.de>
15110 PR tree-optimization/66623
15111 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15112 refactor check_reduction into two parts, properly computing
15113 whether we have to check reduction validity for outer loop
15116 2017-06-09 Richard Biener <rguenther@suse.de>
15118 PR tree-optimization/79483
15119 * graphite-scop-detection.c (order): New global.
15120 (get_order): Compute bb to order mapping that satisfies code
15121 generation constraints.
15122 (cmp_pbbs): New helper.
15123 (build_scops): Start domwalk at entry block, sort generated
15126 2017-06-09 Richard Biener <rguenther@suse.de>
15128 PR middle-end/81007
15129 * ipa-polymorphic-call.c
15130 (ipa_polymorphic_call_context::restrict_to_inner_class):
15131 Skip FIELD_DECLs with error_mark_node type.
15132 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15135 2017-06-09 Martin Liska <mliska@suse.cz>
15137 * predict.c (struct branch_predictor): New struct.
15138 (test_prediction_value_range): New test.
15139 (predict_c_tests): New function.
15140 * selftest-run-tests.c (selftest::run_tests): Run the function.
15141 * selftest.h: Declare new tests.
15143 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15146 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15147 gen_add3_insn did not fail.
15148 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15149 r0, construct that number in a temporary reg and add that reg to r0.
15150 If asked to put the result in r0 as well, fail.
15152 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15154 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15155 for early expansion of vec_eqv.
15157 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15159 PR middle-end/81005
15160 * ubsan.c (instrument_null): Avoid pointless code temporary.
15161 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15163 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15167 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15170 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15173 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15175 * predict.c (unlikely_executed_edge_p): Move ahead.
15176 (probably_never_executed_edge_p): Use it.
15178 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15180 PR middle-end/79988
15181 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15182 gimple_call_builtin_p call.
15184 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15186 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15187 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15188 rtl_check_failed_type2, rtl_check_failed_code1,
15189 rtl_check_failed_code2, rtl_check_failed_code_mode,
15190 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15191 rtvec_check_failed_bounds, rtl_check_failed_flag,
15192 _fatal_insn_not_found, _fatal_insn): Likewise.
15193 * tree.h (tree_contains_struct_check_failed,
15194 tree_check_failed, tree_not_check_failed,
15195 tree_class_check_failed, tree_range_check_failed,
15196 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15197 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15198 tree_operand_check_failed, omp_clause_check_failed,
15199 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15202 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15204 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15205 flag_branch_probabilities.
15206 * ipa-inline.c (edge_badness): Likewise.
15207 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15208 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15209 * predict.c (maybe_hot_frequency_p): Likewise.
15210 (probably_never_executed): Likewise.
15211 * sched-ebb.c (schedule_ebbs): Likewise.
15212 * sched-rgn.c (find_single_block_region): Likewise.
15213 * tracer.c (tail_duplicate): Likewise.
15215 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15217 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15218 longer requires x_flag_profile_use.
15220 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15222 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15223 instead of flag_reorder_blocks_and_partition.
15224 * dbxout.c (dbxout_function_end): Likewise.
15225 * dwarf2out.c (gen_subprogram_die): Likewise.
15226 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15227 * hw-doloop.c (reorg_loops): Likewise.
15228 * varasm.c (assemble_start_function,
15229 assemble_end_function): Likewise.
15230 (decide_function_section): Do not check for
15231 flag_reorder_blocks_and_partition.
15233 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15235 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15237 (chkp_get_hard_register_fake_addr_expr): Ditto.
15238 (chkp_build_addr_expr): Add check for hard reg case.
15239 (chkp_parse_array_and_component_ref): Ditto.
15240 (chkp_find_bounds_1): Ditto.
15241 (chkp_process_stmt): Don't generate bounds store for
15244 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15246 * predict.c (maybe_hot_bb_p): Do not check profile status.
15247 (maybe_hot_edge_p): Likewise.
15248 (probably_never_executed): Check for zero counts even if profile
15250 (unlikely_executed_edge_p): New function.
15251 (unlikely_executed_stmt_p): New function.
15252 (unlikely_executed_bb_p): New function.
15253 (set_even_probabilities): Use unlikely predicates.
15254 (combine_predictions_for_bb): Likewise.
15255 (predict_paths_for_bb): Likewise.
15256 (predict_paths_leading_to_edge): Likewise.
15257 (determine_unlikely_bbs): New function.
15258 (estimate_bb_frequencies): Use it.
15259 (compute_function_frequency): Use zero counts even if profile is
15261 * profile-count.h: Fix typo.
15263 2017-08-08 Julia Koval <julia.koval@intel.com>
15265 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15266 _mm512_mask_cvtsepi16_storeu_epi8,
15267 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15268 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15269 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15270 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15271 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15272 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15273 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15274 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15275 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15276 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15277 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15278 __builtin_ia32_pmovuswb256mem_mask,
15279 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15280 __builtin_ia32_pmovwb512mem_mask): New builtins.
15282 2017-08-08 Julia Koval <julia.koval@intel.com>
15284 PR target/73350,80862
15285 * config/i386/subst.md (round): Fix round pattern.
15286 * config/i386/i386.c (ix86_erase_embedded_rounding):
15287 Fix erasing rounding for the fixed pattern.
15289 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15291 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15293 2017-06-08 Martin Liska <mliska@suse.cz>
15295 PR gcov-profile/80911
15296 * gcov.c (block_info::block_info): New constructor.
15298 2017-06-07 Carl Love <cel@us.ibm.com>
15300 * config/rs6000/rs6000-c: The return type of the following
15301 built-in functions was implemented as int not long long. Fix sign
15302 of return value for the unsigned version of vec_mulo and vec_mule.
15303 vector unsigned long long vec_bperm (vector unsigned long long,
15304 vector unsigned char)
15305 vector signed long long vec_mule (vector signed int,
15307 vector unsigned long long vec_mule (vector unsigned int,
15308 vector unsigned int)
15309 vector signed long long vec_mulo (vector signed int,
15311 vector unsigned long long vec_mulo (vector unsigned int,
15312 vector unsigned int)
15313 * doc/extend.texi: Fix the documentation for the built-in
15316 2017-06-07 Carl Love <cel@us.ibm.com>
15319 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15322 2017-06-07 Carl Love <cel@us.ibm.com>
15324 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15325 support, Generate doublehv for signed int/float for BE case only.
15327 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15329 * doc/invoke.texi (mcx16): Rewrite.
15331 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15333 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15334 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15335 *mov<mode>_softfloat, and an anonymous splitter): Use
15336 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15338 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15340 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15342 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15343 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15344 (rs6000_debug_reg_global): Adjust.
15345 (rs6000_init_hard_regno_mode_ok): Adjust.
15346 (rs6000_dbx_register_number): Adjust.
15347 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15348 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15349 Remove SPE_ACC and SPEFSCR.
15350 (REG_ALLOC_ORDER): Ditto.
15351 (FRAME_POINTER_REGNUM): Change to 111.
15352 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15353 (REG_CLASS_NAMES): Ditto.
15354 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15355 (REGISTER_NAMES): Ditto.
15356 (ADDITIONAL_REG_NAMES): Ditto.
15357 (rs6000_reg_names): Ditto.
15358 * config/rs6000/rs6000.md: Renumber some register number
15361 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15363 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15365 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15366 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15368 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15369 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15370 Delete the SPE high registers.
15371 (REG_ALLOC_ORDER): Ditto.
15372 (enum reg_class): Remove SPE_HIGH_REGS.
15373 (REG_CLASS_NAMES): Ditto.
15374 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15375 (REGISTER_NAMES): Ditto.
15376 (rs6000_reg_names): Ditto.
15377 * doc/tm.texi.in: Remove SPE as example.
15378 * doc/tm.texi: Regenerate.
15380 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15382 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15383 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15384 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15385 * config/rs6000/rs6000.md (type): Remove "brinc".
15387 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15389 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15390 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15391 * config/rs6000/linuxspe.h: Delete file.
15392 * config/rs6000/rs6000.md: Don't include spe.md.
15393 * config/rs6000/spe.h: Delete file.
15394 * config/rs6000/spe.md: Delete file.
15395 * config/rs6000/t-rs6000: Remove spe.md.
15397 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15399 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15400 (reg_or_none500mem_operand): Delete.
15401 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15402 instead of reg_or_none500mem_operand.
15404 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15406 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15407 handling of SPE flags.
15408 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15410 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15412 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15414 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15415 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15416 paired_divv2sf3): Similar.
15417 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15418 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15419 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15421 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15422 Rename the paired_* instruction patterns.
15423 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15425 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15426 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15427 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15428 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15429 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15430 PAIRED_VECTOR_MODE.
15431 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15432 (spe_func_has_64bit_regs_p): Delete.
15433 (spe_expand_predicate_builtin): Delete.
15434 (spe_expand_evsel_builtin): Delete.
15435 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15436 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15437 (invalid_e500_subreg): Delete.
15438 (rs6000_legitimize_address): Always force_reg op2 as well, for
15439 paired single memory accesses.
15440 (rs6000_member_type_forces_blk): Delete.
15441 (rs6000_spe_function_arg): Delete.
15442 (rs6000_expand_unop_builtin): Delete SPE handling.
15443 (rs6000_expand_binop_builtin): Ditto.
15444 (spe_expand_stv_builtin): Delete.
15445 (bdesc_2arg_spe): Delete.
15446 (spe_expand_builtin): Delete.
15447 (spe_expand_predicate_builtin): Delete.
15448 (spe_expand_evsel_builtin): Delete.
15449 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15450 (spe_init_builtins): Delete.
15451 (spe_func_has_64bit_regs_p): Delete.
15452 (savres_routine_name): Delete "info" parameter. Adjust callers.
15453 (rs6000_emit_stack_reset): Ditto.
15454 (rs6000_dwarf_register_span): Delete.
15455 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15456 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15457 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15459 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15461 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15462 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15463 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15464 mulv2sf3, divv2sf3): Delete expanders.
15466 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15468 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15470 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15472 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15473 * config/rs6000/rs6000.c: Ditto.
15475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15477 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15478 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15479 comparison_operator.
15481 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15483 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15484 * config/rs6000/rs6000.opt: Ditto.
15485 * config/rs6000/t-rtems: Ditto.
15487 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15489 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15490 TARGET_E500_SINGLE by 0, simplify.
15491 * config/rs6000/rs6000.c: Ditto.
15492 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15493 (spe_build_register_parallel): Delete.
15494 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15495 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15496 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15497 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15498 (E500_CONVERT): Delete.
15499 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15501 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15503 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15504 * config/rs6000/dfp.md: Ditto.
15505 (negdd2, *negdd2_fpr): Merge.
15506 (absdd2, *absdd2_fpr): Merge.
15507 (negtd2, *negtd2_fpr): Merge.
15508 (abstd2, *abstd2_fpr): Merge.
15509 * config/rs6000/e500.h: Delete file.
15510 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15511 TARGET_FPRS by 1 and simplify.
15512 * config/rs6000/rs6000-c.c: Ditto.
15513 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15514 TARGET_DF_SPE by 0.
15515 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15517 * config/rs6000/rs6000.md: Ditto.
15518 (floatdidf2, *floatdidf2_fpr): Merge.
15519 (move_from_CR_gt_bit): Delete.
15520 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15521 (E500_CR_IOR_COMPARE): Delete.
15522 (All patterns that require !TARGET_FPRS): Delete.
15523 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15525 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15527 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15529 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15531 * graphds.c (add_edge): Intitialize edge's attached data.
15532 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15533 pointer parameter. Call pointed function on each edge during
15534 graph traversing. Skip traversing the edge when the function
15536 (graphds_dfs, graphds_scc): Ditto.
15537 (for_each_edge): New parameter. Pass the new parameter to callback
15539 * graphds.h (skip_edge_callback): New function pointer type.
15540 (graphds_dfs, graphds_scc): New function pointer parameter.
15541 (graphds_edge_callback, for_each_edge): New parameter.
15543 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15545 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15546 out code checking if runtime alias check is possible to below ...
15547 Call the new function.
15548 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15549 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15551 2017-06-07 Marek Polacek <polacek@redhat.com>
15554 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15555 TYPE_OVERFLOW_WRAPS check.
15557 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15559 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15560 if versioning is required.
15561 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15562 peeling with the check for versioning.
15564 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15566 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15567 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15568 Set true to new parameter if new ssa variable is defined.
15569 (vect_gen_vector_loop_niters): Refactor. Set range information
15570 for the new vector loop bound variable.
15571 (vect_do_peeling): Ditto.
15573 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15575 * tree-affine.c (ssa.h): Include header file.
15576 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15577 has wrapping overflow behavior.
15579 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15581 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15583 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15585 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15586 (tree_to_aff_combination): ... here.
15588 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15590 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15591 reg_pressure model function.
15592 (ivopts_global_cost_for_size): Delete.
15593 (determine_set_costs, iv_ca_recount_cost): Call new model function
15594 ivopts_estimate_reg_pressure.
15596 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15598 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15599 expensive than udiv. Remove floating point cases from mod.
15601 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15603 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15604 Increase idiv cost.
15606 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15608 * config/aarch64/aarch64.md
15609 (copysignsf3): Fix mask generation.
15611 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15613 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15615 (class dump_manager): Add register_dumps method.
15616 * dumpfile.c: Include langhooks.h.
15617 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15618 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15619 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15620 (dump_manager::dump_register): Start with 512 entries instead of 32.
15621 (dump_manager::register_dumps): New method.
15622 * toplev.c (general_init): Instead of invoking register_dumps
15623 langhook, invoke register_dumps method on the dump manager.
15624 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15627 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15629 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15630 Say that # requires an associated define_split to exist, and that
15631 the define_split must be suitable for use after register allocation.
15633 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15635 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15636 (compute_outgoing_frequencies): Also initialize zero counts.
15637 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15638 around loops; preserve more of profile when nothing changes.
15640 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15642 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15644 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15645 * config/arm/arm-cpu-cdata.h: Regenerate.
15646 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15647 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15648 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15649 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15650 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15652 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15654 * config/arm/t-rmprofile: Likewise.
15655 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15657 2017-06-06 David S. Miller <davem@davemloft.net>
15660 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15661 blockage if function uses alloca.
15663 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15665 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15666 New "uid" fields to hold pretty-print IDs of group and ref.
15667 Memory references are now identified as <group_id>:<ref_id>
15668 instead of using [random] addresses.
15669 (dump_mem_details): Simplify, no functional change.
15670 (dump_mem_ref): Simplify and make output more concise.
15671 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15672 (find_or_create_group): Initialize group uid.
15673 (record_ref): Initialize ref uid. Improve debug output.
15674 (prune_group_by_reuse, should_issue_prefetch_p,)
15675 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15676 (mark_nontemporal_store, determine_loop_nest_reuse):
15677 Improve debug output.
15679 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15681 * dbgcnt.def (prefetch): New debug counter.
15682 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15683 (schedule_prefetches): Stop issueing prefetches if debug counter
15686 2017-06-06 Tom de Vries <tom@codesourcery.com>
15688 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15689 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15691 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15693 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15694 Use aarch64_reg_or_zero predicate for operand 4.
15695 (aarch64_compare_and_swap<mode> define_insn_and_split):
15696 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15697 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15699 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15701 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15702 (arm_compute_save_core_reg_mask): This.
15703 (thumb1_compute_save_reg_mask): Rename into ...
15704 (thumb1_compute_save_core_reg_mask): This.
15705 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15706 (arm_compute_frame_layout): Likewise.
15708 2017-06-06 Richard Biener <rguenther@suse.de>
15710 PR tree-optimization/80974
15711 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15712 keep or clear leaders SSA info.
15714 2017-06-06 Tom de Vries <tom@codesourcery.com>
15716 * config/nvptx/nvptx.c (split_mode_p): New function.
15717 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15719 2017-06-06 Tom de Vries <tom@codesourcery.com>
15721 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15723 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15726 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15729 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15731 * shrink-wrap.c (handle_simple_exit): Update profile.
15732 (try_shrink_wrapping): Upate profile.
15734 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15736 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15737 (tree_guess_outgoing_edge_probabilities): New.
15738 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15739 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15741 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15743 * ipa-split.c (split_function): Initialize return bb profile.
15745 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15747 * profile.c (compute_branch_probabilities): Also initialize
15748 EXIT_BLOCK profile.
15750 2017-06-06 Richard Biener <rguenther@suse.de>
15752 PR tree-optimization/80928
15753 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15754 (vect_analyze_loop_operations): Properly guard analysis for
15756 (vect_transform_loop): Likewise.
15757 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15758 (vect_model_induction_cost): Do not cost for pure SLP.
15759 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15760 of induction in inner loop vectorization.
15761 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15762 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15763 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15765 (vect_analyze_slp_cost_1): Cost induction.
15766 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15767 (vect_get_slp_vect_defs): Likewise.
15768 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15769 (vect_transform_stmt): Handle SLP reductions.
15770 * tree-vectorizer.h (vectorizable_induction): Adjust.
15772 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15774 * config/rs6000/rs6000.c (make_resolver_func): Update
15775 init_lowered_empty_function call.
15777 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15779 * doc/invoke.texi: Document the -fprofile-abs-path option.
15780 * common.opt (fprofile-abs-path): New option.
15781 * gcov-io.h (gcov_write_filename): Declare.
15782 * gcov-io.c (gcov_write_filename): New function.
15783 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15784 * profile.c (output_location): Likewise.
15786 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15788 * shring-wrap.c: Revert accidental commit.
15790 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15792 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15794 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15796 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15798 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15799 profile in callgraph edge.
15800 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15801 (apply_scale): Likewise.
15802 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15803 Also scale profile when inlining function with zero profile.
15804 (initialize_cfun): Update exit block profile even when it is zero.
15805 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15806 when profile is read.
15808 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15810 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15811 (CLONE_*): New constants to define the processors we can generate
15812 code for with the target_clone attribute.
15813 (rs6000_clone_map): New array to identify which clone processors
15814 the current program is running on.
15815 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15816 target_clone attribute.
15817 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15818 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15819 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15820 (cpu_expand_builtin): Add support for target_clone attribute.
15821 (rs6000_valid_attribute_p): Allow "default" attribute.
15822 (get_decl_name): New debug function to simplify printing the
15823 current function name in debugging statements.
15824 (rs6000_clone_priority): New functions to support the target_clone
15825 attribute, and be able to generate code to switch between ISA 2.05
15826 through ISA 3.0 (power6 through power9).
15827 (rs6000_compare_version_priority): Likewise.
15828 (rs6000_get_function_versions_dispatcher): Likewise.
15829 (make_resolver_func): Likewise.
15830 (add_condition_to_bb): Likewise.
15831 (dispatch_function_versions): Likewise.
15832 (rs6000_generate_version_dispatcher_body): Likewise.
15833 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15834 (fusion_gpr_load_p): Fix a spacing issue.
15835 * doc/extend.texi (Common Function Attributes): Document that the
15836 PowerPC supports the target_clone attribute.
15838 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15840 * config/arm/arm.h: explain F symbol found in description of ARM
15841 register allocation in its legend.
15843 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15845 * config/mips/frame-header-opt.c: Include profile-count.h.
15846 * config/riscv/riscv.c: Include profile-count.h
15848 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15850 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15852 (sm_set_flag_if_changed): Add bbs field.
15853 (execute_sm_if_changed_flag_set): Pass BBS.
15854 (execute_sm): Update.
15856 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15858 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15861 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15863 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15864 (peephole2): New peephole2 to emit the above.
15865 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15867 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15869 * config/aarch64/aarch64.c (define_peephole2 above
15870 *sub_<shift>_<mode>): New peephole.
15872 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15874 * config/i386/i386.c (make_resolver_func): Update.
15875 * Makefile.in: Add profile-count.h and profile-count.o
15876 * auto-profile.c (afdo_indirect_call): Update to new API.
15877 (afdo_set_bb_count): Update.
15878 (afdo_propagate_edge): Update.
15879 (afdo_propagate_circuit): Update.
15880 (afdo_calculate_branch_prob): Update.
15881 (afdo_annotate_cfg): Update.
15882 * basic-block.h: Include profile-count.h
15883 (struct edge_def): Turn count to profile_count.
15884 (struct basic_block_def): Likewie.
15885 (REG_BR_PROB_BASE): Move to profile-count.h
15886 (RDIV): Move to profile-count.h
15887 * bb-reorder.c (max_entry_count): Turn to profile_count.
15888 (find_traces): Update.
15889 (rotate_loop):Update.
15890 (connect_traces):Update.
15891 (sanitize_hot_paths):Update.
15892 * bt-load.c (migrate_btr_defs): Update.
15893 * cfg.c (RDIV): Remove.
15894 (init_flow): Use alloc_block.
15895 (alloc_block): Uninitialize count.
15896 (unchecked_make_edge): Uninitialize count.
15897 (check_bb_profile): Update.
15898 (dump_edge_info): Update.
15899 (dump_bb_info): Update.
15900 (update_bb_profile_for_threading): Update.
15901 (scale_bbs_frequencies_int): Update.
15902 (scale_bbs_frequencies_gcov_type): Update.
15903 (scale_bbs_frequencies_profile_count): New.
15904 * cfg.h (update_bb_profile_for_threading): Update.
15905 (scale_bbs_frequencies_profile_count): Declare.
15906 * cfgbuild.c (compute_outgoing_frequencies): Update.
15907 (find_many_sub_basic_blocks): Update.
15908 * cfgcleanup.c (try_forward_edges): Update.
15909 (try_crossjump_to_edge): Update.
15910 * cfgexpand.c (expand_gimple_tailcall): Update.
15911 (construct_exit_block): Update.
15912 * cfghooks.c (verify_flow_info): Update.
15913 (dump_bb_for_graph): Update.
15914 (split_edge): Update.
15915 (make_forwarder_block): Update.
15916 (duplicate_block): Update.
15917 (account_profile_record): Update.
15918 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15919 (get_estimated_loop_iterations): Update.
15920 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15921 (single_likely_exit): Update.
15922 * cfgloopmanip.c (scale_loop_profile): Update.
15924 (set_zero_probability): Update.
15925 (lv_adjust_loop_entry_edge): Update.
15926 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15927 (purge_dead_edges): Update.
15928 (rtl_account_profile_record): Update.
15929 * cgraph.c (cgraph_node::create): Uninitialize count.
15930 (symbol_table::create_edge): Uninitialize count.
15931 (cgraph_update_edges_for_call_stmt_node): Update.
15932 (cgraph_edge::dump_edge_flags): Update.
15933 (cgraph_node::dump): Update.
15934 (cgraph_edge::maybe_hot_p): Update.
15935 * cgraph.h: Include profile-count.h
15936 (create_clone), create_edge, create_indirect_edge): Update.
15937 (cgraph_node): Turn count to profile_count.
15938 (cgraph_edge0: Likewise.
15939 (make_speculative, clone): Update.
15940 (create_edge): Update.
15941 (init_lowered_empty_function): Update.
15942 * cgraphclones.c (cgraph_edge::clone): Update.
15943 (duplicate_thunk_for_node): Update.
15944 (cgraph_node::create_clone): Update.
15945 * cgraphunit.c (cgraph_node::analyze): Update.
15946 (cgraph_node::expand_thunk): Update.
15947 * final.c (dump_basic_block_info): Update.
15948 * gimple-streamer-in.c (input_bb): Update.
15949 * gimple-streamer-out.c (output_bb): Update.
15950 * graphite.c (print_global_statistics): Update.
15951 (print_graphite_scop_statistics): Update.
15952 * hsa-brig.c: Include basic-block.h.
15953 * hsa-dump.c: Include basic-block.h.
15954 * hsa-gen.c (T sum_slice): Update.
15955 (convert_switch_statements):Update.
15956 * hsa-regalloc.c: Include basic-block.h.
15957 * ipa-chkp.c (chkp_produce_thunks): Update.
15958 * ipa-cp.c (struct caller_statistics): Update.
15959 (init_caller_stats): Update.
15960 (gather_caller_stats): Update.
15961 (ipcp_cloning_candidate_p): Update.
15962 (good_cloning_opportunity_p): Update.
15963 (get_info_about_necessary_edges): Update.
15964 (dump_profile_updates): Update.
15965 (update_profiling_info): Update.
15966 (update_specialized_profile): Update.
15967 (perhaps_add_new_callers): Update.
15968 (decide_about_value): Update.
15969 (ipa_cp_c_finalize): Update.
15970 * ipa-devirt.c (struct odr_type_warn_count): Update.
15971 (struct decl_warn_count): Update.
15972 (struct final_warning_record): Update.
15973 (possible_polymorphic_call_targets): Update.
15974 (ipa_devirt): Update.
15975 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15976 * ipa-icf.c (sem_function::merge): Update.
15977 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15978 * ipa-inline.c (compute_uninlined_call_time): Update.
15979 (compute_inlined_call_time): Update.
15980 (want_inline_small_function_p): Update.
15981 (want_inline_self_recursive_call_p): Update.
15982 (edge_badness): Update.
15983 (lookup_recursive_calls): Update.
15984 (recursive_inlining): Update.
15985 (inline_small_functions): Update.
15986 (dump_overall_stats): Update.
15987 (dump_inline_stats): Update.
15988 * ipa-profile.c (ipa_profile_generate_summary): Update.
15989 (ipa_propagate_frequency): Update.
15990 (ipa_profile): Update.
15991 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15992 * ipa-utils.c (ipa_merge_profiles): Update.
15993 * loop-doloop.c (doloop_modify): Update.
15994 * loop-unroll.c (report_unroll): Update.
15995 (unroll_loop_runtime_iterations): Update.
15996 * lto-cgraph.c (lto_output_edge): Update.
15997 (lto_output_node): Update.
15998 (input_node): Update.
15999 (input_edge): Update.
16000 (merge_profile_summaries): Update.
16001 * lto-streamer-in.c (input_cfg): Update.
16002 * lto-streamer-out.c (output_cfg): Update.
16003 * mcf.c (create_fixup_graph): Update.
16004 (adjust_cfg_counts): Update.
16005 (sum_edge_counts): Update.
16006 * modulo-sched.c (sms_schedule): Update.
16007 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16008 * predict.c (maybe_hot_count_p): Update.
16009 (probably_never_executed): Update.
16010 (dump_prediction): Update.
16011 (combine_predictions_for_bb): Update.
16012 (propagate_freq): Update.
16013 (handle_missing_profiles): Update.
16014 (counts_to_freqs): Update.
16015 (rebuild_frequencies): Update.
16016 (force_edge_cold): Update.
16017 * predict.h: Include profile-count.h
16018 (maybe_hot_count_p, counts_to_freqs): UPdate.
16019 * print-rtl-function.c: Do not include cfg.h
16020 * print-rtl.c: Include basic-block.h
16021 * profile-count.c: New file.
16022 * profile-count.h: New file.
16023 * profile.c (is_edge_inconsistent): Update.
16024 (correct_negative_edge_counts): Update.
16025 (is_inconsistent): Update.
16026 (set_bb_counts): Update.
16027 (read_profile_edge_counts): Update.
16028 (compute_frequency_overlap): Update.
16029 (compute_branch_probabilities): Update; Initialize and deinitialize
16031 (branch_prob): Update.
16032 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16033 (edge_gcov_count): New.
16034 (bb_gcov_count): New.
16035 * shrink-wrap.c (try_shrink_wrapping): Update.
16036 * tracer.c (better_p): Update.
16037 * trans-mem.c (expand_transaction): Update.
16038 (ipa_tm_insert_irr_call): Update.
16039 (ipa_tm_insert_gettmclone_call): Update.
16040 * tree-call-cdce.c: Update.
16041 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16042 (gimple_duplicate_sese_tail): Update.
16043 (gimple_account_profile_record): Update.
16044 (execute_fixup_cfg): Update.
16045 * tree-inline.c (copy_bb): Update.
16046 (copy_edges_for_bb): Update.
16047 (initialize_cfun): Update.
16048 (freqs_to_counts): Update.
16049 (copy_cfg_body): Update.
16050 (expand_call_inline): Update.
16051 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16052 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16053 (try_unroll_loop_completely): Update.
16054 (try_peel_loop): Update.
16055 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16056 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16057 * tree-ssa-loop-split.c (connect_loops): Update.
16058 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16059 * tree-ssa-reassoc.c (branch_fixup): Update.
16060 * tree-ssa-tail-merge.c (replace_block_by): Update.
16061 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16062 (compute_path_counts): Update.
16063 (update_profile): Update.
16064 (recompute_probabilities): Update.
16065 (update_joiner_offpath_counts): Update.
16066 (estimated_freqs_path): Update.
16067 (freqs_to_counts_path): Update.
16068 (clear_counts_path): Update.
16069 (ssa_fix_duplicate_block_edges): Update.
16070 (duplicate_thread_path): Update.
16071 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16072 (struct switch_conv_info): Update.
16073 * tree-tailcall.c (decrease_profile): Update.
16074 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16075 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16076 * value-prof.c (check_counter): Update.
16077 (gimple_divmod_fixed_value): Update.
16078 (gimple_mod_pow2): Update.
16079 (gimple_mod_subtract): Update.
16080 (gimple_ic_transform): Update.
16081 (gimple_stringop_fixed_value): Update.
16082 * value-prof.h (gimple_ic): Update.
16084 2017-06-02 Carl Love <cel@us.ibm.com>
16086 * config/rs6000/rs6000-c: Add support for built-in functions
16087 vector double vec_doublee (vector signed int);
16088 vector double vec_doublee (vector unsigned int);
16089 vector double vec_doublee (vector float);
16090 vector double vec_doubleh (vector signed int);
16091 vector double vec_doubleh (vector unsigned int);
16092 vector double vec_doubleh (vector float);
16093 vector double vec_doublel (vector signed int);
16094 vector double vec_doublel (vector unsigned int);
16095 vector double vec_doublel (vector float);
16096 vector double vec_doubleo (vector signed int);
16097 vector double vec_doubleo (vector unsigned int);
16098 vector double vec_doubleo (vector float);.
16099 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16100 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16102 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16103 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16104 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16106 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16107 vec_doublel, vec_doubleh.
16108 * doc/extend.texi: Update the built-in documentation file for the
16109 new built-in functions.
16111 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16114 * ipa-inline-analysis.c (free_growth_caches): Set
16115 edge_removal_hook_holder to NULL after removing it.
16117 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16119 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16120 comparision with zero.
16122 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16123 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16124 for early expansion of vec_min and vec_max builtins.
16125 (builtin_function_type): Add min/max unsigned variants to those
16126 identified as having unsigned arguments.
16128 2017-06-02 Olivier Hainque <hainque@adacore.com>
16130 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16132 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16134 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16135 Use VALL_F16 iterator rather than VALL.
16137 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16139 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16140 Emit CBNZ inside loop when doing a strong exchange and comparing
16141 against zero. Generate the CC flags after the loop.
16143 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16145 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16146 (dl_section_ref): New.
16147 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16148 On AIX, append an expression to subtract the size of the
16149 section length to dl_section_ref.
16151 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16153 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16154 for early expansion of vector absolute builtins.
16156 2017-06-02 Richard Biener <rguenther@suse.de>
16158 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16159 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16161 2017-06-02 Richard Biener <rguenther@suse.de>
16163 PR tree-optimization/80948
16164 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16165 stmt order as well.
16167 2017-06-02 Richard Biener <rguenther@suse.de>
16169 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16171 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16172 for inductions as relevant.
16174 2017-06-02 Richard Biener <rguenther@suse.de>
16176 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16177 (vectorizable_induction): ... this. Remove dead code.
16179 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16181 * builtins. (expand_builtin_alloca): Remove second parameter and
16182 infer its value from the first parameter instead.
16183 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16185 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16187 PR rtl-optimization/80903
16188 * loop-doloop.c (add_test): Unshare sequence.
16190 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16192 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16194 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16196 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16198 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16199 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16200 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16201 (xlogue_layout::get_stub_rtx): Make static.
16202 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16203 (xlogue_layout::compute_stub_managed_regs): Rename to...
16204 (xlogue_layout::count_stub_managed_regs): ...this.
16205 (xlogue_layout::is_stub_managed_reg): New function.
16206 (xlogue_layout::m_stub_names): Rename to...
16207 (xlogue_layout::s_stub_names): ...this, make static.
16208 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16209 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16210 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16211 xlogue_layout::s_stub_names): Instantiate statics.
16212 (stub_managed_regs): Remove.
16213 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16214 (disable_call_ms2sysv_xlogues): Rename to...
16215 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16216 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16218 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16219 change after reload_completed.
16220 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16222 (ix86_expand_prologue): Likewise.
16223 (ix86_expand_epilogue): Likewise.
16224 (ix86_expand_split_stack_prologue): Likewise.
16225 (ix86_compute_frame_layout): Remove frame parameter ...
16226 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16227 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16229 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16230 (ix86_frame): Move from here ...
16231 * config/i386/i386.h (ix86_frame): ... to here.
16232 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16233 complete ix86_frame data structure instead. Remove some_ld_name.
16235 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16237 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16238 symbols that hold a DECL_VALUE_EXPR.
16240 2017-06-01 Martin Jambor <mjambor@suse.cz>
16242 PR tree-optimization/80898
16243 * tree-sra.c (process_subtree_disqualification): Removed.
16244 (disqualify_candidate): Do not acll
16245 process_subtree_disqualification.
16246 (subtree_mark_written_and_enqueue): New function.
16247 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16248 RHS has been disqualified and re-queue LHS if necessary. Apart
16249 from that, ignore disqualified RHS.
16251 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16253 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16254 address fetch for z10 or later.
16256 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16258 * config/arc/arc.md (tst_movb): Add guard when splitting.
16260 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16262 * config/arc/arc.c (arc_can_eliminate): Test against
16263 arc_frame_pointer_needed.
16265 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16267 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16268 to prevent store reordering.
16269 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16270 (type): Add block type.
16271 (stack_tie): Define special instruction to be used in
16274 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16276 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16277 constraint. It is not valid for the pattern.
16278 (noncommutative_binary_comparison): Likewise.
16280 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16282 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16285 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16287 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16288 be used by the reg-alloc.
16290 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16292 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16293 reg-alloc when having mul64 or mul32x16 instructions.
16294 (mulsidi3): Likewise.
16295 (umulsidi3): Likewise.
16296 (mulsi32x16): New pattern.
16297 (mulsi64): Likewise.
16298 (mulsidi64): Likewise.
16299 (umulsidi64): Likewise.
16300 (MUL32x16_REG): Define.
16301 (mul64_600): Use MUL32x16_REG.
16302 (mac64_600): Likewise.
16303 (umul64_600): Likewise.
16304 (umac64_600): Likewise.
16306 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16308 * config/arc/arc.md (mulsi3_700): Make it commutative.
16310 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16312 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16314 (*sign_extendsidi2_insn): Likewise for movstosw.
16316 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16318 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16319 the type of the input discriminant value. Convert the
16320 discriminant value of signedness vary.
16322 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16324 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16325 Add to -Wall section.
16327 2017-06-01 Richard Biener <rguenther@suse.de>
16329 PR middle-end/66313
16330 * fold-const.c (fold_plusminus_mult_expr): If the factored
16331 factor may be zero use a wrapping type for the inner operation.
16332 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16333 and handle moved defs.
16334 (process_assignment): Properly guard the unary op case. Return a
16335 tri-state indicating that moving the stmt before the call may allow
16336 to continue. Pass through to_move.
16337 (find_tail_calls): Handle moving unrelated defs before
16340 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16343 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16344 splitter result in the canonical way.
16346 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16348 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16349 also for 32bit target. Update insn attributes.
16350 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16352 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16354 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16355 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16356 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16357 (_mm_maskz_min_ss): New intrinsics.
16359 2017-05-31 Martin Liska <mliska@suse.cz>
16361 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16362 Change comment style to one we normally use.
16363 (vectorizable_reduction): Likewise.
16364 (vectorizable_induction): Likewise.
16365 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16366 (vectorizable_call): Likewise.
16367 (vectorizable_simd_clone_call): Likewise.
16368 (vectorizable_conversion): Likewise.
16369 (vectorizable_assignment): Likewise.
16370 (vectorizable_shift): Likewise.
16371 (vectorizable_operation): Likewise.
16372 (vectorizable_store): Likewise.
16373 (vectorizable_load): Likewise.
16374 * tree-vectorizer.h: Likewise.
16376 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16378 * passes.c (emergency_dump_function): New.
16379 * tree-pass.h (emergency_dump_function): Declare.
16380 * plugin.c (plugins_internal_error_function): Remove.
16381 * plugin.h (plugins_internal_error_function): Remove declaration.
16382 * toplev.c (internal_error_function): New static function. Use it...
16383 (general_init): ...here.
16385 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16387 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16388 (arc_rtx_costs): Add costs for new patterns.
16389 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16390 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16392 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16394 * tree-ssa-strlen.c (get_next_strinfo): New function.
16395 (get_stridx_plus_constant): Use it.
16396 (zero_length_string): Likewise.
16397 (adjust_related_strinfos): Likewise.
16398 (adjust_last_stmt): Likewise.
16400 2017-05-31 Richard Biener <rguenther@suse.de>
16403 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16404 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16406 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16408 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16409 loop_vinfo argument and use of dependence distance vectors.
16410 Check instead whether the two references differ only in their
16411 initial value and assume that they have the same alignment if the
16412 difference is a multiple of the vector alignment.
16413 (vect_analyze_data_refs_alignment): Update call accordingly.
16415 2017-05-31 Martin Liska <mliska@suse.cz>
16418 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16420 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16422 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16423 (create_intersect_range_checks): Move from ...
16424 * tree-data-ref.c (create_intersect_range_checks_index)
16425 (create_intersect_range_checks): ... to here.
16426 (create_runtime_alias_checks): New function factored from ...
16427 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16428 here. Call above function.
16429 * tree-data-ref.h (create_runtime_alias_checks): New function.
16431 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16433 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16434 segment length for dr_b and compute it in wide_int.
16436 2017-05-31 Richard Biener <rguenther@suse.de>
16438 PR tree-optimization/80906
16439 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16440 and pass through iv_map.
16441 (copy_bb_and_scalar_dependences): Adjust.
16442 (translate_pending_phi_nodes): Likewise.
16443 (copy_loop_close_phi_args): Handle code-generating IVs instead
16446 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16448 * diagnostic-color.c (color_dict): Add "type-diff".
16449 (parse_gcc_colors): Update comment.
16450 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16451 -fdiagnostics-show-template-tree and -fno-elide-type.
16452 (GCC_COLORS): Add type-diff to example.
16454 (-fdiagnostics-show-template-tree): New.
16455 (-fno-elide-type): New.
16456 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16457 the pp_format_decoder callback. Call any m_format_postprocessor's
16459 (pretty_printer::pretty_printer): Initialize
16460 m_format_postprocessor.
16461 (pretty_printer::~pretty_printer): Delete any
16462 m_format_postprocessor.
16463 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16464 (class format_postprocessor): New class.
16465 (struct pretty_printer::format_decoder): Document the new parameters.
16466 (struct pretty_printer::m_format_postprocessor): New field.
16467 * tree-diagnostic.c (default_tree_printer): Update for new
16468 bool and const char ** params.
16469 * tree-diagnostic.h (default_tree_printer): Likewise.
16471 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16473 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16474 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16475 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16476 rs6000_gen_cell_microcode code.
16477 (rs6000_final_prescan_insn): Delete.
16478 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16479 "warn-cell-microcode" entries.
16480 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16481 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16482 throughout. Change cc_reg_not_micro_cr0_operand to
16483 cc_reg_not_cr0_operand throughout.
16484 (*extendhi<mode>2_noload): Delete.
16485 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16486 (mwarn-cell-microcode): Delete.
16487 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16488 -mgen-cell-microcode and -mwarn-cell-microcode.
16490 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16493 * config/i386/constraints.md (Yd): New constraint.
16495 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16496 and (?Yd, r) alternatives. Update insn attributes.
16497 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16498 and (?*Yd, r) alternatives. Update insn attributes.
16499 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16501 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16503 * gimplify.c (gimplify_modify_expr): Don't create a
16504 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16507 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16509 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16511 2017-05-30 Richard Biener <rguenther@suse.de>
16513 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16514 and reduc_def fields.
16515 (STMT_VINFO_REDUC_TYPE): New define.
16516 (STMT_VINFO_REDUC_DEF): Likewise.
16517 (vect_force_simple_reduction): Adjust prototype.
16518 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16519 (vect_is_simple_reduction): Remove check_reduction argument.
16520 (vect_force_simple_reduction): Adjust and set
16521 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16522 (vectorizable_reduction): Do not re-do reduction analysis
16523 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16524 * tree-parloops.c (gather_scalar_reductions): Adjust.
16526 2017-05-30 Richard Biener <rguenther@suse.de>
16528 PR middle-end/80901
16529 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16532 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16534 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16535 Introduce unknown_misalignment parameter and remove vf.
16536 (vect_peeling_hash_get_lowest_cost):
16537 Pass unknown_misalignment parameter.
16538 (vect_enhance_data_refs_alignment):
16539 Fix unsupportable data ref treatment.
16541 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16543 * tree-vect-data-refs.c (vect_get_data_access_cost):
16544 Workaround for SLP handling.
16545 (vect_enhance_data_refs_alignment):
16546 Compute costs for doing no peeling at all, compare to the best
16547 peeling costs so far and avoid peeling if cheaper.
16549 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16551 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16552 Return peeling info and set costs to zero for unlimited cost
16554 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16555 with unknown misalignment. Compute and costs for unknown
16556 misalignment, compare them to the costs for known misalignment
16557 and choose the cheapest for peeling.
16559 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16561 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16562 (vect_get_peeling_costs_all_drs): Create function.
16563 (vect_peeling_hash_get_lowest_cost):
16564 Use vect_get_peeling_costs_all_drs.
16565 (vect_peeling_supportable): Create function.
16566 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16568 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16570 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16571 DR_HAS_NEGATIVE_STEP.
16572 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16573 (vect_enhance_data_refs_alignment): Use.
16574 (vect_duplicate_ssa_name_ptr_info): Use.
16575 * tree-vectorizer.h (dr_misalignment): Use.
16576 (known_alignment_for_access_p): Use.
16578 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16581 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16583 (has_section_name): New function.
16585 2017-05-30 Martin Liska <mliska@suse.cz>
16588 * auto-profile.c (get_function_decl_from_block): Fix
16591 2017-05-30 Richard Biener <rguenther@suse.de>
16593 PR middle-end/80876
16594 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16596 2017-05-30 Martin Liska <mliska@suse.cz>
16598 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16599 * dumpfile.h (struct dump_file_info): Remove ctors.
16601 2017-05-30 Martin Liska <mliska@suse.cz>
16603 * predict.def: Fix GNU coding style.
16605 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16607 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16608 Mark 'to' argument with ATTRIBUTE_UNUSED.
16610 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16612 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16613 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16614 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16617 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16619 * doc/install.texi (Options specification): Restore entry of
16620 --enable-sjlj-exceptions.
16622 2017-05-27 Michael Eager <eager@eagercon.com>
16625 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16627 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16629 * config/microblaze/microblaze.h
16630 (FIXED_REGISTERS): Update in macro.
16631 (CALL_USED_REGISTERS): Update in macro.
16633 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16635 * doc/install.texi: Add links to macOS binary distributions.
16637 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16641 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16643 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16645 2017-05-26 Martin Liska <mliska@suse.cz>
16647 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16649 2017-05-26 Martin Liska <mliska@suse.cz>
16651 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16652 always leading ';; '.
16653 (dump_bb_info): Likewise.
16654 (brief_dump_cfg): Likewise.
16655 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16656 * dumpfile.c: Remove usage of TDF_VERBOSE.
16657 * dumpfile.h (enum dump_kind): Likewise.
16658 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16659 * print-tree.c (debug_verbose): Remove.
16660 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16661 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16662 * tree-diagnostic.c (default_tree_printer): Replace
16663 TDF_DIAGNOSTIC with TDF_SLIM.
16665 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16667 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16668 in parameter loop, rather than loop_vinfo.
16669 (create_intersect_range_checks): Ditto.
16670 (vect_create_cond_for_alias_checks): Update call to above functions.
16672 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16674 PR tree-optimization/80815
16675 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16676 for merging runtime alias checks. Handle negative DR_STEPs.
16678 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16680 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16682 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16683 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16684 out code pruning runtime alias checks.
16685 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16686 factored out from above.
16687 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16689 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16691 (prune_runtime_alias_test_list): New decalaration.
16693 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16695 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16696 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16697 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16698 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16699 (operator==, comp_dr_with_seg_len_pair): Ditto.
16700 (vect_prune_runtime_alias_test_list): Ditto.
16702 2017-05-26 Martin Liska <mliska@suse.cz>
16705 * params.def: Bound partial-inlining-entry-probability param.
16707 2017-05-26 Marek Polacek <polacek@redhat.com>
16710 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16713 2017-05-26 Richard Biener <rguenther@suse.de>
16715 PR tree-optimization/80842
16716 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16719 2017-05-26 Richard Biener <rguenther@suse.de>
16721 PR tree-optimization/80844
16722 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16724 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16726 * doc/md.texi (Machine Constraints): Update x86 family
16727 machine constraints section to match 'config/i386/constraints.md'.
16729 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16731 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16733 2017-05-25 Nathan Sidwell <nathan@acm.org>
16735 * doc/invoke.texi (--enable-languages): Update documentation.
16737 2017-05-25 Martin Liska <mliska@suse.cz>
16739 * dumpfile.c: Add TDF_FOLDING.
16740 * dumpfile.h (enum dump_kind): Likewise.
16741 * genmatch.c (dt_simplify::gen_1): Use it.
16743 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16745 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16747 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16749 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16750 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16752 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16754 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16755 * match.pd (X == C): Rewrite it here.
16756 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16757 with_certain_nonzero_bits2): New predicates.
16758 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16760 2017-05-24 Nathan Sidwell <nathan@acm.org>
16762 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16765 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16768 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16770 * config/powerpcspe: New port. Files are copied from the rs6000
16771 port, with "rs6000" in filenames replaced by "powerpcspe".
16773 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16775 PR rtl-optimization/80754
16776 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16778 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16780 * config/sparc/sparc.md (length): Return the correct value for -mflat
16781 sibcalls to match output_sibcall.
16783 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16787 * config/rs6000/rs6000.c (struct machine_function): Add new field
16789 (rs6000_get_separate_components): Init that field, use it.
16790 (rs6000_components_for_bb): Use the field.
16792 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16794 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16796 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16798 PR middle-end/80823
16799 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16801 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16804 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16805 address against address_operand predicate.
16806 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16808 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16810 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16811 parameters passed indirectly.
16813 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16815 * config/i386/i386.md (*movdi_internal): Remove SSE4
16816 alternative 18 (?r, *v). Update insn attributes.
16817 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16818 Update insn attributes.
16819 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16820 Update insn attributes.
16821 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16822 alternative 1 (r, v). Remove isa attribute.
16823 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16824 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16825 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16827 2017-05-23 Tom de Vries <tom@codesourcery.com>
16829 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16832 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16834 * cgraphunit.c (symbol_table::process_new_functions): Update.
16835 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16836 (inline_generate_summary): Rename to ...
16837 (ipa_fn_summary_generate): ... this one.
16838 (inline_read_summary): Rename to ...
16839 (ipa_fn_summary_read): ... this one.
16840 (inline_write_summary): Rename to ...
16841 (ipa_fn_summary_write): ... this one.
16842 (inline_free_summary): Rename to ...
16843 (ipa_free_fn_summary): ... this one.
16844 (pass_data_local_fn_summary, pass_local_fn_summary,
16845 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16846 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16847 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16848 make_pass_ipa_fn_summary): New.
16849 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16850 inline_write_summary, inline_free_summary): Remove.
16851 (ipa_free_fn_summary) : New.
16852 * ipa-inline.c (ipa_inline): Update.
16853 (pass_ipa_inline): Do not generate summaries.
16854 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16856 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16857 and add pass_ipa_fn_summary.
16858 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16860 (make_pass_inline_parameters): Remove.
16862 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16864 * omp-low.c (struct omp_context): Remove "default_kind" member.
16867 * omp-offload.c (execute_oacc_device_lower): Remove the
16868 parallelism dimensions function attributes for unparallelized
16869 OpenACC kernels constructs.
16871 2017-05-23 Martin Liska <mliska@suse.cz>
16873 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16875 (cgraph_edge::make_speculative): Likewise.
16876 (cgraph_edge::resolve_speculation): Likewise.
16877 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16878 (cgraph_node::dump): Likewise.
16879 * cgraph.h: Likewise.
16880 * cgraphunit.c (analyze_functions): Likewise.
16881 (symbol_table::compile): Likewise.
16882 * ipa-cp.c (print_all_lattices): Likewise.
16883 (determine_versionability): Likewise.
16884 (initialize_node_lattices): Likewise.
16885 (ipcp_verify_propagated_values): Likewise.
16886 (estimate_local_effects): Likewise.
16887 (update_profiling_info): Likewise.
16888 (create_specialized_node): Likewise.
16889 (perhaps_add_new_callers): Likewise.
16890 (decide_about_value): Likewise.
16891 (decide_whether_version_node): Likewise.
16892 (identify_dead_nodes): Likewise.
16893 (ipcp_store_bits_results): Likewise.
16894 * ipa-devirt.c (dump_targets): Likewise.
16895 (ipa_devirt): Likewise.
16896 * ipa-icf.c (sem_item::dump): Likewise.
16897 (sem_function::equals): Likewise.
16898 (sem_variable::equals): Likewise.
16899 (sem_item_optimizer::read_section): Likewise.
16900 (sem_item_optimizer::execute): Likewise.
16901 (congruence_class::dump): Likewise.
16902 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16903 (dump_inline_summary): Likewise.
16904 (estimate_node_size_and_time): Likewise.
16905 (inline_analyze_function): Likewise.
16906 * ipa-inline-transform.c (inline_call): Likewise.
16907 * ipa-inline.c (report_inline_failed_reason): Likewise.
16908 (want_early_inline_function_p): Likewise.
16909 (edge_badness): Likewise.
16910 (update_edge_key): Likewise.
16911 (inline_small_functions): Likewise.
16912 * ipa-profile.c (ipa_profile): Likewise.
16913 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16914 (ipa_make_edge_direct_to_target): Likewise.
16915 (remove_described_reference): Likewise.
16916 (ipa_impossible_devirt_target): Likewise.
16917 (propagate_controlled_uses): Likewise.
16918 (ipa_print_node_params): Likewise.
16919 (ipcp_transform_function): Likewise.
16920 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16921 (propagate_pure_const): Likewise.
16922 * ipa-reference.c (generate_summary): Likewise.
16923 (read_write_all_from_decl): Likewise.
16924 (propagate): Likewise.
16925 (ipa_reference_read_optimization_summary): Likewise.
16926 * ipa-utils.c (ipa_merge_profiles): Likewise.
16927 * ipa.c (walk_polymorphic_call_targets): Likewise.
16928 (symbol_table::remove_unreachable_nodes): Likewise.
16929 (ipa_single_use): Likewise.
16930 * passes.c (execute_todo): Likewise.
16931 * predict.c (drop_profile): Likewise.
16932 * symtab.c (symtab_node::get_dump_name): New function.
16933 (symtab_node::dump_name): Likewise.
16934 (symtab_node::dump_asm_name): Likewise.
16935 (symtab_node::dump_references): Likewise.
16936 (symtab_node::dump_referring): Likewise.
16937 (symtab_node::dump_base): Likewise.
16938 (symtab_node::debug_symtab): Likewise.
16939 * tree-sra.c (convert_callers_for_node): Likewise.
16940 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16941 * value-prof.c (init_node_map): Likewise.
16943 2017-05-23 Martin Liska <mliska@suse.cz>
16945 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16946 and symtab_node::debug_symtab to symbol_table::debug.
16947 * cgraphunit.c (analyze_functions): Use the renamed function.
16948 (symbol_table::compile): Likewise.
16949 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16950 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16951 * passes.c (execute_todo): Likewise.
16952 * symtab.c (symbol_table::dump): New function.
16953 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16955 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16957 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16958 that nonconst implies exec.
16960 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16962 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16963 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16964 (inline_edge_summary_vec): Turn into ...
16965 (ipa_call_summaries): ... this one.
16966 (redirect_to_unreachable, edge_set_predicate,
16967 evaluate_properties_for_edge, inline_summary_alloc,
16968 reset_ipa_call_summary, reset_inline_summary,
16969 inline_summary_t::duplicate): Update.
16970 (inline_edge_duplication_hook): Turn to ...
16971 (ipa_call_summary_t::duplicate): ... this one.
16972 (inline_edge_removal_hook): Turn to ...
16973 (ipa_call_summary_t::remove): ... this one.
16974 (dump_inline_edge_summary): Turn to ...
16975 (dump_ipa_call_summary): ... this one.
16976 (estimate_function_body_sizes): Update.
16977 (inline_update_callee_summaries): Update.
16978 (remap_edge_change_prob): Update.
16979 (remap_edge_summaries): Update.
16980 (inline_merge_summary): Update.
16981 (do_estimate_edge_time): Update.
16982 (inline_generate_summary): Update.
16983 (inline_read_section): Update.
16984 (inline_read_summary): Update.
16985 (inline_free_summary): Update.
16986 * ipa-inline.c (can_inline_edge_p): Update.
16987 (compute_inlined_call_time): Update.
16988 (want_inline_small_function_p): Update.
16989 (edge_badness): Update.
16990 (early_inliner): Update.
16991 * ipa-inline.h (inline_edge_summary): Turn to ...
16992 (ipa_call_summary): ... this one.
16993 (ipa_call_summary_t): New class.
16994 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16995 (ipa_call_summaries): New.
16996 (inline_edge_summary): Remove.
16997 (estimate_edge_growth): Update.
16998 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16999 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17000 * ipa-split.c (execute_split_functions): Update.
17001 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17003 2017-05-23 Tom de Vries <tom@codesourcery.com>
17005 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17006 attributes): Document rdrand effective target.
17008 2017-05-23 Tom de Vries <tom@codesourcery.com>
17010 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17011 attributes): Sort alphabetically.
17013 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17015 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17017 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17020 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17021 V2DF/V2DI splat into two separate patterns, one that handles
17022 registers, and the other that only handles memory. Drop support
17023 for splatting from a GPR on ISA 2.07 and then splitting the
17024 splat into direct move and splat.
17025 (vsx_splat_<mode>_reg): Likewise.
17026 (vsx_splat_<mode>_mem): Likewise.
17028 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17030 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17032 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17034 PR middle-end/80809
17035 * omp-low.c (finish_taskreg_remap): New function.
17036 (finish_taskreg_scan): If unit size of ctx->record_type
17037 is non-constant, unshare the size expression and replace
17038 decls in it with possible outer var refs.
17040 PR middle-end/80809
17041 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17042 GOVD_SHARED rather than GOVD_PRIVATE with it.
17043 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17044 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17046 PR middle-end/80853
17047 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17048 as last argument to build_outer_var_ref for pointer bases of array
17049 section reductions.
17051 2017-05-19 Martin Sebor <msebor@redhat.com>
17053 * print-tree.c (print_node): Print DECL_READ_P flag.
17055 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17057 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17058 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17059 * cgraph.c: Likewise.
17060 * cgraphunit.c: Likewise.
17061 * gengtype.c: Likewise.
17062 * ipa-cp.c: Likewise.
17063 * ipa-devirt.c: Likewise.
17064 * ipa-icf.c: Likewise.
17065 * ipa-predicate.c: Likewise.
17066 * ipa-profile.c: Likewise.
17067 * ipa-prop.c: Likewise.
17068 * ipa-split.c: Likewise.
17070 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17071 edge_predicate_pool, dump_inline_hints,
17072 inline_summary::account_size_time, redirect_to_unreachable,
17073 edge_set_predicate, set_hint_predicate,
17074 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17075 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17076 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17077 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17078 ipa_call_summary_t::remove, initialize_growth_caches,
17079 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17080 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17081 mark_modified, unmodified_parm_1, unmodified_parm,
17082 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17083 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17084 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17085 will_be_nonconstant_predicate, record_modified_bb_info,
17086 get_minimal_bb, record_modified, param_change_prob,
17087 phi_result_unknown_predicate, predicate_for_phi_result,
17088 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17089 estimate_function_body_sizes, compute_inline_parameters,
17090 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17091 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17092 inline_update_callee_summaries, remap_edge_change_prob,
17093 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17094 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17095 inline_analyze_function, inline_summary_t::insert,
17096 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17097 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17098 inline_free_summary): Move to ipa-fnsummary.h
17099 (predicate_t): Remove.
17100 * ipa-fnsummary.c: New file.
17101 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17102 (enum inline_hints_vals, inline_hints, agg_position_info,
17103 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17104 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17105 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17106 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17107 inline_read_summary, inline_write_summary, inline_free_summary,
17108 inline_analyze_function, initialize_inline_failed,
17109 inline_merge_summary, inline_update_overall_summary,
17110 compute_inline_parameters): Move to ipa-fnsummary.h
17111 * ipa-fnsummary.h: New file.
17112 * ipa-inline-transform.h: Include ipa-inline.h.
17113 * ipa-inline.c: LIkewise.
17115 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17117 * ipa-inline.c (edge_badness): Use inlined_time instead of
17118 inline_summaries->get.
17120 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17122 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17124 2017-05-22 Nathan Sidwell <nathan@acm.org>
17126 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17127 (fdump-lang): Document 'raw' option.
17128 * dumpfile.h (TDI_tu): Delete.
17129 * dumpfile.c (dump_files): Remove translation-unit.
17130 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17132 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17134 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17135 command option from $(AWK) call.
17136 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17137 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17138 [FORMAT]: Remove handling of variable.
17139 * config/avr/t-multilib: Regenerate.
17141 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17143 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17145 (dump_inline_summary): Do not print self_time.
17146 (estimate_function_body_sizes): Do not set self_time.
17147 (compute_inline_parameters): Likewise.
17148 (inline_read_section, inline_write_summary): Do not stream self_time.
17149 * ipa-inline.h (inline_summary): Drop self_time.
17151 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17153 * ipa-inline-analysis.c (account_size_time): Rename to ...
17154 (inline_summary::account_size_time): ... this one.
17155 (reset_ipa_call_summary): Turn to ...
17156 (ipa_call_summary::reset): ... this one.
17157 (reset_inline_summary): Turn to ...
17158 (inline_summary::reset): ... this one.
17159 (inline_summary_t::remove): Update.
17160 (inline_summary_t::duplicate): Update.
17161 (ipa_call_summary_t::remove): Update.
17162 (dump_inline_summary): Update.
17163 (estimate_function_body_sizes): Update.
17164 (compute_inline_parameters): Update.
17165 (estimate_node_size_and_time): Update.
17166 (inline_merge_summary): Update.
17167 (inline_update_overall_summary): Update.
17168 (inline_read_section): Update.
17169 (inline_write_summary): Update.
17170 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17171 add account_size_time and reset member functions.
17172 (ipa_call_summary): Add reset function.
17173 * ipa-predicate.h (predicate::operator &): Constify.
17175 2017-05-22 Richard Biener <rguenther@suse.de>
17177 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17179 2017-05-19 Jason Merrill <jason@redhat.com>
17181 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17183 2017-05-19 Marek Polacek <polacek@redhat.com>
17186 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17187 TYPE_OVERFLOW_WRAPS checks.
17189 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17191 * tree-core.h (enum omp_clause_default_kind): Add
17192 "OMP_CLAUSE_DEFAULT_PRESENT".
17193 * tree-pretty-print.c (dump_omp_clause): Handle it.
17194 * gimplify.c (enum gimplify_omp_var_data): Add
17195 "GOVD_MAP_FORCE_PRESENT".
17196 (gimplify_adjust_omp_clauses_1): Map it to
17197 "GOMP_MAP_FORCE_PRESENT".
17198 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17200 * gimplify.c (oacc_default_clause): Clarify.
17202 2017-05-19 Nathan Sidwell <nathan@acm.org>
17204 LANG_HOOK_REGISTER_DUMPS
17205 * toplev.c (general_init): Call register dump lang hook.
17206 * doc/invoke.texi: Document -fdump-lang option family.
17207 * dumpfile.c (dump_files): Remove class dump here.
17208 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17209 * dumpfile.h (tree_dump_index): Remove TDI_class.
17210 * langhooks-def.h (lhd_register_dumps): Declare.
17211 (LANG_HOOKS_REGISTER_DUMPS): Define.
17212 (LANG_HOOKS_INITIALIZER): Add it.
17213 * langhooks.c (lhd_register_dumps): Define.
17214 * langhooks.h (struct lang_hooks): Add register_dumps.
17216 2017-05-19 Nathan Sidwell <nathan@acm.org>
17218 * context.h (context::set_passes): New.
17219 * context.c (context::context): Do not create pass manager.
17220 * toplev.c (general_init): Create pass manager here.
17222 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17224 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17225 use this splitter if two add or or instructions would also work for
17226 the constant we want to generate.
17228 2017-05-19 Richard Biener <rguenther@suse.de>
17231 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17232 predicate evaluation.
17234 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17236 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17238 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17241 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17243 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17244 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17245 (inline_edge_summary_vec): Turn into ...
17246 (ipa_call_summaries): ... this one.
17247 (redirect_to_unreachable, edge_set_predicate,
17248 evaluate_properties_for_edge, inline_summary_alloc,
17249 reset_ipa_call_summary, reset_inline_summary,
17250 inline_summary_t::duplicate): Update.
17251 (inline_edge_duplication_hook): Turn to ...
17252 (ipa_call_summary_t::duplicate): ... this one.
17253 (inline_edge_removal_hook): Turn to ...
17254 (ipa_call_summary_t::remove): ... this one.
17255 (dump_inline_edge_summary): Turn to ...
17256 (dump_ipa_call_summary): ... this one.
17257 (estimate_function_body_sizes): Update.
17258 (inline_update_callee_summaries): Update.
17259 (remap_edge_change_prob): Update.
17260 (remap_edge_summaries): Update.
17261 (inline_merge_summary): Update.
17262 (do_estimate_edge_time): Update.
17263 (inline_generate_summary): Update.
17264 (inline_read_section): Update.
17265 (inline_read_summary): Update.
17266 (inline_free_summary): Update.
17267 * ipa-inline.c (can_inline_edge_p): Update.
17268 (compute_inlined_call_time): Update.
17269 (want_inline_small_function_p): Update.
17270 (edge_badness): Update.
17271 (early_inliner): Update.
17272 * ipa-inline.h (inline_edge_summary): Turn to ...
17273 (ipa_call_summary): ... this one.
17274 (ipa_call_summary_t): New class.
17275 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17276 (ipa_call_summaries): New.
17277 (inline_edge_summary): Remove.
17278 (estimate_edge_growth): Update.
17279 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17280 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17281 * ipa-split.c (execute_split_functions): Update.
17282 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17284 2017-05-19 Richard Biener <rguenther@suse.de>
17286 PR middle-end/80764
17287 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17289 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17291 * config/rs6000/rs6000.c (struct machine_function): Add field
17292 fpr_is_wrapped_separately.
17293 (rs6000_get_separate_components): Use 64 components. Handle the
17294 new FPR components.
17295 (rs6000_components_for_bb): Handle the FPR components.
17296 (rs6000_emit_prologue_components): Handle the FPR components.
17297 (rs6000_emit_epilogue_components): Handle the FPR components.
17298 (rs6000_set_handled_components): Handle the FPR components.
17299 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17300 that are already separately shrink-wrapped.
17301 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17302 that are already separately shrink-wrapped.
17304 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17307 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17310 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17311 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17312 cases where the register allocator uses a move and an offsettable
17313 memory operation to/from a FPR register on ISA 2.06/2.07.
17314 (define_peephole2 for Altivec d-form store): Likewise.
17316 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17319 * config/i386/mmx.md (*mov<mode>_internal): Enable
17320 alternatives 11, 12, 13 and 14 also for 32bit targets.
17321 Remove alternatives 15, 16, 17 and 18.
17322 * config/i386/sse.md (vec_concatv2di): Change
17323 alternative (!x, *y) to (x, ?!*Yn).
17325 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17327 * dumpfile.h (enum dump_kind): Remove stray comma.
17329 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17331 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17332 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17333 predicate::num_conditions
17334 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17335 (CHANGED): turn into predicate::changed.
17336 (agg_position_info): Move to ipa-predicate.h
17337 (add_condition, predicate::add_clause, predicate::operator &=,
17338 predicate::or_with, predicate::evaluate, predicate::probability,
17339 dump_condition, dump_clause, predicate::dump,
17340 predicate::remap_after_duplication, predicate::remap_after_inlining,
17341 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17342 (evaluate_conditions_for_known_args): Update.
17343 (set_cond_stmt_execution_predicate): Update.
17344 * ipa-inline.h: Include ipa-predicate.h
17345 (condition, inline_param_summary, conditions, agg_position_info,
17346 predicate): Move to ipa-predicate.h
17347 * ipa-predicate.c: New file.
17348 * ipa-predicate.h: New file.
17350 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17352 * final.c (leaf_function_p): Check we are not in a sequence.
17354 2017-05-18 Martin Liska <mliska@suse.cz>
17356 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17357 * dumpfile.c (dump_register): Use new enum dump_kind.
17358 (get_dump_file_name): Likewise.
17359 (dump_enable_all): Likewise.
17360 (dump_switch_p_1): Likewise.
17361 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17362 * dumpfile.h (enum dump_kind): New enum type.
17363 (struct dump_file_info): Create constructor and
17364 format fields and comments.
17365 * passes.c (pass_manager::register_one_dump_file):
17367 * statistics.c (statistics_early_init): Likewise.
17368 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17369 TDF_TREE with TDF_SLIM.
17370 (gather_memory_references_ref): Likewise.
17372 2017-05-18 Martin Liska <mliska@suse.cz>
17374 * vec.h (struct vnull): Use it.
17376 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17378 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17379 (true_predicate, false_predicate, true_predicate_p,
17380 false_predicate_p): Remove.
17381 (single_cond_predicate, not_inlined_predicate): Turn to member function
17383 (add_condition): Update.
17384 (add_clause): Turn to...
17385 (predicate::add_clause): ... this one; update; allow passing NULL
17387 (and_predicates): Turn to ...
17388 (predicate::operator &=): ... this one.
17389 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17390 (or_predicates): Turn to ...
17391 (predicate::or_with): ... this one.
17392 (evaluate_predicate): Turn to ...
17393 (predicate::evaluate): ... this one.
17394 (predicate_probability): Turn to ...
17395 (predicate::probability): ... this one.
17396 (dump_condition): Update.
17397 (dump_predicate): Turn to ...
17398 (predicate::dump): ... this one.
17399 (account_size_time): Update.
17400 (edge_set_predicate): Update.
17401 (set_hint_predicate): UPdate.
17402 (evaluate_conditions_for_known_args): Update.
17403 (evaluate_properties_for_edge): Update.
17404 (remap_predicate_after_duplication): Turn to...
17405 (predicate::remap_after_duplication): ... this one.
17406 (remap_hint_predicate_after_duplication): Update.
17407 (inline_summary_t::duplicate): UPdate.
17408 (dump_inline_edge_summary): Update.
17409 (dump_inline_summary): Update.
17410 (set_cond_stmt_execution_predicate): Update.
17411 (set_switch_stmt_execution_predicate): Update.
17412 (compute_bb_predicates): Update.
17413 (will_be_nonconstant_expr_predicate): Update.
17414 (will_be_nonconstant_predicate): Update.
17415 (phi_result_unknown_predicate): Update.
17416 (predicate_for_phi_result): Update.
17417 (array_index_predicate): Update.
17418 (estimate_function_body_sizes): Update.
17419 (estimate_node_size_and_time): Update.
17420 (estimate_ipcp_clone_size_and_time): Update.
17421 (remap_predicate): Rename to ...
17422 (predicate::remap_after_inlining): ... this one.
17423 (remap_hint_predicate): Update.
17424 (inline_merge_summary): Update.
17425 (inline_update_overall_summary): Update.
17426 (estimate_size_after_inlining): Update.
17427 (read_predicate): Rename to ...
17428 (predicate::stream_in): ... this one.
17429 (read_inline_edge_summary): Update.
17430 (write_predicate): Rename to ...
17431 (predicate::stream_out): ... this one.
17432 (write_inline_edge_summary): Update.
17433 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17434 (clause_t): Turn to uint32_t
17435 (predicate): Turn to class; implement constructor and operators
17437 (size_time_entry): Update.
17438 (inline_summary): Update.
17439 (inline_edge_summary): Update.
17441 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17443 * fold-const.c (fold_binary_loc): Move transformation...
17444 * match.pd (C - X CMP X): ... here.
17446 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17448 * config/sparc/sparc.c (sparc_option_override): Set function
17449 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17450 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17452 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17454 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17456 2017-05-18 Marek Polacek <polacek@redhat.com>
17459 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17460 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17461 gimple_assign_load_p.
17463 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17465 PR middle-end/80692
17466 * real.c (do_compare): Give decimal_do_compare preference over
17467 comparing just the signs.
17469 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17471 * doc/md.texi (Canonicalization of Instructions): Describe the
17472 canonical form of instructions that inherently set a condition
17475 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17477 PR middle-end/80775
17478 * tree-cfg.c: Move deletion of unreachable case statements to after
17479 the merging of consecutive case labels.
17481 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17483 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17484 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17485 restoring of callee-saved registers.
17487 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17489 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17490 * config/visium/visium.c (single_set_and_flags): Likewise.
17491 * config/visium/visium.md (Substitutions): Likewise.
17493 2017-05-17 Martin Liska <mliska@suse.cz>
17495 * cfg.c: Introduce dump_flags_t type and
17496 use it instead of int type.
17498 * cfghooks.c: Likewise.
17499 * cfghooks.h (struct cfg_hooks): Likewise.
17500 * cfgrtl.c: Likewise.
17501 * cfgrtl.h: Likewise.
17502 * cgraph.c (cgraph_node::get_body): Likewise.
17503 * coretypes.h: Likewise.
17504 * domwalk.c: Likewise.
17505 * domwalk.h: Likewise.
17506 * dumpfile.c (struct dump_option_value_info): Likewise.
17507 (dump_enable_all): Likewise.
17508 (dump_switch_p_1): Likewise.
17509 (opt_info_switch_p): Likewise.
17510 * dumpfile.h (enum tree_dump_index): Likewise.
17511 (struct dump_file_info): Likewise.
17512 * genemit.c: Likewise.
17513 * generic-match-head.c: Likewise.
17514 * gengtype.c (open_base_files): Likewise.
17515 * gimple-pretty-print.c: Likewise.
17516 * gimple-pretty-print.h: Likewise.
17517 * graph.c (print_graph_cfg): Likewise.
17518 * graphite-scop-detection.c (dot_all_sese): Likewise.
17519 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17520 * loop-unroll.c (report_unroll): Likewise.
17521 * passes.c (pass_manager::register_one_dump_file): Likewise.
17522 * print-tree.c: Likewise.
17523 * statistics.c: Likewise.
17524 * tree-cfg.c: Likewise.
17525 * tree-cfg.h: Likewise.
17526 * tree-dfa.c: Likewise.
17527 * tree-dfa.h: Likewise.
17528 * tree-dump.c (dump_function): Likewise.
17529 * tree-dump.h (struct dump_info): Likewise.
17530 * tree-pretty-print.c: Likewise.
17531 * tree-pretty-print.h: Likewise.
17532 * tree-ssa-live.c: Likewise.
17533 * tree-ssa-live.h: Likewise.
17534 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17535 * tree-vect-loop.c: Likewise.
17536 * tree-vect-slp.c: Likewise.
17538 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17539 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17541 PR tree-optimization/80457
17542 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17543 of all arguments to a statement as scalar_to_vec operations.
17544 (vectorizable_call): Adjust call to vect_model_simple_cost for
17546 (vectorizable_conversion): Likewise.
17547 (vectorizable_assignment): Likewise.
17548 (vectorizable_shift): Likewise.
17549 (vectorizable_operation): Likewise.
17550 (vectorizable_comparison): Likewise.
17551 (vect_is_simple_cond): Record the def types for operands.
17552 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17553 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17554 for statement argument count.
17556 2017-05-16 Carl Love <cel@us.ibm.com>
17558 * config/rs6000/rs6000-c: Add support for built-in functions
17559 vector unsigned long long vec_bperm (vector unsigned long long,
17560 vector unsigned char)
17561 vector signed long long vec_mule (vector signed int,
17563 vector unsigned long long vec_mule (vector unsigned int,
17564 vector unsigned int)
17565 vector signed long long vec_mulo (vector signed int,
17567 vector unsigned long long vec_mulo (vector unsigned int,
17568 vector unsigned int)
17569 vector signed char vec_sldw (vector signed char,
17570 vector signed char,
17572 vector unsigned char vec_sldw (vector unsigned char,
17573 vector unsigned char,
17575 vector signed short vec_sldw (vector signed short,
17576 vector signed short,
17578 vector unsigned short vec_sldw (vector unsigned short,
17579 vector unsigned short,
17581 vector signed int vec_sldw (vector signed int,
17584 vector unsigned int vec_sldw (vector unsigned int,
17585 vector unsigned int,
17587 vector signed long long vec_sldw (vector signed long long,
17588 vector signed long long,
17590 vector unsigned long long vec_sldw (vector unsigned long long,
17591 vector unsigned long long,
17593 * config/rs6000/rs6000-c: Add support for built-in functions
17594 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17595 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17596 * doc/extend.texi: Update the built-in documentation for the
17597 new built-in functions.
17599 2017-05-16 Marek Polacek <polacek@redhat.com>
17603 * tree.c (save_expr): Don't fold the expression.
17605 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17607 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17608 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17609 and (?*y,m). Update insn attributes.
17611 2017-05-16 Martin Liska <mliska@suse.cz>
17613 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17614 flags argument of print_gimple_stmt, print_gimple_expr,
17615 print_generic_stmt and print_generic_expr.
17616 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17617 * coretypes.h: Likewise.
17618 * except.c (dump_eh_tree): Likewise.
17619 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17620 * gimple-pretty-print.h: Likewise.
17621 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17622 (backprop::push_to_worklist): Likewise.
17623 (backprop::pop_from_worklist): Likewise.
17624 (backprop::process_use): Likewise.
17625 (backprop::intersect_uses): Likewise.
17626 (note_replacement): Likewise.
17627 * gimple-ssa-store-merging.c
17628 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17629 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17630 (pass_store_merging::execute): Likewise.
17631 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17632 (ssa_base_cand_dump_callback): Likewise.
17633 (dump_incr_vec): Likewise.
17634 (replace_refs): Likewise.
17635 (replace_mult_candidate): Likewise.
17636 (create_add_on_incoming_edge): Likewise.
17637 (create_phi_basis): Likewise.
17638 (insert_initializers): Likewise.
17639 (all_phi_incrs_profitable): Likewise.
17640 (introduce_cast_before_cand): Likewise.
17641 (replace_one_candidate): Likewise.
17642 * gimplify.c (gimplify_expr): Likewise.
17643 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17644 (set_rename): Likewise.
17645 (rename_uses): Likewise.
17646 (copy_loop_phi_nodes): Likewise.
17647 (add_close_phis_to_merge_points): Likewise.
17648 (copy_loop_close_phi_args): Likewise.
17649 (copy_cond_phi_args): Likewise.
17650 (graphite_copy_stmts_from_block): Likewise.
17651 (translate_pending_phi_nodes): Likewise.
17652 * graphite-poly.c (print_pdr): Likewise.
17653 (dump_gbb_cases): Likewise.
17654 (dump_gbb_conditions): Likewise.
17655 (print_scop_params): Likewise.
17656 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17657 (build_cross_bb_scalars_use): Likewise.
17658 (gather_bbs::before_dom_children): Likewise.
17659 * hsa-dump.c (dump_hsa_immed): Likewise.
17660 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17661 (get_replacement_map): Likewise.
17662 * ipa-inline-analysis.c (dump_condition): Likewise.
17663 (estimate_function_body_sizes): Likewise.
17664 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17665 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17666 * ipa-prop.c (ipa_dump_param): Likewise.
17667 (ipa_print_node_jump_functions_for_edge): Likewise.
17668 (ipa_modify_call_arguments): Likewise.
17669 (ipa_modify_expr): Likewise.
17670 (ipa_dump_param_adjustments): Likewise.
17671 (ipa_dump_agg_replacement_values): Likewise.
17672 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17673 * ipa-pure-const.c (check_stmt): Likewise.
17674 (pass_nothrow::execute): Likewise.
17675 * ipa-split.c (execute_split_functions): Likewise.
17676 * omp-offload.c (dump_oacc_loop_part): Likewise.
17677 (dump_oacc_loop): Likewise.
17678 * trans-mem.c (tm_log_emit): Likewise.
17679 (tm_memopt_accumulate_memops): Likewise.
17680 (dump_tm_memopt_set): Likewise.
17681 (dump_tm_memopt_transform): Likewise.
17682 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17683 (print_loop): Likewise.
17684 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17685 (chkp_gather_checks_info): Likewise.
17686 (chkp_get_check_result): Likewise.
17687 (chkp_remove_check_if_pass): Likewise.
17688 (chkp_use_outer_bounds_if_possible): Likewise.
17689 (chkp_reduce_bounds_lifetime): Likewise.
17690 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17691 (chkp_mark_completed_bounds): Likewise.
17692 (chkp_register_incomplete_bounds): Likewise.
17693 (chkp_mark_invalid_bounds): Likewise.
17694 (chkp_maybe_copy_and_register_bounds): Likewise.
17695 (chkp_build_returned_bound): Likewise.
17696 (chkp_get_bound_for_parm): Likewise.
17697 (chkp_build_bndldx): Likewise.
17698 (chkp_get_bounds_by_definition): Likewise.
17699 (chkp_generate_extern_var_bounds): Likewise.
17700 (chkp_get_bounds_for_decl_addr): Likewise.
17701 * tree-chrec.c (chrec_apply): Likewise.
17702 * tree-data-ref.c (dump_data_reference): Likewise.
17703 (dump_subscript): Likewise.
17704 (dump_data_dependence_relation): Likewise.
17705 (analyze_overlapping_iterations): Likewise.
17706 * tree-inline.c (expand_call_inline): Likewise.
17707 (tree_function_versioning): Likewise.
17708 * tree-into-ssa.c (dump_defs_stack): Likewise.
17709 (dump_currdefs): Likewise.
17710 (dump_names_replaced_by): Likewise.
17711 (dump_update_ssa): Likewise.
17712 (update_ssa): Likewise.
17713 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17714 * tree-parloops.c (build_new_reduction): Likewise.
17715 (try_create_reduction_list): Likewise.
17716 (ref_conflicts_with_region): Likewise.
17717 (oacc_entry_exit_ok_1): Likewise.
17718 (oacc_entry_exit_single_gang): Likewise.
17719 * tree-pretty-print.h: Likewise.
17720 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17721 (get_scalar_evolution): Likewise.
17722 (add_to_evolution): Likewise.
17723 (get_loop_exit_condition): Likewise.
17724 (analyze_evolution_in_loop): Likewise.
17725 (analyze_initial_condition): Likewise.
17726 (analyze_scalar_evolution): Likewise.
17727 (instantiate_scev): Likewise.
17728 (number_of_latch_executions): Likewise.
17729 (gather_chrec_stats): Likewise.
17730 (final_value_replacement_loop): Likewise.
17731 (scev_const_prop): Likewise.
17732 * tree-sra.c (dump_access): Likewise.
17733 (disqualify_candidate): Likewise.
17734 (create_access): Likewise.
17735 (reject): Likewise.
17736 (maybe_add_sra_candidate): Likewise.
17737 (create_access_replacement): Likewise.
17738 (analyze_access_subtree): Likewise.
17739 (analyze_all_variable_accesses): Likewise.
17740 (sra_modify_assign): Likewise.
17741 (initialize_constant_pool_replacements): Likewise.
17742 (find_param_candidates): Likewise.
17743 (decide_one_param_reduction): Likewise.
17744 (replace_removed_params_ssa_names): Likewise.
17745 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17746 * tree-ssa-copy.c (dump_copy_of): Likewise.
17747 (copy_prop_visit_cond_stmt): Likewise.
17748 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17749 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17750 (record_equivalences_from_stmt): Likewise.
17751 * tree-ssa-dse.c (compute_trims): Likewise.
17752 (delete_dead_call): Likewise.
17753 (delete_dead_assignment): Likewise.
17754 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17755 (forward_propagate_into_cond): Likewise.
17756 (pass_forwprop::execute): Likewise.
17757 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17758 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17760 (move_computations_worker): Likewise.
17761 (execute_sm): Likewise.
17762 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17763 (remove_exits_and_undefined_stmts): Likewise.
17764 (remove_redundant_iv_tests): Likewise.
17765 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17766 (adjust_iv_update_pos): Likewise.
17767 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17768 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17769 (value_replacement): Likewise.
17770 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17771 * tree-ssa-pre.c (print_pre_expr): Likewise.
17772 (get_representative_for): Likewise.
17773 (create_expression_by_pieces): Likewise.
17774 (insert_into_preds_of_block): Likewise.
17775 (eliminate_insert): Likewise.
17776 (eliminate_dom_walker::before_dom_children): Likewise.
17777 (eliminate): Likewise.
17778 (remove_dead_inserted_code): Likewise.
17779 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17780 * tree-ssa-reassoc.c (get_rank): Likewise.
17781 (eliminate_duplicate_pair): Likewise.
17782 (eliminate_plus_minus_pair): Likewise.
17783 (eliminate_not_pairs): Likewise.
17784 (undistribute_ops_list): Likewise.
17785 (eliminate_redundant_comparison): Likewise.
17786 (update_range_test): Likewise.
17787 (optimize_range_tests_var_bound): Likewise.
17788 (optimize_vec_cond_expr): Likewise.
17789 (rewrite_expr_tree): Likewise.
17790 (rewrite_expr_tree_parallel): Likewise.
17791 (linearize_expr): Likewise.
17792 (break_up_subtract): Likewise.
17793 (linearize_expr_tree): Likewise.
17794 (attempt_builtin_powi): Likewise.
17795 (attempt_builtin_copysign): Likewise.
17796 (transform_stmt_to_copy): Likewise.
17797 (transform_stmt_to_multiply): Likewise.
17798 (dump_ops_vector): Likewise.
17799 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17800 (print_scc): Likewise.
17801 (set_ssa_val_to): Likewise.
17802 (visit_reference_op_store): Likewise.
17803 (visit_use): Likewise.
17804 (sccvn_dom_walker::before_dom_children): Likewise.
17805 (run_scc_vn): Likewise.
17806 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17808 (expr_hash_elt::print): Likewise.
17809 (const_and_copies::pop_to_marker): Likewise.
17810 (const_and_copies::record_const_or_copy_raw): Likewise.
17811 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17812 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17813 (dump_predicates): Likewise.
17814 (find_uninit_use): Likewise.
17815 (warn_uninitialized_phi): Likewise.
17816 (pass_late_warn_uninitialized::execute): Likewise.
17817 * tree-ssa.c (verify_vssa): Likewise.
17818 (verify_ssa): Likewise.
17819 (maybe_optimize_var): Likewise.
17820 * tree-vrp.c (dump_value_range): Likewise.
17821 (dump_all_value_ranges): Likewise.
17822 (dump_asserts_for): Likewise.
17823 (register_edge_assert_for_2): Likewise.
17824 (vrp_visit_cond_stmt): Likewise.
17825 (vrp_visit_switch_stmt): Likewise.
17826 (vrp_visit_stmt): Likewise.
17827 (vrp_visit_phi_node): Likewise.
17828 (simplify_cond_using_ranges_1): Likewise.
17829 (fold_predicate_in): Likewise.
17830 (evrp_dom_walker::before_dom_children): Likewise.
17831 (evrp_dom_walker::push_value_range): Likewise.
17832 (evrp_dom_walker::pop_value_range): Likewise.
17833 (execute_early_vrp): Likewise.
17835 2017-05-16 Richard Biener <rguenther@suse.de>
17837 * dwarf2out.c (loc_list_from_tree_1): Do not create
17838 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17840 2017-05-16 Richard Biener <rguenther@suse.de>
17842 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17844 (note_variable_value_in_expr): If we resolved the decl ref
17845 do not push to the stack.
17847 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17849 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17850 operations in fast-math mode.
17851 (vaddq_f16): Likewise.
17852 (vmul_f16): Likewise.
17853 (vmulq_f16): Likewise.
17854 (vsub_f16): Likewise.
17855 (vsubq_f16): Likewise.
17856 * config/arm/neon.md (add<mode>3): New.
17858 (fma:<VH:mode>3): New. Also remove outdated comment.
17861 2017-05-16 Martin Liska <mliska@suse.cz>
17865 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17866 (odr_types_equivalent_p): Likewise.
17868 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17870 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17872 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17875 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17876 non-interunit SSE move alternatives with '?'.
17877 (zero-extendsidi peephole2): New peephole to skip intermediate
17878 general register in SSE zero-extend sequence.
17880 2017-05-15 Jeff Law <law@redhat.com>
17882 * reorg.c (relax_delay_slots): Create a new variable to hold
17883 the temporary target rather than clobbering TARGET_LABEL.
17885 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17886 missing argument to extract_bit_field call.
17887 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17889 2017-05-15 Martin Liska <mliska@suse.cz>
17892 * gcc.c (process_command): Do not allow empty argument of -o option.
17894 2017-05-15 Renlin Li <renlin.li@arm.com>
17896 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17897 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17898 * config/aarch64/constraints.md (Usf): Add long call check.
17899 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17900 (call_value): Likewise.
17901 (sibcall): Likewise.
17902 (sibcall_value): Likewise.
17904 (call_value_insn): New.
17905 (sibcall_insn): Update rtx pattern.
17906 (sibcall_value_insn): Likewise.
17907 (call_internal): Remove.
17908 (call_value_internal): Likewise.
17909 (sibcall_internal): Likewise.
17910 (sibcall_value_internal): Likewise.
17911 (call_reg): Likewise.
17912 (call_symbol): Likewise.
17913 (call_value_reg): Likewise.
17914 (call_value_symbol): Likewise.
17916 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17919 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17921 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17923 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17924 compatible with CCGOCmode and with CCZmode.
17926 2017-05-14 Martin Sebor <msebor@redhat.com>
17928 PR middle-end/77671
17929 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17930 (gimple_fold_builtin_snprintf): Same.
17931 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17932 (gimple_fold_builtin_snprintf): Same.
17933 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17934 of character types.
17935 (is_call_safe): New function.
17936 (try_substitute_return_value): Call it.
17937 (try_simplify_call): New function.
17938 (pass_sprintf_length::handle_gimple_call): Call it.
17940 2017-05-14 Martin Sebor <msebor@redhat.com>
17942 PR middle-end/80669
17943 * builtins.c (expand_builtin_stpncpy): Simplify.
17945 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17947 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17948 * config/i386/i386.h
17949 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17950 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17951 (struct machine_function): Add new members call_ms2sysv,
17952 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17953 (struct machine_frame_state): New fields sp_realigned and
17954 sp_realigned_offset.
17955 * config/i386/i386.c
17956 (enum xlogue_stub): New enum.
17957 (enum xlogue_stub_sets): New enum.
17958 (class xlogue_layout): New class.
17959 (struct ix86_frame): New fields stack_realign_allocate_offset,
17960 stack_realign_offset and outlined_save_offset. Modify comments to
17961 detail stack layout when using out-of-line stubs.
17962 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17963 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17964 -mcall-ms2sysv-xlogues.
17965 (stub_managed_regs): New static variable.
17966 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17967 registers managed by out-of-line stub.
17968 (disable_call_ms2sysv_xlogues): New function.
17969 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17970 m->call_ms2sysv when appropriate and compute frame layout for
17972 (sp_valid_at, fp_valid_at): New inline functions.
17973 (choose_basereg): New function.
17974 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17976 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17977 Use align parameter of choose_baseaddr to generated aligned SSE movs
17979 (pro_epilogue_adjust_stack): Modify to track
17980 machine_frame_state::sp_realigned.
17981 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17982 (ix86_nsaved_sseregs): Likewise.
17983 (ix86_emit_save_regs): Likewise.
17984 (ix86_emit_save_regs_using_mov): Likewise.
17985 (ix86_emit_save_sse_regs_using_mov): Likewise.
17986 (get_scratch_register_on_entry): Likewise.
17987 (gen_frame_set): New function.
17988 (gen_frame_load): Likewise.
17989 (gen_frame_store): Likewise.
17990 (emit_outlined_ms2sysv_save): Likewise.
17991 (emit_outlined_ms2sysv_restore): Likewise.
17992 (ix86_expand_prologue): Modify stack re-alignment code and call
17993 emit_outlined_ms2sysv_save when appropriate.
17994 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17995 parameter rtx_insn *insn, which allows the function to be used to only
17996 generate the notes.
17997 (ix86_expand_epilogue): Modify validity checks of frame and stack
17998 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17999 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18000 * config/i386/predicates.md
18001 (save_multiple): New predicate.
18002 (restore_multiple): Likewise.
18003 * config/i386/sse.md
18004 (save_multiple<mode>): New pattern.
18005 (save_multiple_realign<mode>): Likewise.
18006 (restore_multiple<mode>): Likewise.
18007 (restore_multiple_and_return<mode>): Likewise.
18008 (restore_multiple_leave_return<mode>): Likewise.
18009 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18011 2017-05-14 Julia Koval <julia.koval@intel.com>
18013 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18014 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18015 (__builtin_ia32_xsetbv): New builtins.
18016 * config/i386/i386.c (ix86_expand_special_args_builtin):
18018 (ix86_expand_builtin): Special expand for new intrinsics.
18019 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18020 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18021 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18023 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18025 * cfganal.c (inverted_post_order_compute): Change argument type
18027 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18028 * df-core.c (rest_of_handle_df_initialize): Adjust.
18029 (rest_of_handle_df_finish): Likewise.
18030 (df_analyze_1): Likewise.
18031 (df_analyze): Likewise.
18032 (loop_inverted_post_order_compute): Change argument to be a vec *.
18033 (df_analyze_loop): Adjust.
18034 (df_get_n_blocks): Likewise.
18035 (df_get_postorder): Likewise.
18036 * df.h (struct df_d): Change field to be a vec.
18037 * lcm.c (compute_laterin): Adjust.
18038 (compute_available): Likewise.
18039 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18040 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18041 * tree-ssa-pre.c (compute_antic): Likewise.
18043 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18045 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18046 (depth_first_search::depth_first_search): Change structure init
18047 function to this constructor.
18048 (depth_first_search::add_bb): Rename function to this member.
18049 (depth_first_search::execute): Likewise.
18050 (flow_dfs_compute_reverse_finish): Adjust.
18052 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18054 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18055 (longest_simple_path): Likewise.
18056 * shrink-wrap.c (spread_components): Likewise.
18057 (disqualify_problematic_components): Likewise.
18058 (emit_common_heads_for_components): Likewise.
18059 (emit_common_tails_for_components): Likewise.
18060 (insert_prologue_epilogue_for_components): Likewise.
18062 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18064 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18067 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18069 * df-core.c (df_set_blocks): Start using auto_bitmap.
18070 (df_compact_blocks): Likewise.
18071 * df-problems.c (df_rd_confluence_n): Likewise.
18072 * df-scan.c (df_insn_rescan_all): Likewise.
18073 (df_process_deferred_rescans): Likewise.
18074 (df_update_entry_block_defs): Likewise.
18075 (df_update_exit_block_uses): Likewise.
18076 (df_entry_block_bitmap_verify): Likewise.
18077 (df_exit_block_bitmap_verify): Likewise.
18078 (df_scan_verify): Likewise.
18079 * lra-constraints.c (lra_constraints): Likewise.
18080 (undo_optional_reloads): Likewise.
18081 (lra_undo_inheritance): Likewise.
18082 * lra-remat.c (calculate_gen_cands): Likewise.
18083 (do_remat): Likewise.
18084 * lra-spills.c (assign_spill_hard_regs): Likewise.
18085 (spill_pseudos): Likewise.
18086 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18087 (bitmap_set_subtract_values): Likewise.
18089 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18091 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18092 management with auto_bitmap.
18093 (fix_inter_tick): Likewise.
18094 (fix_recovery_deps): Likewise.
18095 * ira.c (add_store_equivs): Likewise.
18096 (find_moveable_pseudos): Likewise.
18097 (split_live_ranges_for_shrink_wrap): Likewise.
18098 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18099 (rtx_reuse_manager::seen_def_p): Likewise.
18100 (rtx_reuse_manager::set_seen_def): Likewise.
18101 * print-rtl.h (class rtx_reuse_manager): Likewise.
18103 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18105 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18107 (migrate_btr_def): Likewise.
18108 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18109 * df-core.c (loop_post_order_compute): Likewise.
18110 (loop_inverted_post_order_compute): Likewise.
18111 * hsa-common.h: Likewise.
18112 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18113 * init-regs.c (initialize_uninitialized_regs): Likewise.
18114 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18115 (inline_small_functions): Likewise.
18116 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18117 * ira.c (combine_and_move_insns): Likewise.
18118 (build_insn_chain): Likewise.
18119 * loop-invariant.c (find_invariants): Likewise.
18120 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18121 * predict.c (tree_predict_by_opcode): Likewise.
18122 (predict_paths_leading_to): Likewise.
18123 (predict_paths_leading_to_edge): Likewise.
18124 (estimate_loops_at_level): Likewise.
18125 (estimate_loops): Likewise.
18126 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18127 (spread_components): Likewise.
18128 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18129 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18130 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18131 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18132 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18133 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18134 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18135 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18136 (mark_threaded_blocks): Likewise.
18137 (thread_through_all_blocks): Likewise.
18138 * tree-ssa.c (verify_ssa): Likewise.
18139 (execute_update_addresses_taken): Likewise.
18140 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18142 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18144 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18146 (post_order_compute): Likewise.
18147 (inverted_post_order_compute): Likewise.
18148 (pre_and_rev_post_order_compute_fn): Likewise.
18150 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18152 * genrecog.c (int_set::int_set): Explicitly construct our
18153 auto_vec base class.
18154 * vec.h (auto_vec::auto_vec): New constructor.
18156 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18158 * bitmap.h (class auto_bitmap): New constructor taking
18159 bitmap_obstack * argument.
18161 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18163 * bitmap.h (class auto_bitmap): Change type of m_bits to
18164 bitmap_head, and adjust ctor / dtor and member operators.
18166 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18168 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18169 when returned register mode doesn't match original mode.
18171 2017-05-12 Jeff Law <law@redhat.com>
18172 Jakub Jelinek <jakub@redhat.com>
18174 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18175 we look for cc setter after the compare-elim changes.
18176 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18177 within the vector to match what compare-elim now expects.
18178 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18179 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18181 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18182 after the compare-elim changes.
18183 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18184 the vector to match what compare-elim now expects.
18185 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18186 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18187 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18188 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18189 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18191 * config/visium/visium.c (single_set_and_flags): Fix where
18192 we look for cc setter after the compare-elim changes.
18193 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18194 with the vector to match what compare-elim now expects.
18195 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18196 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18197 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18198 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18199 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18200 (neg<mode>2_insn_set_overflow): Likewise.
18202 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18204 PR middle-end/79794
18205 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18206 maybe_expand_insn call, set ops[0].target. If still set after call,
18207 set alt_rtl. Add extra arg to recursive calls.
18208 (extract_bit_field): Add alt_rtl argument. Pass to
18210 * expmed.h (extract_bit_field): Fix prototype.
18211 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18212 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18213 to extract_bit_field_calls.
18214 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18215 Pass alt_rtl to extract_bit_field calls.
18216 * calls.c (store_unaligned_arguments_into_psuedos)
18217 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18218 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18220 * optabs.h (struct expand_operand): Add target bitfield.
18222 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18224 * compare-elim.c (try_eliminate_compare): Canonicalize
18225 operation with embedded compare to
18226 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18227 (set (reg) (operation)].
18229 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18231 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18234 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18235 cost of adding a carry flag for ADC instruction.
18236 [case MINUS]: Ignore the cost of subtracting a carry flag
18237 for SBB instruction.
18239 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18241 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18243 * config/rs6000/bmiintrin.h: New file.
18244 * config/rs6000/bmi2intrin.h: New file.
18245 * config/rs6000/x86intrin.h: New file.
18247 2017-05-12 Jeff Law <law@redhat.com>
18249 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18252 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18254 PR middle-end/80707
18255 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18257 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18259 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18260 early expansion of vector divide builtins.
18261 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18262 builtins identified as having unsigned arguments.
18264 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18266 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18267 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18268 expansion of vector logical operations (and, andc, or, xor,
18271 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18273 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18274 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18276 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18278 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18279 early GIMPLE expansion of vector multiplies.
18281 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18283 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18284 TARGET_HAVE_MOVT conditional.
18285 (movt splitter): Likewise.
18287 2017-05-12 Richard Biener <rguenther@suse.de>
18289 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18290 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18291 Fold all stmts not inplace.
18293 2017-05-12 Richard Biener <rguenther@suse.de>
18295 PR tree-optimization/80713
18296 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18297 inserted_exprs bit for not removed stmts.
18299 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18301 PR middle-end/69921
18302 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18303 parallelized" attribute for parallelized OpenACC kernels.
18304 * omp-offload.c (execute_oacc_device_lower): Use it.
18306 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18307 Set "oacc kernels" attribute.
18308 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18309 parameter. Adjust all users.
18310 (oacc_fn_attrib_kernels_p): Remove function.
18311 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18312 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18313 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18314 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18315 assert "oacc kernels" attribute is set.
18317 2017-05-11 Carl Love <cel@us.ibm.com>
18319 * config/rs6000/rs6000-c: Add support for built-in functions
18320 vector unsigned char vec_popcnt (vector signed char)
18321 vector unsigned char vec_popcnt (vector unsigned char)
18322 vector unsigned short vec_popcnt (vector signed short)
18323 vector unsigned short vec_popcnt (vector unsigned short)
18324 vector unsigned int vec_popcnt (vector signed int)
18325 vector unsigned int vec_popcnt (vector unsigned int)
18326 vector unsigned long long vec_popcnt (vector signed long long)
18327 vector unsigned long long vec_popcnt (vector unsigned long long)
18328 vector signed long long vec_slo (vector signed long long,
18329 vector signed char)
18330 vector signed long long vec_slo (vector signed long long,
18331 vector unsigned char)
18332 vector unsigned long long vec_slo (vector unsigned long long,
18333 vector signed char)
18334 vector unsigned long long vec_slo (vector unsigned long long,
18335 vector unsigned char)
18336 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18337 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18338 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18339 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18340 * doc/extend.texi: Update the built-in documentation file for the
18341 new built-in functions.
18343 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18345 * attribs.h (sorted_attr_string): Move machine independent
18346 functions for target clone support from the i386 port to common
18347 code. Rename ix86_function_versions to common_function_versions.
18348 Rename make_name to make_unique_name.
18349 (common_function_versions): Likewise.
18350 (make_unique_name): Likewise.
18351 (make_dispatcher_decl): Likewise.
18352 (is_function_default_version): Likewise.
18353 * attribs.c (attr_strcmp): Likewise.
18354 (sorted_attr_string): Likewise.
18355 (common_function_versions): Likewise.
18356 (make_unique_name): Likewise.
18357 (make_dispatcher_decl): Likewise.
18358 (is_function_default_version): Likewise.
18359 * config/i386/i386.c (attr_strcmp): Likewise.
18360 (sorted_attr_string): Likewise.
18361 (ix86_function_versions): Likewise.
18362 (make_name): Likewise.
18363 (make_dispatcher_decl): Likewise.
18364 (is_function_default_version): Likewise.
18365 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18367 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18370 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18371 Account for direct move costs for vec_construct of integer
18374 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18377 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18378 (UNSPEC_STX_ATOMIC): Ditto.
18379 (loaddi_via_sse): New insn.
18380 (storedi_via_sse): Ditto.
18381 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18382 Update corresponding peephole2 patterns.
18383 (atomic_storedi_fpu): Ditto.
18385 2017-05-11 Julia Koval <julia.koval@intel.com>
18387 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18388 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18390 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18391 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18392 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18394 2017-05-11 Nathan Sidwell <nathan@acm.org>
18396 * graphite-poly.c: Include dumpfile.h.
18398 * dumpfle.h (dump_function): Declare here ...
18399 * tree-dump.h (dump_function): ... not here.
18400 * dumpfile.c: #include tree-cfg.h.
18401 (dump_function): Move here from ...
18402 * tree-dump.c (dump_function): ... here.
18403 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18404 * graphite-poly.c: Don't include tree-dump.h.
18405 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18406 * print-tree.c: Likewise.
18407 * stor-layout.c: Likewise.
18408 * tree-nested.c: Likewise.
18410 * dumpfile.c (dump_start): Use TDF_FLAGS.
18411 (dump_enable_all): Fix TDF_KIND check thinko.
18413 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18415 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18416 array entries to represent two legal parameterizations of the
18417 overloaded __builtin_cmpb function, as represented by the
18418 P6_OV_BUILTIN_CMPB constant.
18419 (altivec_resolve_overloaded_builtin): Add special case handling
18420 for the __builtin_cmpb function, as represented by the
18421 P6_OV_BUILTIN_CMPB constant.
18422 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18423 (BU_P6_64BIT_2): New macro.
18424 (BU_P6_OVERLOAD_2): New macro
18425 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18426 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18427 (CMPB): Add overload support to represent both 32-bit and 64-bit
18428 compare-bytes function.
18429 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18430 support for TARGET_CMPB.
18431 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18432 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18433 documentation of the __builtin_cmpb overloaded built-in function.
18435 2017-05-11 Richard Biener <rguenther@suse.de>
18437 PR tree-optimization/80705
18438 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18439 bases are not vectorizable.
18441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18443 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18444 when counting register pressure.
18446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18448 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18449 (iv_ca_more_deps): Renamed to ...
18450 (iv_ca_compare_deps): ... this.
18451 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18455 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18457 (determine_group_iv_costs): ... here.
18458 (find_inv_vars_cb): Record inv var if it's not recorded before.
18460 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18462 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18463 (get_shiftadd_cost): Ditto.
18465 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18467 * tree-ssa-address.c: Include header file.
18468 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18470 (add_to_parts): Refactor.
18471 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18472 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18475 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18477 PR tree-optimization/53090
18478 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18480 (extract_cond_operands): Detect condition with IV on both sides
18481 and return COMP_IV_EXPR_2.
18482 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18483 (rewrite_use_compare): Simplify by removing call to function
18484 extract_cond_operands.
18486 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18488 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18489 (extract_cond_operands): Detect condition comparing against non-
18490 invariant bound and return appropriate enum value.
18491 (find_interesting_uses_cond): Update use of extract_cond_operands.
18492 Handle its return value accordingly.
18493 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18495 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18497 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18498 nonlinear iv_use computation in loop invariant sensitive way.
18500 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18502 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18503 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18505 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18507 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18508 (dump_cand): Support iv_cand.inv_exprs.
18509 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18511 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18514 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18516 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18518 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18519 as local function. Include necessary header files.
18520 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18522 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18524 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18526 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18528 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18529 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18530 RSHIFT_EXPR and BIT_NOT_EXPR.
18532 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18534 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18535 (adjust_setup_cost): New parameter supporting round up adjustment.
18536 (struct address_cost_data): Delete.
18537 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18538 (split_address_cost, ptr_difference_cost): Delete.
18539 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18540 (struct ainc_cost_data): New struct.
18541 (get_address_cost_ainc): New function.
18542 (get_address_cost, get_computation_cost): Reimplement.
18543 (determine_group_iv_cost_address): Record inv_expr for all uses of
18545 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18546 (iv_ca_has_deps): Reimplemented to ...
18547 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18549 (iv_ca_extend): Call iv_ca_more_deps.
18551 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18553 * tree-ssa-address.c (struct mem_address): Move to header file.
18554 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18555 * tree-ssa-address.h (struct mem_address): Move from C file.
18556 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18558 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18560 * tree-affine.h (aff_combination_type): New interface.
18561 (aff_combination_zero_p): Remove static.
18562 (aff_combination_const_p): New interface.
18563 (aff_combination_singleton_var_p): New interfaces.
18565 2017-05-11 Richard Biener <rguenther@suse.de>
18567 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18568 Skip unreachable blocks and destinations.
18569 (eliminate): Move stmt removal and fixup ...
18570 (fini_eliminate): ... here. Skip inserted exprs.
18571 (pass_pre::execute): Move fini_pre after fini_eliminate.
18572 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18573 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18574 PRE to get rid of dead code that has invalid SSA form and
18575 split critical edges again.
18577 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18579 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18581 2017-05-11 Richard Biener <rguenther@suse.de>
18583 * passes.c (execute_function_todo): Verify loops if they are
18584 said to be up-to-date.
18585 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18586 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18588 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18591 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18592 handle calling assemble_external ourself.
18595 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18596 modes with zero size. Enhance comment.
18598 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18600 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18601 built-ins for vec_xl and vec_xst with short and char pointer
18604 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18606 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18607 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18608 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18609 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18610 (_mm_maskz_min_round_ss): New intrinsics.
18611 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18612 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18613 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18614 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18615 (__builtin_ia32_minss_mask_round): New builtins.
18616 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18617 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18618 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18620 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18621 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18623 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18626 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18628 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18629 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18630 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18631 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18632 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18633 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18634 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18635 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18636 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18637 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18638 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18639 (__builtin_ia32_mulss_mask_round): New builtins.
18640 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18641 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18642 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18644 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18645 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18647 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18650 2017-05-10 Julia Koval <julia.koval@intel.com>
18652 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18653 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18654 (_mm256_setr_m128i): New intrinsics.
18656 2017-05-10 Julia Koval <julia.koval@intel.com>
18658 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18659 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18660 (_mm_maskz_rcp14_ss): New intrinsics.
18661 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18662 (__builtin_ia32_rcp14ss_mask): New builtins.
18663 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18665 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18667 PR tree-optimization/51513
18668 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18669 (assert_unreachable_fallthru_edge_p): Use it.
18670 (group_case_labels_stmt): Likewise.
18671 * tree-cfg.h: Prototype it.
18672 * stmt.c: Include cfghooks.h and tree-cfg.h.
18673 (emit_case_dispatch_table) <gap_label>: New local variable.
18674 Use it to fill dispatch table gaps.
18675 Test for default_label before updating probabilities.
18676 (expand_case) <default_label>: Remove unneeded initialization.
18677 Test for unreachable default case statement and remove its edge.
18678 Set default_label accordingly.
18679 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18681 2017-05-10 Carl Love <cel@us.ibm.com>
18683 * config/rs6000/rs6000-c: Add support for built-in functions
18684 vector signed char vec_neg (vector signed char)
18685 vector signed short int vec_neg (vector short int)
18686 vector signed int vec_neg (vector signed int)
18687 vector signed long long vec_neg (vector signed long long)
18688 vector float vec_neg (vector float)
18689 vector double vec_neg (vector double)
18690 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18692 * config/rs6000/altivec.h: Add define for vec_neg
18693 * doc/extend.texi: Update the built-in documentation for the
18694 new built-in functions.
18696 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18698 PR tree-optimization/77644
18699 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18701 2017-05-10 Nathan Sidwell <nathan@acm.org>
18703 * dumpfile.h (TDI_lang_all): New.
18704 (TDF_KIND): New. Renumber others
18705 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18707 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18709 (get_dump_file_name): Adjust suffix generation.
18710 (dump_enable_all): Use TDF_KIND.
18711 * doc/invoke.texi (-fdump-lang-all): Document.
18713 * dumpfile.h: Tabify.
18715 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18718 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18719 Move member access before delete.
18721 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18723 * tree-inline.c (expand_call_inline): Split block at stmt
18726 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18729 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18730 are now unused after splitting mov{sf,sd}_hardfloat.
18731 (f32_lr2): Likewise.
18732 (f32_lm): Likewise.
18733 (f32_lm2): Likewise.
18734 (f32_li): Likewise.
18735 (f32_li2): Likewise.
18736 (f32_lv): Likewise.
18737 (f32_sr): Likewise.
18738 (f32_sr2): Likewise.
18739 (f32_sm): Likewise.
18740 (f32_sm2): Likewise.
18741 (f32_si): Likewise.
18742 (f32_si2): Likewise.
18743 (f32_sv): Likewise.
18744 (f32_dm): Likewise.
18745 (f32_vsx): Likewise.
18746 (f32_av): Likewise.
18747 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18748 For movsf, order stores so the VSX stores occur before the GPR
18749 store which encourages the register allocator to use a traditional
18750 FPR instead of a GPR. For movsd, order the stores so that the GPR
18751 store comes before the VSX stores to allow the power6 to work.
18752 This is due to the power6 not having a 32-bit integer store
18753 instruction from a FPR.
18754 (movsf_hardfloat): Likewise.
18755 (movsd_hardfloat): Likewise.
18757 2017-05-09 Martin Sebor <msebor@redhat.com>
18759 PR translation/80280
18760 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18763 PR translation/80280
18764 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18765 data member added in r247778.
18766 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18768 2017-05-09 Nathan Sidwell <nathan@acm.org>
18770 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18772 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18775 2017-05-09 Marek Polacek <polacek@redhat.com>
18777 * doc/invoke.texi: Fix typo.
18779 2017-05-09 Richard Biener <rguenther@suse.de>
18781 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18782 (vrp_val_is_min): Likewise.
18783 (set_value_range_to_value): Likewise.
18784 (set_value_range_to_nonnegative): Likewise.
18785 (gimple_assign_nonzero_p): Likewise.
18786 (gimple_stmt_nonzero_p): Likewise.
18787 (vrp_int_const_binop): Likewise. Remove unreachable case.
18788 (adjust_range_with_scev): Adjust comments.
18789 (compare_range_with_value): Likewise.
18790 (extract_range_from_phi_node): Likewise.
18791 (test_for_singularity): Likewise.
18793 2017-05-09 Richard Biener <rguenther@suse.de>
18795 * tree-vrp.c (get_single_symbol): Add assert that we don't
18796 get overflowed constants as invariant part.
18797 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18798 checks. Use wi::cmp instead of recursing for integer constants.
18799 (compare_values): Just ignore whether we assumed undefined
18800 overflow instead of failing the compare.
18801 (extract_range_for_var_from_comparison_expr): Add comment before the
18802 TREE_NO_WARNING sets.
18803 (test_for_singularity): Likewise.
18804 (extract_range_from_comparison): Do not disable optimization
18805 when we assumed undefined overflow.
18806 (extract_range_basic): Remove init of unused var.
18808 2017-05-09 Richard Biener <rguenther@suse.de>
18810 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18811 (extract_range_from_multiplicative_op_1): Adjust.
18812 (extract_range_from_binary_expr_1): Use int_const_binop.
18814 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18817 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18818 rs6000_store_data_bypass_p in seven define_bypass directives and
18819 in several comments.
18820 * config/rs6000/rs6000-protos.h: Add prototype for
18821 rs6000_store_data_bypass_p function.
18822 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18823 function implements slightly different (rs6000-specific) semantics
18824 than store_data_bypass_p, returning false rather than aborting
18825 with assertion error when arguments do not satisfy the
18826 requirements of store data bypass.
18827 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18828 rs6000_store_data_bypass_p.
18830 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18832 * config/xtensa/xtensa-protos.h
18833 (xtensa_initial_elimination_offset): New declaration.
18834 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18835 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18836 macro definition, add case for FRAME_POINTER_REGNUM when
18837 FRAME_GROWS_DOWNWARD.
18838 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18839 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18840 xtensa_initial_elimination_offset.
18842 2017-05-08 Nathan Sidwell <nathan@acm.org>
18844 * doc/invoke.texi: Alphabetize -fdump options.
18846 2017-05-08 Martin Sebor <msebor@redhat.com>
18848 PR translation/80280
18849 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18851 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18853 * target.def (compute_frame_layout): New optional target hook.
18854 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18855 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18856 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18858 * reload1.c (verify_initial_elim_offsets): Likewise.
18859 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18860 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18861 (arm_get_frame_offsets): Split up into this ...
18862 (arm_compute_frame_layout): ... and this function.
18864 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18866 * config/aarch64/constraints.md (Usa): New constraint.
18867 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18869 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18871 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18872 with_multilib_list after it has been checked.
18874 2017-05-08 Richard Biener <rguenther@suse.de>
18876 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18877 (bitmap_set_subtract_values): Likewise.
18879 2017-05-08 Richard Biener <rguenther@suse.de>
18881 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18882 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18884 (gimple_stmt_nonzero_warnv_p): Rename to ...
18885 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18887 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18888 (extract_range_basic): Adjust, do not disable propagation on
18889 strict overflow sensitive simplification.
18890 (vrp_visit_cond_stmt): Likewise.
18892 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18894 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18895 body size unconditionally.
18897 2017-05-07 Jeff Law <law@redhat.com>
18900 2017-05-06 Jeff Law <law@redhat.com>
18901 PR tree-optimization/78496
18902 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18905 PR tree-optimization/78496
18906 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18907 (simplify_stmt_using_ranges): Call it.
18908 (vrp_dom_walker::before_dom_children): Extract equivalences
18909 from an ASSERT_EXPR with an equality comparison against a
18912 2017-05-06 Jeff Law <law@redhat.com>
18914 PR tree-optimization/78496
18915 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18918 PR tree-optimization/78496
18919 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18920 (simplify_stmt_using_ranges): Call it.
18921 (vrp_dom_walker::before_dom_children): Extract equivalences
18922 from an ASSERT_EXPR with an equality comparison against a
18925 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18927 * lra-constraints.c (lra_copy_reg_equiv): New function.
18928 (split_reg): Use it to copy equivalence information from the
18929 original register to the spill register.
18931 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18933 PR rtl-optimization/75964
18934 * simplify-rtx.c (simplify_const_relational_operation): Remove
18935 invalid handling of comparisons of integer ABS.
18937 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18939 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18940 initialize to zero.
18941 (init_regs): Remove declaration.
18942 (function_arg_advance_32): Initialize error_p as boolean variable.
18944 2017-05-05 Nathan Sidwell <nathan@acm.org>
18946 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18947 lines. Use for (;;).
18949 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18951 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18952 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18953 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18954 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18955 VF=2 that require versioning.
18957 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18959 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18962 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18964 * diagnostic.h (diagnostic_override_option_index): Convert from
18965 macro to inline function.
18967 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18969 * diagnostic.c (last_module_changed_p): New function.
18970 (set_last_module): New function.
18971 (diagnostic_report_current_module): Convert macro usage to
18972 the above functions.
18973 * diagnostic.h (diagnostic_context::last_module): Strengthen
18974 from const line_map * to const line_map_ordinary *.
18975 (diagnostic_last_module_changed): Delete macro.
18976 (diagnostic_set_last_module): Delete macro.
18978 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18980 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18981 with diagnostic_report_diagnostic.
18982 (diagnostic_n_impl_richloc): Likewise.
18983 * diagnostic.h (report_diagnostic): Delete macro.
18984 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18985 with diagnostic_report_diagnostic.
18986 * substring-locations.c (format_warning_va): Likewise.
18988 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18990 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18991 save/restor of format_spec. Move option-printing code to...
18992 (print_option_information): ...this new function, and
18993 reimplement by simply printing to the pretty_printer,
18994 rather than appending to the format string.
18996 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18998 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18999 handling logic into...
19000 (update_effective_level_from_pragmas): ...this new function.
19002 2017-05-04 Andrew Waterman <andrew@sifive.com>
19004 * config/riscv/riscv.opt (mstrict-align): New option.
19005 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19006 (SLOW_UNALIGNED_ACCESS): Define.
19007 (riscv_slow_unaligned_access): Declare.
19008 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19010 (riscv_slow_unaligned_access): New variable.
19011 (rocket_tune_info): Set slow_unaligned_access to true.
19012 (optimize_size_tune_info): Set slow_unaligned_access to false.
19013 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19014 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19015 (riscv_option_override): Set riscv_slow_unaligned_access.
19016 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19018 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19020 * config/riscv/riscv.md: Unify indentation.
19022 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19027 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19029 (extendsi<mode>2): Add support for doing sign extension via
19030 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19031 don't have ISA 3.0 instructions.
19032 (extendsi<mode>2 splitter): Likewise.
19033 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19034 generate the normal insns since SImode can now go in vector
19035 registers. Disallow the special UNSPECs needed for previous
19036 machines to hide SImode being used. Add new insns
19037 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19038 (fix_trunc<mode>si2_stfiwx): Likewise.
19039 (fix_trunc<mode>si2_internal): Likewise.
19040 (fixuns_trunc<mode>si2): Likewise.
19041 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19042 (fctiw<u>z_<mode>_smallint): Likewise.
19043 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19044 of floating point to 32-bit integer from doing a direct move to
19045 the GPR registers to do a store.
19046 (fctiwz_<mode>): Break long line.
19048 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19050 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19051 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19052 (addr_list, addr_offset_valid_p): New.
19053 (split_address_groups): Check offset validity with above function.
19054 (gt-tree-ssa-loop-ivopts.h): Include header file.
19056 2017-05-05 Nathan Sidwell <nathan@acm.org>
19058 * config.gcc (arm*-*-*): Add missing 'fi'.
19060 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19062 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19063 included in -fopt-info does not matter.
19064 * doc/optinfo.texi (-fopt-info): Fix description of default
19065 behavour. Explicitly say order of options included in -fopt-info
19068 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19070 * config.gcc: Allow combinations of aprofile and rmprofile values for
19071 --with-multilib-list.
19072 * config/arm/t-multilib: New file.
19073 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19074 variables. Remove setting of ISA and floating-point ABI in
19075 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19076 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19077 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19078 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19080 * config/arm/t-rmprofile: Likewise except for the matches changes.
19081 * doc/install.texi (--with-multilib-list): Document the combination of
19082 aprofile and rmprofile values and warn about pitfalls in doing that.
19084 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19086 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19087 (movdi_aarch64): Likewise.
19089 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19091 PR tree-optimization/80632
19092 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19094 (build_arrays): Initialize it for virtual phis.
19095 (fix_phi_nodes): Use it for virtual phis.
19097 PR tree-optimization/80558
19098 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19099 [x, y] op z into [x op, y op z] for op & or | if conditions
19102 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19103 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19106 * config/arm/arm.md (use_literal_pool): Remove.
19107 (64-bit immediate split): No longer takes cost into consideration
19108 if arm_disable_literal_pool is enabled.
19109 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19110 used when arm_disable_literal_pool is enabled.
19111 (arm_max_const_double_inline_cost): Remove use of
19112 arm_disable_literal_pool.
19113 (push_minipool_fix): Add assert.
19114 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19115 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19116 (no_literal_pool_sf_immediate): New.
19118 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19120 PR tree-optimization/80613
19121 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19122 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19124 2017-05-05 Richard Biener <rguenther@suse.de>
19126 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19128 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19130 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19131 of this flag from insn conditions due to removal from r247495.
19133 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19135 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19137 (arm_early_store_addr_dep_ptr): Likewise.
19138 * config/arm/aarch-common-protos.h
19139 (arm_early_load_addr_dep_ptr): Add prototype.
19140 (arm_early_store_addr_dep_ptr): Likewise.
19141 * config/arm/cortex-a53.md: Add new bypasses.
19143 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19145 * tree.c (next_type_uid): Change type to unsigned.
19146 (type_hash_canon): Decrement back next_type_uid if
19147 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19148 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19151 2017-05-04 Martin Sebor <msebor@redhat.com>
19153 * builtins.c: Fix a trivial typo in a comment.
19155 PR middle-end/79234
19156 * builtins.c (check_sizes): Adjust to handle reading past the end.
19157 Avoid printing excessive upper bound of ranges. Use %E to print
19158 tree nodes instead of converting them to %wu.
19159 (expand_builtin_memchr): New function.
19160 (compute_dest_size): Rename...
19161 (compute_objsize): ...to this.
19162 (expand_builtin_memcpy): Adjust.
19163 (expand_builtin_mempcpy): Adjust.
19164 (expand_builtin_strcat): Adjust.
19165 (expand_builtin_strcpy): Adjust.
19166 (check_strncat_sizes): Adjust.
19167 (expand_builtin_strncat): Adjust.
19168 (expand_builtin_strncpy): Adjust and simplify.
19169 (expand_builtin_memset): Adjust.
19170 (expand_builtin_bzero): Adjust.
19171 (expand_builtin_memcmp): Adjust.
19172 (expand_builtin): Handle memcmp.
19173 (maybe_emit_chk_warning): Check strncat just once.
19175 2017-05-04 Martin Sebor <msebor@redhat.com>
19177 PR preprocessor/79214
19178 PR middle-end/79222
19179 PR middle-end/79223
19180 * builtins.c (check_sizes): Add inlining context and issue
19181 warnings even when -Wno-system-headers is set.
19182 (check_strncat_sizes): Same.
19183 (expand_builtin_strncat): Same.
19184 (expand_builtin_memmove): New function.
19185 (expand_builtin_stpncpy): Same.
19186 (expand_builtin): Handle memmove and stpncpy.
19188 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19190 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19191 which is not used any more.
19193 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19195 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19197 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19199 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19200 (cortexa53_tunings): Likewise.
19201 (cortexa57_tunings): Likewise.
19202 (cortexa72_tunings): Likewise.
19203 (cortexa73_tunings): Likewise.
19205 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19207 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19208 Set loop alignment to 8.
19210 2017-05-04 Martin Sebor <msebor@redhat.com>
19212 PR translation/80280
19213 * builtins.c (expand_builtin_object_size): Add missing quoting to
19214 %D and like directives.
19215 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19216 (hsa_type_for_tree_type): Same.
19217 (verify_function_arguments): Same.
19218 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19219 * varasm.c (get_section): Same.
19222 2017-05-04 Martin Sebor <msebor@redhat.com>
19224 PR translation/80280
19225 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19227 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19229 * config/aarch64/aarch64.c (generic_addrcost_table):
19230 Change HI/TI mode setting.
19232 2017-05-04 Martin Jambor <mjambor@suse.cz>
19234 PR tree-optimization/80622
19235 * tree-sra.c (comes_initialized_p): New function.
19236 (build_accesses_from_assign): Only set write lazily when
19237 comes_initialized_p is false.
19238 (analyze_access_subtree): Use comes_initialized_p.
19239 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19240 instead of testing for PARM_DECL.
19242 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19244 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19245 constraint on operand 0 to allow more general addressing modes.
19246 Adjust output template.
19247 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19249 * config/aarch64/aarch64-protos.h
19250 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19251 * config/aarch64/constraints.md (Dp): New address constraint.
19252 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19255 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19257 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19258 update use of estimate_ipcp_clone_size_and_time.
19259 (estimate_local_effects): Update use of
19260 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19261 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19262 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19263 Return nonspecialized time.
19265 2017-05-04 Richard Biener <rguenther@suse.de>
19267 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19268 for the last VUSE which def dominates the PHI. Directly call
19270 (get_continuation_for_phi_1): Remove.
19272 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19274 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19275 to explain the use of truncating division. Cap the number of
19276 iterations to the maximum given by nb_iterations_upper_bound,
19279 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19281 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19282 * configure: Regenerate.
19283 * config.in: Regenerate.
19284 * config/i386/driver-mingw32.c: new file.
19285 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19286 * config.host: Link driver-mingw32.o on MinGW host.
19287 * doc/install.texi: Document new --enable-mingw-wildcard configure
19290 2017-05-04 Marek Polacek <polacek@redhat.com>
19292 PR tree-optimization/80612
19293 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19295 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19296 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19298 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19299 (movt splitter): Likewise.
19300 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19301 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19302 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19303 block for Thumb-1 with MOVT.
19304 (thumb2_legitimate_address_p): Move code block ...
19305 (can_avoid_literal_pool_for_label_p): ... into this new function.
19306 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19308 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19309 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19310 "M-profile targets with the MOVT instruction".
19312 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19314 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19315 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19316 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19318 2017-05-04 Martin Liska <mliska@suse.cz>
19320 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19321 variable cond_code.
19323 2017-05-04 Richard Biener <rguenther@suse.de>
19325 * tree.c (array_at_struct_end_p): Handle arrays at struct
19326 end with flexarrays more conservatively. Refactor and treat
19327 arrays of arrays or aggregates more strict. Fix
19328 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19329 * tree.c (array_at_struct_end_p): Adjust prototype.
19330 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19331 * gimple-fold.c (get_range_strlen): Likewise.
19332 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19334 2017-05-04 Richard Biener <rguenther@suse.de>
19336 PR tree-optimization/31130
19337 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19339 (supports_overflow_infinity): Likewise.
19340 (is_negative_overflow_infinity): Likewise.
19341 (is_positive_overflow_infinity): Likewise.
19342 (is_overflow_infinity): Likewise.
19343 (stmt_overflow_infinity): Likewise.
19344 (overflow_infinity_range_p): Likewise.
19345 (usable_range_p): Remove as always returning true.
19346 (make_overflow_infinity): Remove.
19347 (negative_overflow_infinity): Likewise.
19348 (positive_overflow_infinity): Likewise.
19349 (avoid_overflow_infinity): Likewise.
19350 (set_value_range): Adjust accordingly.
19351 (set_value_range_to_nonnegative): Likewise, remove now unused
19352 overflow_infinity arg.
19353 (vrp_operand_equal_p): Adjust.
19354 (update_value_range): Likewise.
19355 (range_int_cst_singleton_p): Likewise.
19356 (operand_less_p): Likewise.
19357 (compare_values_warnv): Likewise.
19358 (extract_range_for_var_from_comparison_expr): Likewise.
19359 (vrp_int_const_binop): Likewise.
19360 (zero_nonzero_bits_from_vr): Likewise.
19361 (extract_range_from_multiplicative_op_1): Likewise.
19362 (extract_range_from_binary_expr_1): Likewise.
19363 (extract_range_from_unary_expr): Likewise.
19364 (extract_range_from_comparison): Likewise.
19365 (extract_range_basic): Likewise.
19366 (adjust_range_with_scev): Likewise.
19367 (compare_ranges): Likewise.
19368 (compare_range_with_value): Likewise.
19369 (dump_value_range): Likewise.
19370 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19372 (simplify_cond_using_ranges): Adjust.
19374 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19376 * brig-builtins.def: Added a builtin for class_f64.
19377 * builtin-types.def: Added a builtin type needed by class_f64.
19379 2017-05-03 Jason Merrill <jason@redhat.com>
19381 * timevar.def: Add TV_CONSTEXPR.
19383 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19385 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19387 2017-05-03 Martin Jambor <mjambor@suse.cz>
19389 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19390 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19391 * ipa-cp.c (ipcp_propagate_stage): Do not call
19392 ipa_update_after_lto_read.
19393 * ipa-inline.c (ipa_inline): Likewise.
19395 2017-05-03 Martin Jambor <mjambor@suse.cz>
19397 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19398 tag. Added a default constructor and a destructor.
19399 (ipa_edge_args_sum_t): New class;
19400 (ipa_edge_args_sum): Declare.
19401 (ipa_edge_args_vector): Remove declaration.
19402 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19403 (ipa_free_edge_args_substructures): Remove declaration.
19404 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19405 (ipa_edge_args_info_available_for_edge_p): Likewise.
19406 * ipa-prop.c (ipa_edge_args_vector): Removed.
19407 (edge_removal_hook_holder): Likewise.
19408 (edge_duplication_hook_holder): Likewise.
19409 (ipa_edge_args_sum): New variable.
19410 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19411 ipa_edge_args_vector.
19412 (ipa_free_edge_args_substructures): Likewise.
19413 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19414 ipa_edge_args_vector.
19415 (ipa_edge_removal_hook): Turned into method
19416 ipa_edge_args_sum_t::remove.
19417 (ipa_edge_duplication_hook): Turned into method
19418 ipa_edge_args_sum_t::duplicate.
19419 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19420 registering edge hooks.
19421 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19422 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19423 ipa_edge_args_sum instead of ipa_edge_args_vector.
19424 * ipa-profile.c (ipa_profile): Likewise.
19426 2017-05-03 Martin Jambor <mjambor@suse.cz>
19428 * symbol-summary.h (function_summary): New method exists.
19429 (function_summary::symtab_removal): Deallocate through release.
19430 (call_summary): New class.
19431 (gt_ggc_mx): New overload.
19432 (gt_pch_nx): Likewise.
19433 (gt_pch_nx): Likewise.
19435 2017-05-03 Jeff Law <law@redhat.com>
19437 PR tree-optimization/78496
19438 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19439 from simplify_cond_using_ranges. Split off code to walk
19440 backwards through casts into ...
19441 (simplify_cond_using_ranges_2): New function.
19442 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19443 (execute_vrp): After identifying jump threads, call
19444 simplify_cond_using_ranges_2.
19446 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19449 * ipa-inline.h (inline_summary): Add ctor.
19450 (create_ggc): Do not use ggc_cleared_alloc.
19452 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19453 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19455 * gcc.c (handle_braces): Support escaping in switch matching
19457 * doc/invoke.texi (Spec Files): Document it.
19458 Remove superfluous @code markup in items.
19460 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19462 * diagnostic-show-locus.c (struct column_range): New struct.
19463 (get_affected_columns): New function.
19464 (get_printed_columns): New function.
19465 (struct correction): New struct.
19466 (correction::ensure_capacity): New function.
19467 (correction::ensure_terminated): New function.
19468 (struct line_corrections): New struct.
19469 (line_corrections::~line_corrections): New dtor.
19470 (line_corrections::add_hint): New function.
19471 (layout::print_trailing_fixits): Reimplement in terms of the new
19473 (selftest::test_overlapped_fixit_printing): New function.
19474 (selftest::diagnostic_show_locus_c_tests): Call it.
19476 2017-05-03 Nathan Sidwell <nathan@acm.org>
19478 Canonicalize canonical type hashing
19479 * tree.h (type_hash_canon_hash): Declare.
19480 * tree.c (type_hash_list, attribute_hash_list): Move into
19481 type_hash_canon_hash.
19482 (build_type_attribute_qual_variant): Break out hash code calc into
19483 type_hash_canon_hash.
19484 (type_hash_canon_hash): New. Generic type hash computation.
19485 (build_range_type_1, build_array_type_1, build_function_type,
19486 build_method_type_directly, build_offset_type, build_complex_type,
19487 make_vector_type): Call it.
19489 2017-05-03 Richard Biener <rguenther@suse.de>
19491 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19492 When all DRs have unknown misaligned do not always peel
19493 when there is a store but apply the same costing model as if
19494 there were only loads.
19496 2017-05-03 Richard Biener <rguenther@suse.de>
19499 PR tree-optimization/80492
19500 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19501 compare_base_decls returning dont-know properly.
19503 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19505 * config/arm/iterators.md (CCSI): New mode iterator.
19506 (arch): New mode attribute.
19507 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19508 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19509 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19510 code iterator for success result mode.
19511 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19512 the corresponding new insn generators.
19514 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19517 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19518 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19520 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19522 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19523 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19524 (DDR_A): Wrap DDR argument in brackets.
19525 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19526 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19527 (DDR_REVERSED_P): Likewise.
19529 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19531 PR tree-optimization/79472
19532 * tree-switch-conversion.c (struct switch_conv_info): Add
19533 contiguous_range and default_case_nonstandard fields.
19534 (collect_switch_conv_info): Compute contiguous_range and
19535 default_case_nonstandard fields, don't clear final_bb if
19536 contiguous_range and only the default case doesn't have the required
19538 (check_all_empty_except_final): Set default_case_nonstandard instead
19539 of failing if contiguous_range and the default case doesn't have empty
19541 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19542 and only the default case doesn't have the required constants. Skip
19544 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19545 if default_case_nonstandard.
19546 (build_constructors): Build constant 1 just once. Assert that default
19547 values aren't inserted in between cases if contiguous_range. Skip
19549 (build_arrays): Skip virtual phis.
19550 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19551 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19552 Handle virtual phis.
19553 (gen_inbound_check): Handle default_case_nonstandard case.
19554 (process_switch): Adjust check_final_bb caller. Call
19555 gather_default_values with the first non-default case instead of
19556 default case if default_case_nonstandard.
19558 2017-05-02 Nathan Sidwell <nathan@acm.org>
19560 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19561 check. Fix formatting.
19563 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19565 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19566 errors when comparing specialized and unspecialized times.
19568 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19570 * diagnostic-show-locus.c
19571 (layout::should_print_annotation_line_p): Make private.
19572 (layout::print_annotation_line): Make private.
19573 (layout::annotation_line_showed_range_p): Make private.
19574 (layout::show_ruler): Make private.
19575 (layout::print_source_line): Make private. Pass in line and
19576 line_width, rather than calling location_get_source_line. Drop
19578 (layout::print_leading_fixits): New method.
19579 (layout::print_any_fixits): Rename to...
19580 (layout::print_trailing_fixits): ...this, and make private.
19581 Don't print newline fixits.
19582 (diagnostic_show_locus): Move logic for printing one row into...
19583 (layout::print_line): ...this new function. Move the
19584 location_get_source_line call and error-handling from
19585 print_source_line to here. Call print_leading_fixits, and rename
19586 print_any_fixits to print_trailing_fixits.
19587 (selftest::test_fixit_insert_containing_newline): Update now that
19588 newlines are partially supported.
19589 (selftest::test_fixit_insert_containing_newline_2): New test.
19590 (selftest::test_fixit_replace_containing_newline): Update comments.
19591 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19592 * edit-context.c (class added_line): New class.
19593 (class edited_line): Describe newline handling in comment.
19594 (edited_line::actually_edited_p): New method.
19595 (edited_line::print_content): Delete redundant decl.
19596 (edited_line::m_predecessors): New field.
19597 (edited_file::print_content): Call edited_line::print_content.
19598 (edited_file::print_diff): Update to support newlines.
19599 (edited_file::print_diff_hunk): Likewise.
19600 (edited_file::print_run_of_changed_lines): New function.
19601 (edited_file::print_diff_line): Convert to...
19602 (print_diff_line): ...this.
19603 (edited_file::get_effective_line_count): New function.
19604 (edited_line::edited_line): Initialize new field m_predecessors.
19605 (edited_line::~edited_line): Clean up m_predecessors.
19606 (edited_line::apply_fixit): Handle newlines.
19607 (edited_line::get_effective_line_count): New function.
19608 (edited_line::print_content): New function.
19609 (edited_line::print_diff_lines): New function.
19610 (selftest::test_applying_fixits_insert_containing_newline): New
19612 (selftest::test_applying_fixits_replace_containing_newline): New
19614 (selftest::insert_line): New function.
19615 (selftest::test_applying_fixits_multiple_lines): Add example of
19617 (selftest::edit_context_c_tests): Call the new tests.
19619 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19621 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19622 parameter cand. Update dump information.
19623 (get_computation_cost): Update uses.
19625 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19627 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19628 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19629 (get_computation_at, rewrite_use_address): Update use of
19630 get_computation_aff.
19632 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19634 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19635 (get_computation): Delete.
19636 (get_computation_cost): Implement like get_computation_cost_at.
19637 Use get_computation_at.
19638 (get_computation_cost_at): Delete.
19639 (rewrite_use_nonlinear_expr): Use get_computation_at.
19640 (rewrite_use_compare, remove_unused_ivs): Ditto.
19642 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19644 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19646 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19648 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19649 (ivopts_global_cost_for_size): Rename parameter and update uses.
19650 (iv_ca_recount_cost): Update uses.
19651 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19652 candidates seperately in n_invs and n_cands.
19653 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19655 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19657 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19658 (find_inv_vars_cb): New.
19659 (find_depends): Renamed to ...
19660 (find_inv_vars): ... this.
19661 (add_candidate_1, force_var_cost): Call find_inv_vars.
19662 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19664 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19666 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19667 inv_vars. Add inv_exprs.
19668 (struct iv_cand): Rename depends_on to inv_vars.
19669 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19670 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19671 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19672 (dump_cand): Dump inv_vars.
19673 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19674 (record_invariant, find_depends, add_candidate_1): Ditto.
19675 (set_group_iv_cost, force_var_cost): Ditto.
19676 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19677 (get_computation_cost_at, get_computation_cost): Ditto.
19678 (determine_group_iv_cost_generic): Ditto.
19679 (determine_group_iv_cost_address): Ditto.
19680 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19681 (determine_group_iv_costs): Ditto.
19682 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19683 (iv_ca_set_remove_invariants): Renamed to ...
19684 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19685 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19686 (iv_ca_set_add_invariants): Renamed to ...
19687 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19688 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19689 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19690 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19691 (create_new_ivs): Remove useless dump.
19693 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19695 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19697 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19698 (iv_ca_set_no_cp, create_new_iv): Ditto.
19700 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19702 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19704 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19706 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19707 function tree_check2.
19709 2017-05-02 Martin Liska <mliska@suse.cz>
19711 * doc/gcov.texi: Add missing preposition.
19712 * gcov.c (function_info::function_info): Properly fill up
19713 all member variables.
19715 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19717 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19719 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19721 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19723 2017-05-02 Martin Liska <mliska@suse.cz>
19726 * lto-streamer-in.c (lto_read_tree_1): Remove
19727 LTO_STREAMER_DEBUG.
19728 * lto-streamer.c (struct tree_hash_entry): Likewise.
19729 (struct tree_entry_hasher): Likewise.
19730 (tree_entry_hasher::hash): Likewise.
19731 (tree_entry_hasher::equal): Likewise.
19732 (lto_streamer_init): Likewise.
19733 (lto_orig_address_map): Likewise.
19734 (lto_orig_address_get): Likewise.
19735 (lto_orig_address_remove): Likewise.
19736 * lto-streamer.h: Likewise.
19737 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19738 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19740 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19742 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19743 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19744 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19745 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19746 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19747 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19748 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19749 (mm_maskz_sub_ss): New intrinsics.
19750 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19751 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19752 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19753 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19754 (__builtin_ia32_subss_mask_round): New builtins.
19755 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19756 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19757 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19759 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19760 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19762 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19765 2017-05-02 Martin Jambor <mjambor@suse.cz>
19767 PR tree-optimization/78687
19768 * tree-sra.c (access): New field parent.
19769 (process_subtree_disqualification): New function.
19770 (disqualify_candidate): Call it.
19771 (build_accesses_from_assign): Reset write flag if creating an
19773 (build_access_subtree): Fill in parent field and also prpagate
19774 down grp_write flag.
19775 (create_artificial_child_access): New parameter set_grp_write, set
19776 grp_write to its value.
19777 (propagate_subaccesses_across_link): Also propagate grp_write flag
19779 (propagate_all_subaccesses): Push the closest parent back to work
19780 queue if add_access_to_work_queue returned true.
19782 2017-05-02 Richard Biener <rguenther@suse.de>
19784 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19785 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19786 -fstrict-overflow documentation.
19787 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19788 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19789 flag_strict_overflow.
19790 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19791 * lto-opts.c (lto_write_options): Do not stream it.
19792 * lto-wrapper.c (merge_and_complain): Do not handle it.
19793 * opts.c (default_options_table): Do not set -fstrict-overflow.
19794 (finish_options): Likewise do not clear it when sanitizing.
19795 * simplify-rtx.c (simplify_const_relational_operation): Do not
19796 test flag_strict_overflow.
19798 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19800 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19801 using enabled attribute.
19802 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19803 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19804 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19805 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19806 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19807 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19808 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19809 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19810 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19811 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19813 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19815 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19817 2017-05-02 Richard Biener <rguenther@suse.de>
19819 PR tree-optimization/80591
19821 2017-04-10 Richard Biener <rguenther@suse.de>
19823 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19826 2017-05-02 Richard Biener <rguenther@suse.de>
19828 PR tree-optimization/80549
19829 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19830 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19831 headers if they do not have a preheader.
19833 2017-05-02 Martin Liska <mliska@suse.cz>
19836 * common.opt: Fix typo.
19837 * doc/invoke.texi: Likewise.
19839 2017-05-01 Jan Beulich <jbeulich@suse.com>
19841 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19842 swapping, add (x,x,m,x,n) alternative.
19844 2017-05-01 Nathan Sidwell <nathan@acm.org>
19846 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19847 unnecessary unadjusted_alignment check.
19849 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19852 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19854 * gimplify.c (gimplify_cilk_detach): New function.
19855 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19856 * tree-core.h: Document EXPR_CILK_SPAWN.
19857 * tree.h (EXPR_CILK_SPAWN): Define.
19859 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19861 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19862 to use new fixit_hint representation, using the "replace" logic.
19863 (get_line_span_for_fixit_hint): Likewise.
19864 (layout::print_any_fixits): Likewise.
19865 (selftest::test_one_liner_many_fixits): Rename to...
19866 (selftest::test_one_liner_many_fixits_1): ...this, and update
19867 comment and expected output to reflect that the multiple fix-it
19868 hints are now consolidated into one insertion.
19869 (selftest::test_one_liner_many_fixits_2): New test.
19870 (selftest::test_diagnostic_show_locus_one_liner): Update for
19872 (selftest::test_fixit_consolidation): Update for fix-it API
19874 * diagnostic.c (print_parseable_fixits): Likewise.
19875 * edit-context.c (edited_line::m_line_events): Convert from
19876 auto_vec <line_event *> to auto_vec <line_event>.
19877 (class line_event): Convert from abstract base class to a concrete
19878 class, taking over the role of replace_event.
19879 (class insert_event): Delete.
19880 (class replace_event): Rename to class line_event. Convert to
19882 (edit_context::add_fixits): Reimplement.
19883 (edit_context::apply_insert): Delete.
19884 (edit_context::apply_replace): Rename to...
19885 (edit_context::apply_fixit): ...this. Convert to half-open range.
19886 (edited_file::apply_insert): Delete.
19887 (edited_file::apply_replace): Rename to...
19888 (edited_file::apply_fixit): ...this.
19889 (edited_line::~edited_line): Drop deletion of events.
19890 (edited_line::apply_insert): Delete.
19891 (edited_line::apply_replace): Rename to...
19892 (edited_line::apply_fixit): ...this. Convert to half-open range.
19893 Update for change to type of m_line_events.
19894 * edit-context.h (edit_context::apply_insert): Delete.
19895 (edit_context::apply_replace): Rename to...
19896 (edit_context::apply_fixit): ...this.
19898 2017-05-01 Martin Sebor <msebor@redhat.com>
19900 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19903 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19906 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19907 __get_cpuid_max returns 0.
19908 (__get_cpuid_count): Ditto.
19910 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19912 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19913 replacement expression is another instance of one of its arguments.
19915 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19918 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19919 check for stack push/pop autoinc.
19920 * config/i386/i386.c (ix86_agi_dependent): Return false
19921 if the only reason why modified_in_p returned true is that
19922 addr is SP based and set_insn is a push or pop.
19924 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19926 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19929 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19932 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19933 (account_size_time): Use two predicates - exec_pred and
19935 (evaluate_conditions_for_known_args): Compute both clause and
19937 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19938 (inline_summary_t::duplicate): Update.
19939 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19941 (compute_inline_parameters): Likewise.
19942 (estimate_edge_size_and_time): Update caluclation of time.
19943 (estimate_node_size_and_time): Compute both time and nonspecialized
19945 (estimate_ipcp_clone_size_and_time): Update.
19946 (inline_merge_summary): Update.
19947 (do_estimate_edge_time): Update.
19948 (do_estimate_edge_size): Update.
19949 (do_estimate_edge_hints): Update.
19950 (inline_read_section, inline_write_summary): Stream both new predicates.
19951 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19953 (compute_inlined_call_time): Cleanup.
19954 (big_speedup_p): Update.
19955 (edge_badness): Update.
19956 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19957 (size_time_entry): Replace predicate by exec_predicate and
19958 nonconst_predicate.
19959 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19960 (estimate_edge_time): Return also nonspec_time.
19961 (reset_edge_growth_cache): Update.
19963 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19965 PR rtl-optimization/80491
19966 * ifcvt.c (noce_process_if_block): When looking for x setter
19967 with missing else_bb, don't check only the insn right before
19968 cond_earliest, but look for the last insn that x is modified in
19969 within the same bb.
19971 PR rtl-optimization/80491
19972 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19974 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19976 PR tree-optimization/80487
19977 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19979 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19981 PR tree-optimization/79697
19982 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19983 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19984 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19986 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19987 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19989 2017-04-28 Martin Sebor <msebor@redhat.com>
19991 PR tree-optimization/80523
19992 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19993 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19995 (maybe_warn, format_directive, parse_directive): Use new functions.
19996 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19998 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20000 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20002 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20004 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20005 target_header_dir): Set correctly.
20006 * configure: Regenerated.
20007 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20008 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20009 instead of SYSTEM_HEADER_DIR.
20011 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20013 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20014 (estimate_local_effects): Likewise.
20015 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20016 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20017 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20018 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20019 do_estimate_edge_time, estimate_edge_time): Likewise.
20020 * ipa-inline-analysis.c (estimate_node_size_and_time,
20021 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20022 (estimate_time_after_inlining): Remove.
20024 2017-04-28 Martin Liska <mliska@suse.cz>
20026 * doc/gcov.texi: Enhance documentation of gcov.
20028 2017-04-28 Martin Liska <mliska@suse.cz>
20030 * doc/gcov.texi: Sort options in alphabetic order.
20031 * doc/gcov-dump.texi: Likewise.
20032 * doc/gcov-tool.texi: Likewise.
20033 * gcov.c (print_usage): Likewise.
20034 * gcov-dump.c (print_usage): Likewise.
20035 * gcov-tool.c (print_merge_usage_message): Likewise.
20036 (print_rewrite_usage_message): Likewise.
20037 (print_overlap_usage_message): Likewise.
20039 2017-04-28 Martin Liska <mliska@suse.cz>
20041 PR gcov-profile/53915
20042 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20044 2017-04-28 Martin Liska <mliska@suse.cz>
20046 PR gcov-profile/79891
20047 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20048 is marked by compiler as living on a line.
20049 (get_cycles_count): Remove usage of the union.
20050 (output_intermediate_file): Likewise.
20051 (find_source): Fix GNU coding style.
20052 (accumulate_line_counts): Remove old non-all block mode.
20053 (output_lines): Remove usage of the union.
20054 * profile.c (output_location): Include all BBs, even if
20055 belonging to a same line (and file) as a previous BB.
20057 2017-04-28 Martin Liska <mliska@suse.cz>
20059 * gcov.c (process_args): Handle new argument 'w'.
20060 (read_graph_file): Assign ID to BBs.
20061 (output_branch_count): Display BB # if verbose flag is set.
20062 (output_lines): Likewise for arcs.
20063 (print_usage): Add '--verbose' option help.
20064 * doc/gcov.texi: Document --verbose (-w) option.
20066 2017-04-28 Martin Liska <mliska@suse.cz>
20068 * gcov.c (struct block_location_info): New struct.
20069 (process_file): Fill up the new structure.
20070 (read_graph_file): Replace usage of encoding by the newly added
20072 (add_line_counts): Likewise.
20073 (accumulate_line_counts): Remove usage of the union.
20074 (function_info::function_info): New function.
20075 (function_info::~function_info): Likewise.
20076 (process_file): Call delete instead of release_function.
20077 (release_function): Release the function.
20078 (release_structures): Call delete instead of release_function.
20079 (solve_flow_graph): Replace usage of num_blocks.
20080 (find_exception_blocks): Likewise.
20081 (output_lines): Fix GNU coding style.
20083 2017-04-28 Martin Liska <mliska@suse.cz>
20086 * coverage.c (coverage_remove_note_file): New function.
20087 * coverage.h: Declare the function.
20088 * toplev.c (finalize): Clean if an error has been seen.
20090 2017-04-28 Martin Liska <mliska@suse.cz>
20092 PR gcov-profile/80031
20093 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20094 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20095 * gcov.c (read_graph_file): Read just number of blocks.
20096 * profile.c (branch_prob): Do not stream 0 flags per a basic
20099 2017-04-28 Martin Liska <mliska@suse.cz>
20101 * gcov-dump.c (tag_*): Add new argument to declarations.
20102 (dump_gcov_file): Likewise.
20103 (tag_blocks): Add and use new argument depth.
20104 (tag_arcs): Likewise.
20105 (tag_lines): Likewise.
20106 (tag_counters): Likewise.
20107 (tag_summary): Likewise.
20108 (dump_working_sets): Use depth to do a proper indentation.
20110 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20113 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20114 * symtab.c (symtab_node::debug_symtab): Move definition here.
20116 2017-04-28 Richard Biener <rguenther@suse.de>
20118 * lto-streamer.h (LTO_major_version): Bump to 7.
20120 2017-04-28 Richard Biener <rguenther@suse.de>
20122 * tree-vrp.c (assert_info): New struct.
20123 (add_assert_info): New helper.
20124 (register_edge_assert_for_2): Refactor to add asserts to a vector
20126 (register_edge_assert_for_1): Likewise.
20127 (register_edge_assert_for): Likewise.
20128 (finish_register_edge_assert_for): New helper actually registering
20129 asserts where live on edge.
20130 (find_conditional_asserts): Adjust.
20131 (find_switch_asserts): Likewise.
20132 (evrp_dom_walker::try_find_new_range): Generalize.
20133 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20135 2017-04-27 Marek Polacek <polacek@redhat.com>
20138 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20139 arg10 and arg11 to itype.
20141 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20143 * doc/extend.texi (Object Size Checking): Improve grammar.
20145 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20148 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20149 that the logic for permitting reciprocal estimates matches that
20152 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20155 * tree.c (type_cache_hasher::equal): Only compare
20156 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20157 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20158 non-aggregate element types.
20159 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20160 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20162 2017-04-27 Richard Biener <rguenther@suse.de>
20164 PR middle-end/80533
20165 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20166 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20167 keeping a reference to a trailing array.
20169 2017-04-27 Richard Biener <rguenther@suse.de>
20171 PR middle-end/80539
20172 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20173 being in loop-closed SSA form conservatively.
20174 (chrec_fold_multiply_poly_poly): Likewise.
20176 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20178 PR middle-end/79665
20179 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20180 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20182 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20185 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20186 (aarch64_function_arg_alignment): Return unsigned int again, but still
20187 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20188 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20189 Don't emit -Wpsabi note.
20190 (aarch64_function_arg_boundary): Likewise.
20191 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20194 2017-04-26 Nathan Sidwell <nathan@acm.org>
20196 * tree.h (crc32_unsigned_n): Declare.
20197 (crc32_unsigned, crc32_unsigned): Make inline.
20198 * tree.c (crc32_unsigned_bits): Replace with ...
20199 (crc32_unsigned_n): ... this.
20200 (crc32_unsigned, crc32_byte): Remove.
20201 (crc32_string): Remove unnecessary braces.
20203 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20205 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20206 * ipa-inline-analysis.c (MAX_TIME): Remove.
20207 (account_size_time): Use sreal for time.
20208 (dump_inline_summary): Update.
20209 (estimate_function_body_sizes): Update.
20210 (estimate_edge_size_and_time): Update.
20211 (estimate_calls_size_and_time): Update.
20212 (estimate_node_size_and_time): Update.
20213 (inline_merge_summary): Update.
20214 (inline_update_overall_summary): Update.
20215 (estimate_time_after_inlining): Update.
20216 (inline_read_section): Update.
20217 (inline_write_summary): Update.
20218 * ipa-inline.c (compute_uninlined_call_time): Update.
20219 (compute_inlined_call_time): Update.
20220 (recursive_inlining): Update.
20221 (inline_small_functions): Update.
20222 (dump_overall_stats): Update.
20223 * ipa-inline.h: Include sreal.h.
20224 (size_time_entry): Turn time to sreal.
20225 (inline_summary): Turn self_time nad time to sreal.
20227 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20229 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20231 (sreal::stream_out, sreal::stream_in): New.
20232 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20234 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20236 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20239 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20242 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20244 (dimode_scalar_chain::compute_convert_gain): Ditto.
20245 (dimode_scalar_chain::make_vector_copies): Ditto.
20246 (dimode_scalar_chain::convert_reg): Ditto.
20247 (dimode_scalar_chain::convert_insn): Ditto.
20248 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20249 (VI248_AVX512BW_1): New mode iterator.
20250 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20251 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20254 2017-04-25 Martin Sebor <msebor@redhat.com>
20256 PR tree-optimization/80497
20257 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20258 constants are representable in HOST_WIDE_INT.
20259 (parse_directive): Ditto.
20261 2017-04-25 Martin Sebor <msebor@redhat.com>
20264 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20265 (new_zero_array): Adjust signature.
20266 (dom_info::dom_init): Used unsigned rather that size_t.
20267 (dom_info::dom_info): Same.
20269 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20270 Jakub Jelinek <jakub@redhat.com>
20273 * config/arm/arm.c: Include gimple.h.
20274 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20275 returns negative, increment ncrn only if it returned positive.
20276 (arm_needs_doubleword_align): Return int instead of bool,
20277 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20278 members, but if there is any such non-FIELD_DECL
20279 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20280 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20281 returns negative, increment nregs only if it returned positive.
20282 (arm_setup_incoming_varargs): Likewise.
20283 (arm_function_arg_boundary): Emit -Wpsabi note if
20284 arm_needs_doubleword_align returns negative, return
20285 DOUBLEWORD_ALIGNMENT only if it returned positive.
20287 2017-04-25 Marek Polacek <polacek@redhat.com>
20290 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20291 first argument to type.
20293 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20296 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20297 type checks to test for compatibility instead of equality.
20299 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20300 Jakub Jelinek <jakub@redhat.com>
20303 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20305 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20306 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20307 the alignment computation, but return their maximum in warn_alignment.
20308 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20309 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20311 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20312 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20315 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20317 * config/arc/simdext.md (dmpyh): Fix typo.
20319 2017-04-25 Richard Biener <rguenther@suse.de>
20321 PR tree-optimization/80492
20322 * alias.c (compare_base_decls): Handle registers with asm
20323 specification conservatively.
20324 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20325 compare_base_decls returning dont-know properly.
20327 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20329 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20330 (legitimate_offset_address_p): New function.
20331 (arc_legitimate_address_p): Use above function.
20333 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20335 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20337 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20339 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20340 ACCH registers whenever they are available.
20342 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20344 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20345 double regs fix when not used.
20347 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20349 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20351 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20352 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20354 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20356 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20357 when emitting short ADD instructions.
20359 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20361 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20363 (cmpsi_cc_c_insn): Likewise.
20364 (cbranchsi4_scratch): Compute proper instruction length using
20365 compact_hreg_operand.
20366 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20368 2017-04-25 Richard Biener <rguenther@suse.de>
20370 PR middle-end/80509
20371 * passes.c (pass_manager::pass_manager): Initialize
20372 m_name_to_pass_map.
20374 2017-04-25 Richard Biener <rguenther@suse.de>
20376 PR tree-optimization/79201
20377 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20379 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20382 * config/s390/vector.md: Split MEM->GPR vector moves for
20383 non-s_operand addresses.
20385 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20388 * config/s390/predicates.md (reload_const_wide_int_operand): New
20390 * config/s390/s390.md ("movti"): Remove d/P alternative.
20391 ("movti_bigconst"): New pattern definition.
20393 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20396 * s390-protos.h (s390_expand_cs_hqi): Removed.
20397 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20398 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20399 modes as well as CCZ1mode and CCZmode.
20400 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20401 signature of s390_emit_compare_and_swap.
20402 (s390_expand_cs_hqi): Likewise, make static.
20403 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20404 compare-and-swap, in some cases.
20405 (s390_expand_cs): Wrapper function.
20406 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20408 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20409 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20410 patterns for small and large integers. Forbid symref memory operands.
20411 Move expander to s390.c. Require cc register.
20412 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20413 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20414 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20415 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20416 symref memory operands. Remove CC mode and call s390_match_ccmode
20418 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20420 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20422 * config/s390/s390.md (define_peephole2): New peephole to help
20423 combining the load-and-test pattern with volatile memory.
20425 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20427 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20428 with CCZmode for TARGET_Z196.
20430 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20432 PR rtl-optimization/80501
20433 * combine.c (make_compound_operation_int): Set subreg_code to SET
20434 even for AND with mask of the sign bit of mode.
20436 PR rtl-optimization/80500
20437 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20438 sum's initial value.
20440 2017-04-25 Julian Brown <julian@codesourcery.com>
20441 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20443 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20445 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20447 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20449 2017-04-25 Julian Brown <julian@codesourcery.com>
20450 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20452 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20453 (thunderx2t99_sha): New Reservation.
20455 2017-04-25 Julian Brown <julian@codesourcery.com>
20456 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20458 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20459 type for 1-element load.
20461 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20463 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20465 2017-04-24 Martin Jambor <mjambor@suse.cz>
20467 PR tree-optimization/80293
20468 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20469 char arrays not totally scalarizable if it is false.
20470 (analyze_all_variable_accesses): Pass correct value in the new
20471 parameter. Add a statistics counter.
20473 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20475 PR middle-end/79931
20476 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20478 2017-04-24 Richard Biener <rguenther@suse.de>
20480 PR tree-optimization/80494
20481 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20482 out for complex types.
20484 2017-04-24 Richard Biener <rguenther@suse.de>
20486 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20487 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20488 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20489 (DFS): Adjust and never fail.
20490 (sccvn_dom_walker::fail): Remove.
20491 (sccvn_dom_walker::before_dom_children): Adjust.
20492 (run_scc_vn): Likewise and never fail.
20493 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20494 (pass_fre::execute): Likewise.
20496 2017-04-24 Richard Biener <rguenther@suse.de>
20498 PR tree-optimization/79725
20499 * tree-ssa-sink.c (statement_sink_location): Return whether
20500 failure reason was zero uses. Move that check later.
20501 (sink_code_in_bb): Deal with zero uses by removing the stmt
20504 2017-04-24 Richard Biener <rguenther@suse.de>
20507 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20508 pointer-based references.
20510 2017-04-24 Richard Biener <rguenther@suse.de>
20513 * pass_manager.h (pass_manager::operator new): Remove.
20514 (pass_manager::operator delete): Likewise.
20515 * passes.c (pass_manager::operator new): Remove.
20516 (pass_manager::operator delete): Likewise.
20517 (pass_manager::pass_manager): Zero individual pass members.
20519 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20522 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20523 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20524 Check "XEXP (src, 1)" operand here.
20525 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20526 Check "XEXP (src, 1)" operand here.
20527 (dimode_scalar_chain::make_vector_copies): Detect count register
20528 of a shift instruction. Zero extend count register from QImode
20529 to DImode to satisfy vector shift pattern count operand predicate.
20530 Substitute vector shift count operand with a DImode copy.
20531 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20534 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20536 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20537 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20538 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20539 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20540 (UNSPEC_NOREX_MEM): Remove definition.
20542 2017-04-21 Richard Biener <rguenther@suse.de>
20544 PR tree-optimization/79547
20545 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20546 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20547 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20548 without any constraints.
20550 2017-04-21 Richard Biener <rguenther@suse.de>
20552 PR tree-optimization/78847
20553 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20555 2017-04-21 Richard Biener <rguenther@suse.de>
20557 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20558 (build_distinct_type_copy): Likewise.
20559 (build_variant_type_copy): Likewise.
20560 * tree.c (build_qualified_type): Pass down mem-stat info.
20561 (build_distinct_type_copy): Likewise.
20562 (build_variant_type_copy): Likewise.
20564 2017-04-21 Richard Biener <rguenther@suse.de>
20566 PR tree-optimization/80237
20567 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20569 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20570 for a simplified result.
20572 2016-04-21 Richard Biener <rguenther@suse.de>
20574 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20575 sth as strict as a simple_iv but a chrec without symbols and an
20576 operand defined in the loop we are peeling (and not some subloop).
20577 (propagate_constants_for_unrolling): Propagate all constants.
20579 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20582 * config/i386/i386.c (print_reg): Remove assert for disalowed
20583 regno values, call output_operand_lossage instead.
20585 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20588 * config/i386/constraints.md (Yc): New register constraint.
20589 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20590 Use Yc constraint for alternative 2 of operand 0. Remove
20591 preferred_for_speed attribute.
20593 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20595 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20596 lastprivate clauses in SIMT case.
20598 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20600 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20602 2017-04-20 Richard Biener <rguenther@suse.de>
20604 PR tree-optimization/57796
20605 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20606 as N scalar stores.
20607 (vect_model_load_cost): Cost gathers as N scalar loads.
20609 2017-04-20 Richard Biener <rguenther@suse.de>
20611 * ggc-page.c (ggc_allocated_p): Rename to ...
20612 (safe_lookup_page_table_entry): ... this and return the lookup
20614 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20616 2017-04-20 Richard Biener <rguenther@suse.de>
20618 PR tree-optimization/80453
20619 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20620 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20621 from the conditions.
20622 (vn_phi_eq): Pass them down.
20623 (vn_phi_lookup): Record them.
20624 (vn_phi_insert): Likewise.
20626 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20628 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20629 uninitialized variable warning to avoid buffer overrun.
20631 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20634 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20635 is suppressed for '{ 0 }' in C.
20637 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20639 * BASE-VER: Set to 8.0.0.
20641 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20643 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20644 priority .init_array and .fini_array section with SECTION_NOTYPE
20647 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20649 PR middle-end/80423
20650 * tree.h (build_array_type): Add typeless_storage default argument.
20651 * tree.c (type_cache_hasher::equal): Also compare
20652 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20653 (build_array_type): Add typeless_storage argument, set
20654 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20656 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20657 (build_array_type): Likewise. Add typeless_storage argument.
20659 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20660 Jakub Jelinek <jakub@redhat.com>
20662 PR tree-optimization/80426
20663 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20664 operation on symbolic operands, also compute the overflow for the
20665 invariant part when the operation degenerates into a negation.
20667 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20670 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20671 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20674 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20676 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20679 * config/avr/avr.c (tree.h): Include it.
20680 (cgraph.h): Include it.
20681 (avr_encode_section_info): Don't warn for uninitialized progmem
20682 variable if it's just an alias.
20684 2017-04-19 Richard Biener <rguenther@suse.de>
20687 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20688 when needed by AutoPGO.
20690 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20693 * doc/lto.texi: Remove an extra 'that'.
20695 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20697 PR rtl-optimization/80429
20698 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20699 are only used in debug insns.
20701 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20702 Vladimir Makarov <vmakarov@redhat.com>
20704 * config/sparc/predicates.md (input_operand): Add comment. Return
20705 true for any memory operand when LRA is in progress.
20706 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20708 2017-04-18 Jeff Law <law@redhat.com>
20711 * mips.md ({return,simple_return}_internal): Do not overwrite
20714 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20716 PR tree-optimization/80443
20717 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20718 instead of adding 1, subtract -1 and similarly instead of subtracting
20721 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20723 PR rtl-optimization/80357
20724 * haifa-sched.c (tmp_bitmap): New variable.
20725 (model_recompute): Handle duplicate use records.
20726 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20727 (free_global_sched_pressure_data): Free it.
20729 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20732 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20733 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20734 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20735 instead of SYSTEM_HEADER_DIR.
20737 2017-04-18 Jeff Law <law@redhat.com>
20739 PR middle-end/80422
20740 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20741 predecessors after walking up the insn chain.
20743 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20746 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20747 sizetype type into debug info.
20749 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20752 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20753 unneeded test for TARGET_UPPER_REGS_SF.
20754 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20756 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20759 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20760 instead of gsi_after_labels.
20762 2017-04-18 Jeff Law <law@redhat.com>
20764 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20768 2017-04-13 Jeff Law <law@redhat.com>
20769 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20770 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20772 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20775 * config/avr/avr.c (intl.h): Include it.
20776 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20778 2017-04-18 Martin Liska <mliska@suse.cz>
20780 PR gcov-profile/78783
20781 * gcov-tool.c (gcov_output_files): Validate that destination
20782 file is either removed by the tool or by a user.
20784 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20785 Guy Benyei <guybe@mellanox.com>
20787 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20788 block, and do not negate it, the stored id is already negative.
20790 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20792 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20794 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20797 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20798 masks of options that should be turned off if the VSX vector
20799 options are turned off.
20800 (OTHER_P8_VECTOR_MASKS): Likewise.
20801 (OTHER_VSX_VECTOR_MASKS): Likewise.
20802 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20803 rs6000_disable_incompatible_switches to validate no type switches
20805 (rs6000_incompatible_switch): New function to disallow turning on
20806 other vector options if -mno-vsx, -mno-power8-vector, or
20807 -mno-power9-vector are specified.
20809 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20811 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20813 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20815 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20816 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20817 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20818 (ARG_POINTER_CFA_OFFSET): Likewise.
20820 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20822 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20823 conditions to take advantage of various optimizations.
20825 2017-04-13 Jeff Law <law@redhat.com>
20827 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20828 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20829 (zero_extendsidi2_dext): Likewise.
20831 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20834 * fold-const.c (fold_ternary_loc): Revert
20835 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20838 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20840 PR rtl-optimization/80343
20841 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20842 new scratch pseudo.
20844 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20847 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20848 to ubsan_encode_value.
20850 2017-04-13 Jeff Law <law@redhat.com>
20852 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20853 appearing in DEBUG_INSNs.
20855 2017-04-13 Martin Liska <mliska@suse.cz>
20857 PR gcov-profile/80413
20858 * gcov-io.c (gcov_write_string): Copy to buffer just when
20859 allocated size is greater than zero.
20861 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20864 * dwarf2out.c (decls_for_scope): Ignore declarations of
20865 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20867 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20870 * ipa-visibility.c (non_local_p): Fix typos.
20871 (localize_node): When localizing symbol in same comdat group,
20872 dissolve the group only when we know external symbols are going
20874 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20876 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20878 PR tree-optimization/79390
20879 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20880 order does not result in usable sequence, retry with reversed operand
20886 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20887 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20888 op0 instead of fold_convert_loc (loc, type, arg0).
20890 2017-04-12 Jeff Law <law@redhat.com>
20892 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20893 has a delay slot in the generated code.
20895 * config/cris/cris.md (cris_preferred_reload_class): Return
20896 GENNONACR_REGS rather than GENERAL_REGS.
20898 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20901 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20902 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20903 signedness of the result type.
20905 2017-04-12 Richard Biener <rguenther@suse.de>
20906 Jeff Law <law@redhat.com>
20908 PR tree-optimization/80359
20909 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20910 trim stores to TARGET_MEM_REFs.
20912 2017-04-12 Richard Biener <rguenther@suse.de>
20914 PR tree-optimization/79390
20915 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20916 threading case even more.
20918 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20921 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20922 for quad_address_p for TImode, instead of just not indexed_address.
20924 2017-04-12 Richard Biener <rguenther@suse.de>
20925 Bernd Edlinger <bernd.edlinger@hotmail.de>
20927 PR middle-end/79671
20928 * alias.c (component_uses_parent_alias_set_from): Handle
20929 TYPE_TYPELESS_STORAGE.
20930 (get_alias_set): Likewise.
20931 * tree-core.h (tree_type_common): Add typeless_storage flag.
20932 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20933 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20934 for types containing members with TYPE_TYPELESS_STORAGE.
20935 (place_field): Likewise.
20936 (layout_type): Likewise for ARRAY_TYPE.
20937 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20938 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20939 TYPE_TYPELESS_STORAGE.
20940 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20942 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20945 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20946 first argument to type.
20948 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20952 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20953 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20954 (rs6000_expand_binop_builtin): Likewise.
20955 (rs6000_expand_ternop_builtin): Likewise; also add missing
20956 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20957 vshasigma built-ins.
20958 * doc/extend.texi: Document that vec_xxpermdi's third argument
20959 must be a constant.
20961 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20963 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20964 Use shift_const cost parameter when calculating gain of STV shifts.
20966 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20968 PR rtl-optimization/70478
20969 * lra-constraints.c (process_alt_operands): Check memory for
20970 disfavoring memory insn operand.
20972 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20974 PR middle-end/80100
20975 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20976 left shift in unsigned HOST_WIDE_INT type.
20978 PR rtl-optimization/80385
20979 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20980 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20983 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20984 if they have any depend clauses.
20986 2017-04-11 Martin Liska <mliska@suse.cz>
20989 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20990 * ipa-split.c (split_function): Create a local comdat symbol
20991 if caller is in a comdat group.
20993 2017-04-11 Martin Liska <mliska@suse.cz>
20996 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20999 2017-04-11 Martin Sebor <msebor@redhat.com>
21001 PR middle-end/80364
21002 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21003 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21005 (directive::set_width, directive::set_precision, format_character):
21007 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21010 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21013 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21014 conflict, set target->arch_name instead of target->cpu_name.
21016 2017-04-11 Richard Biener <rguenther@suse.de>
21018 PR tree-optimization/80374
21019 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21020 build_zero_cst, remove fold_convertible_p check again.
21022 2017-04-11 Martin Liska <mliska@suse.cz>
21025 * ubsan.c (instrument_object_size): Do not instrument register
21028 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21031 * config/i386/i386-builtin-types.def
21032 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21033 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21034 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21035 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21036 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21037 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21038 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21039 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21040 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21041 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21042 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21043 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21044 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21045 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21046 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21047 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21048 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21049 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21050 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21051 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21052 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21053 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21054 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21055 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21056 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21057 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21058 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21059 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21060 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21061 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21062 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21063 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21064 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21065 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21066 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21067 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21068 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21069 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21070 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21071 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21072 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21073 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21074 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21075 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21076 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21077 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21079 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21080 flag to second_arg_count, handle 4 argument function type _COUNT
21081 aliases, handle second_arg_count on second argument rather than last.
21083 2017-04-10 Jeff Law <law@redhat.com>
21085 PR tree-optimization/80374
21086 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21087 record anything if we can not convert integer_zero_node to the
21090 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21093 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21094 Enhance special handling given to the TARGET_P9_MINMAX option in
21095 relation to certain other options.
21097 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21099 PR tree-optimization/80153
21100 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21101 remove POINTER_PLUS_EXPR's base part directly, rather than through
21104 2017-04-10 Richard Biener <rguenther@suse.de>
21105 Bin Cheng <bin.cheng@arm.com>
21107 PR tree-optimization/80153
21108 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21109 the first element of pointer type aff_tree. Build result expr in
21111 (add_elt_to_tree): Convert to type unconditionally. Remove other
21112 fold_convert calls.
21113 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21114 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21116 2017-04-10 Richard Biener <rguenther@suse.de>
21118 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21121 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21123 PR rtl-optimization/70478
21124 * lra-constraints.c (curr_small_class_check): New.
21125 (update_and_check_small_class_inputs): New.
21126 (process_alt_operands): Update curr_small_class_check. Disfavor
21127 alternative insn memory operands. Check available regs for small
21130 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21133 * config/mips/mips.opt (-mvirt): Update description.
21134 * doc/invoke.texi (-mvirt): Likewise.
21136 2017-04-10 Richard Biener <rguenther@suse.de>
21138 PR middle-end/80362
21139 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21140 looking for NEGATE_EXPR in -A / -B to A / B folding.
21142 2017-04-10 Martin Liska <mliska@suse.cz>
21144 PR gcov-profile/80224
21145 * gcov.c (print_usage): Fix usage string.
21146 (get_gcov_intermediate_filename): Remove.
21147 (output_gcov_file): Use both for normal and intermediate format.
21148 (generate_results): Do not initialize special file for
21149 intermediate format.
21151 2017-04-10 Richard Biener <rguenther@suse.de>
21153 PR tree-optimization/80304
21154 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21157 2017-04-10 Nathan Sidwell <nathan@acm.org>
21160 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21161 (rs6000_init_builtins): Use it.
21163 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21165 * config/arm/arm.md (<mrc>): Add mode to SET source.
21166 (<mrrc>): Likewise.
21168 2017-04-10 Richard Biener <rguenther@suse.de>
21170 PR middle-end/80344
21171 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21173 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21176 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21177 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21178 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21179 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21180 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21181 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21182 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21183 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21184 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21185 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21186 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21187 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21188 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21189 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21190 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21191 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21192 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21193 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21194 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21195 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21196 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21197 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21198 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21200 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21202 PR rtl-optimization/70478
21203 * lra-constraints.c: Reverse the last patch.
21205 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21207 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21208 Add comment for WCHAR_T.
21210 2017-04-08 Martin Liska <mliska@suse.cz>
21213 2017-04-07 Martin Liska <mliska@suse.cz>
21216 * ipa-split.c (split_function): Add function part to a same comdat
21219 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21222 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21224 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21226 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21227 (vec_pairing): ...to this.
21228 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21229 (rs6000_sched_init): Adjust for name change.
21230 (struct rs6000_sched_context): Likewise.
21231 (rs6000_init_sched_context): Likewise.
21232 (rs6000_set_sched_context): Likewise.
21234 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21240 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21242 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21243 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21244 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21246 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21248 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21250 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21252 PR rtl-optimization/70703
21253 * ira-color.c (update_conflict_hard_regno_costs): Use
21254 int64_t instead of HOST_WIDE_INT.
21256 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21258 PR rtl-optimization/70478
21259 * lra-constraints.c (process_alt_operands): Disfavor alternative
21260 insn memory operands.
21262 2017-04-07 Jeff Law <law@redhat.com>
21264 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21265 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21267 2017-04-07 Martin Liska <mliska@suse.cz>
21270 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21271 Show error message instead of an ICE.
21273 2017-04-07 Martin Liska <mliska@suse.cz>
21276 * ipa-split.c (split_function): Add function part to a same comdat
21279 2017-04-07 Richard Biener <rguenther@suse.de>
21281 PR middle-end/80341
21282 * tree.c (get_unwidened): Also handle ! for_type case for
21284 * convert.c (do_narrow): Split out from ...
21285 (convert_to_integer_1): ... here. Do not pass final truncation
21286 type to get_unwidened for TRUNC_DIV_EXPR.
21288 2017-04-07 Richard Biener <rguenther@suse.de>
21290 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21292 (aff_combination_const): Adjust.
21293 (aff_combination_scale): Likewise.
21294 (aff_combination_add_elt): Likewise.
21295 (aff_combination_add_cst): Likewise.
21296 (aff_combination_convert): Likewise.
21297 (add_elt_to_tree): Likewise. Remove unused argument.
21298 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21300 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21302 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21304 * config/arm/arm.c (arm_default_short_enums): Use
21305 ARM_DEFAULT_SHORT_ENUMS.
21306 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21308 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21311 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21312 members with redundant out-of-class redeclaration.
21314 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21317 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21318 * config/i386/i386.md (*zero_extendsidi2):
21319 Add (?*x,*x) and (?*v,*v) alternatives.
21321 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21324 * config/i386/i386.c (ix86_expand_builtin)
21325 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21326 mode from insn data. Convert operands to insn operand mode.
21327 Copy operands that don't satisfy insn predicate to a register.
21329 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21331 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21334 2017-04-06 Richard Biener <rguenther@suse.de>
21336 PR tree-optimization/80334
21337 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21338 preserve alignment of accesses.
21340 2017-04-06 Richard Biener <rguenther@suse.de>
21342 PR tree-optimization/80262
21343 * tree-sra.c (build_ref_for_offset): Preserve address-space
21345 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21346 Drop useless address-space information on MEM_REF offsets.
21348 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21350 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21352 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21354 PR rtl-optimization/70703
21355 * ira-color.c (update_conflict_hard_regno_costs): Use
21356 HOST_WIDE_INT instead of long.
21358 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21361 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21362 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21364 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21365 for x86_64 target. Handle -m3dnowa option.
21367 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21369 PR rtl-optimization/70703
21370 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21371 (update_conflict_hard_regno_costs): Use long instead of unsigned
21372 arithmetic for cost calculation.
21374 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21375 Bernd Edlinger <bernd.edlinger@hotmail.de>
21378 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21381 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21384 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21385 ptr_mode with Pmode throughout.
21386 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21387 into probe_stack_range and use DImode.
21389 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21392 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21393 call_eh_return is true.
21395 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21397 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21398 Initialize last_match_fntype_index.
21400 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21403 * tree-nvr.c: Include internal-fn.h.
21404 (pass_return_slot::execute): Ignore internal calls without
21407 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21408 Richard Biener <rguenther@suse.de>
21411 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21412 captures used multiple times, except for the last use.
21413 * generic-match-head.c: Include gimplify.h.
21415 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21417 PR tree-optimization/79390
21418 * target.h (struct noce_if_info): Declare.
21419 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21420 * target.def (noce_conversion_profitable_p): New target hook.
21421 * ifcvt.h (struct noce_if_info): New type, moved from ...
21422 * ifcvt.c (struct noce_if_info): ... here.
21423 (noce_conversion_profitable_p): Renamed to ...
21424 (default_noce_conversion_profitable_p): ... this. No longer
21426 (noce_try_store_flag_constants, noce_try_addcc,
21427 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21428 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21429 instead of noce_conversion_profitable_p.
21430 * config/i386/i386.c: Include ifcvt.h.
21431 (ix86_option_override_internal): Don't override
21432 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21433 (ix86_noce_conversion_profitable_p): New function.
21434 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21435 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21436 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21437 * doc/tm.texi: Regenerated.
21439 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21441 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21444 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21447 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21448 instructions for small multiply cores.
21450 2017-04-04 Jeff Law <law@redhat.com>
21452 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21454 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21455 that are not set by the loop over the elements.
21457 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21460 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21461 int mode, convert_modes it to mode as unsigned, otherwise use
21462 lowpart_subreg to mode rather than SImode.
21463 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21464 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21465 Use DImode instead of SImode for the shift count operand.
21466 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21469 2017-04-04 Richard Biener <rguenther@suse.de>
21471 PR middle-end/80281
21472 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21473 arithmetic done for the negate or the plus. Simplify.
21474 (A - (-B) -> A + B): Likewise.
21475 * fold-const.c (split_tree): Make sure to not negate pointers.
21477 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21479 PR rtl-optimization/60818
21480 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21481 a compare of comparisons with the thing compared if this results
21482 in a different machine mode.
21484 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21486 * alias.c (base_alias_check): Fix typo in comment.
21487 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21488 * cgraphunit.c (symbol_table::compile): Likewise.
21489 * collect2.c (maybe_run_lto_and_relink): Likewise.
21490 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21491 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21492 * config/avr/avr.c (avr_map_op_t): Likewise.
21493 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21494 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21495 * config/epiphany/epiphany.md (movcc): Likewise.
21496 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21497 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21499 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21500 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21501 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21502 * config/sh/sh.c (sh_rtx_costs): Likewise.
21503 * fold-const.c (fold_truth_andor): Likewise.
21504 * genautomata.c (collapse_flag): Likewise.
21505 * gengtype.h (struct type::u::s): Likewise.
21506 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21507 * input.c (FORMAT_AMOUNT): Likewise.
21508 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21509 (known_aggs_to_agg_replacement_list): Likewise.
21510 * ipa-inline-analysis.c: Likewise.
21511 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21512 * ipa-polymorphic-call.c
21513 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21514 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21515 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21517 * modulo-sched.c (apply_reg_moves): Likewise.
21518 * omp-expand.c (build_omp_regions_1): Likewise.
21519 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21520 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21521 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21522 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21523 * value-prof.c: Likewise.
21524 * var-tracking.c (val_reset): Likewise.
21526 2017-04-03 Richard Biener <rguenther@suse.de>
21528 PR tree-optimization/80275
21529 * fold-const.c (split_address_to_core_and_offset): Handle
21532 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21534 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21535 descriptors is at least equal to that of functions.
21537 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21539 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21541 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21544 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21545 (mov<IMOD4:mode>): New expander.
21546 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21548 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21550 PR rtl-optimization/79405
21551 * fwprop.c (propagations_left): New variable.
21552 (forward_propagate_into): Decrement it.
21553 (fwprop_init): Initialize it.
21554 (fw_prop): If the variable has reached zero, stop propagating.
21555 (fwprop_addr): Ditto.
21557 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21560 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21561 a FUNCTION_DECL, pass it as decl instead of origin to
21564 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21566 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21569 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21572 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21573 TARGET_VSX_SMALL_INTEGER.
21575 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21577 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21578 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21580 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21582 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21583 extraction from odd-numbered MSA register.
21585 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21587 PR middle-end/80173
21588 * expmed.c (store_bit_field_1): Don't attempt to create
21589 a word subreg out of hard registers wider than word if they
21590 have HARD_REGNO_NREGS of 1 for their mode.
21592 PR middle-end/80163
21593 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21594 conversions to integer types wider than word and pointer.
21597 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21598 (rtx_equal_for_cselib_p): Pass 0 to it.
21599 * cselib.c (cselib_hasher::equal): Likewise.
21600 (rtx_equal_for_cselib_1): Add depth argument. If depth
21601 is 128, don't look up VALUE locs and punt. Increment
21602 depth in recursive calls when walking VALUE locs.
21604 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21606 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21607 (make_gcov_file_name): Use the canonical path name for generating
21609 (read_line): Fix handling of files with ascii null bytes.
21611 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21613 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21614 to initialise a vector register instead
21615 of using a const_int.
21617 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21619 PR translation/80189
21620 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21621 diagnostic messages.
21623 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21626 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21627 (dfp_diex_<mode>): Update mode of operand 1.
21628 * doc/extend.texi (dxex, dxexq): Document change to return type.
21629 (diex, diexq): Document change to argument type.
21631 2017-03-30 Martin Jambor <mjambor@suse.cz>
21634 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21635 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21636 it reflects the signature changes performed at the callee side.
21637 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21638 to cgraph_build_function_type_skip_args.
21639 (build_function_decl_skip_args): Adjust call to the above function.
21641 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21644 * config/i386/sse.md
21645 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21646 register as dest whenever it is a MEM not rtx_equal_p to the
21647 corresponding dup operand, and when forcing into reg move the
21648 reg into the memory afterwards.
21649 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21650 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21651 for the force_reg mode.
21652 (avx512vl_vextractf128<mode>): Use register as dest either
21653 always when a MEM, or when it is a MEM not rtx_equal_p to the
21654 corresponding dup operand, or even not when it is a CONST_VECTOR
21655 depending on the mode and lo vs. hi.
21656 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21658 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21659 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21660 Likewise. Require that operands[2] is even.
21661 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21662 Remove extraneous parens. Require that operands[2] is a multiple
21664 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21665 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21666 disallow memory then.
21668 2017-03-30 Richard Biener <rguenther@suse.de>
21670 PR tree-optimization/77498
21671 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21672 to non-constants over backedges.
21674 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21676 PR rtl-optimization/80233
21677 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21678 as last_combined_insn. Do not test for BARRIER_P separately.
21680 2017-03-29 Andreas Schwab <schwab@suse.de>
21683 * calls.c (prepare_call_address): Convert funexp to Pmode before
21684 copying to temp reg.
21686 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21688 PR tree-optimization/80158
21689 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21690 Handle possible future case of more than one alternate
21692 (replace_rhs_if_not_dup): Likewise.
21693 (replace_one_candidate): Likewise.
21695 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21697 PR rtl-optimization/80193
21698 * ira.c (ira): Do not check allocation for LRA.
21700 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21702 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21703 (nvptx_output_simt_exit): Declare.
21704 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21705 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21706 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21707 (nvptx_declare_function_name): ...here. Emit declaration of local
21708 memory space buffer for omp_simt_enter insn.
21709 (nvptx_output_unisimt_switch): New.
21710 (nvptx_output_softstack_switch): New.
21711 (nvptx_output_simt_enter): New.
21712 (nvptx_output_simt_exit): New.
21713 * config/nvptx/nvptx.h (struct machine_function): New fields
21714 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21715 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21716 (UNSPECV_SIMT_EXIT): Ditto.
21717 (omp_simt_enter_insn): New insn.
21718 (omp_simt_enter): New expansion.
21719 (omp_simt_exit): New insn.
21720 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21722 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21723 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21724 (expand_GOMP_SIMT_EXIT): New.
21725 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21726 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21727 (GOMP_SIMT_EXIT): Ditto.
21728 * target-insns.def (omp_simt_enter): New insn.
21729 (omp_simt_exit): Ditto.
21730 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21732 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21733 (lower_rec_input_clauses): Likewise.
21734 (lower_lastprivate_clauses): Handle SIMT privatization.
21736 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21737 (ompdevlow_adjust_simt_enter): New.
21738 (find_simtpriv_var_op): New.
21739 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21740 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21742 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21743 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21744 (copy_decl_for_dup_finish): Ditto.
21746 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21748 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21751 * config/i386/i386.c (ix86_option_override_internal): Always
21752 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21754 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21756 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21758 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21760 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21761 mark new edge's irreducible flag accordign to it.
21762 (vect_do_peeling): Check loop preheader edge's irreducible flag
21763 and pass it to function slpeel_add_loop_guard.
21765 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21767 PR tree-optimization/80218
21768 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21769 Update block frequencies and counts.
21771 2017-03-28 Richard Biener <rguenther@suse.de>
21773 PR tree-optimization/78644
21774 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21775 of a simplification result we may not use it at all.
21777 2017-03-28 Richard Biener <rguenther@suse.de>
21780 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21781 without arguments, generate default definition of a SSA name.
21783 2017-03-28 Richard Biener <rguenther@suse.de>
21785 PR middle-end/80222
21786 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21787 TYPE_REF_CAN_ALIAS_ALL references.
21788 * fold-const.c (fold_indirect_ref_1): Likewise.
21790 2017-03-28 Martin Liska <mliska@suse.cz>
21793 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21794 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21796 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21797 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21799 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21800 (EXTRA_SPECS): Define.
21801 (SUBTARGET_EXTRA_SPECS): Likewise.
21802 (SUBTARGET_CPP_SPEC): Likewise.
21803 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21804 SUBTARGET_EXTRA_SPECS.
21805 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21807 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21809 * config/arc/simdext.md (vst64_insn): Update pattern.
21810 (vld32wh_insn): Likewise.
21811 (vld32wl_insn): Likewise.
21812 (vld64_insn): Likewise.
21813 (vld32_insn): Likewise.
21815 2017-03-28 Marek Polacek <polacek@redhat.com>
21818 * fold-const.c (fold_comparison): Use protected_set_expr_location
21819 instead of SET_EXPR_LOCATION.
21821 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21823 * tree.c (add_expr): Avoid name lookup warning.
21825 2017-03-27 Jeff Law <law@redhat.com>
21827 PR tree-optimization/80216
21828 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21829 function name. Limit recursion depth.
21830 (record_temporary_equivalences): Corresponding changes.
21832 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21834 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21837 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21840 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21842 * cfgcleanup.c (reg_note_cfa_p): New array.
21843 (insns_have_identical_cfa_notes): New function.
21844 (old_insns_match_p): Don't cross-jump in between /f
21845 and non-/f instructions. If both i1 and i2 are frame related,
21846 verify all CFA notes, their order and content.
21848 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21851 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21852 HImode and SImode with zero extend to DImode to one insn.
21853 (bswap<mode>2_extenddi): Likewise.
21854 (bswapsi2_extenddi): Likewise.
21855 (bswaphi2_extendsi): Likewise.
21856 (bswaphi2): Combine bswap HImode and SImode into one insn.
21857 Separate memory insns from swapping register.
21858 (bswapsi2): Likewise.
21859 (bswap<mode>2): Likewise.
21860 (bswaphi2_internal): Delete, no longer used.
21861 (bswapsi2_internal): Likewise.
21862 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21863 store, and gpr<-gpr swap insns.
21864 (bswap<mode>2_store): Likewise.
21865 (bswaphi2_reg): Register only splitter, combine with the splitter.
21866 (bswaphi2 splitter): Likewise.
21867 (bswapsi2_reg): Likewise.
21868 (bswapsi2 splitter): Likewise.
21869 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21870 the insns into load, store, and register/register insns.
21871 (bswapdi2_ldbrx): Likewise.
21872 (bswapdi2_load): Likewise.
21873 (bswapdi2_store): Likewise.
21874 (bswapdi2_reg): Likewise.
21876 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21878 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21879 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21881 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21884 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21886 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21887 special handling for target option conflicts between dform
21888 options (-mpower9-dform, -mpower9-dform-vector,
21889 -mpower9-dform-scalar) and -mno-direct-move.
21891 2017-03-27 Richard Biener <rguenther@suse.de>
21893 PR tree-optimization/80181
21894 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21896 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21898 * config/arc/predicates.md (move_double_src_operand): Replace the
21899 call to move_double_src_operand with a call to address_operand.
21901 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21903 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21904 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21905 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21907 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21909 * config/arc/predicates.md (long_immediate_loadstore_operand):
21910 Consider scaled addresses cases.
21912 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21914 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21915 restored when in interrupt.
21916 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21917 doesn't have delay slot.
21919 2017-03-27 Richard Biener <rguenther@suse.de>
21922 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21923 inlined thunk clones.
21925 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21928 * asan.c (instrument_derefs): Copy over last operand from
21929 original COMPONENT_REF to the new COMPONENT_REF with
21930 DECL_BIT_FIELD_REPRESENTATIVE.
21931 * ubsan.c (instrument_object_size): Likewise.
21933 2017-03-27 Richard Biener <rguenther@suse.de>
21935 PR tree-optimization/80170
21936 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21937 sure DR/SCEV didnt fold in constants we do not see when looking
21938 at the reference base alignment.
21940 2017-03-27 Richard Biener <rguenther@suse.de>
21942 PR middle-end/80171
21943 * gimple-fold.c (fold_ctor_reference): Properly guard against
21944 NULL return value from canonicalize_constructor_val.
21946 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21949 * config/i386/i386.c (ix86_expand_builtin)
21950 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21951 flags reg setting and flags reg using instructions.
21952 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21953 clobbering instructions to zero extend op2.
21955 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21957 * doc/install.texi (Configuration) <--with-aix-soname>:
21958 Update link to AIX ld.
21960 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21962 PR rtl-optimization/80160
21963 PR rtl-optimization/80159
21964 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21965 reg_alternate_class into account.
21967 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21970 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21971 to consider in curr_insn_transform.
21973 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21975 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21976 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21977 and emit_mode_inner.
21979 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21981 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21982 argument to the overloaded builtin variants. Use the new flag to
21983 deprecate certain builtin variants.
21984 * config/s390/s390-builtin-types.def: Add new builtin types.
21985 * config/s390/s390-builtins.h: Support new flags field for
21986 overloaded builtins.
21987 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21988 (s390_macro_to_expand): Enable vector float data type.
21989 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21990 builtins by incrementing the __VEC__ version number.
21991 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21993 (s390_resolve_overloaded_builtin): Emit error messages depending
21994 on the builtin flags.
21995 * config/s390/s390.c (s390_expand_builtin): Support additional
21996 flags argument. Change error message to match the messages
21997 emitted in s390-c.c.
21998 * config/s390/s390.md: New UNSPEC_* constants.
21999 (op_type): Add new instruction types.
22000 * config/s390/vecintrin.h: Add new builtins and test data class
22002 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22003 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22004 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22005 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22006 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22007 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22009 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22010 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22011 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22012 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22014 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22015 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22016 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22017 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22019 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22020 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22021 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22022 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22023 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22024 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22025 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22027 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22028 ("vec_scatter_element<V_HW_4:mode>_DI")
22029 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22030 ("vec_fpint<mode>", "vflls")
22031 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22032 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22033 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22034 ("*vec_cmphe<mode>_cc"): ... these.
22036 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22037 mode constant instead of magic value.
22039 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22041 * config/s390/s390.c (s390_expand_vec_compare): Support other
22042 vector floating point modes than just V2DF.
22043 (s390_expand_vcond): Likewise.
22044 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22045 (s390_cannot_change_mode_class): Prevent mode changes between TF
22046 and V1TF in vector registers.
22047 * config/s390/s390.md (DF, SF): New mode attributes.
22048 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22049 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22050 SFmode support for VRs.
22051 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22053 (VFT, VF_HW): New mode iterators.
22054 (vw, sdx): New mode attributes.
22055 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22056 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22057 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22058 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22059 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22060 also the new vector floating point modes. Renaming to ...
22062 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22063 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22064 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22065 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22066 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22067 ("vec_unordered<mode>"): ... these.
22069 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22070 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22071 ("*vec_extendv2df"): New insn definitions.
22073 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22075 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22076 ("mulditi3_2", "*muldi3_sign"): New patterns.
22077 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22078 rename the pattern definition.
22080 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22082 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22084 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22086 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22088 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22089 instruction if possible.
22090 * config/s390/vector.md (vec_halfnumelts): New mode
22092 ("*vec_vllezlf<mode>"): New pattern.
22094 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22096 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22097 ("popcountv4si2", "popcountv2di2"): Rename to ...
22098 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22099 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22101 ("popcount<mode>2_vxe"): New pattern.
22103 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22105 * common/config/s390/s390-common.c (processor_flags_table): Add
22107 * config.gcc: Add arch12.
22108 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22109 Default to arch12 for unknown CPU model numbers.
22110 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22111 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22112 PROCESSOR_max sanity check.
22113 * config/s390/s390-opts.h (enum processor_type): Add
22115 * config/s390/s390.c (processor_table): Add arch12.
22116 (s390_expand_builtin): Add check for B_VXE flag.
22117 (s390_issue_rate): Add PROCESSOR_ARCH12.
22118 (s390_get_sched_attrmask): Likewise.
22119 (s390_get_unit_mask): Likewise.
22120 (s390_sched_score): Enable z13 scheduling for arch12.
22121 (s390_sched_reorder): Likewise.
22122 (s390_sched_variable_issue): Likewise.
22123 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22125 (s390_tune_attr): Use z13 scheduling also for arch12.
22126 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22127 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22128 (TARGET_VXE_P): New macros.
22129 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22130 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22131 * config/s390/s390.opt: Add arch12 as processor_type.
22133 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22135 * config/s390/s390.md
22136 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22137 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22138 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22140 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22141 Rename expanders to ...
22143 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22144 ("fixuns_truncdddi2_emu"): ... these.
22146 ("fixuns_trunc<mode>si2_emu"): New expander.
22148 ("*fixuns_truncdfdi2_z13"): Rename to ...
22149 ("*fixuns_truncdfdi2_vx"): ... this.
22151 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22153 * config/s390/2964.md: Remove the single element vector compare
22154 instructions which are no longer used.
22155 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22157 (s390_canonicalize_comparison): Remove handling of DFmode
22159 (s390_expand_vec_compare_scalar): Remove function.
22160 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22161 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22163 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22165 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22167 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22169 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22170 will anyway by matched by mov<mode>_64dfp.
22172 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22174 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22175 vlef/vstef. Add missing operand to vleif.
22177 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22179 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22180 pair for all vector types with 64 bit elements.
22181 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22182 * config/s390/vector.md (V_HW_64): ... here.
22183 (V_128_NOSINGLE): New mode iterator.
22184 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22185 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22186 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22187 ("*vec_load_pairv2di"): Change to ...
22188 ("*vec_load_pair<mode>"): ... this one.
22190 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22192 * config/s390/constraints.md: Add comments.
22193 (jKK): Reject element sizes > 8 bytes.
22194 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22196 * config/s390/s390.md: Add the s_operand checks formerly in
22197 s390_split_ok_p to various splitters where they are still
22199 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22200 for 128 bit vectors. Plus two splitters.
22202 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22204 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22207 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22210 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22211 error if the boundary argument is not constant.
22213 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22215 PR rtl-optimization/80112
22216 * loop-doloop.c (doloop_condition_get): Don't check condition
22217 if cmp isn't SET with IF_THEN_ELSE src.
22219 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22221 PR tree-optimization/80158
22222 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22223 replacing a candidate statement, also replace it for the
22224 candidate's alternate interpretation.
22225 (replace_rhs_if_not_dup): Likewise.
22226 (replace_one_candidate): Likewise.
22228 2017-03-24 Richard Biener <rguenther@suse.de>
22230 PR tree-optimization/80167
22231 * graphite-isl-ast-to-gimple.c
22232 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22234 (translate_isl_ast_to_gimple::get_rename): Likewise.
22236 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22238 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22239 handling of certain combinations of target options, including the
22240 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22241 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22243 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22246 * config/arm/arm.md (*load_multiple): Add reload_completed to
22247 matching condition.
22249 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22250 Richard Biener <rguenth@suse.de>
22252 PR tree-optimization/79908
22253 PR tree-optimization/80136
22254 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22255 been cast away, gimplify_and_add suffices.
22257 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22259 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22261 2017-03-23 Richard Biener <rguenther@suse.de>
22263 PR tree-optimization/80032
22264 * gimplify.c (gimple_push_cleanup): Forced unconditional
22265 cleanups still have to go to the conditional_cleanups
22268 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22270 PR tree-optimization/80072
22271 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22273 (next_operand_entry_id): Change type to unsigned int.
22274 (sort_by_operand_rank): Make sure to return the right return value
22275 even if unsigned fields are bigger than INT_MAX.
22276 (struct oecount): Change cnt and id type to unsigned int.
22277 (oecount_hasher::equal): Formatting fix.
22278 (oecount_cmp): Make sure to return the right return value
22279 even if unsigned fields are bigger than INT_MAX.
22280 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22283 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22284 TREE_READONLY on result if writing it more than once.
22287 * doc/invoke.texi (-fsanitize=thread): Document that with
22288 -fnon-call-exceptions atomics are not able to throw
22292 * tsan.c: Include tree-eh.h.
22293 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22294 maybe_clean_or_replace_eh_stmt where needed.
22295 (instrument_memory_accesses): Add cfg_changed argument.
22296 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22297 if it returned true.
22298 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22300 PR rtl-optimization/63191
22301 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22302 wrapper function, moved the whole old content into ...
22303 (ix86_delegitimize_address_1): ... this. New inline function.
22304 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22305 true as last argument instead of ix86_delegitimize_address.
22307 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22309 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22310 cortexa57_branch_cost.
22312 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22314 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22316 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22319 * doc/md.texi (Constraints): Document wA constraint.
22320 * config/rs6000/constraints.md (wA): New.
22321 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22322 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22323 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22324 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22326 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22329 * gimplify.c (is_oacc_declared): New function.
22330 (oacc_default_clause): Use it to set default flags for acc declared
22331 variables inside parallel regions.
22332 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22333 declared variables.
22334 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22335 declare attribute to any decl as necessary.
22337 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22340 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22341 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22342 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22343 (arm_arch_lpae): This.
22344 * config/arm/arm.c (arm_arch7ve): Rename into ...
22345 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22346 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22349 2017-03-22 Martin Liska <mliska@suse.cz>
22352 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22353 error message instead of an ICE.
22355 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22357 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22359 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22361 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22363 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22366 2017-03-21 Martin Sebor <msebor@redhat.com>
22368 * doc/extend.texi: Use "cannot" instead of "can't."
22369 * doc/hostconfig.texi: Same.
22370 * doc/install.texi: Same.
22371 * doc/invoke.texi: Same.
22372 * doc/loop.texi: Same.
22373 * doc/md.texi: Same.
22374 * doc/objc.texi: Same.
22375 * doc/rtl.texi: Same.
22376 * doc/tm.texi: Same.
22377 * doc/tm.texi.in: Same.
22378 * doc/trouble.texi: Same.
22380 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22383 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22384 (collect_checksum_attributes): Set it.
22385 (die_checksum_ordered): Use it.
22387 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22389 PR tree-optimization/79908
22390 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22391 change: For a VA_ARG whose LHS has been cast away, use
22392 force_gimple_operand to construct the side effects.
22394 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22396 PR translation/80001
22397 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22398 more amenable to translation.
22399 (oacc_loop_auto_partitions): Likewise.
22401 2017-03-21 Marek Polacek <polacek@redhat.com>
22402 Martin Sebor <msebor@redhat.com>
22404 PR tree-optimization/80109
22405 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22406 on INTEGRAL_TYPE_P.
22408 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22409 Segher Boessenkool <segher@kernel.crashing.org>
22412 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22413 check reg_used_between_p between insn and one of succ or succ2
22414 depending on if succ is artificial insn not inserted into insn
22417 2017-03-21 Martin Liska <mliska@suse.cz>
22419 PR gcov-profile/80081
22420 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22421 * doc/gcc.texi: Include gcov-dump stuff.
22422 * doc/gcov-dump.texi: New file.
22424 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22426 PR rtl-optimization/79150
22427 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22428 conditional jump, if the jump is the last insn of the loop.
22430 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22431 Richard Biener <rguenth@suse.de>
22433 PR tree-optimization/79908
22434 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22435 been cast away, use force_gimple_operand to construct the side
22438 2017-03-21 Martin Liska <mliska@suse.cz>
22440 PR libfortran/79956
22441 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22444 2017-03-21 Brad Spengler <spender@grsecurity.net>
22447 * plugin.c (htab_hash_plugin): New function.
22448 (add_new_plugin): Use it and adjust.
22449 (parse_plugin_arg_opt): Adjust.
22450 (init_one_plugin): Likewise.
22452 2017-03-21 Richard Biener <rguenther@suse.de>
22454 PR tree-optimization/80032
22455 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22456 if set force the cleanup to happen unconditionally.
22457 (gimplify_target_expr): Push inserted clobbers with force_uncond
22458 to avoid them being removed by control-dependent DCE.
22460 2017-03-21 Richard Biener <rguenther@suse.de>
22462 PR tree-optimization/80122
22463 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22464 va_arg_pack_len when the inlined call stmt requires pack
22466 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22468 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22471 * tsan.c (instrument_builtin_call): If the memory model argument
22472 is not a constant, assume it is valid.
22475 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22478 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22480 PR rtl-optimization/79910
22481 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22482 if its dest is used by an insn before I2 (other than the combined
22483 insns themselves, which are properly handled already).
22485 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22488 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22490 * combine.c (record_used_regs): New static function.
22491 (try_combine): Handle situations where there is an additional
22492 instruction between I2 and I3 which needs to have a LOG_LINK
22496 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22498 * combine.c (try_combine): Delete redundant i1 test. Call
22499 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22501 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22504 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22505 alternatives 13/14.
22507 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22509 PR tree-optimization/80054
22510 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22511 the optimization if a PHI or any of its arguments is not dominated
22512 by the candidate's basis. Use gphi* rather than gimple* as
22514 (replace_profitable_candidates): Clean up a gimple* variable that
22515 should be a gphi* variable.
22517 2017-03-20 Martin Sebor <msebor@redhat.com>
22520 * doc/extend.texi (attribute constructor): Document present limitation.
22522 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22525 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22526 __POWER9_VECTOR__ #ifdef control, change template definition to
22527 use Power9-specific built-in function.
22528 (vec_any_eq): Likewise.
22529 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22530 to control outcomes from this test.
22531 (vector_ae_<mode>p): For VEC_F modes, likewise.
22533 2017-03-20 Ian Lance Taylor <iant@google.com>
22535 * config/i386/i386.c (ix86_function_regparm): Save an extra
22536 register for -fsplit-stack with DECL_STATIC_CHAIN.
22538 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22541 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22542 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22544 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22546 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22548 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22551 2017-03-20 Marek Polacek <polacek@redhat.com>
22554 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22556 2017-03-20 Richard Biener <rguenther@suse.de>
22558 PR tree-optimization/80113
22559 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22560 allocate extra SSA name for PHI def.
22561 (add_close_phis_to_outer_loops): Likewise.
22562 (add_close_phis_to_merge_points): Likewise.
22563 (copy_loop_close_phi_args): Likewise.
22564 (copy_cond_phi_nodes): Likewise.
22566 2017-03-20 Martin Liska <mliska@suse.cz>
22568 PR middle-end/79753
22569 * tree-chkp.c (chkp_build_returned_bound): Do not build
22570 returned bounds for a LHS that's not a BOUNDED_P type.
22572 2017-03-20 Martin Liska <mliska@suse.cz>
22576 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22577 COMPLEX_CST and VECTOR_CST.
22579 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22582 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22583 target operand. A new splitter adds the clobber statement in case
22584 the target operand is dead anyway.
22586 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22588 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22589 to age-old versions of binutils and glibc.
22591 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22593 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22595 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22597 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22599 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22601 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22602 requirement for binutils 2.13.
22604 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22606 * combine.c (try_combine): Delete redundant i1 test. Call
22607 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22609 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22611 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22612 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22614 <riscv64-*-elf>: Re-arrange section
22615 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22616 <riscv32-*-linux>: Likewise.
22617 <riscv64-*-elf>: Likewise
22618 <riscv64-*-linux>: Likewise.
22620 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22623 * aarch64.opt(verbose-cost-dump): Fix typo.
22625 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22628 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22629 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22631 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22633 * reload.c (find_reloads): When reloading a nonoffsettable address,
22634 use RELOAD_OTHER for it and its address reloads.
22636 PR rtl-optimization/79910
22637 * combine.c (record_used_regs): New static function.
22638 (try_combine): Handle situations where there is an additional
22639 instruction between I2 and I3 which needs to have a LOG_LINK
22642 2017-03-17 Jeff Law <law@redhat.com>
22644 PR tree-optimization/71437
22645 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22646 conditional in the hash table first.
22647 (vrp_dom_walker::before_dom_children): Extract condition from
22648 ASSERT_EXPR. Record condition, its inverion and any implied
22649 conditions as well.
22651 2017-03-17 Marek Polacek <polacek@redhat.com>
22652 Markus Trippelsdorf <markus@trippelsdorf.de>
22654 PR tree-optimization/80079
22655 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22658 2017-03-17 Richard Biener <rguenther@suse.de>
22660 PR middle-end/80075
22661 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22662 Properly verify the LHS before the RHS possibly claims to be
22664 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22667 2017-03-17 Martin Jambor <mjambor@suse.cz>
22669 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22670 (List of -O2 options): Likewise.
22671 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22674 2017-03-17 Tom de Vries <tom@codesourcery.com>
22676 * gcov-dump.c (print_usage): Print bug_report_url.
22678 2017-03-17 Richard Biener <rguenther@suse.de>
22680 PR middle-end/80050
22681 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22682 (parser::peek): Likewise.
22684 2017-03-17 Richard Biener <rguenther@suse.de>
22686 PR tree-optimization/80048
22687 * sese.c (free_sese_info): Properly release rename_map and
22688 copied_bb_map elements.
22690 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22692 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22693 Add linked-list forward and backlinks. Insert on
22694 construction, remove on destruction.
22695 (class pass_store_merging): Add m_stores_head field.
22696 (pass_store_merging::terminate_and_process_all_chains):
22697 Iterate over m_stores_head list.
22698 (pass_store_merging::terminate_all_aliasing_chains):
22700 (pass_store_merging::execute): Check for debug stmts first.
22701 Push new chains onto the m_stores_head stack.
22703 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22706 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22707 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22708 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22710 2017-03-16 Jeff Law <law@redhat.com>
22712 PR tree-optimization/71437
22713 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22714 member function. Implementation moved into after_dom_children
22715 member function and into the threader's thread_outgoing_edges
22717 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22718 some code into new thread_outgoing_edges.
22719 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22720 definition. Simplify marker handling (do it here). Assume we always
22721 have the available expression and the const/copies tables.
22722 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22724 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22725 * tree-vrp.c (equiv_stack): No longer file scoped.
22726 (vrp_dom_walker): New class.
22727 (vrp_dom_walker::before_dom_children): New member function.
22728 (vrp_dom_walker::after_dom_children): Likewise.
22729 (identify_jump_threads): Setup domwalker. Use it rather than
22730 walking edges in a random order by hand. Simplify setup/finalization.
22731 (finalize_jump_threads): Remove.
22732 (vrp_finalize): Do not call identify_jump_threads here.
22733 (execute_vrp): Do it here instead and call thread_through_all_blocks
22736 PR tree-optimization/71437
22737 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22739 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22741 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22742 (dom_opt_dom_walker::thread_across_edge): Remove
22743 handle_dominating_asserts argument. All callers changed.
22744 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22745 changes. Remove calls to lhs_of_dominating_assert. Other
22746 uses of handle_dominating_asserts turn into unconditional code
22747 (simplify_control_stmt_condition_1): Likewise.
22748 (simplify_control_stmt_condition): Likewise.
22749 (thread_through_normal_block, thread_across_edge): Likewise.
22750 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22751 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22752 object if it is not an SSA_NAME.
22753 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22754 before calling into the VRP specific simplifiers.
22755 (identify_jump_threads): Remove handle_dominating_asserts
22758 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22761 * tree-diagnostic.c (default_tree_printer): No longer static.
22762 * tree-diagnostic.h (default_tree_printer): New prototype.
22764 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22766 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22767 Change ins into fmov.
22769 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22771 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22772 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22773 Use h_con constraint for operand 1.
22774 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22775 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22777 2017-03-15 Jeff Law <law@redhat.com>
22779 PR tree-optimization/71437
22780 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22781 (record_temporary_equivalences): Use it.
22783 PR tree-optimization/71437
22784 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22785 tree-ssa-scopedtables.
22786 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22787 (record_conditions, record_cond, vuse_eq): Likewise.
22788 (record_edge_info): Adjust to API tweak of record_conditions.
22789 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22790 (record_temporary_equivalences, optimize_stmt): Likewise.
22791 (eliminate_redundant_computations): Likewise.
22792 (record_equivalences_from_stmt): Likewise.
22793 * tree-ssa-scopedtables.c: Include options.h and params.h.
22794 (vuse_eq): New function, moved from tree-ssa-dom.c
22795 (build_and_record_new_cond): Likewise.
22796 (record_conditions): Likewise. Accept vector of conditions rather
22797 than edge_equivalence structure for first argument.
22798 for the first argument.
22799 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22800 from tree-ssa-dom.c.
22801 (avail_exprs_stack::record_cond): Likewise.
22802 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22803 from tree-ssa-dom.c.
22804 (avail_exprs_stack): Add new member functions lookup_avail_expr
22806 (record_conditions): Declare.
22808 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22811 * lra-constraints.c (process_alt_operands): Increase reject for
22812 reloading an input/output operand.
22814 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22817 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22818 insns to convert from signed/unsigned char/short to IEEE 128-bit
22820 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22822 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22825 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22826 subreg of inner mode for values already in registers.
22828 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22830 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22831 iteration reg is used after the loop.
22833 2017-03-14 Martin Sebor <msebor@redhat.com>
22835 PR tree-optimization/79800
22836 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22837 precision in negative-positive range.
22838 (format_floating): Call non-const overload with adjusted precision.
22840 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22843 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22846 2017-03-14 Martin Sebor <msebor@redhat.com>
22848 PR middle-end/80020
22849 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22850 * builtins.def (aligned_alloc): Use it.
22853 * Makefile.in (GTFILES): Add calls.c.
22854 * calls.c: Include "gt-calls.h".
22856 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22858 PR rtl-optimization/79728
22859 * regs.h (struct target_regs): New field
22860 x_contains_allocatable_regs_of_mode.
22861 (contains_allocatable_regs_of_mode): New macro.
22862 * reginfo.c (init_reg_sets_1): Initialize it, and change
22863 contains_reg_of_mode so it includes global regs as well.
22864 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22865 rather than contains_regs_of_mode.
22867 2017-03-14 Martin Liska <mliska@suse.cz>
22869 * doc/invoke.texi: Document options that can't be combined with
22870 -fcheck-pointer-bounds.
22872 2017-03-14 Martin Liska <mliska@suse.cz>
22874 PR middle-end/79831
22875 * doc/invoke.texi (-Wchkp): Document the option.
22877 2017-03-14 Martin Liska <mliska@suse.cz>
22879 * Makefile.in: Install gcov-dump.
22881 2017-03-14 Martin Liska <mliska@suse.cz>
22883 * multiple_target.c (expand_target_clones): Bail out for
22884 an invalid attribute.
22886 2017-03-14 Richard Biener <rguenther@suse.de>
22888 * alias.c (struct alias_set_entry): Pack properly.
22889 * cfgloop.h (struct loop): Likewise.
22890 * cse.c (struct set): Likewise.
22891 * ipa-utils.c (struct searchc_env): Likewise.
22892 * loop-invariant.c (struct invariant): Likewise.
22893 * lra-remat.c (struct cand): Likewise.
22894 * recog.c (struct change_t): Likewise.
22895 * rtl.h (struct address_info): Likewise.
22896 * symbol-summary.h (function_summary): Likewise.
22897 * tree-loop-distribution.c (struct partition): Likewise.
22898 * tree-object-size.c (struct object_size_info): Likewise.
22899 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22900 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22901 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22902 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22903 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22904 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22905 (struct _stmt_vec_info): Likewise.
22907 2017-03-14 Martin Liska <mliska@suse.cz>
22910 * multiple_target.c (create_dispatcher_calls): Check that
22911 a target can create a function dispatcher.
22913 2017-03-14 Martin Liska <mliska@suse.cz>
22916 * multiple_target.c (expand_target_clones): Drop local.local
22917 flag for default implementation.
22919 2017-03-14 Richard Biener <rguenther@suse.de>
22921 PR tree-optimization/80030
22922 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22924 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22926 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22927 gcc_fallthrough() instead of __attribute__((fallthrough));
22929 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22931 * doc/gcc.texi: Remove "up" link to (DIR).
22932 * doc/gccint.texi: Ditto.
22934 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22936 * doc/install.texi (Specific) <avr>: Remove reference to
22939 2017-03-13 Jeff Law <law@redhat.com>
22941 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22942 attribute rather than comments.
22944 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22945 match_scratch operand is highest.
22947 2017-03-13 Martin Liska <mliska@suse.cz>
22949 PR middle-end/78339
22950 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22951 is a CHKP clone, use original declaration.
22953 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22955 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22956 (arc_conditional_register_usage): Use a different allocation order
22957 when optimizing for size.
22958 * common/config/arc/arc-common.c (arc_option_optimization_table):
22959 Section anchors default on when optimizing for size.
22961 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22963 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22965 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22967 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22968 * config/arc/arc.md (cpu_facility): Add cd variant.
22969 (*movqi_insn): Add code density variant.
22970 (*movhi_insn): Likewise.
22971 (*movqi_insn): Likewise.
22972 (*addsi3_mixed): Likewise.
22973 (subsi3_insn): Likewise.
22975 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22977 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22979 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22981 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22982 expressions with MINUS and UNARY ops.
22984 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22987 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22989 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22990 between vec_select and vector argument.
22991 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22992 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22993 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22994 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22995 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22996 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22998 2017-03-13 Richard Biener <rguenther@suse.de>
23001 * params.def (vect-max-peeling-for-alignment): Fix typo.
23003 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23005 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23006 issue that only occurred with binutils below 2.18.
23008 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23010 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23011 refer to binutils 2.11/2.12 minimum.
23013 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23015 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23016 ftp.kernel.org and simplify binutils requirement.
23018 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23020 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23022 (Optimize Options): Ditto. Also remove redundancy.
23024 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23026 PR translation/79848
23027 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23029 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23030 to G_ to avoid double translation.
23032 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23034 PR translation/79923
23035 * auto-profile.c (get_combined_location): Convert leading
23036 character of diagnostics to lower case and remove trailing period.
23037 (read_profile): Likewise for various diagnostics.
23038 * config/arm/arm.c (arm_option_override): Remove trailing period
23039 from various diagnostics.
23040 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23041 (msp430_expand_delay_cycles): Likewise.
23043 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23046 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23047 full command-line argument, rather than just "str".
23048 (aarch64_validate_march): Likewise.
23049 (aarch64_validate_mtune): Likewise.
23051 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23053 PR rtl-optimization/78911
23054 * lra-assigns.c (must_not_spill_p): New function.
23055 (spill_for): Use it.
23057 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23059 PR tree-optimization/79981
23060 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23061 ATOMIC_COMPARE_EXCHANGE ifn result.
23062 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23063 IFN_ATOMIC_COMPARE_EXCHANGE.
23065 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23068 * opts.c (parse_sanitizer_options): Add missing question mark to
23069 "did you mean" message.
23071 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23073 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23075 (VMULEUH_UNS): Likewise.
23076 (VMULOUB_UNS): Likewise.
23077 (VMULOUH_UNS): Likewise.
23078 * config/rs6000/rs6000.c (builtin_function_type): Remove
23079 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23081 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23084 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23085 x with result of extra_parsing_for_operand_code_0.
23086 (function_reader::extra_parsing_for_operand_code_0): Convert
23087 return type from void to rtx, returning x. When reading
23088 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23089 larger size containing struct block_symbol.
23091 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23093 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23094 -mfloat128-hardware without -m64.
23096 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23099 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23100 entries to the case statement that marks unsigned arguments to
23101 overloaded functions.
23103 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23105 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23106 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23108 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23111 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23112 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23114 2017-03-10 Martin Liska <mliska@suse.cz>
23118 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23119 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23122 2017-03-10 Olivier Hainque <hainque@adacore.com>
23124 * tree-switch-conversion (array_value_type): Start by resetting
23125 candidate type to it's main variant.
23127 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23129 PR rtl-optimization/79909
23130 * combine.c (try_combine): Use simplify_replace_rtx on individual
23131 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23132 of the whole CALL_INSN_FUNCTION_USAGE.
23134 PR tree-optimization/79972
23135 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23136 get_range_info on SSA_NAMEs. Formatting fixes.
23138 2017-03-10 Richard Biener <rguenther@suse.de>
23139 Jakub Jelinek <jakub@redhat.com>
23141 PR tree-optimization/77975
23142 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23143 edge to be constant.
23144 (get_val_for): For constant x return it. Formatting fix.
23145 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23146 would use the same bases as the current one.
23148 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23150 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23151 instead of vec_select for V1TImode.
23152 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23154 (VSX_LE_128): Add V1TI to this mode iterator.
23155 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23156 (*vsx_le_perm_store_<mode>): Likewise.
23157 (pre-reload splitter for VSX stores): Likewise.
23158 (post-reload splitter for VSX stores): Likewise.
23159 (*vsx_xxpermdi2_le_<mode>): Likewise.
23160 (*vsx_lxvd2x2_le_<mode>): Likewise.
23161 (*vsx_stxvd2x2_le_<mode>): Likewise.
23163 2017-03-09 Michael Eager <eager@eagercon.com>
23165 Correct failures with --enable-checking=yes,rtl.
23167 * config/microblaze/microblaze.c (microblaze_expand_shift):
23168 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23169 test for const0_rtx.
23170 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23171 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23173 2017-03-09 Richard Biener <rguenther@suse.de>
23175 PR tree-optimization/79977
23176 * graphite-scop-detection.c (scop_detection::merge_sese):
23177 Handle the case of extra exits to blocks dominating the entry.
23179 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23181 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23184 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23186 PR rtl-optimization/79949
23187 * lra-constraints.c (process_alt_operands): Check memory when
23188 trying to predict a cycle. Print about the overall increase.
23190 2017-03-09 Richard Biener <rguenther@suse.de>
23192 PR middle-end/79971
23193 * gimple-expr.c (useless_type_conversion_p): Preserve
23194 TYPE_SATURATING for fixed-point types.
23196 2017-03-09 Richard Biener <rguenther@suse.de>
23199 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23200 alignment of BLKmode params.
23202 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23205 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23206 (VALL_NO_V2Q): Likewise.
23208 * config/aarch64/aarch64-simd.md
23209 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23211 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23212 VALL_NO_V2Q mode iterator.
23213 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23215 2017-03-09 Martin Liska <mliska@suse.cz>
23217 PR tree-optimization/79631
23218 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23219 tree_int_cst_sign_bit just for INTEGER constants.
23221 2017-03-09 Martin Liska <mliska@suse.cz>
23225 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23228 2017-03-09 Marek Polacek <polacek@redhat.com>
23231 * tree.c (inchash::add_expr): Handle TREE_VEC.
23233 2017-03-09 Martin Liska <mliska@suse.cz>
23236 (chkp_narrow_size_and_offset): New function.
23237 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23238 (void chkp_parse_bit_field_ref): New function.
23239 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23240 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23242 2017-03-09 Martin Liska <mliska@suse.cz>
23245 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23246 (chkp_find_bounds_1): Remove gcc_unreachable.
23248 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23251 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23252 BUILT_IN_SYNC*, determine the access type from the size suffix and
23253 always build a MEM_REF with that type. Handle forgotten
23254 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23257 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23258 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23259 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23260 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23261 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23262 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23263 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23264 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23265 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23266 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23267 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23268 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23269 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23270 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23271 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23272 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23273 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23274 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23275 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23276 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23277 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23278 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23279 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23280 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23281 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23282 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23283 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23284 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23285 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23286 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23287 definitions outside of __OPTIMIZE__ guarded section.
23290 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23291 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23292 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23293 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23296 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23298 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23299 ("vfenez<mode>"): Add missing constraints.
23301 2017-03-08 Martin Sebor <msebor@redhat.com>
23304 * config/nds32/nds32.c (nds32_option_override):
23305 Fix misspelled diagnostic.
23307 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23310 * gimplify.c (gimplify_omp_for): Replace index var in outer
23311 taskloop statement with an artificial variable and add
23312 OMP_CLAUSE_PRIVATE clause for it.
23314 2017-03-08 Richard Biener <rguenther@suse.de>
23316 PR tree-optimization/79955
23317 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23318 for accesses that are completely outside of the variable.
23320 2017-03-08 Andrew Haley <aph@redhat.com>
23322 PR tree-optimization/79943
23323 * tree-ssa-loop-split.c (compute_new_first_bound): When
23324 calculating the new upper bound, (END-BEG) should be added, not
23327 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23329 * config/avr/avr.md (setmemhi): Make sure match_dup
23330 operand number comes before match_scratch.
23332 2017-03-08 Richard Biener <rguenther@suse.de>
23334 PR tree-optimization/79920
23335 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23336 with ncopies == 1 to ...
23337 (vect_transform_slp_perm_load): ... here. Properly compute
23338 all element loads by iterating VF times over the group. Do
23339 not handle ncopies (computed in a broken way) in
23340 vect_create_mask_and_perm.
23342 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23345 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23346 is a uniform vector, use uniform_vector_p return value instead of
23347 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23349 2017-03-07 Marek Polacek <polacek@redhat.com>
23351 PR middle-end/79809
23352 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23353 (alloca_call_type): Likewise.
23355 2017-03-07 Martin Liska <mliska@suse.cz>
23357 * gcov.c (process_args): Put comment to correct location.
23359 2017-03-07 Martin Liska <mliska@suse.cz>
23361 PR middle-end/68270
23362 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23363 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23364 (chkp_narrow_bounds_for_field): Likewise.
23365 (chkp_parse_array_and_component_ref): Pass one more argument to
23368 2017-03-07 Richard Biener <rguenther@suse.de>
23370 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23373 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23375 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23376 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23378 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23381 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23382 to end of description.
23383 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23385 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23387 PR rtl-optimization/79901
23388 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23390 (*avx512f_<code><mode>3<mask_name>): ... this.
23391 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23392 iterator instead of VI8_AVX2_AVX512BW.
23394 PR rtl-optimization/79901
23395 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23396 min/max expander, expand it using expand_vec_cond_expr.
23399 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23402 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23405 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23406 to void * for PCH reasons.
23407 * dwarf2out.c (output_loc_operands, output_die): Cast
23408 v.val_vec.array to unsigned char *.
23410 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23413 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23416 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23418 PR rtl-optimization/79571
23419 * lra-constraints.c (process_alt_operands): Calculate static
23420 reject and subtract it from overall when only addresses will be
23423 2017-03-06 Julia Koval <julia.koval@intel.com>
23426 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23427 incoming stack boundary to 128 for 64-bit targets.
23429 2017-03-06 Richard Biener <rguenther@suse.de>
23431 PR tree-optimization/79894
23432 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23433 to NULL after folding it.
23435 2017-03-06 Richard Biener <rguenther@suse.de>
23437 PR tree-optimization/79824
23438 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23439 check disabling peeling for gaps.
23441 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23443 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23444 attributes): Document gettimeofday.
23446 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23448 * config/s390/s390.c (s390_option_override_internal): Set
23449 PARAM_MIN_VECT_LOOP_BOUND
23451 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23453 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23454 * config/s390/s390.md: Likewise.
23456 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23459 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23460 (<avx2_avx512>_perm<mode>): Rename to ...
23461 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23463 (<avx512>_perm<mode>_mask): Rename to ...
23464 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23466 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23467 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23468 instead of VI8F_256_512.
23469 (avx512f_perm<mode>): New define_expand.
23470 (avx512f_perm<mode>_mask): Likewise.
23471 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23472 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23474 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23476 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23477 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23479 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23481 2017-03-06 Martin Liska <mliska@suse.cz>
23484 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23485 when having a SSA NAME w/o VAR_DECL assigned to it.
23487 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23489 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23490 msa_dpsub_<su>_d): Fix MODE for vec_select.
23492 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23494 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23496 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23498 2017-03-06 Richard Biener <rguenther@suse.de>
23500 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23501 * plugin.c (register_plugin_info): Likewise.
23502 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23504 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23506 * config/i386/sse.md (sse_storehps, sse_storelps,
23507 avx_<castmode><avxsizesuffix>_<castmode>,
23508 avx512f_<castmode><avxsizesuffix>_<castmode>,
23509 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23510 in condition that at least one operand is not a MEM.
23512 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23514 PR middle-end/79805
23515 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23516 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23518 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23519 gimple_call_nothrow_p flag based on whether original builtin can throw.
23520 If it can, emit following stmts on the fallthrough edge.
23521 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23522 don't create new bb if inserting just debug stmts on the edge, try to
23523 insert them on the fallthru bb or just reset debug stmts.
23525 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23528 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23529 restore recog_data (including the operand rtxes inside it) around
23530 the call to get_insn_template.
23532 2017-03-03 Martin Sebor <msebor@redhat.com>
23534 PR tree-optimization/79699
23535 * context.c (context::~context): Free MPFR caches to avoid
23536 a memory leak on program exit.
23538 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23540 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23541 Use wide_int::ulow () instead of .elt (0).
23543 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23545 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23546 (*pushxf): Limit oF constraint to 32bit targets and add oC
23547 constraint for 64bit targets.
23548 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23549 (*pushdf): Change rmF constraint to rmC.
23551 2017-03-03 Martin Liska <mliska@suse.cz>
23553 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23554 Remove unused variable.
23556 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23559 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23560 is a memory operand, increase num_memory.
23561 (ix86_expand_args_builtin): Likewise.
23563 2017-03-03 Jan Hubicka <jh@suse.cz>
23566 * ipa-devirt.c (maybe_record_node): Properly handle
23567 __cxa_pure_virtual visibility.
23569 2017-03-03 Martin Liska <mliska@suse.cz>
23571 PR tree-optimization/79803
23572 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23574 (pass_loop_prefetch::execute): Disabled optimization if an
23575 assumption about L1 cache size is not met.
23577 2017-03-03 Martin Liska <mliska@suse.cz>
23579 PR rtl-optimization/79574
23580 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23581 (hash_scan_set): Likewise.
23582 (dump_hash_table): Likewise.
23583 (hoist_code): Likewise.
23585 2017-03-03 Richard Biener <rguenther@suse.de>
23587 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23588 in place of uhigh ().
23589 (fixed_convert_from_real): Likewise.
23591 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23594 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23596 2017-03-03 Richard Biener <rguenther@suse.de>
23598 PR middle-end/79818
23599 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23600 TYPE_OVERFLOW_UNDEFINED check.
23602 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23604 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23606 (vector_ae_<mode>_p): Likewise.
23607 (vector_nez_<mode>_p): Likewise.
23608 (vector_ne_v2di_p): Likewise.
23609 (vector_ae_v2di_p): Likewise.
23610 (vector_ne_<mode>_p): Likewise.
23611 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23613 (vsx_tsqrt<mode>2_fe): Likewise.
23615 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23618 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23620 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23622 PR rtl-optimization/79780
23623 * cprop.c (one_cprop_pass): When second and further conditional trap
23624 in a single basic block is turned into an unconditional trap, turn it
23625 into a deleted note to avoid RTL verification failures.
23627 2017-03-02 Richard Biener <rguenther@suse.de>
23629 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23631 2017-03-02 Richard Biener <rguenther@suse.de>
23633 PR tree-optimization/79345
23635 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23636 param and abort the walk, returning -1 if it is hit.
23637 (walk_aliased_vdefs): Take a limit param and pass it on.
23638 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23639 defaulting to 0 and return a signed int.
23640 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23641 (check_defs): New helper.
23642 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23643 about uninitialized memory.
23644 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23645 bogus uninitialized warning.
23646 (fixed_convert_from_real): Likewise.
23648 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23650 PR tree-optimization/66768
23651 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23652 iv_use if base object can't be determined.
23654 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23656 PR tree-optimization/79345
23657 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23658 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23659 (get_pattern_stats): Initialize it.
23660 * genemit.c (gen_expand): Verify match_scratch numbers come after
23661 match_operand/match_dup numbers.
23662 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23663 match_scratch numbers.
23664 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23666 * config/s390/s390.md (trunctdsd2): Likewise.
23668 2017-03-02 Richard Biener <rguenther@suse.de>
23670 * wide-int.h (wide_int_storage::operator=): Implement in terms
23673 2017-03-02 Richard Biener <rguenther@suse.de>
23675 PR tree-optimization/79777
23676 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23677 the to insert expression to sth existing.
23679 2017-03-01 Martin Sebor <msebor@redhat.com>
23681 PR middle-end/79692
23682 * gimple-ssa-sprintf.c
23683 (directive::known_width_and_precision): New function.
23684 (format_integer): Use it.
23685 (get_mpfr_format_length): Consider the full range of precision
23686 when computing %g output with the # flag. Set the likely byte
23687 count to 3 rather than 1 when precision is indeterminate.
23688 (format_floating): Correct the lower bound of precision.
23690 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23692 * doc/invoke.texi: Document default code model for 64-bit Linux.
23694 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23697 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23698 udiv rather than div since input pattern is unsigned.
23700 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23702 * config/i386/i386.c (print_reg): Warn for values of
23703 unsupported size in integer register.
23705 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23708 * config/rs6000/predicates.md (current_file_function_operand): Do
23709 not allow self calls to be local if the function is replaceable.
23711 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23714 * config/rs6000/altivec.h (vec_ctz and others): Change the
23715 preprocessor macro that controls conditional compilation from
23716 _ARCH_PWR9 to __POWER9_VECTOR__.
23717 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23718 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23719 control (instead of _ARCH_PWR9 control) so that template
23720 definition uses power9-specific function.
23721 (vec_any_eq): Likewise.
23722 (vec_all_ne): Change macro definition to use a power9-specific
23723 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23724 _ARCH_PWR9 control).
23725 (vec_any_eq) Likewise.
23726 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23727 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23728 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23729 support for xvcmpnedp instruction.
23730 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23731 macro expansion so that Power9 implementation of vec_all_ne does
23732 not use the AltiVec predicate framework.
23733 (VCMPNEH_P): Likewise.
23734 (VCMPNEW_P): Likewise.
23735 (VCMPNED_P): Likewise.
23736 (VCMPNEFP_P): Likewise.
23737 (VCMPNEDP_P): Likewise.
23738 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23739 implementation of vec_any_eq to not use AltiVec predicate
23741 (VCMPAEH_P): Likewise.
23742 (VCMPAEW_P): Likewise.
23743 (VCMPAED_P): Likewise.
23744 (VCMPAEFP_P): Likewise.
23745 (VCMPAEDP_P): Likewise.
23746 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23747 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23748 not use the AltiVec predicate framework.
23749 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23750 of vec_any_eq to not use AltiVec predicate framework.
23751 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23752 support for predefined __POWER9_VECTOR__ macro to indicate that
23753 Power9 instruction selection is enabled.
23754 (altivec_overloaded_builtins): Remove extraneous
23755 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23756 function argument types RS6000_BTI_bool_V16QI and
23757 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23758 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23759 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23760 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23761 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23762 Power9 for implementations of vec_cmpne. Change the signature for
23763 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23764 (representing vec_all_ne) to remove the previously described first
23765 argument of type RS6000_BTI_INTSI, as this was an artifact of
23766 reliance on the AltiVec predicate framework, which is no longer
23767 used in the implementation of these functions. Add
23768 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23769 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23770 since, unlike the AltiVec predicate framework implementation, we
23771 do not share function descriptors between vec_alle and vec_anyeq.
23772 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23773 set of modes that receive special treatment even when
23774 TARGET_P9_VECTOR is true. The special treatment emits code that
23775 does not depend on Power9 instructions.
23776 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23777 define_expand to not rely on AltiVec predicate framework.
23778 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23780 (vector_ne_v2di_p): Change this define_expand to not rely on
23781 AltiVec predicate framework.
23782 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23784 (vector_ne_<mode>_p): Change this define_expand to not rely on
23785 AltiVec predicate framework.
23786 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23788 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23789 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23790 define_insn pattern.
23791 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23792 define_insn pattern because the xvcmpne<VSs>. instruction is not
23794 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23795 instruction is not supported.
23797 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23799 * config/nvptx/nvptx.c: Include intl.h.
23801 2017-03-01 Martin Jambor <mjambor@suse.cz>
23804 * ipa-prop.h (ipa_bits): Removed field known.
23805 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23806 to pointers. Adjusted their comments to warn about their sharing.
23807 (ipcp_transformation_summary): Change bits to a vector of pointers.
23808 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23809 (ipa_get_ipa_bits_for_value): Declare.
23810 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23811 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23812 (ipa_bits_hash_table): Likewise.
23813 (ipa_vr_ggc_hash_traits): Likewise.
23814 (ipa_vr_hash_table): Likewise.
23815 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23816 being pointers and vr_known being removed.
23817 (ipa_set_jf_unknown): Likewise.
23818 (ipa_get_ipa_bits_for_value): New function.
23819 (ipa_set_jfunc_bits): Likewise.
23820 (ipa_get_value_range): New overloaded functions.
23821 (ipa_set_jfunc_vr): Likewise.
23822 (ipa_compute_jump_functions_for_edge): Use the above functions to
23823 construct bits and vr parts of jump functions.
23824 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23825 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23827 (ipcp_grow_transformations_if_necessary): Also allocate
23828 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23830 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23831 them. Fix too long lines.
23832 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23833 vr_known being removed.
23834 (ipa_read_jump_function): Use new setter functions to construct bits
23835 and vr parts of jump functions or set them to NULL.
23836 (write_ipcp_transformation_info): Adjust for bits being pointers.
23837 (read_ipcp_transformation_info): Likewise.
23838 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23840 Include gt-ipa-prop.h.
23841 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23843 (ipcp_store_bits_results): Likewise.
23844 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23845 Do not write to existing jump functions but use a temporary instead.
23847 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23850 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23851 attempt to use its first operand as BIT_FIELD_REF base.
23853 2017-03-01 Richard Biener <rguenther@suse.de>
23855 PR middle-end/79721
23856 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23857 interpolating formula in wrapping arithmetic.
23858 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23860 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23862 PR tree-optimization/79734
23863 * tree-vect-generic.c (expand_vector_condition): Optimize
23864 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23865 Handle VEC_COND_EXPR where comparison has different inner width from
23866 type's inner width.
23868 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23870 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23871 markup, and similar issues. Remove @opindex entries for things
23872 that aren't options. Add missing -mmpy-option entries.
23874 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23876 PR tree-optimization/79737
23877 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23878 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23879 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23880 instead of byte_size. Formatting fix.
23881 (shift_bytes_in_array_right): Formatting fix.
23883 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23886 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23887 condition on optimize for the leaf function test.
23889 2017-02-28 Martin Liska <mliska@suse.cz>
23892 * read-rtl-function.c (function_reader::handle_unknown_directive):
23893 Bail out when one uses -flto.
23895 2017-02-28 Martin Liska <mliska@suse.cz>
23897 * common.opt: Replace space with tabular for options of <number>
23899 * config/i386/i386.opt: Show <number> value for
23900 -mlarge-data-threshold.
23901 * opts.c (print_filtered_help): Do not display number in hexadecimal
23904 2017-02-28 Martin Liska <mliska@suse.cz>
23906 * common.opt: Fix --help=option -Q for options which are of
23909 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23911 * config/i386/i386.c (print_reg): Error out for values
23912 of 8-bit size in invalid integer register.
23914 2017-02-28 Martin Sebor <msebor@redhat.com>
23916 PR tree-optimization/79691
23917 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23919 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23922 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23923 gcc_unreachable with output_operand_lossage.
23925 2017-02-28 Richard Biener <rguenther@suse.de>
23927 PR tree-optimization/79740
23928 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23930 (visit_nary_op): Insert the nary into the hashtable if we
23931 pattern-matched sth.
23932 * tree-ssa-pre.c (eliminate_insert): Robustify.
23934 2017-02-28 Richard Biener <rguenther@suse.de>
23936 PR middle-end/79731
23937 * fold-const.c (decode_field_reference): Reject out-of-bound
23940 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23942 * config/i386/i386.c: Include intl.h.
23943 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23944 instead of just cond ? "..." : "...".
23945 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23946 * coverage.c (read_counts_file): Likewise.
23947 * omp-offload.c: Include intl.h.
23948 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23949 of just cond ? "..." : "...".
23950 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23951 of just cond ? "..." : "...".
23953 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23956 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23958 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23959 'tune for' CPU name.
23960 * config/arm/arm-cpu-data.h: Regenerated.
23962 2017-02-28 Richard Biener <rguenther@suse.de>
23964 PR tree-optimization/79732
23965 * tree-inline.c (expand_call_inline): Do not shadow var.
23967 2017-02-28 Richard Biener <rguenther@suse.de>
23969 PR tree-optimization/79723
23970 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23971 address-space properly.
23973 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23975 * doc/optinfo.texi (Optimization groups): Fix option used for
23977 * doc/invoke.texi (-fopt-info): Document "omp".
23978 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23979 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23980 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23981 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23982 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23984 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23986 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23987 OPTGROUP_OMP with "omp".
23989 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23992 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23993 for arithmetic shift of unsigned V2DI.
23995 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23997 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23998 arc/linux.h headers.
23999 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24000 (LINK_SPEC): Likewise.
24001 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24002 (EXTRA_SPECS): Likewise.
24003 (STARTFILE_SPEC): Likewise.
24004 (ENDFILE_SPEC): Likewise.
24005 (LIB_SPEC): Likewise.
24006 (TARGET_SDATA_DEFAULT): Likewise.
24007 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24008 (MULTILIB_DEFAULTS): Likewise.
24009 (DWARF2_UNWIND_INFO): Likewise.
24010 * config/arc/big.h: New file.
24011 * config/arc/elf.h: Likewise.
24012 * config/arc/linux.h: Likewise.
24013 * config/arc/t-uClibc: Remove.
24015 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24017 PR tree-optimization/77536
24018 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24019 (tree_transform_and_unroll_loop): Use above function to compute the
24020 estimated niter of unrolled loop and use it when scaling profile.
24021 Also use count info rather than frequency if it's non-zero.
24022 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24023 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24024 (vect_transform_loop): Call above function.
24026 2017-02-27 Richard Biener <rguenther@suse.de>
24028 PR tree-optimization/45397
24029 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24030 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24031 (visit_nary_op): Add pattern matching for CSEing sign-changed
24032 or truncated operations with wider ones.
24034 2017-02-27 Richard Biener <rguenther@suse.de>
24036 PR tree-optimization/79690
24037 * tree-vect-stmts.c (vectorizable_store): Use vector type
24038 built from the DR with address-space.
24040 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24042 * doc/invoke.texi (Optimize Options): Refine the description
24043 of asan-use-after-return.
24045 2017-02-25 Alan Modra <amodra@gmail.com>
24047 PR rtl-optimization/79584
24048 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24049 base, not ad->base_term, the reg within base. Remove assertion
24050 that ad->base == ad->base_term. Replace gen_int_mode using
24051 bogus mode with const0_rtx.
24053 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24055 PR middle-end/79396
24056 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24057 FMA_EXPR like tcc_binary or tcc_unary.
24059 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24062 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24064 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24065 (output_loc_operands): Handle DW_OP_call_ref and
24066 DW_OP_GNU_variable_value.
24067 (struct variable_value_struct): New type.
24068 (struct variable_value_hasher): Likewise.
24069 (variable_value_hash): New variable.
24070 (string_types): Remove.
24071 (copy_loc_descr): New function.
24072 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24073 (prepend_loc_descr_to_each): New function.
24074 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24075 instead of add_loc_descr_to_each if the first argument is single
24076 location list and the second has multiple.
24077 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24078 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24079 when looking for variable value which doesn't have other location info.
24080 (loc_list_from_tree): Formatting fix.
24081 (gen_array_type_die): Simplify DW_AT_string_length handling.
24082 (adjust_string_types): Remove.
24083 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24084 string_types. Call resolve_variable_values.
24085 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24086 (resolve_addr_in_expr): Likewise. Add A argument.
24087 (copy_deref_exprloc): Remove deref argument. Adjust for the
24088 original expression being DW_OP_GNU_variable_value with optionally
24089 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24090 optionally after it.
24091 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24092 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24093 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24094 (variable_value_hasher::hash, variable_value_hasher::equal): New
24096 (resolve_variable_value_in_expr, resolve_variable_value,
24097 resolve_variable_values, note_variable_value_in_expr,
24098 note_variable_value): New functions.
24099 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24102 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24105 * opts.h (handle_generated_option): Add GENERATED_P argument.
24106 * opts-common.c (handle_option): Adjust function comment.
24107 (handle_generated_option): Add GENERATED_P argument, pass it to
24109 (control_warning_option): Pass false to handle_generated_option
24111 * opts.c (maybe_default_option): Pass true to handle_generated_option
24113 * optc-gen.awk: Likewise.
24115 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24117 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24118 a REG, look at the REG it is a SUBREG of.
24119 (splitter for cmpeqsi_t): Ditto.
24121 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24123 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24124 the special USEs with the pattern of the insn, not the insn itself.
24126 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24129 * doc/invoke.texi: Document -mload-store-pairs.
24131 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24132 Sandra Loosemore <sandra@codesourcery.com>
24134 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24135 argument isn't a CONST_INT.
24136 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24137 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24138 (nios2_validate_compare): Bypass alternate compare logic if *op2
24139 is not a CONST_INT.
24140 (ldstwm_operation_p): Return false if first_base is not a REG or
24141 if first_offset is not a CONST_INT.
24143 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24145 * config/cris/cris.md: Use correct operand in a define_peephole2.
24147 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24149 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24151 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24153 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24154 this_insn if it is an INSN or JUMP_INSN.
24155 (force_offsettable): Look at base, not at addr.
24156 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24157 on things that aren't necessarily CONST_INTs.
24159 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24161 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24162 -mfpmath=sse is the default also for x86-32 targets with SSE2
24163 instruction set when @option{-ffast-math} is enabled
24165 2017-02-24 Jeff Law <law@redhat.com>
24167 PR rtl-optimizatoin/79286
24168 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24171 2017-02-24 Richard Biener <rguenther@suse.de>
24173 PR tree-optimization/79389
24174 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24177 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24179 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24180 function comment to reflect reality.
24181 (loop_exits_before_overflow): Fix typo in function description.
24183 2017-02-24 Richard Biener <rguenther@suse.de>
24185 PR tree-optimization/79389
24186 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24187 properly that a threading opportunity exists. Detect conditional
24188 copy/constant propagation opportunities.
24190 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24192 * config/visium/visium.md (type): Add trap.
24193 (b): New mode attribute.
24194 (*btst): Rename into...
24195 (*btst<mode>): ...this and adjust.
24196 (*cbranchsi4_btst_insn): Rename into...
24197 (*cbranch<mode>4_btst_insn): ...this and adjust.
24198 (trap): New define_insn.
24200 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24202 PR tree-optimization/79389
24203 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24204 (noce_reversed_cond_code): New function.
24205 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24206 reversed_comparison_code. Formatting fix.
24207 (noce_try_store_flag): Test rev_cond != NULL in addition to
24208 reversed_comparison_code.
24209 (noce_try_store_flag_constants): Likewise.
24210 (noce_try_store_flag_mask): Likewise.
24211 (noce_try_addcc): Use rev_cond if non-NULL instead of
24212 reversed_comparison_code.
24213 (noce_try_cmove_arith): Likewise. Formatting fixes.
24214 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24215 (noce_find_if_block): Initialize rev_cond.
24216 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24217 instead of false as last argument never attempt to reverse it
24220 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24222 PR tree-optimization/79663
24223 * tree-predcom.c (combine_chains): Process refs in reverse order
24224 only for ZERO length chains, and add explaining comment.
24226 2017-02-23 Jeff Law <law@redhat.com>
24228 PR tree-optimization/79578
24229 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24230 in call to operand_equal_p.
24232 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24235 * config/i386/cpuid.h: Fix another undefined behavior.
24237 2017-02-23 Richard Biener <rguenther@suse.de>
24239 PR tree-optimization/79683
24240 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24241 vector types for data-refs.
24243 2017-02-23 Martin Liska <mliska@suse.cz>
24245 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24247 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24249 PR middle-end/79665
24250 * internal-fn.c (get_range_pos_neg): Moved to ...
24251 * tree.c (get_range_pos_neg): ... here. No longer static.
24252 * tree.h (get_range_pos_neg): New prototype.
24253 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24254 are known to be in between 0 and signed maximum inclusive, try to
24255 expand both unsigned and signed divmod and use the cheaper one from
24258 2017-02-22 Jeff Law <law@redhat.com>
24260 PR tree-optimization/79578
24261 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24262 to compare base operands.
24264 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24267 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24268 gpc_reg_operand instead of fpr_reg_operand.
24270 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24272 * config/mips/mips.c (mips_return_in_memory): Force FP
24273 vector types to be returned in memory for o32 ABI.
24275 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24277 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24278 instead of DW_TAG_member for static data member declarations and don't
24279 set no_linkage_name for static inline data members.
24280 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24283 2017-02-22 Martin Liska <mliska@suse.cz>
24285 * doc/invoke.texi: Replace inequality signs with square brackets
24288 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24290 PR tree-optimization/68644
24291 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24293 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24296 * lra-constraints.c (simplify_operand_subreg): Handle
24297 WORD_REGISTER_OPERATIONS targets.
24299 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24302 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24303 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24304 elimination by swapping fld*.
24306 2017-02-22 Richard Biener <rguenther@suse.de>
24308 PR tree-optimization/79673
24309 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24310 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24311 irrelevant address-space qualifiers and avoiding a
24312 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24314 2017-02-22 Richard Biener <rguenther@suse.de>
24316 PR tree-optimization/79666
24317 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24318 to not symbolically negate if that may introduce undefined
24321 2017-02-22 Martin Liska <mliska@suse.cz>
24324 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24325 * data-streamer-out.c (streamer_write_gcov_count_stream):
24327 * value-prof.c (stream_out_histogram_value): Make assert more
24328 precise based on type of counter.
24330 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24333 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24334 Use nonimmediate_operand instead of memory_operand for operand 1.
24335 (float-extend standard_x87sse_constant_load splitter): Ditto.
24337 2017-02-21 Jeff Law <law@redhat.com>
24339 PR tree-optimization/79621
24340 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24341 blocks with edges to themselves.
24343 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24346 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24347 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24348 Use gimple_call_builtin_p.
24351 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24352 on temporarily removed DEBUG_INSNs.
24354 PR tree-optimization/79649
24355 * tree-loop-distribution.c (classify_partition): Give up on
24356 non-generic address space loads/stores.
24358 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24360 * doc/loop.texi (Loop manipulation): Remove nonexistent
24361 tree_ssa_loop_version from the documentation.
24362 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24364 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24367 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24368 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24369 * config/rs6000/rs6000.c: Include except.h.
24370 (rs6000_expand_split_stack_prologue): Call
24371 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24373 2017-02-21 Martin Jambor <mjambor@suse.cz>
24376 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24377 have been analyzed.
24379 2017-02-21 Martin Jambor <mjambor@suse.cz>
24381 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24382 for backward compatibility only.
24383 * doc/invoke.texi (Option Summary): Remove all references to
24384 -fipa-cp-alignment.
24386 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24390 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24392 * lra-constraints.c (curr_insn_transform): Handle
24393 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24395 2017-02-21 Martin Liska <mliska@suse.cz>
24397 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24399 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24401 PR translation/79638
24402 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24404 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24407 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24408 (arm_function_ok_for_sibcall): Return false for an indirect call by
24409 descriptor if all the argument registers are used.
24410 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24411 alignment of the function.
24413 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24415 PR tree-optimization/61441
24416 * simplify-rtx.c (simplify_const_unary_operation): For
24417 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24418 the sNaN unmodified.
24420 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24422 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24423 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24424 instead of SYSTEM_HEADER_DIR.
24426 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24427 Martin Liška <mliska@suse.cz>
24429 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24430 Fix typos and grammar, use active voice, and clarify.
24432 2017-02-20 Marek Polacek <polacek@redhat.com>
24434 PR middle-end/79537
24435 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24438 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24440 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24443 * config/i386/i386.c (ix86_expand_builtin): Handle
24444 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24445 ix86_builtins_isa[fcode].isa as a requirement of those
24446 flags and any other flag in the bitmask.
24447 (ix86_init_mmx_sse_builtins): Use 0 instead of
24448 ~OPTION_MASK_ISA_64BIT as mask.
24449 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24450 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24451 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24452 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24454 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24457 * lra-constraints.c (split_reg): Check requested split mode
24458 is supported by the register.
24460 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24462 * lra-constraints.c (simplify_operand_subreg): Remove early
24465 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24468 * lra-constraints.c (curr_insn_transform): Tighten condition
24469 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24471 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24474 * lra-constraints.c (curr_insn_transform): Handle
24475 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24477 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24480 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24482 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24484 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24487 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24490 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24492 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24495 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24497 * final.c (last_columnnum, override_columnnum): New variables.
24498 (final_start_function): Set last_columnnum, pass it to begin_prologue
24499 hook and pass 0 to dwarf2out_begin_prologue.
24500 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24501 to source_line debug hook.
24502 (notice_source_line): Compute last_columnnum and for debug_column_info
24503 return true on column changes.
24504 * debug.h (struct gcc_debug_hooks): Add column argument to
24505 source_line and begin_prologue hooks.
24506 (debug_nothing_int_charstar_int_bool): Remove prototype.
24507 (debug_nothing_int_int_charstar,
24508 debug_nothing_int_int_charstar_int_bool): New prototypes.
24509 (dwarf2out_begin_prologue): Add column argument.
24510 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24511 begin_prologue hooks.
24512 (debug_nothing_int_charstar_int_bool): Remove.
24513 (debug_nothing_int_int_charstar,
24514 debug_nothing_int_int_charstar_int_bool): New functions.
24515 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24516 through to dwarf2out_source_line.
24517 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24518 (dwarf2out_source_line): Add column argument, emit it if requested.
24519 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24521 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24522 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24523 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24524 through to dwarf2out_begin_prologue.
24525 (vmsdbgout_source_line): Add column argument, pass it through to
24526 dwarf2out_source_line.
24527 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24528 dbxout_source_line caller.
24529 (dbxout_source_line): Add column argument.
24531 * common.opt (gno-column-info, gcolumn-info): New options.
24532 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24533 (check_die): Also test for multiple DW_AT_decl_column attributes.
24534 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24535 DW_AT_decl_column if requested.
24536 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24538 (gen_variable_die): Likewise.
24539 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24540 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24543 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24544 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24545 (ix86_handle_option): Handle OPT_m3dnowa.
24546 * doc/invoke.texi (-m3dnowa): Document.
24547 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24548 -m3dnowa instead of -m3dnow -march=athlon.
24551 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24552 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24554 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24557 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24558 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24559 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24560 generator for vsx_xxpermdi_<mode>_be.
24561 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24562 force big-endian semantics.
24563 (vsx_xxpermdi_<mode>_be): New define_expand with same
24564 implementation as previous version of vsx_xxpermdi_<mode>.
24566 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24568 PR tree-optimization/79327
24569 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24570 variable, its initialization and use.
24572 2017-02-17 Julia Koval <julia.koval@intel.com>
24574 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24575 (OPTION_MASK_ISA_PKU_UNSET): New.
24576 (ix86_handle_option): Handle -mrdpid.
24577 * config/i386/cpuid.h (bit_RDPID): New.
24578 * config/i386/driver-i386.c (host_detect_local_cpu):
24579 Detect RDPID feature.
24580 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24581 * config/i386/i386-c.c (ix86_target_macros_internal):
24583 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24584 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24585 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24586 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24587 * config/i386/i386.md (define_insn "rdpid"): New.
24588 * config/i386/i386.opt Add -mrdpid.
24589 * config/i386/immintrin.h (_rdpid_u32): New.
24591 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24593 PR rtl-optimization/79541
24594 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24595 instead of transforming it into USE.
24597 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24599 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24600 If HONOR_SNANS (SFmode) force the input to a register.
24601 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24602 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24603 an frsp or similar insn.
24605 2017-02-17 Martin Liska <mliska@suse.cz>
24607 PR rtl-optimization/79577
24608 * params.def (selsched-max-sched-times): Increase minimum to 1.
24610 2017-02-17 Martin Liska <mliska@suse.cz>
24612 PR rtl-optimization/79574
24613 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24615 2017-02-17 Martin Liska <mliska@suse.cz>
24617 PR tree-optimization/79529
24618 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24619 ssa_defined_default_def_p to handle cases which are implicitly
24621 * tree-ssa.c (ssa_defined_default_def_p): New function.
24622 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24623 which are implicitly defined.
24624 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24626 2017-02-17 Richard Biener <rguenther@suse.de>
24628 PR middle-end/79576
24629 * params.def (max-ssa-name-query-depth): Limit to 10.
24631 2017-02-17 Richard Biener <rguenther@suse.de>
24633 PR tree-optimization/79552
24634 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24637 2017-02-17 Richard Biener <rguenther@suse.de>
24640 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24642 2017-02-17 Marek Polacek <polacek@redhat.com>
24644 PR middle-end/79536
24645 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24646 (fold_negate_expr): New wrapper.
24648 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24650 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24651 Correct terminology and de-emphasize pre-standard behavior.
24653 2017-02-16 Alan Modra <amodra@gmail.com>
24655 PR rtl-optimization/79286
24656 * ira.c (def_dominates_uses): New function.
24657 (update_equiv_regs): Don't create an equivalence for insns that
24658 may trap where the register def does not dominate the use.
24660 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24662 PR rtl-optimization/78127
24663 * lra.c (lra): Call lra_eliminate before finish the loop after
24666 2017-02-16 Richard Biener <rguenther@suse.de>
24668 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24670 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24671 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24672 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24673 (isl_val_int_from_wi): New function.
24674 (extract_affine_gmp): Rename to ...
24675 (extract_affine_wi): ... this, take a widest_int.
24676 (extract_affine_int): Just wrap extract_affine_wi.
24677 (add_param_constraints): Use isl_val_int_from_wi.
24678 (add_loop_constraints): Likewise, and extract_affine_wi.
24680 2017-02-15 Jeff Law <law@redhat.com>
24682 PR middle-end/79521
24683 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24684 ira_init_register_move_cost_if_necessary.
24686 2017-02-15 Martin Sebor <msebor@redhat.com>
24688 PR middle-end/32003
24689 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24690 removed in a prior commit.
24692 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24694 PR tree-optimization/79347
24695 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24696 counters during peeling.
24698 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24700 * Makefile.in (site.exp): Remove "set ISLVER".
24702 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24705 * real.c (real_from_integer): Call real_convert even for decimal.
24707 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24710 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24712 2017-02-14 Andrew Pinski <apinski@cavium.com>
24714 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24715 cores and change the partno/implementer to be correct.
24716 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24717 the 'B" as the implementer.
24718 * config/aarch64/aarch64-tune.md: Regenerate.
24720 2017-02-14 Carl Love <cel@us.ibm.com>
24722 * config/rs6000/rs6000.c: Add case statement entry to make the
24723 xvcvuxdsp built-in argument unsigned.
24724 * config/rs6000/vsx.md: Fix the source and return operand types so they
24725 match the instruction definitions from the ISA document. Fix typo
24726 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24729 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24732 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24733 member early_clobber_alts.
24734 * lra-lives.c (reg_early_clobber_p): New.
24735 (process_bb_lives): Use it.
24736 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24737 (debug_operand_data): Initialize early_clobber_alts.
24738 (setup_operand_alternative): Set up early_clobber_alts.
24739 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24740 alternatives to new_insn_reg.
24741 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24743 (lra_update_insn_regno_info): Pass the new arg.
24745 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24747 PR middle-end/79505
24748 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24749 (new_oacc_loop_raw): Don't clear already cleared fields.
24752 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24753 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24754 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24756 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24759 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24761 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24764 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24765 the extra instruction to the right place to store 128-bit constant
24768 2017-02-14 Martin Sebor <msebor@redhat.com>
24770 PR middle-end/79448
24771 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24772 warning for strings of unknown length.
24774 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24776 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24778 2017-02-14 Jeff Law <law@redhat.com>
24781 * ira-costs.c (scan_one_insn): Initialize register move costs
24782 for pseudos seen in USE/CLOBBER insns.
24784 PR tree-optimization/79095
24785 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24786 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24787 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24788 if the operands are known to be not equal, then the resulting range
24790 (intersect_ranges): If the new range is ~[0,0] and the old range is
24791 wide, then prefer ~[0,0].
24792 * tree-vrp.c (overflow_comparison_p_1): New function.
24793 (overflow_comparison_p): New function.
24794 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24795 if NAME is used in an overflow test.
24796 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24797 overflow check that can be expressed as an equality test, then adjust
24798 ops to be that equality test.
24800 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24802 * config/s390/s390-builtin-types.def: Remove flags argument.
24803 * config/s390/s390.c (s390_init_builtins): Likewise.
24805 2017-02-14 Martin Liska <mliska@suse.cz>
24807 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24808 vector. Fix trailing white spaces.
24810 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24812 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24815 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24817 PR rtl-optimization/68664
24818 * config/arm/arm.c (arm_sched_can_speculate_insn):
24819 New function. Declare prototype.
24820 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24822 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24824 PR rtl-optimization/68664
24825 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24827 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24829 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24831 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24832 max skip bytes for function, loop and jump.
24834 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24836 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24837 ABS_EXPR for gimple dump.
24839 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24842 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24844 PR tree-optimization/79408
24845 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24846 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24847 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24848 also if rhs1 is INTEGER_CST.
24850 2017-02-14 Richard Biener <rguenther@suse.de>
24852 PR middle-end/79432
24853 * tree-into-ssa.c (insert_phi_nodes): When the function can
24854 have abnormal edges rewrite SSA names with broken use-def
24855 dominance out of SSA and register them for PHI insertion.
24857 2017-02-13 Martin Sebor <msebor@redhat.com>
24859 PR middle-end/79496
24860 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24861 clearing info.nowrite flag when snprintf size argument is a range.
24863 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24865 * cprop.c (cprop_jump): Add missing space in string literal.
24866 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24867 (get_constraint_for_component_ref): Likewise.
24868 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24869 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24870 * lra-constraints.c (process_alt_operands): Likewise.
24871 * ipa-inline.c (inline_small_functions): Likewise.
24872 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24873 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24874 * trans-mem.c (diagnose_tm_1_op): Likewise.
24875 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24876 (grid_parallel_clauses_gridifiable): Likewise.
24878 * config/nvptx/mkoffload.c (process): Add space in between
24881 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24882 "MOD4_SSE_REGS" and "ALL_REGS".
24884 * spellcheck.c (test_data): Add , in between "foo" and "food".
24886 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24889 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24890 boundary crossing check and subsequent code generation agree.
24892 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24894 * config/aarch64/aarch64.c (has_memory_op): Delete.
24895 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24898 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24900 PR rtl-optimization/79388
24901 PR rtl-optimization/79450
24902 * combine.c (distribute_notes): When removing TEM_INSN for which
24903 corresponding dest has last value recorded, invalidate that last
24906 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24908 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24909 of explicit '@'. Add missing assembly comment marker on branch costs
24912 2017-02-13 Nathan Sidwell <nathan@acm.org>
24914 * gengtype-lex.l (<in_struct>): Add '/'.
24916 2017-02-13 Martin Liska <mliska@suse.cz>
24919 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24921 2017-02-13 Richard Biener <rguenther@suse.de>
24923 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24925 * configure: Re-generate.
24926 * config.in: Likewise.
24927 * graphite-dependences.c: Simplify as if
24928 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24929 * graphite-isl-ast-to-gimple.c: Likewise.
24930 * graphite-optimize-isl.c: Likewise.
24931 * graphite-poly.c: Likewise.
24932 * graphite-sese-to-poly.c: Likewise.
24933 * graphite.h: Likewise.
24934 * toplev.c: Include isl/version.h and use isl_version () for
24935 printing the ISL version.
24936 * doc/install.texi: Update ISL requirement.
24938 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24940 * doc/standards.texi (Standards): Update reference to
24943 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24945 * doc/extend.texi (Named Address Spaces): sourceware.org now
24947 * doc/install.texi (Binaries): Ditto.
24950 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24952 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24953 standard terminology "stringize"/"stringizing" throughout.
24954 * doc/cppinternals.texi: Likewise.
24956 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24958 * doc/extend.texi: Fix some spelling mistakes and typos.
24959 * doc/invoke.texi: Likewise.
24961 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24964 * params.def (inline-min-speedup) Change from 10 to 8.
24966 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24968 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24971 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24974 * ipa-inline-analysis.c (get_minimal_bb): New function.
24975 (record_modified): Use it.
24976 (remap_edge_change_prob): Handle also ancestor functions.
24978 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24980 * doc/contrib.texi (Contributors): Remove broken link into
24981 the Mauve CVS repository.
24983 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24985 PR middle-end/79454
24986 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24987 result computation whenever lhs doesn't have vector mode, not
24988 just when it has BLKmode.
24990 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24992 * doc/makefile.texi (profiledbootstrap): Refer to the
24993 installation instructions only in textual form.
24995 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24998 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25000 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25002 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25003 (Specific): Update mingw-w64 reference.
25005 (Specific): Remove broken link to Renesas RX processor.
25007 2017-02-10 Richard Biener <rguenther@suse.de>
25009 * toplev.c (process_options): Do not mention obsolete graphite
25010 options when printing sorry message about missing graphite support.
25011 Mention -floop-nest-optimize.
25013 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25015 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25016 (vtst_p16): Likewise.
25017 (vtstq_p8): Likewise.
25018 (vtstq_p16): Likewise.
25020 (vtstq_p64): Likewise.
25021 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25022 (vset_lane_p64): New.
25023 (vsetq_lane_p64): New.
25025 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25027 PR tree-optimization/79411
25028 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25029 stmt operands are SSA_NAMEs used in abnormal phis.
25030 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25033 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25036 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25039 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25041 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25043 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25045 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25049 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25050 not arbitrary TREE_CONSTANT.
25053 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25054 "omp declare target link" attribute unless is_global_var.
25055 * omp-offload.c (find_link_var_op): Likewise.
25057 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25058 Chung-Lin Tang <cltang@codesourcery.com>
25060 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25062 (gimplify_adjust_omp_clauses): Don't delete TILE.
25063 (gimplify_omp_for): Deal with TILE.
25064 * internal-fn.c (expand_GOACC_TILE): New function.
25065 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25067 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25068 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25070 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25071 avoid DIV for outermost collapse var.
25072 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25073 Remove out of date comments, fix whitespace.
25074 * omp-general.c (omp_extract_for_data): Deal with tiling.
25075 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25076 adjust OLF_DIM_BASE value.
25077 (struct omp_for_data): Add tiling field.
25078 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25079 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25080 for auto loops. Remove default auto determining, moved to
25081 oacc_loop_fixed_partitions.
25082 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25083 stmts, add e_mask field.
25084 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25085 (oacc_thread_numbers): Use oacc_dim_call.
25086 (oacc_xform_tile): New.
25087 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25088 (finish_oacc_loop): Adjust for ifns vector.
25089 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25090 add case for GOACC_TILE fns.
25091 (oacc_loop_xform_loop): Delete.
25092 (oacc_loop_process): Iterate over call list directly, and add
25093 handling for GOACC_TILE fns.
25094 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25096 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25097 vector partitioning to outer loops. Assign 2 partitions to loops
25098 when available. Add TILE handling.
25099 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25100 (execite_oacc_device_lower): Process GOACC_TILE fns,
25101 ignore unknown specs.
25102 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25103 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25104 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25106 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25108 * configure.ac (ACX_BUGURL): Update.
25109 * configure: Regenerate.
25111 2017-02-09 Richard Biener <rguenther@suse.de>
25113 PR tree-optimization/69823
25114 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25115 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25117 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25119 * config/arc/arc-c.def: Add __NPS400__ definition.
25120 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25121 (TARGET_NPS400): Define.
25123 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25125 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25127 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25128 pointer, arch_info.
25129 (arc_cpu_types): Fill the arch_info field with a pointer into the
25130 arc_arch_types table.
25131 (arc_selected_cpu): Declare.
25132 * config/arc/arc.c (arc_selected_cpu): Make global.
25133 (arc_selected_arch): Delete.
25134 (arc_base_cpu): Delete.
25135 (arc_override_options): Remove references to deleted variables,
25136 update access to arch information.
25137 (ARC_OPT): Update access to arch information.
25138 (ARC_OPTX): Likewise.
25139 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25140 (TARGET_ARC600): Update access to arch information.
25141 (TARGET_ARC601): Likewise.
25142 (TARGET_ARC700): Likewise.
25143 (TARGET_EM): Likewise.
25144 (TARGET_HS): Likewise.
25145 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25148 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25151 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25152 condition/operands for integer GE/LE/GEU/LEU operations.
25154 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25156 PR translation/79397
25157 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25160 2017-02-08 Martin Jambor <mjambor@suse.cz>
25163 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25164 whether allocation happened.
25165 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25166 nothing was allocated.
25168 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25170 PR tree-optimization/79408
25171 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25172 constant, but SSA_NAME with a known integer range, use the minimum
25173 of that range instead of op1 to determine if modulo can be replaced
25174 with its first operand.
25176 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25178 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25180 2017-02-08 Richard Biener <rguenther@suse.de>
25182 PR tree-optimization/71824
25183 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25184 Check all loops contained in the merged region.
25186 2017-02-07 Andrew Pinski <apinski@cavium.com>
25188 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25190 2017-02-07 Andrew Pinski <apinski@cavium.com>
25192 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25193 (thunderxt88): Likewise.
25194 (thunderxt81): Disable LSE and change v8.1 to v8.
25195 (thunderxt83): Likewise.
25197 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25198 Richard Biener <rguenther@suse.de>
25200 PR middle-end/79399
25201 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25202 type from int to size_t.
25203 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25205 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25207 PR rtl-optimization/79386
25208 * cprop.c (bypass_conditional_jumps): Initialize
25209 bypass_last_basic_block already before splitting bbs after
25210 unconditional traps...
25211 (bypass_conditional_jumps): ... rather than here.
25214 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25215 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25216 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25217 fix -masm=intel patterns.
25219 2017-02-07 Richard Biener <rguenther@suse.de>
25221 PR tree-optimization/79256
25222 PR middle-end/79278
25223 * builtins.c (get_object_alignment_2): Use min_align_of_type
25224 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25225 and ADJUST_FIELD_ALIGN.
25227 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25229 * doc/tm.texi: Regenerate.
25230 * stor-layout.c (layout_decl): Adjust.
25231 (update_alignment_for_field): Likewise.
25232 (place_field): Likewise.
25233 (min_align_of_type): Likewise.
25234 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25235 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25236 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25237 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25238 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25239 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25240 * config/i386/i386.c (x86_field_alignment): Likewise.
25241 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25242 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25243 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25244 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25245 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25246 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25250 2017-01-30 Richard Biener <rguenther@suse.de>
25252 PR tree-optimization/79256
25253 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25254 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25257 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25259 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25260 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25261 builtins to SImode and emit a zero-extend, if necessary.
25263 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25265 * docs/invoke.texi (RISC-V Options): Alphabetize.
25267 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25269 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25272 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25274 * config/riscv/riscv.c: New file.
25275 * common/config/riscv/riscv-common.c: Likewise.
25276 * config.gcc: Likewise.
25277 * config/riscv/constraints.md: Likewise.
25278 * config/riscv/elf.h: Likewise.
25279 * config/riscv/generic.md: Likewise.
25280 * config/riscv/linux.h: Likewise.
25281 * config/riscv/multilib-generator: Likewise.
25282 * config/riscv/peephole.md: Likewise.
25283 * config/riscv/pic.md: Likewise.
25284 * config/riscv/predicates.md: Likewise.
25285 * config/riscv/riscv-builtins.c: Likewise.
25286 * config/riscv/riscv-c.c: Likewise.
25287 * config/riscv/riscv-ftypes.def: Likewise.
25288 * config/riscv/riscv-modes.def: Likewise.
25289 * config/riscv/riscv-opts.h: Likewise.
25290 * config/riscv/riscv-protos.h: Likewise.
25291 * config/riscv/riscv.h: Likewise.
25292 * config/riscv/riscv.md: Likewise.
25293 * config/riscv/riscv.opt: Likewise.
25294 * config/riscv/sync.md: Likewise.
25295 * config/riscv/t-elf-multilib: Likewise.
25296 * config/riscv/t-linux: Likewise.
25297 * config/riscv/t-linux-multilib: Likewise.
25298 * config/riscv/t-riscv: Likewise.
25299 * configure.ac: Likewise.
25300 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25301 Waterman as RISC-V maintainers.
25302 * doc/install.texi: Add RISC-V entries.
25303 * doc/invoke.texi: Add RISC-V options section.
25304 * doc/md.texi: Add RISC-V constraints section.
25305 * configure: Regenerated.
25307 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25310 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25311 false values to be constant vectors with all 0 or all 1 bits set.
25312 (vcondu<mode><mode>): Likewise.
25313 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25315 (fpmask_comparison_operator): Update comment.
25316 (vecint_comparison_operator): New predicate.
25317 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25318 vector conditionals when the true and false values are constant
25319 vectors with all 0 bits or all 1 bits set.
25321 2017-02-06 Martin Sebor <msebor@redhat.com>
25323 PR tree-optimization/79376
25324 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25326 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25328 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25329 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25330 to simplify split condition.
25332 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25334 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25335 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25338 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25340 PR rtl-optimization/68664
25341 * target.def (can_speculate_insn): New hook.
25342 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25343 * doc/tm.texi: Regenerate.
25344 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25345 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25346 (rs6000_sched_can_speculate_insn): New function.
25348 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25350 PR tree-optimization/79284
25351 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25352 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25353 vectorizable_mask_load_store, vectorizable_operation,
25354 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25355 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25356 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25357 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25358 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25359 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25360 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25361 is_gimple_assign (stmt). Replace another such test with
25362 is_gimple_assign (stmt).
25364 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25367 * config/avr/avr.c (rtl-iter.h): Include it.
25368 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25369 (avr_legitimate_combined_insn): ...and implementation.
25371 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25373 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25374 * config/s390/s390.c (s390_const_operand_ok)
25375 (s390_canonicalize_comparison, s390_extract_part)
25376 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25377 (s390_contiguous_bitmask_p, s390_rtx_costs)
25378 (legitimize_pic_address): Likewise.
25379 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25380 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25381 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25382 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25383 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25385 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25387 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25388 REGNO($0) == REGNO($1).
25390 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25392 * config/s390/linux.h(SIZE_TYPE): Add comment.
25394 2017-02-06 Julian Brown <julian@codesourcery.com>
25395 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25396 Virendra Pathak <virendra.pathak@broadcom.com>
25398 * config/aarch64/aarch64-cores.def: Change the scheduler
25400 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25401 * config/aarch64/thunderx2t99.md: New file.
25403 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25405 * doc/standards.texi (Go Language): Update link to language
25408 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25410 * tree-eh.c (lower_resx): Sanitize profile.
25411 (cleanup_empty_eh_move_lp): Likewise.
25413 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25416 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25418 * cfgloopmanip.h (loop_version): Update prototype.
25419 * modulo-sched.c (sms_schedule): Update call of loop_version.
25420 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25421 * tree-parloops.c (gen_parallel_loop): Likewise.
25422 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25423 * tree-ssa-loop-split.c (split_loop): Likewise.
25424 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25425 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25427 2017-02-05 Martin Liska <mliska@suse.cz>
25430 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25432 (print_operand_address): Initialize a struct to zero.
25434 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25436 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25437 garbage collector only in textual form.
25439 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25441 * doc/extend.texi (x86 specific memory model extensions for
25442 transactional memory): Simplify a phrase.
25444 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25447 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25448 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25449 (atomic_storedi_1): Likewise.
25451 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25453 PR tree-optimization/79338
25454 * tree-parloops.c (gather_scalar_reductions): Don't call
25455 vect_analyze_loop_form for loop->inner before destroying loop's
25458 2017-02-03 Martin Sebor <msebor@redhat.com>
25460 PR tree-optimization/79327
25461 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25462 when precision has resulted in leading zeros.
25463 (format_integer): Adjust the likely counter to assume an unknown
25464 argument that may be zero is non-zero.
25466 2017-02-03 Jason Merrill <jason@redhat.com>
25469 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25470 avoid copying non-taken branch.
25472 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25474 PR tree-optimization/79340
25475 * tree-vect-loop.c (vectorizable_reduction): Release
25476 vec_defs elements after safe_splicing them into other vectors.
25479 PR tree-optimization/79327
25480 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25481 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25483 (format_integer): Use wide_int_to_tree instead of build_int_cst
25484 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25485 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25486 of shortest and longest sequence.
25488 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25490 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25491 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25493 2017-02-03 Walter Lee <walt@tilera.com>
25496 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25497 after initial stackframe link reg save.
25498 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25500 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25503 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25504 wu for stxssp alternative.
25506 2017-02-03 Martin Sebor <msebor@redhat.com>
25508 PR tree-optimization/79352
25509 * gimple-fold.c (get_range_strlen): Add argument.
25510 (get_range_strlen): Change return type to bool.
25511 (get_maxval_strlen): Pass in a dummy argument.
25512 * gimple-fold.h (get_range_strlen): Change return type to bool.
25513 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25514 * tree.h (array_at_struct_end_p): Add argument.
25515 * tree.c (array_at_struct_end_p): Handle it.
25517 2017-02-03 Martin Liska <mliska@suse.cz>
25520 * multiple_target.c (create_dispatcher_calls): Redirect edge
25521 from a caller of a dispatcher.
25522 (expand_target_clones): Make the clones local.
25523 (ipa_target_clone): Do both target clones and resolvers.
25524 (ipa_dispatcher_calls): Remove the pass.
25525 (pass_dispatcher_calls::gate): Likewise.
25526 (make_pass_dispatcher_calls): Likewise.
25527 * passes.def (pass_target_clone): Put as very first IPA early
25530 2017-02-03 Martin Liska <mliska@suse.cz>
25532 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25533 in case of a function with ifunc attribute.
25535 2017-02-03 Martin Liska <mliska@suse.cz>
25537 * cgraph.c (cgraph_node::dump): Dump function version info.
25538 * symtab.c (symtab_node::dump_base): Add missing new line.
25540 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25542 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25543 (ifcombine_ifandif): Use it.
25545 2017-02-03 Martin Liska <mliska@suse.cz>
25547 * doc/invoke.texi: Document default value for
25548 use-after-scope-direct-emission-threshold.
25550 2017-02-03 Martin Liska <mliska@suse.cz>
25552 PR tree-optimization/79339
25553 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25554 (format_floating): Likewise.
25556 2017-02-03 Martin Liska <mliska@suse.cz>
25559 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25561 (ipa_node_params_t::remove): Likewise.
25562 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25563 initialization from removed ipa_node_params_t::insert.
25564 (ipa_node_params::~ipa_node_params): Move from removed
25565 ipa_node_params_t::release.
25566 * symbol-summary.h (symbol_summary::m_released): New member.
25567 Do not release a summary twice. Do not allow to call finalizer
25568 for types of a summary that live in GGC memory.
25570 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25572 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25575 2017-02-02 Martin Sebor <msebor@redhat.com>
25577 PR middle-end/79275
25578 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25579 (format_string): Tighten up the range of output for non-constant
25580 strings and correct the expected range for wide non-constant strings.
25582 2017-02-02 Martin Sebor <msebor@redhat.com>
25584 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25586 PR middle-end/32003
25587 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25589 (-fdump-tree-@var): Add to index and document how to come up
25590 with pass-specific option and dump file names.
25591 (-fdump-passes): Clarify where to look for output.
25593 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25595 PR middle-end/77445
25596 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25597 statistics of the analyzed path; allow threading for speed when
25598 any of BBs along the path are optimized for speed.
25600 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25602 PR middle-end/78468
25603 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25604 settings of the virtual registers.
25607 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25609 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25610 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25613 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25615 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25616 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25618 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25620 * config/s390/s390.md: Add missing comments with the expanded
25622 * config/s390/vector.md: Likewise.
25623 * config/s390/vx-builtins.md: Likewise.
25625 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25628 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25629 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25630 conditions on a single line.
25632 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25634 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25635 __S390_VX__ to __VX__.
25637 2017-02-01 Andrew Pinski <apinski@cavium.com>
25639 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25640 stmt_info to record_stmt_cost.
25641 (vect_get_known_peeling_cost): Pass stmt_info if known to
25643 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25644 cpu_vector_cost field into
25645 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25646 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25647 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25648 splitting of scalar_stmt_cost and vec_stmt_cost.
25649 (thunderx_vector_cost): Likewise.
25650 (cortexa57_vector_cost): LIkewise.
25651 (exynosm1_vector_cost): Likewise.
25652 (xgene1_vector_cost): Likewise.
25653 (thunderx2t99_vector_cost): Improve after the splitting of the two
25655 (aarch64_builtin_vectorization_cost): Update for the splitting of
25656 scalar_stmt_cost and vec_stmt_cost.
25658 2017-02-01 Torvald Riegel <triegel@redhat.com>
25659 Richard Henderson <rth@redhat.com>
25661 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25662 conditional on existance of a fast atomic load.
25663 * optabs-query.c (can_atomic_load_p): New function.
25664 * optabs-query.h (can_atomic_load_p): Declare it.
25665 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25666 no fast atomic load is available for the particular size of access.
25667 (expand_atomic_compare_and_swap): Likewise.
25668 (expand_atomic_load): Likewise.
25669 (expand_atomic_store): Likewise.
25670 (expand_atomic_fetch_op): Likewise.
25671 * testsuite/lib/target-supports.exp
25672 (check_effective_target_sync_int_128): Remove x86 because it provides
25673 no fast atomic load.
25674 (check_effective_target_sync_int_128_runtime): Likewise.
25676 2017-02-01 Richard Biener <rguenther@suse.de>
25678 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25679 (graphite_transform_loops): Provide opt-info for optimized nests.
25680 * tree-parloop.c (parallelize_loops): Provide opt-info for
25681 parallelized loops.
25683 2017-02-01 Richard Biener <rguenther@suse.de>
25685 PR middle-end/79315
25686 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25687 was not set before.
25689 2017-02-01 Richard Biener <rguenther@suse.de>
25691 PR tree-optimization/71824
25692 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25693 Verify the loops are valid in the merged SESE region.
25694 (scop_detection::can_represent_loop_1): Check analyzing the
25695 evolution of the number of iterations in the region succeeds.
25697 2017-01-31 Ian Lance Taylor <iant@golang.org>
25699 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25700 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25702 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25704 PR preprocessor/79210
25705 * input.c (get_substring_ranges_for_loc): Replace line_width
25706 assertion with error-handling.
25708 2017-01-31 Richard Biener <rguenther@suse.de>
25710 PR tree-optimization/77318
25711 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25712 (create_pw_aff_from_tree): Take loop parameter.
25713 (add_condition_to_pbb): Pass loop of the condition to
25714 create_pw_aff_from_tree.
25716 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25718 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25719 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25721 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25725 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25727 (convert_int_to_float128): Likewise.
25728 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25729 (convert_int_to_float128): Likewise.
25730 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25731 (UNSPEC_IEEE128_CONVERT): Likewise.
25732 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25733 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25734 Use local variables for IBM extended format.
25735 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25736 (fix_trunc<mode>si2_fprs): Likewise.
25737 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25738 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25739 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25740 to know that we can now have integers of all sizes in vector
25742 (fix<uns>_<mode>di2_hw): Likewise.
25743 (float<uns>_<mode>si2_hw): Likewise.
25744 (fix_<mode>si2_hw): Likewise.
25745 (fixuns_<mode>si2_hw): Likewise.
25746 (float<uns>_<mode>di2_hw): Likewise.
25747 (float_<mode>di2_hw): Likewise.
25748 (float_<mode>si2_hw): Likewise.
25749 (floatuns_<mode>di2_hw): Likewise.
25750 (floatuns_<mode>si2_hw): Likewise.
25751 (xscvqp<su>wz_<mode>): Delete, no longer used.
25752 (xscvqp<su>dz_<mode>): Likewise.
25753 (xscv<su>dqp_<mode>): Likewise.
25754 (ieee128_mfvsrd_64bit): Likewise.
25755 (ieee128_mfvsrd_32bit): Likewise.
25756 (ieee128_mfvsrwz): Likewise.
25757 (ieee128_mtvsrw): Likewise.
25758 (ieee128_mtvsrd_64bit): Likewise.
25759 (ieee128_mtvsrd_32bit): Likewise.
25761 2017-01-31 Martin Liska <mliska@suse.cz>
25764 * ipa-prop.c (ipa_free_all_node_params): Call release method
25765 instead of ~sumbol_summary to not to trigger double times
25768 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25770 PR tree-optimization/71691
25771 * bitmap.h (class auto_bitmap): New.
25772 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25773 is_maybe_undefined instead of ssa_undefined_value_p.
25775 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25777 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25778 __S390_ARCH_LEVEL__ to __ARCH__.
25780 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25782 PR tree-optimization/79267
25783 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25784 if should_remove_lhs_p is true.
25786 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25789 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25790 (add_alignment_attribute): New.
25791 (base_type_die): Add alignment attribute.
25792 (subrange_type_die): Likewise.
25793 (modified_type_die): Likewise.
25794 (gen_array_type_die): Likewise.
25795 (gen_descr_array_type_die: Likewise.
25796 (gen_enumeration_type_die): Likewise.
25797 (gen_subprogram_die): Likewise.
25798 (gen_variable_die): Likewise.
25799 (gen_field_die): Likewise.
25800 (gen_ptr_to_mbr_type_die): Likewise.
25801 (gen_struct_or_union_type_die): Likewise.
25802 (gen_subroutine_type_die): Likewise.
25803 (gen_typedef_die): Likewise.
25804 (base_type_cmp): Compare alignment attribute.
25806 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25809 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25810 (setb_unsigned) New pattern for setb with CCUNS.
25811 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25812 subfc./subfe sequence to avoid overflow problems. Generate a
25813 shorter sequence with cmpld/setb for power9.
25814 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25815 for generating subfc. instruction.
25816 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25817 now uses this instruction.
25819 2017-01-30 Ian Lance Taylor <iant@google.com>
25822 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25823 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25825 2017-01-30 Martin Sebor <msebor@redhat.com>
25827 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25828 Move constant to the right of a relational operator.
25829 (get_mpfr_format_length, format_character, format_string): Ditto.
25830 (should_warn_p, maybe_warn): Same.
25832 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25834 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25837 * asan.c (get_translation_unit_decl): Remove function.
25838 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25840 2017-01-30 Martin Liska <mliska@suse.cz>
25842 PR gcov-profile/79259
25843 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25844 -fprofile-generate.
25846 2017-01-30 Martin Liska <mliska@suse.cz>
25849 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25850 Initialize variables with NULL value.
25852 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25855 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25857 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25859 2017-01-30 Richard Biener <rguenther@suse.de>
25861 PR tree-optimization/79276
25862 * tree-vrp.c (process_assert_insertions): Properly adjust common
25863 when removing a duplicate.
25865 * gcc.dg/torture/pr79276.c: New testcase.
25867 2017-01-30 Richard Biener <rguenther@suse.de>
25869 PR tree-optimization/79256
25870 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25871 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25873 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25875 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25878 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25879 ("*r<noxa>sbg_<mode>_sll_bitmask")
25880 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25881 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25882 Use contiguous_bitmask_nowrap_operand.
25884 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25887 * config/rs6000/altivec.h (vec_xl): Revise #define.
25888 (vec_xst): Likewise.
25890 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25892 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25894 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25896 PR rtl-optimization/79194
25897 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25898 traps before call to bypass_conditional_jumps.
25900 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25902 PR tree-optimization/71374
25903 * lra-constraints.c (check_conflict_input_operands): New.
25904 (match_reload): Use it.
25906 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25909 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25910 account to calculate conflict_set.
25912 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25914 PR rtl-optimization/78559
25915 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25916 other_insn in combine.
25918 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25920 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25921 uint16_type_node for BT_UINT16.
25923 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25925 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25926 "RTL Tests" to menu.
25927 (GIMPLE Tests): New node.
25928 (RTL Tests): New node.
25930 2017-01-27 Richard Biener <rguenther@suse.de>
25932 PR tree-optimization/79245
25933 * tree-loop-distribution.c (distribute_loop): Apply cost
25934 modeling also to detected patterns.
25936 2017-01-27 Richard Biener <rguenther@suse.de>
25938 PR tree-optimization/71433
25939 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25940 (compare_assert_loc): New function.
25941 (process_assert_insertions): Sort and optimize assert locations
25942 to remove duplicates and push down identical assertions on
25943 edges to their destination block.
25945 2017-01-27 Richard Biener <rguenther@suse.de>
25947 PR tree-optimization/79244
25948 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25949 out SSA names even if abnormal.
25951 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25953 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25954 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25955 instead of MPFR_RNDN.
25957 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25960 * arm.c (arm_option_override): Don't call build_target_option_node
25961 until after doing all option overrides.
25962 (arm_valid_target_attribute_tree): Likewise.
25964 2017-01-27 Martin Liska <mliska@suse.cz>
25966 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25967 for {cd}tors and C++ {cd}tors.
25969 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25971 * config/s390/s390.md ("*setmem_long_and")
25972 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25974 2017-01-26 Martin Sebor <msebor@redhat.com>
25976 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25979 2017-01-26 Martin Sebor <msebor@redhat.com>
25981 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25982 HAVE_DFmode before using XFmode or DFmode.
25983 (parse_directive): Avoid using the z length modifier to avoid
25984 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25986 PR middle-end/78703
25987 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25988 to accept adjustment as an array.
25989 (get_int_range): New function.
25990 (struct directive): Make width and prec arrays.
25991 (directive::set_width, directive::set_precision): Call get_int_range.
25992 (format_integer, format_floating): Handle width and precision ranges.
25993 (format_string, parse_directive): Same.
25995 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25998 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25999 comdat_type_p set, just clone them, but keep the children in the
26003 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26004 which have direct callers with -fvar-tracking-assignments enabled
26006 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26007 inside of type units.
26009 2017-01-26 Martin Sebor <msebor@redhat.com>
26011 PR middle-end/78703
26012 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26014 (struct format_result): Replace number_chars, number_chars_min,
26015 and number_chars_max with a single member of struct result_range.
26017 (format_result::operator+=): Adjust.
26018 (struct fmtresult): Remove bounded. Handle likely and unlikely
26020 (fmtresult::adjust_for_width_or_precision): New function.
26021 (fmtresult:type_max_digits): New function.
26022 (bytes_remaining): Handle likely and unlikely counters.
26023 (min_bytes_remaining): Remove.
26024 (format_percent): Simplify.
26025 (format_integer, format_floating): Set likely and unlikely counters.
26026 (get_string_length, format_character, format_string): Same.
26027 (format_plain, should_warn_p): New function.
26028 (maybe_warn): Call should_warn_p. Update diagnostic messages
26029 and handle those for all directives, including plain strings.
26030 (format_directive): Handle likely and unlikely counters.
26031 Remove unnecessary quoting from diagnostics. Add an informational
26033 (add_bytes): Remove.
26034 (pass_sprintf_length::compute_format_length): Simplify.
26035 (try_substitute_return_value): Handle likely and unlikely counters.
26037 2017-01-26 Carl Love <cel@us.ibm.com>
26039 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26040 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26042 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26045 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26046 endianess for subregs into account.
26047 * lra-constraints.c (lra_constraints): Do risky transformations
26048 always on the first iteration.
26049 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26050 last_call_used_reg_set.
26051 (process_bb_lives): Define and use last_call_used_reg_set.
26052 * lra.c (lra): Always continue after lra_constraints on the first
26055 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26057 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26059 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26061 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26063 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26064 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26065 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26066 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26067 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26068 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26069 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26070 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26071 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26073 2017-01-26 Marek Polacek <polacek@redhat.com>
26076 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26077 for the third operand.
26079 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26081 PR middle-end/79236
26082 * omp-low.c (struct omp_context): Add simt_stmt field.
26083 (scan_omp_for): Return omp_context *.
26084 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26085 context to the _simt_ SIMD stmt.
26086 (lower_omp_for): For combined SIMD with sibling _simt_
26087 SIMD, make sure to use the same decls in _looptemp_
26088 clauses as in the sibling.
26090 2017-01-26 David Sherwood <david.sherwood@arm.com>
26092 PR middle-end/79212
26093 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26096 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26099 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26100 emit fld b; fld a; if possible.
26102 * brig-builtins.def: Update copyright years.
26103 * config/arm/arm_acle_builtins.def: Update copyright years.
26105 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26108 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26109 constraint instead of o for the stxsd instruction.
26111 2017-01-25 Carl Love <cel@us.ibm.com>
26113 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26114 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26116 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26118 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26120 2017-01-25 Richard Biener <rguenther@suse.de>
26122 PR tree-optimization/69264
26123 * target.def (vector_alignment_reachable): Improve documentation.
26124 * doc/tm.texi: Regenerate.
26125 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26127 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26128 earlier changes with respect to TYPE_USER_ALIGN.
26129 (vector_alignment_reachable_p): Likewise. Improve dumping.
26131 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26134 * config/arm/arm.md (xordi3): Force constant operand into a register
26137 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26139 * doc/invoke.texi (-fstore-merging): Correct default optimization
26140 levels at which it is enabled.
26141 (-O): Move -fstore-merging from list to...
26144 2017-01-25 Richard Biener <rguenther@suse.de>
26147 * omp-expand.c: Include debug.h.
26148 (expand_omp_taskreg): Make sure to generate early debug before
26149 outlining anything from a function.
26150 (expand_omp_target): Likewise.
26151 (grid_expand_target_grid_body): Likewise.
26153 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26156 * asan.c (get_translation_unit_decl): New function.
26157 (asan_add_global): Extract modules file name from globals
26158 TRANSLATION_UNIT_DECL name.
26160 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26163 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26164 for long calls with APCS frame and VFP.
26166 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26168 * cfg.c (original_copy_tables_initialized_p): New function.
26169 * cfg.h (original_copy_tables_initialized_p): New decl.
26170 * cfgrtl.c (relink_block_chain): Guard the call to
26171 free_original_copy_tables with a call to
26172 original_copy_tables_initialized_p.
26173 * cgraph.h (symtab_node::native_rtl_p): New decl.
26174 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26175 (symtab_node::needed_p): Don't assert for early assembly output
26176 for __RTL functions.
26177 (cgraph_node::finalize_function): Set "force_output" for __RTL
26179 (cgraph_node::analyze): Bail out early for __RTL functions.
26180 (analyze_functions): Update assertion to support __RTL functions.
26181 (cgraph_node::expand): Bail out early for __RTL functions.
26182 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26184 * function.h (struct function): Update comment for field
26186 * gimple-expr.c: Include "tree-pass.h".
26187 (gimple_has_body_p): Return false for __RTL functions.
26188 * Makefile.in (OBJS): Add run-rtl-passes.o.
26189 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26191 (gcc::pass_manager::get_clean_slate): New accessor.
26192 * passes.c: Include "insn-addr.h".
26193 (should_skip_pass_p): Add logging. Update logic for running
26194 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26195 property-provider override so it is only done for gimple passes.
26197 (skip_pass): New function.
26198 (execute_one_pass): Call skip_pass when skipping passes.
26199 * read-md.c (md_reader::read_char): Support filtering
26200 the input to a subset of line numbers.
26201 (md_reader::md_reader): Initialize fields
26202 m_first_line and m_last_line.
26203 (md_reader::read_file_fragment): New function.
26204 * read-md.h (md_reader::read_file_fragment): New decl.
26205 (md_reader::m_first_line): New field.
26206 (md_reader::m_last_line): New field.
26207 * read-rtl-function.c (function_reader::create_function): Only
26208 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26209 curr_properties. Set DECL_INITIAL to a dummy block.
26210 (read_rtl_function_body_from_file_range): New function.
26211 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26213 * run-rtl-passes.c: New file.
26214 * run-rtl-passes.h: New file.
26216 2017-01-24 Jeff Law <law@redhat.com>
26218 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26221 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26223 PR tree-optimization/79159
26224 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26225 (record_nonwrapping_iv): Improve boundary using above function if no
26226 value range information.
26228 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26229 Martin Jambor <mjambor@suse.cz>
26231 * brig-builtins.def: New file.
26232 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26233 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26234 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26235 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26236 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26237 * builtin-types.def (BT_INT8): New.
26238 (BT_INT16): Likewise.
26239 (BT_UINT8): Likewise.
26240 (BT_UINT16): Likewise.
26241 (BT_FN_ULONG): Likewise.
26242 (BT_FN_UINT_INT): Likewise.
26243 (BT_FN_UINT_ULONG): Likewise.
26244 (BT_FN_UINT_LONG): Likewise.
26245 (BT_FN_UINT_PTR): Likewise.
26246 (BT_FN_ULONG_PTR): Likewise.
26247 (BT_FN_INT8_FLOAT): Likewise.
26248 (BT_FN_INT16_FLOAT): Likewise.
26249 (BT_FN_UINT32_FLOAT): Likewise.
26250 (BT_FN_UINT16_FLOAT): Likewise.
26251 (BT_FN_UINT8_FLOAT): Likewise.
26252 (BT_FN_UINT64_FLOAT): Likewise.
26253 (BT_FN_UINT16_UINT32): Likewise.
26254 (BT_FN_UINT32_UINT16): Likewise.
26255 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26256 (BT_FN_INT_PTR_INT): Likewise.
26257 (BT_FN_UINT_PTR_UINT): Likewise.
26258 (BT_FN_LONG_PTR_LONG): Likewise.
26259 (BT_FN_ULONG_PTR_ULONG): Likewise.
26260 (BT_FN_VOID_UINT64_UINT64): Likewise.
26261 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26262 (BT_FN_INT8_INT8_INT8): Likewise.
26263 (BT_FN_INT16_INT16_INT16): Likewise.
26264 (BT_FN_INT_INT_INT): Likewise.
26265 (BT_FN_UINT_FLOAT_UINT): Likewise.
26266 (BT_FN_FLOAT_UINT_UINT): Likewise.
26267 (BT_FN_ULONG_UINT_UINT): Likewise.
26268 (BT_FN_ULONG_UINT_PTR): Likewise.
26269 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26270 (BT_FN_UINT_UINT_UINT): Likewise.
26271 (BT_FN_VOID_UINT_PTR): Likewise.
26272 (BT_FN_UINT_UINT_PTR: Likewise.
26273 (BT_FN_UINT32_UINT64_PTR): Likewise.
26274 (BT_FN_INT_INT_UINT_UINT): Likewise.
26275 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26276 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26277 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26278 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26279 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26280 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26281 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26282 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26283 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26284 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26285 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26286 * doc/frontends.texi: List BRIG FE.
26287 * doc/install.texi (Testing): Add BRIG tesring requirements.
26288 * doc/invoke.texi (Overall Options): Mention BRIG.
26289 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26291 2017-01-24 Richard Biener <rguenther@suse.de>
26293 PR translation/79208
26294 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26296 2017-01-24 Martin Jambor <mjambor@suse.cz>
26299 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26300 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26301 and known_contexts.
26303 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26305 PR middle-end/79123
26306 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26307 casts from signed to unsigned really don't have a range.
26309 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26311 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26312 GMP_RNDx for compatiblity.
26314 2017-01-24 Martin Liska <mliska@suse.cz>
26317 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26318 that would prevent us to call alloca with -1 as argument.
26320 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26322 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26323 Avoid -Wformat-security warning.
26325 2017-01-23 Andrew Pinski <apinski@cavium.com>
26327 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26330 2017-01-23 Martin Sebor <msebor@redhat.com>
26332 PR middle-end/78703
26333 * gimple-ssa-sprintf.c (warn_level): New global.
26334 (format_integer): Use it here and throughout the rest of the file.
26335 Use the same switch to compute sign as base.
26336 (maybe_warn): New function.
26337 (format_directive): Factor out warnings into maybe_warn.
26338 Add debugging output. Use warn_level.
26339 (add_bytes): Use warn_level.
26340 (pass_sprintf_length::compute_format_length): Add debugging output.
26341 (try_substitute_return_value): Same.
26342 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26344 PR middle-end/78703
26345 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26346 (struct fmtresult, format_integer, format_floating): Adjust.
26347 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26348 (get_string_length, format_string,format_directive): Same.
26349 (pass_sprintf_length::compute_format_length): Same.
26350 (try_substitute_return_value): Simplify slightly.
26352 PR middle-end/78703
26353 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26354 (fmtresult::operator+=): Outlined.
26355 (struct fmtresult): Add ctors.
26356 (struct conversion_spec): Rename...
26357 (struct directive): ...to this. Add and remove data members.
26358 (directive::set_width, directive::set_precision): New functions.
26359 (format_percent): Use fmtresult ctor.
26360 (get_width_and_precision): Remove.
26361 (format_integer): Make naming changes. Avoid computing width and
26363 (format_floating): Same. Adjust indentation.
26364 (format_character, format_none): New functions.
26365 (format_string): Moved character handling to format_character.
26366 (format_directive): Remove arguments, change return type.
26367 (parse_directive): New function.
26368 (pass_sprintf_length::compute_format_length): Move directive
26369 parsing to parse_directive.
26371 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26373 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26374 (assign_assembler_name_if_needed): ... this.
26375 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26376 (assign_assembler_name_if_needed): ... this.
26377 (free_lang_data_in_cgraph): Adjust callers.
26378 * cgraphunit.c (cgraph_node::analyze): Likewise.
26379 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26382 2017-01-23 Richard Biener <rguenther@suse.de>
26384 PR tree-optimization/79088
26385 PR tree-optimization/79188
26386 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26387 resetting loop bounds after last path deletion. Reset loop
26388 bounds of the target loop, make code match the comments.
26389 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26390 Make sure loops need no fixups.
26392 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26394 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26395 exponent support with double type for first argument.
26396 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26397 type returned by __builtin_vec_extract_sig,
26398 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26399 functions from "vector int" to "vector unsigned int" or from
26400 "vector long long int" to "vector unsigned long long int".
26401 Changed type returned by __builtin_vec_extract_exp,
26402 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26403 functions from "vector int" to "vector unsigned int" or from
26404 "vector long long int" to "vector unsigned long long int".
26405 Changed return type of __builtin_vec_test_data_class,
26406 __builtin_vec_test_data_class_sp, and
26407 __builtin_vec_test_data_class_dp from "vector int" to
26408 "vector bool int" or from "vector long long int" to "vector bool
26409 long long int" and changed second argument type from "unsigned
26410 int" to "int". Added new overloaded function forms "vector float
26411 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26412 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26413 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26414 double, vector unsigned long long int)" and "vector double
26415 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26416 long int)". Changed return type of
26417 __builtin_scalar_test_data_class and
26418 __builtin_scalar_test_data_class_sp and
26419 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26420 int" and changed second argument from "unsigned int" to "int".
26421 Changed type returned by __builtin_scalar_test_neg,
26422 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26423 from "int" to "bool int". Added new overloaded function form
26424 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26425 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26426 exponent double-precision with floating point first argument.
26427 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26428 documentation of scalar_test_data_class, scalar_test_neg,
26429 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26430 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26431 vec_test_data_class built-in functions to reflect refinements in
26432 their type signatures.
26434 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26436 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26438 (aarch64_elf_asm_destructor): Likewise.
26440 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26442 PR rtl-optimization/78634
26443 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26444 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26445 * ifcvt.c (noce_try_cmove): Add missing cost check.
26447 PR rtl-optimization/71724
26448 * combine.c (if_then_else_cond): Look for situations where it is
26449 beneficial to undo the work of one of the recursive calls.
26451 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26453 PR tree-optimization/70754
26454 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26455 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26456 combined stmt before it if not NULL.
26457 (combine_chains): Process refs reversely and compute dominance point
26460 2017-01-23 Martin Liska <mliska@suse.cz>
26462 PR tree-optimization/79196
26463 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26464 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26466 (strlen_optimize_stmt): Call the renamed function.
26468 2017-01-23 Michael Matz <matz@suse.de>
26470 PR tree-optimization/78384
26471 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26473 2017-01-23 Richard Biener <rguenther@suse.de>
26475 PR tree-optimization/79186
26476 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26477 both incoming edges before moving an assert.
26479 2017-01-23 Martin Jambor <mjambor@suse.cz>
26481 * ipa-prop.c (load_from_param_1): Removed.
26482 (load_from_unmodified_param): Bits from load_from_param_1 put back
26484 (load_from_param): Removed.
26485 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26486 with stmt. Reverted back to use of load_from_unmodified_param.
26488 2017-01-23 Martin Jambor <mjambor@suse.cz>
26491 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26492 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26493 field a pointer to garbage collected vector, mark lattices and
26494 ipcp_orig_node with GTY((skip)).
26495 (ipa_get_param_count): Adjust to descriptors being a pointer.
26496 (ipa_get_param): Likewise.
26497 (ipa_get_type): Likewise.
26498 (ipa_get_param_move_cost): Likewise.
26499 (ipa_set_param_used): Likewise.
26500 (ipa_get_controlled_uses): Likewise.
26501 (ipa_set_controlled_uses): Likewise.
26502 (ipa_is_param_used): Likewise.
26503 (ipa_node_params_t): Move into garbage collector. New methods insert
26505 (ipa_node_params_sum): Annotate wth GTY(()).
26506 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26508 (ipa_load_from_parm_agg): Adjust declaration.
26509 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26510 * ipa-profile.c (ipa_profile): Likewise.
26511 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26512 (ipa_populate_param_decls): Make descriptors parameter garbage
26514 (ipa_dump_param): Adjust to descriptors being a pointer.
26515 (ipa_alloc_node_params): Likewise.
26516 (ipa_initialize_node_params): Likewise.
26517 (load_from_param_1): Make descriptors parameter garbage collected.
26518 (load_from_unmodified_param): Likewise.
26519 (load_from_param): Likewise.
26520 (ipa_load_from_parm_agg): Likewise.
26521 (ipa_node_params::~ipa_node_params): Removed.
26522 (ipa_free_all_node_params): Remove call to delete operator.
26523 (ipa_node_params_t::insert): New.
26524 (ipa_node_params_t::remove): Likewise.
26525 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26526 copy known_csts and known_contexts vectors.
26527 (ipa_read_node_info): Adjust to descriptors being a pointer.
26528 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26530 (ipcp_transform_function): Make descriptors variable garbage
26533 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26535 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26536 * config/i386/avx512dqintrin.h: Ditto.
26537 * config/i386/avx512fintrin.h: Ditto.
26538 * config/i386/i386.c: Handle new builtins.
26539 * config/i386/i386-builtin.def: Add new builtins.
26540 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26541 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26543 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26544 Martin Liska <mliska@suse.cz>
26546 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26547 * asan.c (asan_expand_poison_ifn): Support stores and use
26548 appropriate ASAN report function.
26549 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26550 * internal-fn.def (ASAN_POISON_USE): Declare.
26551 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26552 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26553 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26554 ASAN_POISON calls w/o LHS.
26555 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26556 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26557 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26558 * gimplify.c (asan_poison_variables): Add attribute
26559 use_after_scope_memory to variables that really needs to live
26561 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26562 having the attribute.
26564 2017-01-23 Martin Liska <mliska@suse.cz>
26566 * asan.c (create_asan_shadow_var): New function.
26567 (asan_expand_poison_ifn): Likewise.
26568 * asan.h (asan_expand_poison_ifn): New declaration.
26569 * internal-fn.c (expand_ASAN_POISON): Likewise.
26570 * internal-fn.def (ASAN_POISON): New builtin.
26571 * sanopt.c (pass_sanopt::execute): Expand
26572 asan_expand_poison_ifn.
26573 * tree-inline.c (copy_decl_for_dup_finish): Make function
26575 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26576 * tree-ssa.c (is_asan_mark_p): New function.
26577 (execute_update_addresses_taken): Rewrite local variables
26578 (identified just by use-after-scope as addressable) into SSA.
26580 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26582 * doc/install.texi (Specific): opensource.apple.com uses https
26583 now. Remove trailing slash.
26585 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26587 * README.Portability: Remove note on an Irix compatibility issue.
26589 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26591 * gcov.c (INCLUDE_ALGORITHM): Define.
26592 (INCLUDE_VECTOR): Define.
26593 No longer include <vector> and <algorithm> directly.
26595 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26597 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26599 * doc/invoke.texi (Code Gen Options): Ditto.
26601 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26604 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26606 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26608 rtl-optimization/79125
26609 * cprop.c (local_cprop_pass): Handle cases where we make an
26610 unconditional trap.
26612 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26616 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26617 read from, for big endian.
26619 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26621 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26622 register pauth builtins for LP64 only.
26624 2017-01-20 Marek Polacek <polacek@redhat.com>
26627 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26630 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26632 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26634 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26635 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26636 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26638 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26641 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26642 in big-endian mode when they are not a single duplicated value.
26644 2017-01-20 Richard Biener <rguenther@suse.de>
26646 * BASE-VER: Bump to 7.0.1.
26648 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26650 * omp-low.c (omplow_simd_context): New struct. Use it...
26651 (lower_rec_simd_input_clauses): ...here and...
26652 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26653 references to idx, lane, max_vf, is_simt.
26655 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26657 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26660 2017-01-20 Martin Jambor <mjambor@suse.cz>
26662 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26663 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26665 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26666 (GTFILES): Rename hsa.c to hsa-common.c.
26667 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26668 * hsa-dump.c: Likewise.
26669 * hsa-gen.c: Likewise.
26670 * hsa-regalloc.c: Likewise.
26671 * ipa-hsa.c: Likewise.
26672 * omp-expand.c: Likewise.
26673 * omp-low.c: Likewise.
26674 * toplev.c: Likewise.
26676 2017-01-20 Marek Polacek <polacek@redhat.com>
26679 * doc/invoke.texi: Document -Wduplicated-branches.
26680 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26681 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26682 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26683 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26684 return 0 only when not OEP_LEXICOGRAPHIC.
26685 (fold_build_cleanup_point_expr): Use the expression
26686 location when building CLEANUP_POINT_EXPR.
26687 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26688 * tree.c (add_expr): Handle error_mark_node.
26690 2017-01-20 Martin Liska <mliska@suse.cz>
26693 * tree-profile.c (init_ic_make_global_vars): Do not call
26695 (gimple_init_gcov_profiler): Likewise.
26697 2017-01-20 Martin Liska <mliska@suse.cz>
26700 * cgraph.h (maybe_create_reference): Remove argument and
26702 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26704 * ipa-cp.c (create_specialized_node): Likewise.
26705 * symtab.c (symtab_node::maybe_create_reference): Handle
26706 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26708 2017-01-20 Martin Liska <mliska@suse.cz>
26710 * read-rtl-function.c (function_reader::create_function): Use
26711 build_decl instread of build_decl_stat.
26713 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26715 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26716 * config/i386/avx512dqintrin.h: Ditto.
26717 * config/i386/avx512fintrin.h: Ditto.
26718 * config/i386/i386-builtin-types.def: Add new types.
26719 * config/i386/i386.c: Handle new types.
26720 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26721 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26722 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26723 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26724 (__builtin_ia32_kshiftridi): New.
26725 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26727 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26731 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26732 define to rs6000_init_stack_protect_guard.
26733 (rs6000_init_stack_protect_guard): New function.
26735 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26736 Yunqiang Su <yunqiang.su@imgtec.com>
26738 * config.gcc (supported_defaults): Add madd4.
26739 (with_madd4): Add validation.
26740 (all_defaults): Add madd4.
26741 * config/mips/mips.opt (mmadd4): New option.
26742 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26744 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26746 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26747 (ISA_HAS_FUSED_MADD4): Likewise.
26748 * doc/invoke.texi (-mmadd4): Document the new option.
26749 * doc/install.texi (--with-madd4): Document the new option.
26751 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26753 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26754 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26755 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26756 (aarch64_init_pauth_hint_builtins): New.
26757 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26758 (aarch64_expand_builtin): Expand new builtins.
26760 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26762 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26763 * combine-stack-adj.c (no_unhandled_cfa): Handle
26764 REG_CFA_TOGGLE_RA_MANGLE.
26765 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26766 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26767 info for return address signing.
26768 (aarch64_expand_epilogue): Likewise.
26770 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26772 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26773 * config/aarch64/aarch64-protos.h
26774 (aarch64_return_address_signing_enabled): New declaration.
26775 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26777 (aarch64_expand_prologue): Sign return address before it's pushed onto
26779 (aarch64_expand_epilogue): Authenticate return address fetched from
26781 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26782 (aarch64_attributes): New function attributes for "sign-return-address".
26783 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26784 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26785 ("*do_return"): Generate combined instructions according to key index.
26786 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26787 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26789 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26790 * config/aarch64/aarch64.opt (msign-return-address=): New.
26791 * doc/extend.texi (AArch64 Function Attributes): Documents
26792 "sign-return-address=".
26793 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26795 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26797 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26798 overall option summary.
26800 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26802 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26803 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26804 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26805 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26807 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26809 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26810 -mpower9-minmax by default for -mcpu=power9.
26811 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26812 128-bit floating point.
26814 2017-01-20 Alan Modra <amodra@gmail.com>
26816 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26817 optimizing for size.
26819 2017-01-20 Alan Modra <amodra@gmail.com>
26822 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26823 for strcmp and strncmp from corresponding builtin decl.
26825 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26827 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26828 instead of i386/rtems-64.h.
26829 * config/i386/rtems-64.h: Remove.
26831 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26835 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26837 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26839 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26841 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26842 Change int to HOST_WIDE_INT.
26843 * config/aarch64/aarch64-protos.h
26844 (aarch64_simd_gen_const_vector_dup): Likewise.
26845 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26847 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26849 * langhooks-def.h (lhd_type_for_size): New decl.
26850 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26851 * langhooks.c (lhd_type_for_size): New function, taken from
26854 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26856 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26857 define_bypass for CR latency.
26858 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26859 (power9-alu2): Add define_bypass for CR latency.
26861 (power9-mul): Update insn latency.
26862 (power9-mul-compare): Update insn latency, bypass latency and remove
26865 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26867 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26869 * config/aarch64/aarch64.md
26870 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26871 aarch64_nopcrelative_literal_loads.
26872 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26874 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26876 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26877 TARGET_LOONGSON_3A.
26878 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26880 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26883 * config.gcc (supported_defaults): Add lxc1-sxc1.
26884 (with_lxc1_sxc1): Add validation.
26885 (all_defaults): Add lxc1-sxc1.
26886 * config/mips/mips.opt (mlxc1-sxc1): New option.
26887 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26889 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26890 __mips_no_lxc1_sxc1.
26891 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26892 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26893 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26895 2017-01-19 Richard Biener <rguenther@suse.de>
26897 PR tree-optimization/72488
26898 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26899 sure to restore SSA info.
26900 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26902 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26904 PR rtl-optimization/79121
26905 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26906 of the inner type when shifting an extended value.
26908 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26911 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26912 interposable aliases.
26914 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26917 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26918 Use the evmergelohi instruction.
26919 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26920 (mov_sitf_e500_subreg8_2_be): Likewise.
26921 (mov_sitf_e500_subreg12_2_le): Likewise.
26922 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26923 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26924 (mov_sitf_e500_subreg8_2_le): Likewise.
26925 (mov_sitf_e500_subreg12_2_be): Likewise.
26927 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26929 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26930 attribute from vecsimple to vecperm.
26931 (altivec_vbpermq2): Likewise.
26933 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26936 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26938 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26939 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26940 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26941 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26942 case where N arg is SIZE_MAX.
26943 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26944 (cmpstrsi): Add pattern.
26946 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26948 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26949 __builtin_vec_revb builtins.
26950 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26951 built-in functions to support generation of the ISA 3.0 XXBR<x>
26952 vector byte reverse instructions.
26953 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26954 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26955 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26956 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26957 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26958 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26959 (P9V_BUILTIN_VEC_REVB): Likewise.
26960 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26961 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26962 (p9_xxbrq_v16qi): Likewise.
26963 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26964 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26965 (p9_xxbrh_v8hi): Likewise.
26966 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26967 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26968 vec_revb built-in functions.
26970 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26972 PR rtl-optimization/78952
26973 * config/i386/i386.md (any_extract): New code iterator.
26974 (*insvqi_2): Use any_extract for source operand.
26975 (*insvqi_3): Use any_shiftrt for source operand.
26977 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26979 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26981 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26983 2017-01-18 Matthias Klose <doko@ubuntu.com>
26985 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26987 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26989 * config/rs6000/altivec.h (vec_bperm): Change #define.
26990 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26991 (altivec_vbpermq2): New define_insn.
26992 (altivec_vbpermd): Likewise.
26993 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26994 function interface.
26995 (VBPERMD): Likewise.
26996 (VBPERM): New polymorphic function interface.
26997 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26998 Add entries for P9V_BUILTIN_VEC_VBPERM.
26999 * doc/extend.texi: Add interfaces for vec_bperm.
27001 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27003 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27004 first letter of error messages.
27005 (s390_resolve_overloaded_builtin): Likewise.
27006 * config/s390/s390.c (s390_expand_builtin): Likewise.
27007 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27008 (s390_valid_target_attribute_inner_p): Likewise.
27009 * config/s390/s390.md ("tabort"): Likewise.
27011 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27013 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27014 (ISA_AVOID_DIV_HILO): New macro.
27015 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27016 (ISA_HAS_DDIV): Likewise.
27018 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27020 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27022 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27024 * doc/invoke.texi (fabi-version): Spelling fix.
27026 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27029 * doc/invoke.texi (fabi-version): Mention mangling fix for
27032 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27035 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27037 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27040 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27043 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27045 (stack_protect_test): Ditto.
27046 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27047 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27049 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27050 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27051 -mstack-protector-guard-offset=.
27052 (RS/6000 and PowerPC Options): Ditto.
27054 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27056 * config/i386/i386.h (MASK_CLASS_P): New define.
27057 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27058 there are no registers from different register sets also when
27059 mask registers are used. Update function comment.
27060 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27061 to (*k/*r) and (*k/*km) alternatives.
27063 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27065 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27066 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27067 (EH_RETURN_HANDLER_RTX): New define.
27068 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27069 Force frame pointer in EH return functions.
27070 (aarch64_expand_epilogue): Add barrier for eh_return.
27071 (aarch64_final_eh_return_addr): Remove.
27072 (aarch64_eh_return_handler_rtx): New function.
27073 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27075 (aarch64_eh_return_handler_rtx): New prototype.
27077 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27079 * config/rs6000/altivec.h (vec_rlmi): New #define.
27080 (vec_vrlnm): Likewise.
27081 (vec_rlnm): Likewise.
27082 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27083 (UNSPEC_VRLNM): Likewise.
27084 (VIlong): New mode iterator.
27085 (altivec_vrl<VI_char>mi): New define_insn.
27086 (altivec_vrl<VI_char>nm): Likewise.
27087 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27089 (VRLDNM): Likewise.
27090 (RLNM): New polymorphic function entry.
27091 (VRLWMI): New monomorphic function entry.
27092 (VRLDMI): Likewise.
27093 (RLMI): New polymorphic function entry.
27094 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27095 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27096 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27099 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27102 * dwarf2out.c (field_byte_offset): Restore the
27103 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27104 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27105 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27108 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27111 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27113 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27116 * dwarf2out.c (add_data_member_location_attribute): For constant
27117 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27118 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27119 DW_AT_byte_size attributes.
27121 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27123 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27124 after forcing to constant memory when the code model is medium.
27126 2017-01-17 Julia Koval <julia.koval@intel.com>
27129 * config/i386/avx512fintrin.h
27130 (_mm512_i32gather_ps): Change __addr type to void const*.
27131 (_mm512_mask_i32gather_ps): Ditto.
27132 (_mm512_i32gather_pd): Ditto.
27133 (_mm512_mask_i32gather_pd): Ditto.
27134 (_mm512_i64gather_ps): Ditto.
27135 (_mm512_mask_i64gather_ps): Ditto.
27136 (_mm512_i64gather_pd): Ditto.
27137 (_mm512_mask_i64gather_pd): Ditto.
27138 (_mm512_i32gather_epi32): Ditto.
27139 (_mm512_mask_i32gather_epi32): Ditto.
27140 (_mm512_i32gather_epi64): Ditto.
27141 (_mm512_mask_i32gather_epi64): Ditto.
27142 (_mm512_i64gather_epi32): Ditto.
27143 (_mm512_mask_i64gather_epi32): Ditto.
27144 (_mm512_i64gather_epi64): Ditto.
27145 (_mm512_mask_i64gather_epi64): Ditto.
27146 (_mm512_i32scatter_ps): Change __addr type to void*.
27147 (_mm512_mask_i32scatter_ps): Ditto.
27148 (_mm512_i32scatter_pd): Ditto.
27149 (_mm512_mask_i32scatter_pd): Ditto.
27150 (_mm512_i64scatter_ps): Ditto.
27151 (_mm512_mask_i64scatter_ps): Ditto.
27152 (_mm512_i64scatter_pd): Ditto.
27153 (_mm512_mask_i64scatter_pd): Ditto.
27154 (_mm512_i32scatter_epi32): Ditto.
27155 (_mm512_mask_i32scatter_epi32): Ditto.
27156 (_mm512_i32scatter_epi64): Ditto.
27157 (_mm512_mask_i32scatter_epi64): Ditto.
27158 (_mm512_i64scatter_epi32): Ditto.
27159 (_mm512_mask_i64scatter_epi32): Ditto.
27160 (_mm512_i64scatter_epi64): Ditto.
27161 (_mm512_mask_i64scatter_epi64): Ditto.
27162 * config/i386/avx512pfintrin.h
27163 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27164 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27165 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27166 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27167 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27168 (_mm512_prefetch_i32scatter_ps): Ditto.
27169 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27170 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27171 (_mm512_prefetch_i64scatter_pd): Ditto.
27172 (_mm512_prefetch_i64scatter_ps): Ditto.
27173 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27174 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27175 * config/i386/avx512vlintrin.h
27176 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27177 (_mm_mmask_i32gather_ps): Ditto.
27178 (_mm256_mmask_i32gather_pd): Ditto.
27179 (_mm_mmask_i32gather_pd): Ditto.
27180 (_mm256_mmask_i64gather_ps): Ditto.
27181 (_mm_mmask_i64gather_ps): Ditto.
27182 (_mm256_mmask_i64gather_pd): Ditto.
27183 (_mm_mmask_i64gather_pd): Ditto.
27184 (_mm256_mmask_i32gather_epi32): Ditto.
27185 (_mm_mmask_i32gather_epi32): Ditto.
27186 (_mm256_mmask_i32gather_epi64): Ditto.
27187 (_mm_mmask_i32gather_epi64): Ditto.
27188 (_mm256_mmask_i64gather_epi32): Ditto.
27189 (_mm_mmask_i64gather_epi32): Ditto.
27190 (_mm256_mmask_i64gather_epi64): Ditto.
27191 (_mm_mmask_i64gather_epi64): Ditto.
27192 (_mm256_i32scatter_ps): Change __addr type to void*.
27193 (_mm256_mask_i32scatter_ps): Ditto.
27194 (_mm_i32scatter_ps): Ditto.
27195 (_mm_mask_i32scatter_ps): Ditto.
27196 (_mm256_i32scatter_pd): Ditto.
27197 (_mm256_mask_i32scatter_pd): Ditto.
27198 (_mm_i32scatter_pd): Ditto.
27199 (_mm_mask_i32scatter_pd): Ditto.
27200 (_mm256_i64scatter_ps): Ditto.
27201 (_mm256_mask_i64scatter_ps): Ditto.
27202 (_mm_i64scatter_ps): Ditto.
27203 (_mm_mask_i64scatter_ps): Ditto.
27204 (_mm256_i64scatter_pd): Ditto.
27205 (_mm256_mask_i64scatter_pd): Ditto.
27206 (_mm_i64scatter_pd): Ditto.
27207 (_mm_mask_i64scatter_pd): Ditto.
27208 (_mm256_i32scatter_epi32): Ditto.
27209 (_mm256_mask_i32scatter_epi32): Ditto.
27210 (_mm_i32scatter_epi32): Ditto.
27211 (_mm_mask_i32scatter_epi32): Ditto.
27212 (_mm256_i32scatter_epi64): Ditto.
27213 (_mm256_mask_i32scatter_epi64): Ditto.
27214 (_mm_i32scatter_epi64): Ditto.
27215 (_mm_mask_i32scatter_epi64): Ditto.
27216 (_mm256_i64scatter_epi32): Ditto.
27217 (_mm256_mask_i64scatter_epi32): Ditto.
27218 (_mm_i64scatter_epi32): Ditto.
27219 (_mm_mask_i64scatter_epi32): Ditto.
27220 (_mm256_i64scatter_epi64): Ditto.
27221 (_mm256_mask_i64scatter_epi64): Ditto.
27222 (_mm_i64scatter_epi64): Ditto.
27223 (_mm_mask_i64scatter_epi64): Ditto.
27224 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27225 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27226 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27227 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27228 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27229 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27230 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27231 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27232 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27233 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27234 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27235 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27236 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27237 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27238 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27239 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27240 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27241 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27242 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27243 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27244 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27245 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27246 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27247 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27248 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27249 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27250 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27251 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27252 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27253 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27254 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27255 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27256 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27257 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27258 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27259 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27260 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27261 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27262 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27263 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27264 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27265 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27266 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27267 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27268 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27269 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27270 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27271 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27272 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27273 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27274 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27275 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27276 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27277 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27278 definitions accordingly.
27280 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27281 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27284 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27287 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27290 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27291 hard regs for inner regno.
27293 2017-01-17 Martin Liska <mliska@suse.cz>
27296 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27297 assumption and add comment.
27299 2017-01-17 Nathan Sidwell <nathan@acm.org>
27301 * ipa-visibility.c (localize_node): New function, broken out of ...
27302 (function_and_variable_visibility): ... here. Call it.
27304 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27306 PR middle-end/77445
27307 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27308 correctly set frequency of oudgoing edge.
27309 (duplicate_thread_path): Fix profile updating.
27311 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27314 * configure.ac: Add GCC_BASE_VER.
27315 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27316 version from BASE-VER file.
27317 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27318 (gcc.o): Depend on $(BASEVER).
27319 * common.opt (dumpfullversion): New option.
27320 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27321 * doc/invoke.texi: Document -dumpfullversion.
27322 * doc/install.texi: Document --with-gcc-major-version-only.
27323 * configure: Regenerated.
27325 2017-01-17 Richard Biener <rguenther@suse.de>
27327 PR tree-optimization/71433
27328 * tree-vrp.c (register_new_assert_for): Merge same asserts
27329 on all incoming edges.
27330 (process_assert_insertions_for): Handle insertions at the
27333 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27335 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27336 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27338 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27341 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27344 2017-01-17 Alan Modra <amodra@gmail.com>
27347 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27348 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27349 symbolic stack limit when pic.
27351 2017-01-16 Martin Sebor <msebor@redhat.com>
27353 PR tree-optimization/78608
27354 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27356 2017-01-16 Jeff Law <law@redhat.com>
27359 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27360 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27361 for several include directories that may be relative to sysroot.
27362 * config/i386/x-mingw32 (gplus_includedir): Define.
27363 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27364 (native_system_includedir): Likewise.
27365 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27366 override if TARGET_SYSTEM_ROOT is defined.
27367 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27369 PR tree-optimization/79090
27370 PR tree-optimization/33562
27371 PR tree-optimization/61912
27372 PR tree-optimization/77485
27373 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27374 and computed trims into the dump file.
27376 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27378 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27380 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27383 * gimplify.c (gimplify_init_constructor): If want_value and
27384 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27388 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27389 sequence. Formatting fixes.
27390 (doloop_optimize): Formatting fixes.
27393 * gcc.c (debug_level_greater_than_spec_func): New function.
27394 (static_spec_functions): Add debug-level-gt spec function.
27395 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27397 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27398 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27399 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27400 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27401 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27402 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27404 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27406 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27407 QImode fixups to general and mask registers only.
27409 2017-01-16 Carl Love <cel@us.ibm.com>
27411 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27412 for built-in functions
27413 vector signed char vec_nabs (vector signed char)
27414 vector signed short vec_nabs (vector signed short)
27415 vector signed int vec_nabs (vector signed int)
27416 vector signed long long vec_nabs (vector signed long long)
27417 vector float vec_nabs (vector float)
27418 vector double vec_nabs (vector double)
27419 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27421 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27422 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27423 * doc/extend.texi: Update the documentation file for the new built-in
27426 2017-01-16 Martin Sebor <msebor@redhat.com>
27428 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27431 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27433 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27434 UNSPEC_VSX__XXSPLTD to require special splat handling.
27436 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27439 * system.h: Poison strndup.
27441 2017-01-16 Alan Modra <amodra@gmail.com>
27444 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27447 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27449 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27451 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27453 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27454 call recog here. Assert that INSN_CODE (insn) is non-negative.
27456 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27459 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27461 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27462 in the currently scheduled RTL fragment.
27464 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27466 PR rtl-optimization/78751
27467 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27470 2017-01-14 Jeff Law <law@redhat.com>
27472 PR tree-optimization/79090
27473 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27474 variable length stores.
27475 (compute_trims): Delete dead assignment to *trim_tail.
27476 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27479 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27481 PR rtl-optimization/78626
27482 PR rtl-optimization/78727
27483 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27484 of a block, and split such blocks after everything else is finished.
27486 2017-01-14 Alan Modra <amodra@gmail.com>
27489 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27490 target legitimate_combined_insn.
27491 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27492 (rs6000_legitimate_combined_insn): New function.
27493 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27495 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27496 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27497 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27499 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27501 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27503 2017-01-13 Jeff Law <law@redhat.com>
27505 PR tree-optimization/33562
27506 PR tree-optimization/61912
27507 PR tree-optimization/77485
27508 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27510 (delete_dead_assignment): Likewise.
27511 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27512 statement to delete_dead_call and delete_dead_assignment.
27514 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27517 * substring-locations.c (format_warning_va): Strengthen case 1 so
27518 that both endpoints of the substring must be within the format
27519 range for just the substring to be printed.
27521 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27523 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27524 * config/i386/i386.c (ix86_target_string): Add missing options
27525 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27526 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27527 flags_other and ix86_target_other to flags2_other. Display unknown
27529 (ix86_valid_target_attribute_inner_p): Add missing options and
27530 reorder options by implied ISAs, as in ix86_target_string.
27532 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27534 * hash-table.h (hash_table::too_empty_p): New function.
27535 (hash_table::expand): Use it.
27536 (hash_table::traverse): Likewise.
27537 (hash_table::empty_slot): Use sizeof (value_type) instead of
27538 sizeof (PTR) to convert bytes to elements. Shrink the table
27539 if the current size is excessive for the current number of
27542 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27544 * ira-costs.c (record_reg_classes): Break from the inner loop
27545 early once alt_fail is known to be true. Update outer loop
27546 handling accordingly.
27548 2017-01-13 Jeff Law <law@redhat.com>
27550 * tree-ssa-dse.c (decrement_count): New function.
27551 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27552 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27553 when we know the partially dead statement is a mem* function.
27555 PR tree-optimization/61912
27556 PR tree-optimization/77485
27557 * tree-ssa-dse.c: Include expr.h.
27558 (maybe_trim_constructor_store): New function.
27559 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27561 PR tree-optimization/33562
27562 PR tree-optimization/61912
27563 PR tree-optimization/77485
27564 * doc/invoke.texi: Document new dse-max-object-size param.
27565 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27566 * tree-ssa-dse.c: Include params.h.
27567 (dse_store_status): New enum.
27568 (initialize_ao_ref_for_dse): New, partially extracted from
27570 (valid_ao_ref_for_dse, normalize_ref): New.
27571 (setup_live_bytes_from_ref, compute_trims): Likewise.
27572 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27573 (maybe_trim_partially_dead_store): Likewise.
27574 (maybe_trim_complex_store): Likewise.
27575 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27576 Track what bytes live from the original store. Return tri-state
27577 for dead, partially dead or live.
27578 (dse_dom_walker): Add constructor, destructor and new private members.
27579 (delete_dead_call, delete_dead_assignment): New extracted from
27581 (dse_optimize_stmt): Make a member of dse_dom_walker.
27582 Use initialize_ao_ref_for_dse.
27584 PR tree-optimization/33562
27585 PR tree-optimization/61912
27586 PR tree-optimization/77485
27587 * sbitmap.h (bitmap_count_bits): Prototype.
27588 (bitmap_clear_range, bitmap_set_range): Likewise.
27589 * sbitmap.c (bitmap_clear_range): New function.
27590 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27592 2017-01-13 Martin Liska <mliska@suse.cz>
27595 * function.c (set_cfun): Add new argument force.
27596 * function.h (set_cfun): Likewise.
27597 * ipa-inline-transform.c (inline_call): Use the function when
27598 strict alising from is dropped for function we inline to.
27600 2017-01-13 Richard Biener <rguenther@suse.de>
27602 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27603 for dumping GIMPLE INTEGER_CSTs.
27605 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27607 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27608 to 201112L since C++17.
27610 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27613 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27614 if -fsanitize=kernel-address is present.
27616 2017-01-13 Richard Biener <rguenther@suse.de>
27618 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27619 as _Literal ( type ) number in case usual suffixes do not
27620 preserve all information.
27622 2017-01-13 Richard Biener <rguenther@suse.de>
27624 PR tree-optimization/77283
27625 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27626 and ssa-iterators.h.
27627 (is_feasible_trace): Implement a cost model based on joiner
27630 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27633 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27634 char or short to __float128/_Float128 directly.
27636 2017-01-12 Martin Sebor <msebor@redhat.com>
27638 to -Wformat-overflow.
27639 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27640 (min_bytes_remaining): Same.
27641 (get_string_length): Same.
27642 (format_string): Same.
27643 (format_directive): Same.
27645 (pass_sprintf_length::handle_gimple_call): Same.
27647 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27649 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27650 info.nowrite calls with no lhs that can't throw. Return bool
27651 whether gsi_remove has been called or not.
27652 (pass_sprintf_length::handle_gimple_call): Return bool whether
27653 try_substitute_return_value called gsi_remove. Formatting fix.
27654 (pass_sprintf_length::execute): Don't use gsi_remove if
27655 handle_gimple_call returned true.
27658 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27659 be removed due to side-effects, don't remove following barrier nor
27660 turn the successor edge into fallthru edge.
27662 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27665 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27666 element-reversing loads and stores as not swappable.
27668 2017-01-12 Nathan Sidwell <nathan@acm.org>
27669 Nicolai Stange <nicstange@gmail.com>
27671 * combine.c (try_combine): Don't ignore result of overlap checking
27672 loop. Combine overlap & asm check into single loop.
27674 2017-01-12 Richard Biener <rguenther@suse.de>
27676 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27677 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27679 2017-01-12 Richard Biener <rguenther@suse.de>
27681 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27682 and TS_TARGET_OPTION directly derive from TS_BASE.
27683 * tree-core.h (tree_optimization_option): Derive from tree_base.
27684 (tree_target_option): Likewise.
27686 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27688 * config/i386/i386.c (memory_address_length): Increase len
27689 only when rip_relative_addr_p returns false.
27691 2017-01-11 Julia Koval <julia.koval@intel.com>
27693 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27694 (OPTION_MASK_ISA_SGX_SET): New.
27695 (ix86_handle_option): Handle OPT_msgx.
27696 * config.gcc: Added sgxintrin.h.
27697 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27698 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27699 * config/i386/i386.c (ix86_target_string): Add -msgx.
27701 (ix86_option_override_internal): Handle new options.
27702 (ix86_valid_target_attribute_inner_p): Add sgx.
27703 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27704 * config/i386/i386.opt: Add msgx.
27705 * config/i386/sgxintrin.h: New file.
27706 * config/i386/x86intrin.h: Add sgxintrin.h.
27708 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27711 * fold-const.c (maybe_nonzero_address): Return 1 for function
27713 (tree_single_nonzero_warnv_p): Don't handle function local objects
27717 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27720 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27723 * opts.c: Include "spellcheck.h"
27724 (struct string_fragment): New struct.
27725 (struct edit_distance_traits<const string_fragment &>): New
27727 (get_closest_sanitizer_option): New function.
27728 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27730 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27732 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27734 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27735 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27736 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27737 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27738 for initial die_offset if dwarf_split_debug_info.
27739 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27740 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27742 (output_skeleton_debug_sections): Formatting fix. Use
27743 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27744 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27746 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27748 * config/arm/cortex-a53.md: Add bypasses for
27749 cortex_a53_r2f_cvt.
27750 (cortex_a53_r2f): Only use for transfers.
27751 (cortex_a53_f2r): Likewise.
27752 (cortex_a53_r2f_cvt): Add reservation for conversions.
27753 (cortex_a53_f2r_cvt): Likewise.
27755 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27757 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27758 to all inlined functions, change static to extern.
27760 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27763 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27765 (arm_assemble_integer): Likewise.
27767 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27769 * config.gcc: Use new awk script to check CPU, FPU and architecture
27770 parameters for --with-... options.
27771 * config/arm/parsecpu.awk: New file
27772 * config/arm/arm-cpus.in: New file.
27773 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27775 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27777 * config/arm/t-arm: Update dependency rules.
27778 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27779 of processing .def files.
27780 * config/arm/genopt.sh: Deleted.
27781 * config/arm/gentune.sh: Deleted.
27782 * config/arm/arm-cores.def: Deleted.
27783 * config/arm/arm-arches.def: Deleted.
27784 * config/arm/arm-fpus.def: Deleted.
27785 * config/arm/arm-tune.md: Regenerated.
27786 * config/arm/arm-tables.opt: Regenerated.
27787 * config/arm/arm-cpu.h: New generated file.
27788 * config/arm/arm-cpu-data.h: New generated file.
27789 * config/arm/arm-cpu-cdata.h: New generated file.
27791 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27794 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27796 (input_varpool_node): Unpack dynamically_initialized bit.
27798 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27800 PR rtl-optimization/79032
27801 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27802 the alignment of the adjusted memory reference against that of MODE,
27803 instead of the alignment of the original memory reference.
27805 2017-01-11 Martin Jambor <mjambor@suse.cz>
27807 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27809 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27810 decorated functions.
27812 2017-01-11 Richard Biener <rguenther@suse.de>
27814 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27815 set range/nonnull info for PHI results. Do not set it on
27816 stmts marked for removal.
27818 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27820 * expr.c (store_field): In the bitfield case, fetch the return value
27821 from the registers before applying a single big-endian adjustment.
27822 Always do a final load for a BLKmode value not larger than a word.
27824 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27827 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27828 that we correctly handle column numbers greater than
27829 LINE_MAP_MAX_COLUMN_NUMBER.
27831 2017-01-10 Martin Sebor <msebor@redhat.com>
27833 PR middle-end/78245
27834 * gimple-ssa-sprintf.c (get_destination_size): Call
27835 {init,fini}object_sizes.
27836 * tree-object-size.c (addr_object_size): Adjust.
27837 (pass_through_call): Adjust.
27838 (pass_object_sizes::execute): Adjust.
27839 * tree-object-size.h (fini_object_sizes): Declare.
27841 2017-01-10 Martin Sebor <msebor@redhat.com>
27843 PR tree-optimization/78775
27844 * builtins.c (get_size_range): Move...
27845 * calls.c: ...to here.
27846 (alloc_max_size): Accept zero argument.
27847 (operand_signed_p): Remove.
27848 (maybe_warn_alloc_args_overflow): Call get_size_range.
27849 * calls.h (get_size_range): Declare.
27851 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27853 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27854 from TI's devices.csv file as of September 2016.
27855 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27857 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27859 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27860 * doc/invoke.texi: Likewise.
27861 * doc/md.texi: Likewise.
27862 * doc/objc.texi: Likewise.
27864 2017-01-10 Joshua Conner <joshconner@google.com>
27866 * config/arm/fuchsia-elf.h: New file.
27867 * config/fuchsia.h: New file.
27868 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27869 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27871 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27873 2016-01-10 Richard Biener <rguenther@suse.de>
27875 PR tree-optimization/79034
27876 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27877 Propagate out degenerate PHIs in the joiner.
27879 2017-01-10 Martin Liska <mliska@suse.cz>
27881 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27882 (sort_congruence_classes_by_decl_uid): Likewise.
27883 (sort_congruence_class_groups_by_decl_uid): Likewise.
27884 (sem_item_optimizer::merge_classes): Sort class, groups in these
27885 classes and members in the groups by DECL_UID of declarations.
27886 This would make merge operations stable.
27888 2017-01-10 Martin Liska <mliska@suse.cz>
27890 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27891 usage of m_classes_vec.
27892 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27893 (sem_item_optimizer::get_group_by_hash): Likewise.
27894 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27895 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27896 (sem_item_optimizer::verify_classes): Likewise.
27897 (sem_item_optimizer::process_cong_reduction): Likewise.
27898 (sem_item_optimizer::dump_cong_classes): Likewise.
27899 (sem_item_optimizer::merge_classes): Likewise.
27900 * ipa-icf.h (congruence_class_hash): Rename from
27901 congruence_class_group_hash. Remove declaration of m_classes_vec.
27903 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27905 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27906 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27907 * config.gcc: Add avx512vpopcntdqintrin.h.
27908 * config/i386/avx512vpopcntdqintrin.h: New.
27909 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27910 * config/i386/i386-builtin-types.def: Add new types.
27911 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27912 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27913 __builtin_ia32_vpopcountq_v8di_mask): New.
27914 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27915 __AVX512VPOPCNTDQ__.
27916 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27917 (PTA_AVX512VPOPCNTDQ): Define.
27918 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27919 TARGET_AVX512VPOPCNTDQ_P): Define.
27920 * config/i386/i386.opt: Add mavx512vpopcntdq.
27921 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27922 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27924 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27926 PR middle-end/77484
27927 * predict.def (PRED_CALL): Set to 67.
27929 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27931 * expr.c (store_field): In the bitfield case, if the value comes from
27932 a function call and is of an aggregate type returned in registers, do
27933 not modify the field mode; extract the value in all cases if the mode
27934 is BLKmode and the size is not larger than a word.
27936 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27939 * config/i386/cpuid.h: Fix undefined behavior.
27941 2017-01-04 Jeff Law <law@redhat.com>
27943 PR tree-optimization/79007
27944 PR tree-optimization/67955
27945 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27946 conservative for pt.null when flag_non_call_exceptions is on.
27948 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27950 PR translation/79019
27951 PR translation/79020
27952 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27953 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27954 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27956 * config/avr/avr.opt (maccumulate-args): Likewise.
27957 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27958 * common.opt (freport-bug): Likewise.
27959 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27960 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27961 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27962 translatable string.
27963 * config/i386/i386.c (function_value_32): Likewise.
27964 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27965 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27967 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27968 * common/config/msp430/msp430-common.c (msp430_handle_option):
27970 * symtab.c (symtab_node::verify_base): Likewise.
27971 * opts.c (set_debug_level): Likewise.
27972 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27973 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27974 missing whitespace to translatable strings.
27975 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27976 * config/sh/superh.h: Likewise.
27977 * config/i386/xopintrin.h: Likewise.
27978 * config/i386/znver1.md: Likewise.
27979 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27980 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27981 * double-int.h (struct double_int): Likewise.
27982 * double-int.c (div_and_round_double): Likewise.
27983 * wide-int.cc: Likewise.
27984 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27985 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27986 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27987 (crossjumps_occurred): ... this.
27988 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27991 PR tree-optimization/78899
27992 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27993 returning bool return struct loop *, NULL for failure and the new
27995 (versionable_outer_loop_p): Don't version outer loop if it has
27996 dont_vectorized bit set.
27997 (tree_if_conversion): When versioning outer loop, ensure
27998 tree_if_conversion is performed also on the inner loop of the
27999 non-vectorizable outer loop copy.
28000 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28001 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28002 prevent vectorization of it.
28003 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28004 the outer loop vectorization of the non-scalar version is attempted
28005 before vectorization of the inner loop in scalar version. If
28006 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28007 vectorization of its inner loop.
28008 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28009 has 2 inner loops, rename also on edges from bb whose single pred
28010 is outer_loop->header. Fix typo in function comment.
28012 2017-01-09 Martin Sebor <msebor@redhat.com>
28015 * asan.c (asan_emit_stack_protection): Increase local buffer size
28016 to avoid snprintf truncation warning.
28018 2017-01-09 Andrew Pinski <apinski@cavium.com>
28020 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28021 to reference thunderx2t99 for the tuning structure
28022 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28024 (thunderx2t99_extra_costs): This.
28025 * config/aarch64/aarch64-tune.md: Regenerate.
28026 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28027 (vulcan_addrcost_table): This.
28028 (vulcan_regmove_cost): Rename to ...
28029 (thunderx2t99_regmove_cost): This.
28030 (vulcan_vector_cost): Rename to ...
28031 (thunderx2t99_vector_cost): this.
28032 (vulcan_branch_cost): Rename to ...
28033 (thunderx2t99_branch_cost): This.
28034 (vulcan_tunings): Rename to ...
28035 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28036 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28038 2017-01-09 Martin Jambor <mjambor@suse.cz>
28042 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28043 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28044 (propagate_vr_accross_jump_function): Use the above function for all
28045 value range computations for pass-through jump functions and type
28046 converasion from explicit value range values.
28047 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28048 parameters from TYPE_ARG_TYPES.
28049 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28050 (ipa_write_node_info): Stream type of the actual argument.
28051 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28053 2017-01-09 Martin Liska <mliska@suse.cz>
28056 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28057 (lookup_compiler): Do not show error message with have_E.
28059 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28061 PR tree-optimization/78938
28062 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28063 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28064 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28065 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28068 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28070 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28073 2017-01-09 Richard Biener <rguenther@suse.de>
28075 PR tree-optimization/78997
28076 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28077 name condition properly.
28079 2017-01-09 Richard Biener <rguenther@suse.de>
28082 * dwarf2out.c (is_cxx): New overload with context.
28083 (is_naming_typedef_decl): Use it.
28085 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28087 * invoke.texi (Option Summary): Correct spacing in option lists
28088 and add line breaks to fix over-long lines.
28090 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28092 PR middle-end/17660
28094 * extend.texi (Common Variable Attributes): Add xref to GCC
28095 Internals manual to explain mode attribute keywords.
28097 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28100 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28101 and Preprocessor Options.
28102 (Options for Linking): Document -pthread here....
28103 (RS/6000 and PowerPC Options): ...not here.
28104 (Solaris 2 Options): ...or here.
28105 * doc/cppopts.texi: Document -pthread.
28107 2017-01-08 Martin Sebor <msebor@redhat.com>
28109 PR middle-end/77708
28110 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28111 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28112 New member functions.
28113 (format_directive): Used them.
28115 (pass_sprintf_length::handle_gimple_call): Same.
28116 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28117 to avoid truncation for any argument.
28118 (extract_affine_mul): Same.
28119 * tree.c (get_file_function_name): Same.
28121 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28123 PR middle-end/77484
28124 * predict.def (PRED_INDIR_CALL): Set to 86.
28126 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28128 PR preprocessor/54124
28129 * doc/cppopts.texi: Reformat -d subtable to list the full name
28130 of the options. Add cross-reference to the docs for the general
28131 compiler -d options.
28132 * doc/invoke.texi (Developer Options): Add cross-reference to the
28133 preprocessor-specific -d option documentation.
28135 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28137 PR preprocessor/13498
28138 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28139 redudant material, and reflect new command-line options.
28140 (System Headers): Likewise.
28142 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28144 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28145 -isystem, and -idirafter. Copy-edit.
28146 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28147 default for -ftrack-macro-expansion. Delete obsolete and
28148 badly-formatted implementation details about -fdebug-cpp output.
28149 * doc/cppwarnopts.texi: Copy-edit.
28151 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28154 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28155 that the transition from a max line width >= 1<<10 to narrower
28156 lines works correctly.
28158 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28160 * doc/options.texi (PerFunction): New.
28161 * opt-functions.awk (switch_flags): Map both Optimization and
28162 PerFunction to CL_OPTIMIZATION.
28163 * opth-gen.awk: Test for PerFunction flag along with
28165 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28166 it only when the latter is present. Skip those that don't in
28167 the hash function generator.
28168 * common.opt (fvar-tracking): Mark as PerFunction instead of
28170 (fvar-tracking-assignments): Likewise.
28171 (fvar-tracking-assignments-toggle): Likewise.
28172 (fvar-tracking-uninit): Likewise.
28174 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28176 PR translation/79018
28177 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28180 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28183 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28184 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28185 TARGET_LONG_JUMP_TABLE_OFFSETS.
28186 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28187 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28188 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28189 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28190 * config/m68k/m68k.md (tablejump expander): Likewise.
28191 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28192 TARGET_LONG_JUMP_TABLE_OFFSETS.
28193 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28194 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28196 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28197 David Holsgrove <david.holsgrove@xilinx.com>
28199 * common/config/microblaze/microblaze-common.c
28200 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28201 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28203 * config/microblaze/microblaze.c (microblaze_must_save_register)
28204 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28206 (microblaze_eh_return): New function.
28207 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28208 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28209 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28210 * config/microblaze/microblaze.md (eh_return): New pattern.
28212 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28214 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28215 GCC_DIAGNOSTIC_STRINGIFY): Define.
28217 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28219 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28221 * config/arm/arm.md (<mcrr>): New.
28223 * config/arm/arm.c (arm_arch5te): New.
28224 (arm_option_override): Set arm_arch5te.
28225 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28227 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28228 (arm_mcrr_qualifiers): ... this. New.
28229 (MRRC_QUALIFIERS): Define to...
28230 (arm_mrrc_qualifiers): ... this. New.
28231 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28233 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28234 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28235 (MRRCI, mrrc, MRRC): New.
28236 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28237 VUNSPEC_MRRC2): New.
28239 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28241 * config/arm/arm.md (<mcr>): New.
28243 * config/arm/arm.c (arm_coproc_builtin_available): Add
28244 support for mcr, mrc, mcr2 and mrc2.
28245 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28246 (arm_mcr_qualifiers): ... this. New.
28247 (MRC_QUALIFIERS): Define to ...
28248 (arm_mrc_qualifiers): ... this. New.
28249 (MCR_QUALIFIERS): Define to ...
28250 (arm_mcr_qualifiers): ... this. New.
28251 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28253 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28254 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28255 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28256 VUNSPEC_MRC2): New.
28258 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28260 * config/arm/arm.md (*ldc): New.
28264 * config/arm/arm.c (arm_coproc_builtin_available): Add
28265 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28266 (arm_coproc_ldc_stc_legitimate_address): New.
28267 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28268 'qualifier_const_pointer'.
28269 (LDC_QUALIFIERS): Define to...
28270 (arm_ldc_qualifiers): ... this. New.
28271 (STC_QUALIFIERS): Define to...
28272 (arm_stc_qualifiers): ... this. New.
28273 * config/arm/arm-protos.h
28274 (arm_coproc_ldc_stc_legitimate_address): New.
28275 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28276 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28277 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28278 stc2, stcl, stc2l): New.
28279 * config/arm/constraints.md (Uz): New.
28280 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28281 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28282 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28283 VUNSPEC_STC2L): New.
28285 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28287 * config/arm/arm.md (<cdp>): New.
28288 * config/arm/arm.c (neon_const_bounds): Rename this ...
28289 (arm_const_bounds): ... this.
28290 (arm_coproc_builtin_available): New.
28291 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28292 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28293 (CDP_QUALIFIERS): Define to...
28294 (arm_cdp_qualifiers): ... this. New.
28296 (arm_expand_builtin_args): Add case for 6 arguments.
28297 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28298 (arm_const_bounds): ... this.
28299 (arm_coproc_builtin_available): New.
28300 * config/arm/arm_acle.h (__arm_cdp): New.
28302 * config/arm/arm_acle_builtins.def (cdp): New.
28304 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28305 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28306 'arm_const_bounds'.
28307 * config/arm/types.md (coproc): New.
28308 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28309 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28310 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28311 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28313 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28315 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28316 (UBINOP_QUALIFIERS): New.
28318 (acle_builtin_data): New. Change comment.
28319 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28320 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28321 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28322 arm_acle_builtins.def.
28323 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28324 (arm_init_acle_builtins): New.
28325 (CRC32_BUILTIN): Remove.
28326 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28327 crc32cb, crc32ch and crc32cw.
28328 (arm_init_crc32_builtins): Remove.
28329 (arm_init_builtins): Use arm_init_acle_builtins rather
28330 than arm_init_crc32_builtins.
28331 (arm_expand_acle_builtin): New.
28332 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28333 * config/arm/arm_acle_builtins.def: New.
28335 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28337 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28338 (arm_builtin_datum): ... this.
28339 (arm_init_neon_builtin): Rename to ...
28340 (arm_init_builtin): ... this. Add a new parameters PREFIX
28341 and USE_SIG_IN_NAME.
28342 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28343 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28344 'arm_builtin_datum'.
28345 (arm_init_vfp_builtins): Likewise.
28346 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28347 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28348 (arm_expand_neon_args): Rename to ...
28349 (arm_expand_builtin_args): ... this. Rename builtin_arg
28350 enum values and differentiate between ARG_BUILTIN_MEMORY
28351 and ARG_BUILTIN_NEON_MEMORY.
28352 (arm_expand_neon_builtin_1): Rename to ...
28353 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28354 values, arm_expand_builtin_args and add bool parameter NEON.
28355 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28356 (arm_expand_vfp_builtin): Likewise.
28357 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28359 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28361 PR middle-end/77484
28362 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28363 * predict.c (tree_estimate_probability_bb): Reverse direction of
28364 polymorphic call predictor.
28366 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28368 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28369 (determine_pass_name_match): ...this new function and...
28370 (should_skip_pass_p): ...this new function.
28372 2017-01-06 Nathan Sidwell <nathan@acm.org>
28374 * ipa-visibility.c (function_and_variable_visibility): Reformat
28375 comments and long lines. Remove extrneous if.
28376 * symtab.c (symtab_node::make_decl_local): Fix code format.
28377 (symtab_node::set_section_for_node): Fix comment typo.
28379 2017-01-06 Martin Liska <mliska@suse.cz>
28382 * lra-constraints.c: Rename invariant to lra_invariant.
28383 * predict.c (set_even_probabilities): Initialize e to NULL.
28385 2017-01-05 Martin Sebor <msebor@redhat.com>
28387 PR tree-optimization/78910
28388 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28389 (format_integer): Correct off-by-one error in the handling
28390 of precision with negative numbers in signed conversions..
28392 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28394 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28396 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28398 PR tree-optimization/71016
28399 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28400 factor_out_conditional_conversion. Formatting fix.
28401 (factor_out_conditional_conversion): Add cond_stmt argument.
28402 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28403 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28406 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28408 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28409 read-rtl-function.o, and selftest-rtl.o.
28410 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28411 (selftest::aarch64_test_loading_full_dump): New function.
28412 (selftest::aarch64_run_selftests): New function.
28413 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28414 selftest::aarch64_run_selftests.
28415 * config/i386/i386.c
28416 (selftest::ix86_test_loading_dump_fragment_1): New function.
28417 (selftest::ix86_test_loading_call_insn): New function.
28418 (selftest::ix86_test_loading_full_dump): New function.
28419 (selftest::ix86_test_loading_unspec): New function.
28420 (selftest::ix86_run_selftests): Call the new functions.
28421 * emit-rtl.c (maybe_set_max_label_num): New function.
28422 * emit-rtl.h (maybe_set_max_label_num): New decl.
28423 * function.c (instantiate_decls): Guard call to
28424 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28425 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28427 * gensupport.c (gen_reader::gen_reader): Pass "false"
28428 for new "compact" param of rtx_reader.
28429 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28430 rather than an empty string for NULL strings.
28431 * read-md.c: Potentially include config.h rather than bconfig.h.
28432 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28433 (have_error): New global, copied from errors.c.
28434 (md_reader::read_name): Rename to...
28435 (md_reader::read_name_1): ...this, adding "out_loc" param,
28436 and converting "missing name or number" to returning false, rather
28438 (md_reader::read_name): Reimplement in terms of read_name_1.
28439 (md_reader::read_name_or_nil): New function.
28440 (md_reader::read_string): Handle "(nil)" by returning NULL.
28441 (md_reader::md_reader): Add new param "compact".
28442 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28443 (md_reader::read_file): New method.
28444 * read-md.h (md_reader::md_reader): Add new param "compact".
28445 (md_reader::read_file): New method.
28446 (md_reader::is_compact): New accessor.
28447 (md_reader::read_name): Convert return type from void to file_location.
28448 (md_reader::read_name_or_nil): New decl.
28449 (md_reader::read_name_1): New decl.
28450 (md_reader::m_compact): New field.
28451 (noop_reader::noop_reader): Pass "false" for new "compact" param
28453 (rtx_reader::rtx_reader): Add new "compact" param.
28454 (rtx_reader::read_rtx_operand): Make virtual and convert return
28455 type from void to rtx.
28456 (rtx_reader::read_until): New decl.
28457 (rtx_reader::handle_any_trailing_information): New virtual function.
28458 (rtx_reader::postprocess): New virtual function.
28459 (rtx_reader::finalize_string): New virtual function.
28460 (rtx_reader::m_in_call_function_usage): New field.
28461 (rtx_reader::m_reuse_rtx_by_id): New field.
28462 * read-rtl-function.c: New file.
28463 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28464 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28465 (selftest::verify_three_block_rtl_cfg): New decl.
28466 * read-rtl-function.h: New file.
28467 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28468 For host, include function.h, memmodel.h, and emit-rtl.h.
28469 (one_time_initialization): New function.
28470 (struct compact_insn_name): New struct.
28471 (compact_insn_names): New array.
28472 (find_code): Handle insn codes in compact dumps.
28473 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28474 (bind_subst_iter_and_attr): Likewise.
28475 (add_condition_to_string): Likewise.
28476 (add_condition_to_rtx): Likewise.
28477 (apply_attribute_uses): Likewise.
28478 (add_current_iterators): Likewise.
28479 (apply_iterators): Likewise.
28480 (initialize_iterators): Guard usage of apply_subst_iterator with
28481 #ifdef GENERATOR_FILE.
28482 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28483 (md_reader::read_mapping): Likewise.
28484 (add_define_attr_for_define_subst): Likewise.
28485 (add_define_subst_attr): Likewise.
28486 (read_subst_mapping): Likewise.
28487 (check_code_iterator): Likewise.
28488 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28490 (one_time_initialization): New function.
28491 (rtx_reader::read_until): New method.
28492 (read_flags): New function.
28493 (parse_reg_note_name): New function.
28494 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28495 Handle reuse_rtx ids.
28496 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28497 Add parsing support for RTL dumps, mirroring the special-cases in
28498 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28499 values, and calling handle_any_trailing_information.
28500 (rtx_reader::read_rtx_operand): Convert return type from void
28501 to rtx, returning return_rtx. Handle case 'e'. Call
28502 finalize_string on XSTR and XTMPL fields.
28503 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28504 "(nil)" values were omitted. Call the postprocess vfunc on the
28506 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28507 class ctor. Initialize m_in_call_function_usage. Call
28508 one_time_initialization.
28509 * rtl-tests.c (selftest::test_uncond_jump): Call
28510 set_new_first_and_last_insn.
28511 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28512 * selftest-rtl.c: New file.
28513 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28514 (selftest::get_insn_by_uid): New decl.
28515 * selftest-run-tests.c (selftest::run_tests): Call
28516 read_rtl_function_c_tests.
28517 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28518 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28521 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28523 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28524 operands in a special way. Assert that pos+len <= mode precision.
28526 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28528 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28529 3 argument Alias with unlimited for the negative form.
28530 (fno-vect-cost-model): Removed.
28532 2017-01-05 Martin Liska <mliska@suse.cz>
28534 * hsa-gen.c (gen_hsa_divmod): New function.
28535 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28537 2017-01-05 Martin Liska <mliska@suse.cz>
28540 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28543 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28545 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28546 small constant length operands.
28548 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28550 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28551 between loop iterations.
28553 2017-01-05 Martin Liska <mliska@suse.cz>
28556 * gimplify.c (gimplify_decl_expr): Compare to
28557 asan_poisoned_variables instread of checking flags.
28558 (gimplify_target_expr): Likewise.
28559 (gimplify_expr): Likewise.
28560 (gimplify_function_tree): Conditionally initialize
28561 asan_poisoned_variables.
28563 2017-01-04 Jeff Law <law@redhat.com>
28565 PR tree-optimizatin/78812
28566 * rtl.h (contains_mem_rtx_p): Prototype.
28567 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28568 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28569 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28570 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28571 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28573 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28575 * input.c (assert_char_at_range): Default-initialize actual_range.
28577 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28579 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28580 to match the caller.
28582 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28584 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28585 insns after final jump in test to emit dummy move.
28587 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28589 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28590 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28592 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28594 * multiple_target.c (create_dispatcher_calls): Init e_next.
28595 * tree-ssa-loop-split.c (split_loop): Init border.
28596 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28599 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28604 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28605 (altivec_register_operand): Do not return true if the operand
28606 contains a SUBREG mixing SImode and SFmode.
28607 (vsx_register_operand): Likewise.
28608 (vsx_reg_sfsubreg_ok): New predicate.
28609 (vfloat_operand): Do not return true if the operand contains a
28610 SUBREG mixing SImode and SFmode.
28611 (vint_operand): Likewise.
28612 (vlogical_operand): Likewise.
28613 (gpc_reg_operand): Likewise.
28614 (int_reg_operand): Likewise.
28615 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28616 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28617 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28619 (rs6000_emit_move_si_sf_subreg): New helper function.
28620 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28621 fixup SUBREGs involving SImode and SFmode.
28622 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28623 numbers for the new peephole2 optimization.
28624 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28625 the GLIBC math library that do AND/IOR/XOR operations on single
28626 precision floating point.
28627 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28628 target macros to say whether we need to avoid SUBREGs mixing
28630 (TARGET_ALLOW_SF_SUBREG): Likewise.
28631 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28632 (UNSPEC_SI_FROM_SF): Likewise.
28633 (iorxor): Change spacing.
28634 (and_ior_xor): New iterator for AND, IOR, and XOR.
28635 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28636 (movdi_from_sf_zero_ext): Likewise.
28637 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28638 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28639 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28640 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28641 (fms<mode>4): Likewise.
28642 (fnma<mode>4): Likewise.
28643 (fnms<mode>4): Likewise.
28644 (nfma<mode>4): Likewise.
28645 (nfms<mode>4): Likewise.
28647 2017-01-04 Marek Polacek <polacek@redhat.com>
28650 * doc/invoke.texi: Document -Wpointer-compare.
28652 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28654 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28657 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28658 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28661 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28664 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28665 documentation of the powerpc_popcntb_ok attribute.
28666 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28667 code to issue warning messages if a requested CPU configuration is
28668 not supported by the binary (assembler and loader) toolchain.
28669 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28670 made to define a built-in function that has been disabled.
28671 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28672 made to define a built-in function that has been disabled.
28673 (altivec_init_builtins): Add comment explaining why definition
28674 of the DST built-in functions is not preceded by an assertion
28675 check. Add assertions to prevent ICE if attempts are made to
28676 define an altivec predicate or an abs* built-in function that has
28678 (htm_init_builtins): Add comment explaining why definition of the
28679 htm built-in functions is not preceded by an assertion check.
28681 2017-01-04 Jeff Law <law@redhat.com>
28683 PR tree-optimizatin/67955
28684 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28685 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28686 the points-to solution does not include pt_null. Use DECL_PT_UID
28689 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28691 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28692 Use gen_int_mode instead of gen_lopwart for const_int operands.
28694 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28696 PR tree-optimization/71563
28697 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28698 out of range value - has low bits known to be zero.
28700 2017-01-04 Alan Modra <amodra@gmail.com>
28702 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28703 * configure: Regenerate.
28704 * config.in: Regenerate.
28706 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28709 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28710 a substring of the message, but strcmp with the whole message. Ifdef
28711 ENABLE_NLS, translate the message first using dgettext.
28713 2017-01-03 Jeff Law <law@redhat.com>
28715 PR tree-optimizatin/78856
28716 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28717 (mark_threaded_blocks): Remove code to truncate thread paths that
28718 cross multiple loop headers. Instead invalidate the cached loop
28719 iteration information and handle case of a thread path walking
28720 into an irreducible region.
28722 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28725 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28726 assertions. Add support for doing the signbit if the IEEE 128-bit
28727 floating point value is in a GPR.
28728 * config/rs6000/rs6000.md (Fsignbit): Delete.
28729 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28730 Update the length attribute if the value is in a GPR.
28731 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28732 the sign or zero extension instruction, since the value is always 0/1.
28733 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28736 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28737 extracting SImode to a GPR register so that we can generate a
28738 store, limit the vector to be in a traditional Altivec register
28739 for the vextuwrx instruction.
28741 2017-01-03 Ian Lance Taylor <iant@google.com>
28743 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28745 2017-01-03 Martin Sebor <msebor@redhat.com>
28747 PR tree-optimization/78696
28748 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28749 precision. Use MPFR for %f for greater fidelity. Correct handling
28751 (pass_sprintf_length::compute_format_length): Set width and precision
28752 specified by asrerisk to void_node for vararg functions.
28753 (try_substitute_return_value): Adjust dump output.
28755 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28757 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28759 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28761 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28762 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28763 -mlra/-mno-lra was passed to the compiler.
28765 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28767 PR rtl-optimization/65618
28768 * emit-rtl.c (try_split): Move initialization of "before" and
28769 "after" to just before the call to emit_insn_after_setloc.
28771 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28773 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28775 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28777 * dwarf2out.c (gen_enumeration_type_die): When
28778 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28780 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28782 PR tree-optimization/78965
28783 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28784 Change first argument from const call_info & to call_info &. For %n
28785 set info.nowrite to false.
28787 PR middle-end/78901
28788 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28789 possibly throwing calls.
28791 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28792 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28793 and fns handling, rather than in a separate case SSA_NAME.
28795 2017-01-02 Jeff Law <law@redhat.com>
28797 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28798 fixes for first_period and second_period variables.
28800 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28803 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28804 (*insvqi_1): New insn pattern.
28805 (*insvqi_1_mem_rex64): Ditto.
28806 (*insvqi_2): Ditto.
28807 (*insvqi_3): Rename from *insvqi.
28809 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28811 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28813 * doc/cfg.texi (Edges): Remove reference to Java.
28814 (Maintaining the CFG): Ditto.
28816 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28818 PR middle-end/77674
28819 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28820 transparent aliases.
28822 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28824 PR middle-end/77484
28825 * predict.def (PRED_CALL): Update hitrate.
28826 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28827 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28828 into direct/indirect/polymorphic variants.
28830 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28832 Update copyright years.
28834 * gcc.c (process_command): Update copyright notice dates.
28835 * gcov-dump.c (print_version): Ditto.
28836 * gcov.c (print_version): Ditto.
28837 * gcov-tool.c (print_version): Ditto.
28838 * gengtype.c (create_file): Ditto.
28839 * doc/cpp.texi: Bump @copying's copyright year.
28840 * doc/cppinternals.texi: Ditto.
28841 * doc/gcc.texi: Ditto.
28842 * doc/gccint.texi: Ditto.
28843 * doc/gcov.texi: Ditto.
28844 * doc/install.texi: Ditto.
28845 * doc/invoke.texi: Ditto.
28847 Copyright (C) 2017 Free Software Foundation, Inc.
28849 Copying and distribution of this file, with or without modification,
28850 are permitted in any medium without royalty provided the copyright
28851 notice and this notice are preserved.