1 2017-11-23 Mike Stump <mikestump@comcast.net>
2 Eric Botcazou <ebotcazou@adacore.com>
4 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5 * cfgloop.h (struct loop): Add unroll field.
6 * function.h (struct function): Add has_unroll bitfield.
7 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
8 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
9 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
10 (pass_rtl_unroll_loops::gate): Likewise.
11 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
12 for which loop->unroll==1.
13 (decide_unroll_constant_iterations): Use note for consistency and
14 take loop->unroll into account. Return early if loop->unroll is set.
15 Fix thinko in existing test.
16 (decide_unroll_runtime_iterations): Use note for consistency and
17 take loop->unroll into account.
18 (decide_unroll_stupid): Likewise.
19 * lto-streamer-in.c (input_cfg): Read loop->unroll.
20 * lto-streamer-out.c (output_cfg): Write loop->unroll.
21 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
23 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
24 (print_loop): Print loop->unroll if set.
25 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
26 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
27 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
29 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
30 loop->unroll is set and smaller than the trip count. Otherwise bypass
31 entirely the heuristics if loop->unroll is set. Remove dead note.
32 Fix off-by-one bug in other note.
33 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
34 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
36 (tree_unroll_loops_completely): Make static.
37 (pass_complete_unroll::execute): Use correct type for variable.
38 (pass_complete_unrolli::execute): Fix formatting.
39 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
41 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
43 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
44 enable when TARGET_PREFER_AVX128 is set.
46 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
48 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
50 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
51 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
53 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
55 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
56 frequencies. Fix estimation of aggregate parameters.
58 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
60 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
63 2017-11-23 Tom de Vries <tom@codesourcery.com>
65 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
67 2017-11-23 Tom de Vries <tom@codesourcery.com>
69 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
72 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
75 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
76 sibcall_value_pcrel_fdpic): Use local variable instead of
78 (calli_tbr_rel): Add missing operand 2.
79 (call_valuei_tbr_rel): Add missing operand 3.
81 2017-11-23 Jakub Jelinek <jakub@redhat.com>
84 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
85 bitpos/bitsize covering the whole destination, use store_expr only if
86 the complex mode is the same. Otherwise, use expand_normal and if
87 it returns CONCAT, subreg each part separately instead of trying to
88 subreg the whole result.
90 2017-11-23 Richard Biener <rguenther@suse.de>
92 PR tree-optimization/23094
93 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
94 come here from walking over backedges in the first iteration.
95 (vn_reference_lookup_3): Skip clobbers that store the same value.
97 2017-11-23 Richard Biener <rguenther@suse.de>
99 PR tree-optimization/81403
100 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
101 a block we need a leader relative to.
102 (phi_translate_1): For nary processing require a leader from
103 get_representative_for given we run expression simplification
104 using match-and-simplify. Remove previous fix.
106 2017-11-22 Jeff Law <law@redhat.com>
108 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
109 Use new method allocate_value_range rather than accessing the
110 vrp_value_range_pool data member directly.
111 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
112 to use extract_range_from_stmt method to avoid need for
113 extract_range_from_assignment method.
114 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
115 method rather than setting values_propgated data member directly.
116 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
117 and values propagated data members and extract_range_from_assignment
118 method. Reorder private data members to conform to standards.
119 Add new methods set_lattice_propagation_complete and
120 allocate_value_range.
122 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
124 PR rtl-optimization/83030
125 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
126 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
127 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
128 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
129 CROSSING_JUMP_P flag is already set before setting it.
130 * cfgrtl.c (fixup_partition_crossing): Likewise.
131 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
134 2017-11-22 Jakub Jelinek <jakub@redhat.com>
136 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
137 Handle the case where both arguments are using gen_const_vec_series.
139 2017-11-22 David Malcolm <dmalcolm@redhat.com>
142 * pretty-print.c (pp_format): Move quoting implementation to
143 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
144 to pass address of "quote" local.
145 (pp_begin_quote): New function.
146 (pp_end_quote): New function.
147 * pretty-print.h (printer_fn): Convert penultimate param from bool
149 (pp_begin_quote): New decl.
150 (pp_end_quote): New decl.
151 * tree-diagnostic.c (default_tree_printer): Convert penultimate
152 param from bool to bool *.
153 * tree-diagnostic.h (default_tree_printer): Likewise.
155 2017-11-22 Jeff Law <law@redhat.com>
157 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
158 Initialize vr_values.
159 (evrp_range_analyzer::try_find_new_range): Call methods attached to
160 vr_values via vr_values class instance rather than delegators.
161 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
162 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
163 (evrp_range_analyzer::push_value_range): Likewise.
164 (evrp_range_analyzer::pop_value_range): Likewise.
165 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
166 most delegators. Those remaining are exposed as public interfaces.
167 Make vr_values a pointer and private.
168 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
170 (evrp_range_analyzer::get_vr_value): New method.
171 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
172 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
173 (class evrp_dom_walker): Attach evrp_folder class, initialize
174 it in the ctor. Remove temporary delegators.
175 (evrp_dom_walker::before_dom_children): Call methods in attached
176 evrp_range_analyzer class via class instance pointer. Use
177 free value_range_constant_singleton to remove need for
178 op_with_constant_singleton_value delegator method. Do not
179 create a vrp_prop class instance for every call! Narrow
180 scope of a couple variables.
181 (evrp_dom_walker::cleanup): Call methods in attached
182 evrp_range_analyzer class via class instance pointer.
183 * vr-values.h (class vr_values): Privatize many methods and
186 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
188 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
189 * cfgexpand.c (expand_debug_expr): Remove handling for them.
190 * expr.c (expand_expr_real_2): Likewise.
191 * fold-const.c (const_unop): Likewise.
192 * optabs-tree.c (optab_for_tree_code): Likewise.
193 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
194 * tree-inline.c (estimate_operator_cost): Likewise.
195 * tree-pretty-print.c (dump_generic_node): Likewise.
196 (op_code_prio): Likewise.
197 (op_symbol_code): Likewise.
198 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
199 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
200 * internal-fn.c (direct_internal_fn_optab): New function.
201 (direct_internal_fn_array, direct_internal_fn_supported_p
202 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
203 * fold-const-call.c (fold_const_reduction): New function.
204 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
206 * tree-vect-loop.c: Include internal-fn.h.
207 (reduction_code_for_scalar_code): Rename to...
208 (reduction_fn_for_scalar_code): ...this and return an internal
210 (vect_model_reduction_cost): Take an internal_fn rather than
212 (vect_create_epilog_for_reduction): Likewise. Build calls rather
214 (vectorizable_reduction): Use internal functions rather than tree
215 codes for the reduction operation. Update calls to the functions
217 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
218 Use calls to internal functions rather than REDUC tree codes.
219 * config/aarch64/aarch64-simd.md: Update comment accordingly.
221 2017-11-22 Olivier Hainque <hainque@adacore.com>
223 * config/vxworks.c (vxworks_override_options): Pick default
224 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
225 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
226 DWARF_GNAT_ENCODINGS_DEFAULT.
227 * config/vxworksae.h: Likewise.
229 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
231 PR tree-optimization/83104
232 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
235 2017-11-22 Marek Polacek <polacek@redhat.com>
236 H.J. Lu <hongjiu.lu@intel.com>
237 Jason Merrill <jason@redhat.com>
242 * calls.c (initialize_argument_information): Call
243 warn_parameter_passing_abi target hook.
244 (store_one_arg): Use 0 for empty record size. Don't push 0 size
246 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
247 * common.opt: Update -fabi-version description.
248 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
249 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
251 (ix86_is_empty_record): New function.
252 (ix86_warn_parameter_passing_abi): New function.
253 (TARGET_EMPTY_RECORD_P): Redefine.
254 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
255 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
256 * doc/tm.texi: Regenerated.
257 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
258 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
259 * dwarf2out.c (get_ultimate_context): Move to tree.c.
260 * explow.c (hard_function_value): Call arg_int_size_in_bytes
261 instead of int_size_in_bytes.
262 * expr.c (copy_blkmode_to_reg): Likewise.
263 * function.c (aggregate_value_p): Return 0 for empty types.
264 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
265 (locate_and_pad_parm): Call arg size_in_bytes instead
267 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
268 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
269 * target.def (empty_record_p, warn_parameter_passing_abi): New target
271 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
272 (std_gimplify_va_arg_expr): Skip empty records. Call
273 arg_size_in_bytes instead size_in_bytes.
274 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
275 * tree-core.h (tree_type_common): Add empty_flag.
276 (tree_decl_common): Update comments.
277 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
279 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
280 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
282 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
283 * tree.c (default_is_empty_type): New function.
284 (default_is_empty_record): New function.
285 (arg_int_size_in_bytes): New function.
286 (arg_size_in_bytes): New function.
287 (get_ultimate_context): New function.
288 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
289 TRANSLATION_UNIT_WARN_EMPTY_P.
290 (default_is_empty_record, arg_int_size_in_bytes,
291 arg_size_in_bytes, get_ultimate_context): Declare.
293 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
295 * config/arm/arm.c (cmse_clear_registers): New function.
296 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
297 code by call to cmse_clear_registers.
298 (cmse_nonsecure_entry_clear_before_return): Likewise.
300 2017-11-22 Tamar Christina <tamar.christina@arm.com>
302 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
303 (vdot_s32, vdotq_s32): New.
304 (vdot_lane_u32, vdotq_lane_u32): New.
305 (vdot_lane_s32, vdotq_lane_s32): New.
308 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
311 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
312 for unsigned values with fewer HWIs than the precision.
313 (test_overflow): New function.
314 (wide_int_cc_tests): Call it.
316 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
318 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
319 has the same mode class as Pmode.
321 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
323 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
324 padding_bits_to_clear_ptr.
325 (cmse_nonsecure_entry_clear_before_return): Likewise.
327 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
329 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
330 auto_sbitap instead of integer bitfield to control register needing
333 2017-11-22 Jakub Jelinek <jakub@redhat.com>
335 PR tree-optimization/83044
336 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
337 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
338 that need the upper bound. Subtract offset from
339 get_addr_base_and_unit_offset only if positive and subtract it
340 before division by eltsize rather than after it.
343 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
344 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
345 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
349 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
350 expand_binop with *mul_widen_optab, make sure at least one of the
351 operands doesn't have VOIDmode.
354 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
356 PR rtl-optimization/82044
357 PR tree-optimization/82042
358 * dse.c (record_store): Check for overflow.
359 (check_mem_read_rtx): Properly check for overflow if width == -1, call
360 add_wild_read instead of clear_rhs_from_active_local_stores on
361 overflow and log it into dump_file.
363 2017-11-22 Richard Biener <rguenther@suse.de>
365 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
366 fake edges to exit when looking for a place to insert.
367 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
369 (insert_into_preds_of_block): Commit edge insertion immediately,
370 assert that doesn't require new BBs.
371 (fini_pre): Release expressions.
372 (pass_pre::execute): Shuffle things around a bit, if the fn
373 is too large do not compute AVAIL either as this is really the
376 2017-11-22 Richard Biener <rguenther@suse.de>
378 PR tree-optimization/83089
379 * tree-if-conv.c (pass_if_conversion::execute): If anything
380 changed reset SCEV and free the number of iteration estimates.
382 2017-11-21 Martin Sebor <msebor@redhat.com>
384 PR tree-optimization/82945
385 * calls.h (warn_nonstring_bound): Remove unused function.
387 2017-11-21 Martin Sebor <msebor@redhat.com>
389 PR tree-optimization/82945
390 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
391 * calls.h (maybe_warn_nonstring_arg): Declare new function.
392 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
394 (initialize_argument_information): Call maybe_warn_nonstring_arg.
395 * calls.h (get_attr_nonstring_decl): Declare new function.
396 * doc/extend.texi (attribute nonstring): Update.
397 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
398 get_attr_nonstring_decl and handle it.
399 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
400 detection of nul-termination.
401 (strlen_to_stridx): Change to a pointer.
402 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
403 (pass_strlen::execute): Same.
405 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
407 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
408 for the new option -mprefer-vector-width=[none|128|256|512].
409 * config/i386/i386.c (ix86_target_string): remove old style options
410 -mprefer-avx256 and make -mprefer-avx128 as alias.
411 (ix86_option_override_internal): Apply defaults for the
412 -mprefer-vector-width=[128|256] option.
413 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
414 Implement macros to work with -mprefer-vector-width=.
415 * config/i386/i386.opt: Implemented option
416 -mprefer-vector-width=[none|128|256|512].
417 * doc/invoke.texi: Documentation for
418 -mprefer-vector-width=[none|128|256|512].
420 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
422 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
424 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
426 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
428 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
430 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
432 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
433 * expr.c (expand_expr_real_2): Likewise.
434 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
435 fold_binary_loc): Likewise.
436 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
437 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
438 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
439 MINUS_EXPR transformations.
440 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
441 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
442 * tree-inline.c (estimate_operator_cost): Likewise.
443 * tree-pretty-print.c (dump_generic_node, op_code_prio,
444 op_symbol_code): Likewise.
445 * tree-vect-stmts.c (vectorizable_operation): Likewise.
446 * vr-values.c (extract_range_from_binary_expr): Likewise.
447 * varasm.c (initializer_constant_valid_p_1): Likewise.
448 * tree.def: New tree code POINTER_DIFF_EXPR.
450 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
452 * config/i386/i386.md (*bswap<mode>2_movbe): Add
453 integer suffix to movbe mnemonic.
454 (*bswaphi2_movbe): Ditto.
455 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
457 2017-11-21 Jakub Jelinek <jakub@redhat.com>
460 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
461 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
462 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
464 2017-11-21 Martin Liska <mliska@suse.cz>
466 * tree-inline.c (expand_call_inline): Remove not needed
469 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
470 Jakub Jelinek <jakub@redhat.com>
473 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
474 Remove static keyword from f variable.
476 2017-11-21 Jakub Jelinek <jakub@redhat.com>
478 PR tree-optimization/83086
479 * gimple-ssa-store-merging.c
480 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
481 rather than n.base_addr.
483 2017-11-21 Martin Liska <mliska@suse.cz>
485 PR rtl-optimization/82044
486 PR tree-optimization/82042
487 * dse.c (check_mem_read_rtx): Check for overflow.
489 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
491 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
494 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
496 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
497 and its corresponding call arg location note.
499 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
500 Andrew Burgess <andrew.burgess@embecosm.com>
502 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
504 (arc_return_slot_offset): Likewise.
505 (arc_eh_return_address_location): New declaration.
506 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
507 (MUST_SAVE_REGISTER): Add exception handler case.
508 (MUST_SAVE_RETURN_ADDR): Likewise.
509 (arc_frame_pointer_required): Likewise.
510 (arc_frame_pointer_needed): New function.
511 (arc_compute_frame_size): Changed.
512 (arc_expand_prologue): Likewise.
513 (arc_expand_epilogue): Likewise.
514 (arc_initial_elimination_offset): Likewise.
515 (arc_return_slot_offset): Delete.
516 (arc_eh_return_address_location): New function.
517 (arc_builtin_setjmp_frame_value): Likewise.
518 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
519 (EH_RETURN_STACKADJ_RTX): Define.
520 (EH_RETURN_HANDLER_RTX): Likewise.
521 * config/arc/arc.md (eh_return): Delete.
523 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
525 * print-tree.h (debug_vec_tree): Remove prototype.
526 * gdbinit.in (pvt): Remove macro.
528 2017-11-21 Jakub Jelinek <jakub@redhat.com>
530 PR tree-optimization/83047
531 * gimple-ssa-store-merging.c
532 (imm_store_chain_info::output_merged_store): If the loads with the
533 same vuse are in different basic blocks, for load_gsi pick a load
534 location that is dominated by the other loads.
537 * config/i386/i386.c (ix86_memmodel_check): Start
538 -Winvalid-memory-model diagnostics with lowercase letter.
541 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
542 set in_cold_section_p to the partition containing loc_list->first.
543 When seeing loc_list->last_before_switch node, update secname and
544 perform range_across_switch second partition handling only after that.
547 * run-rtl-passes.c: Include debug.h.
548 (run_rtl_passes): Call debug_hooks->assembly_start.
549 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
553 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
554 OPTAB_DIRECT in calls to expand_simple_binop.
556 2017-11-20 David Malcolm <dmalcolm@redhat.com>
559 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
561 2017-11-20 David Malcolm <dmalcolm@redhat.com>
564 * spellcheck.h (best_match::blithely_get_best_candidate): New
567 2017-11-20 Jakub Jelinek <jakub@redhat.com>
569 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
570 with lower case letter.
572 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
574 * config/i386/i386.md (bswaphi2): New expander.
575 (*bswaphi2_movbe): New insn pattern.
576 (bswaphi -> rorhi pepehole2): New peephole pattern.
578 2017-11-20 Jeff Law <law@redhat.com>
580 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
581 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
582 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
583 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
584 gimple-ssa-evrp-analyze.h.
586 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
588 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
590 2017-11-20 Bin Cheng <bin.cheng@arm.com>
592 * tree-predcom.c: Add general comment on Store-Store chains.
593 (split_data_refs_to_components): Postpone clearing eliminate_store_p
595 (get_chain_last_ref_at): Rename into...
596 (get_chain_last_write_at): ...this.
597 (get_chain_last_write_before_load): New function.
598 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
599 CT_STORE_STORE when write reference is added.
600 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
601 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
602 (initialize_root_vars_store_elim_1): Ditto.
603 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
604 definition is created.
605 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
606 chain by replacing it with dominant stored value.
608 2017-11-20 Bin Cheng <bin.cheng@arm.com>
610 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
612 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
614 * vr-values.c (extract_range_from_binary_expr): Use a full range
617 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
619 * config/arm/arm.md (R4_REGNUM): Define constant.
620 (nonsecure_call_internal): Remove r4 clobber.
621 (nonsecure_call_value_internal): Likewise.
622 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
623 clobber and resequence match_operands.
624 (nonsecure_call_value_reg_thumb1_v5): Likewise.
625 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
626 (nonsecure_call_value_reg_thumb2): Likewise.
628 2017-11-20 Jakub Jelinek <jakub@redhat.com>
630 PR tree-optimization/78821
631 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
632 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
633 to the address of the base rather than the base itself.
634 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
635 (find_bswap_or_nop_finalize): New function.
636 (find_bswap_or_nop): Use it.
637 (bswap_replace): Return a tree rather than bool, change first
638 argument from gimple * to gimple_stmt_iterator, allow inserting
639 into an empty sequence, allow ins_stmt to be NULL - then emit
640 all stmts into gsi. Fix up MEM_REF address gimplification.
641 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
642 (struct store_immediate_info): Add N and INS_STMT non-static
644 (store_immediate_info::store_immediate_info): Initialize them
645 from newly added ctor args.
646 (merged_store_group::apply_stores): Formatting fixes. Sort by
648 (stmts_may_clobber_ref_p): For stores call also
649 refs_anti_dependent_p.
650 (gather_bswap_load_refs): New function.
651 (imm_store_chain_info::try_coalesce_bswap): New method.
652 (imm_store_chain_info::coalesce_immediate_stores): Use it.
653 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
654 (imm_store_chain_info::output_merged_store): Fail if number of
655 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
656 and NOP_EXPR rhs_code.
657 (pass_store_merging::process_store): Compute n and ins_stmt, if
658 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
659 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
661 (pass_store_merging::execute): Calculate dominators.
663 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
664 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
665 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
666 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
667 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
668 class pass_optimize_bswap, bswap_replace,
669 pass_optimize_bswap::execute): Moved to ...
670 * gimple-ssa-store-merging.c: ... this file.
671 Include optabs-tree.h.
672 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
673 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
674 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
675 anonymous namespace, remove static keywords.
676 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
677 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
679 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
682 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
684 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
686 * vec.h (debug_helper): New function.
687 (DEFINE_DEBUG_VEC): New macro.
688 * hash-set.h (debug_helper): New function.
689 (DEFINE_DEBUG_HASH_SET): New macro.
690 * cfg.c (debug_slim (edge)): New function.
691 Call DEFINE_DEBUG_VEC for edges.
692 Call DEFINE_DEBUG_HASH_SET for edges.
693 * cfghooks.c (debug_slim (basic_block)): New function.
694 Call DEFINE_DEBUG_VEC for basic blocks.
695 Call DEFINE_DEBUG_HASH_SET for basic blocks.
696 * print-tree.c (debug_slim): New function to handle trees.
697 Call DEFINE_DEBUG_VEC for trees.
698 Call DEFINE_DEBUG_HASH_SET for trees.
699 (debug (vec<tree, va_gc>) &): Remove.
700 (debug (<vec<tree, va_gc>) *): Remove.
701 * print-rtl.c (debug_slim): New function to handle const_rtx.
702 Call DEFINE_DEBUG_VEC for rtx_def.
703 Call DEFINE_DEBUG_VEC for rtx_insn.
704 Call DEFINE_DEBUG_HASH_SET for rtx_def.
705 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
706 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
707 (debug (vec<rtx_insn *> *ptr): Remove.
708 (debug_insn_vector): Remove.
709 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
711 2017-11-20 Tom de Vries <tom@codesourcery.com>
713 PR rtl-optimization/82020
714 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
715 IF_THEN_ELSE condition.
717 2017-11-19 Jeff Law <law@redhat.com>
719 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
720 of degenerates resulting from ignoring an edge.
722 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
725 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
727 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
730 * profile-count.c (profile_count::to_sreal_scale): Fix return value
731 for uninitialied counts.
733 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
736 * tree-inline.c (estimate_num_insns): Set to 1 at least.
738 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
741 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
743 2017-11-19 Tom de Vries <tom@codesourcery.com>
745 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
748 2017-11-19 Tom de Vries <tom@codesourcery.com>
750 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
753 2017-11-19 Tom de Vries <tom@codesourcery.com>
755 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
758 2017-11-19 Tom de Vries <tom@codesourcery.com>
760 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
762 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
764 2017-11-19 Tom de Vries <tom@codesourcery.com>
766 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
768 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
769 semicolon after MCORE_EXPORT_NAME call.
771 2017-11-19 Tom de Vries <tom@codesourcery.com>
774 * vmsdbgout.c (vmsdbgout_early_finish): New function.
775 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
777 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
779 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
780 (cgraph_edge::clone): Cleanup updating of profile.
781 * ipa-cp.c (update_profiling_info): Likewise.
782 * ipa-inline-transform.c (inline_transform): Likewise.
783 * ipa-inline.c (inline_small_functions): Add missing space to dump.
784 * ipa-split.c (execute_split_functions): Do not split when function
786 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
787 * profile-count.c (profile_count::dump): Add global0.
788 (profile_count::to_cgraph_frequency): Do not ICE when entry is
790 (profile_count::to_sreal_scale): Likewise.
791 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
792 (profile_count::combine_with_ipa_count): New function.
793 * profile-count.h (profile_guessed_global0adjusted): New.
794 (profile_count::adjusted_zero): New.
795 (profile_count::global0adjusted): New.
796 (profile_count::combine_with_ipa_count): New.
797 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
798 correct profile of return block of split functions.
799 (copy_cfg_body): Remove unused profile_count.
800 (copy_body): Likewise.
801 (expand_call_inline): Update.
802 (tree_function_versioning): Update.
804 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
806 * hash-set.h (hash_set::empty): New.
807 * tree-ssa-threadbackward.h: Delete.
808 * tree-ssa-threadbackward.c (class thread_jumps): New.
809 Move max_threaded_paths into class.
810 (fsm_find_thread_path): Remove arguments that are now in class.
811 (profitable_jump_thread_path): Rename to...
812 (thread_jumps::profitable_jump_thread_path): ...this.
813 (convert_and_register_jump_thread_path): Rename to...
814 (thread_jumps::convert_and_register_current_path): ...this.
815 (check_subpath_and_update_thread_path): Rename to...
816 (thread_jumps::check_subpath_and_update_thread_path): ...this.
817 (register_jump_thread_path_if_profitable): Rename to...
818 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
819 (handle_phi): Rename to...
820 (thread_jumps::handle_phi): ...this.
821 (handle_assignment): Rename to...
822 (thread_jumps::handle_assignment): ...this.
823 (fsm_find_control_statement_thread_paths): Rename to...
824 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
825 (find_jump_threads_backwards): Rename to...
826 (thread_jumps::find_jump_threads_backwards): ...this.
827 Initialize path local data.
828 (pass_thread_jumps::execute): Call find_jump_threads_backwards
829 from within thread_jumps class.
830 (pass_early_thread_jumps::execute): Same.
832 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
834 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
836 2017-11-17 Steve Ellcey <sellcey@cavium.com>
838 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
839 to canonical location.
841 2017-11-17 Steve Ellcey <sellcey@cavium.com>
844 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
846 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
848 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
850 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
851 rcrt1.o%s/grcrt1.o%s for -static-pie.
853 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
855 * i386.c (ix86_multiplication_cost, ix86_division_cost,
856 ix86_shift_rotate_cost): Break out from ...
857 (ix86_rtx_costs): ... here.
858 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
861 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
863 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
864 when entry block was promoted unlikely.
865 (estimate_bb_frequencies): Increase frequency scale.
866 * profile-count.h (profile_count): Export precision info.
868 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
870 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
871 disturbin profile of entry block.
873 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
875 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
878 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
880 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
881 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
883 2017-11-17 Jeff Law <law@redhat.com>
885 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
886 from evrp_dom_walker class. Various methods moved into new class.
887 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
888 (evrp_range_analyzer::enter): New method.
889 (evrp_range_analyzer::leave): New method.
890 (evrp_dom_walker): Remove delegators no longer needed by this class.
891 Replace vr_values data member with evrp_range_analyzer
893 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
894 method extracted from evrp_dom_walker::before_dom_children.
895 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
896 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
898 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
899 Add private copy constructor and move assignment operators.
900 Privatize methods and class data where trivially possible.
901 (evrp_dom_walker::cleanup): New function, extracted from
902 execute_early_vrp. Simplify access to class data.
904 * vr-values.h (get_output_for_vrp): Prototype.
905 * vr-values.c (get_output_for_vrp): New function extracted from
906 vrp_visit_assignment_or_call and extract_range_from_stmt.
907 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
909 2017-11-17 Luis Machado <luis.machado@linaro.org>
911 * config/aarch64/aarch64.c
912 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
913 (qdf24xx_tunings) <autoprefetcher_model>: Set to
914 tune_params::AUTOPREFETCHER_WEAK.
916 2017-11-17 Tamar Christina <tamar.christina@arm.com>
919 * config/arm/arm.c (arm_valid_target_attribute_rec):
920 Parse "arch=" and "+<ext>".
921 (arm_valid_target_attribute_tree): Re-init global options.
922 (arm_option_override): Make non-static.
923 (arm_options_perform_arch_sanity_checks): Make errors fatal.
924 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
925 (__ARM_FEATURE_CRC32): Support undef.
926 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
927 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
929 2017-11-17 David Malcolm <dmalcolm@redhat.com>
931 * gdbinit.in (break-on-diagnostic): New command.
933 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
935 * config/i386/i386.c (ix86_expand_epilogue): Change simple
936 return to indirect jump for EH return if control-flow
937 protection is enabled. Change explicit 'false' argument in
938 pro_epilogue_adjust_stack with a value of
940 * config/i386/i386.md (simple_return_indirect_internal):
941 Remove SImode restriction to support 64-bit.
943 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
945 * combine.c (added_notes_insn): New.
946 (try_combine): Handle added_notes_insn like added_links_insn.
947 Rewrite return value code.
948 (distribute_notes): Set added_notes_insn to the earliest insn we added
951 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
953 PR rtl-optimization/82621
954 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
955 dest of one of those SETs is unused.
957 2017-11-17 Richard Biener <rguenther@suse.de>
960 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
961 * tree-pretty-print.c (dump_generic_node): Handle
962 annot_expr_parallel_kind.
963 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
964 * gimplify.c (gimple_boolify): Likewise.
966 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
968 * config.gcc (extra_headers): Add cet.h for x86 targets.
969 * config/i386/cet.h: New file.
970 * doc/install.texi: Add --enable-cet/--disable-cet.
972 2017-11-17 Richard Biener <rguenther@suse.de>
974 PR tree-optimization/83017
975 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
976 (gen_parallel_loop): Properly count iterations.
977 (parallelize_loops): Handle loop->can_be_parallel independent
978 of flag_loop_parallelize_all. Make static profitability test match
980 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
981 * invoke.texi (parloops-min-per-thread): Document.
983 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
985 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
986 upstreaming review comments.
988 2017-11-17 Tamar Christina <tamar.christina@arm.com>
990 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
991 with fast unaligned access.
992 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
994 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
996 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
997 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
999 2017-11-17 Richard Biener <rguenther@suse.de>
1001 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1002 folding of references.
1004 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1007 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1008 of replacing call to strncmp with corresponding call to strcmp when
1011 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1013 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1014 option prefer-avx256 for skylake-avx512 configuration.
1015 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1016 (get_builtin_code_for_version): Ditto.
1018 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1019 Monk Chiang <sh.chiang04@gmail.com>
1021 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1022 (FIXED_REGISTERS): Reserve more register numbers.
1023 (CALL_USED_REGISTERS): Likewise.
1024 (REG_ALLOC_ORDER): Likewise.
1025 (REG_CLASS_CONTENTS): Likewise.
1026 (REGISTER_NAMES): Likewise.
1028 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1029 Kito Cheng <kito.cheng@gmail.com>
1031 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1033 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1035 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1037 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1038 parameter list for vec_splats.
1040 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1042 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1044 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1046 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1049 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1050 with the required version of automake 1.11.6.
1052 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1054 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1056 (DU_C2_3_power9): Correct reservation combinations.
1057 (FP_DIV_power9, VEC_DIV_power9): New.
1058 (power9-alu): Split out rotate/shift...
1059 (power9-rot): ...to here, correct dispatch resource.
1060 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1062 (power9-fp): Correct latency.
1063 (power9-sdiv): Add div/sqrt resource.
1064 (power9-ddiv): Correct latency, add div/sqrt resource.
1065 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1066 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1068 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1070 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1072 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1073 switch statement mapping KF built-ins to TF built-ins if we don't
1074 have the proper ISA 3.0 assembler support.
1076 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1078 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1079 (lower_emutls_function_body): Do not compute it.
1081 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1083 * ipa-split.c (split_bb_info): Turn time to sreal.
1084 (split_point): Likewise.
1085 (dump_split_point): Likewise.
1086 (fine_split_points): Likewise.
1087 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1089 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1091 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1094 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1096 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1098 (expensive_function_p): Remove useless assert.
1099 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1101 2017-11-16 Martin Sebor <msebor@redhat.com>
1103 PR tree-optimization/82588
1104 PR tree-optimization/82583
1105 * tree-vrp.c (check_array_ref): Handle flexible array members,
1106 string literals, and inner indices.
1107 (search_for_addr_array): Add detail to diagnostics.
1109 2017-11-16 Nathan Sidwell <nathan@acm.org>
1113 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1114 (SET_DECL_ASSEMBLER_NAME): Forward to
1115 overwrite_decl_assembler_name.
1116 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1117 (overwrite_decl_assembler_name): Declare.
1118 * tree.c (overwrite_decl_assembler_name): New.
1119 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1120 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1121 (LANG_HOOKS_INITIALIZER): Add it.
1122 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1123 * langhooks.c (lhd_set_decl_assembler_name): Use
1124 SET_DECL_ASSEMBLER_NAME.
1125 (lhd_overwrite_decl_assembler_name): Default implementation.
1127 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1128 Jackson Woodruff <jackson.woodruff@arm.com>
1130 PR tree-optimization/71026
1131 * match.pd: Canonicalize constant multiplies in division.
1133 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1135 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
1136 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
1137 Remove OPT_fomit_frame_pointer entry.
1138 * common/config/arc/arc-common.c: Likewise.
1139 * common/config/arm/arm-common.c: Likewise.
1140 * common/config/avr/avr-common.c: Likewise.
1141 * common/config/c6x/c6x-common.c: Likewise.
1142 * common/config/cr16/cr16-common.c: Likewise.
1143 * common/config/cris/cris-common.c: Likewise.
1144 * common/config/epiphany/epiphany-common.c: Likewise.
1145 * common/config/fr30/fr30-common.c: Likewise.
1146 * common/config/frv/frv-common.c: Likewise.
1147 * common/config/ia64/ia64-common.c: Likewise.
1148 * common/config/iq2000/iq2000-common.c: Likewise.
1149 * common/config/lm32/lm32-common.c: Likewise.
1150 * common/config/m32r/m32r-common.c: Likewise.
1151 * common/config/mcore/mcore-common.c: Likewise.
1152 * common/config/microblaze/microblaze-common.c: Likewise.
1153 * common/config/mips/mips-common.c: Likewise.
1154 * common/config/mmix/mmix-common.c: Likewise.
1155 * common/config/mn10300/mn10300-common.c: Likewise.
1156 * common/config/nios2/nios2-common.c: Likewise.
1157 * common/config/pa/pa-common.c: Likewise.
1158 * common/config/pdp11/pdp11-common.c: Likewise.
1159 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
1160 * common/config/riscv/riscv-common.c: Likewise.
1161 * common/config/rs6000/rs6000-common.c: Likewise.
1162 * common/config/rx/rx-common.c: Likewise.
1163 * common/config/s390/s390-common.c: Likewise.
1164 * common/config/sh/sh-common.c: Likewise.
1165 * common/config/sparc/sparc-common.c: Likewise.
1166 * common/config/tilegx/tilegx-common.c: Likewise.
1167 * common/config/tilepro/tilepro-common.c: Likewise.
1168 * common/config/v850/v850-common.c: Likewise.
1169 * common/config/visium/visium-common.c: Likewise.
1170 * common/config/xstormy16/xstormy16-common.c: Likewise.
1171 * common/config/xtensa/xtensa-common.c: Likewise.
1172 * invoke.texi (-fomit-frame-pointer): Update documentation.
1174 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1176 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
1178 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1180 * cfg.c (scale_bbs_frequencies_int,
1181 cale_bbs_frequencies_gcov_type): Remove.
1182 * cfg.h (scale_bbs_frequencies_int,
1183 cale_bbs_frequencies_gcov_type): Remove.
1185 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1187 * tree-ssa-loop-manip.c
1188 (scale_dominated_blocks_in_loop): Update to profile counts.
1189 (tree_transform_and_unroll_loop): Likewise.
1191 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1193 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
1194 scale_bbs_frequencies_int.
1196 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1198 * final.c (compute_alignments): Use counts rather than frequencies.
1200 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1202 * cfgloopanal.c: Include sreal.h
1203 (average_num_loop_insns): Use counts and sreal for accounting.
1205 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1207 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
1210 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1211 Kito Cheng <kito.cheng@gmail.com>
1213 * config/nds32/constraints.md: Provide more constraints.
1214 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
1215 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
1216 support constraints usage.
1218 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1220 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
1222 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1223 Kito Cheng <kito.cheng@gmail.com>
1225 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
1226 * config/nds32/nds32.opt: Refine the layout.
1227 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
1228 TARGET_EXT_STRING): Support new options.
1229 * config/nds32/nds32.h: Likewise.
1230 * config/nds32/nds32.md: Likewise.
1231 * config/nds32/nds32-predicates.c: Likewise.
1232 * config/nds32/constraints.md: Likewise.
1233 * common/config/nds32/nds32-common.c: Likewise.
1235 2017-11-16 Julia Koval <julia.koval@intel.com>
1238 * config/i386/gfniintrin.h: Add sse check.
1239 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1241 2017-11-16 Julia Koval <julia.koval@intel.com>
1243 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1244 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1245 (ix86_handle_option): Handle -mavx512vbmi2.
1246 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1247 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1248 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1249 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1250 (ix86_valid_target_attribute_inner_p): Ditto.
1251 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1252 * config/i386/i386.opt (mavx512vbmi2): New option.
1253 * doc/invoke.texi: Add new option.
1255 [2017-11-16 Julia Koval <julia.koval@intel.com>
1257 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1258 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1259 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1260 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1261 _mm512_gf2p8mul_epi8): New intrinsics.
1262 * config/i386/i386-builtin-types.def
1263 (V64QI_FTYPE_V64QI_V64QI): New type.
1264 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1265 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1266 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1267 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1268 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1269 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1271 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1273 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1274 explicitly as a stream of bytes.
1276 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1278 * config/rs6000/altivec.h (vec_xst_be): New #define.
1279 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1280 and externalize from *altivec_vperm_<mode>_internal.
1281 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1283 (XL_BE_V8HI): Likewise.
1284 (XL_BE_V4SI): Likewise.
1285 (XL_BE_V4SI): Likewise.
1286 (XL_BE_V2DI): Likewise.
1287 (XL_BE_V4SF): Likewise.
1288 (XL_BE_V2DF): Likewise.
1289 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1290 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1291 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1292 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1293 VSX_BUILTIN_VEC_XST_BE.
1294 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1295 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1297 (altivec_init_builtins): Replace conditional calls to def_builtin
1298 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1299 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1300 with unconditional calls. Remove calls to def_builtin for
1301 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1302 __builtin_vec_xst_be.
1303 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1304 to define_expand, and add alternate RTL generation for P8.
1305 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1306 vsx_ld_elemrev_v8hi.
1307 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1308 add alternate RTL generation for P8.
1309 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1310 vsx_ld_elemrev_v16qi.
1311 (vsx_st_elemrev_v8hi): Convert define_insn
1312 to define_expand, and add alternate RTL generation for P8.
1313 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1314 vsx_st_elemrev_v8hi.
1315 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1316 add alternate RTL generation for P8.
1317 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1318 vsx_st_elemrev_v16qi.
1320 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1323 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1324 TARGET_AVX512ER check.
1325 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1326 neither -mzeroupper nor -mno-zeroupper is used and
1327 TARGET_EMIT_VZEROUPPER is set.
1328 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1329 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1331 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1333 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1334 folding of vector compares.
1335 (fold_build_vec_cmp): New helper function.
1336 (fold_compare_helper): New helper function.
1337 (builtin_function_type): Add compare builtins to the list of functions
1338 having unsigned arguments. Cosmetic updates to comment indentation.
1339 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1340 the not+eq combination.
1342 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1344 PR tree-optimization/82726
1345 PR tree-optimization/70754
1346 * tree-predcom.c (order_drefs_by_pos): New function.
1347 (combine_chains): Move code setting has_max_use_after to...
1348 (try_combine_chains): ...here. New parameter. Sort combined chains
1349 according to position information.
1350 (tree_predictive_commoning_loop): Update call to above function.
1351 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1353 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1355 PR tree-optimization/82726
1357 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1359 PR tree-optimization/70754
1360 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1361 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1362 combined stmt before it if not NULL.
1363 (combine_chains): Process refs reversely and compute dominance point
1367 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1369 PR tree-optimization/79663
1370 * tree-predcom.c (combine_chains): Process refs in reverse order
1371 only for ZERO length chains, and add explaining comment.
1373 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1375 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1376 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1377 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1379 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1381 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1383 2017-11-15 Martin Liska <mliska@suse.cz>
1385 * tree-cfg.c (pass_warn_function_return::execute):
1386 Compare warn_return_type for greater than zero.
1388 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1392 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1393 to return true on Xeon and not on Xeon Phi.
1394 (ix86_check_avx256_register): Changed to ...
1395 (ix86_check_avx_upper_register): ... this. Add extra check for
1396 VALID_AVX512F_REG_OR_XI_MODE.
1397 (ix86_avx_u128_mode_needed): Changed
1398 ix86_check_avx256_register to ix86_check_avx_upper_register.
1399 (ix86_check_avx256_stores): Changed to ...
1400 (ix86_check_avx_upper_stores): ... this. Changed
1401 ix86_check_avx256_register to ix86_check_avx_upper_register.
1402 (ix86_avx_u128_mode_after): Changed
1403 avx_reg256_found to avx_upper_reg_found. Changed
1404 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1405 (ix86_avx_u128_mode_entry): Changed
1406 ix86_check_avx256_register to ix86_check_avx_upper_register.
1407 (ix86_avx_u128_mode_exit): Ditto.
1408 * config/i386/i386.h: (host_detect_local_cpu): New define.
1410 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1412 * config/arm/xgene1.md (xgene1): Split into automatons
1413 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1414 (xgene1_f_load): Adjust reservations and/or types.
1415 (xgene1_f_store): Likewise.
1416 (xgene1_load_pair): Likewise.
1417 (xgene1_store_pair): Likewise.
1418 (xgene1_fp_load1): Likewise.
1419 (xgene1_load1): Likewise.
1420 (xgene1_store1): Likewise.
1421 (xgene1_move): Likewise.
1422 (xgene1_alu): Likewise.
1423 (xgene1_simd): Likewise.
1424 (xgene1_bfm): Likewise.
1425 (xgene1_neon_load1): Likewise.
1426 (xgene1_neon_store1): Likewise.
1427 (xgene1_neon_logic): Likewise.
1428 (xgene1_neon_st1): Likewise.
1429 (xgene1_neon_ld1r): Likewise.
1430 (xgene1_alu_cond): Added.
1431 (xgene1_shift_reg): Likwise.
1432 (xgene1_bfx): Likewise.
1433 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1435 2017-11-15 Jakub Jelinek <jakub@redhat.com>
1438 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
1440 (can_widen_mult_without_libcall): New function.
1441 (expand_mul_overflow): If only checking unsigned mul overflow,
1442 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
1443 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
1444 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
1445 (expand_DIVMOD): Formatting fix.
1446 * expmed.h (expand_mult): Add NO_LIBCALL argument.
1447 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
1448 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
1450 PR tree-optimization/82977
1451 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
1452 constructed temporary to strlen_to_stridx.put.
1454 2017-11-15 Martin Liska <mliska@suse.cz>
1456 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
1457 * configure: Regenerate.
1459 2017-11-15 Martin Liska <mliska@suse.cz>
1462 * config/sh/sh-mem.cc: Use proper probability for
1465 2017-11-14 Jeff Law <law@redhat.com>
1467 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
1468 the red zone for stack_clash_protection_final_dynamic_probe targets
1469 when the total dynamic stack size is zero bytes.
1471 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
1472 blocks is post order.
1474 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
1476 * dumpfile.h (TDF_COMPARE_DEBUG): New.
1477 * final.c (rest_of_clean_state): Set it for the
1478 -fcompare-debug dump.
1479 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
1480 class when TDF_COMPARE_DEBUG is set.
1482 * dwarf2out.c (gen_producer_string): Discard
1485 2017-11-15 Joseph Myers <joseph@codesourcery.com>
1488 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
1489 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
1490 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
1491 (__TGMATH_REAL_2_3): Remove macros.
1492 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
1493 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
1495 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
1497 (remquo): Define using __TGMATH_REAL_3.
1499 2017-11-14 Jeff Law <law@redhat.com>
1501 * vr-values.c: New file with contents extracted from tree-vrp.c.
1502 * Makefile.in (OBJS): Add vr-values.o
1503 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
1504 (set_value_range, set_and_canonicalize_value_range): Likewise.
1505 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
1506 (value_range_constant_singleton, symbolic_range_p): Likewise.
1507 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
1508 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
1509 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
1510 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
1511 (find_case_label_range, find_case_label_index): Likewise.
1512 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
1513 (range_int_cst_singleton_p, value_inside_range): Likewise.
1514 (get_single_symbol): Likewise.
1515 (switch_update): Move structure definition here.
1516 (to_remove_edges, to_update_switch_stmts): Provide externs.
1517 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
1518 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
1519 (vrp_val_is_min, set_value_range): Likewise.
1520 (set_and_canonicalize_value_range, copy_value_range): Likewise.
1521 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
1522 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
1523 (range_is_nonnull, range_int_cst_p): Likewwise.
1524 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
1525 (get_single_symbol, operand_less_p): Likewise
1526 (compare_values_warnv, compare_values): Likewise.
1527 (value_inside_range, value_range_constant_singleton): Likewise.
1528 (zero_nonzero_bitgs_from_vr): Likewise.
1529 (extract_range_from_binary_expr_1): Likewise.
1530 (overflow_comparison_p): Likewise.
1531 (to_remove_edges, to_update_switch_stmts): Likewise.
1532 (find_case_label-index, find_case_label_range): Likewise.
1533 (switch_update, set_value_range_to_nonnegative): Remove.
1534 (set_value_range_to_truthvalue): Likewise.
1535 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
1536 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
1537 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
1538 (find_case_label_ranges, test_for_singularity): Likewise.
1539 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
1540 (x_vr_values): Move to its remaining use site.
1542 2017-11-10 Jeff Law <law@redhat.com>
1544 * vr-values.h (VR_INITIALIZER): Move #define here.
1545 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
1546 * Makefile.in (OBJS): Add tree-evrp.o
1547 * tree-vrp.h (assert_info): Move structure definition here.
1548 (set_value_range_to_varying): Prototype.
1549 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1550 (infer_value_range, register_edge_assert_for): Likewise.
1551 (stmt_interesting_for_vrp): Likewise.
1552 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
1553 (set_value_range_to_varying): No longer static.
1554 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1555 (infer_value_range, register_edge_assert_for): Likewise.
1557 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1559 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
1560 generate the XXBRD instruction.
1562 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
1563 (rs6000_builtin_type_compatible): Treat _Float128 and long double
1564 as being compatible if -mabi=ieeelongdouble.
1565 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
1566 to setup float128 built-ins with hardware support.
1567 (BU_FLOAT128_HW_2): Likewise.
1568 (BU_FLOAT128_HW_3): Likewise.
1569 (BU_FLOAT128_HW_VSX_1): Likewise.
1570 (BU_FLOAT128_HW_VSX_2): Likewise.
1571 (scalar_extract_expq): Change float128 built-in functions to
1572 accommodate having both KFmode and TFmode functions. Use the
1573 KFmode variant as the default.
1574 (scalar_extract_sigq): Likewise.
1575 (scalar_test_neg_qp): Likewise.
1576 (scalar_insert_exp_q): Likewise.
1577 (scalar_insert_exp_qp): Likewise.
1578 (scalar_test_data_class_qp): Likewise.
1579 (sqrtf128_round_to_odd): Delete processing the round to odd
1580 built-in functions as special built-in functions, and define them
1581 as float128 built-ins. Use the KFmode variant as the default.
1582 (truncf128_round_to_odd): Likewise.
1583 (addf128_round_to_odd): Likewise.
1584 (subf128_round_to_odd): Likewise.
1585 (mulf128_round_to_odd): Likewise.
1586 (divf128_round_to_odd): Likewise.
1587 (fmaf128_round_to_odd): Likewise.
1588 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1589 support for KFmode and TFmode xststdcqp calls.
1590 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1591 point, switch the built-in handlers for the get/set float128
1592 exponent, get float128 mantissa, float128 test built-ins, and the
1593 float128 round to odd built-in functions. Eliminate creating the
1594 float128 round to odd built-in functions as special built-ins.
1595 (rs6000_init_builtins): Eliminate special creation of the float128
1596 round to odd built-in functions.
1597 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1598 function insns to support both TFmode and KFmode variants.
1599 (xsxsigqp_<mode>): Likewise.
1600 (xsiexpqpf_<mode>): Likewise.
1601 (xsiexpqp_<mode>): Likewise.
1602 (xststdcqp_<mode>): Likewise.
1603 (xststdcnegqp_<mode>): Likewise.
1604 (xststdcqp_<mode>): Likewise.
1606 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1608 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1609 unused path_in_freq_ptr parameter.
1610 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1612 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1614 * ipa-inline.c (edge_badness): Dump sreal frequency.
1615 (compute_inlined_call_time): Match natural implementaiton ...
1616 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1617 forgotten division by CGRAPH_FREQ_BASE.
1619 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1621 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1622 Solaris 11. Update comment.
1623 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1625 * config/sol2.h (STARTFILE_SPEC): Likewise.
1626 * configure: Regenerate.
1628 2017-11-14 Carl Love <cel@us.ibm.com>
1630 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1631 le_ and be_ prefixes to swap* variables. Remove
1632 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1635 2017-11-14 Jason Merrill <jason@redhat.com>
1637 Support GTY((cache)) on hash_map.
1638 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1639 (ggc_cache_remove): Override it instead of ggc_mx.
1640 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1641 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1642 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1643 (simple_cache_map_traits): Override maybe_mx.
1644 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1645 (hash_map): Friend gt_cleare_cache.
1646 (gt_cleare_cache): New.
1647 * tree.h (tree_cache_traits): New hash_map traits class.
1648 (tree_cache_map): New typedef.
1650 2017-11-14 Richard Biener <rguenther@suse.de>
1652 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1653 paramter and handling.
1654 (cleanup_control_flow_bb): Likewise.
1655 (cleanup_control_flow_pre): New helper performing a DFS walk
1656 to call cleanup_control_flow_bb in PRE order.
1657 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1658 via cleanup_control_flow_pre.
1660 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1662 * config/aarch64/aarch64-simd.md
1663 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1664 (*aarch64_simd_bsl<mode>_alt): Likewise.
1665 (aarch64_simd_bsldi_internal): New.
1666 (aarch64_simd_bsldi_alt): Likewise.
1668 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1670 * tracer.c (better_p): Do not compare frequencies.
1671 * reg-stack.c (better_edge): Likewise.
1672 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1675 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1677 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1678 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1679 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1680 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1681 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1682 * predict.c (maybe_hot_frequency_p): Inline to ...
1683 (maybe_hot_count_p): ... here; rewrite to counts.
1684 (counts_to_freqs): Rename to ...
1685 (update_max_bb_count): ... this one.
1686 (expensive_function_p): Use counts.
1687 (estimate_bb_frequencies): Update.
1688 (rebuild_frequencies): Update.
1689 * predict.h (counts_to_freqs): Rename to ...
1690 (update_max_bb_count): ... this one.
1691 * profile.c (compute_branch_probabilities): Add debug info
1692 * tree-inline.c (expand_call_inline): Update debug info.
1693 (optimize_inline_calls): Use update_max_bb_count..
1694 (tree_function_versioning): Use update_max_bb_count..
1695 * value-prof.c (gimple_value_profile_transformations):
1696 Do not use update_max_bb_count.
1698 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1700 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1701 always use frequencies.
1703 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1705 * bb-reorder.c: Remove frequencies from comments.
1706 (better_edge_p): Use profile counts.
1707 (find_traces): Dump profile counts.
1708 (rotate_loop): Use profile counts.
1709 (find_traces_1_round): Likewise.
1710 (connect_better_edge_p): Use counts instead of probabilities for
1712 (copy_bb_p): Drop early check for non-0 frequency.
1713 (sanitize_hot_paths): Update comments.
1715 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1717 * ipa-split.c (struct split_point): Add count.
1718 (consider_split): Do not compute incoming frequency; compute incoming
1719 count and store it to split_point.
1720 (split_function): Set count of the call to split part correctly.
1722 2017-11-13 Carl Love <cel@us.ibm.com>
1724 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1726 2017-11-13 Tom Tromey <tom@tromey.com>
1728 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1730 2017-11-13 Carl Love <cel@us.ibm.com>
1732 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1733 Add support for builtins:
1734 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1735 vector {un,}signed {char,int,short},
1736 vector {un,}signed {char,int,short}) arguments.
1737 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1738 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1739 Add BU_P9V_AV_2 expansions for the builtins.
1740 * config/rs6000/altivec.h (vec_first_match_index,
1741 vec_first_mismatch_index, vec_first_match_or_eos_index,
1742 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1743 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1744 new extern declaration.
1745 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1746 * config/rs6000/vsx.md (first_match_index_<mode>,
1747 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1748 first_mismatch_or_eos_index_<mode>): Add define expand.
1749 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1750 * doc/extend.texi: Update the built-in documenation file for the new
1753 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1755 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1756 into the min/max operations for _Float<N> and _Float<N>X types.
1758 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1761 * dwarf2out.c (do_eh_frame): New static variable.
1762 (dwarf2out_begin_prologue): Set it.
1763 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1765 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1767 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1769 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1771 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1774 * debug.h (dwarf2out_do_eh_frame): Declare.
1775 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1776 (dwarf2out_do_frame): Use it.
1777 (dwarf2out_do_cfi_asm): Likewise.
1778 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1779 (dwarf2out_assembly_start): Likewise.
1780 (dwarf2out_begin_prologue): Fix comment.
1781 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1782 if the target needs either debug or unwind DWARF2 info.
1783 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1784 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1785 -ftrapping-math, -ftrapv and -fwrapv.
1787 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1789 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1790 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1791 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1792 (estimate_edge_size_and_time): Likewise.
1793 (ipa_merge_fn_summary_after_inlining): Likewise.
1794 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1795 (compute_uninlined_call_time): Use sreal_frequency.
1796 (compute_inlined_call_time): Likewise.
1797 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1798 * profile-count.c: Include sreal.h.
1799 (profile_count::to_sreal_scale): New.
1800 * profile-count.h: Forward declare sreal.
1801 (profile_count::to_sreal_scale): Declare.
1803 2017-11-13 Nathan Sidwell <nathan@acm.org>
1805 * diagnostic.c (maybe_line_and_column): New.
1806 (diagnostic_get_location_text): Use it.
1807 (diagnostic_report_current_module): Likewise.
1808 (test_diagnostic_get_location_text): Add tests.
1810 2017-11-13 Luis Machado <luis.machado@linaro.org>
1812 * doc/md.texi (Specifying processor pipeline description): Fix
1813 incorrect latency for the div instruction example.
1815 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1817 PR tree-optimization/78821
1818 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1819 that bit_not_p is the same.
1820 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1821 (split_group): Count precisely bit_not_p bits in each statement.
1822 (invert_op): New function.
1823 (imm_store_chain_info::output_merged_store): Use invert_op to
1824 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1825 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1827 2017-11-13 Martin Liska <mliska@suse.cz>
1829 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1830 (struct source_info): Likewise.
1831 (add_branch_counts): Likewise.
1832 (add_line_counts): Likewise.
1833 (function_summary): Likewise.
1834 (output_intermediate_line): Likewise.
1835 (generate_results): Likewise.
1837 2017-11-13 Martin Liska <mliska@suse.cz>
1839 * gcov.c (struct block_info): Remove typedef for block_t.
1840 (struct line_info): Likewise.
1841 (line_info::has_block): Likewise.
1842 (EXIT_BLOCK): Likewise.
1843 (unblock): Likewise.
1844 (circuit): Likewise.
1845 (get_cycles_count): Likewise.
1846 (process_file): Likewise.
1847 (read_graph_file): Likewise.
1848 (solve_flow_graph): Likewise.
1849 (find_exception_blocks): Likewise.
1850 (add_line_counts): Likewise.
1851 (accumulate_line_info): Likewise.
1852 (output_line_details): Likewise.
1854 2017-11-13 Martin Liska <mliska@suse.cz>
1856 * gcov.c (struct arc_info): Remove typedef for arc_t.
1857 (struct line_info): Likewise.
1858 (add_branch_counts): Likewise.
1859 (output_branch_count): Likewise.
1860 (function_info::~function_info): Likewise.
1861 (circuit): Likewise.
1862 (output_intermediate_line): Likewise.
1863 (read_graph_file): Likewise.
1864 (solve_flow_graph): Likewise.
1865 (find_exception_blocks): Likewise.
1866 (add_line_counts): Likewise.
1867 (accumulate_line_info): Likewise.
1868 (output_line_details): Likewise.
1869 (output_function_details): Likewise.
1871 2017-11-13 Martin Liska <mliska@suse.cz>
1873 * gcov.c (struct function_info): Remove typedef for function_t.
1874 (struct source_info): Likewise.
1875 (source_info::get_functions_at_location): Likewise.
1876 (solve_flow_graph): Likewise.
1877 (find_exception_blocks): Likewise.
1878 (add_line_counts): Likewise.
1879 (output_intermediate_file): Likewise.
1880 (process_file): Likewise.
1881 (generate_results): Likewise.
1882 (release_structures): Likewise.
1883 (read_graph_file): Likewise.
1884 (read_count_file): Likewise.
1885 (accumulate_line_counts): Likewise.
1886 (output_lines): Likewise.
1888 2017-11-13 Martin Liska <mliska@suse.cz>
1890 * gcov.c (function_info::function_info): Remove num_counts
1891 and add vector<gcov_type>.
1892 (function_info::~function_info): Use the vector.
1893 (process_file): Likewise.
1894 (read_graph_file): Likewise.
1895 (read_count_file): Likewise.
1896 (solve_flow_graph): Likewise.
1898 2017-11-13 Martin Liska <mliska@suse.cz>
1900 * gcov.c (function_info::is_artificial): New function.
1901 (process_file): Erase all artificial early.
1902 (generate_results): Skip as all artificial are already
1905 2017-11-13 Martin Liska <mliska@suse.cz>
1907 * gcov.c (read_graph_file): Store to global vector of functions.
1908 (read_count_file): Iterate the vector.
1909 (process_file): Likewise.
1910 (generate_results): Likewise.
1911 (release_structures): Likewise.
1913 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1915 PR tree-optimization/82954
1916 * gimple-ssa-store-merging.c
1917 (imm_store_chain_info::coalesce_immediate_stores): If
1918 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1920 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1922 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1923 Upddate call to ENDIAN_LANE_N.
1924 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1925 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1926 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1927 and use aarch64_endian_lane_rtx.
1928 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1930 2017-11-12 Tom de Vries <tom@codesourcery.com>
1932 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1934 2017-11-12 Tom de Vries <tom@codesourcery.com>
1936 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1938 2017-11-12 Tom de Vries <tom@codesourcery.com>
1940 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1942 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1943 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1944 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1946 2017-11-11 Martin Sebor <msebor@redhat.com>
1949 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1952 * prefic.c (translate_name): Replace strncpy with memcpy to
1953 avoid -Wstringop-truncation.
1955 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1957 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1959 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1961 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1963 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1965 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1968 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1970 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1971 enable generating XXBRH if the value is in a vector register.
1972 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1973 value is in a vector register.
1974 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1975 register to register bswap64's instead of doing the GPR sequence
1976 used on previous machines.
1977 (bswapdi2_xxbrd): New insn.
1978 (bswapdi2_reg): Disallow on ISA 3.0.
1979 (register to register bswap64 splitter): Do not split the insn on
1980 ISA 3.0 systems that use XXBRD.
1982 2017-11-10 Martin Sebor <msebor@redhat.com>
1985 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1987 (find_subframework_file): Same.
1989 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1991 * auto-profile.c (afdo_indirect_call): Drop frequency.
1992 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1993 (cgraph_node::create_edge): Drop frequency argument.
1994 (cgraph_node::create_indirect_edge): Drop frequency argument.
1995 (cgraph_edge::make_speculative): Drop frequency arguments.
1996 (cgraph_edge::resolve_speculation): Do not update frequencies
1997 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1998 (cgraph_node::dump): Check consistency in IPA mode.
1999 (cgraph_edge::maybe_hot_p): Use IPA counter.
2000 (cgraph_edge::verify_count_and_frequency): Rename to ...
2001 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2002 (cgraph_node::verify_node): Update.
2003 * cgraph.h (struct cgraph_edge): Drop frequency.
2004 (cgraph_edge::frequency): New function.
2005 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2007 (cgraph_edge::rebuild_edges): Likewise.
2008 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2009 (duplicate_thunk_for_node): Do not pass frequency.
2010 (cgraph_node::create_clone): Scale only counts.
2011 (cgraph_node::create_virtual_clone): Do not pass frequency.
2012 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2013 (cgraph_node::create_version_clone): Do not pass frequency.
2014 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2015 (cgraph_node::expand_thunk): Do not pass frequency.
2016 (cgraph_node::create_wrapper): Do not pass frequency.
2017 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2019 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2020 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2021 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2022 (gather_caller_stats): Use frequency function.
2023 (ipcp_cloning_candidate_p): Use frequency function.
2024 (ipcp_propagate_stage): Use frequency function.
2025 (get_info_about_necessary_edges): Use frequency function.
2026 (update_profiling_info): Update only IPA profile.
2027 (update_specialized_profile): Use frequency functoin.
2028 (perhaps_add_new_callers): Update only IPA profile.
2029 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2030 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2031 (dump_ipa_call_summary): Use frequency function.
2032 (estimate_edge_size_and_time): Use frequency function.
2033 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2034 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2035 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2036 (update_noncloned_counts): ... ths one; scale counts only.
2037 (clone_inlined_nodes): Do not scale frequency.
2038 (inline_call): Do not pass frequency.
2039 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2040 (compute_inlined_call_time): Use IPA profile.
2041 (want_inline_small_function_p): Use IPA profile.
2042 (want_inline_self_recursive_call_p): Use IPA profile.
2043 (edge_badness): Use IPA profile.
2044 (lookup_recursive_calls): Use IPA profile.
2045 (recursive_inlining): Do not pass frequency.
2046 (resolve_noninline_speculation): Do not update frequency.
2047 (inline_small_functions): Collect max of IPA profile.
2048 (dump_overall_stats): Dump IPA porfile.
2049 (dump_inline_stats): Dump IPA porfile.
2050 (ipa_inline): Collect IPA stats.
2051 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2052 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2053 (ipa_propagate_frequency): Use frequency function.
2054 (ipa_profile): Cleanup.
2055 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2056 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2057 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2058 (input_node): Do not stream frequency.
2059 (input_edge): Do not stream frequency.
2060 (merge_profile_summaries): Scale only IPA profiles.
2061 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2062 * predict.c (drop_profile): Do not recompute frequency.
2063 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2064 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2065 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2066 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2067 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2068 * tree-inline.c (copy_bb): Do not scale frequency.
2069 (expand_call_inline): Do not scale frequency.
2070 (tree_function_versioning): Do not scale frequency.
2071 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2073 2017-11-10 Julia Koval <julia.koval@intel.com>
2075 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2076 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2077 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2078 (_mm256_maskz_gf2p8affine_epi64_epi8)
2079 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2080 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2081 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2082 (__builtin_ia32_vgf2p8affineqb_v32qi)
2083 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2084 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2086 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2090 (arm_option_override): Refactor.
2091 (arm_option_reconfigure_globals): New.
2092 (arm_options_perform_arch_sanity_checks): New.
2093 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2095 (arm_options_perform_arch_sanity_checks): Likewise
2097 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2099 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2100 (power9-qpmul): New.
2101 * rs6000/rs6000.md ("type" attr): Add qmul.
2102 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2103 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2104 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2106 2017-11-10 Martin Sebor <msebor@redhat.com>
2109 * builtins.c (compute_objsize): Handle arrays that
2110 compute_builtin_object_size likes to fail for. Make extern.
2111 * builtins.h (compute_objsize): Declare.
2112 (check_strncpy_sizes): New function.
2113 (expand_builtin_strncpy): Call check_strncpy_sizes.
2114 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2115 -Wstringop-truncation.
2116 (gimple_fold_builtin_strncat): Same.
2117 * gimple.c (gimple_build_call_from_tree): Set call location.
2118 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2119 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2120 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2121 (handle_builtin_strlen): Use strlen_to_stridx.
2122 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2124 Use strlen_to_stridx.
2125 (pass_strlen::execute): Release strlen_to_stridx.
2126 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2127 (-Wstringop-truncation): Document new option.
2129 2017-11-10 Martin Liska <mliska@suse.cz>
2131 PR gcov-profile/82702
2132 * gcov.c (main): Handle intermediate files in a different
2134 (get_gcov_intermediate_filename): New function.
2135 (output_gcov_file): Remove support of intermediate files.
2136 (generate_results): Allocate intermediate file.
2137 (release_structures): Clean-up properly fn_end.
2138 (output_intermediate_file): Start iterating with line 1.
2140 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2142 PR tree-optimization/82929
2143 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2144 ops_swapped_p non-static data member.
2145 (store_immediate_info::store_immediate_info): Clear it.
2146 (imm_store_chain_info::coalesce_immediate_stores): If swapping
2147 ops set ops_swapped_p.
2148 (count_multiple_uses): Handle ops_swapped_p.
2150 2017-11-10 Martin Liska <mliska@suse.cz>
2152 * coverage.c (coverage_init): Stream information about
2153 support of has_unexecuted_blocks.
2154 * doc/gcov.texi: Document that.
2155 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
2156 * gcov.c (read_graph_file): Likewise.
2157 (output_line_beginning): Fix a small issue with
2160 2017-11-10 Bin Cheng <bin.cheng@arm.com>
2162 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
2163 reference of trivial component.
2165 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2168 * gimple-ssa-store-merging.c
2169 (pass_store_merging::terminate_all_aliasing_chains): For
2170 gimple_store_p stmts also call refs_output_dependent_p.
2172 PR rtl-optimization/82913
2173 * compare-elim.c (try_merge_compare): Punt if def_insn is not
2176 2017-11-09 Jeff Law <law@redhat.com>
2178 * vr-values.h: New file with vr_values class.
2179 * tree-vrp.c: Include vr-values.h
2180 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
2181 data objects into the vr_values class.
2182 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
2183 (get_value_range): Make it a member function within vr_values class.
2184 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
2185 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
2186 (op_with_constant_singleton_value_range): Likewise.
2187 (extract_range_for_var_from_comparison_expr): Likewise.
2188 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
2189 (extract_range_from_binary_expr): Likewise.
2190 (extract_range_from_unary_expr): Likewise.
2191 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
2192 (check_for_binary_op_overflow, extract_range_basic): Likewise.
2193 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
2194 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
2195 (compare_name_with_value, compare_names): Likewise.
2196 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
2197 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
2198 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
2199 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
2200 (extract_range_from_phi_node): Likewise.
2201 (simplify_truth_ops_using_ranges): Likewise.
2202 (simplify_div_or_mod_using_ranges): Likewise.
2203 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
2204 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
2205 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
2206 (simplify_float_conversion_using_ranges): Likewise.
2207 (simplify_internal_call_using_ranges): Likewise.
2208 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
2209 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
2210 poitner via x_vr_values for calls into gimple folder.
2211 (vrp_initialize_lattice): Make this the vr_values ctor.
2212 (vrp_free_lattice): Make this the vr_values dtor.
2213 (set_vr_value): New function.
2214 (class vrp_prop): Add vr_values data member. Add various member
2215 functions as well as member functions that delegate to vr_values.
2216 (check_array_ref): Make a member function within vrp_prop class.
2217 (search_for_addr_array, vrp_initialize): Likewise.
2218 (vrp_finalize): Likewise. Revamp to avoid direct access to
2219 vr_value, values_propagated, etc.
2220 (check_array_bounds): Extract vrp_prop class instance pointer from
2221 walk info structure. Use it to call member functions.
2222 (check_all_array_refs): Make a member function within vrp_prop class.
2223 Smuggle class instance pointer via walk info structure.
2224 (x_vr_values): New local static.
2225 (vrp_valueize): Use x_vr_values to get class instance.
2226 (vr_valueize_1): Likewise.
2227 (class vrp_folder): Add vr_values data member. Add various member
2228 functions as well as member functions that delegate to vr_values.
2229 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
2230 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
2231 class instance from vr_values. Use it to call member functions.
2232 (vrp_dom_walker): Add vr_values data member.
2233 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2234 instance via x_vr_values.
2235 (identify_jump_threads): Accept vr_values as argument. Store
2236 it into the walker structure.
2237 (evrp_dom_walker): Add vr_values class data member. Add various
2239 (evrp_dom_walker::try_find_new_range): Use vr_values data
2240 member to access the memory allocator.
2241 (evrp_dom_walker::before_dom_children): Store vr_values class
2242 instance into the vrp_folder class.
2243 (evrp_dom_walker::push_value_range): Rework to avoid direct
2244 access to num_vr_values and vr_value.
2245 (evrp_dom_walker::pop_value_range): Likewise.
2246 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2247 Use vr_values to get to dump_all_value_ranges member function.
2248 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2249 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2250 Pass vr_values class instance down to identify_jump_threads.
2251 Remove call to vrp_free_lattice.
2252 (debug_all_value_ranges): Remove.
2254 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2255 (vrp_folder): Likewise.
2257 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2258 Get it from the existing bitmap instead.
2259 (vrp_intersect_ranges_1): Likewise.
2261 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2263 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2265 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2266 set bit_not_p to it.
2267 (imm_store_chain_info::coalesce_immediate_stores): Break group
2268 if bit_not_p is different.
2269 (count_multiple_uses, split_group,
2270 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2271 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2272 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2273 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2274 to store_immediate_info ctor.
2276 2017-11-09 Jim Wilson <jimw@sifive.com>
2278 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2279 (scan_prog_file): Likewise.
2281 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2283 * bb-reorder.c (max_entry_frequency): Remove.
2284 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2285 connect_traces, push_to_next_round_p): Remove prototypes.
2286 (find_traces_1_round): Use counts only.
2287 (push_to_next_round_p): Likewise.
2288 (find_traces): Likewise.
2289 (rotate_loop): Likewise.
2290 (find_traces_1_round): Likewise.
2291 (connect_traces): Likewise.
2292 (edge_order): Likewise.
2294 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2296 * config/arm/arm.c (output_return_instruction): Add comments to
2297 indicate requirement for cmse_nonsecure_entry return to account
2298 for the size of clearing instruction output here.
2299 (thumb_exit): Likewise.
2300 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2301 return in hardfloat mode.
2303 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2305 * config/rs6000/rs6000.c (machine_function): Add a bool,
2306 "toc_is_wrapped_separately".
2307 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2308 if it wasn't explicitly set or unset, we are optimizing for speed, and
2309 doing separate shrink-wrapping.
2310 (rs6000_get_separate_components): Enable the TOC component if
2311 saving the TOC register in the prologue.
2312 (rs6000_components_for_bb): Handle the TOC component.
2313 (rs6000_emit_prologue_components): Store the TOC register where needed.
2314 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2315 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2317 2017-11-09 Martin Jambor <mjambor@suse.cz>
2319 * ipa-param-manipulation.c: New file.
2320 * ipa-param-manipulation.h: Likewise.
2321 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2322 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2323 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2324 (ipa_parm_adjustment): Likewise.
2325 (ipa_parm_adjustment_vec): Likewise.
2326 (ipa_get_vector_of_formal_parms): Moved declaration to
2327 ipa-param-manipulation.h.
2328 (ipa_get_vector_of_formal_parm_types): Likewise.
2329 (ipa_modify_formal_parameters): Likewise.
2330 (ipa_modify_call_arguments): Likewise.
2331 (ipa_combine_adjustments): Likewise.
2332 (ipa_dump_param_adjustments): Likewise.
2333 (ipa_modify_expr): Likewise.
2334 (ipa_get_adjustment_candidate): Likewise.
2335 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2336 ipa-param-manipulation.c.
2337 (ipa_get_vector_of_formal_parm_types): Likewise.
2338 (ipa_modify_formal_parameters): Likewise.
2339 (ipa_modify_call_arguments): Likewise.
2340 (ipa_modify_expr): Likewise.
2341 (get_ssa_base_param): Likewise.
2342 (ipa_get_adjustment_candidate): Likewise.
2343 (index_in_adjustments_multiple_times_p): Likewise.
2344 (ipa_combine_adjustments): Likewise.
2345 (ipa_dump_param_adjustments): Likewise.
2346 * tree-sra.c: Also include ipa-param-manipulation.h
2347 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2350 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2351 Alan Hayward <alan.hayward@arm.com>
2352 David Sherwood <david.sherwood@arm.com>
2354 * doc/sourcebuild.texi (vect_masked_store): Document.
2356 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2357 Alan Hayward <alan.hayward@arm.com>
2358 David Sherwood <david.sherwood@arm.com>
2360 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2362 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2363 Alan Hayward <alan.hayward@arm.com>
2364 David Sherwood <david.sherwood@arm.com>
2366 * doc/sourcebuild.texi (vect_variable_length): Document.
2368 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2369 Alan Hayward <alan.hayward@arm.com>
2370 David Sherwood <david.sherwood@arm.com>
2372 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2374 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2375 Alan Hayward <alan.hayward@arm.com>
2376 David Sherwood <david.sherwood@arm.com>
2378 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2380 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2381 Alan Hayward <alan.hayward@arm.com>
2382 David Sherwood <david.sherwood@arm.com>
2384 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2385 previously undocumented selectors.
2386 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2388 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2390 * doc/rtl.texi (const_vector): Say that elements can be
2391 const_wide_ints too.
2392 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2393 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2394 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2395 * optabs.c (expand_vector_broadcast): Likewise.
2397 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2398 Alan Hayward <alan.hayward@arm.com>
2399 David Sherwood <david.sherwood@arm.com>
2401 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2402 scaled index even if the unscaled address was invalid.
2403 Don't increase the complexity of using a scale in that case.
2405 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2406 Alan Hayward <alan.hayward@arm.com>
2407 David Sherwood <david.sherwood@arm.com>
2409 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2410 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2411 UNITS_PER_WORD bytes.
2412 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2413 based on REGMODE_NATURAL_SIZE of the inner mode.
2414 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2415 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2416 * expmed.c (lowpart_bit_field_p): Divide the value up into
2417 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2418 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2419 whether something is likely to occupy more than one register.
2421 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2424 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2425 profile_count::adjust_for_ipa_scaling.
2426 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2427 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2429 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2431 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2433 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2434 (split_group): Add total_orig and total_new arguments, estimate the
2435 number of statements related to the store group without store merging
2436 and with store merging.
2437 (imm_store_chain_info::output_merged_store): Adjust split_group
2438 callers, punt if estimated number of statements with store merging
2439 is not smaller than estimated number of statements without it.
2441 (handled_load): Remove has_single_use checks.
2442 (pass_store_merging::process_store): Likewise.
2444 2017-11-09 Richard Biener <rguenther@suse.de>
2446 PR tree-optimization/82902
2447 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
2449 2017-11-09 Martin Liska <mliska@suse.cz>
2452 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
2455 2017-11-09 Martin Liska <mliska@suse.cz>
2457 PR tree-optimization/82669
2458 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
2460 2017-11-09 Martin Liska <mliska@suse.cz>
2462 PR gcov-profile/48463
2463 * coverage.c (coverage_begin_function): Output also end locus
2464 of a function and information whether the function is
2466 * gcov-dump.c (tag_function): Parse and print the information.
2467 * gcov.c (INCLUDE_MAP): Add include.
2468 (INCLUDE_SET): Likewise.
2469 (struct line_info): Move earlier in the source file because
2470 of vector<line_info> in function_info structure.
2471 (line_info::line_info): Likewise.
2472 (line_info::has_block): Likewise.
2473 (struct source_info): Add new member index.
2474 (source_info::get_functions_at_location): New function.
2475 (function_info::group_line_p): New function.
2476 (output_intermediate_line): New function.
2477 (output_intermediate_file): Use the mentioned function.
2478 (struct function_start): New.
2479 (struct function_start_pair_hash): Likewise.
2480 (process_file): Add code that identifies group functions.
2481 Assign lines either to global or function scope.
2482 (generate_results): Skip artificial functions.
2483 (find_source): Assign index for each source file.
2484 (read_graph_file): Read new flag artificial and end_line.
2485 (add_line_counts): Assign it either to global of function scope.
2486 (accumulate_line_counts): Isolate core of the function to
2487 accumulate_line_info and call it for both function and global
2489 (accumulate_line_info): New function.
2490 (output_line_beginning): Fix GNU coding style.
2491 (print_source_line): New function.
2492 (output_line_details): Likewise.
2493 (output_function_details): Likewise.
2494 (output_lines): Iterate both source (global) scope and function
2496 (struct function_line_start_cmp): New class.
2497 * doc/gcov.texi: Reflect changes in documentation.
2499 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2502 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
2503 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
2504 and similarly for not instead of neg.
2506 2017-11-08 Andi Kleen <ak@linux.intel.com>
2508 * config/i386/i386.opt: Add -mforce-indirect-call.
2509 * config/i386/predicates.md: Check for flag_force_indirect_call.
2510 * doc/invoke.texi: Document -mforce-indirect-call
2512 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
2514 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
2516 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
2517 config/riscv/riscv.c (predict.h): New include.
2518 (riscv_slow_unaligned_access_p): No longer static.
2519 (riscv_block_move_straight): Add require.
2520 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2522 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2525 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
2526 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
2527 nonimmediate_operand predicate for operand 1 instead of
2530 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2532 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
2534 * config/aarch64/constraints.md (Uml): New constraint.
2535 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
2538 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2540 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
2541 of two vec_duplicates into a vec_concat.
2543 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2545 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2546 Simplify vec_merge of vec_duplicate and vec_concat.
2547 * config/aarch64/constraints.md (Utq): New constraint.
2548 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
2551 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2553 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2554 Simplify vec_merge of vec_duplicate and const_vector.
2555 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
2557 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
2558 mode iterator. Update predicate on operand 1 to
2559 handle non-const_vec constants. Delete constraints.
2560 (*aarch64_combinez_be<mode>): Likewise for operand 2.
2562 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2564 PR tree-optimization/78821
2565 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
2567 (store_operand_info::store_operand_info): Initialize it to false.
2568 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
2569 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
2570 store in the group, and if chain_info is non-NULL, to ignore altogether
2572 (compatible_load_p): Fail if bit_not_p does not match.
2573 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2574 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2575 (pass_store_merging::process_store): Adjust
2576 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2577 call terminate_all_aliasing_chains newly when adding a store into
2578 a chain with non-NULL chain_info.
2580 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2582 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2584 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2586 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2588 (aarch64_layout_frame): Initialise emit_frame_chain.
2589 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2590 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2592 2017-11-08 Martin Liska <mliska@suse.cz>
2594 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2595 of gimple_call_internal_p.
2597 2017-11-07 Tom de Vries <tom@codesourcery.com>
2599 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2601 2017-11-07 Tom de Vries <tom@codesourcery.com>
2603 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2606 2017-11-08 Martin Liska <mliska@suse.cz>
2609 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2611 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2613 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2615 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2618 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2619 (*cmp<mode>_ccz_1): New insn with $k alternative.
2622 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2623 RTX_COMM_COMPARE as commutative as well.
2624 (ix86_binary_operator_ok): Formatting fix.
2625 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2626 *<code><mode>3<mask_name><round_saeonly_name>,
2627 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2628 *<s>mul<mode>3_highpart<mask_name>,
2629 *vec_widen_umult_even_v16si<mask_name>,
2630 *vec_widen_umult_even_v8si<mask_name>,
2631 *vec_widen_umult_even_v4si<mask_name>,
2632 *vec_widen_smult_even_v16si<mask_name>,
2633 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2634 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2635 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2636 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2637 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2638 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2639 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2640 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2641 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2642 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2643 ix86_binary_operator_ok. Formatting fixes.
2644 (*<plusminus_insn><mode>3<mask_name><round_name>,
2645 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2648 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2650 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2651 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2652 eq and ne if TARGET_ISEL.
2653 (cmp): New code_iterator.
2654 (UNS, UNSU_, UNSIK): New code_attrs.
2655 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2656 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2658 ("eq<mode>3"): ... this.
2659 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2661 ("ne<mode>3"): ... this.
2663 2017-11-07 Julia Koval <julia.koval@intel.com>
2666 * common/config/i386/i386-common.c
2667 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2668 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2669 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2670 * config/i386/i386.opt: Ditto.
2671 * config/i386/i386.c (ix86_target_string): Ditto.
2672 (ix86_option_override_internal): Ditto.
2673 (ix86_init_mpx_builtins): Move MPX to args2.
2674 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2675 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2676 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2677 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2678 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2679 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2680 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2682 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2685 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2686 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2687 (zero-extendsidi peephole2): Remove peephole.
2689 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2692 * stor-layout.c: Include attribs.h.
2693 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2694 explicit lookup of "aligned" attribute.
2696 2017-11-07 Andrew Waterman <andrew@sifive.com>
2698 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2699 (riscv_expand_block_move): Likewise.
2700 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2702 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2703 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2704 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2705 (riscv_adjust_block_mem): Likewise.
2706 (riscv_block_move_loop): Likewise.
2707 (riscv_expand_block_move): Likewise.
2708 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2710 2017-11-07 Michael Clark <michaeljclark@mac.com>
2712 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2713 (MUSL_DYNAMIC_LINKER): Likewise.
2715 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2717 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2718 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2721 2017-11-07 Richard Biener <rguenther@suse.de>
2723 * match.pd: Fix build.
2725 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2726 Jackson Woodruff <jackson.woodruff@arm.com>
2728 PR tree-optimization/71026
2729 * match.pd: Canonicalize negate in division.
2731 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2734 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2736 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2738 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2739 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2741 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2743 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2744 non-scalar integral types.
2745 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2746 (-(A-B), -(~A)): New transformations.
2748 2017-11-07 Tom de Vries <tom@codesourcery.com>
2750 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2751 semicolon after "do {} while (0)".
2752 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2753 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2754 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2755 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2756 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2758 2017-11-07 Tom de Vries <tom@codesourcery.com>
2760 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2761 after "do {} while (0)".
2762 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2763 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2764 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2765 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2766 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2768 2017-11-07 Tom de Vries <tom@codesourcery.com>
2771 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2773 (arm_rtx_costs_internal): Add missing semicolon after
2774 HANDLE_NARROW_SHIFT_ARITH call.
2776 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2778 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2779 disable isel if it was not set explicitly.
2781 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2783 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
2784 (add_type_attribute) likewise.
2786 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2788 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2789 of struct ix86_frame.
2790 (ix86_initial_elimination_offset): Likewise.
2791 (ix86_expand_split_stack_prologue): Likewise.
2793 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2795 * tree-vrp.h (enum value_range_type): Update stale comment.
2797 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2798 Alan Hayward <alan.hayward@arm.com>
2799 David Sherwood <david.sherwood@arm.com>
2801 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2802 (aarch64_expand_vec_perm_const): Take the number of units too.
2803 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2804 (aarch64_expand_vec_perm_const): Likewise.
2805 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2806 (vec_perm<mode>): Update accordingly.
2808 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2809 Alan Hayward <alan.hayward@arm.com>
2810 David Sherwood <david.sherwood@arm.com>
2812 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2813 Take the number of units too.
2814 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2815 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2816 but check for a vector mode before rather than after the call.
2817 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2818 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2819 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2820 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2821 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2822 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2823 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2824 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2825 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2826 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2827 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2828 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2829 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2830 (aarch64_sqdmull2_n<mode>): Update accordingly.
2832 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2833 Alan Hayward <alan.hayward@arm.com>
2834 David Sherwood <david.sherwood@arm.com>
2836 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2837 the number of units too.
2838 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2839 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2840 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2841 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2842 (vec_store_lanesxi<mode>): Update accordingly.
2844 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2845 Alan Hayward <alan.hayward@arm.com>
2846 David Sherwood <david.sherwood@arm.com>
2848 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2849 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2850 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2851 of units rather than the mode.
2852 * config/aarch64/iterators.md (nunits): New mode attribute.
2853 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2854 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2855 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2856 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2857 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2858 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2859 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2860 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2861 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2862 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2863 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2864 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2865 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2866 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2867 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2868 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2869 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2870 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2871 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2872 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2873 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2874 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2875 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2876 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2877 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2878 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2879 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2880 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2881 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2882 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2883 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2884 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2885 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2886 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2887 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2888 (aarch64_simd_vec_setv2di): Likewise.
2890 2017-11-06 Carl Love <cel@us.ibm.com>
2892 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2894 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2895 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2897 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2899 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2900 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2901 Add power 8 macro expansions.
2902 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2903 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2904 power 8 instructions. (VSX_XXBR): Add iterator.
2906 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2908 * config/arm/arm.md (predicable_short_it): Change default to "no",
2909 improve documentation, remove uses that are identical to the default.
2910 (enabled_for_depr_it): Rename to enabled_for_short_it.
2911 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2912 Remove default uses.
2913 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2914 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2915 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2916 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2918 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2921 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2922 float128 helper macros, which are no longer used after deleting
2923 the old 'q' built-in functions, and moving the round to odd
2924 built-in functions to being special built-in functions.
2925 (BU_FLOAT128_2): Likewise.
2926 (BU_FLOAT128_1_HW): Likewise.
2927 (BU_FLOAT128_2_HW): Likewise.
2928 (BU_FLOAT128_3_HW): Likewise.
2929 (FABSQ): Delete old 'q' built-in functions.
2930 (COPYSIGNQ): Likewise.
2931 (SQRTF128_ODD): Move round to odd built-in functions to be
2932 special built-in functions, so that we can handle
2933 -mabi=ieeelongdouble.
2934 (TRUNCF128_ODD): Likewise.
2935 (ADDF128_ODD): Likewise.
2936 (SUBF128_ODD): Likewise.
2937 (MULF128_ODD): Likewise.
2938 (DIVF128_ODD): Likewise.
2939 (FMAF128_ODD): Likewise.
2940 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2941 built-in names to 'f128'.
2942 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2943 old 'q' built-in functions, as the machine independent code for
2944 'f128' built-in functions handles this.
2945 (rs6000_expand_builtin): Add expansion for float128 round to odd
2946 functions, keying off on -mabi=ieeelongdouble of whether to use
2947 the KFmode or TFmode variant.
2948 (rs6000_init_builtins): Initialize the _Float128 round to odd
2950 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2951 _Float128 'q' built-in functions are now mapped into the new
2952 'f128' built-in functions.
2954 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2956 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2957 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2959 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2962 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2963 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2964 * toplev.c: Include "ipa-fnsummary.h".
2965 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2967 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2969 PR tree-optimization/82838
2970 * gimple-ssa-store-merging.c
2971 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2972 on a separate gimple_seq which is then appended to seq.
2974 2017-11-06 Jeff Law <law@redhat.com>
2977 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2978 (get_probe_interval): New functions.
2979 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2980 (ix86_adjust_stack_and_probe): Likewise.
2981 (output_adjust_stack_and_probe): Likewise.
2982 (ix86_emit_probe_stack_range): Likewise.
2983 (ix86_expand_prologue): Likewise.
2985 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2987 PR tree-optimization/82816
2988 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2989 if the modes of the two types are the same.
2990 (convert_plusminus_to_widen): Likewise.
2992 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2994 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2996 (usadv16qi): New define_expand.
2997 (usadv8hi): New define_expand.
2999 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3002 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3004 (inline_transform): Likewise.
3005 * predict.c (counts_to_freqs): Remove useless conditional.
3006 * profile-count.h (profile_count::apply_scale): Move sanity check.
3007 * tree-inline.c (copy_bb): Always scale.
3008 (copy_cfg_body): Likewise.
3010 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3013 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3015 (*cmp_ite0): Add enabled_for_depr_it attribute.
3016 (*cmp_ite1): Likewise.
3018 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3020 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3023 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3025 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3026 return the value by pointer.
3027 (extract_range_from_multiplicative_op_1): Update accordingly.
3028 Return as soon as an operation fails.
3030 2017-11-05 Tom de Vries <tom@codesourcery.com>
3033 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3034 (DEF_SANITIZER_BUILTIN): ... here.
3035 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3036 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3038 2017-11-05 Tom de Vries <tom@codesourcery.com>
3041 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3043 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3044 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3045 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3047 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3048 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3049 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3051 2017-11-05 Tom de Vries <tom@codesourcery.com>
3054 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3057 2017-11-04 Michael Clark <michaeljclark@mac.com>
3059 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3060 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3062 (*addsi3_extended): Likewise.
3063 (*addsi3_extended2): Likewise.
3064 (<optab>si3): Likewise.
3065 (<optab>di3): Likewise.
3066 (<optab><mode>3): Likewise.
3067 (<*optabe>si3_internal): Likewise.
3068 (zero_extendqi<SUPERQI:mode>2): Likewise.
3069 (*add<mode>hi3): Likewise.
3070 (*xor<mode>hi3): Likewise.
3071 (<optab>di3): Likewise.
3072 (*<optab>si3_extend): Likewise.
3073 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3074 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3075 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3077 2017-11-04 Andrew Waterman <andrew@sifive.com>
3079 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3080 TARGET_STRICT_ALIGN based upon -mtune argument.
3082 2017-11-04 Andrew Waterman <andrew@sifive.com>
3084 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3086 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3088 config/i386/i386.c (choose_basereg): Use optional scratch
3089 register and add assertion.
3090 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3091 needed, and don't allocate stack.
3092 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3093 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3094 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3096 2017-11-03 Jeff Law <law@redhat.com>
3098 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3099 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3100 to probe at the start of a noreturn function.
3102 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3104 PR tree-optimization/78821
3105 * gimple-ssa-store-merging.c: Update the file comment.
3106 (MAX_STORE_ALIAS_CHECKS): Define.
3107 (struct store_operand_info): New type.
3108 (store_operand_info::store_operand_info): New constructor.
3109 (struct store_immediate_info): Add rhs_code and ops data members.
3110 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3111 and op1r arguments to the ctor, initialize corresponding data members.
3112 (struct merged_store_group): Add load_align_base and load_align
3114 (merged_store_group::merged_store_group): Initialize them.
3115 (merged_store_group::do_merge): Update them.
3116 (merged_store_group::apply_stores): Pick the constant for
3117 encode_tree_to_bitpos from one of the two operands, or skip
3118 encode_tree_to_bitpos if neither operand is a constant.
3119 (class pass_store_merging): Add process_store method decl. Remove
3120 bool argument from terminate_all_aliasing_chains method decl.
3121 (pass_store_merging::terminate_all_aliasing_chains): Remove
3122 var_offset_p argument and corresponding handling.
3123 (stmts_may_clobber_ref_p): New function.
3124 (compatible_load_p): New function.
3125 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3126 if there is overlap and rhs_code is not INTEGER_CST. For
3127 non-overlapping stores terminate group if rhs is not mergeable.
3128 (get_alias_type_for_stmts): Change first argument from
3129 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3130 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3131 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3133 (get_location_for_stmts): Change first argument from
3134 auto_vec<gimple *> & to vec<gimple *> &.
3135 (struct split_store): Remove orig_stmts data member, add orig_stores.
3136 (split_store::split_store): Create orig_stores rather than orig_stmts.
3137 (find_constituent_stmts): Renamed to ...
3138 (find_constituent_stores): ... this. Change second argument from
3139 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
3140 to info structures rather than the statements.
3141 (split_group): Rename ALLOW_UNALIGNED argument to
3142 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
3143 it. Adjust find_constituent_stores caller.
3144 (imm_store_chain_info::output_merged_store): Handle rhs_code other
3145 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
3146 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
3147 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
3148 (mem_valid_for_store_merging): New function.
3149 (handled_load): New function.
3150 (pass_store_merging::process_store): New method.
3151 (pass_store_merging::execute): Use process_store method. Adjust
3152 terminate_all_aliasing_chains caller.
3154 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3156 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3157 Return true for more constants, symbols and label references.
3158 (aarch64_valid_floating_const): Remove unused function.
3160 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3163 * config/aarch64/aarch64.c (aarch64_layout_frame):
3164 Undo forcing of LR at bottom of frame.
3166 2017-11-03 Jeff Law <law@redhat.com>
3169 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
3170 for int_registers_saved.
3172 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
3173 extracted from tree-ssa-dom.c.
3174 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
3175 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
3176 (record_equivalences_from_incoming_edge): Add additional argument
3177 to single_pred_edge_ignoring_loop_edges call.
3178 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
3179 (uncprop_dom_walker::before_dom_children): Add additional argument
3180 to single_pred_edge_ignoring_loop_edges call.
3181 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
3182 single_pred_edge_ignoring_loop_edges rather than open coding.
3183 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
3185 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
3187 * match.pd (-(-A)): Rewrite.
3189 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3191 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
3192 (rs6000_emit_int_cmove): New declaration.
3193 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
3194 (rs6000_emit_sISEL): Delete.
3195 (rs6000_emit_int_cmove): Make non-static.
3196 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
3197 instead of rs6000_emit_sISEL.
3199 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3201 * asan.c (create_cond_insert_point): Maintain profile.
3202 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
3204 * basic-block.h (struct basic_block_def): Remove frequency.
3205 (EDGE_FREQUENCY): Use to_frequency
3206 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
3208 (find_traces): Update to use to_frequency.
3209 (find_traces_1_round): Likewise; use only IPA counts.
3210 (bb_to_key): Likewise.
3211 (connect_traces): Use IPA counts only.
3212 (copy_bb_p): Update to use to_frequency.
3213 (fix_up_crossing_landing_pad): Likewise.
3214 (sanitize_hot_paths): Likewise.
3215 * bt-load.c (basic_block_freq): Likewise.
3216 * cfg.c (init_flow): Set count_max to uninitialized.
3217 (check_bb_profile): Remove frequencies; check counts.
3218 (dump_bb_info): Do not dump frequencies.
3219 (update_bb_profile_for_threading): Update counts only.
3220 (scale_bbs_frequencies_int): Likewise.
3221 (MAX_SAFE_MULTIPLIER): Remove.
3222 (scale_bbs_frequencies_gcov_type): Update counts only.
3223 (scale_bbs_frequencies_profile_count): Update counts only.
3224 (scale_bbs_frequencies): Update counts only.
3225 * cfg.h (struct control_flow_graph): Add count-max.
3226 (update_bb_profile_for_threading): Update prototype.
3227 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
3228 (find_many_sub_basic_blocks): Likewise.
3229 * cfgcleanup.c (try_forward_edges): Likewise.
3230 (try_crossjump_to_edge): Likewise.
3231 * cfgexpand.c (expand_gimple_cond): Likewise.
3232 (expand_gimple_tailcall): Likewise.
3233 (construct_init_block): Likewise.
3234 (construct_exit_block): Likewise.
3235 * cfghooks.c (verify_flow_info): Check consistency of counts.
3236 (dump_bb_for_graph): Do not dump frequencies.
3237 (split_block_1): Do not update frequencies.
3238 (split_edge): Do not update frequencies.
3239 (make_forwarder_block): Do not update frequencies.
3240 (duplicate_block): Do not update frequencies.
3241 (account_profile_record): Do not update frequencies.
3242 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3243 for global heuristics.
3244 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3245 (expected_loop_iterations_unbounded): Use counts only.
3246 * cfgloopmanip.c (scale_loop_profile): Simplify.
3247 (create_empty_loop_on_edge): Simplify
3249 (duplicate_loop_to_header_edge): Simplify
3250 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3251 (update_br_prob_note): Take care of removing note when profile
3253 (relink_block_chain): Do not dump frequency.
3254 (rtl_account_profile_record): Use to_frequency.
3255 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3256 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3257 (cgraph_update_edges_for_call_stmt_node): Likewise.
3258 (cgraph_edge::verify_count_and_frequency): Update.
3259 (cgraph_node::verify_node): Temporarily disable frequency verification.
3260 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3261 to_cgraph_frequency.
3262 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3263 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3265 (cgraph_node::expand_thunk): Update profile.
3266 * except.c (dw2_build_landing_pads): Do not update frequency.
3267 * final.c (compute_alignments): Use to_frequency.
3268 (dump_basic_block_info): Do not dump frequency.
3269 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3270 (dump_gimple_bb_header): Do not dump frequency.
3271 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3273 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3274 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3275 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3276 (init_before_recovery): Do not update frequency.
3277 (sched_create_recovery_edges): Do not update frequency.
3278 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3279 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3280 (ipa_cp_c_finalize): Set max_count to uninitialized.
3281 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3282 (param_change_prob): Use counts.
3283 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3285 * ipa-split.c (consider_split): Use to_frequency.
3286 (split_function): Use to_frequency.
3287 * ira-build.c (loop_compare_func): Likewise.
3288 (mark_loops_for_removal): Likewise.
3289 (mark_all_loops_for_removal): Likewise.
3290 * loop-doloop.c (doloop_modify): Do not update frequency.
3291 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3293 * lto-streamer-in.c (input_function): Update count_max.
3294 * omp-expand.c (expand_omp_taskreg): Update count_max.
3295 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3296 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3297 (maybe_hot_count_p): Use ipa counts only.
3298 (maybe_hot_bb_p): Simplify.
3299 (maybe_hot_edge_p): Simplify.
3300 (probably_never_executed): Do not take frequency argument.
3301 (probably_never_executed_bb_p): Do not pass frequency.
3302 (probably_never_executed_edge_p): Likewise.
3303 (combine_predictions_for_bb): Check that profile is nonzero.
3304 (propagate_freq): Do not set frequency.
3305 (drop_profile): Simplify.
3306 (counts_to_freqs): Simplify.
3307 (expensive_function_p): Use to_frequency.
3308 (propagate_unlikely_bbs_forward): Simplify.
3309 (determine_unlikely_bbs): Simplify.
3310 (estimate_bb_frequencies): Add hack to silence graphite issues.
3311 (compute_function_frequency): Use ipa counts.
3312 (pass_profile::execute): Update.
3313 (rebuild_frequencies): Use counts only.
3314 (force_edge_cold): Use counts only.
3315 * profile-count.c (profile_count::dump): Dump new count types.
3316 (profile_count::differs_from_p): Check compatiblity.
3317 (profile_count::to_frequency): New function.
3318 (profile_count::to_cgraph_frequency): New function.
3319 * profile-count.h (struct function): Declare.
3320 (enum profile_quality): Add profile_guessed_local and
3321 profile_guessed_global0.
3322 (class profile_proability): Decrease number of bits to 29;
3323 update from_reg_br_prob_note and to_reg_br_prob_note.
3324 (class profile_count: Update comment; decrease number of bits
3325 to 61. Check compatibility.
3326 (profile_count::compatible_p): New private member function.
3327 (profile_count::ipa_p): New member function.
3328 (profile_count::operator<): Handle global zero correctly.
3329 (profile_count::operator>): Handle global zero correctly.
3330 (profile_count::operator<=): Handle global zero correctly.
3331 (profile_count::operator>=): Handle global zero correctly.
3332 (profile_count::nonzero_p): New member function.
3333 (profile_count::force_nonzero): New member function.
3334 (profile_count::max): New member function.
3335 (profile_count::apply_scale): Handle IPA scalling.
3336 (profile_count::guessed_local): New member function.
3337 (profile_count::global0): New member function.
3338 (profile_count::ipa): New member function.
3339 (profile_count::to_frequency): Declare.
3340 (profile_count::to_cgraph_frequency): Declare.
3341 * profile.c (OVERLAP_BASE): Delete.
3342 (compute_frequency_overlap): Delete.
3343 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3344 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3345 * sched-ebb.c (rank): Use counts only.
3346 * shrink-wrap.c (handle_simple_exit): Use counts only.
3347 (try_shrink_wrapping): Use counts only.
3348 (place_prologue_for_one_component): Use counts only.
3349 * tracer.c (find_best_predecessor): Use to_frequency.
3350 (find_trace): Use to_frequency.
3351 (tail_duplicate): Use to_frequency.
3352 * trans-mem.c (expand_transaction): Do not update frequency.
3353 * tree-call-cdce.c: Do not update frequency.
3354 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3355 (gimple_merge_blocks): Likewise.
3356 (gimple_split_edge): Likewise.
3357 (gimple_duplicate_sese_region): Likewise.
3358 (gimple_duplicate_sese_tail): Likewise.
3359 (move_sese_region_to_fn): Likewise.
3360 (gimple_account_profile_record): Likewise.
3361 (insert_cond_bb): Likewise.
3362 * tree-complex.c (expand_complex_div_wide): Likewise.
3363 * tree-eh.c (lower_resx): Update profile.
3364 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3366 (initialize_cfun): Do not initialize frequencies
3367 (freqs_to_counts): Delete.
3368 (copy_cfg_body): Ignore count parameter.
3369 (copy_body): Update.
3370 (expand_call_inline): Update count_max.
3371 (optimize_inline_calls): Update count_max.
3372 (tree_function_versioning): Update count_max.
3373 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3374 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3376 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3377 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3378 (try_peel_loop): Likewise.
3379 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3381 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3382 (tree_transform_and_unroll_loop): Do not use frequencies
3383 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3384 Use reliable prediction only.
3385 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3386 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3387 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3388 probability scaling.
3389 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3390 not update frequency
3391 (any_remaining_duplicated_blocks): Likewise.
3392 (update_profile): Likewise.
3393 (estimated_freqs_path): Delete.
3394 (freqs_to_counts_path): Delete.
3395 (clear_counts_path): Delete.
3396 (ssa_fix_duplicate_block_edges): Likewise.
3397 (duplicate_thread_path): Likewise.
3398 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3399 * tree-tailcall.c (decrease_profile): Do not update frequency.
3400 (eliminate_tail_call): Likewise.
3401 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3402 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3403 (optimize_mask_stores): Likewise.
3404 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3405 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3406 (ubsan_expand_ptr_ifn): Update profile.
3407 * value-prof.c (gimple_ic): Simplify.
3408 * value-prof.h (gimple_ic): Update prototype.
3409 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3410 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3412 (want_inline_self_recursive_call_p): Likewise.
3413 (resolve_noninline_speculation): Only cummulate defined counts.
3414 (inline_small_functions): Use nonzero_p.
3415 (ipa_inline): Do not access freed node.
3417 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3419 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3420 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3422 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3424 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3425 non-legitimate address.
3427 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3429 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3430 (*lt0_<mode>di, *lt0_<mode>si): New.
3432 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3434 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3435 TARGET_PAIRED_FLOAT.
3437 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
3438 Jim Wilson <jim.wilson@linaro.org>
3440 * config/aarch64/aarch64-cores.def (saphira): New CPU.
3441 * config/aarch64/aarch64-tune.md: Regenerated.
3442 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
3443 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
3445 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
3447 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
3448 (arc_expand_prologue): Restore blink for millicode.
3449 * config/arc/linux.h (LINK_EH_SPEC): Defined.
3451 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
3454 * config/i386/i386.c (ix86_vector_duplicate_value): Use
3455 gen_vec_duplicate after forcing the scalar into a register.
3457 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
3459 * combine (try_combine): Print the insns input to try_combine to the
3462 2017-11-02 Steve Ellcey <sellcey@cavium.com>
3465 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
3466 Remove second argument from aarch64_process_target_attr call.
3467 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
3469 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
3471 (aarch64_handle_attr_arch): Remove second argument.
3472 (aarch64_handle_attr_cpu): Ditto.
3473 (aarch64_handle_attr_tune): Ditto.
3474 (aarch64_handle_attr_isa_flags): Ditto.
3475 (aarch64_process_one_target_attr): Ditto.
3476 (aarch64_process_target_attr): Ditto.
3477 (aarch64_option_valid_attribute_p): Remove second argument.
3478 on aarch64_process_target_attr call.
3480 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3482 * diagnostic.c: Include "selftest-diagnostic.h".
3483 (selftest::assert_location_text): New function.
3484 (selftest::test_diagnostic_get_location_text): New function.
3485 (selftest::diagnostic_c_tests): Call it.
3487 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3489 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
3490 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
3491 (class selftest::test_diagnostic_context): Move to...
3492 * selftest-diagnostic.c: New file.
3493 * selftest-diagnostic.h: New file.
3495 2017-11-02 James Bowman <james.bowman@ftdichip.com>
3497 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
3498 offset range for FT32B.
3499 * config/ft32/ft32.h: option "mcompress" enables relaxation.
3500 * config/ft32/ft32.md: Add TARGET_NOPM.
3501 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
3502 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
3504 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
3506 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
3508 2017-11-02 Jeff Law <law@redhat.com>
3510 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
3511 virtual keyword on FINAL OVERRIDE members.
3513 * tree-ssa-propagate.h (ssa_propagation_engine): Group
3514 virtuals together. Add virtual destructor.
3515 (substitute_and_fold_engine): Similarly.
3517 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
3519 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
3521 2017-11-02 Richard Biener <rguenther@suse.de>
3523 PR tree-optimization/82795
3524 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
3526 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3528 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
3530 (gcc_GAS_CHECK_FEATURE): Remove.
3531 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
3532 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
3533 * configure: Regenerate.
3535 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
3537 * config/arc/arc.c (hwloop_optimize): Account for empty
3540 2017-11-02 Richard Biener <rguenther@suse.de>
3543 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
3544 Truncate ARRAY_REF index and element size.
3546 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3548 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
3550 2017-11-01 Jeff Law <law@redhat.com>
3552 * tree-ssa-ccp.c (ccp_folder): New class derived from
3553 substitute_and_fold_engine.
3554 (ccp_folder::get_value): New member function.
3555 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
3556 (ccp_fold_stmt): Remove prototype.
3557 (ccp_finalize): Call substitute_and_fold from the ccp_class.
3558 * tree-ssa-copy.c (copy_folder): New class derived from
3559 substitute_and_fold_engine.
3560 (copy_folder::get_value): Renamed from get_value.
3561 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
3562 * tree-vrp.c (vrp_folder): New class derived from
3563 substitute_and_fold_engine.
3564 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
3565 (vrp_folder::get_value): New member function.
3566 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
3567 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
3568 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
3569 provide a class interface to folder/substitute routines.
3570 (ssa_prop_fold_stmt_fn): Remove typedef.
3571 (ssa_prop_get_value_fn): Likewise.
3572 (subsitute_and_fold): Remove prototype.
3573 (replace_uses_in): Likewise.
3574 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3575 Renamed from replace_uses_in. Call the virtual member function
3576 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3577 (substitute_and_fold_dom_walker): Remove initialization of
3578 data member entries for calbacks. Add substitute_and_fold_engine
3579 member and initialize it.
3580 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3581 member functions for get_value, replace_phi_args_in c
3582 replace_uses_in, and fold_stmt calls.
3583 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3584 substitute_and_fold. Remove assert. Update ctor call.
3586 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3587 (ssa_prop_visit_phi_fn): Likewise.
3588 (class ssa_propagation_engine): New class to provide an interface
3590 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3592 (ssa_prop_visit_phi): Likewise.
3593 (ssa_propagation_engine::simulate_stmt): Moved into class.
3594 Call visit_phi/visit_stmt from the class rather than via
3595 file scoped static variables.
3596 (ssa_propagation_engine::simulate_block): Moved into class.
3597 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3598 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3599 set file scoped statics for the visit_stmt/visit_phi callbacks.
3600 * tree-complex.c (complex_propagate): New class derived from
3601 ssa_propagation_engine.
3602 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3603 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3604 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3606 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3607 ssa_propagation_engine.
3608 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3609 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3610 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3611 * tree-ssa-copy.c (copy_prop): New class derived from
3612 ssa_propagation_engine.
3613 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3614 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3615 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3616 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3617 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3618 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3619 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3621 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3623 PR rtl-optimization/82778
3624 PR rtl-optimization/82597
3625 * compare-elim.c (struct comparison): Add in_a_setter field.
3626 (find_comparison_dom_walker::before_dom_children): Remove killed
3627 bitmap and df_simulate_find_defs call, instead walk the defs.
3628 Compute last_setter and initialize in_a_setter. Merge definitions
3629 with first initialization for a few variables.
3630 (try_validate_parallel): Use insn_invalid_p instead of
3631 recog_memoized. Return insn rather than just the pattern.
3632 (try_merge_compare): Fix up comment. Don't uselessly test if
3633 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3635 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3638 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3639 Alan Hayward <alan.hayward@arm.com>
3640 David Sherwood <david.sherwood@arm.com>
3642 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3643 aarch64_hard_regno_nregs to get the number of registers
3646 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3647 Alan Hayward <alan.hayward@arm.com>
3648 David Sherwood <david.sherwood@arm.com>
3650 * config/aarch64/constraints.md (Upl): Rename to...
3652 * config/aarch64/aarch64.md
3653 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3656 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3657 Alan Hayward <alan.hayward@arm.com>
3658 David Sherwood <david.sherwood@arm.com>
3660 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3661 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3664 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3665 Alan Hayward <alan.hayward@arm.com>
3666 David Sherwood <david.sherwood@arm.com>
3668 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3669 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3670 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3672 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3673 of a single element.
3674 * config/aarch64/iterators.md: Add a comment above the permute
3675 unspecs to say that they are generated directly by
3676 aarch64_expand_vec_perm_const.
3677 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3679 2017-11-01 Nathan Sidwell <nathan@acm.org>
3681 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3683 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3685 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3686 and medany code models, and describe what they do.
3688 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3690 Revert accidental duplicate:
3692 * combine.c (can_change_dest_mode): Reject changes in
3693 REGMODE_NATURAL_SIZE.
3695 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3697 PR rtl-optimization/64682
3698 PR rtl-optimization/69567
3699 PR rtl-optimization/69737
3700 PR rtl-optimization/82683
3701 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3702 register mentioned in the note, drop the note, unless it came from I3,
3703 in which case it should go to I3 again.
3705 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3707 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3708 and return false if not.
3709 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3711 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3713 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3714 range is known to be empty.
3716 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3717 Alan Hayward <alan.hayward@arm.com>
3718 David Sherwood <david.sherwood@arm.com>
3720 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3721 and CONST_VECTOR_NUNITS instead of computing the number of units from
3722 the byte sizes of the vector and element.
3723 (simplify_binary_operation_1): Likewise.
3724 (simplify_const_binary_operation): Likewise.
3725 (simplify_ternary_operation): Likewise.
3727 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3728 Alan Hayward <alan.hayward@arm.com>
3729 David Sherwood <david.sherwood@arm.com>
3731 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3732 (int_mem_offset): ...this new function.
3733 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3734 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3735 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3738 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3739 Alan Hayward <alan.hayward@arm.com>
3740 David Sherwood <david.sherwood@arm.com>
3742 * lower-subreg.c (interesting_mode_p): New function.
3743 (compute_costs, find_decomposable_subregs, decompose_register)
3744 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3745 (resolve_clobber, dump_choices): Use it.
3747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3748 Alan Hayward <alan.hayward@arm.com>
3749 David Sherwood <david.sherwood@arm.com>
3751 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3753 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3754 Alan Hayward <alan.hayward@arm.com>
3755 David Sherwood <david.sherwood@arm.com>
3757 * alias.c (find_base_value, find_base_term): Only process integer
3758 truncations. Check the precision rather than the size.
3760 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3761 Alan Hayward <alan.hayward@arm.com>
3762 David Sherwood <david.sherwood@arm.com>
3764 * machmode.h (is_narrower_int_mode): New function
3765 * optabs.c (expand_float, expand_fix): Use it.
3766 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3768 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3769 Alan Hayward <alan.hayward@arm.com>
3770 David Sherwood <david.sherwood@arm.com>
3772 * rtl.h (narrower_subreg_mode): New function.
3773 * ira-color.c (update_costs_from_allocno): Use it.
3775 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3776 Alan Hayward <alan.hayward@arm.com>
3777 David Sherwood <david.sherwood@arm.com>
3779 * optabs-query.h (convert_optab_p): New function, split out from...
3780 (convert_optab_handler): ...here.
3781 (widening_optab_handler): Delete.
3782 (find_widening_optab_handler): Remove permit_non_widening parameter.
3783 (find_widening_optab_handler_and_mode): Likewise. Provide an
3784 override that operates on mode class wrappers.
3785 * optabs-query.c (widening_optab_handler): Delete.
3786 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3787 parameter. Assert that the two modes are the same class and that
3788 the "from" mode is narrower than the "to" mode. Use
3789 convert_optab_handler instead of widening_optab_handler.
3790 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3791 instead of widening_optab_handler.
3792 * expr.c (expand_expr_real_2): Update calls to
3793 find_widening_optab_handler.
3794 * optabs.c (expand_widen_pattern_expr): Likewise.
3795 (expand_binop_directly): Take the insn_code as a parameter.
3796 (expand_binop): Only call find_widening_optab_handler for
3797 conversion optabs; use optab_handler otherwise. Update calls
3798 to find_widening_optab_handler and expand_binop_directly.
3799 Use convert_optab_handler instead of widening_optab_handler.
3800 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3801 find_widening_optab_handler and use scalar_mode rather than
3803 (convert_plusminus_to_widen): Likewise.
3805 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3806 Alan Hayward <alan.hayward@arm.com>
3807 David Sherwood <david.sherwood@arm.com>
3809 * machmode.h (fixed_size_mode): New class.
3810 * rtl.h (get_pool_mode): Return fixed_size_mode.
3811 * gengtype.c (main): Add fixed_size_mode.
3812 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3813 (get_raw_arg_mode): Likewise.
3814 * doc/tm.texi: Regenerate.
3815 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3816 * targhooks.c (default_get_reg_raw_mode): Likewise.
3817 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3818 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3819 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3820 (msp430_get_raw_result_mode): Likewise.
3821 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3822 * dbxout.c (dbxout_parms): Require fixed-size modes.
3823 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3824 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3825 * omp-low.c (lower_oacc_reductions): Likewise.
3826 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3827 (simplify_subreg): Update accordingly.
3828 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3829 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3830 that aren't fixed-size.
3831 (get_pool_mode): Return a fixed_size_mode.
3832 (output_constant_pool_2): Take a fixed_size_mode.
3834 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3835 Alan Hayward <alan.hayward@arm.com>
3836 David Sherwood <david.sherwood@arm.com>
3838 * doc/rtl.texi (vec_series): Document.
3839 (const): Say that the operand can be a vec_series.
3840 * rtl.def (VEC_SERIES): New rtx code.
3841 * rtl.h (const_vec_series_p_1): Declare.
3842 (const_vec_series_p): New function.
3843 * emit-rtl.h (gen_const_vec_series): Declare.
3844 (gen_vec_series): Likewise.
3845 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3846 (gen_vec_series): Likewise.
3847 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3848 * simplify-rtx.c (simplify_unary_operation): Handle negations
3850 (simplify_binary_operation_series): New function.
3851 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3852 (test_vector_ops_series): New function.
3853 (test_vector_ops): Call it.
3854 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3855 gen_const_vec_series.
3856 (altivec_lvsr): Likewise.
3857 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3859 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3860 Alan Hayward <alan.hayward@arm.com>
3861 David Sherwood <david.sherwood@arm.com>
3863 * doc/rtl.texi (const): Update description of address constants.
3864 Say that vector constants are allowed too.
3865 * common.md (E, F): Use CONSTANT_P instead of checking for
3867 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3868 checking for CONST_VECTOR.
3869 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3871 * expr.c (expand_expr_real_2): Check for vector modes instead
3872 of checking for CONST_VECTOR.
3873 * rtl.h (const_vec_p): New function.
3874 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3875 (unwrap_const_vec_duplicate): Handle them here too.
3877 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3878 David Malcolm <dmalcolm@redhat.com>
3879 Alan Hayward <alan.hayward@arm.com>
3880 David Sherwood <david.sherwood@arm.com>
3882 * rtl.h (vec_duplicate_p): New function.
3883 * selftest-rtl.c (assert_rtx_eq_at): New function.
3884 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3885 (assert_rtx_eq_at): Declare.
3886 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3887 * selftest-run-tests.c (selftest::run_tests): Call it.
3888 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3889 (simplify_unary_operation_1): Recursively handle vector duplicates.
3890 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3892 (simplify_subreg): Handle subregs of vector duplicates.
3893 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3894 (selftest::simplify_rtx_c_tests): New functions.
3896 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3897 Alan Hayward <alan.hayward@arm.com>
3898 David Sherwood <david.sherwood@arm.com>
3900 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3901 (gen_vec_duplicate): Likewise.
3902 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3904 (gen_const_vector): ...here.
3905 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3906 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3907 whose elements are all equal.
3908 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3909 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3910 (simplify_relational_operation): Likewise.
3911 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3913 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3914 (aarch64_expand_vector_init): Likewise.
3915 * config/arm/arm.c (neon_vdup_constant): Likewise.
3916 (neon_expand_vector_init): Likewise.
3917 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3918 (arm_block_set_unaligned_vect): Likewise.
3919 (arm_block_set_aligned_vect): Likewise.
3920 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3921 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3922 (expand_vec_perm_even_odd_pack): Likewise.
3923 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3924 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3925 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3926 gen_const_vec_duplicate.
3927 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3928 * config/mips/mips.c (mips_gen_const_int_vector): Use
3929 gen_const_vec_duplicate.
3930 (mips_expand_vector_init): Use CONST0_RTX.
3931 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3932 (define_split): Use gen_const_vec_duplicate.
3933 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3934 (define_split): Use gen_const_vec_duplicate.
3935 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3936 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3937 * config/spu/spu.c (spu_const): Likewise.
3939 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3940 Alan Hayward <alan.hayward@arm.com>
3941 David Sherwood <david.sherwood@arm.com>
3943 * combine.c (can_change_dest_mode): Reject changes in
3944 REGMODE_NATURAL_SIZE.
3946 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3948 * configure.ac (--enable-libssp): New.
3949 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3950 trying to determine target-specific default. Adjust indentation.
3951 * configure: Regenerated.
3952 * doc/install.texi (Configuration): Expand --disable-libssp
3955 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3957 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3960 2017-10-31 Martin Jambor <mjambor@suse.cz>
3963 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3965 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3967 * auto-profile.c (autofdo_source_profile::read): Use
3968 UNKNOWN_LOCATION rather than 0.
3969 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3970 (warning_at): ...this overload.
3971 (warning_at_rich_loc_n): Rename to...
3972 (warning_n): ...this overload.
3973 (error_at_rich_loc): Rename to...
3974 (error_at): ...this overload.
3975 (pedwarn_at_rich_loc): Rename to...
3976 (pedwarn): ...this overload.
3977 (permerror_at_rich_loc): Rename to...
3978 (permerror): ...this overload.
3979 (inform_at_rich_loc): Rename to...
3980 (inform): ...this overload.
3981 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3982 (diagnostic_n_impl_richloc): Rename to...
3983 (diagnostic_n_impl): ...this rich_location *-based decl.
3984 (inform_at_rich_loc): Rename to...
3985 (inform): ...this, and add an assertion.
3986 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3987 (warning_at_rich_loc): Rename to...
3988 (warning_at): ...this, and add an assertion.
3989 (warning_at_rich_loc_n): Rename to...
3990 (warning_n): ...this, and add an assertion.
3991 (warning_n): Update location_t-based implementation for removal of
3992 location_t-based diagnostic_n_impl.
3993 (pedwarn_at_rich_loc): Rename to...
3994 (pedwarn): ...this, and add an assertion.
3995 (permerror_at_rich_loc): Rename to...
3996 (permerror): ...this, and add an assertion.
3997 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3998 (error_at_rich_loc): Rename to...
3999 (error_at): ...this, and add an assertion.
4000 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4001 (driver::do_spec_on_infiles): Likewise.
4002 * substring-locations.c (format_warning_va): Update for renaming
4003 of inform_at_rich_loc.
4005 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4007 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4008 _Float<N>X built-in functions so that the variant without the
4009 "__builtin_" prefix is only enabled for the GNU C and Objective C
4010 languages when they are in non-strict ANSI/ISO mode.
4011 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4012 * target.def (floatn_builtin_p): Add a target hook to control
4013 whether _Float<N> and _Float<N>X built-in functions without the
4014 "__builtin_" prefix are enabled, and return true for C and
4015 Objective C in the default hook. Include langhooks.h in
4017 * targhooks.h (default_floatn_builtin_p): Likewise.
4018 * targhooks.c (default_floatn_builtin_p): Likewise.
4019 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4020 floatn_builtin_p target hook.
4021 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4023 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4024 Eric Botcazou <ebotcazou@adacore.com>
4026 PR rtl-optimization/81803
4027 * lra-constraints.c (curr_insn_transform): Also reload the whole
4028 register for a strict subreg no wider than a word if this is for
4029 a WORD_REGISTER_OPERATIONS target.
4031 2017-10-31 Jason Merrill <jason@redhat.com>
4033 * gdbinit.in: Skip over inlines from timevar.h.
4035 2017-10-31 Martin Liska <mliska@suse.cz>
4037 * doc/gcov.texi: Document new option.
4038 * gcov.c (print_usage): Likewise print it.
4039 (process_args): Support the argument.
4040 (format_count): New function.
4041 (format_gcov): Use the function.
4043 2017-10-31 Martin Liska <mliska@suse.cz>
4045 * gcov.c (struct name_map): do not use typedef.
4046 Define operator== and operator<.
4047 (name_search): Remove.
4048 (name_sort): Remove.
4049 (main): Do not allocate names.
4050 (process_file): Add vertical space.
4051 (generate_results): Use std::find.
4052 (release_structures): Do not release memory.
4053 (find_source): Use std::find.
4055 2017-10-31 Martin Liska <mliska@suse.cz>
4057 * gcov.c (struct line_info): Remove it's typedef.
4058 (line_info::line_info): Add proper ctor.
4059 (line_info::has_block): Do not use a typedef.
4060 (struct source_info): Do not use typedef.
4061 (circuit): Likewise.
4062 (get_cycles_count): Likewise.
4063 (output_intermediate_file): Iterate via vector iterator.
4064 (add_line_counts): Use std::vector methods.
4065 (accumulate_line_counts): Likewise.
4066 (output_lines): Likewise.
4068 2017-10-31 Martin Liska <mliska@suse.cz>
4070 * gcov.c (struct source_info): Remove typedef.
4071 (source_info::source_info): Add proper ctor.
4072 (accumulate_line_counts): Use struct, not it's typedef.
4073 (output_gcov_file): Likewise.
4074 (output_lines): Likewise.
4075 (main): Do not allocate an array.
4076 (output_intermediate_file): Use size of vector container.
4077 (process_file): Resize the vector.
4078 (generate_results): Do not preallocate, use newly added vector
4080 (release_structures): Do not release sources.
4081 (find_source): Use vector methods.
4082 (add_line_counts): Do not use typedef.
4084 2017-10-31 Martin Liska <mliska@suse.cz>
4086 * doc/gcov.texi: Document that.
4087 * gcov.c (add_line_counts): Mark lines with a non-executed
4089 (output_line_beginning): Handle such lines.
4090 (output_lines): Pass new argument.
4091 (output_intermediate_file): Print it in intermediate format.
4093 2017-10-31 Martin Liska <mliska@suse.cz>
4095 * color-macros.h: New file.
4096 * diagnostic-color.c: Factor out color related to macros to
4098 * doc/gcov.texi: Document -k option.
4099 * gcov.c (INCLUDE_STRING): Include string.h.
4100 (print_usage): Add -k option.
4101 (process_args): Parse it.
4102 (pad_count_string): New function.
4103 (output_line_beginning): Likewise.
4104 (DEFAULT_LINE_START): New macro.
4105 (output_lines): Support color output.
4107 2017-10-31 Martin Liska <mliska@suse.cz>
4109 PR gcov-profile/82633
4110 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4111 their interaction with GCOV infrastructure.
4112 * configure.ac: Add -fkeep-{inline,static}-functions to
4114 * configure: Regenerate.
4116 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4119 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4121 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4124 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4125 into a register if it does not fit into an immediate offset field.
4127 2017-10-31 Olivier Hainque <hainque@adacore.com>
4129 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4131 2017-10-31 Julia Koval <julia.koval@intel.com>
4133 * config.gcc: Add gfniintrin.h.
4134 * config/i386/gfniintrin.h: New.
4135 * config/i386/i386-builtin-types.def
4136 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4137 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4138 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4139 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4140 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4141 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
4142 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
4143 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
4144 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
4145 V64QI_FTYPE_V64QI_V64QI_INT): New types.
4146 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4147 * config/i386/immintrin.h: Include gfniintrin.h.
4148 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
4150 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
4152 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
4154 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4156 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
4157 (arm_ashldi3_1bit): Remove pattern.
4158 (ashrdi3): Remove shift by 1 expansion.
4159 (arm_ashrdi3_1bit): Remove pattern.
4160 (lshrdi3): Remove shift by 1 expansion.
4161 (arm_lshrdi3_1bit): Remove pattern.
4162 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
4163 cost of ashldi3 by 1.
4164 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
4165 (<shift>di3_neon): Likewise.
4167 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4169 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
4170 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
4171 and (*aarch64_simd_mov<VQ:mode>).
4172 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
4173 pattern alternative.
4174 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
4175 attributes to match pattern alternative.
4177 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
4179 * config.gcc (powerpc*-*-*): Add emmintrin.h.
4180 * config/rs6000/emmintrin.h: New file.
4181 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
4183 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4185 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
4186 * (movdi_vfp_cortexa8): Remove pattern.
4188 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4190 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
4192 (Specific, alpha*-dec-osf5.1): Remove.
4193 (Specific, mips-sgi-irix5): Remove.
4194 (Specific, mips-sgi-irix6): Remove.
4196 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4199 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
4200 arguments to clear_bit_region_be.
4202 2017-10-30 Jim Wilson <wilson@tuliptree.org>
4204 * gimplify.c: Include memmodel.h.
4206 2017-10-30 Martin Jambor <mjambor@suse.cz>
4208 * omp-grid.c (grid_attempt_target_gridification): Also insert a
4209 condition whether loop should be executed at all.
4211 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
4213 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4214 gimple folding of vec_madd() intrinsics.
4215 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
4216 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
4217 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
4219 2017-10-30 Richard Biener <rguenther@suse.de>
4221 PR tree-optimization/82762
4223 2017-10-23 Richard Biener <rguenther@suse.de>
4225 PR tree-optimization/82129
4227 2017-08-01 Richard Biener <rguenther@suse.de>
4229 PR tree-optimization/81181
4230 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4231 (compute_antic): ... end of iteration here.
4233 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4235 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4236 -std=iso9899:2017 and -std=gnu17.
4237 * doc/standards.texi (C Language): Document C17 support.
4238 * doc/cpp.texi (Overview): Mention -std=c17.
4239 (Standard Predefined Macros): Document C11 and C17 values of
4240 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4241 * doc/extend.texi (Inline): Do not list individual options for
4242 standards newer than C99.
4243 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4245 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4248 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4250 * asan.c (asan_finish_file): Align asan globals array by shadow
4253 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4256 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4257 (struct store_immediate_info): Add bitregion_start and bitregion_end
4259 (store_immediate_info::store_immediate_info): Add brs and bre
4260 arguments and initialize bitregion_{start,end} from those.
4261 (struct merged_store_group): Add bitregion_start, bitregion_end,
4262 align_base and mask fields. Drop unnecessary struct keyword from
4263 struct store_immediate_info. Add do_merge method.
4264 (clear_bit_region_be): Use memset instead of loop storing zeros.
4265 (merged_store_group::do_merge): New method.
4266 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4267 stores as long as the surrounding bitregions have no gaps.
4268 (merged_store_group::merge_overlapping): Use do_merge.
4269 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4270 is byte aligned, rather than requiring that start and width are
4271 byte aligned. Drop unnecessary struct keyword from
4272 struct store_immediate_info. Allocate and populate also mask array.
4273 Make start of the arrays relative to bitregion_start rather than
4274 start and size them according to bitregion_{end,start} difference.
4275 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4276 struct store_immediate_info.
4277 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4278 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4279 struct keyword from struct store_immediate_info.
4280 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4281 between stores as long as the surrounding bitregions have no gaps.
4283 (struct split_store): Add orig non-static data member.
4284 (split_store::split_store): Initialize orig to false.
4285 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4286 if there is exactly a single original stmt. Change stmts argument
4287 to pointer from reference, if NULL, don't push anything to it. Add
4288 first argument, use it to optimize skipping over orig stmts that
4289 are known to be before bitpos already. Simplify.
4290 (split_group): Return unsigned int count how many stores are or
4291 would be needed rather than a bool. Add allow_unaligned argument.
4292 Change split_stores argument from reference to pointer, if NULL,
4293 only do a dry run computing how many stores would be produced.
4294 Rewritten algorithm to use both alignment and misalign if
4295 !allow_unaligned and handle bitfield stores with gaps.
4296 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4297 from bitregion_start instead of start. Compute allow_unaligned
4298 here, if true, do 2 split_group dry runs to compute which one
4299 produces fewer stores and prefer aligned if equal. Punt if
4300 new count is bigger or equal than original before emitting any
4301 statements, rather than during that. Remove no longer needed
4302 new_ssa_names tracking. Replace num_stmts with
4303 split_stores.length (). Use 32-bit stack allocated entries
4304 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4305 if possible. Handle bitfields with gaps.
4306 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4307 Compute bitregion_{start,end} for the stores and construct
4308 store_immediate_info with that. Formatting fixes.
4310 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4313 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4314 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4316 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4318 * gimplify.c: Include tm_p.h.
4320 * common.opt (gcoff): Re-add as ignored option.
4321 (gcoff1, gcoff2, gcoff3): Likewise.
4323 * Makefile.in (OBJS): Delete sdbout.o.
4324 (GTFILES): Delete $(srcdir)/sdbout.c.
4325 * debug.h: Delete sdb_debug_hooks.
4326 * final.c: Delete sdbout.h include.
4327 (final_scan_insn): Delete SDB_DEBUG check.
4328 (rest_of_clean_state): Likewise.
4329 * output.h: Delete sdb_begin_function_line.
4332 * toplev.c: Delete sdbout.h include.
4333 (process_options): Delete SDB_DEBUG check.
4334 * tree-core.h (tree_type_common): Delete pointer field of
4336 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4337 TYPE_SYMTAB_POINTER.
4338 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4339 (TYPE_SYMTAB_IS_POINTER): Delete.
4340 (TYPE_SYMTAB_IS_DIE): Renumber.
4341 * xcoffout.c: Refer to former sdbout.c file.
4342 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4344 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4345 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4346 (Debugging Options): Delete -gcoff.
4347 (-gstabs): Delete SDB reference.
4349 (-gcoff@var{level}): Delete.
4350 * doc/passes.texi (Debugging information output): Delete SDB and
4351 sdbout.c references.
4352 * doc/tm.texi: Regenerate.
4353 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4354 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4356 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4357 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4358 (SDB_DEBUGGING_INFO): Delete.
4359 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4360 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4361 * target.def (output_source_filename): Delete COFF reference.
4363 * common.opt (gcoff): Delete.
4364 (gxcoff+): Update Negative chain.
4365 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4367 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4368 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4369 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4371 (expand_function_start): Change sdb reference to past tense.
4372 (expand_function_end): Change sdb reference to past tense.
4373 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4374 * opts.c (debug_type_names): Delete coff entry.
4375 (common_handle_option): Delete OPT_gcoff case.
4376 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4378 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4379 * config/cris/cris.h: Delete SDB reference in comment.
4380 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4381 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4382 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4383 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4385 (ix86_expand_prologue): Likewise.
4386 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4387 * config/ia64/ia64.h: Likewise.
4388 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4389 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4391 * config/mmix/mmix.h: Likewise.
4392 * config/nds32/nds32.c: Likewise.
4393 * config/stormy/storym16.h: Likewise.
4394 * config/visium/visium.h: Likewise.
4395 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4397 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4399 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4400 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4401 FRAME_POINTER_REGNUM point at high end of local var area.
4403 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4405 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4406 Move comment around. Do not reset best_edge for a copiable
4407 destination if the copy would cause a partition change.
4408 (better_edge_p): Remove redundant check.
4410 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4412 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4414 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4416 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4417 for math functions that have _Float<N> and _Float<N>X variants.
4418 (mathfn_built_in_2): Add support for math functions that have
4419 _Float<N> and _Float<N>X variants.
4420 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4421 (expand_builtin_mathfn_ternary): Add support for fma with
4422 _Float<N> and _Float<N>X variants.
4423 (expand_builtin): Likewise.
4424 (fold_builtin_3): Likewise.
4425 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4426 create math function _Float<N> and _Float<N>X variants as external
4428 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4429 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4430 the __builtin_ prefix and if not strict ansi, without the prefix.
4431 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4432 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4433 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4434 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4435 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4436 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4437 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
4438 function signatures for fma _Float<N> and _Float<N>X variants.
4439 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
4440 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
4441 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
4442 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
4443 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
4444 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
4445 * gencfn-macros.c (print_case_cfn): Add support for math functions
4446 that have _Float<N> and _Float<N>X variants.
4447 (print_define_operator_list): Likewise.
4448 (fltfn_suffixes): Likewise.
4450 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
4451 for math functions that have _Float<N> and _Float<N>X variants.
4452 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
4453 and _Float<N>X variants.
4454 (COPYSIGN): Likewise.
4457 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
4458 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
4460 (integer_valued_read_call_p): Likewise.
4461 * fold-const-call.c (fold_const_call_ss): Likewise.
4462 (fold_const_call_sss): Add support for copysign, fmin, and fmax
4463 _Float<N> and _Float<N>X variants.
4464 (fold_const_call_ssss): Add support for fma _Float<N> and
4465 _Float<N>X variants.
4466 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
4467 support for copysign and fma _Float<N> and _Float<N>X variants.
4468 (backprop::process_builtin_call_use): Likewise.
4469 * tree-call-cdce.c (can_test_argument_range); Add support for
4470 sqrt _Float<N> and _Float<N>X variants.
4471 (edom_only_function): Likewise.
4472 (get_no_error_domain): Likewise.
4473 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
4474 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
4475 copysign _Float<N> and _Float<N>X variants.
4476 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
4477 handled by machine independent code.
4478 (FMAF128): Likewise.
4479 * doc/cpp.texi (Common Predefined Macros): Document defining
4480 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
4481 fma _Float<N> and _Float<N>X variants.
4483 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4486 * config/i386/i386-modes.def (CCFPU): Remove definition.
4487 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
4488 (ix86_cc_modes_compatible): Ditto.
4489 (ix86_expand_carry_flag_compare): Ditto.
4490 (ix86_expand_int_movcc): Ditto.
4491 (ix86_expand_int_addcc): Ditto.
4492 (ix86_reverse_condition): Ditto.
4493 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
4494 Return true/false for unordered/ordered fp comparisons.
4495 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
4496 (ix86_prepare_fp_compare_args): Update for rename.
4497 (ix86_expand_fp_compare): Update for rename. Generate unordered
4498 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
4499 (ix86_expand_sse_compare_and_jump): Ditto.
4500 * config/i386/predicates.md (fcmov_comparison_operator):
4501 Remove CCFPU mode handling.
4502 (ix86_comparison_operator): Ditto.
4503 (ix86_carry_flag_operator): Ditto.
4504 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
4505 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
4506 (*cmpu<mode>_cc_i387): Ditto.
4507 (FPCMP): Remove mode iterator.
4508 (unord): Remove mode attribute.
4509 (unord_subst): New define_subst transformation
4510 (unord): New define_subst attribute.
4512 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
4513 (*cmpi<unord>xf_i387): Ditto.
4514 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
4515 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
4516 using unord_subst transformation.
4517 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
4518 (round_saeonly): Also handle CCFP mode.
4519 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
4520 Remove UNSPEC_SAHF unspec handling.
4522 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4524 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
4526 2017-10-27 Jeff Law <law@redhat.com>
4528 * gimple-ssa-sprintf.c: Include domwalk.h.
4529 (class sprintf_dom_walker): New class, derived from dom_walker.
4530 (sprintf_dom_walker::before_dom_children): New function.
4531 (struct call_info): Moved into sprintf_dom_walker class
4532 (compute_formath_length, handle_gimple_call): Likewise.
4533 (sprintf_length::execute): Call the dominator walker rather
4534 than walking the statements.
4536 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
4537 gimple statement locations.
4538 (check_array_bounds): Corresponding changes. Get the statement's
4539 location directly from wi->stmt.
4541 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
4544 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
4546 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4548 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
4549 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
4551 2017-10-27 Jakub Jelinek <jakub@redhat.com>
4554 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
4555 * config/i386/i386.c (maybe_get_pool_constant): Removed.
4556 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
4557 maybe_get_pool_constant.
4558 * config/i386/predicates.md (zero_extended_scalar_load_operand):
4561 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4563 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
4564 2.26 caveat. Update gas and gld versions.
4565 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
4568 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4570 * cgraph.h (set_malloc_flag): Declare.
4571 * cgraph.c (set_malloc_flag_1): New function.
4572 (set_malloc_flag): Likewise.
4573 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4574 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4576 (read_ipa_call_summary): Add support for reading is_return_callee.
4577 (write_ipa_call_summary): Stream is_return_callee.
4578 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4579 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4580 ipa-prop.h, ipa-fnsummary.h.
4581 (pure_const_names): Change to static.
4582 (malloc_state_e): Define.
4583 (malloc_state_names): Define.
4584 (funct_state_d): Add field malloc_state.
4585 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4586 (check_retval_uses): New function.
4587 (malloc_candidate_p): Likewise.
4588 (analyze_function): Add support for malloc attribute.
4589 (pure_const_write_summary): Stream malloc_state.
4590 (pure_const_read_summary): Add support for reading malloc_state.
4591 (dump_malloc_lattice): New function.
4592 (propagate_malloc): New function.
4593 (warn_function_malloc): New function.
4594 (ipa_pure_const::execute): Call propagate_malloc and
4595 ipa_free_fn_summary.
4596 (pass_local_pure_const::execute): Add support for malloc attribute.
4597 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4598 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4600 2017-10-27 Martin Liska <mliska@suse.cz>
4602 PR gcov-profile/82457
4603 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4604 for fork-like functions to be properly instrumented.
4606 2017-10-27 Richard Biener <rguenther@suse.de>
4609 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4610 info when we redirected EH.
4612 2017-10-26 Michael Collison <michael.collison@arm.com>
4614 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4616 (<optab>_trunchf<GPI:mode>2: New pattern.
4617 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4618 * config/aarch64/iterators.md (wv): New mode attribute.
4619 (vf, VF): New mode attributes.
4620 (vgp, VGP): New mode attributes.
4621 (s): Update attribute with SImode and DImode prefixes.
4623 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4625 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4626 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4627 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4628 (nios2_option_overide): Initialize it. Don't allow R0-relative
4629 addressing with PIC.
4630 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4631 (nios2_symbolic_constant_p): Likewise.
4632 (nios2_legitimate_address_p): Likewise.
4633 (nios2_r0rel_section_name_p): New.
4634 (nios2_symbol_ref_in_r0rel_data_p): New.
4635 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4636 (r0rel_constant_p): New.
4637 (nios2_print_operand_address): Handle r0rel_constant_p.
4638 (nios2_cdx_narrow_form_p): Likewise.
4639 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4640 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4641 (Nios II Options): Document -mr0rel-sec.
4643 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4645 * config/nios2/nios2.c: Include xregex.h.
4646 (nios2_gprel_sec_regex): New.
4647 (nios2_option_overide): Initialize it. Don't allow GP-relative
4648 addressing with PIC.
4649 (nios2_small_section_name_p): Check for regex match.
4650 * config/nios2/nios2.opt (mgprel-sec=): New option.
4651 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4652 (Nios II Options): Document -mgprel-sec.
4654 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4656 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4658 2017-10-26 Tom de Vries <tom@codesourcery.com>
4660 PR tree-optimization/82707
4661 * gimple.c (gimple_copy): Fix unsharing of
4662 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4664 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4666 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4667 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4668 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4669 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4670 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4671 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4672 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4673 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4674 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4675 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4676 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4677 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4678 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4679 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4680 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4681 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4682 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4684 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4686 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4688 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4689 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4690 warning to rs6000.c. Remove the Undocumented flag, since it has
4692 (-mabi=ibmlongdouble): Likewise.
4693 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4694 already set, set the default format for long double.
4695 (rs6000_debug_reg_global): Print whether long double is IBM or
4697 (rs6000_option_override_internal): Rework setting long double
4698 format. Only warn if the user is changing the long double default
4699 and they did not use -Wno-psabi.
4700 * doc/invoke.texi (PowerPC options): Update the documentation for
4701 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4703 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4704 Alan Hayward <alan.hayward@arm.com>
4705 David Sherwood <david.sherwood@arm.com>
4707 * rtl.h (wider_subreg_mode): New function.
4708 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4709 rather than an unsigned int *.
4710 * ira-color.c (regno_max_ref_width): Replace with...
4711 (regno_max_ref_mode): ...this new variable.
4712 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4713 Use wider_subreg_mode.
4714 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4715 rather than an unsigned int *.
4716 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4717 (process_alt_operands): Likewise.
4718 (invariant_p): Likewise.
4719 * lra-spills.c (assign_mem_slot): Likewise.
4720 (add_pseudo_to_slot): Likewise.
4721 * lra.c (collect_non_operand_hard_regs): Likewise.
4722 (add_regs_to_insn_regno_info): Likewise.
4723 * reload1.c (regno_max_ref_width): Replace with...
4724 (regno_max_ref_mode): ...this new variable.
4725 (reload): Update accordingly. Update call to
4726 ira_sort_regnos_for_alter_reg.
4727 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4728 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4729 (scan_paradoxical_subregs): Likewise.
4731 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4733 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4734 (aarch64_frame): Add emit_frame_chain boolean.
4735 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4736 Move eh_return case to aarch64_layout_frame.
4737 (aarch64_layout_frame): Initialize emit_frame_chain.
4738 (aarch64_expand_prologue): Use emit_frame_chain.
4740 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4742 * config/aarch64/aarch64.c (aarch64_layout_frame):
4743 Ensure LR is always stored at the bottom of the callee-saves.
4744 Remove rarely used frame layout which saves callee-saves at top of
4745 frame, so the store of LR can be used as a valid probe in all cases.
4747 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4749 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4750 Improve unaligned TImode/TFmode base/offset split.
4752 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4753 Alan Hayward <alan.hayward@arm.com>
4754 David Sherwood <david.sherwood@arm.com>
4756 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4757 * combine.c (find_single_use_1): Likewise.
4758 (expand_field_assignment): Likewise.
4759 (move_deaths): Likewise.
4760 * lra-constraints.c (simplify_operand_subreg): Likewise.
4761 (curr_insn_transform): Likewise.
4762 * lra.c (collect_non_operand_hard_regs): Likewise.
4763 (add_regs_to_insn_regno_info): Likewise.
4764 * rtlanal.c (reg_referenced_p): Likewise.
4765 (covers_regno_no_parallel_p): Likewise.
4767 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4769 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4770 Don't print any bits outside the precision of the value.
4771 * wide-int.cc (test_printing): Add some new tests.
4773 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4775 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4776 supports -xbrace_comment option.
4777 * configure: Regenerate.
4778 * config.in: Regenerate.
4779 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4780 (ASM_CPU_SPEC): Use it.
4782 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4784 * target.def (static_rtx_alignment): New hook.
4785 * targhooks.h (default_static_rtx_alignment): Declare.
4786 * targhooks.c (default_static_rtx_alignment): New function.
4787 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4788 * doc/tm.texi: Regenerate.
4789 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4790 instead of targetm.constant_alignment. Remove call to
4792 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4793 (cris_preferred_mininum_alignment): New function, split out from...
4794 (cris_constant_alignment): ...here.
4795 (cris_static_rtx_alignment): New function.
4796 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4798 (ix86_constant_alignment): ...here.
4799 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4800 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4801 (mmix_static_rtx_alignment): New function.
4802 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4803 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4805 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4808 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4809 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4811 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4813 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4814 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4815 mark as a sign-extending load.
4816 (local_pic_load_u): Define.
4818 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4821 * fold-const.c (operand_equal_for_comparison_p): Also return true
4822 if ARG0 is a simple variant of ARG1 with narrower precision.
4823 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4825 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4827 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4829 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4830 scatter_static, scatter_per_elt.
4831 * x86-tune-costs.h: Add new cost entries.
4833 2017-10-25 Richard Biener <rguenther@suse.de>
4835 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4836 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4837 class pass_fre): Move to ...
4838 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4839 class pass_fre): ... here and adjust for statistics.
4841 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4844 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4845 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4846 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4847 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4850 2017-10-25 Richard Biener <rguenther@suse.de>
4852 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4853 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4854 eliminate_push_avail, eliminate_insert): Move inside...
4855 (class eliminate_dom_walker): ... this class in preparation
4857 (fini_eliminate): Remove by merging with ...
4858 (eliminate): ... this function. Adjust for class changes.
4859 (pass_pre::execute): Remove fini_eliminate call.
4860 (pass_fre::execute): Likewise.
4862 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4865 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4866 (VPERMI2, VPERMI2I): New mode iterators.
4867 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4868 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4870 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4871 mode iterator. Remove 3 old define_insn patterns.
4872 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4873 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4874 VPERMI2 mode iterator, remove the other two expanders.
4875 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4876 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4877 instructions, remove the other two patterns.
4878 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4879 mode iterator, remove the other two patterns.
4880 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4881 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4882 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4883 and adjust argument order accordingly.
4884 (ix86_expand_vec_perm): Adjust caller.
4885 (expand_vec_perm_1): Likewise.
4886 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4887 (expand_vec_perm_vpermt2_vpshub2): ... this.
4888 (ix86_expand_vec_perm_const_1): Adjust caller.
4889 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4892 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4893 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4894 (vec_shl_<mode>): Remove unused expander.
4895 (avx512bw_<shift_insn><mode>3): New define_insn.
4896 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4897 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4899 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4902 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4905 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4907 PR rtl-optimization/82396
4908 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4909 (autopref_multipass_init): Simplify initialization.
4910 (autopref_rank_data): Simplify sort order.
4911 * gcc/sched-int.h (autopref_multipass_data_): Remove
4912 multi_mem_insn_p, min_offset and max_offset.
4914 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4917 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4918 Check special value of flag_omit_frame_pointer.
4919 (aarch64_can_eliminate): Likewise.
4920 (aarch64_override_options_after_change_1): Simplify handling of
4921 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4923 2017-10-24 Richard Biener <rguenther@suse.de>
4925 PR tree-optimization/82697
4926 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4927 zero for conditional load and unconditional store.
4929 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4931 * doc/install.texi: Document bootstrap-cet.
4933 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4936 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4937 ENDBR instruction at function entrance if function is only
4940 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4943 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4944 patterns to better describe from which operation the CF is computed.
4945 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4946 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4947 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4948 is 0, use _0 suffixed expanders instead of emitting a comparison
4951 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4953 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4954 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4956 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4959 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4960 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4961 * loop-iv.c (iv_get_reaching_def): Likewise.
4962 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4963 variable is promoted and the partition contains undefined values.
4965 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4967 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4969 (nios2_address_cost): Define.
4970 (nios2_legitimize_address): Recognize (exp + constant) directly.
4971 (TARGET_ADDRESS_COST): Define.
4973 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4975 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4976 (nios2_symbolic_memory_operand_p): Declare.
4977 (nios2_split_large_constant): Declare.
4978 (nios2_split_symbolic_memory_operand): Declare.
4979 * config/nios2/nios2.c: Adjust includes.
4980 (nios2_symbolic_constant_allowed): New.
4981 (nios2_symbolic_constant_p): New.
4982 (nios2_plus_symbolic_constant_p): New.
4983 (nios2_valid_addr_expr_p): Recognize addresses involving
4985 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4986 (nios2_symbolic_memory_operand_p): New.
4987 (nios2_large_constant_p): New.
4988 (nios2_split_large_constant): New.
4989 (nios2_split_plus_large_constant): New.
4990 (nios2_split_symbolic_memory_operand): New.
4991 (nios2_legitimize_address): Code refactoring. Handle addresses
4992 involving symbolic constants.
4993 (nios2_emit_move_sequence): Likewise.
4994 (nios2_print_operand): Improve error output.
4995 (nios2_print_operand_address): Handle LO_SUM.
4996 (nios2_cdx_narrow_form_p): Likewise.
4997 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4998 operands involving symbolic constants.
4999 (movhi_internal, movsi_internal): Likewise.
5000 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5001 (extendhisi2, extendqi<mode>2): Likewise.
5003 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5005 * tree-pass.h (PROP_rtl_split_insns): Define.
5006 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5008 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5010 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5012 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5015 * target.def (const_not_ok_for_debug_p): Default to
5016 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5017 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5018 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5019 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5020 which targetm.const_not_ok_for_debug_p returned true.
5021 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5023 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5025 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5026 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5027 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5029 (ix86_const_not_ok_for_debug_p): New function.
5030 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5031 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5033 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5036 * system.h: Conditionally include "unique-ptr.h" if
5037 INCLUDE_UNIQUE_PTR is defined.
5038 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5039 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5041 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5043 * config/rl78/rl78.md: New define_expand "subdi3".
5045 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5048 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5049 DF_REF_INSN if DF_REF_INSN_INFO is false.
5051 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5053 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5054 xmm_move instead of sse_move.
5055 (sse_store_index): New function.
5056 (ix86_register_move_cost): Be more sensible about mismatch stall;
5057 model AVX moves correctly; make difference between sse->integer and
5059 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5060 moves; make difference between SSE and AVX.
5061 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5062 and zmm_move. Increase size of sse load and store tables;
5063 add unaligned load and store tables; add ssemmx_to_integer.
5064 * x86-tune-costs.h: Update all entries according to real
5065 move latencies from Agner Fog's manual and chip documentation.
5067 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5070 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5071 * config/i386/constraints.md (Wf): New constraint.
5072 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5073 (cmp<dwi>_doubleword): Removed.
5074 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5075 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5076 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5077 expand with cmp<dwi>_doubleword. For LTU and GEU use
5078 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5080 * common.opt (gcolumn-info): Enable by default.
5081 * doc/invoke.texi (gcolumn-info): Document new default.
5083 2017-10-23 Richard Biener <rguenther@suse.de>
5085 PR tree-optimization/82672
5086 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5087 Fold the stmt if we propagated into it.
5089 2017-10-23 Richard Biener <rguenther@suse.de>
5091 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5092 (bitmap_remove_expr_from_set): ... this. All callers call this
5093 for non-constant values.
5094 (bitmap_set_subtract): Rename to...
5095 (bitmap_set_subtract_expressions): ... this. Adjust and
5097 (bitmap_set_contains_value): Remove superfluous check.
5098 (bitmap_set_replace_value): Inline into single caller ...
5099 (bitmap_value_replace_in_set): ... here and simplify.
5100 (dependent_clean): Merge into ...
5101 (clean): ... this using an overload. Adjust.
5102 (prune_clobbered_mems): Adjust.
5103 (compute_antic_aux): Likewise.
5104 (compute_partial_antic_aux): Likewise.
5106 2017-10-23 Richard Biener <rguenther@suse.de>
5108 PR tree-optimization/82129
5110 2017-08-01 Richard Biener <rguenther@suse.de>
5112 PR tree-optimization/81181
5113 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5114 (compute_antic): ... end of iteration here.
5116 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5118 * target.def (starting_frame_offset): New hook.
5119 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5120 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5121 * doc/tm.texi.in: Regenerate.
5122 * hooks.h (hook_hwi_void_0): Declare.
5123 * hooks.c (hook_hwi_void_0): New function.
5124 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5125 STARTING_FRAME_OFFSET.
5126 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5127 * reload1.c (reload): Likewise.
5128 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5129 instead of STARTING_FRAME_OFFSET.
5130 * function.c (try_fit_stack_local): Likewise.
5131 (assign_stack_local_1): Likewise
5132 (instantiate_virtual_regs): Likewise.
5133 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5134 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
5135 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
5136 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
5137 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
5138 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
5139 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
5140 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
5141 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
5142 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
5143 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
5144 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
5145 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
5146 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
5147 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
5148 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
5149 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
5150 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
5151 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
5152 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
5153 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
5154 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
5155 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
5156 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
5157 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
5158 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
5159 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
5160 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
5161 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
5162 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
5163 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
5164 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
5165 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
5166 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
5167 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
5168 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
5169 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
5170 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
5171 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
5172 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
5173 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
5174 * config/avr/avr.c (avr_starting_frame_offset): Make static and
5175 return a HOST_WIDE_INT.
5176 (avr_builtin_setjmp_frame_value): Use it instead of
5177 STARTING_FRAME_OFFSET.
5178 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5179 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
5180 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
5182 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5183 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
5184 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
5185 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5186 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
5187 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
5188 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5189 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
5190 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
5191 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5192 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
5193 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
5195 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5196 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
5197 * config/mips/mips.c (mips_compute_frame_info): Refer to
5198 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5199 (mips_starting_frame_offset): New function.
5200 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5201 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
5202 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
5203 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
5204 and return a HOST_WIDE_INT.
5205 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5206 (mmix_initial_elimination_offset): Refer to
5207 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5208 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
5209 * config/pa/pa.c (pa_starting_frame_offset): New function.
5210 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
5211 (pa_expand_prologue): Likewise.
5212 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5213 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
5214 !FRAME_GROWS_DOWNWARD handling to...
5215 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5216 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
5217 !FRAME_GROWS_DOWNWARD handling to...
5218 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5219 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
5220 !FRAME_GROWS_DOWNWARD handling to...
5221 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5222 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
5224 (rs6000_starting_frame_offset): New function.
5225 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
5226 !FRAME_GROWS_DOWNWARD handling to...
5227 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5228 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
5229 !FRAME_GROWS_DOWNWARD handling to...
5230 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5231 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
5232 !FRAME_GROWS_DOWNWARD handling to...
5233 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5234 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5235 (rs6000_starting_frame_offset): New function.
5236 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5237 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5238 * config/vax/vax.c (vax_starting_frame_offset): New function.
5239 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5240 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5241 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5242 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5243 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5244 * system.h (STARTING_FRAME_OFFSET): Poison.
5246 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5248 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5249 SCALAR_TYPE_MODE instead of TYPE_MODE.
5251 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5252 Alan Hayward <alan.hayward@arm.com>
5253 David Sherwood <david.sherwood@arm.com>
5255 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5257 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5258 Alan Hayward <alan.hayward@arm.com>
5259 David Sherwood <david.sherwood@arm.com>
5261 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5263 2017-10-23 Richard Biener <rguenther@suse.de>
5265 PR tree-optimization/82129
5266 * tree-ssa-pre.c (bitmap_set_and): Remove.
5267 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5268 canonicalizing expressions in the set to those with lowest
5269 ID rather than taking that from the first edge.
5271 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5273 * combine.c (rtx_equal_for_field_assignment_p): Use
5274 byte_lowpart_offset.
5276 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5277 Alan Hayward <alan.hayward@arm.com>
5278 David Sherwood <david.sherwood@arm.com>
5280 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5281 to a SUBREG_PROMOTED_VAR.
5283 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5284 Alan Hayward <alan.hayward@arm.com>
5285 David Sherwood <david.sherwood@arm.com>
5287 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5288 (expand_debug_source_expr): Likewise.
5289 * combine.c (combine_simplify_rtx): Likewise.
5290 * cse.c (fold_rtx): Likewise.
5291 * optabs.c (expand_float): Likewise.
5292 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5293 (simplify_binary_operation_1): Likewise.
5295 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5296 Alan Hayward <alan.hayward@arm.com>
5297 David Sherwood <david.sherwood@arm.com>
5299 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5300 (record_promoted_value): Likewise.
5301 * expr.c (expand_expr_real_2): Likewise.
5302 * ree.c (update_reg_equal_equiv_notes): Likewise.
5303 (combine_set_extension): Likewise.
5304 * rtlanal.c (low_bitmask_len): Likewise.
5305 * simplify-rtx.c (neg_const_int): Likewise.
5306 (simplify_binary_operation_1): Likewise.
5308 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5309 Alan Hayward <alan.hayward@arm.com>
5310 David Sherwood <david.sherwood@arm.com>
5312 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5313 * regcprop.c (maybe_mode_change): Likewise.
5314 * reload1.c (alter_reg): Likewise.
5316 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5318 * inchash.h (inchash::hash::add_wide_int): New function.
5319 * lto-streamer-out.c (hash_tree): Use it.
5321 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5323 * inchash.h (inchash::hash::add_wide_int): Rename to...
5324 (inchash::hash::add_hwi): ...this.
5325 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5326 (polymorphic_call_target_hasher::hash): Likewise.
5327 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5328 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5329 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5330 * lto-streamer-out.c (hash_tree): Likewise.
5331 * optc-save-gen.awk: Likewise.
5332 * tree.c (add_expr): Likewise.
5334 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5337 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5338 for ordered inequality comparisons even with TARGET_IEEE_FP.
5340 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5343 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5344 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5345 Expand with cmp<dwi>_doubleword.
5347 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5349 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5350 List CET intrinsics.
5351 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5352 specific to -fcf-protection option.
5354 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5356 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5357 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5358 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5359 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5360 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5361 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5362 (extra_objs): Add cet.o for Linux/x86 targets.
5363 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5364 * config/i386/cet.c: New file.
5365 * config/i386/cetintrin.h: Likewise.
5366 * config/i386/t-cet: Likewise.
5367 * config/i386/cpuid.h (bit_SHSTK): New.
5368 (bit_IBT): Likewise.
5369 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5370 pass IBT and SHSTK bits.
5371 * config/i386/i386-builtin-types.def
5372 (VOID_FTYPE_UNSIGNED_PVOID): New.
5373 (VOID_FTYPE_UINT64_PVOID): Likewise.
5374 * config/i386/i386-builtin.def: Add CET intrinsics.
5375 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5376 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5377 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5378 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5380 * config/i386/i386.c (rest_of_insert_endbranch): New.
5381 (pass_data_insert_endbranch): Likewise.
5382 (pass_insert_endbranch): Likewise.
5383 (make_pass_insert_endbranch): Likewise.
5384 (ix86_notrack_prefixed_insn_p): Likewise.
5385 (ix86_target_string): Add -mibt, -mshstk flags.
5386 (ix86_option_override_internal): Add flag_cf_protection
5388 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5389 (ix86_print_operand): Add 'notrack' prefix output.
5390 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5391 (ix86_expand_builtin): Expand CET intrinsics.
5392 (x86_output_mi_thunk): Add 'endbranch' instruction.
5393 * config/i386/i386.h (TARGET_IBT): New.
5394 (TARGET_IBT_P): Likewise.
5395 (TARGET_SHSTK): Likewise.
5396 (TARGET_SHSTK_P): Likewise.
5397 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5398 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5399 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5400 (builtin_setjmp_setup): New pattern.
5401 (builtin_longjmp): Likewise.
5402 (rdssp<mode>): Likewise.
5403 (incssp<mode>): Likewise.
5404 (saveprevssp): Likewise.
5405 (rstorssp): Likewise.
5406 (wrss<mode>): Likewise.
5407 (wruss<mode>): Likewise.
5408 (setssbsy): Likewise.
5409 (clrssbsy): Likewise.
5410 (nop_endbr): Likewise.
5411 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5413 * config/i386/immintrin.h: Include <cetintrin.h>.
5414 * config/i386/linux-common.h
5415 (file_end_indicate_exec_stack_and_cet): New prototype.
5416 (TARGET_ASM_FILE_END): New.
5418 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5420 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5421 latencies instead of having separate table; make difference between
5422 integer and float costs.
5423 * i386.h (processor_costs): Remove scalar_stmt_cost,
5424 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5425 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5427 * x86-tune-costs.h: Remove entries which has been removed in
5428 procesor_costs from all tables; make cond_taken_branch_cost
5429 and cond_not_taken_branch_cost COST_N_INSNS based.
5431 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5433 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5435 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5437 * config/i386/i386.md (isa): Remove fma_avx512f.
5438 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5439 <avx512>_fmadd_<mode>_mask3<round_name>,
5440 <avx512>_fmsub_<mode>_mask<round_name>,
5441 <avx512>_fmsub_<mode>_mask3<round_name>,
5442 <avx512>_fnmadd_<mode>_mask<round_name>,
5443 <avx512>_fnmadd_<mode>_mask3<round_name>,
5444 <avx512>_fnmsub_<mode>_mask<round_name>,
5445 <avx512>_fnmsub_<mode>_mask3<round_name>,
5446 <avx512>_fmaddsub_<mode>_mask<round_name>,
5447 <avx512>_fmaddsub_<mode>_mask3<round_name>,
5448 <avx512>_fmsubadd_<mode>_mask<round_name>,
5449 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
5450 (*vec_widen_umult_even_v16si<mask_name>,
5451 *vec_widen_smult_even_v16si<mask_name>): Likewise.
5452 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
5454 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5456 * extend.texi: Add 'nocf_check' documentation.
5457 * gimple.texi: Add second parameter to
5458 gimple_build_call_from_tree.
5459 * invoke.texi: Add -fcf-protection documentation.
5460 * rtl.texi: Add REG_CALL_NOTRACK documenation.
5462 2017-10-20 Richard Biener <rguenther@suse.de>
5464 PR tree-optimization/82473
5465 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5466 the largest input type.
5468 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5470 * c-attribs.c (handle_nocf_check_attribute): New function.
5471 (c_common_attribute_table): Add 'nocf_check' handling.
5472 * gimple-parser.c: Add second argument NULL to
5473 gimple_build_call_from_tree.
5474 * attrib.c (comp_type_attributes): Check nocf_check attribute.
5475 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
5477 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
5478 * common.opt: Add fcf-protection flag.
5479 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
5480 * flag-types.h: Add enum cf_protection_level.
5481 * gimple.c (gimple_build_call_from_tree): Add second parameter.
5482 Add 'nocf_check' attribute propagation to gimple call.
5483 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
5484 (gimple_build_call_from_tree): Update prototype.
5485 (gimple_call_nocf_check_p): New function.
5486 (gimple_call_set_nocf_check): Likewise.
5487 * gimplify.c: Add second argument to gimple_build_call_from_tree.
5488 * ipa-icf.c: Add nocf_check attribute in statement hash.
5489 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
5490 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
5491 * toplev.c (process_options): Add flag_cf_protection handling.
5493 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5495 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
5497 2017-10-20 Richard Biener <rguenther@suse.de>
5499 PR tree-optimization/82603
5500 * tree-if-conv.c (predicate_mem_writes): Make sure to only
5501 remove false predicated stores.
5503 2017-10-20 Richard Biener <rguenther@suse.de>
5505 * graphite-isl-ast-to-gimple.c
5506 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
5507 Remove return value and simplify, dump copied stmt after lhs
5509 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
5510 Reduce dump verbosity.
5511 (gsi_insert_earliest): Likewise.
5512 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
5513 * graphite.c (print_global_statistics): Adjust dumping.
5514 (print_graphite_scop_statistics): Likewise.
5515 (print_graphite_statistics): Do not dump loops here.
5516 (graphite_transform_loops): But here.
5518 2017-10-20 Nicolas Roche <roche@adacore.com>
5520 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
5521 * configure: Regenerate.
5523 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5526 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
5527 functions when optimizing replace GIMPLE_RETURN stmts with
5528 calls to __builtin_unreachable ().
5531 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
5532 for -fsanitize=thread link of executables.
5533 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
5534 link of executables.
5537 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
5539 (<shift_insn><mode>3<mask_name>): Change the last of the 3
5540 define_insns for logical vector shifts to use VI248_AVX512BW
5541 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
5542 condition, useless isa and prefix attributes. Change the first
5543 2 of these define_insns to ...
5544 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
5545 define_insn for avx512vl.
5546 (<shift_insn><mode>3): ... and this, new define_insn without
5547 masking for non-avx512vl.
5550 * config/i386/sse.md (*andnot<mode>3,
5551 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
5552 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
5553 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
5554 not applied use empty suffix even for TARGET_AVX512VL.
5555 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
5556 is applied, supply evex,evex or evex,evex,evex instead of just
5559 2017-10-20 Julia Koval <julia.koval@intel.com>
5561 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
5562 (OPTION_MASK_ISA_GFNI_UNSET): New.
5563 (ix86_handle_option): Handle OPT_mgfni.
5564 * config/i386/cpuid.h (bit_GFNI): New.
5565 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
5566 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
5567 * config/i386/i386.c (ix86_target_string): Add -mgfni.
5568 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
5569 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
5570 * config/i386/i386.opt: Add mgfni.
5572 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5574 * config/msp430/msp430.c (msp430_option_override): Disable
5575 -fdelete-null-pointer-checks.
5576 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5578 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5580 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5581 of x87 and SSE instructions.
5583 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5585 * asan.c (create_cond_insert_point): Do not update edge count.
5586 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5587 (afdo_propagate_circuit): Likewise.
5588 (afdo_calculate_branch_prob): Likewise.
5589 (afdo_annotate_cfg): Likewise.
5590 * basic-block.h (struct edge_def): Remove count.
5591 (edge_def::count): New accessor.
5592 * bb-reorder.c (rotate_loop): Update.
5593 (find_traces_1_round): Update.
5594 (connect_traces): Update.
5595 (sanitize_hot_paths): Update.
5596 * cfg.c (unchecked_make_edge): Update.
5597 (make_single_succ_edge): Update.
5598 (check_bb_profile): Update.
5599 (dump_edge_info): Update.
5600 (update_bb_profile_for_threading): Update.
5601 (scale_bbs_frequencies_int): Update.
5602 (scale_bbs_frequencies_gcov_type): Update.
5603 (scale_bbs_frequencies_profile_count): Update.
5604 (scale_bbs_frequencies): Update.
5605 * cfganal.c (connect_infinite_loops_to_exit): Update.
5606 * cfgbuild.c (compute_outgoing_frequencies): Update.
5607 (find_many_sub_basic_blocks): Update.
5608 * cfgcleanup.c (try_forward_edges): Update.
5609 (try_crossjump_to_edge): Update
5610 * cfgexpand.c (expand_gimple_cond): Update
5611 (expand_gimple_tailcall): Update
5612 (construct_exit_block): Update
5613 * cfghooks.c (verify_flow_info): Update
5614 (redirect_edge_succ_nodup): Update
5615 (split_edge): Update
5616 (make_forwarder_block): Update
5617 (duplicate_block): Update
5618 (account_profile_record): Update
5619 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5620 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5621 * cfgloopmanip.c (scale_loop_profile): Update.
5623 (lv_adjust_loop_entry_edge): Update.
5624 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5625 (force_nonfallthru_and_redirect): Update.
5626 (purge_dead_edges): Update.
5627 (rtl_flow_call_edges_add): Update.
5628 * cgraphunit.c (init_lowered_empty_function): Update.
5629 (cgraph_node::expand_thunk): Update.
5630 * gimple-pretty-print.c (dump_probability): Update.
5631 (dump_edge_probability): Update.
5632 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5633 * haifa-sched.c (sched_create_recovery_edges): Update.
5634 * hsa-gen.c (convert_switch_statements): Update.
5635 * ifcvt.c (dead_or_predicable): Update.
5636 * ipa-inline-transform.c (inline_transform): Update.
5637 * ipa-split.c (split_function): Update.
5638 * ipa-utils.c (ipa_merge_profiles): Update.
5639 * loop-doloop.c (add_test): Update.
5640 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5641 * lto-streamer-in.c (input_cfg): Update.
5642 (input_function): Update.
5643 * lto-streamer-out.c (output_cfg): Update.
5644 * modulo-sched.c (sms_schedule): Update.
5645 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5646 * predict.c (maybe_hot_edge_p): Update.
5647 (unlikely_executed_edge_p): Update.
5648 (probably_never_executed_edge_p): Update.
5649 (dump_prediction): Update.
5650 (drop_profile): Update.
5651 (propagate_unlikely_bbs_forward): Update.
5652 (determine_unlikely_bbs): Update.
5653 (force_edge_cold): Update.
5654 * profile.c (compute_branch_probabilities): Update.
5655 * reg-stack.c (better_edge): Update.
5656 * shrink-wrap.c (handle_simple_exit): Update.
5657 * tracer.c (better_p): Update.
5658 * trans-mem.c (expand_transaction): Update.
5659 (split_bb_make_tm_edge): Update.
5660 * tree-call-cdce.c: Update.
5661 * tree-cfg.c (gimple_find_sub_bbs): Update.
5662 (gimple_split_edge): Update.
5663 (gimple_duplicate_sese_region): Update.
5664 (gimple_duplicate_sese_tail): Update.
5665 (gimple_flow_call_edges_add): Update.
5666 (insert_cond_bb): Update.
5667 (execute_fixup_cfg): Update.
5668 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5669 * tree-complex.c (expand_complex_div_wide): Update.
5670 * tree-eh.c (lower_resx): Update.
5671 (unsplit_eh): Update.
5672 (cleanup_empty_eh_move_lp): Update.
5673 * tree-inline.c (copy_edges_for_bb): Update.
5674 (freqs_to_counts): Update.
5675 (copy_cfg_body): Update.
5676 * tree-ssa-dce.c (remove_dead_stmt): Update.
5677 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5678 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5679 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5680 (unloop_loops): Update.
5681 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5682 * tree-ssa-loop-split.c (connect_loops): Update.
5683 (split_loop): Update.
5684 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5685 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5686 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5687 * tree-ssa-reassoc.c (branch_fixup): Update.
5688 * tree-ssa-tail-merge.c (replace_block_by): Update.
5689 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5690 (compute_path_counts): Update.
5691 (update_profile): Update.
5692 (recompute_probabilities): Update.
5693 (update_joiner_offpath_counts): Update.
5694 (estimated_freqs_path): Update.
5695 (freqs_to_counts_path): Update.
5696 (clear_counts_path): Update.
5697 (ssa_fix_duplicate_block_edges): Update.
5698 (duplicate_thread_path): Update.
5699 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5700 (case_bit_test_cmp): Update.
5701 (collect_switch_conv_info): Update.
5702 (gen_inbound_check): Update.
5703 (do_jump_if_equal): Update.
5704 (emit_cmp_and_jump_insns): Update.
5705 * tree-tailcall.c (decrease_profile): Update.
5706 (eliminate_tail_call): Update.
5707 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5708 (vect_do_peeling): Update.
5709 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5710 * ubsan.c (ubsan_expand_null_ifn): Update.
5711 (ubsan_expand_ptr_ifn): Update.
5712 * value-prof.c (gimple_divmod_fixed_value): Update.
5713 (gimple_mod_pow2): Update.
5714 (gimple_mod_subtract): Update.
5715 (gimple_ic): Update.
5716 (gimple_stringop_fixed_value): Update.
5718 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5721 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5723 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5725 PR rtl-optimization/82395
5726 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5727 based on non_spilled_static_chain_regno_p.
5729 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5731 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5732 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5733 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5735 2017-10-19 Martin Sebor <msebor@redhat.com>
5737 PR tree-optimization/82596
5738 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5740 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5742 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5743 (get_mem_refs_of_builtin_call): Likewise.
5744 * builtins.c (expand_builtin_apply): Adjust call to
5745 allocate_dynamic_stack_space.
5746 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5747 the third argument to allocate_dynamic_stack_space, otherwise -1.
5748 (expand_builtin): Deal with all alloca variants.
5749 (is_inexpensive_builtin): Likewise.
5750 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5751 * calls.c (special_function_p): Deal with all alloca variants.
5752 (initialize_argument_information): Adjust call to
5753 allocate_dynamic_stack_space.
5754 (expand_call): Likewise.
5755 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5756 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5757 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5758 use it for the stack usage computation.
5759 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5760 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5761 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5762 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5763 (in_loop_p): Remove first argument and useless check.
5764 (pass_walloca::execute): Remove useless test and adjust call to above.
5765 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5766 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5767 (gimplify_call_expr): Deal with all alloca variants.
5768 * hsa-gen.c (gen_hsa_alloca): Likewise.
5769 (gen_hsa_insns_for_call): Likewise.
5770 * ipa-pure-const.c (special_builtin_state): Likewise.
5771 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5772 * tree-object-size.c (alloc_object_size): Likewise.
5773 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5774 (call_may_clobber_ref_p_1): Likewise.
5775 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5776 (ccp_fold_stmt): Likewise.
5777 (optimize_stack_restore): Likewise.
5778 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5779 (mark_all_reaching_defs_necessary_1): Likewise.
5780 (propagate_necessity): Likewise.
5781 (eliminate_unnecessary_stmts): Likewise.
5782 * tree.c (build_common_builtin_nodes): Build
5783 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5784 (build_alloca_call_expr): New function.
5785 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5786 (CASE_BUILT_IN_ALLOCA): Likewise.
5787 (build_alloca_call_expr): Declare.
5788 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5790 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5793 * dwarf2out.c (new_die_raw): New static inline function.
5794 (new_die): Use it to create the DIE.
5795 (add_AT_external_die_ref): Likewise.
5796 (clone_die): Likewise.
5797 (clone_as_declaration): Likewise.
5798 (dwarf2out_vms_debug_main_pointer): Likewise.
5799 (base_type_die): Likewise. Remove early return for corner cases.
5800 Do not call add_pubtype on the DIE here.
5801 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5802 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5803 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5804 native order exists for base types, attach the DIE manually and call
5805 add_pubtype on it. Do not equate a reverse order DIE to the type.
5807 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5809 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5810 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5812 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5814 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5816 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5817 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5818 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5819 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5820 * opts.c (sanitizer_opts): Add builtin.
5821 * ubsan.c (instrument_builtin): New function.
5822 (pass_ubsan::execute): Call it.
5823 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5824 * doc/invoke.texi: Document -fsanitize=builtin.
5826 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5827 builtins, store max (log2 (align), 0) into uchar field instead of
5828 align into uptr field.
5829 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5830 store uchar 0 field instead of uptr 0 field.
5831 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5832 instead of passing one address of struct with 2 locations pass
5833 two addresses of structs with 1 location each.
5834 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5835 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5836 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5837 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5838 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5839 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5840 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5841 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5843 2017-10-19 Martin Liska <mliska@suse.cz>
5846 * file-find.c (remove_prefix): Remove.
5847 * file-find.h (remove_prefix): Likewise.
5848 * gcc-ar.c: Remove smartness of lookup.
5850 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5852 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5853 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5854 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5856 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5859 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5860 (setcc + and to xor + setcc): New peephole2.
5862 2017-10-19 Tom de Vries <tom@codesourcery.com>
5864 * doc/sourcebuild.texi (Test Directives, Variants of
5865 dg-require-support): Add dg-require-stack-size.
5867 2017-10-19 Martin Liska <mliska@suse.cz>
5870 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5871 that have a large alignment.
5872 (gimplify_target_expr): Likewise.
5874 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5876 PR rtl-optimization/82602
5877 * ira.c (rtx_moveable_p): Return false for volatile asm.
5879 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5882 * config/i386/i386-modes.def (CCGZ): New CC mode.
5883 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5884 * config/i386/predicates.md (ix86_comparison_operator):
5886 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5887 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5888 with double-word subtraction.
5889 (put_condition_code): Handle CCGZmode.
5891 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5893 * wide-int.cc (debug (const wide_int &)): New.
5894 (debug (const wide_int *)): New.
5895 (debug (const widest_int &)): New.
5896 (debug (const widest_int *)): New.
5898 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5901 * lra-constraints.c (curr_insn_transform): Use non-input operand
5902 instead of output one for matched reload.
5904 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5906 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5907 (tree-ssa-loop-ivopts.h): New header file.
5908 (struct builtin_info): New fields.
5909 (classify_builtin_1): Compute and record base and offset parts for
5910 memset builtin partition by calling strip_offset.
5911 (offset_cmp, fuse_memset_builtins): New functions.
5912 (finalize_partitions): Fuse adjacent memset partitions by calling
5914 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5915 Expose the interface.
5916 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5918 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5920 PR tree-optimization/82574
5921 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5922 that data reference must be executed exactly once per iteration
5923 against the outermost loop in nest.
5924 (classify_partition): Update call to above function.
5926 2017-10-18 Richard Biener <rguenther@suse.de>
5928 PR tree-optimization/82591
5929 * graphite.c (graphite_transform_loops): Move code gen message
5931 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5932 Here. Handle scop_to_isl_ast failing.
5933 (scop_to_isl_ast): Limit the number of ISL operations.
5935 2017-10-18 Richard Biener <rguenther@suse.de>
5937 * graphite-isl-ast-to-gimple.c
5938 (translate_isl_ast_to_gimple::set_rename): Simplify.
5939 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5940 (graphite_copy_stmts_from_block): ... here.
5941 (copy_bb_and_scalar_dependences): Simplify.
5942 (add_parameters_to_ivs_params): Canonicalize.
5943 (generate_entry_out_of_ssa_copies): Simplify.
5944 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5945 by passing in ISL dimension.
5946 (parameter_index_in_region_1): Rename to ...
5947 (parameter_index_in_region): ... this.
5948 (extract_affine): Adjust assert, pass down parameter index.
5949 (add_param_constraints): Use range-info when available.
5950 (build_scop_context): Adjust.
5951 * sese.c (new_sese_info): Adjust.
5952 (free_sese_info): Likewise.
5953 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5954 Remove unused typedefs.
5955 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5957 2017-10-18 Martin Liska <mliska@suse.cz>
5959 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5961 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5963 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5964 (s390_sched_init): Do not reset s390_sched_state if we entered the
5965 current basic block via a fallthru edge and all others are unlikely.
5967 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5969 * config/s390/s390.c (NUM_SIDES): New variable.
5970 (LONGRUNNING_THRESHOLD): New variable.
5971 (LATENCY_FACTOR): New variable.
5972 (s390_sched_score): Decrease score for long-running instructions on
5974 (s390_sched_variable_issue): Perform bookkeeping for long-running
5977 2017-10-18 Richard Biener <rguenther@suse.de>
5979 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5980 Simplify with removal of the parameter rename map.
5981 (set_rename): Likewise.
5982 (should_copy_to_new_region): Likewise.
5983 (graphite_copy_stmts_from_block): Likewise.
5984 (copy_bb_and_scalar_dependences): Remove initialization of
5985 unused copied_bb_map.
5987 (copy_internal_parameters): Likewise.
5988 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5989 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5990 Use INTEGRAL_TYPE_P.
5991 (parameter_index_in_region_1): Rename to ...
5992 (assign_parameter_index_in_region): ... this. Assert we have
5993 a parameter we handle.
5994 (scan_tree_for_params): Adjust.
5995 * sese.h (parameter_rename_map_t): Remove.
5996 (struct sese_info_t): Remove unused parameter_rename_map and
5997 copied_bb_map members.
5998 * sese.c (new_sese_info): Adjust.
5999 (free_sese_info): Likewise.
6001 2017-10-18 Martin Liska <mliska@suse.cz>
6004 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6005 on an abnormal edge.
6007 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6009 * doc/invoke.texi (ffunction-sections and fdata-sections):
6012 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6014 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6015 the use statement can throw internally.
6017 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6019 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6020 any RTX present on the RHS of a SET.
6021 * compare-elim.c (try_eliminate_compare): Restore comment.
6023 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6025 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6026 may be also called on tcc_type nodes.
6027 * langhooks.c (lhd_tree_size): Likewise.
6029 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6031 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6032 format_warning_at_substring.
6033 (maybe_warn): Convert source_range * param to a location_t. Pass
6034 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6035 (format_directive): Remove code to extract source_ranges and
6036 source_range * in favor of just a location_t.
6037 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6039 * substring-locations.c (format_warning_va): Convert
6040 source_range * param to a location_t.
6041 (format_warning_at_substring): Likewise.
6042 * substring-locations.h (format_warning_va): Likewise.
6043 (format_warning_at_substring): Likewise.
6045 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6047 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6049 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6050 and scatter/gather ops.
6052 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6053 vec_gather_load and vec_scatter_store.
6054 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6055 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6056 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6057 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6058 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6059 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6061 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6063 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6064 Detect FTST instruction and handle its register pops. Only pop
6065 second operand if can_pop_second_op is true.
6066 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6067 set can_pop_second_op to false in the compare_for_stack_reg call.
6069 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6070 output_fp_compare for stack register operands.
6071 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6072 instructions here. Do not emit stack register pops here. Assert
6073 that FCOMPP pops next to top stack register. Rewrite function.
6075 2017-10-17 Nathan Sidwell <nathan@acm.org>
6078 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6079 use DECL_ASSEMBLER_NAME_RAW.
6082 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6085 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6086 Wilco Dijkstra <wilco.dijkstra@arm.com>
6088 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6089 converstion to Pmode from the buf_addr.
6091 2017-10-17 Richard Biener <rguenther@suse.de>
6093 * graphite-dependences.c (scop_get_reads_and_writes): Change
6094 output parameters to references.
6096 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6098 PR 71026/tree-optimization
6099 * fold-const.c (distribute_real_division): Removed.
6100 (fold_binary_loc): Remove calls to distribute_real_divison.
6102 2017-10-17 Richard Biener <rguenther@suse.de>
6104 * graphite-scop-detection.c
6105 (scop_detection::stmt_has_simple_data_refs_p): Always use
6106 the full nest as region.
6107 (try_generate_gimple_bb): Likewise.
6108 * sese.c (scalar_evolution_in_region): Simplify now that
6109 SCEV can handle instantiation in regions.
6110 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6111 in the non-loop part of a function if requested.
6113 2017-10-17 Richard Biener <rguenther@suse.de>
6115 PR tree-optimization/82563
6116 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6118 (graphite_regenerate_ast_isl): Call it.
6119 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6121 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6123 PR tree-optimization/82549
6124 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6125 Formatting fixes. Instead of calling make_bit_field_ref with negative
6128 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6130 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6131 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6132 _mm_maskz_reduce_ss): New.
6133 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6134 __builtin_ia32_reducess_mask): Ditto..
6135 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
6136 * config/i386/sse.md (reduces<mode>): Renamed to ...
6137 (reduces<mode><mask_scalar_name>): ... this.
6138 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
6140 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
6141 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
6143 2017-10-16 David Malcolm <dmalcolm@redhat.com>
6145 * Makefile.in (OBJS): Add unique-ptr-tests.o.
6146 * selftest-run-tests.c (selftest::run_tests): Call
6147 selftest::unique_ptr_tests_cc_tests.
6148 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
6149 * unique-ptr-tests.cc: New file.
6151 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
6154 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6156 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6157 (make_hard_regno_born, make_hard_regno_dead): Update
6158 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
6160 2017-10-16 Jeff Law <law@redhat.com>
6162 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6164 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
6166 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
6168 2017-10-16 Olivier Hainque <hainque@adacore.com>
6170 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
6171 with_cpu if we were configured for an e500v2 target cpu name.
6173 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
6175 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
6176 * doc/invoke.texi: Document +nodsp as a valid extension for
6179 2017-10-16 Martin Liska <mliska@suse.cz>
6181 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
6182 (test_set_range): Likewise.
6183 (test_range_functions): Rename to ...
6184 (test_bit_in_range): ... this.
6185 (sbitmap_c_tests): Add new test.
6187 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6189 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
6191 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
6192 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
6194 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6196 * config/aarch64/aarch64-builtins.c
6197 (aarch64_types_quadopu_lane_qualifiers): New.
6198 (TYPES_QUADOPU_LANE): New.
6199 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
6200 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
6201 (aarch64_<sur>dot_laneq<vsi2qi>): New.
6202 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
6203 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
6204 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
6205 (Vdottype, DOTPROD): New.
6206 (sur): Add SDOT and UDOT.
6208 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6210 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
6211 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
6212 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
6214 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
6215 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
6216 Enable TARGET_DOTPROD.
6217 (cortex-a75.cortex-a55): Likewise.
6218 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
6220 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6222 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
6223 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
6225 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
6227 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
6228 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
6229 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
6230 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
6231 * config/arm/types.md (neon_dot, neon_dot_q): New.
6232 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6234 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6236 * config/arm/arm.h (TARGET_DOTPROD): New.
6237 * config/arm/arm.c (arm_arch_dotprod): New.
6238 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6239 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6240 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6241 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6242 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6243 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6244 * doc/invoke.texi (armv8.2-a): Document dotprod
6246 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6248 * i386.c (ix86_vec_cost): New function.
6249 (ix86_rtx_costs): Handle vector operations better.
6250 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6251 * x86-tune-costs.h: Add new costs to all tables.
6253 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6255 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6257 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6258 divsd, sqrtss and sqrtsd
6259 * x86-tune-costs.h: Add new entries to all costs.
6260 (znver1_cost): Fix to match real instruction latencies.
6262 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6263 Michael Collison <michael.collison@arm.com>
6265 * compare-elim.c: Include emit-rtl.h.
6266 (can_merge_compare_into_arith): New function.
6267 (try_validate_parallel): Likewise.
6268 (try_merge_compare): Likewise.
6269 (try_eliminate_compare): Call the above when no previous clobber
6271 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6274 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6278 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6279 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6283 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6284 to be any operand_equal_p operands. For & (B - 1) require
6285 B to be power of 2. Recognize
6286 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6288 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6291 * optabs.c (expand_memory_blockage): Fix call of
6292 targetm.have_memory_blockage.
6294 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6297 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6298 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6299 objects to extra_objs instead of overwriting it.
6301 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6303 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6304 Use any_fp_register_operand as operand[3] predicate. Simplify
6305 equality test for operands[2] and operands[4] memory location.
6306 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6307 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6308 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6309 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6310 any_fp_register_operand as operand[1] predicate. Simplify
6311 equality test for operands[0] and operands[3] memory location.
6312 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6313 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6314 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6316 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6318 * target-insns.def: Add memory_blockage.
6319 * optabs.c (expand_memory_blockage): New function.
6320 (expand_asm_memory_barrier): Rename ...
6321 (expand_asm_memory_blockage): ... to this.
6322 (expand_mem_thread_fence): Call expand_memory_blockage
6323 instead of expand_asm_memory_barrier.
6324 (expand_mem_singnal_fence): Ditto.
6325 (expand_atomic_load): Ditto.
6326 (expand_atomic_store): Ditto.
6327 * doc/md.texi (Standard Pattern Names For Generation):
6328 Document memory_blockage instruction pattern.
6330 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6332 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6333 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6334 * config/rl78/rl78.md: New define_expand "adddi3".
6336 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6338 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6341 2017-10-13 Jeff Law <law@redhat.com>
6343 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6345 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6348 * internal-fn.c (expand_mul_overflow): If both operands have
6349 the same highpart of -1 or 0 and the topmost bit of lowpart
6350 is different, overflow is if res <= 0 rather than res < 0.
6352 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6354 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6355 TARGET_P9_VECTOR code for unaligned_load case.
6357 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6359 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6361 2017-10-13 Nathan Sidwell <nathan@acm.org>
6363 * tree-core.h (tree_contains_struct): Make bool.
6364 * tree.c (tree_contains_struct): Likewise.
6365 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6366 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6367 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6368 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6369 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6371 2017-10-13 Richard Biener <rguenther@suse.de>
6373 * graphite-isl-ast-to-gimple.c
6374 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6375 parameters and dominance check.
6376 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6377 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6378 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6379 Do not update SSA form here or do intermediate IL verification.
6380 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6381 (graphite_initialize): Remove check on the number of loops in
6382 the function and inline into graphite_transform_loops.
6383 (graphite_finalize): Inline into graphite_transform_loops.
6384 (graphite_transform_loops): Perform SSA update and IL verification
6386 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6388 2017-10-13 Richard Biener <rguenther@suse.de>
6390 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6391 graphite_expression_type_precision): Avoid global constructor
6393 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6394 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6395 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6396 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6397 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6398 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6400 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6403 * config/i386/i386.h (ix86_red_zone_size): New.
6404 * config/i386/i386.md (push peephole2s): Replace
6405 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6407 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6408 Alan Hayward <alan.hayward@arm.com>
6409 David Sherwood <david.sherwood@arm.com>
6411 * combine.c (can_change_dest_mode): Reject changes in
6412 REGMODE_NATURAL_SIZE.
6414 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6415 Alan Hayward <alan.hayward@arm.com>
6416 David Sherwood <david.sherwood@arm.com>
6418 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6419 (expand_debug_source_expr): Likewise.
6420 * combine.c (combine_simplify_rtx): Likewise.
6421 * cse.c (fold_rtx): Likewise.
6422 * fwprop.c (canonicalize_address): Likewise.
6423 * targhooks.c (default_shift_truncation_mask): Likewise.
6425 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6426 Alan Hayward <alan.hayward@arm.com>
6427 David Sherwood <david.sherwood@arm.com>
6429 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6430 (widened_mode): Likewise.
6431 (expand_unop): Likewise.
6432 * ree.c (transform_ifelse): Likewise.
6433 (merge_def_and_ext): Likewise.
6434 (combine_reaching_defs): Likewise.
6435 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6437 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6438 Alan Hayward <alan.hayward@arm.com>
6439 David Sherwood <david.sherwood@arm.com>
6441 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
6442 * combine.c (gen_lowpart_for_combine): Likewise.
6443 * dwarf2out.c (rtl_for_decl_location): Likewise.
6444 * final.c (alter_subreg): Likewise.
6445 * rtlhooks.c (gen_lowpart_general): Likewise.
6446 (gen_lowpart_if_possible): Likewise.
6448 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6449 Alan Hayward <alan.hayward@arm.com>
6450 David Sherwood <david.sherwood@arm.com>
6452 * calls.c (expand_call): Use subreg_lowpart_offset.
6453 * cse.c (cse_insn): Likewise.
6454 * regcprop.c (copy_value): Likewise.
6455 (copyprop_hardreg_forward_1): Likewise.
6457 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6460 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
6461 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
6462 =Q constraints to +Q and into insn condition add check
6463 that operands[0] and operands[1] are equal.
6464 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
6465 =Q constraints to +Q and into insn condition add check
6466 that operands[0] is equal to either operands[1] or operands[2].
6469 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
6470 instead of handling MINUS_EXPR twice (once for each argument),
6471 canonicalize operand order and handle just once, use rtype where
6472 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
6475 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
6476 any values of __C while still being pattern recognizable as a simple
6479 2017-10-13 Richard Biener <rguenther@suse.de>
6481 PR tree-optimization/82451
6483 2017-10-02 Richard Biener <rguenther@suse.de>
6485 PR tree-optimization/82355
6486 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6487 a mapping for the enclosing loop but avoid generating one for
6489 (copy_bb_and_scalar_dependences): Remove premature codegen
6490 error on PHIs in blocks duplicated into multiple places.
6491 * graphite-scop-detection.c
6492 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6493 in the region use it as loop and nest to analyze the DR in.
6494 (try_generate_gimple_bb): Likewise.
6495 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6496 (add_loop_constraints): For blocks in a loop not in the region
6497 create a dimension with a single iteration.
6498 * sese.h (gbb_loop_at_index): Remove assert.
6500 * cfgloop.c (loop_preheader_edge): For the loop tree root
6501 return the single successor of the entry block.
6502 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6503 Reset the SCEV hashtable and niters.
6504 * graphite-scop-detection.c
6505 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
6506 assert that we only have POLYNOMIAL_CHREC that vary in loops
6507 contained in the region.
6508 (scop_detection::graphite_can_represent_expr): Adjust.
6509 (scop_detection::stmt_has_simple_data_refs_p): For loops
6510 not in the region set loop to NULL. The nest is now the
6511 entry edge to the region.
6512 (try_generate_gimple_bb): Likewise.
6513 * sese.c (scalar_evolution_in_region): Adjust for
6514 instantiate_scev change.
6515 * tree-data-ref.h (graphite_find_data_references_in_stmt):
6516 Make nest parameter the edge into the region.
6517 (create_data_ref): Likewise.
6518 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
6519 entry edge into a region and adjust instantiate_scev calls.
6520 (create_data_ref): Likewise.
6521 (graphite_find_data_references_in_stmt): Likewise.
6522 (find_data_references_in_stmt): Pass the loop preheader edge
6523 from the nest argument.
6524 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
6525 parameter the edge into the region.
6526 (instantiate_parameters): Use the loop preheader edge as entry.
6527 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
6529 (get_instantiated_value_entry): Make instantiate_below parameter
6530 the edge into the region.
6531 (instantiate_scev_name): Likewise. Adjust dominance checks,
6532 when we cannot use loop-based instantiation instantiate by
6533 walking use-def chains.
6534 (instantiate_scev_poly): Adjust.
6535 (instantiate_scev_binary): Likewise.
6536 (instantiate_scev_convert): Likewise.
6537 (instantiate_scev_not): Likewise.
6538 (instantiate_array_ref): Remove.
6539 (instantiate_scev_3): Likewise.
6540 (instantiate_scev_2): Likewise.
6541 (instantiate_scev_1): Likewise.
6542 (instantiate_scev_r): Do not blindly handle N-operand trees.
6543 Do not instantiate array-refs. Handle all constants and invariants.
6544 (instantiate_scev): Make instantiate_below parameter
6545 the edge into the region.
6546 (resolve_mixers): Use the loop preheader edge for the region
6547 parameter to instantiate_scev_r.
6548 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
6550 2017-10-13 Richard Biener <rguenther@suse.de>
6552 PR tree-optimization/82525
6553 * graphite-isl-ast-to-gimple.c
6554 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
6556 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
6557 Fail code generation when we cannot represent the isl integer.
6558 (binary_op_to_tree): Elide modulo operations that are no-ops
6559 in the type we code generate. Remove now superfluous code
6562 2017-10-13 Richard Biener <rguenther@suse.de>
6564 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
6565 (scop_detection::harmful_loop_in_region): Remove premature
6566 IV type restriction.
6567 (scop_detection::graphite_can_represent_scev): We can handle
6568 pointer IVs just fine.
6570 2017-10-13 Alan Modra <amodra@gmail.com>
6572 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6573 "Clobbers and Scratch Registers". Add paragraph on
6574 alternative to clobbers for scratch registers and OpenBLAS
6577 2017-10-13 Alan Modra <amodra@gmail.com>
6579 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6580 example of a memory input for a string of known length. Move
6581 commentary out of table. Add a number of new examples
6582 covering array memory inputs.
6584 2017-10-12 Martin Liska <mliska@suse.cz>
6586 PR tree-optimization/82493
6587 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6588 (test_range_functions): New function.
6589 (sbitmap_c_tests): Likewise.
6590 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6591 * selftest.h (sbitmap_c_tests): New function.
6593 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6595 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6597 * config/rs6000/amo.h: Fix spacing issue.
6599 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6602 * config/i386/i386.md (*ashl<mode>3_mask_1,
6603 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6604 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6607 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6609 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6610 (profile_probability): Set max_probability
6611 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6613 * profile-count.c (profile_probability::differs_from_p): Do not
6614 rely on max_probaiblity == 10000
6616 2017-10-12 Jeff Law <law@redhat.com>
6618 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6621 2017-10-12 Martin Sebor <msebor@redhat.com>
6625 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6626 (handle_alias_pairs): Call it.
6627 * common.opt (-Wattribute-alias): New option.
6628 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6629 * doc/invoke.texi (-Wattribute-alias): Document.
6631 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6634 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6636 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6638 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6639 (make_hard_regno_born, make_hard_regno_dead): Update
6640 bb_killed_pseudos and bb_gen_pseudos.
6642 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6644 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6646 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6648 * config/alpha/alpha.c (alpha_split_conditional_move):
6649 Use std::swap instead of manually swapping.
6650 (alpha_stdarg_optimize_hook): Ditto.
6651 (alpha_canonicalize_comparison): Ditto.
6653 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6655 * tree-loop-distribution.c (struct builtin_info): New struct.
6656 (struct partition): Refactor fields into struct builtin_info.
6657 (partition_free): Free struct builtin_info.
6658 (build_size_arg_loc, build_addr_arg_loc): Delete.
6659 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6660 information from struct builtin_info.
6661 (find_single_drs): New function refactored from classify_partition.
6662 Also moved builtin validity checks to this function.
6663 (compute_access_range, alloc_builtin): New functions.
6664 (classify_builtin_st, classify_builtin_ldst): New functions.
6665 (classify_partition): Refactor code into functions find_single_drs,
6666 classify_builtin_st and classify_builtin_ldst.
6667 (distribute_loop): Don't do runtime alias check when distributing
6669 (find_seed_stmts_for_distribution): New function.
6670 (pass_loop_distribution::execute): Refactor code finding seed
6671 stmts into above function. Support distribution for the innermost
6672 two-level loop nest. Adjust dump information.
6674 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6676 * tree-loop-distribution.c: Adjust the general comment.
6677 (NUM_PARTITION_THRESHOLD): New macro.
6678 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6679 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6680 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6681 in call to build_partition_graph.
6682 (finalize_partitions): New parameter. Make loop distribution more
6683 conservative by fusing more partitions.
6684 (distribute_loop): Don't do runtime alias check in case of loop nest
6686 (find_seed_stmts_for_distribution): New function.
6687 (prepare_perfect_loop_nest): New function.
6688 (pass_loop_distribution::execute): Refactor code finding seed stmts
6689 and loop nest into above functions. Support loop nest distribution.
6690 Adjust dump information accordingly.
6692 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6694 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6695 and set PTYPE_SEQUENTIAL for merged partition.
6697 2017-10-12 Richard Biener <rguenther@suse.de>
6699 PR tree-optimization/69728
6701 2017-09-19 Richard Biener <rguenther@suse.de>
6703 PR tree-optimization/69728
6704 * graphite-sese-to-poly.c (schedule_error): New global.
6705 (add_loop_schedule): Handle empty domain by failing the
6707 (build_original_schedule): Handle schedule_error.
6709 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6710 domain by returning an unchanged schedule.
6712 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6714 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6715 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6717 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6719 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6722 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6725 * expr.c (store_field): Don't optimize away bitsize == 0 store
6726 from CALL_EXPR with addressable return type.
6728 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6730 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6731 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6732 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6733 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6734 TARGET_ISEL instead of TARGET_ISEL<sel>.
6736 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6738 * config/rs6000/rs6000.c
6739 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6741 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6743 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6745 (reg_or_cint_operand): Fix comment.
6746 (reg_or_zero_operand): New predicate.
6747 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6748 * config/rs6000/rs6000.c (output_isel): Delete.
6749 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6750 instead of reg_or_cint_operand. Output instruction directly (not via
6752 (isel_unsigned_<mode>): Ditto.
6753 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6754 gpc_reg_operand. Add an instruction alternative for this. Output
6755 instruction directly.
6756 (*isel_reversed_unsigned_<mode>): Ditto.
6758 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6760 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6761 (TARGET_CANONICALIZE_COMPARISON): Define.
6763 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6766 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6767 Check whether the dest is REG before adding REG_EQUIV note.
6769 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6772 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6774 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6775 (make_hard_regno_born, make_hard_regno_dead): Update
6776 bb_killed_pseudos and bb_gen_pseudos.
6778 2017-10-11 Nathan Sidwell <nathan@acm.org>
6780 * incpath.h (enum incpath_kind): Name enum, prefix values.
6781 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6782 * incpath.c (heads, tails): Use INC_MAX.
6783 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6784 (merge_include_chains, split_quote_chain,
6785 register_include_chains): Update incpath_kind names.
6786 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6787 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6789 (add_framework_path, darwin_register_objc_includes): Likewise.
6790 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6792 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6794 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6795 Do not use float_operator operator predicate.
6796 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6797 * config/i386/predicates.md (float_operator): Remove predicate.
6799 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6801 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6802 (*jccxf_i387): Ditto.
6803 (*jcc<mode>_i387): Ditto.
6804 (*jccu<mode>_i387): Ditto.
6805 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6806 (*jcc_*_i387 splitters): Remove.
6807 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6808 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6809 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6812 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6814 * profile-count.h (slow_safe_scale_64bit): New function.
6815 (safe_scale_64bit): New inline.
6816 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6817 * profile-count.c: Include wide-int.h
6818 (slow_safe_scale_64bit): New.
6820 2017-10-11 Nathan Sidwell <nathan@acm.org>
6822 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6823 HAS_DECL_ASSEMBLER_NAME_P.
6824 * gimple-expr.c (gimple_decl_printable_name: Check
6825 HAS_DECL_ASSEMBLER_NAME_P too.
6826 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6827 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6828 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6829 * passes.c (rest_of_decl_compilation): Check
6830 HAS_DECL_ASSEMBLER_NAME_P too.
6831 * recog.c (verify_changes): Likewise.
6832 * tree-pretty-print.c (dump_decl_name): Likewise.
6833 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6835 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6836 (SET_DECL_ASSEMBLER_NAME): Use it.
6837 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6838 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6839 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6841 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6843 * config.gcc (i386, x86_64): Add extra objects.
6844 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6845 (ix86_min_insn_size): Declare.
6846 (ix86_issue_rate): Declare.
6847 (ix86_adjust_cost): Declare.
6848 (ia32_multipass_dfa_lookahead): Declare.
6849 (ix86_macro_fusion_p): Declare.
6850 (ix86_macro_fusion_pair_p): Declare.
6851 (ix86_bd_has_dispatch): Declare.
6852 (ix86_bd_do_dispatch): Declare.
6853 (ix86_core2i7_init_hooks): Declare.
6854 (ix86_atom_sched_reorder): Declare.
6855 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6856 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6857 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6858 (rip_relative_addr_p): Rename to ...
6859 (ix86_rip_relative_addr_p): ... this one; export.
6860 (memory_address_length): Update.
6861 (ix86_issue_rate): Move to x86-tune-sched.c.
6862 (ix86_flags_dependent): Move to x86-tune-sched.c.
6863 (ix86_agi_dependent): Move to x86-tune-sched.c.
6864 (exact_dependency_1): Move to x86-tune-sched.c.
6865 (exact_store_load_dependency): Move to x86-tune-sched.c.
6866 (ix86_adjust_cost): Move to x86-tune-sched.c.
6867 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6868 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6869 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6870 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6871 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6872 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6873 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6874 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6875 (min_insn_size): Rename to ...
6876 (ix86_min_insn_size): ... this one; export.
6877 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6878 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6879 (core2i7_first_cycle_multipass_backtrack): Move to
6880 x86-tune-sched-core.c.
6881 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6882 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6883 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6884 (ix86_avoid_jump_mispredicts): Update.
6885 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6886 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6887 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6888 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6889 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6890 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6891 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6892 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6893 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6894 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6895 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6896 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6897 (BIG): Move to ix86-tune-sched-bd.c.
6898 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6899 (enum insn_path): Move to ix86-tune-sched-bd.c.
6900 (get_mem_group): Move to ix86-tune-sched-bd.c.
6901 (is_cmp): Move to ix86-tune-sched-bd.c.
6902 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6903 (is_branch): Move to ix86-tune-sched-bd.c.
6904 (is_prefetch): Move to ix86-tune-sched-bd.c.
6905 (init_window): Move to ix86-tune-sched-bd.c.
6906 (allocate_window): Move to ix86-tune-sched-bd.c.
6907 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6908 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6909 (process_end_window): Move to ix86-tune-sched-bd.c.
6910 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6911 (find_constant): Move to ix86-tune-sched-bd.c.
6912 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6913 (has_immediate): Move to ix86-tune-sched-bd.c.
6914 (get_insn_path): Move to ix86-tune-sched-bd.c.
6915 (get_insn_group): Move to ix86-tune-sched-bd.c.
6916 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6917 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6918 (add_insn_window): Move to ix86-tune-sched-bd.c.
6919 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6920 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6921 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6922 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6923 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6924 (do_dispatch): Move to ix86-tune-sched-bd.c.
6925 (has_dispatch): Move to ix86-tune-sched-bd.c.
6926 * config/i386/t-i386: Add new object files.
6927 * config/i386/x86-tune-costs.h: New file.
6928 * config/i386/x86-tune-sched-atom.c: New file.
6929 * config/i386/x86-tune-sched-bd.c: New file.
6930 * config/i386/x86-tune-sched-core.c: New file.
6931 * config/i386/x86-tune-sched.c: New file.
6933 2017-10-11 Liu Hao <lh_mouse@126.com>
6935 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6936 the generic version below instead.
6937 (should_colorize): Recognize Windows consoles as terminals
6939 * pretty-print.c [__MINGW32__] (write_all): New function.
6940 [__MINGW32__] (find_esc_head): Likewise.
6941 [__MINGW32__] (find_esc_terminator): Likewise.
6942 [__MINGW32__] (eat_esc_sequence): Likewise.
6943 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6945 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6948 2017-10-11 Richard Biener <rguenther@suse.de>
6950 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6951 Properly call analyze_scalar_evolution with the loop of the stmt.
6953 2017-10-11 Richard Biener <rguenther@suse.de>
6955 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6956 * tree-core.h (tree_base): Add chrec_var union member.
6957 * tree.h (CHREC_VAR): Remove.
6958 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6959 * tree-chrec.h (build_polynomial_chrec): Adjust.
6960 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6961 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6963 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6965 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6966 * match.pd: ... here.
6967 ((T) X == (T) Y): Relax condition.
6969 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6971 PR tree-optimization/82472
6972 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6974 (break_alias_scc_partitions): Update postorder number.
6976 2017-10-11 Martin Liska <mliska@suse.cz>
6979 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6981 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6983 2017-10-11 Martin Liska <mliska@suse.cz>
6985 * print-rtl.c (print_insn): Move declaration of idbuf
6986 to same scope as name.
6988 2017-10-11 Martin Liska <mliska@suse.cz>
6993 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6996 2017-10-11 Martin Liska <mliska@suse.cz>
6999 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7002 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7004 * config.gcc (default_gnu_indirect_function): Default to yes for
7005 arm*-*-linux* with glibc.
7007 2017-10-11 Richard Biener <rguenther@suse.de>
7009 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7010 default-defs and types we do not want to analyze.
7011 (interpret_loop_phi): Replace unreachable code with an assert.
7012 (compute_scalar_evolution_in_loop): Remove and inline ...
7013 (analyze_scalar_evolution_1): ... here, replacing condition with
7014 what makes the intent clearer. Remove handling of cases
7015 get_scalar_evolution now handles.
7017 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7019 PR rtl-optimization/81434
7020 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7021 comment for main loop. In sched_group_found if, also add checks for
7022 pass and min_cost_group.
7024 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7026 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7027 (rs6000_insn_cost): New function.
7028 * config/rs6000/rs6000.md (cost): New attribute.
7030 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7031 H.J. Lu <hongjiu.lu@intel.com>
7035 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7036 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7037 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7038 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7039 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7040 OPTION_MASK_ISA_MMX and some other bit set require both
7041 mmx and the ISAs without the mmx bit.
7042 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7043 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7044 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7045 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7046 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7047 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7048 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7049 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7050 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7051 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7052 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7053 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7054 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7055 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7056 Add OPTION_MASK_ISA_MMX.
7058 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7060 * config.gcc (armv7*-*-freebsd*): New target.
7061 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7063 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7065 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7066 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7069 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7071 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7072 a template parameter.
7073 (WIDE_INT_REF_FOR): Update accordingly.
7074 * tree.h (wi::int_traits <const_tree>): Delete.
7075 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7076 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7077 (wi::tree_to_wide_ref): New typedef.
7078 (wi::to_wide): New function.
7079 * calls.c (get_size_range): Use wi::to_wide when operating on
7081 * cgraph.c (cgraph_node::create_thunk): Likewise.
7082 * config/i386/i386.c (ix86_data_alignment): Likewise.
7083 (ix86_local_alignment): Likewise.
7084 * dbxout.c (stabstr_O): Likewise.
7085 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7086 * expr.c (const_vector_from_tree): Likewise.
7087 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7088 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7089 (fold_negate_expr_1, int_const_binop_1, const_binop)
7090 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7091 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7092 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7093 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7094 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7095 (fold_not_const, round_up_loc): Likewise.
7096 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7097 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7098 (alloca_call_type): Likewise.
7099 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7100 * godump.c (go_output_typedef): Likewise.
7101 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7102 * internal-fn.c (get_min_precision): Likewise.
7103 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7104 * ipa-polymorphic-call.c
7105 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7106 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7107 (ipa_modify_call_arguments): Likewise.
7108 * match.pd: Likewise.
7109 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7110 * print-tree.c (print_node_brief, print_node): Likewise.
7111 * stmt.c (expand_case): Likewise.
7112 * stor-layout.c (layout_type): Likewise.
7113 * tree-affine.c (tree_to_aff_combination): Likewise.
7114 * tree-cfg.c (group_case_labels_stmt): Likewise.
7115 * tree-data-ref.c (dr_analyze_indices): Likewise.
7116 (prune_runtime_alias_test_list): Likewise.
7117 * tree-dump.c (dequeue_and_dump): Likewise.
7118 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7119 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7120 * tree-pretty-print.c (dump_generic_node): Likewise.
7121 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7122 (simple_iv_with_niters): Likewise.
7123 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7124 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7125 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7126 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7127 (refine_value_range_using_guard, number_of_iterations_ne_max)
7128 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7129 (get_cst_init_from_scev, record_nonwrapping_iv)
7130 (scev_var_range_cant_overflow): Likewise.
7131 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7132 * tree-ssa-pre.c (compute_avail): Likewise.
7133 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7134 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
7135 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
7136 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
7137 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
7138 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
7139 (dump_case_nodes, try_switch_expansion): Likewise.
7140 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
7141 (vect_do_peeling): Likewise.
7142 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
7143 * tree-vect-stmts.c (vectorizable_load): Likewise.
7144 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
7145 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
7146 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
7147 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
7148 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
7149 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
7150 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
7151 (evrp_dom_walker::before_dom_children): Likewise.
7152 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
7153 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
7154 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
7155 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
7156 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
7157 (get_range_pos_neg): Likewise.
7158 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
7159 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
7160 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
7161 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
7162 * config/avr/avr.c (avr_fold_builtin): Likewise.
7163 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
7164 * config/msp430/msp430.c (msp430_attr): Likewise.
7165 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
7166 * config/powerpcspe/powerpcspe-c.c
7167 (altivec_resolve_overloaded_builtin): Likewise.
7168 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
7169 (rs6000_expand_ternop_builtin): Likewise.
7170 * config/rs6000/rs6000-c.c
7171 (altivec_resolve_overloaded_builtin): Likewise.
7172 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
7173 (rs6000_expand_ternop_builtin): Likewise.
7174 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
7176 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7178 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
7179 when copying loop nest with only one inner loop.
7181 2017-10-10 Richard Biener <rguenther@suse.de>
7183 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
7184 blocks if SCEV is active.
7185 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
7187 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
7188 (scev_initialize): Assert we are not yet initialized.
7190 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7192 * tree-loop-distribution.c (generate_loops_for_partition): Remove
7193 inner loop's exit stmt by making it always exit the loop, otherwise
7194 we would generate an infinite empty loop.
7196 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7198 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
7199 renaming variables in new preheader if it's deleted.
7201 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7203 * tree-loop-distribution.c (struct partition): Remove unused field
7204 loops of the structure.
7205 (partition_alloc, partition_free): Ditto.
7206 (build_rdg_partition_for_vertex): Ditto.
7208 2017-10-09 Jeff Law <law@redhat.com>
7210 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
7211 return type to match prototype and documentation.
7213 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7215 * config/rs6000/rs6000.c (processor_costs): Move to ...
7216 * config/rs6000/rs6000.h: ... here.
7217 (rs6000_cost): Declare.
7219 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
7221 * except.c (setjmp_fn): New global variable.
7222 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
7223 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
7224 if DONT_USE_BUILTIN_SETJMP is defined.
7226 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7228 * target.def (insn_cost): New hook.
7229 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
7230 * doc/tm.texi: Regenerate.
7231 * rtlanal.c (insn_cost): Use the new hook.
7233 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7235 * combine.c (combine_validate_cost): Compute the new insn_cost,
7236 not just pattern_cost.
7237 (try_combine): Adjust comment.
7239 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7241 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7243 * combine.c (uid_insn_cost): Adjust comment.
7244 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7246 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7247 * dse.c (find_shift_sequence): Ditto.
7248 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7249 (bb_valid_for_noce_process_p): Use pattern_cost.
7250 * rtl.h (insn_rtx_cost): Delete.
7251 (pattern_cost): New prototype.
7252 (insn_cost): New prototype.
7253 * rtlanal.c (insn_rtx_cost): Rename to...
7254 (pattern_cost): ... this.
7257 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7259 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7260 (*jcc<mode>_0_r_i387): Ditto.
7261 (*jccxf_r_i387): Ditto.
7262 (*jcc<mode>_r_i387): Ditto.
7263 (*jccu<mode>_r_i387): Ditto.
7264 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7265 (*jcc): Rename from *jcc_1.
7267 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7269 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7270 deferred rescans after the lvx/stvx recombination pre-pass.
7272 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7274 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7275 memory operation instruction support.
7276 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7277 (rs6000-ibm-aix[789]*): Likewise.
7278 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7279 Document new functions.
7281 2017-10-09 Richard Biener <rguenther@suse.de>
7283 PR tree-optimization/82397
7284 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7285 equality only for semantically equal trees.
7287 2017-10-09 Richard Biener <rguenther@suse.de>
7289 PR tree-optimization/82449
7290 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7291 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7292 allow constant addresses.
7293 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7296 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7298 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7301 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7304 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7307 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7310 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7312 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7315 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7316 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7318 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7320 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7321 (WI_BINARY_PREDICATE_RESULT): Likewise.
7322 (wi::binary_traits::operator_result): New type.
7323 (wi::binary_traits::predicate_result): Likewise.
7324 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7325 (generic_wide_int::operator==, generic_wide_int::operator!=)
7326 (generic_wide_int::operator&, generic_wide_int::and_not)
7327 (generic_wide_int::operator|, generic_wide_int::or_not)
7328 (generic_wide_int::operator^, generic_wide_int::operator+
7329 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7331 (operator~, unary operator-, operator==, operator!=, operator&)
7332 (operator|, operator^, operator+, binary operator-, operator*): New
7334 * expr.c (get_inner_reference): Use wi::bit_and_not.
7335 * fold-const.c (fold_binary_loc): Likewise.
7336 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7337 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7338 (bit_value_binop): Likewise.
7339 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7340 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7341 (extract_range_from_binary_expr_1): Likewise.
7342 (masked_increment): Likewise.
7343 (simplify_bit_ops_using_ranges): Likewise.
7345 2017-10-09 Martin Jambor <mjambor@suse.cz>
7348 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7349 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7350 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7352 (hsa_fixup_mov_insn_type): New function.
7353 (hsa_op_with_type::get_in_type): Use it.
7354 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7355 immediates in an assert.
7356 (hsa_op_with_type::extend_int_to_32bit): New method.
7357 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7358 types. Convert to dest type if necessary.
7359 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7360 (reg_for_gimple_ssa): Pass false as min32int to
7361 hsa_type_for_scalar_tree_type.
7362 (gen_hsa_addr): Fixup type when creating addresable temporary.
7363 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7364 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7365 necessary. Call hsa_fixup_mov_insn_type.
7366 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7367 extend operands and convert to dest type if necessary.
7368 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7369 to dest type if necessary.
7370 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7371 if conversion nt necessary and size matches.
7372 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7373 to dest type if necessary.
7374 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7375 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7377 (gen_hsa_clrsb): Likewise.
7378 (gen_hsa_ffs): Likewise.
7379 (gen_hsa_divmod): Extend operands and convert to dest type if
7381 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7383 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7385 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7386 Remove empty default arguments. Use a brace block as output
7388 (conditional return): Ditto.
7390 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7391 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7393 (group_ending_nop): Ditto.
7394 (doloop_end): Ditto.
7395 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7396 (splitters for those): Ditto.
7398 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7400 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7401 a conditional jump (and the compare for it) so that pc_rtx is the
7403 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7404 for the deleted and renamed ctr<mode>_internal[234] patterns.
7405 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7406 Delete second conditional return pattern.
7407 (ctr<mode>_internal2): Delete this second bdnz pattern.
7408 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7409 (ctr<mode>_internal4): Delete this second bdz pattern.
7411 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7413 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7414 (always_initialized_rtx_for_ssa_name_p): New predicate.
7415 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7416 (finish_out_of_ssa): Free new field of SA.
7417 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7418 * tree-ssa-coalesce.c: Include tree-ssa.h.
7419 (get_parm_default_def_partitions): Remove extern keyword.
7420 (get_undefined_value_partitions): New function.
7421 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7422 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7424 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7426 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7428 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7430 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7432 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7433 for targets that preffer 128bit.
7435 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7437 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
7439 2017-10-08 Olivier Hainque <hainque@adacore.com>
7441 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
7442 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
7444 (thumb_set_return_address): Likewise.
7446 2017-10-08 Olivier Hainque <hainque@adacore.com>
7448 * common/config/arm/arm-common.c (arm_except_unwind_info):
7449 Handle DWARF2_UNWIND_INFO.
7451 2017-10-07 Michael Collison <michael.collison@arm.com>
7453 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
7456 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
7458 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
7459 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
7460 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
7461 defined, force the creation of a new block for a dispatch label.
7463 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
7465 * invoke.texi (Wsuggest-attribute=cold): Document.
7466 * common.opt (Wsuggest-attribute=cold): New
7467 * ipa-pure-const.c (warn_function_cold): New function.
7468 * predict.c (compute_function_frequency): Use it.
7469 * predict.h (warn_function_cold): Declare.
7471 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
7473 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
7476 2017-10-06 Martin Liska <mliska@suse.cz>
7478 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
7479 keyword for member functions.
7480 (struct sanopt_tree_couple): New struct.
7481 (struct sanopt_tree_couple_hash): New function.
7482 (struct sanopt_ctx): Add new hash_map.
7483 (has_dominating_ubsan_ptr_check): New function.
7484 (record_ubsan_ptr_check_stmt): Likewise.
7485 (maybe_optimize_ubsan_ptr_ifn): Likewise.
7486 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
7487 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
7489 2017-10-06 Sudakshina Das <sudi.das@arm.com>
7492 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
7493 aarch64_simd_valid_immediate on CONST_VECTORs.
7494 (aarch64_reg_or_bic_imm): Likewise.
7496 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
7498 PR rtl-optimization/82396
7499 * haifa-sched.c (ready_sort_real): Disable qsort checking.
7501 2017-10-06 Sebastian Pop <sebpop@gmail.com>
7503 * graphite-dependences.c (scop_get_reads): Move code to...
7504 (scop_get_must_writes): Move code to...
7505 (scop_get_may_writes): Move code to...
7506 (scop_get_reads_and_writes): ... here.
7507 (scop_get_dependences): Call scop_get_reads_and_writes.
7509 2017-10-06 Jakub Jelinek <jakub@redhat.com>
7511 PR tree-optimization/82434
7512 * fold-const.h (can_native_encode_type_p,
7513 can_native_encode_string_p): Remove.
7514 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
7515 don't encode anything, just return what would be otherwise returned.
7516 (native_encode_fixed, native_encode_complex, native_encode_vector):
7518 (native_encode_string): Likewise. Inline by hand
7519 can_native_encode_string_p.
7520 (can_native_encode_type_p): Remove.
7521 (can_native_encode_string_p): Remove.
7522 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
7523 STRING_CSTs using can_native_encode_string_p, test all
7524 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
7525 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
7526 argument from native_encode_expr.
7527 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
7528 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
7531 2017-10-06 Richard Biener <rguenther@suse.de>
7533 PR tree-optimization/82397
7534 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
7535 operand_equal_p but rely on data_ref_compare_tree for detecting
7537 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
7538 to match up with dr_group_sort_cmp.
7540 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7543 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
7545 * config/s390/s390-builtin-types.def: Regenerate.
7547 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7550 * config/s390/s390-builtin-types.def: Regenerate.
7551 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
7552 Change flag from B_VXE to B_VX.
7553 (s390_vec_min_dbl): Remove B_VXE flag.
7555 2017-10-06 Richard Biener <rguenther@suse.de>
7557 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
7558 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
7559 translate_isl_ast_to_gimple::is_valid_rename,
7560 translate_isl_ast_to_gimple::get_rename,
7561 translate_isl_ast_to_gimple::get_def_bb_for_const,
7562 translate_isl_ast_to_gimple::get_new_name,
7563 translate_isl_ast_to_gimple::collect_all_ssa_names,
7564 translate_isl_ast_to_gimple::copy_loop_phi_args,
7565 translate_isl_ast_to_gimple::collect_all_ssa_names,
7566 translate_isl_ast_to_gimple::copy_loop_phi_args,
7567 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
7568 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
7569 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
7570 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
7571 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
7572 translate_isl_ast_to_gimple::copy_cond_phi_args,
7573 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7574 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7575 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7576 translate_isl_ast_to_gimple::rename_uses,
7577 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7578 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7579 (set_rename_for_each_def): Likewise.
7580 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7581 here. Handle rewriting SCEV analyzable uses here.
7582 (copy_bb_and_scalar_dependences): Generate code for PHI
7584 (graphite_regenerate_ast_isl): Adjust.
7585 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7586 (add_write, add_read): New functions.
7587 (build_cross_bb_scalars_def): Use it and simplify.
7588 (build_cross_bb_scalars_use): Likewise.
7589 (graphite_find_cross_bb_scalar_vars): Inline into...
7590 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7591 simulating out-of-SSA. Compute liveout and add dependencies.
7592 (build_scops): Force an empty entry block.
7593 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7595 (sese_build_liveouts): Declare.
7596 (sese_trivially_empty_bb_p): Likewise.
7597 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7598 compute liveout and debug_liveout.
7599 (sese_bad_liveouts_use): Remove.
7600 (sese_reset_debug_liveouts_bb): Likewise.
7601 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7602 (sese_build_liveouts): Build liveout and debug_liveout and store
7604 (new_sese_info): Adjust.
7605 (free_sese_info): Likewise.
7606 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7607 do not build liveout here.
7608 (move_sese_in_condition): Adjust region entry.
7609 (scev_analyzable_p): Match up with chrec_apply requirements.
7610 (sese_trivially_empty_bb_p): New.
7611 * tree-into-ssa.c (get_reaching_def): Properly support generating
7612 default-defs for incremental rewrite of anonymous names.
7614 2017-10-06 Richard Biener <rguenther@suse.de>
7616 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7617 precision do not perform modulo reduction.
7619 2017-10-06 Richard Biener <rguenther@suse.de>
7621 PR tree-optimization/82436
7622 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7623 conservatively choose the vectorization factor when checking
7624 whether we can perform the required load permutation.
7625 (vect_transform_slp_perm_load): Assert when we may not fail.
7627 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7629 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7630 message for incompatible -msdata=* and -mcall-* options.
7632 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7634 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7635 rate for post-reload scheduling.
7637 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7639 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7641 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7643 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7644 to improve monte carlo in scimark.
7646 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7648 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7649 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7650 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7651 pentium4_cost, nocona_cost): Set reassociation width to 1.
7652 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7653 width to 2 for fp operations and 1 otherwise.
7654 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7656 (atom_cost): Set reassociation width to 2.
7657 (slm_cost, generic_cost): Set fp reassociation width
7658 to 2 and 1 otherwise.
7659 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7660 (core_cost): Set fp reassociation width to 4 and vector to 2.
7661 (ix86_reassociation_width): Rewrite using cost table; special case
7662 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7663 and TARGET_AVX128_OPTIMAL.
7664 * config/i386/i386.h (processor_costs): Add
7665 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7666 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7667 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7668 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7669 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7670 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7672 2017-10-05 Nathan Sidwell <nathan@acm.org>
7674 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7676 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7678 * config/arm/arm.c (arm_test_fpu_data): New.
7679 (arm_run_selftests): Call arm_test_fpu_data.
7681 2017-10-04 Nathan Sidwell <nathan@acm.org>
7683 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7685 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7687 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7689 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7690 check type for aarch64_simd_valid_immediate.
7691 (aarch64_output_simd_mov_immediate): Update prototype.
7692 (aarch64_simd_valid_immediate): Update prototype.
7693 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7694 support for ORR-immediate.
7695 (and<mode>3): modified pattern to add support for BIC-immediate.
7696 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7697 now checks for valid immediate for BIC and ORR based on new enum
7699 (aarch64_output_simd_mov_immediate): Function now used to output
7700 BIC/ORR imm as well based on new enum argument.
7701 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7703 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7704 (aarch64_reg_or_bic_imm): Likewise.
7706 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7708 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7709 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7711 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7715 PR rtl-optimization/82396
7716 * haifa-sched.c (autopref_multipass_init): Simplify
7718 (autopref_rank_data): Simplify sort order.
7719 * sched-int.h (autopref_multipass_data_): Remove
7720 multi_mem_insn_p, min_offset and max_offset.
7722 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7724 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7726 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7728 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7731 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7733 * doc/sourcebuild.texi: Document vect_long_mult.
7735 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7737 PR tree-optimization/82413
7738 * fold-const.c (build_range_check): Use widest_int when comparing
7739 the maximum ETYPE value with HIGH.
7741 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7743 PR rtl-optimization/82396
7744 * haifa-sched.c (autopref_multipass_init): Simplify
7746 (autopref_rank_data): Simplify sort order.
7747 * sched-int.h (autopref_multipass_data_): Remove
7748 multi_mem_insn_p, min_offset and max_offset.
7750 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7752 PR tree-optimization/82381
7753 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7754 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7757 PR tree-optimization/82374
7758 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7759 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7760 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7761 current_function_decl to the new decl.
7763 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7765 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7766 helper macro for IEEE float128 hardware built-in functions.
7767 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7769 (TRUNCF128_ODD): Likewise.
7770 (ADDF128_ODD): Likewise.
7771 (SUBF128_ODD): Likewise.
7772 (MULF128_ODD): Likewise.
7773 (DIVF128_ODD): Likewise.
7774 (FMAF128_ODD): Likewise.
7775 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7776 UNSPEC_TRUNC_ROUND_TO_ODD.
7777 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7778 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7779 floating point round to odd instructions.
7780 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7781 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7782 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7783 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7784 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7785 (trunc<mode>sf2_hw): Change the truncate with round to odd
7786 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7787 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7788 to odd hardware instructions.
7789 (sub<mode>3_odd): Likewise.
7790 (mul<mode>3_odd): Likewise.
7791 (div<mode>3_odd): Likewise.
7792 (sqrt<mode>2_odd): Likewise.
7793 (fma<mode>4_odd): Likewise.
7794 (fms<mode>4_odd): Likewise.
7795 (nfma<mode>4_odd): Likewise.
7796 (nfms<mode>4_odd): Likewise.
7797 (trunc<mode>df2_odd): Change the truncate with round to odd
7798 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7800 * doc/extend.texi (PowerPC built-in functions): Update documentation
7801 for existing IEEE float128-bit built-in functions. Add built-in
7802 functions that generate the IEEE 128-bit floating point round to
7805 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7807 PR rtl-optimization/77729
7808 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7809 to (X&(C1&~C2))|C2 transformations.
7811 2017-10-03 Martin Jambor <mjambor@suse.cz>
7813 PR tree-optimization/82363
7814 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7815 mismatch, mark lacc written regardless of racc.
7817 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7819 PR tree-optimization/82381
7820 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7821 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7822 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7823 fallthrough into reassoc_stmt_dominates_stmt_p.
7826 * combine.c (combine_instructions): Don't combine in unreachable
7829 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7832 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7833 function to not use the have_cpu variable. Do not set cpu_index,
7834 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7836 (rs6000_valid_attribute_p): Remove duplicate initializations of
7837 old_optimize and func_optimize.
7838 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7839 (rs6000_activate_target_options): Make global.
7840 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7843 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7845 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7846 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7847 Set *poffset to 0 and *psize and *pmax_size to -1 if
7848 *poffset + *psize overflows in HOST_WIDE_INT.
7850 PR tree-optimization/82387
7851 PR tree-optimization/82388
7852 PR tree-optimization/82389
7853 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7854 instead of live_bytes non-NULL.
7856 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7859 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7862 2017-10-02 Richard Biener <rguenther@suse.de>
7864 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7865 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7866 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7868 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7870 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7871 requested precision matches the type's.
7872 * calls.c (alloc_max_size): Calculate the new candidate size as
7873 a widest_int and use wi::to_widest when comparing it with the
7874 current candidate size.
7875 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7876 zero rather than integer_zero_node.
7877 * match.pd: Check for a no-op conversion before using wi::add
7878 rather than after. Use tree_to_uhwi when summing small shift
7879 counts into an unsigned int.
7881 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7882 Alan Hayward <alan.hayward@arm.com>
7883 David Sherwood <david.sherwood@arm.com>
7886 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7887 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7888 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7889 POINTER_AND_FP_REGS.
7891 2017-10-02 Richard Biener <rguenther@suse.de>
7893 PR tree-optimization/82355
7894 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7895 a mapping for the enclosing loop but avoid generating one for
7897 (copy_bb_and_scalar_dependences): Remove premature codegen
7898 error on PHIs in blocks duplicated into multiple places.
7899 * graphite-scop-detection.c
7900 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7901 in the region use it as loop and nest to analyze the DR in.
7902 (try_generate_gimple_bb): Likewise.
7903 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7904 (add_loop_constraints): For blocks in a loop not in the region
7905 create a dimension with a single iteration.
7906 * sese.h (gbb_loop_at_index): Remove assert.
7908 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7910 * omp-expand.c (adjust_context_scope): New function.
7911 (expand_parallel_call): Call adjust_context_scope.
7913 2017-10-01 Jeff Law <law@redhat.com>
7915 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7916 dom_opt_dom_walker class with direct access to private members.
7917 Add comments. Call test_for_singularity.
7918 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7919 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7920 m_dummy_cond anymore.
7921 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7923 (pass_dominator:execute): Build the dummy_cond here and pass it
7924 to the dom_opt_dom_walker ctor.
7925 (test_for_singularity): New function.
7927 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7928 Maya Rashish <coypu@sdf.org>
7930 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7931 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7932 (alpha*-*-netbsd*) Use nbsd_tm_file.
7933 (arm*-*-netbsdelf*) Likewise.
7934 (i[34567]86-*-netbsdelf*) Likewise.
7935 (x86_64-*-netbsd*) Likewise.
7936 (mips*-*-netbsd*) Likewise.
7937 (powerpc-*-netbsd*) Likewise.
7938 (sh*-*-netbsd*) Likewise.
7939 (sparc-*-netbsdelf*) Likewise.
7940 (sparc64-*-netbsd*) Likewise.
7941 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7943 (vax-*-netbsdelf*) Likewise.
7944 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7945 (UINT_FAST8_TYPE) Likewise.
7946 (INT_FAST16_TYPE) Check CHAR_FAST16.
7947 (UINT_FAST16_TYPE) Likewise.
7949 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7952 * config/i386/i386.md
7953 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7954 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7955 *divmodsi4_zext_2): New define_insn_and_split.
7956 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7957 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7958 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7959 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7960 New define_insn_and_split.
7961 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7962 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7963 operands[1] having DImode when mode is SImode.
7965 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7966 always SImode for DIV and MOD in REG_EQUAL notes.
7968 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7971 * match.pd: Fix handling of NaNs in pattern.
7973 2017-09-29 Jeff Law <law@redhat.com>
7975 * sbitmap.c (bitmap_bit_in_range_p): New function.
7976 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7977 * tree-ssa-dse.c (live_bytes_read): New function.
7978 (dse_classify_store): Ignore reads of dead bytes.
7980 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7981 typos and whitespace errors.
7982 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7983 * config/s390/s390.c (s390_emit_prologue): Likewise.
7985 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7988 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7989 with a symbol for LRA.
7991 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7993 PR rtl-optimization/82338
7994 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7996 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7998 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7999 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8000 (qsort_chk): Declare.
8001 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8002 (qsort_chk): New function.
8004 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8006 PR tree-optimization/82337
8007 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8008 phi definition if the PHI result appears in an abnormal PHI.
8009 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8010 basis appears in an abnormal PHI.
8012 2017-09-29 Richard Biener <rguenther@suse.de>
8014 * graphite-isl-ast-to-gimple.c
8015 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8016 (binary_op_to_tree): Use it.
8017 (get_rename_from_scev): Likewise.
8018 (copy_loop_phi_nodes): Likewise.
8019 (copy_bb_and_scalar_dependences): Likewise.
8020 (translate_pending_phi_nodes): Likewise.
8022 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8025 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8026 for movabsq $(i32 << shift), r64.
8028 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8030 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8031 index when encoding %esp as %rsp to avoid 0x67 prefix.
8033 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8035 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8036 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8038 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8040 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8041 Extensions with more than 16 double VFP registers.
8042 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8043 to_clear_mask and all code related to it. Replace the remaining
8044 entry by a sbitmap and adapt code accordingly.
8046 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8048 * brig-builtins.def: Change pure attributes to const.
8050 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8052 * config.gcc (default_gnu_indirect_function): Default to yes for
8053 sparc*-*-linux* with glibc.
8055 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8057 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8058 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8059 when creating .init_array and .fini_array sections with priority
8062 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8065 * config/aarch64/aarch64.c
8066 (aarch64_builtin_support_vector_misalignment): Always return false
8067 when misalignment is unknown.
8069 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8071 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8072 this function to return false if the definition used by the swap
8073 instruction is artificial, or if the memory address from which the
8074 constant value is loaded is not represented by a base address held
8075 in a register or if the base address register is a frame or stack
8076 pointer. Additionally, return false if the base address of the
8077 loaded constant is a SYMBOL_REF but is not considered to be a
8079 (replace_swapped_load_constant): New function.
8080 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8081 loaded constant vector with a load of a swapped constant vector.
8083 2017-09-27 Carl Love <cel@us.ibm.com>
8085 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8086 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8087 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8090 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8092 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8093 first, always call autopref_rank_data otherwise.
8095 2017-09-27 Richard Biener <rguenther@suse.de>
8097 * graphite-scop-detection.c (find_scop_parameters): Move
8098 loop bound handling ...
8099 (gather_bbs::before_dom_children): ... here, avoiding the need
8100 to build scop_info->loop_nest.
8101 (record_loop_in_sese): Remove.
8102 * sese.h (sese_info_t::loop_nest): Remove.
8103 * sese.c (new_sese_info): Do not allocate loop_nest.
8104 (free_sese_info): Do not free loop_nest.
8106 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8109 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8110 lhs from calls if the lhs has addressable type.
8112 2017-09-27 Richard Biener <rguenther@suse.de>
8114 * graphite.h (scop::max_alias_set): New member.
8115 * graphite-scop-detection.c: Remove references to non-existing
8116 --param in comments.
8117 (build_alias_sets): Record the maximum alias set used for drs.
8118 (build_scops): Support zero as unlimited for
8119 --param graphite-max-arrays-per-scop.
8120 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8122 (build_poly_sr_1): ... here. Compute alias set based on the
8123 maximum alias set used for drs rather than
8124 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8126 2017-09-27 Richard Biener <rguenther@suse.de>
8128 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8129 --param loop-block-tile-size=0 to disable tiling.
8131 2017-09-27 Richard Biener <rguenther@suse.de>
8133 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8134 (graphite-max-nb-scop-params): Document special value zero.
8135 * domwalk.h (dom_walker::STOP): New symbolical constant.
8136 (dom_walker::dom_walker): Add optional parameter for bb to
8138 (dom_walker::~dom_walker): Declare.
8139 (dom_walker::before_dom_children): Document STOP return value.
8140 (dom_walker::m_user_bb_to_rpo): New member.
8141 (dom_walker::m_bb_to_rpo): Likewise.
8142 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
8143 mapping here if not provided by the user.
8144 (dom_walker::~dom_walker): Free bb to RPO mapping if not
8145 provided by the user.
8146 (dom_walker::STOP): Define.
8147 (dom_walker::walk): Do not compute bb to RPO mapping here.
8148 Support STOP return value from before_dom_children to stop
8150 * graphite-optimize-isl.c (optimize_isl): If the schedule
8151 is the same still generate code if -fgraphite-identity
8152 or -floop-parallelize-all are given.
8153 * graphite-scop-detection.c: Include cfganal.h.
8154 (gather_bbs::gather_bbs): Get and pass through bb to RPO
8156 (gather_bbs::before_dom_children): Return STOP for BBs
8158 (build_scops): Compute bb to RPO mapping and pass it to
8159 the domwalk. Treat --param graphite-max-nb-scop-params=0
8160 as not limiting the number of params.
8161 * graphite.c (graphite_initialize): Remove limit on the
8162 number of basic-blocks in a function.
8163 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
8164 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
8165 default value of 10.
8167 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8169 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
8170 Adjust code to eliminate needing to do the shift right 32-bits
8171 operation after XSCVDPSPN.
8173 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8175 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
8176 ((X / Y) != 0 -> X >= Y): Likewise.
8178 2017-09-26 Carl Love <cel@us.ibm.com>
8180 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
8181 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
8182 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
8183 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
8184 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
8185 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
8186 definitions and overloading.
8187 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
8188 statement for P9V_BUILTIN_XST_LEN_R.
8189 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
8190 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
8191 define_expand and define_insn for the instructions and builtins.
8192 * doc/extend.texi: Update the built-in documentation file for the new
8194 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
8195 define_insn for the instructions
8197 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
8200 * gcc/config/netbsd-protos.h: New file.
8201 * gcc/config/netbsd.c: New file.
8202 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
8203 * gcc/config/t-netbsd: New file.
8204 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
8205 (tmake_file) Add t-netbsd.
8206 (extra_objs) Add netbsd.o.
8208 2017-09-26 Janus Weil <janus@gcc.gnu.org>
8212 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
8214 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8216 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
8217 sign extension from a vector register to a GPR by doing a 32-bit
8218 direct move and then an EXTSW.
8219 (extendsi<mode>2 splitter): Likewise.
8220 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
8221 right or vector extract after doing XSCVDPSPN. Use
8222 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
8224 (movdi_from_sf_zero_ext): Likewise.
8225 (reload_gpr_from_vsxsf): Likewise.
8226 (p8_mfvsrd_4_disf): Delete, no longer used.
8227 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
8228 then needing to move the SFmode to a GPR to use the XSCVDPSP
8229 instruction instead of FRSP and XSCVDPSPN.
8230 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
8231 it is adjacent to the other XSCVSPDP insns.
8232 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
8233 SFmode to be in traditional Altivec registers.
8234 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8235 (vsx_xscvspdpn): Likewise.
8236 (vsx_xscvspdpn_scalar): Likewise.
8238 2017-09-26 Martin Jambor <mjambor@suse.cz>
8240 * tree-sra.c (compare_access_positions): Put integral types first,
8241 stabilize sorting of integral types, remove conditions putting
8242 non-full-precision integers last.
8243 (sort_and_splice_var_accesses): Disable scalarization if a
8244 non-integert would be represented by a non-full-precision integer.
8246 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8248 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8249 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8250 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8251 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8252 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8253 conditionals inside the function instead of around it. Call
8254 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8255 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8257 2017-09-26 Richard Biener <rguenther@suse.de>
8259 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8261 (scop_detection::build_scop_breadth): ... this. Removed.
8262 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8263 (scop_detection::harmful_stmt_in_bb): Likewise.
8264 (scop_detection::graphite_can_represent_stmt): Likewise.
8265 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8266 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8267 (scop_detection::can_represent_loop_1): ... this. Removed.
8268 (scop_detection::harmful_loop_in_region): Simplify after inlining
8269 the above and remove more quadraticness.
8270 (build_scops): Adjust.
8271 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8274 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8277 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8278 REGNO (base) == SP_REG if base is a REG.
8281 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8282 if it is different SSA_NAME.
8283 (optimize_range_tests_cmp_bitwise): New function.
8284 (optimize_range_tests): Call it.
8286 2017-09-26 Richard Biener <rguenther@suse.de>
8288 PR tree-optimization/82321
8289 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8290 for the def being inside the loop.
8292 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8294 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8296 * config/s390/s390-builtins.def: Fix constraint on op4.
8298 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8300 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8301 independent expanders.
8302 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8303 ("vec_ordered", "vec_unordered"): New expanders.
8305 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8307 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8310 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8312 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8313 vec_unpacks_lo_v16qi.
8314 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8316 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8318 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8319 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8320 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8322 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8324 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8326 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8328 ("*vec_slb<mode>"): New insn pattern.
8329 ("vec_shr_<mode>"): New expander.
8330 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8331 and force the shift count operand to V16QImode.
8332 ("vec_srb<mode>"): Set shift count mode to V16QI.
8334 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8336 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8337 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8338 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8340 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8343 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8345 2017-09-26 Richard Biener <rguenther@suse.de>
8347 PR tree-optimization/82320
8348 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8351 2017-09-25 Jeff Law <law@redhat.com>
8353 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8354 prototype for new argument.
8355 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8356 mostly extracted from rs6000_emit_allocate_stack.
8357 (rs6000_emit_probe_stack_range_stack_clash): New function.
8358 (rs6000_emit_allocate_stack): Call
8359 rs6000_emit_probe_stack_range_stack_clash as needed.
8360 (rs6000_emit_probe_stack_range): Add additional argument
8361 to call to gen_probe_stack_range{si,di}.
8362 (output_probe_stack_range): New.
8363 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8364 (output_probe_stack_range_stack_clash): New.
8365 (rs6000_emit_prologue): Emit notes into dump file as requested.
8366 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8367 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8368 Add additional operand and pass it to output_probe_stack_range.
8370 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8372 PR tree-optimization/82163
8373 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8374 (checking_verify_loop_closed_ssa): New parameter.
8375 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8376 (check_loop_closed_ssa_stmt): Delete.
8377 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8378 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8379 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8382 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8384 * brig-builtins.def: Treat HSAIL barrier builtins as
8385 setjmp/longjump style functions.
8387 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8389 * target.def (constant_alignment): New hook.
8390 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8391 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8392 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8393 * doc/tm.texi: Regenerate.
8394 * targhooks.h (default_constant_alignment): Declare.
8395 (constant_alignment_word_strings): Likewise.
8396 * targhooks.c (default_constant_alignment): New function.
8397 (constant_alignment_word_strings): Likewise.
8398 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8399 instead of CONSTANT_ALIGNMENT.
8400 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8401 (force_const_mem): Likewise.
8402 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8403 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8404 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8405 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8406 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8408 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8409 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8410 constant_alignment_word_strings.
8411 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8412 (CONSTANT_ALIGNMENT): Likewise.
8413 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8414 (arm_constant_alignment): New function.
8415 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8416 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8417 constant_alignment_word_strings.
8418 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8419 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8420 constant_alignment_word_strings.
8421 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8422 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8423 (cris_constant_alignment): New function.
8424 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8425 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8426 (epiphany_constant_alignment): New function.
8427 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8428 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8429 constant_alignment_word_strings.
8430 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8431 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8432 constant_alignment_word_strings.
8433 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8434 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8435 constant_alignment_word_strings.
8436 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8437 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
8438 * config/i386/i386.c (ix86_constant_alignment): Make static.
8439 Use the same interface as the target hook.
8440 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8441 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
8442 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8443 constant_alignment_word_strings.
8444 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
8445 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
8446 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8447 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
8448 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8449 constant_alignment_word_strings.
8450 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
8451 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8452 constant_alignment_word_strings.
8453 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
8454 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8455 constant_alignment_word_strings.
8456 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
8457 * config/microblaze/microblaze.c (microblaze_constant_alignment):
8459 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8460 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
8461 * config/mips/mips.c (mips_constant_alignment): New function.
8462 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8463 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
8464 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
8465 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8466 (mmix_constant_alignment): Make static. Use the same interface
8468 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
8469 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8470 constant_alignment_word_strings.
8471 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
8472 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8473 constant_alignment_word_strings.
8474 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
8475 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8476 constant_alignment_word_strings.
8477 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
8478 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8479 (rs6000_constant_alignment): New function.
8480 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
8481 * config/riscv/riscv.c (riscv_constant_alignment): New function.
8482 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8483 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
8484 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8485 (rs6000_constant_alignment): New function.
8486 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
8487 * config/s390/s390.c (s390_constant_alignment): New function.
8488 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8489 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
8490 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8491 constant_alignment_word_strings.
8492 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
8493 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8494 (sparc_constant_alignment): New function.
8495 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
8496 * config/spu/spu.c (spu_constant_alignment): New function.
8497 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8498 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
8499 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8500 constant_alignment_word_strings.
8501 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
8502 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8503 constant_alignment_word_strings.
8504 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
8505 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8506 constant_alignment_word_strings.
8507 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
8508 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8509 (visium_constant_alignment): New function.
8510 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
8511 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8512 (xtensa_constant_alignment): New function.
8513 * system.h (CONSTANT_ALIGNMENT): Poison.
8515 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
8517 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8518 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
8519 (rs6000_builtin_valid_without_lhs): New helper function.
8520 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8521 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
8523 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8525 * target.h (vec_perm_indices): Use unsigned short rather than
8527 (auto_vec_perm_indices): Likewise.
8528 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
8529 Use unsigned int rather than unsigned char.
8530 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
8532 2017-09-25 Richard Biener <rguenther@suse.de>
8534 * cfgloop.h (sort_sibling_loops): Declare.
8535 * cfgloop.c (sort_sibling_loops_cmp): New helper.
8536 (sort_sibling_loops): New function sorting the sibling loop list
8538 * graphite.c (graphite_transform_loops): Sort sibling loops.
8540 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
8542 * target.def (vec_perm_const_ok): Change sel parameter to
8544 * optabs-query.c (can_vec_perm_p): Update accordingly.
8545 * doc/tm.texi: Regenerate.
8546 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
8547 auto_vec_perm_indices and remove separate nelt field.
8548 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
8549 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
8550 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
8551 (aarch64_expand_vec_perm_const): Update accordingly.
8552 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
8553 to vec_perm_indices.
8554 * config/arm/arm.c (expand_vec_perm_d): Change perm to
8555 auto_vec_perm_indices and remove separate nelt field.
8556 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
8557 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
8558 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
8560 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
8561 to vec_perm_indices.
8562 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
8563 sel to vec_perm_indices.
8564 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
8565 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
8566 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
8568 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
8571 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8574 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8575 on the FUNCTION_DECL function context if it has a DIE that is a
8578 2017-09-25 Richard Biener <rguenther@suse.de>
8580 PR tree-optimization/82285
8581 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8584 2017-09-25 Tom de Vries <tom@codesourcery.com>
8588 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8591 2017-09-25 Richard Biener <rguenther@suse.de>
8593 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8594 ISL errors other than isl_error_quota happen. Dump if the
8595 schedule is the same.
8596 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8597 errors instead of aborting inside ISL.
8599 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8602 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8603 of libgcc_eh for m64.
8604 * config/i386/darwin64.h: Likewise.
8606 2017-09-25 Richard Biener <rguenther@suse.de>
8609 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8610 attribute for incomplete types nor twice for complete ones.
8612 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8615 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8616 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8619 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8622 * config/i386/i386.opt (mprefer-avx256): Use
8623 ix86_target_flags variable.
8624 * config/i386/i386.c (ix86_target_string): Move
8625 -mprefer-avx256 to flag2_opts.
8627 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8630 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8631 and x != -1 | y != -1 into (x & y) != -1.
8633 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8635 * config.gcc: Add new case statement to set
8636 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8637 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8638 s390x-*-linux* case statements. Added aarch64 to the list of
8639 supported architectures.
8641 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8643 PR tree-optimization/82289
8644 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8645 STMT_VINFO_RELEVANT_P.
8647 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8648 Alan Hayward <alan.hayward@arm.com>
8649 David Sherwood <david.sherwood@arm.com>
8651 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8652 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8653 (extract_range_from_binary_expr_1): Don't call
8654 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8656 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8657 Alan Hayward <alan.hayward@arm.com>
8658 David Sherwood <david.sherwood@arm.com>
8660 * target.def (preferred_vector_alignment): New hook.
8661 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8663 * doc/tm.texi: Regenerate.
8664 * targhooks.h (default_preferred_vector_alignment): Declare.
8665 * targhooks.c (default_preferred_vector_alignment): New function.
8666 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8668 (DR_TARGET_ALIGNMENT): New macro.
8669 (aligned_access_p): Update commentary.
8670 (vect_known_alignment_in_bytes): New function.
8671 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8673 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8674 Calculate the misalignment based on the target alignment rather than
8676 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8677 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8678 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8679 the target alignment, rather than masking the element misalignment
8680 with the number of elements in a vector. Also use the target
8681 alignment when calculating the maximum number of peels.
8682 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8683 instead of TYPE_ALIGN_UNIT.
8684 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8685 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8686 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8687 (vect_create_data_ref_ptr): Likewise.
8688 (vect_setup_realignment): Realign by ANDing with
8689 -DR_TARGET_MISALIGNMENT.
8690 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8691 the number of peels based on DR_TARGET_ALIGNMENT.
8692 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8693 with the guaranteed alignment boundary when deciding whether
8695 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8696 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8697 (ensure_base_align): Remove stmt_info parameter. Get the
8698 target base alignment from DR_TARGET_ALIGNMENT.
8699 (vectorizable_store): Update call accordingly. Interpret
8700 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8702 (vectorizable_load): Likewise.
8704 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8705 Alan Hayward <alan.hayward@arm.com>
8706 David Sherwood <david.sherwood@arm.com>
8708 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8709 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8710 (vect_enhance_data_refs_alignment): Likewise.
8712 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8714 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8715 error. Only quit immediately if parsing is complete.
8716 (BEGIN): Initialize fatal_err and parse_done.
8717 (begin fpu, end fpu): Check number of arguments.
8718 (begin arch, end arch): Likewise.
8719 (begin cpu, end cpu): Likewise.
8720 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8721 (optalias): Likewise.
8723 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8725 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8726 * config/arm/arm-isa.h: Delete. Move definitions to ...
8727 * arm-cpus.in: ... here. Use new feature and fgroup values.
8728 * config/arm/arm.c (arm_option_override): Use lower case for feature
8730 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8731 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8732 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8734 (print_isa_bits_for): New function.
8735 (gen_isa): New function.
8736 (gen_comm_data): Use print_isa_bits_for.
8737 (define feature): New keyword.
8738 (define fgroup): New keyword.
8739 * config/arm/t-arm (TM_H): Remove.
8740 (GTM_H): Add arm-isa.h.
8741 (arm-isa.h): Add rule to generate file.
8742 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8743 case for feature bit names.
8745 2017-09-22 Richard Biener <rguenther@suse.de>
8747 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8749 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8750 print of no dependency loops ...
8751 * graphite.c (graphite_transform_loops): ... here.
8752 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8754 (same_close_phi_node, remove_duplicate_close_phi,
8755 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8756 (canonicalize_loop_closed_ssa): ... here and simplify.
8757 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8758 (optimize_isl): Use dump_printf_loc to tell when we stopped
8759 optimizing because of an ISL timeout.
8761 2017-09-22 Richard Biener <rguenther@suse.de>
8763 PR tree-optimization/82291
8764 * tree-if-conv.c (predicate_mem_writes): Make sure to
8765 remove writes in blocks predicated with false.
8767 2017-09-22 Richard Biener <rguenther@suse.de>
8769 * sese.c: Include cfganal.h.
8770 (if_region_set_false_region): Remove.
8771 (create_if_region_on_edge): Likewise.
8772 (move_sese_in_condition): Re-implement without destroying
8775 2017-09-22 Richard Biener <rguenther@suse.de>
8777 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8778 Verify both BBs contain loop PHI nodes before dispatching to
8780 (graphite_regenerate_ast_isl): Do not recompute dominators,
8781 do not verify three times. Restructure for clarity.
8782 * graphite-scop-detection.c (same_close_phi_node,
8783 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8784 defined_in_loop_p, canonicalize_loop_closed_ssa,
8785 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8786 checking and SSA rewrite, move to ...
8787 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8788 (graphite_initialize): Do not pass in ctx, do not reset the
8789 SCEV cache, compute only dominators.
8790 (graphite_transform_loops): Allocate ISL ctx after
8791 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8792 Maintain post-dominators only around build_scops.
8793 * sese.c (if_region_set_false_region): Make static. Free
8794 and recompute dominators.
8795 (move_sese_in_condition): Assert we don't get called with
8796 post-dominators computed.
8797 * sese.h (if_region_set_false_region): Remove.
8799 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8801 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8802 mode attribute for TARGET_AVX512VL.
8804 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8806 * config/i386/i386.opt (mprefer-avx256): New option.
8807 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8809 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8810 for TARGET_PREFER_AVX256.
8811 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8813 2017-09-21 Jeff Law <law@redhat.com>
8815 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8816 Fix dump output if the only stack space is for pushed registers.
8818 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8820 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8823 2017-09-21 Martin Sebor <msebor@redhat.com>
8826 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8827 code (in C++) or code that triggers warnings.
8829 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8831 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8833 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8835 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8836 * sched-rgn.c: Ditto.
8837 * sel-sched-ir.c: Ditto.
8839 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8841 * toplev.h (set_random_seed): Adjust return type.
8842 * toplev.c (init_local_tick): Move eager initialization of random_seed
8843 to get_random_seed. Adjust comment.
8844 (init_random_seed): Inline to get_random_seed, delete.
8845 (get_random_seed): Initialize random_seed lazily.
8846 (set_random_seed): Do not return previous value.
8847 (print_switch_value): Do not call get_random_seed.
8849 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8851 * cgraph.c (delete_function_version): New, broken out from...
8852 (cgraph_node::delete_function_version): ...here. Rename to
8853 cgraph_node::delete_function_version_by_decl. Update all uses.
8854 (cgraph_node::remove): Call delete_function_version.
8856 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8859 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8860 VAR_DECLs to which addressable non-volatile parameters are mapped
8861 and for id->retvar after the return value assignment. Clear
8862 id->retval and id->retbnd after inlining.
8864 2017-09-21 Richard Biener <rguenther@suse.de>
8866 PR tree-optimization/82276
8867 PR tree-optimization/82244
8868 * tree-vrp.c (build_assert_expr_for): Set
8869 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8871 (remove_range_assertions): Revert earlier change.
8873 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8876 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8878 2017-09-21 Richard Biener <rguenther@suse.de>
8880 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8881 Restore valid IL after code generation errors.
8882 * graphite.c (graphite_transform_loops): Diagnose code
8883 generation issues as MSG_MISSED_OPTIMIZATION and continue
8884 with processing SCOPs.
8886 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8887 Alan Hayward <alan.hayward@arm.com>
8888 David Sherwood <david.sherwood@arm.com>
8890 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8891 rather than choosing between plus_constant and gen_rtx_<CODE>.
8892 * expr.c (emit_push_insn): Likewise.
8893 (expand_expr_real_2): Likewise.
8895 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8896 Alan Hayward <alan.hayward@arm.com>
8897 David Sherwood <david.sherwood@arm.com>
8899 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8901 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8902 Alan Hayward <alan.hayward@arm.com>
8903 David Sherwood <david.sherwood@arm.com>
8905 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8906 calling tree_to_uhwi.
8908 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8909 Alan Hayward <alan.hayward@arm.com>
8910 David Sherwood <david.sherwood@arm.com>
8912 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8913 INTEGER_CST rather than a negative test for ADDR_EXPR.
8915 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8916 Alan Hayward <alan.hayward@arm.com>
8917 David Sherwood <david.sherwood@arm.com>
8919 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8920 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8922 2017-09-21 Richard Biener <rguenther@suse.de>
8924 PR tree-optimization/71351
8925 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8926 graphite_create_new_loop_guard): Remove, fold remaining parts
8928 (translate_isl_ast_node_for): ... here and simplify.
8930 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8933 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8934 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8935 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8936 alternative always use QI mode, for -Os imov (=R,R) alternative
8937 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8940 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8941 Jeff Law <law@redhat.com>
8943 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8944 (allocate_stack_space): New function, partially extracted from
8946 (s390_emit_prologue): Track offset to most recent stack probe.
8947 Code to allocate space moved into allocate_stack_space.
8948 Dump actions when no stack is allocated.
8949 (s390_prologue_plus_offset): New function.
8950 (s390_emit_stack_probe): Likewise.
8952 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8954 * common.opt (Wa, Wl, Wp, g, gz=): Add
8956 (gno-column-info): Remove.
8957 (gcolumn-info): Drop RejectNegative.
8959 (gno-record-gcc-switches): Remove.
8960 (grecord-gcc-switches): Drop RejectNegative.
8961 (gno-split-dwarf): Remove.
8962 (gsplit-dwarf): Drop RejectNegative.
8963 (gno-strict-dwarf): Remove.
8964 (gstrict-dwarf): Drop RejectNegative.
8965 * config/darwin.opt (gfull, gused): Add RejectNegative.
8966 * dwarf2out.c (gen_producer_string): Drop
8967 gno-record-gcc-switches handler.
8968 * optc-gen.awk: Add g to prefixes with negative forms.
8969 * opts-common.c (remapping_prefix_p): New.
8970 (find_opt): Check it.
8971 (generate_canonical_option): Test g prefix.
8972 (option_map): Add -gno- mapping.
8973 (add_misspelling_candidates): Check remapping_prefix_p.
8975 2017-09-20 Jeff Law <law@redhat.com>
8977 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8978 thinko in stack clash protection support.
8980 * explow.c (compute_stack_clash_protection_loop_data): Use
8981 CONST_INT_P instead of explicit test. Verify object is a
8982 CONST_INT_P before looking at INTVAL.
8983 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8984 instead of explicit test.
8986 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8989 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8990 address instead of to r1 and r11.
8992 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8994 * config.gcc: Support "knm".
8995 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8996 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8998 * config/i386/i386.c (m_KNM): Define.
8999 (processor_target_table): Add "knm".
9001 (ix86_option_override_internal): Add "knm".
9002 (ix86_issue_rate): Add PROCESSOR_KNM.
9003 (ix86_adjust_cost): Ditto.
9004 (ia32_multipass_dfa_lookahead): Ditto.
9005 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9006 (fold_builtin_cpu): Add M_INTEL_KNM.
9007 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9008 (processor_type): Add PROCESSOR_KNM.
9009 * config/i386/x86-tune.def: Add m_KNM.
9010 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9012 2017-09-20 Richard Biener <rguenther@suse.de>
9014 PR tree-optimization/80213
9015 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9016 are allowed in empty BBs as well.
9017 (canonicalize_loop_closed_ssa): Also look for other complex
9019 (scop_detection::get_sese): Include the loop-closed PHI block
9021 (scop_detection::merge_sese): Remove code adding extra blocks.
9022 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9023 (build_scops): Assert the final returned scop is invalid.
9025 2017-09-20 Richard Biener <rguenther@suse.de>
9027 PR tree-optimization/82264
9028 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9030 (vn_phi_lookup): Likewise.
9031 (vn_phi_insert): Likewise.
9033 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9035 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9036 that fits into uhwi or shwi, add DW_AT_const_value regardless
9037 of early_dwarf without going through RTL, using add_AT_unsigned
9040 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9041 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9042 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9043 (DEBUG_MACRO_SECTION): Likewise.
9044 (DEBUG_LTO_MACRO_SECTION): Likewise.
9045 (DEBUG_STR_DWO_SECTION): Likewise.
9046 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9047 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9048 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9049 (DEBUG_LTO_LINE_STR_SECTION): Define.
9050 (init_sections_and_labels): Initialize debug_line_str_section
9051 variable. Initialize debug_loc_section for -gdwarf-5 to
9052 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9054 2017-09-20 Richard Biener <rguenther@suse.de>
9056 * graphite-sese-to-poly.c (extract_affine): Properly handle
9057 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9059 2017-09-20 Richard Biener <rguenther@suse.de>
9061 PR tree-optimization/81373
9062 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9063 Force SESE live-out defs to be handled even if they are
9066 2017-09-19 Jeff Law <law@redhat.com>
9068 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9069 nothing for stack adjustments with REG_STACK_CHECK.
9070 * sched-deps.c (parse_add_or_inc): Reject insns with
9071 REG_STACK_CHECK from dependency breaking.
9072 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9073 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9074 * reg-notes.def (STACK_CHECK): New note.
9076 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9077 (ix86_expand_prologue): Dump stack clash info as needed.
9078 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9080 * function.c (dump_stack_clash_frame_info): New function.
9081 * function.h (dump_stack_clash_frame_info): Prototype.
9082 (enum stack_clash_probes): New enum.
9084 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9085 flag_stack_clash_protection.
9086 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9087 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9088 (arm_frame_pointer_required): Likewise.
9089 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9090 (ia64_expand_prologue): Likewise.
9091 * config/mips/mips.c (mips_expand_prologue): Likewise.
9092 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9093 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9094 (sparc_flat_expand_prologue): Likewise.
9095 * config/spu/spu.c (spu_expand_prologue): Likewise.
9097 * explow.c: Include "params.h".
9098 (anti_adjust_stack_and_probe_stack_clash): New function.
9099 (get_stack_check_protect): Likewise.
9100 (compute_stack_clash_protection_loop_data): Likewise.
9101 (emit_stack_clash_protection_loop_start): Likewise.
9102 (emit_stack_clash_protection_loop_end): Likewise.
9103 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9104 Use anti_adjust_stack_and_probe_stack_clash.
9105 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9106 (emit_stack_clash_protection_loop_start): Likewise.
9107 (emit_stack_clash_protection_loop_end): Likewise.
9108 * rtl.h (get_stack_check_protect): Prototype.
9109 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9110 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9111 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9113 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9115 * doc/tm.texi: Rebuilt.
9116 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9117 get_stack_check_protect.
9118 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9119 * config/arm/arm.c (arm_expand_prologue): Likewise.
9120 (arm_frame_pointer_required): Likewise.
9121 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9122 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9123 * config/mips/mips.c (mips_expand_prologue): Likewise.
9124 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9125 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9126 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9127 (sparc_flat_expand_prologue): Likewise.
9129 * common.opt (-fstack-clash-protection): New option.
9130 * flag-types.h (enum stack_check_type): Note difference between
9131 -fstack-check= and -fstack-clash-protection.
9132 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9133 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9134 * toplev.c (process_options): Issue warnings/errors for cases
9135 not handled with -fstack-clash-protection.
9136 * doc/invoke.texi (-fstack-clash-protection): Document new option.
9137 (-fstack-check): Note additional problem with -fstack-check=generic.
9138 Note that -fstack-check is primarily for Ada and refer users
9139 to -fstack-clash-protection for stack-clash-protection.
9140 Document new params for stack clash protection.
9142 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
9144 * config/i386/i386.c (ix86_split_long_move): Do not handle
9145 address used for LEA in a special way.
9147 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
9149 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
9151 2017-09-19 Martin Sebor <msebor@redhat.com>
9154 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
9155 of incompatible types.
9157 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
9159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9160 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
9161 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9162 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
9164 2017-09-19 Richard Biener <rguenther@suse.de>
9166 PR tree-optimization/82244
9167 * tree-vrp.c (remove_range_assertions): Do not propagate
9168 a constant to abnormals but replace the assert with a copy.
9170 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
9172 PR rtl-optimization/57878
9173 PR rtl-optimization/68988
9174 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
9175 avoidance test involving non_reload_pseudos. Move frequency test
9176 below the general fragmentation avoidance test.
9178 2017-09-19 Richard Biener <rguenther@suse.de>
9180 PR tree-optimization/69728
9181 * graphite-sese-to-poly.c (schedule_error): New global.
9182 (add_loop_schedule): Handle empty domain by failing the
9184 (build_original_schedule): Handle schedule_error.
9186 2017-09-19 Richard Biener <rguenther@suse.de>
9188 * graphite-scop-detection.c (scop_detection::can_represent_loop):
9189 Do not iterate to sibling loops but only to siblings of inner
9192 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
9195 * config/m68k/m68k.md (moveq feeding equality comparison): Check
9196 that the registers are different.
9198 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
9200 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
9201 to processor_model and "amdfam17h" to arch_names_table.
9202 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
9204 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9207 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
9209 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9210 Alan Hayward <alan.hayward@arm.com>
9211 David Sherwood <david.sherwood@arm.com>
9213 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
9215 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
9217 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
9218 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
9219 vect_schedule_slp_instance.
9220 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
9221 Update call to vect_slp_analyze_node_operations. Simplify return
9223 (vect_slp_analyze_bb_1): Update call accordingly.
9224 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
9225 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
9226 (vect_schedule_slp): Update call accordingly.
9228 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9229 Alan Hayward <alan.hayward@arm.com>
9230 David Sherwood <david.sherwood@arm.com>
9232 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
9233 with types that aren't in fact scalar.
9235 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9237 * tree-vect-slp.c (vect_record_max_nunits): New function,
9239 (vect_build_slp_tree_1): ...here.
9240 (vect_build_slp_tree_2): Call it for phis too.
9242 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9244 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9245 to vect_get_vec_def_for_operand when getting the mask operand.
9247 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9248 Alan Hayward <alan.hayward@arm.com>
9249 David Sherwood <david.sherwood@arm.com>
9251 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9254 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9255 Alan Hayward <alan.hayward@arm.com>
9256 David Sherwood <david.sherwood@arm.com>
9258 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9259 calculation for vector booleans.
9261 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9262 Alan Hayward <alan.hayward@arm.com>
9263 David Sherwood <david.sherwood@arm.com>
9265 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9267 (vect_transform_stmt): ...here.
9268 (vect_analyze_stmt): Use it instead of calling
9269 vectorizable_live_operation directly.
9271 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9273 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9274 non-SIMT targets in acc vector loops.
9276 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9278 * configure.ac: Add arc and check if assembler supports gdwarf2.
9279 * configure: Regenerate.
9281 2017-09-18 Richard Biener <rguenther@suse.de>
9283 PR tree-optimization/82220
9284 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9285 epilogue niters from the min_profitable_iters compute.
9287 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9290 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9291 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9292 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9294 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9297 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9298 switching to a new text section.
9300 2017-09-18 Richard Biener <rguenther@suse.de>
9302 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9304 (build_alias_set): Reject aliases with no access function.
9306 2017-09-18 Richard Biener <rguenther@suse.de>
9308 PR tree-optimization/79622
9309 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9311 (build_cross_bb_scalars_use): Likewise.
9313 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9315 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9317 2017-09-18 Alan Modra <amodra@gmail.com>
9320 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9321 stack_pointer_rtx for count 0. Update comments. Break up
9322 large rtl expression.
9324 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9326 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9327 Increase to 20 bytes.
9328 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9329 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9330 or avx version of the stub.
9332 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9335 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9336 compute the minimum stack alignment. Also update preferred stack
9337 boundary for leaf functions.
9339 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9341 PR tree-optimization/82228
9342 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9345 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9347 * common/config/nds32/nds32-common.c
9348 (nds32_option_optimization_table): Refine formatting.
9349 (nds32_option_optimization_table): Use -fsched-pressure and
9350 -fomit-frame-pointer for specific optimization level.
9352 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9354 * config/nds32/nds32.c: Refine formatting and comments.
9355 * config/nds32/nds32.h: Likewise.
9356 * config/nds32/nds32.md: Likewise.
9357 * config/nds32/nds32-cost.c: Likewise.
9358 * config/nds32/nds32-isr.c: Likewise.
9359 * config/nds32/nds32-md-auxiliary.c: Likewise.
9360 * config/nds32/nds32-multiple.md: Likewise.
9361 * config/nds32/nds32-predicates.c: Likewise.
9363 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9364 Jakub Jelinek <jakub@redhat.com>
9366 Add support for -std=c++2a.
9367 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9369 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9371 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9374 * doc/extend.texi (Common Function Attributes): Add
9375 references to ARM, AArch64, and S/390 specific attributes.
9376 (Function Specific Option Pragmas): Add AArch64 and S/390
9377 to list of back ends that support the target pragma.
9379 2017-09-15 Nathan Sidwell <nathan@acm.org>
9381 * doc/standards.texi: Fix C++17 description. Update URLs for
9384 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9386 * common.opt (Wcast-align=strict): New warning option.
9387 * doc/invoke.texi: Document -Wcast-align=strict.
9389 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9391 * cgraph.h (cgraph_thunk_info): Add comments.
9392 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9393 assert for VIRTUAL_* arguments stricter.
9395 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9397 PR tree-optimization/71026
9398 * match.pd: Move RDIV patterns from fold-const.c
9399 * fold-const.c (distribute_real_division): Removed.
9400 (fold_binary_loc): Remove calls to distribute_real_divison.
9402 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9404 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9405 c++1z and gnu++1z as deprecated. Change other references to
9406 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9407 Change -Wc++1z-compat to -Wc++17-compat.
9408 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9409 * dwarf2out.c (highest_c_language): Handle C++17.
9410 (gen_compile_unit_die): Likewise.
9412 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9414 PR rtl-optimization/82192
9415 * combine.c (make_extraction): Don't look through non-paradoxical
9416 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9419 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9420 Alan Hayward <alan.hayward@arm.com>
9421 David Sherwood <david.sherwood@arm.com>
9423 * target.def (function_arg_offset): New hook.
9424 * targhooks.h (default_function_arg_offset): Declare.
9425 * targhooks.c (default_function_arg_offset): New function.
9426 * function.c (locate_and_pad_parm): Use
9427 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9428 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9429 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9430 * doc/tm.texi: Regenerate.
9431 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9432 * config/spu/spu.c (spu_function_arg_offset): New function.
9433 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9434 * system.h (FUNCTION_ARG_OFFSET): Poison.
9436 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9437 Alan Hayard <alan.hayward@arm.com>
9438 David Sherwood <david.sherwood@arm.com>
9440 * target.def (truly_noop_truncation): New hook.
9441 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
9442 than TRULY_NOOP_TRUNCATION.
9443 * hooks.h (hook_bool_uint_uint_true): Declare.
9444 * hooks.c (hook_bool_uint_uint_true): New function.
9445 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
9446 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
9447 * doc/tm.texi: Regenerate.
9448 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
9449 rather than TRULY_NOOP_TRUNCATION in comments.
9450 (simplify_comparison): Likewise.
9451 (record_truncated_value): Likewise.
9452 * expmed.c (extract_bit_field_1): Likewise.
9453 (extract_split_bit_field): Likewise.
9454 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
9455 instead of TRULY_NOOP_TRUNCATION.
9456 * function.c (assign_parm_setup_block): Likewise.
9457 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
9458 * rtlhooks.c: Include target.h.
9459 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
9460 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
9461 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
9462 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
9463 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
9464 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
9465 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
9466 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
9467 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
9468 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
9469 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
9470 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
9471 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
9472 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
9473 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
9474 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
9475 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
9476 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
9477 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
9478 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
9479 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
9480 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
9481 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
9482 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
9483 * config/mips/mips.c (mips_truly_noop_truncation): New function.
9484 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9485 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9486 rather than TRULY_NOOP_TRUNCATION in comments.
9487 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
9488 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
9489 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
9490 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
9491 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
9492 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
9493 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
9494 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
9495 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
9496 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
9497 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
9498 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9499 rather than TRULY_NOOP_TRUNCATION in comments.
9500 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
9501 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
9502 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
9503 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
9504 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
9505 TRULY_NOOP_TRUNCATION condition.
9506 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
9507 (TRULY_NOOP_TRUNCATION): Delete.
9508 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
9509 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
9510 * config/spu/spu.c (spu_truly_noop_truncation): New function.
9511 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9512 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
9513 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
9514 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
9515 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9516 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9517 rather than TRULY_NOOP_TRUNCATION in comments.
9518 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
9519 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
9520 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
9521 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
9522 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
9523 * system.h (TRULY_NOOP_TRUNCATION): Poison.
9525 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
9528 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
9529 (*cmp_ior): Likewise.
9530 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
9531 (*ior_scc_scc_cmp): Likewise.
9532 (*and_scc_scc): Likewise.
9533 (*and_scc_scc_cmp): Likewise.
9535 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9536 Alan Hayard <alan.hayward@arm.com>
9537 David Sherwood <david.sherwood@arm.com>
9539 * target.def (can_change_mode_class): New hook.
9540 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
9541 (hard_regno_nregs): Likewise.
9542 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
9543 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
9544 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
9545 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
9546 (LOAD_EXTEND_OP): Update accordingly.
9547 * doc/tm.texi: Regenerate.
9548 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
9549 CANNOT_CHANGE_MODE_CLASS.
9550 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
9551 (REG_CAN_CHANGE_MODE_P): ...this new macro.
9552 * combine.c (simplify_set): Update accordingly.
9553 * emit-rtl.c (validate_subreg): Likewise.
9554 * recog.c (general_operand): Likewise.
9555 * regcprop.c (mode_change_ok): Likewise.
9556 * reload1.c (choose_reload_regs): Likewise.
9557 (inherit_piecemeal_p): Likewise.
9558 * rtlanal.c (simplify_subreg_regno): Likewise.
9559 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
9560 instead of CANNOT_CHANGE_MODE_CLASS.
9561 (reload_cse_simplify_operands): Likewise.
9562 * reload.c (push_reload): Use targetm.can_change_mode_class
9563 instead of CANNOT_CHANGE_MODE_CLASS.
9564 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
9565 REG_CANNOT_CHANGE_MODE_P.
9566 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9567 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
9568 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9569 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9570 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9571 (arm_can_change_mode_class): New function.
9572 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9573 than CANNOT_CHANGE_MODE_CLASS in comments.
9574 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9575 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9576 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9577 (ix86_can_change_mode_class): ...this new function, inverting the
9578 sense of the return value.
9579 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9580 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9581 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9582 (ia64_can_change_mode_class): New function.
9583 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9584 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9585 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9586 (m32c_can_change_mode_class): ...this new function, inverting the
9587 sense of the return value.
9588 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9589 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9590 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9591 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9592 (mips_can_change_mode_class): ...this new function, inverting the
9593 sense of the return value.
9594 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9595 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9596 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9597 (msp430_can_change_mode_class): New function.
9598 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9599 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9600 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9601 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9602 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9603 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9604 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9605 (pa_cannot_change_mode_class): Replace with...
9606 (pa_can_change_mode_class): ...this new function, inverting the
9607 sense of the return value.
9608 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9609 than CANNOT_CHANGE_MODE_CLASS in comments.
9610 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9611 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9612 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9613 (pdp11_cannot_change_mode_class): Replace with...
9614 (pdp11_can_change_mode_class): ...this new function, inverting the
9615 sense of the return value.
9616 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9617 * config/powerpcspe/powerpcspe-protos.h
9618 (rs6000_cannot_change_mode_class_ptr): Delete.
9619 * config/powerpcspe/powerpcspe.c
9620 (rs6000_cannot_change_mode_class_ptr): Delete.
9621 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9622 (rs6000_option_override_internal): Assign to
9623 targetm.can_change_mode_class instead of
9624 rs6000_cannot_change_mode_class_ptr.
9625 (rs6000_cannot_change_mode_class): Replace with...
9626 (rs6000_can_change_mode_class): ...this new function, inverting the
9627 sense of the return value.
9628 (rs6000_debug_cannot_change_mode_class): Replace with...
9629 (rs6000_debug_can_change_mode_class): ...this new function.
9630 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9631 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9632 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9633 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9634 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9636 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9637 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9638 (rs6000_option_override_internal): Assign to
9639 targetm.can_change_mode_class instead of
9640 rs6000_cannot_change_mode_class_ptr.
9641 (rs6000_cannot_change_mode_class): Replace with...
9642 (rs6000_can_change_mode_class): ...this new function, inverting the
9643 sense of the return value.
9644 (rs6000_debug_cannot_change_mode_class): Replace with...
9645 (rs6000_debug_can_change_mode_class): ...this new function.
9646 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9647 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9648 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9649 (s390_can_change_mode_class): ...this new function, inverting the
9650 sense of the return value.
9651 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9652 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9653 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9654 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9655 (sh_cannot_change_mode_class): Replace with...
9656 (sh_can_change_mode_class): ...this new function, inverting the
9657 sense of the return value.
9658 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9659 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9660 (sparc_can_change_mode_class): New function.
9661 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9662 * config/spu/spu.c (spu_can_change_mode_class): New function.
9663 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9664 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9665 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9666 (visium_can_change_mode_class): New function.
9667 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9669 2017-09-15 Richard Biener <rguenther@suse.de>
9671 PR tree-optimization/82217
9672 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9673 but not undefined case.
9675 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9678 * postreload.c (reload_cse_simplify_operands): Skip
9679 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9681 2017-09-15 Richard Biener <rguenther@suse.de>
9683 PR tree-optimization/68823
9684 * graphite-scop-detection.c (build_alias_set): If we have a
9685 possible dependence check whether we can handle them by just
9686 looking at the DRs DR_ACCESS_FNs.
9687 (build_scops): If build_alias_set fails, fail the SCOP.
9689 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9691 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9692 to support float128 built-in functions that require the ISA 3.0
9694 (BU_FLOAT128_3_HW): Likewise.
9695 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9697 (FMAF128): Likewise.
9699 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9700 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9701 floating point instructions.
9702 (rs6000_invalid_builtin): Likewise.
9703 (rs6000_builtin_mask_names): Likewise.
9704 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9705 (RS6000_BTM_FLOAT128_HW): Likewise.
9706 (RS6000_BTM_COMMON): Likewise.
9707 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9709 * doc/extend.texi (RS/6000 built-in functions): Document the
9710 IEEE 128-bit floating point square root and fused multiply-add
9713 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9715 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9716 reg (r2) isn't in the set of registers defined in the prologue.
9718 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9719 Alan Hayward <alan.hayward@arm.com>
9720 David Sherwood <david.sherwood@arm.com>
9722 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9723 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9724 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9725 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9726 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9728 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9729 max_vectorization_factor.
9730 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9732 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9733 Alan Hayward <alan.hayward@arm.com>
9734 David Sherwood <david.sherwood@arm.com>
9736 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9737 (vect_worthwhile_without_simd_p): Declare.
9738 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9739 (vectorizable_reduction): Use it.
9740 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9741 (vectorizable_operation): Likewise.
9743 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9744 Alan Hayward <alan.hayward@arm.com>
9745 David Sherwood <david.sherwood@arm.com>
9747 * tree-vectorizer.h (vect_get_num_copies): New function.
9748 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9749 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9750 (vectorizable_induction): Likewise.
9751 (vectorizable_live_operation): Likewise.
9752 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9753 (vectorizable_bswap): Likewise.
9754 (vectorizable_call): Likewise.
9755 (vectorizable_conversion): Likewise.
9756 (vectorizable_assignment): Likewise.
9757 (vectorizable_shift): Likewise.
9758 (vectorizable_operation): Likewise.
9759 (vectorizable_store): Likewise.
9760 (vectorizable_load): Likewise.
9761 (vectorizable_condition): Likewise.
9762 (vectorizable_comparison): Likewise.
9763 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9765 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9766 Alan Hayward <alan.hayward@arm.com>
9767 David Sherwood <david.sherwood@arm.com>
9769 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9770 of vect_init_vector.
9772 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9773 Alan Hayward <alan.hayward@arm.com>
9774 David Sherwood <david.sherwood@arm.com>
9776 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9777 an inline wrapper that provides a location.
9778 (gimple_build_vector): Likewise.
9779 * gimple-fold.c (gimple_build_vector_from_val): New function.
9780 (gimple_build_vector): Likewise.
9781 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9782 functions to build the initial value. Always return a gimple value.
9783 (get_initial_defs_for_reduction): Likewise. Only compute
9785 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9786 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9787 (vectorizable_induction): Use gimple_build_vector rather than
9790 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9791 Alan Hayward <alan.hayward@arm.com>
9792 David Sherwood <david.sherwood@arm.com>
9794 * target.h (vec_perm_indices): New typedef.
9795 (auto_vec_perm_indices): Likewise.
9796 * optabs-query.h: Include target.h
9797 (can_vec_perm_p): Take a vec_perm_indices *.
9798 * optabs-query.c (can_vec_perm_p): Likewise.
9799 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9800 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9801 * tree-vect-generic.c (lower_vec_perm): Likewise.
9802 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9803 (vect_grouped_load_supported): Likewise.
9804 (vect_shift_permute_load_chain): Likewise.
9805 (vect_permute_store_chain): Use auto_vec_perm_indices.
9806 (vect_permute_load_chain): Likewise.
9807 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9808 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9809 Update uses of can_vec_perm_p.
9810 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9811 mode with a number of elements. Take a vec_perm_indices *.
9812 (vect_create_epilog_for_reduction): Update accordingly.
9813 Use auto_vec_perm_indices.
9814 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9815 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9816 (vect_transform_slp_perm_load): Likewise.
9817 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9818 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9819 (vect_gen_perm_mask_checked): Likewise.
9820 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9821 (vect_gen_perm_mask_checked): Likewise.
9822 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9823 (vectorizable_store): Likewise.
9824 (vectorizable_load): Likewise.
9825 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9826 (vectorizable_bswap): Likewise.
9828 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9829 Alan Hayward <alan.hayward@arm.com>
9830 David Sherwood <david.sherwood@arm.com>
9832 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9833 * tree.c (build_vector): Likewise.
9834 (build_vector_from_ctor): Update accordingly.
9835 (build_vector_from_val): Likewise.
9836 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9837 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9838 * tree-vect-generic.c (add_rshift): Likewise.
9839 (expand_vector_divmod): Likewise.
9840 (optimize_vector_constructor): Likewise.
9841 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9842 (vect_transform_slp_perm_load): Likewise.
9843 (vect_schedule_slp_instance): Likewise.
9844 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9845 (vectorizable_call): Likewise.
9846 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9847 * expmed.c (make_tree): Likewise.
9848 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9849 a vector passed to build_vector.
9850 (fold_convert_const): Likewise.
9851 (exact_inverse): Likewise.
9852 (fold_ternary_loc): Likewise.
9853 (fold_relational_const): Likewise.
9854 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9855 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9856 (const_unop): Likewise. Store the reduction accumulator in a
9857 variable rather than an array.
9858 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9859 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9860 the new vector, rather than constructing it after the input arrays.
9861 (native_interpret_vector): Use auto_vec<tree> when building
9862 a vector passed to build_vector. Add elements in order.
9863 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9864 auto_vec<tree> when building a vector passed to build_vector.
9865 (vect_create_epilog_for_reduction): Likewise.
9866 (vectorizable_induction): Likewise.
9867 (get_initial_def_for_reduction): Likewise. Fix indentation of
9869 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9871 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9872 passed to build_vector.
9874 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9875 Alan Hayward <alan.hayward@arm.com>
9876 David Sherwood <david.sherwood@arm.com>
9878 * tree-core.h (tree_base::u): Add an "nelts" field.
9879 (tree_vector): Use VECTOR_CST_NELTS as the length.
9880 * tree.c (tree_size): Likewise.
9881 (make_vector): Initialize VECTOR_CST_NELTS.
9882 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9883 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9884 TYPE_VECTOR_SUBPARTS.
9885 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9886 as the number of units, setting it from VECTOR_CST_NELTS.
9887 (const_vector_from_tree): Likewise.
9888 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9889 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9890 (fold_negate_expr_1): Likewise.
9891 (fold_convert_const): Likewise.
9892 (const_binop): Likewise. Differentiate the number of output and
9894 (const_unop): Likewise.
9895 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9896 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9897 in cases that did the opposite.
9899 2017-09-14 Richard Biener <rguenther@suse.de>
9901 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9904 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9906 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9908 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9911 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9912 if and where to split a bb, except for splitting before debug insn
9913 sequences followed by non-label real insn. Delete debug insns
9914 in between basic blocks.
9916 * combine.c (make_compound_operation_int): Formatting fixes.
9918 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9919 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9920 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9921 * config/sol2.h (LINK_EH_SPEC): Likewise.
9922 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9923 * config/s390/linux.h (LINK_SPEC): Likewise.
9924 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9925 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9926 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9927 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9928 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9929 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9930 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9931 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9933 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9935 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9937 (ENDFILE_LINUX_SPEC): Likewise.
9938 (LINK_EH_SPEC): Likewise.
9939 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9940 (LINK_OS_LINUX_SPEC32): Likewise.
9941 (LINK_OS_LINUX_SPEC64): Likewise.
9942 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9943 (LINK_OS_LINUX_SPEC): Likewise.
9945 2017-09-13 Martin Liska <mliska@suse.cz>
9948 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9949 CASE_HIGH is NULL_TREE.
9951 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9952 Alan Hayward <alan.hayward@arm.com>
9953 David Sherwood <david.sherwood@arm.com>
9955 * target.def (secondary_memory_needed): New hook.
9956 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9957 instead of SECONDARY_MEMORY_NEEDED.
9958 (secondary_memory_needed_mode): Likewise.
9959 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9960 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9961 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9962 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9963 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9964 * doc/tm.texi: Regenerate.
9965 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9966 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9967 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9968 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9969 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9970 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9971 mode argument first and change the reg_class arguments to reg_class_t.
9972 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9973 Make static. Update the call to inline_secondary_memory_needed.
9974 (ix86_register_move_cost): Update the call to
9975 inline_secondary_memory_needed.
9976 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9977 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9979 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9980 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9982 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9983 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9984 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9985 and match hook interface. Add comment from mips.h.
9986 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9987 * config/mmix/mmix.md (truncdfsf2): Refer to
9988 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9990 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9991 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9992 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9993 (pa_secondary_memory_needed): New function.
9994 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9995 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9996 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9997 (pdp11_secondary_memory_needed): Make static and match hook interface.
9998 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9999 * config/powerpcspe/powerpcspe-protos.h
10000 (rs6000_secondary_memory_needed_ptr): Delete.
10001 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10003 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10004 (rs6000_option_override_internal): Assign to
10005 targetm.secondary_memory_needed rather than
10006 rs6000_secondary_memory_needed_ptr.
10007 (rs6000_secondary_memory_needed): Match hook interface.
10008 (rs6000_debug_secondary_memory_needed): Likewise.
10009 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10010 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10011 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10012 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10013 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10014 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10016 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10017 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10018 (rs6000_option_override_internal): Assign to
10019 targetm.secondary_memory_needed rather than
10020 rs6000_secondary_memory_needed_ptr.
10021 (rs6000_secondary_memory_needed): Match hook interface.
10022 (rs6000_debug_secondary_memory_needed): Likewise.
10023 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10024 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10025 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10026 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10027 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10028 (sparc_secondary_memory_needed): New function.
10029 * lra-constraints.c (check_and_process_move): Refer to
10030 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10032 (curr_insn_transform): Likewise.
10033 (process_alt_operands): Use targetm.secondary_memory_needed
10034 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10035 (check_secondary_memory_needed_p): Likewise.
10036 (choose_split_class): Likewise.
10037 * reload.c: Unconditionally include code that was previously
10038 conditional on SECONDARY_MEMORY_NEEDED.
10039 (push_secondary_reload): Use targetm.secondary_memory_needed
10040 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10041 (push_reload): Likewise.
10042 * reload1.c: Unconditionally include code that was previously
10043 conditional on SECONDARY_MEMORY_NEEDED.
10044 (choose_reload_regs): Use targetm.secondary_memory_needed
10045 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10046 (gen_reload): Likewise.
10047 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10049 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10050 Alan Hayward <alan.hayward@arm.com>
10051 David Sherwood <david.sherwood@arm.com>
10053 * target.def (secondary_memory_needed_mode): New hook:
10054 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10055 * targhooks.h (default_secondary_memory_needed_mode): New function.
10056 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10057 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10058 * doc/tm.texi: Regenerate.
10059 * lra-constraints.c (check_and_process_move): Use
10060 targetm.secondary_memory_needed_mode instead of
10061 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10062 (curr_insn_transform): Likewise.
10063 * reload.c (get_secondary_mem): Likewise.
10064 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10065 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10067 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10068 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10069 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10070 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10071 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10073 * config/powerpcspe/powerpcspe-protos.h
10074 (rs6000_secondary_memory_needed_mode): Delete.
10075 * config/powerpcspe/powerpcspe.c
10076 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10077 (rs6000_secondary_memory_needed_mode): Make static.
10078 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10079 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10081 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10083 (rs6000_secondary_memory_needed_mode): Make static.
10084 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10085 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10086 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10087 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10088 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10090 (sparc_secondary_memory_needed_mode): New function.
10091 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10093 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10095 * config/aarch64/constraints.md (Umq): New constraint.
10096 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10098 (mov<mode>): Update condition.
10100 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10102 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10103 when bitposition is the same.
10105 2017-09-13 Richard Biener <rguenther@suse.de>
10107 * dwarf2out.c (output_die_symbol): Remove.
10108 (output_die): Do not output a DIEs symbol.
10110 2017-09-13 Richard Biener <rguenther@suse.de>
10112 PR middle-end/82128
10113 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10114 default-def to avoid breaking iterator update with the weird
10115 interaction with cgraph_update_edges_for_call_stmt_node.
10117 2017-09-13 Richard Biener <rguenther@suse.de>
10119 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10120 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10121 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10122 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10124 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10126 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10127 Disable pc relative literal load irrespective of
10128 TARGET_FIX_ERR_A53_84341 for default.
10130 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10132 * config/sparc/sparc.c (output_return): Output the source location of
10133 the insn in the delay slot, if any.
10134 (output_sibcall): Likewise.
10136 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
10139 * common.opt (-static-pie): New alias.
10140 (shared): Negate static-pie.
10141 (-no-pie): Update help text.
10143 (static-pie): New option.
10144 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
10145 -static-pie support.
10146 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
10147 (LINK_EH_SPEC): Likewise.
10148 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10149 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10150 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10151 * gcc.c (LINK_COMMAND_SPEC): Likewise.
10152 (init_gcc_specs): Likewise.
10153 (init_spec): Likewise.
10154 (display_help): Update help message for -pie.
10155 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
10158 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
10160 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
10161 (movdi_aarch64): Likewise.
10162 (movti_aarch64): Likewise.
10164 2017-09-12 Simon Wright <simon@pushface.org>
10167 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
10168 calculation of the minor version, always output as 0.
10170 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10173 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10174 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
10175 on it early, rather than manual conversion late. For
10176 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
10177 instead of performing manual conversion.
10179 2017-09-12 Carl Love <cel@us.ibm.com>
10181 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
10182 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
10184 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10185 VMULOSW): Add definitions.
10186 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10187 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10188 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10189 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10190 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10192 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10194 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
10196 (movti_aarch64): Likewise.
10197 (movdf_aarch64): Likewise.
10198 (movtf_aarch64): Likewise.
10199 (load_pairdi): Likewise.
10200 (store_pairdi): Likewise.
10201 (load_pairdf): Likewise.
10202 (store_pairdf): Likewise.
10203 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
10204 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
10205 (ldr_got_small_<mode>): Likewise.
10206 (ldr_got_small_28k_<mode>): Likewise.
10207 (ldr_got_tiny): Likewise.
10208 * config/aarch64/iterators.md (ldst_sz): New.
10209 (ldpstp_sz): Likewise.
10210 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
10212 (thunderx_load): Split load_8 to load_16.
10213 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
10215 (thunderx2t99_storepair_basic): Split store_8 to store_16.
10216 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
10217 (xgene1_store_pair): Split store_8 to store_16.
10218 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
10219 (falkor_st_0_st_sd): Split store_8 to store_16.
10221 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10223 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
10224 and store1/2/3/4 to store_4/8/12/16.
10225 * config/aarch64/aarch64.md: Update for rename.
10226 * config/arm/arm.md: Likewise.: Likewise.
10227 * config/arm/arm.c: Likewise.
10228 * config/arm/thumb1.md: Likewise.
10229 * config/arm/thumb2.md: Likewise.
10230 * config/arm/vfp.md: Likewise.
10231 * config/arm/arm-generic.md: Likewise.
10232 * config/arm/arm1020e.md: Likewise.
10233 * config/arm/arm1026ejs.md: Likewise.
10234 * config/arm/arm1136jfs.md: Likewise.
10235 * config/arm/arm926ejs.md: Likewise.
10236 * config/arm/cortex-a15.md: Likewise.
10237 * config/arm/cortex-a17.md: Likewise.
10238 * config/arm/cortex-a5.md: Likewise.
10239 * config/arm/cortex-a53.md: Likewise.
10240 * config/arm/cortex-a57.md: Likewise.
10241 * config/arm/cortex-a7.md: Likewise.
10242 * config/arm/cortex-a8.md: Likewise.
10243 * config/arm/cortex-a9.md: Likewise.
10244 * config/arm/cortex-m4.md: Likewise.
10245 * config/arm/cortex-m7.md: Likewise.
10246 * config/arm/cortex-r4.md: Likewise.
10247 * config/arm/exynos-m1.md: Likewise.
10248 * config/arm/fa526.md: Likewise.
10249 * config/arm/fa606te.md: Likewise.
10250 * config/arm/fa626te.md: Likewise.
10251 * config/arm/fa726te.md: Likewise.
10252 * config/arm/fmp626.md: Likewise.
10253 * config/arm/iwmmxt.md: Likewise.
10254 * config/arm/ldmstm.md: Likewise.
10255 * config/arm/marvell-pj4.md: Likewise.
10256 * config/arm/xgene1.md: Likewise.
10257 * config/aarch64/thunderx.md: Likewise.
10258 * config/aarch64/thunderx2t99.md: Likewise.
10259 * config/aarch64/falkor.md: Likewise.
10261 2017-09-12 Martin Liska <mliska@suse.cz>
10263 * attribs.c (private_lookup_attribute): New function.
10264 * attribs.h (private_lookup_attribute): Declared here.
10265 (lookup_attribute): Called from this place.
10267 2017-09-12 Richard Biener <rguenther@suse.de>
10269 PR tree-optimization/82157
10270 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10271 stmts with side-effects.
10273 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10274 Alan Hayward <alan.hayward@arm.com>
10275 David Sherwood <david.sherwood@arm.com>
10277 * target.def (hard_regno_nregs): New hook.
10278 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10279 * targhooks.h (default_hard_regno_nregs): Declare.
10280 * targhooks.c (default_hard_regno_nregs): New function.
10281 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10282 (TARGET_HARD_REGNO_NREGS): ...this hook.
10283 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10284 (CLASS_MAX_NREGS): Likewise.
10285 * doc/tm.texi: Regenerate.
10286 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10287 instead of HARD_REGNO_NREGS.
10288 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10289 HARD_REGNO_NREGS in the comment.
10290 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10291 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10292 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10293 Return an unsigned int.
10294 (TARGET_HARD_REGNO_NREGS): Redefine.
10295 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10296 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10297 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10298 (arc_hard_regno_nregs): New function.
10299 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10300 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10301 (arm_hard_regno_nregs): New function.
10302 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10303 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10304 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10305 (TARGET_HARD_REGNO_NREGS): Redefine.
10306 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10307 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10308 (HARD_REGNO_NREGS): Delete.
10309 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10310 (cr16_hard_regno_nregs): New function.
10311 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10312 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10313 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10314 (cris_hard_regno_nregs): New function.
10315 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10316 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10317 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10318 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10319 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10320 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10321 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10322 (frv_hard_regno_nregs): Make static. Take and return an
10324 (frv_class_max_nregs): Remove outdated copy of documentation.
10325 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10326 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10327 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10328 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10329 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10330 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10331 (TARGET_HARD_REGNO_NREGS): Redefine.
10332 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10333 (CLASS_MAX_NREGS): Update comment.
10334 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10335 (ia64_hard_regno_nregs): New function.
10336 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10337 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10338 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10339 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10340 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10342 (m32c_hard_regno_nregs): Likewise. Make static.
10343 (TARGET_HARD_REGNO_NREGS): Redefine.
10344 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10345 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10346 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10347 (m68k_hard_regno_nregs): New function.
10348 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10349 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10350 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10351 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10352 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10353 Take and return an unsigned int.
10354 (TARGET_HARD_REGNO_NREGS): Redefine.
10355 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10356 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10357 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10358 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10359 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10360 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10361 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10362 (msp430_hard_regno_nregs): Make static. Take and return an
10364 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10365 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10366 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10367 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10368 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10369 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10370 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10371 (TARGET_HARD_REGNO_NREGS): Redefine.
10372 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10373 (PA_HARD_REGNO_NREGS): ...this.
10374 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10375 (PA_HARD_REGNO_NREGS): ...this.
10376 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10377 (pa_hard_regno_nregs): New function.
10378 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10379 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10380 (pdp11_hard_regno_nregs): New function.
10381 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10382 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10383 (rs6000_hard_regno_nregs_hook): New function.
10384 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10385 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10386 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10387 Take and return an unsigned int. Move earlier in file.
10388 (TARGET_HARD_REGNO_NREGS): Redefine.
10389 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10390 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10391 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10392 (rl78_hard_regno_nregs): Make static. Take and return an
10394 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10395 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10396 (rs6000_hard_regno_nregs_hook): New function.
10397 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10398 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10399 (TARGET_HARD_REGNO_NREGS): Redefine.
10400 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10401 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10402 instead of HARD_REGNO_NREGS.
10403 (s390_hard_regno_nregs): New function.
10404 (s390_hard_regno_mode_ok): Add comment from s390.h.
10405 (TARGET_HARD_REGNO_NREGS): Redefine.
10406 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10407 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10408 (sh_hard_regno_nregs): New function.
10409 (sh_pass_in_reg_p): Use it.
10410 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10411 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10412 (sparc_hard_regno_nregs): New function.
10413 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10414 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10415 (spu_function_arg_advance): Use it, supplying a valid register number.
10416 (TARGET_HARD_REGNO_NREGS): Redefine.
10417 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10418 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10419 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10420 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10421 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10422 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10423 (CLASS_MAX_NREGS): Remove copy of old documentation.
10424 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10425 (visium_hard_regno_nregs): New function.
10426 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10427 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10428 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10429 xtensa_hard_regno_nregs): New function.
10430 * system.h (HARD_REGNO_NREGS): Poison.
10432 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10434 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10435 hard_regno_nregs instead of HARD_REGNO_NREGS.
10436 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10437 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
10438 (c6x_expand_epilogue): Likewise.
10439 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
10440 (frv_read_iacc_argument): Likewise.
10441 * config/sh/sh.c: Include regs.h.
10442 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10443 (regs_used): Likewise.
10444 (output_stack_adjust): Likewise.
10445 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
10446 * expmed.c: Include regs.h.
10447 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10448 * ree.c: Include regs.h.
10449 (combine_reaching_defs): Use hard_regno_nregs instead of
10451 (add_removable_extension): Likewise.
10453 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10455 * regs.h (hard_regno_nregs): Turn into a function.
10456 (end_hard_regno): Update accordingly.
10457 * caller-save.c (setup_save_areas): Likewise.
10458 (save_call_clobbered_regs): Likewise.
10459 (replace_reg_with_saved_mem): Likewise.
10460 (insert_restore): Likewise.
10461 (insert_save): Likewise.
10462 * combine.c (can_change_dest_mode): Likewise.
10463 (move_deaths): Likewise.
10464 (distribute_notes): Likewise.
10465 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
10466 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
10467 (rs6000_split_multireg_move): Likewise.
10468 (rs6000_register_move_cost): Likewise.
10469 (rs6000_memory_move_cost): Likewise.
10470 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
10471 (rs6000_split_multireg_move): Likewise.
10472 (rs6000_register_move_cost): Likewise.
10473 (rs6000_memory_move_cost): Likewise.
10474 * cselib.c (cselib_reset_table): Likewise.
10475 (cselib_lookup_1): Likewise.
10476 * emit-rtl.c (set_mode_and_regno): Likewise.
10477 * function.c (aggregate_value_p): Likewise.
10478 * ira-color.c (setup_profitable_hard_regs): Likewise.
10479 (check_hard_reg_p): Likewise.
10480 (calculate_saved_nregs): Likewise.
10481 (assign_hard_reg): Likewise.
10482 (improve_allocation): Likewise.
10483 (calculate_spill_cost): Likewise.
10484 * ira-emit.c (modify_move_list): Likewise.
10485 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
10486 (ira_hard_reg_in_set_p): Likewise.
10487 * ira.c (setup_reg_mode_hard_regset): Likewise.
10488 (clarify_prohibited_class_mode_regs): Likewise.
10489 (check_allocation): Likewise.
10490 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10491 (lra_setup_reg_renumber): Likewise.
10492 (setup_try_hard_regno_pseudos): Likewise.
10493 (spill_for): Likewise.
10494 (assign_hard_regno): Likewise.
10495 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
10496 * lra-constraints.c (in_class_p): Likewise.
10497 (lra_constraint_offset): Likewise.
10498 (simplify_operand_subreg): Likewise.
10499 (lra_constraints): Likewise.
10500 (split_reg): Likewise.
10501 (split_if_necessary): Likewise.
10502 (invariant_p): Likewise.
10503 (inherit_in_ebb): Likewise.
10504 * lra-lives.c (process_bb_lives): Likewise.
10505 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
10506 (get_hard_regs): Likewise.
10507 (do_remat): Likewise.
10508 * lra-spills.c (assign_spill_hard_regs): Likewise.
10509 * mode-switching.c (create_pre_exit): Likewise.
10510 * postreload.c (reload_combine_recognize_pattern): Likewise.
10511 * recog.c (peep2_find_free_register): Likewise.
10512 * regcprop.c (kill_value_regno): Likewise.
10513 (set_value_regno): Likewise.
10514 (copy_value): Likewise.
10515 (maybe_mode_change): Likewise.
10516 (find_oldest_value_reg): Likewise.
10517 (copyprop_hardreg_forward_1): Likewise.
10518 * regrename.c (check_new_reg_p): Likewise.
10519 (regrename_do_replace): Likewise.
10520 * reload.c (push_reload): Likewise.
10521 (combine_reloads): Likewise.
10522 (find_dummy_reload): Likewise.
10523 (operands_match_p): Likewise.
10524 (find_reloads): Likewise.
10525 (find_equiv_reg): Likewise.
10526 (reload_adjust_reg_for_mode): Likewise.
10527 * reload1.c (count_pseudo): Likewise.
10528 (count_spilled_pseudo): Likewise.
10529 (find_reg): Likewise.
10530 (clear_reload_reg_in_use): Likewise.
10531 (free_for_value_p): Likewise.
10532 (allocate_reload_reg): Likewise.
10533 (choose_reload_regs): Likewise.
10534 (reload_adjust_reg_for_temp): Likewise.
10535 (emit_reload_insns): Likewise.
10536 (delete_output_reload): Likewise.
10537 * rtlanal.c (subreg_get_info): Likewise.
10538 * sched-deps.c (sched_analyze_reg): Likewise.
10539 * sel-sched.c (init_regs_for_mode): Likewise.
10540 (mark_unavailable_hard_regs): Likewise.
10541 (choose_best_reg_1): Likewise.
10542 (verify_target_availability): Likewise.
10543 * valtrack.c (dead_debug_insert_temp): Likewise.
10544 * var-tracking.c (track_loc_p): Likewise.
10545 (emit_note_insn_var_location): Likewise.
10546 * varasm.c (make_decl_rtl): Likewise.
10547 * reginfo.c (choose_hard_reg_mode): Likewise.
10548 (init_reg_modes_target): Refer directly to
10549 this_target_regs->x_hard_regno_nregs.
10551 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10553 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
10554 instead of hard_regno_nregs.
10556 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10558 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
10559 end_hard_regno instead of hard_regno_nregs.
10560 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
10561 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
10562 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
10563 * ira-color.c (improve_allocation): Likewise.
10564 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10565 * lra-lives.c (mark_regno_live): Likewise.
10566 (mark_regno_dead): Likewise.
10567 * lra-remat.c (operand_to_remat): Likewise.
10568 * lra.c (collect_non_operand_hard_regs): Likewise.
10569 * postreload.c (reload_combine_note_store): Likewise.
10570 (move2add_valid_value_p): Likewise.
10571 * reload.c (regno_clobbered_p): Likewise.
10573 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10575 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10577 * config/v850/v850.c (v850_reorg): Likewise.
10578 * reload.c (refers_to_regno_for_reload_p): Likewise.
10579 (find_equiv_reg): Likewise.
10580 * reload1.c (reload_reg_reaches_end_p): Likewise.
10582 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10584 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10586 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10587 * config/arm/arm.c (output_move_neon): Likewise.
10588 (arm_attr_length_move_neon): Likewise.
10589 (neon_split_vcombine): Likewise.
10590 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10591 (c6x_mark_reg_written): Likewise.
10592 (c6x_dwarf_register_span): Likewise.
10593 * config/i386/i386.c (ix86_save_reg): Likewise.
10594 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10595 (rws_access_reg): Likewise.
10596 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10597 * mode-switching.c (create_pre_exit): Likewise.
10598 * ree.c (combine_reaching_defs): Likewise.
10599 (add_removable_extension): Likewise.
10600 * regcprop.c (find_oldest_value_reg): Likewise.
10601 (copyprop_hardreg_forward_1): Likewise.
10602 * reload.c (reload_inner_reg_of_subreg): Likewise.
10603 (push_reload): Likewise.
10604 (combine_reloads): Likewise.
10605 (find_dummy_reload): Likewise.
10606 (reload_adjust_reg_for_mode): Likewise.
10607 * reload1.c (find_reload_regs): Likewise.
10608 (forget_old_reloads_1): Likewise.
10609 (reload_reg_free_for_value_p): Likewise.
10610 (reload_adjust_reg_for_temp): Likewise.
10611 (emit_reload_insns): Likewise.
10612 (delete_output_reload): Likewise.
10613 * sel-sched.c (choose_best_reg_1): Likewise.
10614 (choose_best_pseudo_reg): Likewise.
10616 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10617 Alan Hayward <alan.hayward@arm.com>
10618 David Sherwood <david.sherwood@arm.com>
10620 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10621 * target.def (slow_unaligned_access): New hook.
10622 * targhooks.h (default_slow_unaligned_access): Declare.
10623 * targhooks.c (default_slow_unaligned_access): New function.
10624 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10625 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10626 * doc/tm.texi: Regenerate.
10627 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10628 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10629 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10631 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10632 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10634 (rs6000_slow_unaligned_access): New function.
10635 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10636 (expand_block_compare): Likewise.
10637 (expand_strn_compare): Likewise.
10638 (rs6000_rtx_costs): Likewise.
10639 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10640 (riscv_slow_unaligned_access): Likewise.
10641 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10642 (riscv_slow_unaligned_access_p): ...this and make static.
10643 (riscv_option_override): Update accordingly.
10644 (riscv_slow_unaligned_access): New function.
10645 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10646 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10647 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10648 (rs6000_slow_unaligned_access): New function.
10649 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10650 (rs6000_rtx_costs): Likewise.
10651 * config/rs6000/rs6000-string.c (expand_block_compare)
10652 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10653 of SLOW_UNALIGNED_ACCESS.
10654 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10655 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10656 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10657 of SLOW_UNALIGNED_ACCESS.
10658 * expmed.c (simple_mem_bitfield_p): Likewise.
10659 * expr.c (alignment_for_piecewise_move): Likewise.
10660 (emit_group_load_1): Likewise.
10661 (emit_group_store): Likewise.
10662 (copy_blkmode_from_reg): Likewise.
10663 (emit_push_insn): Likewise.
10664 (expand_assignment): Likewise.
10665 (store_field): Likewise.
10666 (expand_expr_real_1): Likewise.
10667 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10668 * lra-constraints.c (simplify_operand_subreg): Likewise.
10669 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10670 * gimple-ssa-store-merging.c: Likewise in block comment at start
10672 * tree-ssa-strlen.c: Include target.h.
10673 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10674 of SLOW_UNALIGNED_ACCESS.
10675 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10677 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10679 PR rtl-optimization/82185
10680 * expmed.c (emit_store_flag_int): Only test tem if it has been
10683 2017-09-12 Richard Biener <rguenther@suse.de>
10685 PR middle-end/82149
10686 * match.pd ((FTYPE) N CMP CST): Fix typo.
10688 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10690 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10692 (mips_near_type_p): Add 'short_call' attribute as a synonym
10694 * doc/extend.texi (short_call): Document new function attribute.
10696 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10699 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10700 assertion check that in the condition.
10701 (get_atomic_generic_size): Likewise. Before testing if parameter
10702 has pointer type, if it has array type, call for C++
10703 default_conversion to perform array-to-pointer conversion.
10705 2017-09-12 Richard Biener <rguenther@suse.de>
10707 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10708 for operations we cannot scalarize.
10710 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10712 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10713 vectors heap vectors. Clean up comments.
10714 Make visited_bbs a reference.
10715 (profitable_jump_thread_path): Make GC
10716 vectors heap vectors. Clean up comments.
10718 (convert_and_register_jump_thread_path): Make GC vectors heap
10720 (check_subpath_and_update_thread_path): Same. Clean up comments.
10721 Make visited_bbs a reference.
10722 (handle_phi): Abstract common code to to
10723 register_jump_thread_path_if_profitable.
10724 Rename VAR_BB to DEF_BB.
10726 Make GC vectors heap vectors.
10727 Make visited_bbs a reference.
10728 (handle_assignment): Same.
10729 (register_jump_thread_path_if_profitable): New.
10730 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10732 Make GC vectors heap vectors. Clean up comments.
10733 Make visited_bbs a reference.
10734 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10735 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10738 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10741 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10742 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10744 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10746 Revert r251800 and r251799.
10748 2017-09-11 Martin Jambor <mjambor@suse.cz>
10751 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10752 arguments in advance.
10753 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10754 use it to find predecessor edges.
10755 (naive_outof_ssa): Collect vector of predecessors.
10757 2017-09-08 Jason Merrill <jason@redhat.com>
10759 PR c++/70029 - ICE with ref-qualifier and -flto
10760 * langhooks.h (struct lang_hooks_for_types): Add
10761 copy_lang_qualifiers.
10762 * attribs.c (build_type_attribute_qual_variant): Use it.
10763 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10765 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10766 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10768 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10771 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10772 (*mulsi3_sp64): New instruction.
10773 (mulsi3): New expander.
10775 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10777 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10779 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10781 * sancov.c: Include memmodel.h.
10783 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10786 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10789 2017-09-07 Carl Love <cel@us.ibm.com>
10791 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10792 the sldi instruction.
10794 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10796 * sancov.c: Include tm_p.h.
10798 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10801 * output.h (switch_to_other_text_partition): New declaration.
10802 * varasm.c (switch_to_other_text_partition): New function.
10803 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10804 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10805 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10806 to the other text partition before emitting LCL label and switch back
10807 after emitting the word after it.
10809 2017-09-07 Richard Biener <rguenther@suse.de>
10811 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10812 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10813 critical edges here, after loop init.
10814 (pass_data_pre): Remove PROP_no_crit_edges flags.
10815 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10816 for valueization of call args to avoid leaking VN_TOP.
10817 (visit_use): Assert we do not visit default defs.
10818 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10819 Use error_mark_node to more easily detect leaking VN_TOP.
10820 All default-defs are varying, not VN_TOP. Mark them visited.
10821 (run_scc_vn): Make code match comment.
10823 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10825 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10826 OPTION_MASK_FLOAT128_KEYWORD.
10827 (POWERPC_MASKS): Likewise.
10828 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10829 support for the -mfloat128-type option, and make -mfloat128
10830 default on PowerPC Linux systems. Define or undefine
10831 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10832 Define __float128 to be __ieee128 if IEEE 128-bit support is
10833 enabled, or undefine it.
10834 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10835 Delete defining __FLOAT128_TYPE__.
10836 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10837 -mfloat128-type option and make -mfloat128 default on PowerPC
10839 (TARGET_FLOAT128_TYPE): Likewise.
10840 (-mfloat128-type): Likewise.
10841 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10842 Delete the -mfloat128-type option and make -mfloat128 default on
10843 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10844 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10845 128-bit floating point is enabled. Change tests from using
10846 -mfloat128-type to -mfloat128.
10847 (rs6000_mangle_type): Use the correct mangling for the __float128
10848 type even if normal long double is restricted to 64-bits.
10849 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10851 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10852 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10853 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10855 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10856 documentation for -mfloat128.
10858 2017-09-06 Olivier Hainque <hainque@adacore.com>
10860 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10862 2017-09-06 Wish Wu <wishwu007@gmail.com>
10863 Jakub Jelinek <jakub@redhat.com>
10865 * asan.c (initialize_sanitizer_builtins): Add
10866 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10867 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10868 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10869 BT_FN_VOID_UINT64_PTR variables.
10870 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10871 (BT_FN_VOID_UINT16_UINT16): Likewise.
10872 (BT_FN_VOID_UINT32_UINT32): Likewise.
10873 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10874 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10875 (BT_FN_VOID_UINT64_PTR): Likewise.
10876 * common.opt (flag_sanitize_coverage): New variable.
10877 (fsanitize-coverage=trace-pc): Remove.
10878 (fsanitize-coverage=): Add.
10879 * flag-types.h (enum sanitize_coverage_code): New enum.
10880 * fold-const.c (fold_range_test): Disable non-short-circuit
10881 optimization if flag_sanitize_coverage.
10882 (fold_truth_andor): Likewise.
10883 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10884 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10885 (coverage_sanitizer_opts): New array.
10886 (get_closest_sanitizer_option): Add OPTS argument, handle also
10887 OPT_fsanitize_coverage_.
10888 (parse_sanitizer_options): Adjusted to also handle
10889 OPT_fsanitize_coverage_.
10890 (common_handle_option): Add OPT_fsanitize_coverage_.
10891 * sancov.c (instrument_comparison, instrument_switch): New function.
10892 (sancov_pass): Add trace-cmp support.
10893 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10894 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10895 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10896 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10897 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10898 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10899 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10900 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10901 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10902 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10904 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10906 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10907 error. Only quit immediately if parsing is complete.
10908 (BEGIN): Initialize fatal_err and parse_done.
10909 (begin fpu, end fpu): Check number of arguments.
10910 (begin arch, end arch): Likewise.
10911 (begin cpu, end cpu): Likewise.
10912 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10913 (optalias): Likewise.
10915 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10917 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10918 * config/arm/arm-isa.h: Delete. Move definitions to ...
10919 * arm-cpus.in: ... here. Use new feature and fgroup values.
10920 * config/arm/arm.c (arm_option_override): Use lower case for feature
10922 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10923 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10924 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10926 (print_isa_bits_for): New function.
10927 (gen_isa): New function.
10928 (gen_comm_data): Use print_isa_bits_for.
10929 (define feature): New keyword.
10930 (define fgroup): New keyword.
10931 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10932 (arm-isa.h): Add rule to generate file.
10933 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10934 case for feature bit names.
10936 2017-09-06 Richard Biener <rguenther@suse.de>
10938 * tree-ssa-pre.c (NECESSARY): Remove.
10939 (create_expression_by_pieces): Do not touch pass-local flags.
10940 (insert_into_preds_of_block): Likewise.
10941 (do_pre_regular_insertion): Likewise.
10942 (eliminate_insert): Likewise.
10943 (eliminate_dom_walker::before_dom_children): Likewise.
10944 (fini_eliminate): Do not look at inserted_exprs.
10945 (mark_operand_necessary): Remove.
10946 (remove_dead_inserted_code): Replace with simple work-list
10947 algorithm based on inserted_exprs and SSA uses.
10948 (pass_pre::execute): Re-order fini_eliminate and
10949 remove_dead_inserted_code.
10951 2017-09-06 Olivier Hainque <hainque@adacore.com>
10953 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10954 for VxWorks 7. Adjust surrounding comments.
10956 2017-09-06 Richard Biener <rguenther@suse.de>
10958 * gimple-ssa-strength-reduction.c
10959 (find_candidates_dom_walker::before_dom_children): Also allow
10962 2017-09-06 Richard Biener <rguenther@suse.de>
10964 PR tree-optimization/82108
10965 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10966 for gap in the non-permutation SLP case.
10968 2017-09-06 Martin Jambor <mjambor@suse.cz>
10970 PR tree-optimization/82078
10971 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10972 add_access_to_work_queue...
10973 (build_accesses_from_assign): ...here.
10974 (propagate_all_subaccesses): Make sure racc is the group
10975 representative, if there is one.
10977 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10979 PR middle-end/82095
10980 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10983 2017-09-06 Richard Biener <rguenther@suse.de>
10985 * gimple-ssa-strength-reduction.c
10986 (find_candidates_dom_walker::before_doom_children): Use a
10987 type and not a mode check.
10989 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10992 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10994 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10996 2017-09-05 Jeff Law <law@redhat.com>
10998 PR tree-optimization/64910
10999 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11000 cases where we have 3 or more operands.
11002 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11004 PR middle-end/81768
11005 * omp-low.c (lower_omp_for): Recompute tree invariant if
11006 gimple_omp_for_initial/final is ADDR_EXPR.
11008 PR middle-end/81768
11009 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11010 into gimple val before gimplification fo the COND_EXPR.
11012 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11014 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11015 REGION_COPY argument.
11016 (thread_through_all_blocks): Remove unused argument to
11017 duplicate_thread_path.
11019 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11020 Alan Hayward <alan.hayward@arm.com>
11021 David Sherwood <david.sherwood@arm.com>
11023 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11024 Take a scalar_mode rather than a machine_mode.
11025 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11026 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11027 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11028 (aarch64_gen_adjusted_ldpstp): Likewise.
11029 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11031 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11032 Alan Hayward <alan.hayward@arm.com>
11033 David Sherwood <david.sherwood@arm.com>
11035 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11036 Take a scalar_int_mode instead of a machine_mode.
11037 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11038 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11039 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11040 (aarch64_simd_attr_length_rglist): Delete.
11041 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11042 a scalar_int_mode instead of a machine_mode.
11043 (aarch64_add_offset): Likewise.
11044 (aarch64_internal_mov_immediate): Likewise
11045 (aarch64_add_constant_internal): Likewise.
11046 (aarch64_add_constant): Likewise.
11047 (aarch64_movw_imm): Likewise.
11048 (aarch64_rtx_arith_op_extract_p): Likewise.
11049 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11050 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11051 Remove assert that the mode isn't a vector.
11052 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11053 (aarch64_expand_mov_immediate): Update calls after above changes.
11054 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11055 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11056 (aarch64_move_imm): Likewise.
11057 (aarch64_can_const_movi_rtx_p): Likewise.
11058 (aarch64_strip_extend): Likewise.
11059 (aarch64_extr_rtx_p): Likewise.
11060 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11061 a CONST_INT when the mode parameter is VOIDmode.
11062 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11064 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11066 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11067 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11068 (bitwise_type_for_mode): Update accordingly.
11070 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11072 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11073 * stor-layout.c (mode_for_size_tree): Likewise.
11074 (mode_for_array): Update accordingly.
11075 (layout_decl): Likewise.
11076 (compute_record_mode): Likewise. Only set the mode once.
11078 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11080 * target.def (get_mask_mode): Change return type to opt_mode.
11082 * doc/tm.texi: Regenerate.
11083 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11084 * targhooks.c (default_get_mask_mode): Likewise.
11085 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11086 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11087 targetm.get_mask_mode.
11088 * tree.c (build_truth_vector_type): Likewise.
11090 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11092 * machmode.h (mode_for_vector): Return an opt_mode.
11093 * stor-layout.c (mode_for_vector): Likewise.
11094 (mode_for_int_vector): Update accordingly.
11095 (layout_type): Likewise.
11096 * config/i386/i386.c (emit_memmov): Likewise.
11097 (ix86_expand_set_or_movmem): Likewise.
11098 (ix86_expand_vector_init): Likewise.
11099 (ix86_get_mask_mode): Likewise.
11100 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11102 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11103 * expmed.c (extract_bit_field_1): Likewise.
11104 * expr.c (expand_expr_real_2): Likewise.
11105 * optabs-query.c (can_vec_perm_p): Likewise.
11106 (can_vec_mask_load_store_p): Likewise.
11107 * optabs.c (expand_vec_perm): Likewise.
11108 * targhooks.c (default_get_mask_mode): Likewise.
11109 * tree-vect-stmts.c (vectorizable_store): Likewise.
11110 (vectorizable_load): Likewise.
11111 (get_vectype_for_scalar_type_and_size): Likewise.
11113 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11115 * machmode.h (mode_for_int_vector): New function.
11116 * stor-layout.c (mode_for_int_vector): Likewise.
11117 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11118 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11119 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11120 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11121 (s390_expand_vcond): Likewise.
11123 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11125 * machmode.h (opt_machine_mode): New type.
11126 (opt_mode<T>): Allow construction from anything that can be
11128 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11129 (mode_for_size): Return an opt_machine_mode.
11130 * stor-layout.c (mode_for_size): Likewise.
11131 (mode_for_size_tree): Update call accordingly.
11132 (bitwise_mode_for_mode): Likewise.
11133 (make_fract_type): Likewise.
11134 (make_accum_type): Likewise.
11135 * caller-save.c (replace_reg_with_saved_mem): Update call
11137 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11138 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11139 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11140 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11141 * expmed.c (extract_bit_field_1): Likewise.
11142 * reload.c (get_secondary_mem): Likewise.
11143 * varasm.c (assemble_integer): Likewise.
11144 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
11147 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11149 * machmode.h (decimal_float_mode_for_size): New function.
11150 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
11151 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
11152 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
11153 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
11154 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
11155 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11157 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11159 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
11160 (get_builtin_sync_mode): Likewise.
11161 (expand_ifn_atomic_compare_exchange): Likewise.
11162 (expand_builtin_atomic_clear): Likewise.
11163 (expand_builtin_atomic_test_and_set): Likewise.
11164 (fold_builtin_atomic_always_lock_free): Likewise.
11165 * calls.c (compute_argument_addresses): Likewise.
11166 (emit_library_call_value_1): Likewise.
11167 (store_one_arg): Likewise.
11168 * combine.c (combine_instructions): Likewise.
11169 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
11170 * config/arm/arm.c (arm_function_value): Likewise.
11171 (aapcs_allocate_return_reg): Likewise.
11172 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
11173 * config/i386/i386.c (construct_container): Likewise.
11174 (ix86_gimplify_va_arg): Likewise.
11175 (ix86_expand_sse_cmp): Likewise.
11176 (emit_memmov): Likewise.
11177 (emit_memset): Likewise.
11178 (expand_small_movmem_or_setmem): Likewise.
11179 (ix86_expand_pextr): Likewise.
11180 (ix86_expand_pinsr): Likewise.
11181 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
11182 * config/microblaze/microblaze.c (microblaze_block_move_straight):
11184 * config/mips/mips.c (mips_function_value_1) Likewise.
11185 (mips_block_move_straight): Likewise.
11186 (mips_expand_ins_as_unaligned_store): Likewise.
11187 * config/powerpcspe/powerpcspe.c
11188 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11189 (rs6000_darwin64_record_arg_flush): Likewise.
11190 * config/rs6000/rs6000.c
11191 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11192 (rs6000_darwin64_record_arg_flush): Likewise.
11193 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
11194 (sparc_function_value_1): Likewise.
11195 * config/spu/spu.c (adjust_operand): Likewise.
11196 (spu_emit_branch_or_set): Likewise.
11197 (arith_immediate_p): Likewise.
11198 * emit-rtl.c (gen_lowpart_common): Likewise.
11199 * expr.c (expand_expr_real_1): Likewise.
11200 * function.c (assign_parm_setup_block): Likewise.
11201 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
11202 * reload1.c (alter_reg): Likewise.
11203 * stor-layout.c (mode_for_vector): Likewise.
11204 (layout_type): Likewise.
11206 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11208 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
11209 (spu_convert_move): Likewise.
11210 * lower-subreg.c (resolve_simple_move): Likewise.
11212 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11215 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
11216 define_insn to a define_expand.
11217 (altivec_vsum2sws_direct): New define_insn.
11218 (altivec_vsumsws): Convert from a define_insn to a define_expand.
11220 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11222 * config/arm/arm.c (arm_option_params_internal): Improve setting of
11225 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
11230 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
11231 realign stack if stack alignment needed is less than incoming
11234 2017-09-05 Marek Polacek <polacek@redhat.com>
11237 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11240 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11242 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11244 2017-09-05 Richard Biener <rguenther@suse.de>
11246 PR tree-optimization/82084
11247 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11249 2017-09-05 Richard Biener <rguenther@suse.de>
11251 PR tree-optimization/82102
11252 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11254 2017-09-05 Martin Liska <mliska@suse.cz>
11256 PR tree-optimization/82032
11257 * tree-cfg.c (generate_range_test): New function.
11258 * tree-cfg.h (generate_range_test): Declared here.
11259 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11260 (cleanup_control_expr_graph): Use it.
11261 * tree-switch-conversion.c (try_switch_expansion): Remove
11263 (emit_case_nodes): Use generate_range_test.
11265 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11268 * config/i386/i386.md (*<btsc><mode>_mask): Add
11269 TARGET_USE_BT to insn constraint.
11270 (*btr<mode>_mask): Ditto.
11272 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11274 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11275 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11277 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11280 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11281 TARGET_NEON and TARGET_IWMMXT.
11282 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11283 TARGET_NEON and TARGET_IWMMXT.
11284 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11286 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11288 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11289 (ix86_rewrite_tls_address): Ditto.
11290 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11291 (ix86_rewrite_tls_address_1): Ditto.
11292 (ix86_rewrite_tls_address): Ditto.
11293 * config/i386/predicates.md (tls_address_pattern): New predicate.
11294 * config/i386/i386.md (TLS address splitter): New splitter.
11296 2017-09-04 Richard Biener <rguenther@suse.de>
11298 PR tree-optimization/82084
11299 * fold-const.h (can_native_encode_string_p): Declare.
11300 * fold-const.c (can_native_encode_string_p): Factor out from ...
11301 (native_encode_string): ... here.
11302 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11303 vectorizing stores from constants we later cannot handle.
11305 2017-09-04 Marek Polacek <polacek@redhat.com>
11308 * doc/invoke.texi: Update -Wtautological-compare documentation.
11310 2017-09-04 Jeff Law <law@redhat.com>
11312 PR tree-optimization/64910
11313 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11314 swap the first and last operand if the last is a constant.
11316 2017-09-04 Marek Polacek <polacek@redhat.com>
11319 * convert.c (do_narrow): When sanitizing signed integer overflows,
11320 bail out for signed types.
11321 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11323 2017-09-04 Richard Biener <rguenther@suse.de>
11325 PR tree-optimization/82060
11326 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11327 Move devirtualization after stmt folding and before EH/AB/noreturn
11328 cleanup to get the stmt refs canonicalized. Use a bool instead
11329 of gimple_modified_p since that doesn't work for NOPs. Schedule
11330 NOPs generated by folding for removal.
11332 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11333 Alan Hayward <alan.hayward@arm.com>
11334 David Sherwood <david.sherwood@arm.com>
11336 * coretypes.h (pad_direction): New enum.
11337 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11338 (FUNCTION_ARG_PADDING): Likewise.
11339 * target.def (function_arg_padding): New hook.
11340 * targhooks.h (default_function_arg_padding): Declare.
11341 * targhooks.c (default_function_arg_padding): New function.
11342 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11343 (TARGET_FUNCTION_ARG_PADDING): ...this.
11344 * doc/tm.texi: Regenerate.
11345 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11346 instead of direction.
11347 (compute_argument_addresses): Likewise.
11348 (load_register_parameters): Likewise.
11349 (emit_library_call_value_1): Likewise.
11350 (store_one_arg): Use targetm.calls.function_arg_padding instead
11351 of FUNCTION_ARG_PADDING.
11352 (must_pass_in_stack_var_size_or_pad): Likewise.
11353 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11354 (emit_group_store): Likewise.
11355 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11356 instead of FUNCTION_ARG_PADDING.
11357 (emit_push_insn): Likewise, and propagate enum change throughout
11359 * function.h (direction): Delete.
11360 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11362 * function.c (assign_parm_find_stack_rtl): Likewise.
11363 (assign_parm_setup_block_p): Likewise.
11364 (assign_parm_setup_block): Likewise.
11365 (gimplify_parameters): Likewise.
11366 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11367 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11369 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11370 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11371 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11372 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11373 (aarch64_function_arg_padding): ...this new function.
11374 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11375 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11376 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11377 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11378 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11379 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11380 (arm_pad_arg_upward): Replace with...
11381 (arm_function_arg_padding): ...this new function.
11382 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11384 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11385 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11386 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11387 (ia64_hpux_function_arg_padding): Replace with...
11388 (ia64_function_arg_padding): ...this new function. Use pad_direction
11389 instead of direction. Check for TARGET_HPUX.
11390 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11391 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11392 (iq2000_function_arg_padding): New function.
11393 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11394 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11395 (mips_function_arg_padding): ...this new function.
11396 (mips_pad_reg_upward): Update accordingly.
11397 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11398 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11399 targetm.calls.function_arg_padding.
11400 (FUNCTION_ARG_PADDING): Delete.
11401 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11402 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11403 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11404 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11405 (nios2_block_reg_padding): Return pad_direction instead of direction.
11406 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11407 instead of direction.
11408 (nios2_function_arg_padding): Likewise. Make static.
11409 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11410 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11411 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11412 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11413 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11414 (pa_function_arg_padding): Make static. Return pad_direction instead
11416 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11417 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11418 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11419 instead of direction. Use targetm.calls.function_arg_padding.
11420 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11421 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11422 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11423 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11424 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11426 (function_arg_padding): Rename to...
11427 (rs6000_function_arg_padding): ...this. Make static. Return
11428 pad_direction instead of direction.
11429 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11430 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11431 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11432 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11433 instead of direction. Use targetm.calls.function_arg_padding.
11434 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11435 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11436 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11437 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
11438 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11439 (function_arg_padding): Rename to...
11440 (rs6000_function_arg_padding): ...this. Make static. Return
11441 pad_direction instead of direction.
11442 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11443 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
11444 * config/s390/s390.c (s390_function_arg_padding): New function.
11445 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11446 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
11447 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
11448 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11449 (function_arg_padding): Rename to...
11450 (sparc_function_arg_padding): ...this. Make static. Return
11451 pad_direction instead of direction.
11452 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
11453 * config/spu/spu.c (spu_function_arg_padding): New function.
11454 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11455 * system.h (FUNCTION_ARG_PADDING): Poison.
11457 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11458 Alan Hayward <alan.hayward@arm.com>
11459 David Sherwood <david.sherwood@arm.com>
11461 * target.def (modes_tieable_p): New hook.
11462 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
11463 (TARGET_MODES_TIEABLE_P): ...this.
11464 * doc/tm.texi.in: Regenerate.
11465 * hooks.h (hook_bool_mode_mode_true): Declare.
11466 * hooks.c (hook_bool_mode_mode_true): New function.
11467 * combine.c (subst): Use targetm.modes_tieable_p instead of
11469 * dse.c (find_shift_sequence): Likewise.
11470 * expmed.c (extract_low_bits): Likewise.
11471 * lower-subreg.c: Include target.h.
11472 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
11474 * rtlanal.c (rtx_cost): Likewise.
11475 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
11476 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
11477 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
11478 (TARGET_MODES_TIEABLE_P): Redefine.
11479 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
11480 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
11481 (TARGET_MODES_TIEABLE_P): Redefine.
11482 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
11483 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
11484 (arc_modes_tieable_p): New function.
11485 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
11486 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
11487 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
11488 (arm_modes_tieable_p): Make static.
11489 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
11490 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
11491 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
11492 (TARGET_MODES_TIEABLE_P): Redefine.
11493 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
11494 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
11495 (TARGET_MODES_TIEABLE_P): Redefine.
11496 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
11497 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
11498 (cr16_modes_tieable_p): New function.
11499 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
11500 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
11501 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
11502 (TRULY_NOOP_TRUNCATION): Update comment.
11503 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
11504 (TRULY_NOOP_TRUNCATION): Update comment.
11505 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
11506 (frv_modes_tieable_p): New function.
11507 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
11508 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
11509 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
11510 (TARGET_MODES_TIEABLE_P): Redefine.
11511 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
11512 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
11513 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
11514 (TARGET_MODES_TIEABLE_P): Redefine.
11515 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
11516 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
11517 (ia64_modes_tieable_p): New function.
11518 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
11519 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
11520 (iq2000_modes_tieable_p): New function.
11521 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
11522 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
11523 (lm32_modes_tieable_p): New function.
11524 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
11525 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
11526 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
11527 (TARGET_MODES_TIEABLE_P): Redefine.
11528 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
11529 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
11530 (m32r_modes_tieable_p): New function.
11531 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
11532 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
11533 (m68k_modes_tieable_p): New function.
11534 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
11535 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
11536 (mcore_modes_tieable_p): New function.
11537 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
11538 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
11540 (TARGET_MODES_TIEABLE_P): Redefine.
11541 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
11542 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
11543 * config/mips/mips.c (mips_modes_tieable_p): Make static.
11544 (TARGET_MODES_TIEABLE_P): Redefine.
11545 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
11546 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
11547 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
11548 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
11549 (mn10300_modes_tieable_p): ...this and make static.
11550 (TARGET_MODES_TIEABLE_P): Redefine.
11551 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
11552 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
11553 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
11554 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
11555 (msp430_modes_tieable_p): Make static.
11556 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
11557 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
11558 (TARGET_MODES_TIEABLE_P): Redefine.
11559 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
11560 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
11561 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
11562 (TARGET_MODES_TIEABLE_P): Redefine.
11563 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
11564 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
11565 * config/pa/pa.c (pa_modes_tieable_p): Make static.
11566 (TARGET_MODES_TIEABLE_P): Redefine.
11567 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
11568 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
11569 (pdp11_modes_tieable_p): New function.
11570 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
11571 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
11572 (rs6000_modes_tieable_p): New function.
11573 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11574 * config/powerpcspe/powerpcspe.md: Update comment.
11575 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11576 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11577 (TARGET_MODES_TIEABLE_P): Redefine.
11578 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11579 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11580 (rl78_modes_tieable_p): New function.
11581 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11582 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11583 (rs6000_modes_tieable_p): New function.
11584 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11585 * config/rs6000/rs6000.md: Update comment.
11586 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11587 * config/rx/rx.c (rx_modes_tieable_p): New function.
11588 (TARGET_MODES_TIEABLE_P): Redefine.
11589 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11590 * config/s390/s390.c (s390_modes_tieable_p): New function.
11591 (TARGET_MODES_TIEABLE_P): Redefine.
11592 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11593 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11594 (sh_modes_tieable_p): New function.
11595 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11596 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11597 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11598 (sparc_modes_tieable_p): Make static.
11599 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11600 * config/spu/spu.c (spu_modes_tieable_p): New function.
11601 (TARGET_MODES_TIEABLE_P): Redefine.
11602 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11603 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11604 (TARGET_MODES_TIEABLE_P): Redefine.
11605 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11606 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11607 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11608 * config/v850/v850.c (v850_modes_tieable_p): New function.
11609 (TARGET_MODES_TIEABLE_P): Redefine.
11610 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11611 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11612 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11613 (visium_modes_tieable_p): New function.
11614 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11615 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11616 (xtensa_modes_tieable_p): New function.
11617 * system.h (MODES_TIEABLE_P): Poison.
11619 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11620 Alan Hayward <alan.hayward@arm.com>
11621 David Sherwood <david.sherwood@arm.com>
11623 * target.def (hard_regno_mode_ok): New hook.
11624 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11625 (TARGET_HARD_REGNO_MODE_OK): ...this.
11626 * doc/tm.texi.in: Regenerate.
11627 * hooks.h (hook_bool_uint_mode_true): Declare.
11628 * hooks.c (hook_bool_uint_mode_true): New function.
11629 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11630 HARD_REGNO_MODE_OK.
11631 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11632 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11633 instead of HARD_REGNO_MODE_OK.
11634 * caller-save.c: Include target.h.
11635 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11636 HARD_REGNO_MODE_OK.
11637 * combine.c (can_combine_p): Likewise.
11638 (combinable_i3pat): Likewise.
11639 (can_change_dest_mode): Likewise.
11640 * expr.c (init_expr_target): Likewise.
11641 (convert_move): Likewise.
11642 (convert_modes): Likewise.
11643 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11644 (setup_prohibited_mode_move_regs): Likewise.
11645 * ira.h (target_ira): Likewise.
11646 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11647 * lra-constraints.c (process_alt_operands): Likewise.
11648 (split_reg): Likewise.
11649 * recog.c (peep2_find_free_register): Likewise.
11650 * ree.c (combine_reaching_defs): Likewise.
11651 * regcprop.c (maybe_mode_change): Likewise.
11652 * reginfo.c (init_reg_sets_1): Likewise.
11653 (choose_hard_reg_mode): Likewise.
11654 (simplifiable_subregs): Likewise.
11655 * regrename.c (check_new_reg_p): Likewise.
11656 * reload.c (find_valid_class): Likewise.
11657 (find_valid_class_1): Likewise.
11658 (reload_inner_reg_of_subreg): Likewise.
11659 (push_reload): Likewise.
11660 (combine_reloads): Likewise.
11661 (find_dummy_reload): Likewise.
11662 (find_reloads): Likewise.
11663 * reload1.c (find_reg): Likewise.
11664 (set_reload_reg): Likewise.
11665 (allocate_reload_reg): Likewise.
11666 (choose_reload_regs): Likewise.
11667 (reload_adjust_reg_for_temp): Likewise.
11668 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11669 (simplify_subreg_regno): Likewise.
11670 * sel-sched.c (init_regs_for_mode): Likewise.
11671 * varasm.c (make_decl_rtl): Likewise.
11672 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11673 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11674 HARD_REGNO_MODE_OK.
11675 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11676 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11677 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11678 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11679 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11681 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11682 (arc_mode_class): Delete.
11683 (HARD_REGNO_MODE_OK): Delete.
11684 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11685 (arc_hard_regno_mode_ok): Rename old array to...
11686 (arc_hard_regno_mode_ok_modes): ...this.
11687 (arc_conditional_register_usage): Update accordingly.
11688 (arc_mode_class): Make static.
11689 (arc_hard_regno_mode_ok): New function.
11690 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11691 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11692 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11693 (arm_hard_regno_mode_ok): Make static.
11694 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11695 HARD_REGNO_MODE_OK.
11696 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11697 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11698 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11700 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11701 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11702 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11703 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11704 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11705 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11706 * config/bfin/predicates.md (valid_reg_operand): Use
11707 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11708 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11709 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11711 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11712 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11713 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11714 (cr16_hard_regno_mode_ok): Make static and return a bool.
11715 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11716 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11717 (cris_hard_regno_mode_ok): New function.
11718 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11719 (epiphany_mode_class): Delete.
11720 (HARD_REGNO_MODE_OK): Delete.
11721 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11722 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11723 (hard_regno_mode_ok): Rename to...
11724 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11725 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11726 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11727 HARD_REGNO_MODE_OK.
11728 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11729 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11730 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11731 (frv_hard_regno_mode_ok): Make static and return a bool.
11732 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11733 HARD_REGNO_MODE_OK.
11734 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11735 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11736 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11737 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11739 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11740 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11741 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11742 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11745 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11746 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11747 (ia64_hard_regno_mode_ok): New function.
11748 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11749 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11750 (iq2000_hard_regno_mode_ok): New function.
11751 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11752 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11753 (lm32_hard_regno_mode_ok): New function.
11754 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11755 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11756 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11757 instead of HARD_REGNO_MODE_OK.
11758 (m32c_hard_regno_ok): Rename to...
11759 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11760 (m32c_cannot_change_mode_class): Update accordingly.
11761 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11762 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11763 (m32r_mode_class): Delete.
11764 (HARD_REGNO_MODE_OK): Delete.
11765 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11766 (m32r_hard_regno_mode_ok): Rename to...
11767 (m32r_hard_regno_modes): ...this.
11768 (m32r_mode_class): Make static.
11769 (m32r_hard_regno_mode_ok): New function.
11770 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11771 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11772 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11773 (m68k_hard_regno_mode_ok): Make static.
11774 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11775 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11776 (mcore_hard_regno_mode_ok): New function.
11777 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11778 (HARD_REGNO_MODE_OK): Delete.
11779 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11781 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11782 (microblaze_hard_regno_mode_ok): New function.
11783 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11784 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11785 (mips_hard_regno_mode_ok): Delete.
11786 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11787 (mips_hard_regno_mode_ok_p): ...this and make static.
11788 (mips_hard_regno_mode_ok_p): Rename to...
11789 (mips_hard_regno_mode_ok_uncached): ...this.
11790 (mips_hard_regno_mode_ok): New function.
11791 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11792 of HARD_REGNO_MODE_OK.
11793 (mips_option_override): Update after above name changes.
11794 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11795 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11796 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11797 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11798 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11799 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11800 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11801 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11802 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11803 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11804 (msp430_hard_regno_mode_ok): Make static and return a bool.
11805 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11806 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11807 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11809 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11810 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11811 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11812 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11813 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11814 (PA_HARD_REGNO_MODE_OK): ...this
11815 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11816 (PA_HARD_REGNO_MODE_OK): ...this.
11817 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11818 (pa_hard_regno_mode_ok): New function.
11819 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11820 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11821 (pdp11_hard_regno_mode_ok): New function.
11822 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11823 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11825 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11827 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11828 (rs6000_hard_regno_mode_ok): Rename to...
11829 (rs6000_hard_regno_mode_ok_uncached): ...this.
11830 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11831 (rs6000_hard_regno_mode_ok): New function.
11832 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11833 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11834 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11835 (riscv_hard_regno_mode_ok): ...this and make static.
11836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11837 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11838 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11839 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11840 (rl78_hard_regno_mode_ok): Make static and return bool.
11841 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11842 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11844 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11846 (rs6000_hard_regno_mode_ok): Rename to...
11847 (rs6000_hard_regno_mode_ok_uncached): ...this.
11848 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11849 (rs6000_hard_regno_mode_ok): New function.
11850 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11851 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11852 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11853 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11854 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11855 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11856 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11857 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11858 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11859 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11860 (sh_hard_regno_mode_ok): Make static.
11861 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11862 instead of HARD_REGNO_MODE_OK.
11863 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11864 (sparc_mode_class): Delete.
11865 (HARD_REGNO_MODE_OK): Delete.
11866 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11867 (hard_regno_mode_classes): Make static.
11868 (sparc_mode_class): Likewise.
11869 (sparc_hard_regno_mode_ok): New function.
11870 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11871 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11872 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11874 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11875 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11876 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11877 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11878 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11879 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11880 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11881 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11882 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11883 (visium_hard_regno_mode_ok): New function.
11884 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11885 instead of HARD_REGNO_MODE_OK.
11886 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11887 (HARD_REGNO_MODE_OK): Delete.
11888 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11889 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11890 (xtensa_option_override): Update accordingly.
11891 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11892 (xtensa_hard_regno_mode_ok): New function.
11893 * system.h (HARD_REGNO_MODE_OK): Poison.
11895 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11896 Alan Hayward <alan.hayward@arm.com>
11897 David Sherwood <david.sherwood@arm.com>
11899 * target.def (hard_regno_call_part_clobbered): New hook.
11900 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11901 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11902 * doc/tm.texi: Regenerate.
11903 * hooks.h (hook_bool_uint_mode_false): Declare.
11904 * hooks.c (hook_bool_uint_mode_false): New function.
11905 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11906 * cselib.c (cselib_process_insn): Use
11907 targetm.hard_regno_call_part_clobbered instead of
11908 HARD_REGNO_CALL_PART_CLOBBERED.
11909 * ira-conflicts.c (ira_build_conflicts): Likewise.
11910 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11911 * lra-constraints.c (need_for_call_save_p): Likewise.
11912 * lra-lives.c: Include target.h.
11913 (check_pseudos_live_through_calls): Use
11914 targetm.hard_regno_call_part_clobbered instead of
11915 HARD_REGNO_CALL_PART_CLOBBERED.
11916 * regcprop.c: Include target.h.
11917 (copyprop_hardreg_forward_1): Use
11918 targetm.hard_regno_call_part_clobbered instead of
11919 HARD_REGNO_CALL_PART_CLOBBERED.
11920 * reginfo.c (choose_hard_reg_mode): Likewise.
11921 * regrename.c (check_new_reg_p): Likewise.
11922 * reload.c (find_equiv_reg): Likewise.
11923 * reload1.c (emit_reload_insns): Likewise.
11924 * sched-deps.c (deps_analyze_insn): Likewise.
11925 * sel-sched.c (init_regs_for_mode): Likewise.
11926 (mark_unavailable_hard_regs): Likewise.
11927 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11928 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11929 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11931 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11932 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11933 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11935 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11937 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11938 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11939 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11941 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11942 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11943 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11945 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11946 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11948 * config/powerpcspe/powerpcspe.c
11949 (rs6000_hard_regno_call_part_clobbered): New function.
11950 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11951 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11952 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11954 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11955 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11956 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11958 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11959 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11960 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11962 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11963 Alan Hayward <alan.hayward@arm.com>
11964 David Sherwood <david.sherwood@arm.com>
11966 * rtl.h (subreg_memory_offset): Declare.
11967 * emit-rtl.c (subreg_memory_offset): New function.
11968 * expmed.c (store_bit_field_1): Use it.
11969 * expr.c (undefined_operand_subword_p): Likewise.
11970 * simplify-rtx.c (simplify_subreg): Likewise.
11972 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11974 PR rtl-optimization/57448
11977 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11978 using atomic_load pattern.
11979 (expand_atomic_store): Likewise.
11981 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11984 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11985 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11986 handling. Use replace_call_with_value with NULL instead of
11987 gsi_replace, unlink_stmt_vdef and release_defs.
11989 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11992 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11994 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11997 * rtl.h (emit_library_call_value_1): Declare.
11998 (emit_library_call): Replace declaration with a series of overloads.
11999 Remove the parameter count argument.
12000 (emit_library_call_value): Likewise.
12001 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12002 with an "rtx_mode_t *".
12003 (emit_library_call_value): Delete.
12004 (emit_library_call): Likewise.
12005 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12006 (asan_emit_allocas_unpoison): Likewise.
12007 * builtins.c (expand_builtin_powi): Likewise.
12008 (expand_asan_emit_allocas_unpoison): Likewise.
12009 * cfgexpand.c (expand_main_function): Likewise.
12010 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12011 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12012 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12013 * config/arm/arm.c (arm_trampoline_init): Likewise.
12014 (arm_call_tls_get_addr): Likewise.
12015 (arm_expand_divmod_libfunc): Likewise.
12016 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12017 (smulsi3_highpart): Likewise.
12018 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12019 (c6x_expand_compare): Likewise.
12020 (c6x_expand_movmem): Likewise.
12021 * config/frv/frv.c (frv_trampoline_init): Likewise.
12022 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12023 (ix86_expand_divmod_libfunc): Likewise.
12024 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12025 (ia64_expand_compare): Likewise.
12026 (ia64_profile_hook): Likewise.
12027 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12028 (nonlocal_goto): Likewise.
12029 (restore_stack_nonlocal): Likewise.
12030 * config/m32r/m32r.c (block_move_call): Likewise.
12031 (m32r_trampoline_init): Likewise.
12032 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12033 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12034 (m68k_call_m68k_read_tp): Likewise.
12035 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12036 (microblaze_expand_divide): Likewise.
12037 * config/mips/mips.h (mips_args): Likewise.
12038 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12039 (MIPS_ICACHE_SYNC): Likewise.
12040 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12041 (nios2_trampoline_init): Likewise.
12042 * config/pa/pa.c (hppa_tls_call): Likewise.
12043 (pa_trampoline_init): Likewise.
12044 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12045 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12046 (expand_strn_compare): Likewise.
12047 (rs6000_generate_compare): Likewise.
12048 (rs6000_expand_float128_convert): Likewise.
12049 (output_profile_hook): Likewise.
12050 (rs6000_trampoline_init): Likewise.
12051 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12052 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12053 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12054 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12055 (rs6000_generate_compare): Likewise.
12056 (rs6000_expand_float128_convert): Likewise.
12057 (output_profile_hook): Likewise.
12058 (rs6000_trampoline_init): Likewise.
12059 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12060 * config/sh/sh.c (sh_trampoline_init): Likewise.
12061 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12062 (sparc_emit_float_lib_cmp): Likewise.
12063 (sparc32_initialize_trampoline): Likewise.
12064 (sparc64_initialize_trampoline): Likewise.
12065 (sparc_profile_hook): Likewise.
12066 * config/spu/spu.c (ea_load_store): Likewise.
12067 * config/spu/spu.md (floatunssidf2): Likewise.
12068 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12069 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12070 * config/visium/visium.c (expand_block_move_4): Likewise.
12071 (expand_block_move_2): Likewise.
12072 (expand_block_move_1): Likewise.
12073 (expand_block_set_4): Likewise.
12074 (expand_block_set_2): Likewise.
12075 (expand_block_set_1): Likewise.
12076 (visium_trampoline_init): Likewise.
12077 (visium_profile_hook): Likewise.
12078 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12079 (xtensa_setup_frame_addresses): Likewise.
12080 (xtensa_trampoline_init): Likewise.
12081 * except.c (sjlj_emit_function_enter): Likewise.
12082 (sjlj_emit_function_exit): Likewise.
12083 * explow.c (allocate_dynamic_stack_space): Likewise.
12084 (probe_stack_range): Likewise.
12085 * expr.c (convert_mode_scalar): Likewise.
12086 * optabs.c (expand_binop): Likewise.
12087 (expand_twoval_binop_libfunc): Likewise.
12088 (expand_unop): Likewise.
12089 (prepare_cmp_insn): Likewise.
12090 (prepare_float_lib_cmp): Likewise.
12091 (expand_float): Likewise.
12092 (expand_fix): Likewise.
12093 (expand_fixed_convert): Likewise.
12094 (maybe_emit_sync_lock_test_and_set): Likewise.
12095 (expand_atomic_compare_and_swap): Likewise.
12096 (expand_mem_thread_fence): Likewise.
12097 (expand_atomic_fetch_op): Likewise.
12099 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12101 * doc/generic.texi (OpenACC): Adjust URL.
12102 * doc/invoke.texi (C Dialect Options): Ditto.
12104 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12106 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12107 predicate for operand 1. Add (m,<S>) constraint.
12108 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12109 Prevent memory operand 1 with register operand 2.
12111 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12113 PR rtl-optimization/82024
12114 * combine.c (try_combine): If the combination result is a PARALLEL,
12115 and we only need to retain the SET in there that would be placed
12116 at I2, check that we can place that at I3 instead, before doing so.
12118 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12121 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12123 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12124 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12127 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12128 Jeff Law <law@redhat.com>
12130 * varasm.c (bss_initializer_p): Do not put constants into .bss
12131 (categorize_decl_for_section): Handle bss_initializer_p returning
12132 false when DECL_INITIAL is NULL.
12134 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12137 * config/s390/s390.c (s390_can_inline_p): New function.
12139 2017-09-01 Jeff Law <law@redhat.com>
12141 PR tree-optimization/82052
12142 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12143 Always initialize the returned slot after a hash table miss
12144 when INSERT is true.
12146 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
12148 * config/s390/s390.md (mem_signal_fence): Remove.
12149 * doc/md.texi (mem_signal_fence): Remove.
12150 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
12152 * target-insns.def (mem_signal_fence): Remove.
12154 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12157 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
12160 * asan.c (create_odr_indicator): Strip name encoding from assembler
12161 name before appending it after __odr_asan_.
12163 2017-09-01 Martin Liska <mliska@suse.cz>
12165 PR tree-optimization/82059
12166 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
12167 frequency only when an edge is redirected.
12169 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12171 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
12172 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
12173 (arc_conditional_register_usage): Remove ARC600 lp_count
12175 (arc_file_start): Emit Tag_ARC_CPU_variation.
12176 (arc_can_use_doloop_p): New conditions to use ZOLs.
12177 (hwloop_fail): New function.
12178 (hwloop_optimize): Likewise.
12179 (hwloop_pattern_reg): Likewise.
12180 (arc_doloop_hooks): New struct, to be used with reorg_loops.
12181 (arc_reorg_loops): New function, calls reorg_loops.
12182 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
12183 (arc600_corereg_hazard): Remove ZOL checking, case handled by
12185 (arc_loop_hazard): Remove function, functionality moved into
12187 (arc_hazard): Remove arc_loop_hazard call.
12188 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
12189 into hwloop_optimize.
12190 (arc_label_align): Remove ZOL handling.
12191 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
12192 * config/arc/arc.md (doloop_begin): Remove pattern.
12193 (doloop_begin_i): Likewise.
12194 (doloop_end_i): Likewise.
12195 (doloop_fallback): Likewise.
12196 (doloop_fallback_m): Likewise.
12197 (doloop_end): Reimplement expand.
12198 (arc_lp): New pattern for LP instruction.
12199 (loop_end): New pattern.
12200 (loop_fail): Likewise.
12201 (decrement_and_branch_until_zero): Likewise.
12202 * config/arc/arc.opt (mlpc-width): New option.
12203 * doc/invoke.texi (mlpc-width): Document option.
12205 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12207 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
12208 (arc_ccfsm_advance): Fix checking for delay slots.
12209 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
12211 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12213 * config/arc/arc.md (movqi_insn): Add stores to save constant long
12215 (movhi_insn): Update store instruction constraint which are saving
12216 6-bit short immediates.
12217 (movsi_insn): Consider also short scaled load operations.
12218 (zero_extendhisi2_i): Use Usd constraint instead of T.
12219 (extendhisi2_i): Add q constraint.
12220 (arc_clzsi2): Add type and length attributes.
12221 (arc_ctzsi2): Likewise.
12222 * config/arc/constraints.md (Usc): Update constraint, the
12223 assembler can parse two relocations for a single instruction.
12225 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12227 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
12228 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
12230 2017-08-31 Olivier Hainque <hainque@adacore.com>
12232 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
12233 match as powerpc-wrs-vxworks*.
12235 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12237 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12238 register constraint for by-element operand.
12239 (aarch64_mls_elt_merge<mode>): Likewise.
12241 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12243 * config/arc/arc.c (arc_can_follow_jump): Check for short
12246 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12248 * config.gcc: Use g.opt for arc.
12249 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12250 functionality moved to ...
12251 (legitimate_scaled_address_p): New function, ...here.
12252 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12253 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12254 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12256 (arc_override_options): Handle G option.
12257 (arc_output_pic_addr_const): Correct function definition.
12258 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12259 (arc_decl_anon_ns_mem_p): Delete.
12260 (arc_in_small_data_p): Overhaul this function to take into
12261 consideration the value given via G option.
12262 (arc_rewrite_small_data_1): Renamed and corrected old
12263 arc_rewrite_small_data function.
12264 (arc_rewrite_small_data): New function.
12265 (small_data_pattern): Don't use pic_offset_table_rtx.
12266 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12267 * config/arc/simdext.md (movmisalignv2hi): Use
12268 prepare_move_operands function.
12270 (movmisalign*): Likewise.
12271 * doc/invoke.texi (ARC options): Document -G option.
12273 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12275 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12277 * config/arc/arc.c (arc_print_operand): Output scalled address for
12278 sdata whenever is possible.
12279 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12280 load/stores are available.
12281 (compact_sda_memory_operand): Check for the alignment required by
12282 code density instructions.
12283 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12285 * config/arc/constraints.md (Usd): Update constraint.
12286 (Us0): New constraint.
12287 (Usc): Update constraint.
12289 2017-08-31 Richard Biener <rguenther@suse.de>
12291 PR middle-end/82054
12292 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12293 function only once.
12295 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12297 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12298 Resize type_signature.
12300 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12301 Alan Hayward <alan.hayward@arm.com>
12302 David Sherwood <david.sherwood@arm.com>
12304 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12305 subregs whose inner modes can be stored in GPRs.
12306 (aarch64_classify_index): Likewise.
12308 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12309 Alan Hayward <alan.hayward@arm.com>
12310 David Sherwood <david.sherwood@arm.com>
12312 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12313 (V_INT_EQUIV): ...this.
12314 (v_cmp_result): Rename to...
12315 (v_int_equiv): ...this.
12316 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12317 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12318 (copysign<mode>3): Likewise.
12319 (aarch64_simd_bsl<mode>_internal): Likewise.
12320 (aarch64_simd_bsl<mode>): Likewise.
12321 (vec_cmp<mode><mode>): Likewise.
12322 (vcond<mode><mode>): Likewise.
12323 (vcond<v_cmp_mixed><mode>): Likewise.
12324 (vcondu<mode><v_cmp_mixed>): Likewise.
12325 (aarch64_cm<optab><mode>): Likewise.
12326 (aarch64_cmtst<mode>): Likewise.
12327 (aarch64_fac<optab><mode>): Likewise.
12328 (vec_perm_const<mode>): Likewise.
12329 (vcond_mask_<mode><v_cmp_result>): Rename to...
12330 (vcond_mask_<mode><v_int_equiv>): ...this.
12331 (vec_cmp<mode><v_cmp_result>): Rename to...
12332 (vec_cmp<mode><v_int_equiv>): ...this.
12334 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12335 Alan Hayward <alan.hayward@arm.com>
12336 David Sherwood <david.sherwood@arm.com>
12338 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12340 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12341 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12342 (UNSPEC_LD4_DREG): New unspecs.
12343 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12344 (aarch64_ld2<mode>_dreg_be): Replace with...
12345 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12347 (aarch64_ld3<mode>_dreg_le)
12348 (aarch64_ld3<mode>_dreg_be): Replace with...
12349 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12351 (aarch64_ld4<mode>_dreg_le)
12352 (aarch64_ld4<mode>_dreg_be): Replace with...
12353 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12356 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12358 PR tree-optimization/81987
12359 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12360 insert an initializer in a location not dominated by the stride
12363 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12365 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12366 on the entire header of the finally block in the fallthru case.
12368 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12370 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12372 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12374 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12375 rs6000_emit_move_from_cr and call renamed function.
12376 (rs6000_emit_prologue): Call renamed functions.
12377 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12378 movesi_from_cr, remove volatile CRs.
12380 2017-08-30 Jon Beniston <jon@beniston.com>
12381 Richard Biener <rguenther@suse.de>
12383 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12384 of VECTOR_MODE_P check.
12385 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12386 element vector types.
12388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12390 * df.h (df_read_modify_subreg_p): Remove in favor of...
12391 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12392 const_rtx instead of an rtx.
12393 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12394 * df-problems.c (df_word_lr_mark_ref): Likewise.
12395 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12396 (mark_pseudo_reg_dead): Likewise.
12397 (mark_ref_dead): Likewise.
12398 * reginfo.c (init_subregs_of_mode): Likewise.
12399 * sched-deps.c (sched_analyze_1): Likewise.
12400 * df-scan.c (df_def_record_1): Likewise.
12401 (df_uses_record): Likewise.
12402 (df_read_modify_subreg_p): Remove in favor of...
12403 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12404 const_rtx instead of an rtx.
12406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12407 Alan Hayward <alan.hayward@arm.com>
12408 David Sherwood <david.sherwood@arm.com>
12410 * rtl.h (partial_subreg_p): New function.
12411 * caller-save.c (save_call_clobbered_regs): Use it.
12412 * calls.c (expand_call): Likewise.
12413 * combine.c (combinable_i3pat): Likewise.
12414 (simplify_set): Likewise.
12415 (make_extraction): Likewise.
12416 (make_compound_operation_int): Likewise.
12417 (gen_lowpart_or_truncate): Likewise.
12418 (force_to_mode): Likewise.
12419 (make_field_assignment): Likewise.
12420 (reg_truncated_to_mode): Likewise.
12421 (record_truncated_value): Likewise.
12422 (move_deaths): Likewise.
12423 * cse.c (record_jump_cond): Likewise.
12424 (cse_insn): Likewise.
12425 * cselib.c (cselib_lookup_1): Likewise.
12426 * expmed.c (extract_bit_field_using_extv): Likewise.
12427 * function.c (assign_parm_setup_reg): Likewise.
12428 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12429 * ira-build.c (create_insn_allocnos): Likewise.
12430 * lra-coalesce.c (merge_pseudos): Likewise.
12431 * lra-constraints.c (match_reload): Likewise.
12432 (simplify_operand_subreg): Likewise.
12433 (curr_insn_transform): Likewise.
12434 * lra-lives.c (process_bb_lives): Likewise.
12435 * lra.c (new_insn_reg): Likewise.
12436 (lra_substitute_pseudo): Likewise.
12437 * regcprop.c (mode_change_ok): Likewise.
12438 (maybe_mode_change): Likewise.
12439 (copyprop_hardreg_forward_1): Likewise.
12440 * reload.c (push_reload): Likewise.
12441 (find_reloads): Likewise.
12442 (find_reloads_subreg_address): Likewise.
12443 * reload1.c (alter_reg): Likewise.
12444 (eliminate_regs_1): Likewise.
12445 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12447 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
12449 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
12450 back to if statements, including unpack.
12452 2017-08-30 Martin Liska <mliska@suse.cz>
12454 PR inline-asm/82001
12455 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
12457 (func_checker::compare_asm_inputs_outputs): ... this function.
12458 (func_checker::compare_gimple_asm): Use the function to compare
12459 also ASM constrains.
12460 * ipa-icf-gimple.h: Rename the function.
12462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12463 Alan Hayward <alan.hayward@arm.com>
12464 David Sherwood <david.sherwood@arm.com>
12466 * coretypes.h (complex_mode): New type.
12467 * gdbhooks.py (build_pretty_printer): Handle it.
12468 * machmode.h (complex_mode): New class.
12469 (complex_mode::includes_p): New function.
12470 (is_complex_int_mode): Likewise.
12471 (is_complex_float_mode): Likewise.
12472 * genmodes.c (get_mode_class): Handle complex mode classes.
12473 * function.c (expand_function_end): Use is_complex_int_mode.
12475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12476 Alan Hayward <alan.hayward@arm.com>
12477 David Sherwood <david.sherwood@arm.com>
12479 * coretypes.h (scalar_mode_pod): New typedef.
12480 * gdbhooks.py (build_pretty_printer): Handle it.
12481 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
12482 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
12483 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
12484 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
12485 as_a <scalar_mode>.
12487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12488 Alan Hayward <alan.hayward@arm.com>
12489 David Sherwood <david.sherwood@arm.com>
12491 * machmode.h (mode_for_vector): Take a scalar_mode instead
12493 * stor-layout.c (mode_for_vector): Likewise.
12494 * explow.c (promote_mode): Use as_a <scalar_mode>.
12495 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
12497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12498 Alan Hayward <alan.hayward@arm.com>
12499 David Sherwood <david.sherwood@arm.com>
12501 * target.def (preferred_simd_mode): Take a scalar_mode
12502 instead of a machine_mode.
12503 * targhooks.h (default_preferred_simd_mode): Likewise.
12504 * targhooks.c (default_preferred_simd_mode): Likewise.
12505 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
12506 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
12507 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
12508 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
12509 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
12510 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
12511 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
12512 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
12514 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
12515 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
12516 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
12517 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
12518 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
12519 * doc/tm.texi: Regenerate.
12520 * optabs-query.c (can_vec_mask_load_store_p): Return false for
12523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12524 Alan Hayward <alan.hayward@arm.com>
12525 David Sherwood <david.sherwood@arm.com>
12527 * target.def (scalar_mode_supported_p): Take a scalar_mode
12528 instead of a machine_mode.
12529 * targhooks.h (default_scalar_mode_supported_p): Likewise.
12530 * targhooks.c (default_scalar_mode_supported_p): Likewise.
12531 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
12532 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12533 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
12534 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
12535 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
12536 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
12537 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
12538 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
12539 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
12540 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
12541 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
12542 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
12544 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
12545 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
12546 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12547 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12548 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
12550 * doc/tm.texi: Regenerate.
12552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12553 Alan Hayward <alan.hayward@arm.com>
12554 David Sherwood <david.sherwood@arm.com>
12556 * coretypes.h (opt_scalar_mode): New typedef.
12557 * gdbhooks.py (build_pretty_printers): Handle it.
12558 * machmode.h (mode_iterator::get_2xwider): Add overload for
12560 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
12562 * expr.c (convert_mode_scalar): Likewise.
12563 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12564 * optabs.c (expand_float): Likewise.
12565 (expand_fix): Likewise.
12566 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12569 Alan Hayward <alan.hayward@arm.com>
12570 David Sherwood <david.sherwood@arm.com>
12572 * optabs.c (expand_float): Explicitly check for scalars before
12573 using a branching expansion.
12574 (expand_fix): Likewise.
12576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12577 Alan Hayward <alan.hayward@arm.com>
12578 David Sherwood <david.sherwood@arm.com>
12580 * expr.c (convert_mode): Split scalar handling out into...
12581 (convert_mode_scalar): ...this new function. Treat the modes
12584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12585 Alan Hayward <alan.hayward@arm.com>
12586 David Sherwood <david.sherwood@arm.com>
12588 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12590 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12593 Alan Hayward <alan.hayward@arm.com>
12594 David Sherwood <david.sherwood@arm.com>
12596 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12597 rather than a machine_mode.
12598 (fixed_from_string): Likewise.
12599 (fixed_convert): Likewise.
12600 (fixed_convert_from_int): Likewise.
12601 (fixed_convert_from_real): Likewise.
12602 (real_convert_from_fixed): Likewise.
12603 * fixed-value.c (fixed_from_double_int): Likewise.
12604 (fixed_from_string): Likewise.
12605 (fixed_convert): Likewise.
12606 (fixed_convert_from_int): Likewise.
12607 (fixed_convert_from_real): Likewise.
12608 (real_convert_from_fixed): Likewise.
12609 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12612 Alan Hayward <alan.hayward@arm.com>
12613 David Sherwood <david.sherwood@arm.com>
12615 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12616 of separate mode class checks. Do not allow vector modes here.
12617 (immed_wide_int_const): Use as_a <scalar_mode>.
12618 * explow.c (trunc_int_for_mode): Likewise.
12619 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12620 (wi::shwi): Likewise.
12621 (wi::min_value): Likewise.
12622 (wi::max_value): Likewise.
12623 * dwarf2out.c (loc_descriptor): Likewise.
12624 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12625 for CONST_WIDE_INT.
12627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12628 Alan Hayward <alan.hayward@arm.com>
12629 David Sherwood <david.sherwood@arm.com>
12631 * tree.h (SCALAR_TYPE_MODE): New macro.
12632 * expr.c (expand_expr_addr_expr_1): Use it.
12633 (expand_expr_real_2): Likewise.
12634 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12635 (fold_convert_const_fixed_from_int): Likewise.
12636 (fold_convert_const_fixed_from_real): Likewise.
12637 (native_encode_fixed): Likewise
12638 (native_encode_complex): Likewise
12639 (native_encode_vector): Likewise.
12640 (native_interpret_fixed): Likewise.
12641 (native_interpret_real): Likewise.
12642 (native_interpret_complex): Likewise.
12643 (native_interpret_vector): Likewise.
12644 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12645 (simd_clone_adjust_argument_types): Likewise.
12646 (simd_clone_init_simd_arrays): Likewise.
12647 (simd_clone_adjust): Likewise.
12648 * stor-layout.c (layout_type): Likewise.
12649 * tree.c (build_minus_one_cst): Likewise.
12650 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12651 * tree-inline.c (estimate_move_cost): Likewise.
12652 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12653 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12654 (vectorizable_reduction): Likewise.
12655 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12656 (vect_recog_mixed_size_cond_pattern): Likewise.
12657 (check_bool_pattern): Likewise.
12658 (adjust_bool_pattern): Likewise.
12659 (search_type_for_mask_1): Likewise.
12660 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12661 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12662 (vectorizable_load): Likewise.
12663 (vectorizable_store): Likewise.
12664 * ubsan.c (ubsan_encode_value): Likewise.
12665 * varasm.c (output_constant): Likewise.
12667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12668 Alan Hayward <alan.hayward@arm.com>
12669 David Sherwood <david.sherwood@arm.com>
12671 * coretypes.h (scalar_mode): New class.
12672 * machmode.h (scalar_mode): Likewise.
12673 (scalar_mode::includes_p): New function.
12674 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12675 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12676 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12677 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12678 * expmed.c (store_bit_field_1): Likewise.
12679 (extract_bit_field_1): Likewise.
12680 * expr.c (write_complex_part): Likewise.
12681 (read_complex_part): Likewise.
12682 (emit_move_complex_push): Likewise.
12683 (expand_expr_real_2): Likewise.
12684 * function.c (assign_parm_setup_reg): Likewise.
12685 (assign_parms_unsplit_complex): Likewise.
12686 * optabs.c (expand_binop): Likewise.
12687 * rtlanal.c (subreg_get_info): Likewise.
12688 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12689 * varasm.c (output_constant_pool_2): Likewise.
12691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12692 Alan Hayward <alan.hayward@arm.com>
12693 David Sherwood <david.sherwood@arm.com>
12695 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12697 (expmed_mult_highpart_optab): Likewise.
12698 (expmed_mult_highpart): Likewise.
12700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12701 Alan Hayward <alan.hayward@arm.com>
12702 David Sherwood <david.sherwood@arm.com>
12704 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12705 instead of a machine_mode.
12706 (builtin_memset_read_str): Likewise.
12707 * builtins.c (c_readstr): Likewise.
12708 (builtin_memcpy_read_str): Likewise.
12709 (builtin_strncpy_read_str): Likewise.
12710 (builtin_memset_read_str): Likewise.
12711 (builtin_memset_gen_str): Likewise.
12712 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12713 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12714 instead of a machine_mode.
12715 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12717 (make_extraction): Likewise.
12718 (try_widen_shift_mode): Take and return scalar_int_modes instead
12720 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12721 a scalar_int_mode instead of a machine_mode.
12722 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12723 (avr_addr_space_pointer_mode): Likewise.
12724 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12725 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12726 (msp430_unwind_word_mode): Likewise.
12727 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12728 (spu_addr_space_pointer_mode): Likewise.
12729 (spu_addr_space_address_mode): Likewise.
12730 (spu_libgcc_cmp_return_mode): Likewise.
12731 (spu_libgcc_shift_count_mode): Likewise.
12732 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12733 (rl78_addr_space_pointer_mode): Likewise.
12734 (fl78_unwind_word_mode): Likewise.
12735 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12737 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12738 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12739 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12740 (mips_valid_pointer_mode): Likewise.
12741 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12742 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12743 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12745 (ft32_addr_space_address_mode): Likewise.
12746 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12747 scalar_int_mode instead of a machine_mode.
12748 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12750 (m32c_addr_space_address_mode): Likewise.
12751 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12752 (rs6000_eh_return_filter_mode): Likewise.
12753 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12754 (rs6000_eh_return_filter_mode): Likewise.
12755 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12756 (s390_libgcc_shift_count_mode): Likewise.
12757 (s390_unwind_word_mode): Likewise.
12758 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12760 * target.def (mode_rep_extended): Likewise.
12761 (valid_pointer_mode): Likewise.
12762 (addr_space.valid_pointer_mode): Likewise.
12763 (eh_return_filter_mode): Return a scalar_int_mode rather than
12765 (libgcc_cmp_return_mode): Likewise.
12766 (libgcc_shift_count_mode): Likewise.
12767 (unwind_word_mode): Likewise.
12768 (addr_space.pointer_mode): Likewise.
12769 (addr_space.address_mode): Likewise.
12770 * doc/tm.texi: Regenerate.
12771 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12773 (do_jump): Use scalar_int_mode for local variables.
12774 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12775 rather than a machine_mode.
12776 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12777 (scompare_loc_descriptor_wide): Likewise.
12778 (scompare_loc_descriptor_narrow): Likewise.
12779 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12781 * except.c (sjlj_emit_dispatch_table): Likewise.
12782 (expand_builtin_eh_copy_values): Likewise.
12783 * explow.c (convert_memory_address_addr_space_1): Likewise.
12784 Take a scalar_int_mode rather than a machine_mode.
12785 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12786 than a machine_mode.
12787 (memory_address_addr_space): Use scalar_int_mode for local variables.
12788 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12789 rather than a machine_mode.
12790 * expmed.c (mask_rtx): Likewise.
12791 (init_expmed_one_conv): Likewise.
12792 (expand_mult_highpart_adjust): Likewise.
12793 (extract_high_half): Likewise.
12794 (expmed_mult_highpart_optab): Likewise.
12795 (expmed_mult_highpart): Likewise.
12796 (expand_smod_pow2): Likewise.
12797 (expand_sdiv_pow2): Likewise.
12798 (emit_store_flag_int): Likewise.
12799 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12801 (extract_low_bits): Likewise.
12802 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12804 * expr.c (pieces_addr::adjust): Likewise.
12805 (can_store_by_pieces): Likewise.
12806 (store_by_pieces): Likewise.
12807 (clear_by_pieces_1): Likewise.
12808 (expand_expr_addr_expr_1): Likewise.
12809 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12810 (expand_expr_real_1): Likewise.
12811 (try_casesi): Likewise.
12812 * final.c (shorten_branches): Likewise.
12813 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12814 type of "mode" to machine_mode.
12815 * internal-fn.c (expand_arith_overflow_result_store): Take a
12816 scalar_int_mode rather than a machine_mode.
12817 (expand_mul_overflow): Use scalar_int_mode for local variables.
12818 * loop-doloop.c (doloop_modify): Likewise.
12819 (doloop_optimize): Likewise.
12820 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12821 than a machine_mode.
12822 (expand_doubleword_shift_condmove): Likewise.
12823 (expand_doubleword_shift): Likewise.
12824 (expand_doubleword_clz): Likewise.
12825 (expand_doubleword_popcount): Likewise.
12826 (expand_doubleword_parity): Likewise.
12827 (expand_absneg_bit): Use scalar_int_mode for local variables.
12828 (prepare_float_lib_cmp): Likewise.
12829 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12830 rather than a machine_mode.
12831 (convert_memory_address_addr_space): Likewise.
12832 (get_mode_bounds): Likewise.
12833 (get_address_mode): Return a scalar_int_mode rather than a
12835 * rtlanal.c (get_address_mode): Likewise.
12836 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12837 than a machine_mode.
12838 * targhooks.c (default_mode_rep_extended): Likewise.
12839 (default_valid_pointer_mode): Likewise.
12840 (default_addr_space_valid_pointer_mode): Likewise.
12841 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12842 than a machine_mode.
12843 (default_libgcc_cmp_return_mode): Likewise.
12844 (default_libgcc_shift_count_mode): Likewise.
12845 (default_unwind_word_mode): Likewise.
12846 (default_addr_space_pointer_mode): Likewise.
12847 (default_addr_space_address_mode): Likewise.
12848 * targhooks.h (default_eh_return_filter_mode): Likewise.
12849 (default_libgcc_cmp_return_mode): Likewise.
12850 (default_libgcc_shift_count_mode): Likewise.
12851 (default_unwind_word_mode): Likewise.
12852 (default_addr_space_pointer_mode): Likewise.
12853 (default_addr_space_address_mode): Likewise.
12854 (default_mode_rep_extended): Take a scalar_int_mode rather than
12856 (default_valid_pointer_mode): Likewise.
12857 (default_addr_space_valid_pointer_mode): Likewise.
12858 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12860 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12861 rather than a machine_mode.
12862 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12863 for local variables.
12864 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12865 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12866 than a machine_mode.
12868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12869 Alan Hayward <alan.hayward@arm.com>
12870 David Sherwood <david.sherwood@arm.com>
12872 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12873 the mode argument to scalar_int_mode.
12874 (do_jump_by_parts_zero_rtx): Likewise.
12875 (do_jump_by_parts_equality_rtx): Likewise.
12876 (do_jump_by_parts_greater): Take a mode argument.
12877 (do_jump_by_parts_equality): Likewise.
12878 (do_jump_1): Update calls accordingly.
12880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12881 Alan Hayward <alan.hayward@arm.com>
12882 David Sherwood <david.sherwood@arm.com>
12884 * is-a.h (safe_dyn_cast): New function.
12885 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12886 (jump_table_for_label): Likewise.
12887 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12888 instead of an rtx_insn *.
12889 (shorten_branches): Use dyn_cast instead of LABEL_P and
12890 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12891 rtx_jump_table_data::get_data_mode.
12892 (final_scan_insn): Likewise.
12894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12895 Alan Hayward <alan.hayward@arm.com>
12896 David Sherwood <david.sherwood@arm.com>
12898 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12899 trying to combine a full-register integer set with a subreg
12902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12903 Alan Hayward <alan.hayward@arm.com>
12904 David Sherwood <david.sherwood@arm.com>
12906 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12907 that is always either address_mode or pointer_mode.
12909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12910 Alan Hayward <alan.hayward@arm.com>
12911 David Sherwood <david.sherwood@arm.com>
12913 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12914 when the two are known to be equal.
12916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12917 Alan Hayward <alan.hayward@arm.com>
12918 David Sherwood <david.sherwood@arm.com>
12920 * simplify-rtx.c (simplify_const_unary_operation): Use
12921 is_a <scalar_int_mode> instead of checking for a nonzero
12922 precision. Forcibly convert op_mode to a scalar_int_mode
12923 in that case. More clearly differentiate the operand and
12924 result modes and use the former when deciding what the value
12925 of a count-bits operation should be. Use is_int_mode instead
12926 of checking for a MODE_INT. Remove redundant check for whether
12927 this mode has a zero precision.
12929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12930 Alan Hayward <alan.hayward@arm.com>
12931 David Sherwood <david.sherwood@arm.com>
12933 * optabs.c (widen_leading): Change the type of the mode argument
12934 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12935 (widen_bswap): Likewise.
12936 (expand_parity): Likewise.
12937 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12938 (expand_ffs): Likewise.
12939 (epand_unop): Check for scalar integer modes before calling the
12942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12943 Alan Hayward <alan.hayward@arm.com>
12944 David Sherwood <david.sherwood@arm.com>
12946 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12948 (expand_expr_real_1): Update call accordingly.
12950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12951 Alan Hayward <alan.hayward@arm.com>
12952 David Sherwood <david.sherwood@arm.com>
12954 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12955 value_mode arguments. Use scalar_int_mode internally.
12956 (store_bit_field_1): Rename the new integer mode from imode
12957 to op0_mode and use it instead of GET_MODE (op0). Update calls
12958 to store_split_bit_field, store_bit_field_using_insv and
12959 store_fixed_bit_field.
12960 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12961 Use scalar_int_mode internally. Use a bit count rather than a mode
12962 when calculating the largest bit size for get_best_mode.
12963 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12964 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12965 Remove assertion that OP0 has a scalar integer mode.
12966 (store_split_bit_field): Add op0_mode and value_mode arguments.
12967 Update calls to extract_fixed_bit_field.
12968 (extract_bit_field_using_extv): Add an op0_mode argument.
12969 Use scalar_int_mode internally.
12970 (extract_bit_field_1): Rename the new integer mode from imode to
12971 op0_mode and use it instead of GET_MODE (op0). Update calls to
12972 extract_split_bit_field, extract_bit_field_using_extv and
12973 extract_fixed_bit_field.
12974 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12975 to extract_split_bit_field and extract_fixed_bit_field_1.
12976 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12977 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12978 on the target mode.
12979 (extract_split_bit_field): Add an op0_mode argument. Update call
12980 to extract_fixed_bit_field.
12982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12983 Alan Hayward <alan.hayward@arm.com>
12984 David Sherwood <david.sherwood@arm.com>
12986 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12987 * explow.c (hard_function_value): Likewise.
12988 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12989 convert_to_mode call outside the loop.
12990 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12991 for the mode iterator. Require the mode specified by max_pieces
12993 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12995 (copy_blkmode_to_reg): Likewise.
12996 (set_storage_via_setmem): Likewise.
12997 * optabs.c (prepare_cmp_insn): Likewise.
12998 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12999 * stor-layout.c (finish_bitfield_representative): Likewise.
13001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13002 Alan Hayward <alan.hayward@arm.com>
13003 David Sherwood <david.sherwood@arm.com>
13005 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13006 * expr.c (convert_move): Use them.
13007 (convert_modes): Likewise.
13008 (store_expr_with_bounds): Likewise.
13010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13011 Alan Hayward <alan.hayward@arm.com>
13012 David Sherwood <david.sherwood@arm.com>
13014 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13015 parameter for the mode of "x". Remove the "known_x", "known_mode"
13016 and "known_ret" arguments. Change the type of the mode argument
13017 to scalar_int_mode.
13018 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13019 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13020 (reg_num_sign_bit_copies_for_combine): Likewise.
13021 * rtlanal.c (nonzero_bits1): Likewise.
13022 (num_sign_bit_copies1): Likewise.
13023 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13024 (reg_num_sign_bit_copies_general): Likewise.
13025 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13026 (reg_nonzero_bits_general): Likewise.
13028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13029 Alan Hayward <alan.hayward@arm.com>
13030 David Sherwood <david.sherwood@arm.com>
13032 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13033 than in subroutines. Return 1 for non-integer modes.
13034 (cached_num_sign_bit_copies): Change the type of the mode parameter
13035 to scalar_int_mode.
13036 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13037 classes. Handle CONST_INT_P first and then check whether X also
13038 has a scalar integer mode. Check the same thing for inner registers
13039 of a SUBREG and for values that are being extended or truncated.
13041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13042 Alan Hayward <alan.hayward@arm.com>
13043 David Sherwood <david.sherwood@arm.com>
13045 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13046 in subroutines. Return the mode mask for non-integer modes.
13047 (cached_nonzero_bits): Change the type of the mode parameter
13048 to scalar_int_mode.
13049 (nonzero_bits1): Likewise. Remove early exit for other mode
13050 classes. Handle CONST_INT_P first and then check whether X
13051 also has a scalar integer mode.
13053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13054 Alan Hayward <alan.hayward@arm.com>
13055 David Sherwood <david.sherwood@arm.com>
13057 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13058 Return a scalar_int_mode and assert that the size is greater than
13060 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13061 (op_by_pieces_d::op_by_pieces_d): Likewise.
13062 (op_by_pieces_d::run): Likewise.
13063 (can_store_by_pieces): Likewise.
13065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13066 Alan Hayward <alan.hayward@arm.com>
13067 David Sherwood <david.sherwood@arm.com>
13069 * combine.c (extract_left_shift): Add a mode argument and update
13071 (make_compound_operation_int): Change the type of the mode parameter
13072 to scalar_int_mode and update the call to extract_left_shift.
13074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13075 Alan Hayward <alan.hayward@arm.com>
13076 David Sherwood <david.sherwood@arm.com>
13078 * combine.c (simplify_and_const_int): Change the type of the mode
13079 parameter to scalar_int_mode.
13080 (simplify_and_const_int_1): Likewise. Update recursive call.
13082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13083 Alan Hayward <alan.hayward@arm.com>
13084 David Sherwood <david.sherwood@arm.com>
13086 * combine.c (simplify_compare_const): Check that the mode is a
13087 scalar_int_mode (rather than VOIDmode) before testing its
13089 (simplify_comparison): Move COMPARISON_P handling out of the
13090 loop and restrict the latter part of the loop to scalar_int_modes.
13091 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13092 and when considering SUBREG_REGs. Use is_int_mode instead of
13093 checking GET_MODE_CLASS against MODE_INT.
13095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13096 Alan Hayward <alan.hayward@arm.com>
13097 David Sherwood <david.sherwood@arm.com>
13099 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13100 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13101 shift_unit_mode and for modes involved in scalar shifts.
13103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13104 Alan Hayward <alan.hayward@arm.com>
13105 David Sherwood <david.sherwood@arm.com>
13107 * combine.c (force_int_to_mode): New function, split out from...
13108 (force_to_mode): ...here. Keep xmode up-to-date and use it
13109 instead of GET_MODE (x).
13111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13112 Alan Hayward <alan.hayward@arm.com>
13113 David Sherwood <david.sherwood@arm.com>
13115 * optabs-query.h (extraction_insn::struct_mode): Change type to
13116 opt_scalar_int_mode and update comment.
13117 (extraction_insn::field_mode): Change type to scalar_int_mode.
13118 (extraction_insn::pos_mode): Likewise.
13119 * combine.c (make_extraction): Update accordingly.
13120 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13121 (get_optab_extraction_insn): Likewise.
13122 * recog.c (simplify_while_replacing): Likewise.
13123 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13124 parameter to opt_scalar_int_mode.
13126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13127 Alan Hayward <alan.hayward@arm.com>
13128 David Sherwood <david.sherwood@arm.com>
13130 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13131 to a scalar_int_mode instead of a machine_mode.
13132 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13133 (get_best_mode): Return a boolean and use a pointer argument to store
13134 the selected mode. Replace the limit mode parameter with a bit limit.
13135 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
13136 for the values returned by bit_field_mode_iterator::next_mode.
13137 (store_bit_field): Update call to get_best_mode.
13138 (store_fixed_bit_field): Likewise.
13139 (extract_fixed_bit_field): Likewise.
13140 * expr.c (optimize_bitfield_assignment_op): Likewise.
13141 * fold-const.c (optimize_bit_field_compare): Likewise.
13142 (fold_truth_andor_1): Likewise.
13143 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
13144 Update for new type of m_mode.
13145 (get_best_mode): As above.
13147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13148 Alan Hayward <alan.hayward@arm.com>
13149 David Sherwood <david.sherwood@arm.com>
13151 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
13152 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
13153 (store_bit_field): Check is_a <scalar_int_mode> before calling
13154 strict_volatile_bitfield_p.
13155 (extract_bit_field): Likewise.
13157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13158 Alan Hayward <alan.hayward@arm.com>
13159 David Sherwood <david.sherwood@arm.com>
13161 * target.def (cstore_mode): Return a scalar_int_mode.
13162 * doc/tm.texi: Regenerate.
13163 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
13164 * targhooks.h (default_cstore_mode): Likewise.
13165 * targhooks.c (default_cstore_mode): Likewise, using a forced
13167 * expmed.c (emit_cstore): Expect the target of the cstore to be
13170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13171 Alan Hayward <alan.hayward@arm.com>
13172 David Sherwood <david.sherwood@arm.com>
13174 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
13176 (niter_desc): Likewise mode.
13177 (iv_analyze): Add a mode parameter.
13179 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
13180 and change its type to scalar_int_mode.
13181 * loop-iv.c: Update commentary at head of file.
13182 (iv_constant): Pass the mode paraeter before the rtx it describes
13183 and change its type to scalar_int_mode. Remove VOIDmode handling.
13184 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
13185 (iv_extend): Likewise.
13186 (shorten_into_mode): Likewise.
13187 (iv_add): Use scalar_int_mode.
13188 (iv_mult): Likewise.
13189 (iv_shift): Likewise.
13190 (canonicalize_iv_subregs): Likewise.
13191 (get_biv_step_1): Pass the outer_mode parameter before the rtx
13192 it describes and change its mode to scalar_int_mode. Also change
13193 the type of the returned inner_mode to scalar_int_mode.
13194 (get_biv_step): Likewise, turning outer_mode from a pointer
13195 into a direct parameter. Update call to get_biv_step_1.
13196 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
13197 iv_constant and get_biv_step.
13198 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
13199 and change its type to scalar_int_mode. Don't initialise iv->mode
13200 to VOIDmode and remove later checks for its still being VOIDmode.
13201 Update calls to iv_analyze_op and iv_analyze_expr. Check
13202 is_a <scalar_int_mode> when changing the mode under consideration.
13203 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
13204 Update call to iv_analyze_expr.
13205 (iv_analyze_op): Add a mode parameter. Reject subregs whose
13206 inner register is not also a scalar_int_mode. Update call to
13208 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
13209 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
13210 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
13211 separate mode class checks. Update calls to iv_analyze. Remove
13212 fix-up of VOIDmodes after iv_analyze_biv.
13213 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
13214 don't have a scalar_int_mode. Update call to biv_p.
13216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13217 Alan Hayward <alan.hayward@arm.com>
13218 David Sherwood <david.sherwood@arm.com>
13220 * cfgexpand.c (convert_debug_memory_address): Use
13221 as_a <scalar_int_mode>.
13222 * combine.c (expand_compound_operation): Likewise.
13223 (make_extraction): Likewise.
13224 (change_zero_ext): Likewise.
13225 (simplify_comparison): Likewise.
13226 * cse.c (cse_insn): Likewise.
13227 * dwarf2out.c (minmax_loc_descriptor): Likewise.
13228 (mem_loc_descriptor): Likewise.
13229 (loc_descriptor): Likewise.
13230 * expmed.c (init_expmed_one_mode): Likewise.
13231 (synth_mult): Likewise.
13232 (emit_store_flag_1): Likewise.
13233 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13234 of a comparison with size.
13235 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13236 (reduce_to_bit_field_precision): Likewise.
13237 * function.c (expand_function_end): Likewise.
13238 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13239 * loop-doloop.c (doloop_modify): Likewise.
13240 * optabs.c (expand_binop): Likewise.
13241 (expand_unop): Likewise.
13242 (expand_copysign_absneg): Likewise.
13243 (prepare_cmp_insn): Likewise.
13244 (maybe_legitimize_operand): Likewise.
13245 * recog.c (const_scalar_int_operand): Likewise.
13246 * rtlanal.c (get_address_mode): Likewise.
13247 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13248 (simplify_cond_clz_ctz): Likewise.
13249 * tree-nested.c (get_nl_goto_field): Likewise.
13250 * tree.c (build_vector_type_for_mode): Likewise.
13251 * var-tracking.c (use_narrower_mode): Likewise.
13253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13254 Alan Hayward <alan.hayward@arm.com>
13255 David Sherwood <david.sherwood@arm.com>
13257 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13258 * builtins.c (expand_builtin_signbit): Use it.
13259 * cfgexpand.c (expand_debug_expr): Likewise.
13260 * dojump.c (do_jump): Likewise.
13261 (do_compare_and_jump): Likewise.
13262 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13263 * expmed.c (make_tree): Likewise.
13264 * expr.c (expand_expr_real_2): Likewise.
13265 (expand_expr_real_1): Likewise.
13266 (try_casesi): Likewise.
13267 * fold-const-call.c (fold_const_call_ss): Likewise.
13268 * fold-const.c (unextend): Likewise.
13269 (extract_muldiv_1): Likewise.
13270 (fold_single_bit_test): Likewise.
13271 (native_encode_int): Likewise.
13272 (native_encode_string): Likewise.
13273 (native_interpret_int): Likewise.
13274 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13275 * internal-fn.c (expand_addsub_overflow): Likewise.
13276 (expand_neg_overflow): Likewise.
13277 (expand_mul_overflow): Likewise.
13278 (expand_arith_overflow): Likewise.
13279 * match.pd: Likewise.
13280 * stor-layout.c (layout_type): Likewise.
13281 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13282 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13283 * tree-ssanames.c (get_range_info): Likewise.
13284 * tree-switch-conversion.c (array_value_type) Likewise.
13285 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13286 (vect_recog_divmod_pattern): Likewise.
13287 (vect_recog_mixed_size_cond_pattern): Likewise.
13288 * tree-vrp.c (extract_range_basic): Likewise.
13289 (simplify_float_conversion_using_ranges): Likewise.
13290 * tree.c (int_fits_type_p): Likewise.
13291 * ubsan.c (instrument_bool_enum_load): Likewise.
13292 * varasm.c (mergeable_string_section): Likewise.
13293 (narrowing_initializer_constant_valid_p): Likewise.
13294 (output_constant): Likewise.
13296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13297 Alan Hayward <alan.hayward@arm.com>
13298 David Sherwood <david.sherwood@arm.com>
13300 * machmode.h (NARROWEST_INT_MODE): New macro.
13301 * expr.c (alignment_for_piecewise_move): Use it instead of
13302 GET_CLASS_NARROWEST_MODE (MODE_INT).
13303 (push_block): Likewise.
13304 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13306 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13309 Alan Hayward <alan.hayward@arm.com>
13310 David Sherwood <david.sherwood@arm.com>
13312 * postreload.c (move2add_valid_value_p): Change the type of the
13313 mode parameter to scalar_int_mode.
13314 (move2add_use_add2_insn): Add a mode parameter and use it instead
13316 (move2add_use_add3_insn): Likewise.
13317 (reload_cse_move2add): Update accordingly.
13319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13320 Alan Hayward <alan.hayward@arm.com>
13321 David Sherwood <david.sherwood@arm.com>
13323 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13325 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13326 * optabs.c (expand_unop): Likewise.
13328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13329 Alan Hayward <alan.hayward@arm.com>
13330 David Sherwood <david.sherwood@arm.com>
13332 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13333 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13334 (popcount_loc_descriptor): Likewise.
13335 (bswap_loc_descriptor): Likewise.
13336 (rotate_loc_descriptor): Likewise.
13337 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13338 calling the functions above.
13340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13341 Alan Hayward <alan.hayward@arm.com>
13342 David Sherwood <david.sherwood@arm.com>
13344 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13346 (try_combine): Likewise.
13347 (simplify_if_then_else): Likewise.
13348 * cse.c (cse_insn): Likewise.
13349 * dwarf2out.c (mem_loc_descriptor): Likewise.
13350 * emit-rtl.c (gen_lowpart_common): Likewise.
13351 * simplify-rtx.c (simplify_truncation): Likewise.
13352 (simplify_binary_operation_1): Likewise.
13353 (simplify_const_relational_operation): Likewise.
13354 (simplify_ternary_operation): Likewise.
13355 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13358 Alan Hayward <alan.hayward@arm.com>
13359 David Sherwood <david.sherwood@arm.com>
13361 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13362 * reload.c (push_reload): Likewise.
13363 (find_reloads): Likewise.
13365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13366 Alan Hayward <alan.hayward@arm.com>
13367 David Sherwood <david.sherwood@arm.com>
13369 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13370 (make_compound_operation_int): Likewise.
13371 (change_zero_ext): Likewise.
13372 * expr.c (convert_move): Likewise.
13373 (convert_modes): Likewise.
13374 * fwprop.c (forward_propagate_subreg): Likewise.
13375 * loop-iv.c (get_biv_step_1): Likewise.
13376 * optabs.c (widen_operand): Likewise.
13377 * postreload.c (move2add_valid_value_p): Likewise.
13378 * recog.c (simplify_while_replacing): Likewise.
13379 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13380 (simplify_binary_operation_1): Likewise. Remove redundant
13381 mode equality check.
13383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13384 Alan Hayward <alan.hayward@arm.com>
13385 David Sherwood <david.sherwood@arm.com>
13387 * combine.c (combine_simplify_rtx): Add checks for
13388 is_a <scalar_int_mode>.
13389 (simplify_if_then_else): Likewise.
13390 (make_field_assignment): Likewise.
13391 (simplify_comparison): Likewise.
13392 * ifcvt.c (noce_try_bitop): Likewise.
13393 * loop-invariant.c (canonicalize_address_mult): Likewise.
13394 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13397 Alan Hayward <alan.hayward@arm.com>
13398 David Sherwood <david.sherwood@arm.com>
13400 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13401 is_a <scalar_int_mode> instead of != BLKmode.
13403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13404 Alan Hayward <alan.hayward@arm.com>
13405 David Sherwood <david.sherwood@arm.com>
13407 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13408 instead of != VOIDmode.
13409 * combine.c (if_then_else_cond): Likewise.
13410 (change_zero_ext): Likewise.
13411 * dwarf2out.c (mem_loc_descriptor): Likewise.
13412 (loc_descriptor): Likewise.
13413 * rtlanal.c (canonicalize_condition): Likewise.
13414 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13417 Alan Hayward <alan.hayward@arm.com>
13418 David Sherwood <david.sherwood@arm.com>
13420 * simplify-rtx.c (simplify_binary_operation_1): Use
13421 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13424 Alan Hayward <alan.hayward@arm.com>
13425 David Sherwood <david.sherwood@arm.com>
13427 * wide-int.h (int_traits<unsigned char>) New class.
13428 (int_traits<unsigned short>) Likewise.
13429 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13430 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13432 * combine.c (set_nonzero_bits_and_sign_copies): Use
13433 is_a <scalar_int_mode>.
13434 (find_split_point): Likewise.
13435 (combine_simplify_rtx): Likewise.
13436 (simplify_logical): Likewise.
13437 (expand_compound_operation): Likewise.
13438 (expand_field_assignment): Likewise.
13439 (make_compound_operation): Likewise.
13440 (extended_count): Likewise.
13441 (change_zero_ext): Likewise.
13442 (simplify_comparison): Likewise.
13443 * dwarf2out.c (scompare_loc_descriptor): Likewise.
13444 (ucompare_loc_descriptor): Likewise.
13445 (minmax_loc_descriptor): Likewise.
13446 (mem_loc_descriptor): Likewise.
13447 (loc_descriptor): Likewise.
13448 * expmed.c (init_expmed_one_mode): Likewise.
13449 * lra-constraints.c (lra_constraint_offset): Likewise.
13450 * optabs.c (prepare_libcall_arg): Likewise.
13451 * postreload.c (move2add_note_store): Likewise.
13452 * reload.c (operands_match_p): Likewise.
13453 * rtl.h (load_extend_op): Likewise.
13454 * rtlhooks.c (gen_lowpart_general): Likewise.
13455 * simplify-rtx.c (simplify_truncation): Likewise.
13456 (simplify_unary_operation_1): Likewise.
13457 (simplify_binary_operation_1): Likewise.
13458 (simplify_const_binary_operation): Likewise.
13459 (simplify_const_relational_operation): Likewise.
13460 (simplify_subreg): Likewise.
13461 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13462 * var-tracking.c (adjust_mems): Likewise.
13463 (prepare_call_arguments): Likewise.
13465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13466 Alan Hayward <alan.hayward@arm.com>
13467 David Sherwood <david.sherwood@arm.com>
13469 * machmode.h (is_int_mode): New fuction.
13470 * combine.c (find_split_point): Use it.
13471 (combine_simplify_rtx): Likewise.
13472 (simplify_if_then_else): Likewise.
13473 (simplify_set): Likewise.
13474 (simplify_shift_const_1): Likewise.
13475 (simplify_comparison): Likewise.
13476 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
13477 * cse.c (notreg_cost): Likewise.
13478 (cse_insn): Likewise.
13479 * cselib.c (cselib_lookup_1): Likewise.
13480 * dojump.c (do_jump_1): Likewise.
13481 (do_compare_rtx_and_jump): Likewise.
13482 * dse.c (get_call_args): Likewise.
13483 * dwarf2out.c (rtl_for_decl_init): Likewise.
13484 (native_encode_initializer): Likewise.
13485 * expmed.c (emit_store_flag_1): Likewise.
13486 (emit_store_flag): Likewise.
13487 * expr.c (convert_modes): Likewise.
13488 (store_field): Likewise.
13489 (expand_expr_real_1): Likewise.
13490 * fold-const.c (fold_read_from_constant_string): Likewise.
13491 * gimple-ssa-sprintf.c (get_format_string): Likewise.
13492 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
13493 * optabs.c (expand_binop): Likewise.
13494 (expand_unop): Likewise.
13495 (expand_abs_nojump): Likewise.
13496 (expand_one_cmpl_abs_nojump): Likewise.
13497 * simplify-rtx.c (mode_signbit_p): Likewise.
13498 (val_signbit_p): Likewise.
13499 (val_signbit_known_set_p): Likewise.
13500 (val_signbit_known_clear_p): Likewise.
13501 (simplify_relational_operation_1): Likewise.
13502 * tree.c (vector_type_mode): Likewise.
13504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13505 Alan Hayward <alan.hayward@arm.com>
13506 David Sherwood <david.sherwood@arm.com>
13508 * machmode.h (smallest_mode_for_size): Fix formatting.
13509 (smallest_int_mode_for_size): New function.
13510 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
13511 instead of smallest_mode_for_size.
13512 * combine.c (make_extraction): Likewise.
13513 * config/arc/arc.c (arc_expand_movmem): Likewise.
13514 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
13515 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13516 * config/s390/s390.c (s390_expand_insv): Likewise.
13517 * config/sparc/sparc.c (assign_int_registers): Likewise.
13518 * config/spu/spu.c (spu_function_value): Likewise.
13519 (spu_function_arg): Likewise.
13520 * coverage.c (get_gcov_type): Likewise.
13521 (get_gcov_unsigned_t): Likewise.
13522 * dse.c (find_shift_sequence): Likewise.
13523 * expmed.c (store_bit_field_1): Likewise.
13524 * expr.c (convert_move): Likewise.
13525 (store_field): Likewise.
13526 * internal-fn.c (expand_arith_overflow): Likewise.
13527 * optabs-query.c (get_best_extraction_insn): Likewise.
13528 * optabs.c (expand_twoval_binop_libfunc): Likewise.
13529 * stor-layout.c (layout_type): Likewise.
13530 (initialize_sizetypes): Likewise.
13531 * targhooks.c (default_get_mask_mode): Likewise.
13532 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
13534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13535 Alan Hayward <alan.hayward@arm.com>
13536 David Sherwood <david.sherwood@arm.com>
13538 * machmode.h (opt_mode::else_blk): New function.
13539 (int_mode_for_mode): Declare.
13540 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
13541 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
13543 * cfgexpand.c (expand_debug_expr): Likewise.
13544 * combine.c (gen_lowpart_or_truncate): Likewise.
13545 (gen_lowpart_for_combine): Likewise.
13546 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
13547 * config/avr/avr.c (avr_to_int_mode): Likewise.
13548 (avr_out_plus_1): Likewise.
13549 (avr_out_plus): Likewise.
13550 (avr_out_round): Likewise.
13551 * config/i386/i386.c (ix86_split_to_parts): Likewise.
13552 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13553 (s390_expand_vcond): Likewise.
13554 * config/spu/spu.c (spu_split_immediate): Likewise.
13555 (spu_expand_mov): Likewise.
13556 * dse.c (get_stored_val): Likewise.
13557 * expmed.c (store_bit_field_1): Likewise.
13558 (convert_extracted_bit_field): Use int_mode_for_mode instead of
13560 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
13561 (extract_low_bits): Likewise.
13562 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
13563 handling rather than repeating the check.
13564 (emit_group_store): Likewise.
13565 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
13566 * optabs.c (expand_absneg_bit): Likewise.
13567 (expand_copysign_absneg): Likewise.
13568 (expand_copysign_bit): Likewise.
13569 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
13570 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
13571 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
13572 * var-tracking.c (prepare_call_arguments): Likewise.
13573 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13574 int_mode_for_mode instead of mode_for_size.
13575 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13578 Alan Hayward <alan.hayward@arm.com>
13579 David Sherwood <david.sherwood@arm.com>
13581 * machmode.h (int_mode_for_size): New function.
13582 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13583 instead of mode_for_size.
13584 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13586 * combine.c (expand_field_assignment): Use int_mode_for_size
13587 instead of mode_for_size.
13588 (make_extraction): Likewise.
13589 (simplify_shift_const_1): Likewise.
13590 (simplify_comparison): Likewise.
13591 * dojump.c (do_jump): Likewise.
13592 * dwarf2out.c (mem_loc_descriptor): Likewise.
13593 * emit-rtl.c (init_derived_machine_modes): Likewise.
13594 * expmed.c (flip_storage_order): Likewise.
13595 (convert_extracted_bit_field): Likewise.
13596 * expr.c (copy_blkmode_from_reg): Likewise.
13597 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13598 * internal-fn.c (expand_mul_overflow): Likewise.
13599 * lower-subreg.c (simple_move): Likewise.
13600 * optabs-libfuncs.c (init_optabs): Likewise.
13601 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13602 * tree.c (vector_type_mode): Likewise.
13603 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13604 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13605 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13606 * tree-vect-stmts.c (vectorizable_load): Likewise.
13607 (vectorizable_store): Likewise.
13609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13610 Alan Hayward <alan.hayward@arm.com>
13611 David Sherwood <david.sherwood@arm.com>
13613 * coretypes.h (pod_mode): New type.
13614 (scalar_int_mode_pod): New typedef.
13615 * machmode.h (pod_mode): New class.
13616 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13617 * genmodes.c (emit_mode_int_n): Update accordingly.
13618 * lower-subreg.h (target_lower_subreg): Change type to
13619 scalar_int_mode_pod.
13620 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13621 scalar_int_mode_pod.
13623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13624 Alan Hayward <alan.hayward@arm.com>
13625 David Sherwood <david.sherwood@arm.com>
13627 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13628 machine_mode to scalar_int_mode.
13629 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13630 (rs6000_option_override_internal): Remove cast to int.
13631 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13632 machine_mode to scalar_int_mode.
13633 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13634 (rs6000_option_override_internal): Remove cast to int.
13635 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13636 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13638 * config/s390/s390.c (s390_expand_builtin): Likewise.
13639 * coretypes.h (scalar_int_mode): New type.
13640 (opt_scalar_int_mode): New typedef.
13641 * machmode.h (scalar_int_mode): New class.
13642 (scalar_int_mode::includes_p): New function.
13643 (byte_mode): Change type to scalar_int_mode.
13644 (word_mode): Likewise.
13645 (ptr_mode): Likewise.
13646 * emit-rtl.c (byte_mode): Likewise.
13647 (word_mode): Likewise.
13648 (ptr_mode): Likewise.
13649 (init_derived_machine_modes): Update accordingly.
13650 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13651 and MODE_PARTIAL_INT.
13652 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13653 opt_scalar_int_mode.
13655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13656 Alan Hayward <alan.hayward@arm.com>
13657 David Sherwood <david.sherwood@arm.com>
13659 * target.def (libgcc_floating_mode_supported_p): Take a
13661 * doc/tm.texi: Regenerate.
13662 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13664 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13665 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13669 Alan Hayward <alan.hayward@arm.com>
13670 David Sherwood <david.sherwood@arm.com>
13672 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13673 * doc/tm.texi: Regenerate.
13674 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13675 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13676 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13677 * targhooks.h (default_floatn_mode): Likewise.
13678 * targhooks.c (default_floatn_mode): Likewise.
13679 * tree.c (build_common_tree_nodes): Update accordingly.
13681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13682 Alan Hayward <alan.hayward@arm.com>
13683 David Sherwood <david.sherwood@arm.com>
13685 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13686 (mode_iterator::iterate_p): Likewise.
13687 (mode_iterator::get_wider): Likewise.
13688 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13691 Alan Hayward <alan.hayward@arm.com>
13692 David Sherwood <david.sherwood@arm.com>
13694 * coretypes.h (opt_scalar_float_mode): New typedef.
13695 * machmode.h (float_mode_for_size): New function.
13696 * emit-rtl.c (double_mode): Delete.
13697 (init_emit_once): Use float_mode_for_size.
13698 * stor-layout.c (layout_type): Likewise.
13699 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13702 Alan Hayward <alan.hayward@arm.com>
13703 David Sherwood <david.sherwood@arm.com>
13705 * output.h (assemble_real): Take a scalar_float_mode.
13706 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13707 * config/arm/arm.md (consttable_4): Likewise.
13708 (consttable_8): Likewise.
13709 (consttable_16): Likewise.
13710 * config/mips/mips.md (consttable_float): Likewise.
13711 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13712 * varasm.c (assemble_real): Take a scalar_float_mode.
13713 (output_constant_pool_2): Update accordingly.
13714 (output_constant): Likewise.
13716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13717 Alan Hayward <alan.hayward@arm.com>
13718 David Sherwood <david.sherwood@arm.com>
13720 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13721 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13722 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13723 (native_encode_real): Likewise.
13724 (native_interpret_real): Likewise.
13725 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13726 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13729 Alan Hayward <alan.hayward@arm.com>
13730 David Sherwood <david.sherwood@arm.com>
13732 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13733 <scalar_float_mode>. Simplify.
13734 (gen_extend_conv_libfunc): Likewise.
13736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13737 Alan Hayward <alan.hayward@arm.com>
13738 David Sherwood <david.sherwood@arm.com>
13740 * coretypes.h (scalar_float_mode): New type.
13741 * machmode.h (mode_traits::from_int): Use machine_mode if
13742 USE_ENUM_MODES is defined.
13743 (is_a): New function.
13745 (dyn_cast): Likewise.
13746 (scalar_float_mode): New class.
13747 (scalar_float_mode::includes_p): New function.
13748 (is_float_mode): Likewise.
13749 * gdbhooks.py (MachineModePrinter): New class.
13750 (build_pretty_printer): Use it for scalar_float_mode.
13751 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13752 (format_helper::format_helper): Turn into a template.
13753 * genmodes.c (get_mode_class): New function.
13754 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13755 or machine_mode if none.
13756 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13757 as_a <scalar_float_mode>.
13758 * dwarf2out.c (mem_loc_descriptor): Likewise.
13759 (insert_float): Likewise.
13760 (add_const_value_attribute): Likewise.
13761 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13762 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13763 (expand_unop): Update accordingly.
13764 (expand_abs_nojump): Likewise.
13765 (expand_copysign_absneg): Take a scalar_float_mode.
13766 (expand_copysign_bit): Likewise.
13767 (expand_copysign): Update accordingly.
13769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13770 Alan Hayward <alan.hayward@arm.com>
13771 David Sherwood <david.sherwood@arm.com>
13773 * coretypes.h (opt_mode): New class.
13774 * machmode.h (opt_mode): Likewise.
13775 (opt_mode::else_void): New function.
13776 (opt_mode::require): Likewise.
13777 (opt_mode::exists): Likewise.
13778 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13779 (GET_MODE_2XWIDER_MODE): Likewise.
13780 (mode_iterator::get_wider): Update accordingly.
13781 (mode_iterator::get_2xwider): Likewise.
13782 (mode_iterator::get_known_wider): Likewise, turning into a template.
13783 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13784 forcing a wider mode to exist.
13785 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13786 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13787 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13788 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13789 * lower-subreg.c (init_lower_subreg): Likewise.
13790 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13791 on the final iteration.
13792 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13793 a wider mode exists before asking for a move pattern.
13794 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13795 forcing a wider mode to exist.
13796 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13797 returning false if no such mode exists.
13798 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13799 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13800 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13801 Avoid checking for a MODE_INT if we already know the mode is not a
13803 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13804 forcing a wider mode to exist.
13805 (expmed_mult_highpart_optab): Likewise.
13806 (expmed_mult_highpart): Likewise.
13807 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13809 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13810 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13811 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13812 * internal-fn.c (expand_mul_overflow): Update use of
13813 GET_MODE_2XWIDER_MODE.
13814 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13815 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13816 GET_MODE_WIDER_MODE.
13817 (convert_plusminus_to_widen): Likewise.
13818 * tree-switch-conversion.c (array_value_type): Likewise.
13819 * var-tracking.c (emit_note_insn_var_location): Likewise.
13820 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13821 Return false inside rather than outside the loop if no wider mode
13823 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13824 and GET_MODE_2XWIDER_MODE
13825 (can_compare_p): Use else_void.
13826 * gdbhooks.py (OptMachineModePrinter): New class.
13827 (build_pretty_printer): Use it for opt_mode.
13829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13830 Alan Hayward <alan.hayward@arm.com>
13831 David Sherwood <david.sherwood@arm.com>
13833 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13834 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13837 Alan Hayward <alan.hayward@arm.com>
13838 David Sherwood <david.sherwood@arm.com>
13840 * machmode.h (mode_traits): New structure.
13841 (get_narrowest_mode): New function.
13842 (mode_iterator::start): Likewise.
13843 (mode_iterator::iterate_p): Likewise.
13844 (mode_iterator::get_wider): Likewise.
13845 (mode_iterator::get_known_wider): Likewise.
13846 (mode_iterator::get_2xwider): Likewise.
13847 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13848 (FOR_EACH_MODE): Likewise.
13849 (FOR_EACH_MODE_FROM): Likewise.
13850 (FOR_EACH_MODE_UNTIL): Likewise.
13851 (FOR_EACH_WIDER_MODE): Likewise.
13852 (FOR_EACH_2XWIDER_MODE): Likewise.
13853 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13854 * combine.c (simplify_comparison): Likewise
13855 * config/i386/i386.c (type_natural_mode): Likewise.
13856 * cse.c (cse_insn): Likewise.
13857 * dse.c (find_shift_sequence): Likewise.
13858 * emit-rtl.c (init_derived_machine_modes): Likewise.
13859 (init_emit_once): Likewise.
13860 * explow.c (hard_function_value): Likewise.
13861 * expmed.c (extract_fixed_bit_field_1): Likewise.
13862 (extract_bit_field_1): Likewise.
13863 (expand_divmod): Likewise.
13864 (emit_store_flag_1): Likewise.
13865 * expr.c (init_expr_target): Likewise.
13866 (convert_move): Likewise.
13867 (alignment_for_piecewise_move): Likewise.
13868 (widest_int_mode_for_size): Likewise.
13869 (emit_block_move_via_movmem): Likewise.
13870 (copy_blkmode_to_reg): Likewise.
13871 (set_storage_via_setmem): Likewise.
13872 (compress_float_constant): Likewise.
13873 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13874 * optabs-query.c (get_best_extraction_insn): Likewise.
13875 * optabs.c (expand_binop): Likewise.
13876 (expand_twoval_unop): Likewise.
13877 (expand_twoval_binop): Likewise.
13878 (widen_leading): Likewise.
13879 (widen_bswap): Likewise.
13880 (expand_parity): Likewise.
13881 (expand_unop): Likewise.
13882 (prepare_cmp_insn): Likewise.
13883 (prepare_float_lib_cmp): Likewise.
13884 (expand_float): Likewise.
13885 (expand_fix): Likewise.
13886 (expand_sfix_optab): Likewise.
13887 * postreload.c (move2add_use_add2_insn): Likewise.
13888 * reg-stack.c (reg_to_stack): Likewise.
13889 * reginfo.c (choose_hard_reg_mode): Likewise.
13890 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13891 * stor-layout.c (mode_for_size): Likewise.
13892 (smallest_mode_for_size): Likewise.
13893 (mode_for_vector): Likewise.
13894 (finish_bitfield_representative): Likewise.
13895 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13896 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13897 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13898 * var-tracking.c (prepare_call_arguments): Likewise.
13900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13901 Alan Hayward <alan.hayward@arm.com>
13902 David Sherwood <david.sherwood@arm.com>
13904 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13905 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13906 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13907 * machmode.h (mode_size): Move earlier in file.
13908 (mode_precision): Likewise.
13909 (mode_inner): Likewise.
13910 (mode_nunits): Likewise.
13911 (mode_unit_size): Likewise.
13912 (unit_unit_precision): Likewise.
13913 (mode_wider): Likewise.
13914 (mode_2xwider): Likewise.
13915 (machine_mode): New class.
13916 (mode_to_bytes): New function.
13917 (mode_to_bits): Likewise.
13918 (mode_to_precision): Likewise.
13919 (mode_to_inner): Likewise.
13920 (mode_to_unit_size): Likewise.
13921 (mode_to_unit_precision): Likewise.
13922 (mode_to_nunits): Likewise.
13923 (GET_MODE_SIZE): Use mode_to_bytes.
13924 (GET_MODE_BITSIZE): Use mode_to_bits.
13925 (GET_MODE_PRECISION): Use mode_to_precision.
13926 (GET_MODE_INNER): Use mode_to_inner.
13927 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13928 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13929 (GET_MODE_NUNITS): Use mode_to_nunits.
13930 * system.h (ALWAYS_INLINE): New macro.
13931 * config/powerpcspe/powerpcspe-c.c
13932 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13933 int for arg1_mode and arg2_mode.
13935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13936 Alan Hayward <alan.hayward@arm.com>
13937 David Sherwood <david.sherwood@arm.com>
13939 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13940 Prefix mode names with E_ in case statements.
13941 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13942 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13943 (aarch64_split_simd_move): Likewise.
13944 (aarch64_gen_storewb_pair): Likewise.
13945 (aarch64_gen_loadwb_pair): Likewise.
13946 (aarch64_gen_store_pair): Likewise.
13947 (aarch64_gen_load_pair): Likewise.
13948 (aarch64_get_condition_code_1): Likewise.
13949 (aarch64_constant_pool_reload_icode): Likewise.
13950 (get_rsqrte_type): Likewise.
13951 (get_rsqrts_type): Likewise.
13952 (get_recpe_type): Likewise.
13953 (get_recps_type): Likewise.
13954 (aarch64_gimplify_va_arg_expr): Likewise.
13955 (aarch64_simd_container_mode): Likewise.
13956 (aarch64_emit_load_exclusive): Likewise.
13957 (aarch64_emit_store_exclusive): Likewise.
13958 (aarch64_expand_compare_and_swap): Likewise.
13959 (aarch64_gen_atomic_cas): Likewise.
13960 (aarch64_emit_bic): Likewise.
13961 (aarch64_emit_atomic_swap): Likewise.
13962 (aarch64_emit_atomic_load_op): Likewise.
13963 (aarch64_evpc_trn): Likewise.
13964 (aarch64_evpc_uzp): Likewise.
13965 (aarch64_evpc_zip): Likewise.
13966 (aarch64_evpc_ext): Likewise.
13967 (aarch64_evpc_rev): Likewise.
13968 (aarch64_evpc_dup): Likewise.
13969 (aarch64_gen_ccmp_first): Likewise.
13970 (aarch64_gen_ccmp_next): Likewise.
13971 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13972 (alpha_emit_xfloating_libcall): Likewise.
13973 (emit_insxl): Likewise.
13974 (alpha_arg_type): Likewise.
13975 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13976 (arc_preferred_simd_mode): Likewise.
13977 (arc_secondary_reload): Likewise.
13978 (get_arc_condition_code): Likewise.
13979 (arc_print_operand): Likewise.
13980 (arc_legitimate_constant_p): Likewise.
13981 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13982 * config/arc/arc.md (casesi_load): Likewise.
13983 (casesi_compact_jump): Likewise.
13984 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13985 (cc_use_register): Likewise.
13986 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13987 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13988 (arm_init_iwmmxt_builtins): Likewise.
13989 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13990 (neon_expand_vector_init): Likewise.
13991 (arm_attr_length_move_neon): Likewise.
13992 (maybe_get_arm_condition_code): Likewise.
13993 (arm_emit_vector_const): Likewise.
13994 (arm_preferred_simd_mode): Likewise.
13995 (arm_output_iwmmxt_tinsr): Likewise.
13996 (thumb1_output_casesi): Likewise.
13997 (thumb2_output_casesi): Likewise.
13998 (arm_emit_load_exclusive): Likewise.
13999 (arm_emit_store_exclusive): Likewise.
14000 (arm_expand_compare_and_swap): Likewise.
14001 (arm_evpc_neon_vuzp): Likewise.
14002 (arm_evpc_neon_vzip): Likewise.
14003 (arm_evpc_neon_vrev): Likewise.
14004 (arm_evpc_neon_vtrn): Likewise.
14005 (arm_evpc_neon_vext): Likewise.
14006 (arm_validize_comparison): Likewise.
14007 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14008 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14009 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14010 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14011 (c6x_preferred_simd_mode): Likewise.
14012 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14013 (epiphany_rtx_costs): Likewise.
14014 * config/epiphany/predicates.md (proper_comparison_operator):
14016 * config/frv/frv.c (condexec_memory_operand): Likewise.
14017 (frv_emit_move): Likewise.
14018 (output_move_single): Likewise.
14019 (output_condmove_single): Likewise.
14020 (frv_hard_regno_mode_ok): Likewise.
14021 (frv_matching_accg_mode): Likewise.
14022 * config/h8300/h8300.c (split_adds_subs): Likewise.
14023 (h8300_rtx_costs): Likewise.
14024 (h8300_print_operand): Likewise.
14025 (compute_mov_length): Likewise.
14026 (output_logical_op): Likewise.
14027 (compute_logical_op_length): Likewise.
14028 (compute_logical_op_cc): Likewise.
14029 (h8300_shift_needs_scratch_p): Likewise.
14030 (output_a_shift): Likewise.
14031 (compute_a_shift_length): Likewise.
14032 (compute_a_shift_cc): Likewise.
14033 (expand_a_rotate): Likewise.
14034 (output_a_rotate): Likewise.
14035 * config/i386/i386.c (classify_argument): Likewise.
14036 (function_arg_advance_32): Likewise.
14037 (function_arg_32): Likewise.
14038 (function_arg_64): Likewise.
14039 (function_value_64): Likewise.
14040 (ix86_gimplify_va_arg): Likewise.
14041 (ix86_legitimate_constant_p): Likewise.
14042 (put_condition_code): Likewise.
14043 (split_double_mode): Likewise.
14044 (ix86_avx256_split_vector_move_misalign): Likewise.
14045 (ix86_expand_vector_logical_operator): Likewise.
14046 (ix86_split_idivmod): Likewise.
14047 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14048 (ix86_build_const_vector): Likewise.
14049 (ix86_build_signbit_mask): Likewise.
14050 (ix86_match_ccmode): Likewise.
14051 (ix86_cc_modes_compatible): Likewise.
14052 (ix86_expand_branch): Likewise.
14053 (ix86_expand_sse_cmp): Likewise.
14054 (ix86_expand_sse_movcc): Likewise.
14055 (ix86_expand_int_sse_cmp): Likewise.
14056 (ix86_expand_vec_perm_vpermi2): Likewise.
14057 (ix86_expand_vec_perm): Likewise.
14058 (ix86_expand_sse_unpack): Likewise.
14059 (ix86_expand_int_addcc): Likewise.
14060 (ix86_split_to_parts): Likewise.
14061 (ix86_vectorize_builtin_gather): Likewise.
14062 (ix86_vectorize_builtin_scatter): Likewise.
14063 (avx_vpermilp_parallel): Likewise.
14064 (inline_memory_move_cost): Likewise.
14065 (ix86_tieable_integer_mode_p): Likewise.
14066 (x86_maybe_negate_const_int): Likewise.
14067 (ix86_expand_vector_init_duplicate): Likewise.
14068 (ix86_expand_vector_init_one_nonzero): Likewise.
14069 (ix86_expand_vector_init_one_var): Likewise.
14070 (ix86_expand_vector_init_concat): Likewise.
14071 (ix86_expand_vector_init_interleave): Likewise.
14072 (ix86_expand_vector_init_general): Likewise.
14073 (ix86_expand_vector_set): Likewise.
14074 (ix86_expand_vector_extract): Likewise.
14075 (emit_reduc_half): Likewise.
14076 (ix86_emit_i387_round): Likewise.
14077 (ix86_mangle_type): Likewise.
14078 (ix86_expand_round_sse4): Likewise.
14079 (expand_vec_perm_blend): Likewise.
14080 (canonicalize_vector_int_perm): Likewise.
14081 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14082 (expand_vec_perm_1): Likewise.
14083 (expand_vec_perm_interleave3): Likewise.
14084 (expand_vec_perm_even_odd_pack): Likewise.
14085 (expand_vec_perm_even_odd_1): Likewise.
14086 (expand_vec_perm_broadcast_1): Likewise.
14087 (ix86_vectorize_vec_perm_const_ok): Likewise.
14088 (ix86_expand_vecop_qihi): Likewise.
14089 (ix86_expand_mul_widen_hilo): Likewise.
14090 (ix86_expand_sse2_abs): Likewise.
14091 (ix86_expand_pextr): Likewise.
14092 (ix86_expand_pinsr): Likewise.
14093 (ix86_preferred_simd_mode): Likewise.
14094 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14095 * config/i386/sse.md (*andnot<mode>3): Likewise.
14096 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14097 (*<code><mode>3): Likewise.
14098 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14099 (ia64_expand_atomic_op): Likewise.
14100 (ia64_arg_type): Likewise.
14101 (ia64_mode_to_int): Likewise.
14102 (ia64_scalar_mode_supported_p): Likewise.
14103 (ia64_vector_mode_supported_p): Likewise.
14104 (expand_vec_perm_broadcast): Likewise.
14105 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14106 (iq2000_function_arg_advance): Likewise.
14107 (iq2000_function_arg): Likewise.
14108 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14109 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14110 (m68k_libcall_value): Likewise.
14111 (m68k_function_value): Likewise.
14112 (sched_attr_op_type): Likewise.
14113 * config/mcore/mcore.c (mcore_output_move): Likewise.
14114 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14116 (microblaze_function_arg): Likewise.
14117 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14118 (mips_print_operand): Likewise.
14119 (mips_mode_ok_for_mov_fmt_p): Likewise.
14120 (mips_vector_mode_supported_p): Likewise.
14121 (mips_preferred_simd_mode): Likewise.
14122 (mips_expand_vpc_loongson_even_odd): Likewise.
14123 (mips_expand_vec_unpack): Likewise.
14124 (mips_expand_vi_broadcast): Likewise.
14125 (mips_expand_vector_init): Likewise.
14126 (mips_expand_vec_reduc): Likewise.
14127 (mips_expand_msa_cmp): Likewise.
14128 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14129 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14130 (cc_flags_for_mode): Likewise.
14131 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14132 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14133 (nds32_output_casesi_pc_relative): Likewise.
14134 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14135 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
14136 (nvptx_gen_unpack): Likewise.
14137 (nvptx_gen_pack): Likewise.
14138 (nvptx_gen_shuffle): Likewise.
14139 (nvptx_gen_wcast): Likewise.
14140 (nvptx_preferred_simd_mode): Likewise.
14141 * config/pa/pa.c (pa_secondary_reload): Likewise.
14142 * config/pa/predicates.md (base14_operand): Likewise.
14143 * config/powerpcspe/powerpcspe-c.c
14144 (altivec_resolve_overloaded_builtin): Likewise.
14145 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
14147 (rs6000_preferred_simd_mode): Likewise.
14148 (output_vec_const_move): Likewise.
14149 (rs6000_expand_vector_extract): Likewise.
14150 (rs6000_split_vec_extract_var): Likewise.
14151 (reg_offset_addressing_ok_p): Likewise.
14152 (rs6000_legitimate_offset_address_p): Likewise.
14153 (rs6000_legitimize_address): Likewise.
14154 (rs6000_emit_set_const): Likewise.
14155 (rs6000_const_vec): Likewise.
14156 (rs6000_emit_move): Likewise.
14157 (spe_build_register_parallel): Likewise.
14158 (rs6000_darwin64_record_arg_recurse): Likewise.
14159 (swap_selector_for_mode): Likewise.
14160 (spe_init_builtins): Likewise.
14161 (paired_init_builtins): Likewise.
14162 (altivec_init_builtins): Likewise.
14163 (do_load_for_compare): Likewise.
14164 (rs6000_generate_compare): Likewise.
14165 (rs6000_expand_float128_convert): Likewise.
14166 (emit_load_locked): Likewise.
14167 (emit_store_conditional): Likewise.
14168 (rs6000_output_function_epilogue): Likewise.
14169 (rs6000_handle_altivec_attribute): Likewise.
14170 (rs6000_function_value): Likewise.
14171 (emit_fusion_gpr_load): Likewise.
14172 (emit_fusion_p9_load): Likewise.
14173 (emit_fusion_p9_store): Likewise.
14174 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
14175 (fusion_gpr_mem_load): Likewise.
14176 (fusion_addis_mem_combo_load): Likewise.
14177 (fusion_addis_mem_combo_store): Likewise.
14178 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
14179 (fusion_gpr_mem_load): Likewise.
14180 (fusion_addis_mem_combo_load): Likewise.
14181 (fusion_addis_mem_combo_store): Likewise.
14182 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14184 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
14185 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14186 (rs6000_preferred_simd_mode): Likewise.
14187 (output_vec_const_move): Likewise.
14188 (rs6000_expand_vector_extract): Likewise.
14189 (rs6000_split_vec_extract_var): Likewise.
14190 (reg_offset_addressing_ok_p): Likewise.
14191 (rs6000_legitimate_offset_address_p): Likewise.
14192 (rs6000_legitimize_address): Likewise.
14193 (rs6000_emit_set_const): Likewise.
14194 (rs6000_const_vec): Likewise.
14195 (rs6000_emit_move): Likewise.
14196 (rs6000_darwin64_record_arg_recurse): Likewise.
14197 (swap_selector_for_mode): Likewise.
14198 (paired_init_builtins): Likewise.
14199 (altivec_init_builtins): Likewise.
14200 (rs6000_expand_float128_convert): Likewise.
14201 (emit_load_locked): Likewise.
14202 (emit_store_conditional): Likewise.
14203 (rs6000_output_function_epilogue): Likewise.
14204 (rs6000_handle_altivec_attribute): Likewise.
14205 (rs6000_function_value): Likewise.
14206 (emit_fusion_gpr_load): Likewise.
14207 (emit_fusion_p9_load): Likewise.
14208 (emit_fusion_p9_store): Likewise.
14209 * config/rx/rx.c (rx_gen_move_template): Likewise.
14210 (flags_from_mode): Likewise.
14211 * config/s390/predicates.md (s390_alc_comparison): Likewise.
14212 (s390_slb_comparison): Likewise.
14213 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
14214 (s390_vector_mode_supported_p): Likewise.
14215 (s390_cc_modes_compatible): Likewise.
14216 (s390_match_ccmode_set): Likewise.
14217 (s390_canonicalize_comparison): Likewise.
14218 (s390_emit_compare_and_swap): Likewise.
14219 (s390_branch_condition_mask): Likewise.
14220 (s390_rtx_costs): Likewise.
14221 (s390_secondary_reload): Likewise.
14222 (__SECONDARY_RELOAD_CASE): Likewise.
14223 (s390_expand_cs): Likewise.
14224 (s390_preferred_simd_mode): Likewise.
14225 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
14226 * config/sh/sh.c (sh_print_operand): Likewise.
14227 (dump_table): Likewise.
14228 (sh_secondary_reload): Likewise.
14229 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14230 * config/sh/sh.md (casesi_worker_1): Likewise.
14231 (casesi_worker_2): Likewise.
14232 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
14233 (fcc_comparison_operator): Likewise.
14234 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14235 (emit_soft_tfmode_cvt): Likewise.
14236 (sparc_preferred_simd_mode): Likewise.
14237 (output_cbranch): Likewise.
14238 (sparc_print_operand): Likewise.
14239 (sparc_expand_vec_perm_bmask): Likewise.
14240 (vector_init_bshuffle): Likewise.
14241 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14242 (spu_vector_mode_supported_p): Likewise.
14243 (spu_expand_insv): Likewise.
14244 (spu_emit_branch_or_set): Likewise.
14245 (spu_handle_vector_attribute): Likewise.
14246 (spu_builtin_splats): Likewise.
14247 (spu_builtin_extract): Likewise.
14248 (spu_builtin_promote): Likewise.
14249 (spu_expand_sign_extend): Likewise.
14250 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14251 (tilegx_simd_int): Likewise.
14252 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14253 (tilepro_simd_int): Likewise.
14254 * config/v850/v850.c (const_double_split): Likewise.
14255 (v850_print_operand): Likewise.
14256 (ep_memory_offset): Likewise.
14257 * config/vax/vax.c (vax_rtx_costs): Likewise.
14258 (vax_output_int_move): Likewise.
14259 (vax_output_int_add): Likewise.
14260 (vax_output_int_subtract): Likewise.
14261 * config/visium/predicates.md (visium_branch_operator): Likewise.
14262 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14263 (visium_print_operand_address): Likewise.
14264 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14265 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14266 (xtensa_expand_conditional_branch): Likewise.
14267 (xtensa_copy_incoming_a7): Likewise.
14268 (xtensa_output_literal): Likewise.
14269 * dfp.c (decimal_real_maxval): Likewise.
14270 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14273 Alan Hayward <alan.hayward@arm.com>
14274 David Sherwood <david.sherwood@arm.com>
14276 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14277 (mode_nunits_inline): Likewise.
14278 (mode_inner_inline): Likewise.
14279 (mode_unit_size_inline): Likewise.
14280 (mode_unit_precision_inline): Likewise.
14281 (emit_insn_modes_h): Likewise. Also emit a #define of the
14283 (emit_mode_wider): Add an E_ prefix to mode names.
14284 (emit_mode_complex): Likewise.
14285 (emit_mode_inner): Likewise.
14286 (emit_mode_adjustments): Likewise.
14287 (emit_mode_int_n): Likewise.
14288 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14289 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14290 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14291 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14292 (CRC32_BUILTIN, ENTRY): Likewise.
14293 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14294 (aarch64_pop_regs): Likewise.
14295 (aarch64_process_components): Likewise.
14296 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14297 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14298 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14299 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14300 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14301 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14303 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14304 (pcmpistr): Likewise.
14305 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14306 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14307 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14309 * config/rl78/rl78.c (mduc_regs): Likewise.
14310 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14311 (htm_expand_builtin): Likewise.
14312 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14313 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14314 * config/xtensa/xtensa.c (print_operand): Likewise.
14315 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14316 (NUM_MODE_VECTOR_INT): Likewise.
14317 * genoutput.c (null_operand): Likewise.
14318 (output_operand_data): Likewise.
14319 * genrecog.c (print_parameter_value): Likewise.
14320 * lra.c (debug_operand_data): Likewise.
14322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14324 * dwarf2out.c (scompare_loc_descriptor_wide)
14325 (scompare_loc_descriptor_narrow): New functions, split out from...
14326 (scompare_loc_descriptor): ...here.
14327 * expmed.c (emit_store_flag_int): New function, split out from...
14328 (emit_store_flag): ...here.
14330 2017-08-30 Richard Biener <rguenther@suse.de>
14332 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14333 (dwarf2out_early_finish): Move setting of AT_pubnames from
14334 early debug output to early finish.
14336 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14338 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14339 and -mdata-region to the assembler.
14341 2017-08-30 Richard Biener <rguenther@suse.de>
14343 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14345 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14346 (dwarf2out_early_global_decl): Only generate a DIE for the
14347 abstract origin if it doesn't already exist or is a declaration DIE.
14348 (resolve_addr): Do not add the linkage name twice when
14349 generating a stub DIE for the DW_TAG_GNU_call_site target.
14351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14353 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14354 Use machine_mode rather than int for arg1_mode.
14356 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14359 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14360 that the second argument of the built-in functions to unpack
14361 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14362 switch statement instead a lot of if statements.
14363 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14364 Allow 64-bit values to be in Altivec registers as well as
14365 traditional floating point registers.
14366 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14368 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14370 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14371 MAX_REGS_PER_ADDRESS == 1.
14373 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14375 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14376 * config/i386/i386.c (ix86_option_override_internal): Simplify
14377 setting of opts->x_flag_entry.
14379 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14380 Jakub Jelinek <jakub@redhat.com>
14381 Richard Biener <rguenther@suse.de>
14383 PR tree-optimization/81503
14384 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14385 folded constant fits in the target type; reorder tests for clarity.
14387 2017-08-29 Martin Liska <mliska@suse.cz>
14389 * passes.def: Include pass_lower_switch.
14390 * stmt.c (dump_case_nodes): Remove and move to
14391 tree-switch-conversion.
14392 (case_values_threshold): Likewise.
14393 (expand_switch_as_decision_tree_p): Likewise.
14394 (emit_case_decision_tree): Likewise.
14395 (expand_case): Likewise.
14396 (balance_case_nodes): Likewise.
14397 (node_has_low_bound): Likewise.
14398 (node_has_high_bound): Likewise.
14399 (node_is_bounded): Likewise.
14400 (emit_case_nodes): Likewise.
14401 (struct simple_case_node): New struct.
14402 (add_case_node): Remove.
14403 (emit_case_dispatch_table): Use vector instead of case_list.
14404 (reset_out_edges_aux): Remove.
14405 (compute_cases_per_edge): Likewise.
14406 (expand_case): Build list of simple_case_node.
14407 (expand_sjlj_dispatch_table): Use it.
14408 * tree-switch-conversion.c (struct case_node): Moved from
14409 stmt.c and adjusted.
14410 (emit_case_nodes): Likewise.
14411 (node_has_low_bound): Likewise.
14412 (node_has_high_bound): Likewise.
14413 (node_is_bounded): Likewise.
14414 (case_values_threshold): Likewise.
14415 (reset_out_edges_aux): Likewise.
14416 (compute_cases_per_edge): Likewise.
14417 (add_case_node): Likewise.
14418 (dump_case_nodes): Likewise.
14419 (balance_case_nodes): Likewise.
14420 (expand_switch_as_decision_tree_p): Likewise.
14421 (emit_jump): Likewise.
14422 (emit_case_decision_tree): Likewise.
14423 (try_switch_expansion): Likewise.
14424 (do_jump_if_equal): Likewise.
14425 (emit_cmp_and_jump_insns): Likewise.
14426 (fix_phi_operands_for_edge): New function.
14427 (record_phi_operand_mapping): Likewise.
14428 (class pass_lower_switch): New pass.
14429 (pass_lower_switch::execute): New function.
14430 (make_pass_lower_switch): Likewise.
14431 (conditional_probability):
14432 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14433 * tree-pass.h: Add make_pass_lower_switch.
14435 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14438 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
14441 2017-08-29 Richard Biener <rguenther@suse.de>
14443 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
14444 we do not add a DW_AT_inline attribute twice.
14445 (gen_subprogram_die): Remove code setting DW_AT_inline on
14446 DECL_ABSTRACT_P nodes.
14448 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
14450 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
14451 calls to internal functions.
14452 (gimplify_modify_expr): Likewise.
14453 * tree-call-cdce.c (use_internal_fn): Likewise.
14454 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
14455 (convert_to_divmod): Set the nothrow flag.
14456 * tree-if-conv.c (predicate_mem_writes): Likewise.
14457 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14458 (vectorizable_call): Likewise.
14459 (vectorizable_store): Likewise.
14460 (vectorizable_load): Likewise.
14461 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
14462 (vect_recog_mask_conversion_pattern): Likewise.
14464 2017-08-29 Martin Liska <mliska@suse.cz>
14467 * gcc.c (driver_handle_option): Add new argument.
14468 * opts-common.c (handle_option): Pass
14469 target_option_override_hook.
14470 * opts-global.c (lang_handle_option): Add new option.
14471 (set_default_handlers): Add new argument.
14472 (decode_options): Likewise.
14473 * opts.c (target_handle_option): Likewise.
14474 (common_handle_option): Call target_option_override_hook.
14475 * opts.h (struct cl_option_handler_func): Add hook for
14476 target option override.
14477 (struct cl_option_handlers): Likewise.
14478 (set_default_handlers): Add new argument.
14479 (decode_options): Likewise.
14480 (common_handle_option): Likewise.
14481 (target_handle_option): Likewise.
14482 * toplev.c (toplev::main): Pass targetm.target_option.override
14485 2017-08-29 Richard Biener <rguenther@suse.de>
14486 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
14488 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
14489 life to the active subtree.
14491 2017-08-28 Jeff Law <law@redhat.com>
14493 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
14494 derive_equivalences.
14495 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
14496 Code moved into....
14497 (edge_info::derive_equivalences): New private member function
14499 * tree-ssa-dom.c (class edge_info): Changed from a struct
14500 to a class. Add ctor/dtor, methods and data members.
14501 (edge_info::edge_info): Renamed from allocate_edge_info.
14502 Initialize additional members.
14503 (edge_info::~edge_info): New.
14504 (free_dom_edge_info): Delete the edge info.
14505 (record_edge_info): Use new class & associated member functions.
14506 Tighten forms for testing for edge equivalences.
14507 (record_temporary_equivalences): Iterate over the simple
14508 equivalences rather than assuming there's only one per edge.
14509 (cprop_into_successor_phis): Iterate over the simple
14510 equivalences rather than assuming there's only one per edge.
14511 (optimize_stmt): Use operand_equal_p rather than pointer
14512 equality for mini-DSE code.
14514 2017-08-28 Nathan Sidwell <nathan@acm.org>
14516 * gcc.c (execute): Fold SIGPIPE handling into switch
14517 statement. Adjust internal error message.
14519 2017-08-28 Richard Biener <rguenther@suse.de>
14522 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
14523 Do nothing for removed DIEs.
14525 2017-08-28 Richard Biener <rguenther@suse.de>
14527 PR tree-optimization/81977
14528 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
14531 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
14534 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
14535 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
14536 using targetm.gen_mem_thread_fence.
14538 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
14541 * config/i386/i386.md (*<btsc><mode>): Change operand 2
14542 predicate to register_operand. Reorder operands.
14543 (*btr<mode>): Ditto.
14544 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
14545 (*btr<mode>_mask): Ditto.
14547 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
14549 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
14550 * config/rs6000/xmmintrin.h: New file.
14551 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
14553 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14556 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
14557 parameter and_insn and return it.
14558 (recombine_lvx_pattern): Insert a copy to ensure availability of
14559 the base register of the copied masking operation at the point of
14560 the instruction replacement.
14561 (recombine_stvx_pattern): Likewise.
14563 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
14565 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
14566 undocumented switches.
14567 (-mpower9-dform-vector): Likewise.
14568 (-mpower9-dform): Likewise.
14569 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14570 comments to delete references to -mpower9-dform* switches.
14571 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
14572 Delete reference to -mpower9-dform* switches, test for
14573 -mpower9-vector instead.
14574 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14575 (OTHER_P9_VECTOR_MASKS): Likewise.
14576 (POWERPC_MASKS): Likewise.
14577 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14578 tests against -mpower9-dform* to -mpower9-vector. Delete code
14579 that checked for -mpower9-dform* consistancy with other options.
14580 Add test for -mpower9-misc to enable other power9 switches.
14581 (rs6000_init_hard_regno_mode_ok): Likewise.
14582 (rs6000_option_override_internal): Likewise.
14583 (rs6000_emit_prologue): Likewise.
14584 (rs6000_emit_epilogue): Likewise.
14585 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14586 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14587 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14589 (emit_fusion_p9_store): Likewise.
14590 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14591 resetting these macros if the assembler does not support ISA 3.0
14593 (TARGET_P9_DFORM_VECTOR): Likewise.
14594 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14595 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14597 2017-08-25 Alan Modra <amodra@gmail.com>
14599 PR rtl-optimization/81747
14600 * cse.c (cse_extended_basic_block): Don't attempt to record
14601 equivalences for degenerate conditional branches that branch
14602 to their fall-through.
14604 2017-08-24 Martin Sebor <msebor@redhat.com>
14606 PR middle-end/81908
14607 * gimple-fold.c (size_must_be_zero_p): New function.
14608 (gimple_fold_builtin_memory_op): Call it.
14610 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14612 * config/rs6000/mm_malloc.h: New file.
14614 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14616 PR tree-optimization/81913
14617 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14618 analysis when either IVs in condition can wrap.
14620 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14622 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14623 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14625 2017-08-24 Richard Biener <rguenther@suse.de>
14628 * targhooks.c (default_target_can_inline_p): Properly
14629 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14630 is present and always compare.
14631 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14632 infer -mfpmath=sse from TARGET_SSE_P.
14633 (ix86_can_inline_p): Properly use target_option_default_node when
14634 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14636 2017-08-24 Richard Biener <rguenther@suse.de>
14639 * dwarf2out.c (output_die): Handle flag_generate_offload like
14641 (output_comp_unit): Likewise.
14642 (gen_array_type_die): Likewise.
14643 (dwarf2out_early_finish): Likewise.
14644 (note_variable_value_in_expr): Likewise.
14645 (dwarf2out_finish): Likewise. Adjust assert.
14646 * cgraphunit.c (symbol_table::compile): Move setting of
14647 flag_generate_offload earlier ...
14648 (symbol_table::finalize_compilation_unit): ... here, before
14649 early debug finalization.
14651 2017-08-24 Richard Biener <rguenther@suse.de>
14653 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14654 and ipa-fnsummary.h.
14655 (ix86_can_inline_p): When ix86_fpmath flags do not match
14656 check whether the callee uses FP math at all.
14658 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14660 PR middle-end/81931
14661 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14662 instead of TYPE_PRECISION.
14664 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14665 Alan Hayward <alan.hayward@arm.com>
14666 David Sherwood <david.sherwood@arm.com>
14668 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14670 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14672 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14674 (ix86_frame::stack_realign_allocate): New field.
14675 (struct machine_frame_state): Modify comments.
14676 (machine_frame_state::sp_realigned_fp_end): New field.
14677 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14678 layout calculation.
14679 (sp_valid_at): Add assertion to assure no attempt to access invalid
14680 offset of a realigned stack.
14681 (fp_valid_at): Likewise.
14682 (choose_baseaddr): Modify comments.
14683 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14684 ix86_expand_prologue.
14685 (ix86_expand_prologue): Modify stack realignment and allocation.
14686 (ix86_expand_epilogue): Modify comments.
14687 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14688 avx2_runtime, avx512f, and avx512f_runtime.
14690 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14692 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14693 (mstackrealign): Do not init to -1.
14694 * config/i386/i386.c (ix86_option_override_internal):
14695 Check opts_set, not opts when setting default value of
14696 opts->x_ix86_force_align_arg_pointer.
14698 2017-08-23 Richard Biener <rguenther@suse.de>
14700 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14701 lang_hooks.decl_printable_name.
14702 * print-rtl-function.c (print_rtx_function): Likewise.
14703 * tree-pretty-print.c (dump_function_header): Likewise.
14705 2017-08-23 Richard Biener <rguenther@suse.de>
14708 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14709 -g0 at compile-time.
14711 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14713 PR middle-end/19706
14714 * doc/sourcebuild.texi (Other hardware attributes):
14717 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14719 PR middle-end/19706
14720 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14721 Fix single-use check.
14723 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14725 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14727 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14729 * doc/install.texi: Modify to add more details on running selected
14732 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14734 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14735 is combined with -mabi=ms.
14736 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14739 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14741 PR tree-optimization/81488
14742 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14743 and cached_basis fields.
14744 (MAX_SPREAD): New constant.
14745 (alloc_cand_and_find_basis): Initialize new fields.
14746 (clear_visited): New function.
14747 (create_phi_basis_1): Rename from create_phi_basis, set visited
14748 and cached_basis fields.
14749 (create_phi_basis): New wrapper function.
14750 (phi_add_costs_1): Rename from phi_add_costs, add spread
14751 parameter, set visited field, short-circuit when limits reached.
14752 (phi_add_costs): New wrapper function.
14753 (record_phi_increments_1): Rename from record_phi_increments, set
14755 (record_phi_increments): New wrapper function.
14756 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14757 (phi_incr_cost): New wrapper function.
14758 (all_phi_incrs_profitable_1): Rename from
14759 all_phi_incrs_profitable, set visited field.
14760 (all_phi_incrs_profitable): New wrapper function.
14762 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14763 Alan Hayward <alan.hayward@arm.com>
14764 David Sherwood <david.sherwood@arm.com>
14766 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14767 that takes the outer and inner modes.
14768 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14769 comparison as the canonical test for a paradoxical subreg.
14770 * combine.c (simplify_set): Use paradoxical_subreg_p.
14771 (make_extraction): Likewise.
14772 (force_to_mode): Likewise.
14773 (rtx_equal_for_field_assignment_p): Likewise.
14774 (gen_lowpart_for_combine): Likewise.
14775 (simplify_comparison): Likewise.
14776 * cse.c (equiv_constant): Likewise.
14777 * expmed.c (store_bit_field_1): Likewise.
14778 * final.c (alter_subreg): Likewise.
14779 * fwprop.c (propagate_rtx): Likewise.
14780 (forward_propagate_subreg): Likewise.
14781 * ira-conflicts.c (ira_build_conflicts): Likewise.
14782 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14783 * lra-constraints.c (curr_insn_transform): Likewise.
14784 (split_reg): Likewise.
14785 * lra-eliminations.c (move_plus_up): Likewise.
14786 (lra_eliminate_regs_1): Likewise.
14787 * recog.c (general_operand): Likewise.
14788 * ree.c (combine_reaching_defs): Likewise.
14789 * reload.c (push_reload): Likewise.
14790 (find_reloads): Likewise.
14791 * reload1.c (elimination_effects): Likewise.
14792 (compute_reload_subreg_offset): Likewise.
14793 (choose_reload_regs): Likewise.
14794 * rtlanal.c (subreg_lsb_1): Likewise.
14795 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14796 (simplify_subreg): Likewise.
14797 * var-tracking.c (track_loc_p): Likewise.
14798 * emit-rtl.c (byte_lowpart_offset): Likewise.
14799 (paradoxical_subreg_p): Delete out-of-line definition.
14801 2017-08-22 Jeff Law <law@redhat.com>
14803 PR tree-optimization/81741
14804 PR tree-optimization/71947
14805 * tree-ssa-dom.c: Include tree-inline.h.
14806 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14807 equivalences if one is more expensive to compute than the other.
14808 * tree-ssa-scopedtables.h (class const_or_copies): Make
14809 record_const_or_copy_raw method private.
14810 (class avail_exprs_stack): New method simplify_binary_operation.
14811 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14812 avail_exprs_stack::simplify_binary_operation as needed.
14813 (avail_exprs_stack::simplify_binary_operation): New function.
14815 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14817 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14818 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14819 (DOT_SYMBOLS): Likewise.
14820 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14821 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14822 (RS6000_ABI_NAME): Likewise.
14823 (TARGET_CMODEL): Likewise.
14824 (TOC_SECTION_ASM_OP): Likewise.
14825 (SET_CMODEL): New macro.
14826 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14828 2017-08-22 Richard Biener <rguenther@suse.de>
14830 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14831 to allow for free-lang-data replacements similar to verify_type_variant.
14833 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14834 Alan Hayward <alan.hayward@arm.com>
14835 David Sherwood <david.sherwood@arm.com>
14837 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14838 VOIDmode for the LABEL_REF.
14840 2017-08-22 Richard Biener <rguenther@suse.de>
14842 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14844 2017-08-22 Richard Biener <rguenther@suse.de>
14846 * common.opt (feliminate-dwarf2-dups): Ignore.
14847 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14848 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14849 same_die_p_wrap, compute_section_prefix,
14850 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14851 (comdat_symbol_id, comdat_symbol_number): Likewise.
14852 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14854 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14855 (output_die): Mark unreachable path unreachable.
14856 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14857 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14858 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14859 (dwarf2out_early_finish): Likewise.
14861 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14863 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14865 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14868 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14869 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14870 (avr_attribute_table) <io, io_low, address>: Initialize
14871 .decl_required with true.
14873 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14875 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14876 undocumented debugging options.
14877 (-mvsx-scalar-double): Likewise.
14878 (-mallow-df-permute): Likewise.
14879 (-mvectorize-builtins): Likewise.
14880 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14881 (rs6000_builtin_vectorized_function): Likewise.
14882 (rs6000_builtin_md_vectorized_function): Likewise.
14883 (rs6000_opt_vars): Likewise.
14885 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14888 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14889 (*btrq_imm): Rename from *btrq.
14890 (*btcq_imm): Rename from *btcq.
14891 (btsc): New code attribute.
14892 (*<btsc><mode>): New insn pattern.
14893 (*btr<mode>): Ditto.
14894 (*<btsc><mode>_mask): New insn_and_split pattern.
14895 (*btr<mode>_mask): Ditto.
14897 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14898 Alan Hayward <alan.hayward@arm.com>
14899 David Sherwood <david.sherwood@arm.com>
14901 * function.c (pad_below): Simplify padding calculation.
14903 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14904 Alan Hayward <alan.hayward@arm.com>
14905 David Sherwood <david.sherwood@arm.com>
14907 * target.def (function_prologue): Remove frame size argument.
14908 (function_epilogue): Likewise.
14909 * doc/tm.texi: Regenerate.
14910 * final.c (final_start_function): Update call to function_prologue.
14911 (final_end_function): Update call to function_epilogue.
14912 (default_function_pro_epilogue): Remove frame size argument.
14913 * output.h (default_function_pro_epilogue): Likewise.
14914 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14915 (arm_output_function_prologue): Likewise.
14916 * config/frv/frv.c (frv_function_prologue): Likewise.
14917 (frv_function_epilogue): Likewise.
14918 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14919 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14920 (ia64_output_function_epilogue): Likewise.
14921 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14922 (m32r_output_function_epilogue): Likewise.
14923 * config/microblaze/microblaze.c (microblaze_function_prologue)
14924 (microblaze_function_epilogue): Likewise.
14925 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14926 (mips_output_function_epilogue): Likewise.
14927 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14928 (mmix_target_asm_function_epilogue): Likewise.
14929 * config/msp430/msp430.c (msp430_start_function): Likewise.
14930 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14931 (nds32_asm_function_epilogue): Likewise.
14932 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14933 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14934 (pa_output_function_epilogue): Likewise.
14935 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14936 (rs6000_output_function_epilogue): Likewise.
14937 * config/rl78/rl78.c (rl78_start_function): Likewise.
14938 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14939 (rs6000_output_function_epilogue): Likewise.
14940 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14941 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14942 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14943 (sparc_asm_function_epilogue): Likewise.
14945 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14947 * tree.h (type_has_mode_precision_p): New function.
14948 * convert.c (convert_to_integer_1): Use it.
14949 * expr.c (expand_expr_real_2): Likewise.
14950 (expand_expr_real_1): Likewise.
14951 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14952 * match.pd: Likewise.
14953 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14954 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14955 * tree-tailcall.c (process_assignment): Likewise.
14956 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14957 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14958 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14959 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14960 (vectorizable_assignment): Likewise.
14961 (vectorizable_shift): Likewise.
14962 (vectorizable_operation): Likewise.
14963 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14965 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14967 * match.pd: Add pow (C, x) simplification.
14969 2017-08-21 Richard Biener <rguenther@suse.de>
14971 PR tree-optimization/81900
14972 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14973 for blocks with abnormal predecessors.
14974 (compute_antic): Do not set visited flag prematurely.
14976 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14979 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14981 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14983 * stor-layout.h (vector_type_mode): Move to...
14984 * tree.h (vector_type_mode): ...here.
14985 * stor-layout.c (vector_type_mode): Move to...
14986 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14988 2017-08-21 Richard Biener <rguenther@suse.de>
14990 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14991 register_external_die hooks.
14992 (debug_false_tree_charstarstar_uhwistar): Declare.
14993 (debug_nothing_tree_charstar_uhwi): Likewise.
14994 * debug.c (do_nothing_debug_hooks): Adjust.
14995 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14996 (debug_nothing_tree_charstar_uhwi): Likewise.
14997 * dbxout.c (dbx_debug_hooks): Adjust.
14998 (xcoff_debug_hooks): Likewise.
14999 * sdbout.c (sdb_debug_hooks): Likewise.
15000 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15001 * dwarf2out.c (macinfo_label_base): New global.
15002 (dwarf2out_register_external_die): New function for the
15003 register_external_die hook.
15004 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15005 (dwarf2_debug_hooks): Use them.
15006 (dwarf2_lineno_debug_hooks): Adjust.
15007 (struct die_struct): Add with_offset flag.
15008 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15009 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15010 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15011 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15012 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15013 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15014 defining section names for the early LTO debug variants.
15015 (reset_indirect_string): New helper.
15016 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15017 (print_dw_val): Add support for offsetted symbol references.
15018 (get_ultimate_context): Split out from is_cxx.
15019 (is_cxx): Use get_ultimate_context.
15020 (is_fortran): Add decl overload.
15021 (compute_comp_unit_symbol): Split out worker from
15022 compute_section_prefix.
15023 (compute_section_prefix): Call compute_comp_unit_symbol and
15024 set comdat_type_p here.
15025 (output_die): Skip DIE symbol output for the LTO added one.
15026 Handle DIE symbol references with offset.
15027 (output_comp_unit): Guard section name mangling properly.
15028 For LTO debug sections emit a symbol at the section beginning
15029 which we use to refer to its DIEs.
15030 (add_abstract_origin_attribute): For DIEs registered via
15031 dwarf2out_register_external_die directly refer to the early
15032 DIE rather than indirectly through the shadow one we created.
15033 Remove obsolete call to dwarf2out_abstract_function for
15034 non-function/block origins.
15035 (gen_array_type_die): When generating early LTO debug do
15036 not emit DW_AT_string_length.
15037 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15038 late when in LTO. As suggested place a gcc_unreachable for
15039 the DECL_ABSTRACT_P case.
15040 (gen_subprogram_die): Avoid another specification DIE
15041 for early built declarations/definitions for the late LTO case.
15042 (gen_variable_die): Add type references for late duplicated VLA dies
15044 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15045 we have the abstract instance already.
15046 (process_scope_var): Adjust decl DIE contexts in LTO which
15047 first puts them in limbo.
15048 (gen_decl_die): Do not generate type DIEs late apart from
15049 types for VLAs or for decls we do not yet have a DIE. Do not
15050 call dwarf2out_abstract_function late.
15051 (dwarf2out_early_global_decl): Make sure to create DIEs
15052 for abstract instances of a decl first.
15053 (dwarf2out_late_global_decl): Adjust comment.
15054 (output_macinfo_op): With multiple macro sections use
15055 macinfo_label_base to distinguish labels.
15056 (output_macinfo): Likewise. Update macinfo_label_base.
15057 Pass in the line info label.
15058 (note_variable_value_in_expr): When generating LTO resolve
15059 all variable values here by generating DIEs as needed.
15060 (init_sections_and_labels): Add early LTO debug flag parameter
15061 and generate different sections and names if set. Add generation
15062 counter for the labels so we can have multiple of them.
15063 (reset_dies): Helper to allow DIEs to be output multiple times.
15064 (dwarf2out_finish): When outputting DIEs to the fat part of an
15065 LTO object first reset DIEs.
15066 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15067 (modified_type_die): Check for decl_ultimate_origin being self
15069 (gen_type_die_with_usage): Likewise.
15070 (gen_typedef_die): Allow decl_ultimate_origin being self.
15071 (set_decl_abstract_flags): Remove.
15072 (set_block_abstract_flags): Likewise.
15073 (dwarf2out_abstract_function): Treat the early generated DIEs
15074 as the abstract copy and only add DW_AT_inline and
15075 DW_AT_artificial here and call set_decl_origin_self.
15076 If the DIE has an abstract origin don't do anything.
15077 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15078 if we have none yet (Go fails to build one, PR78628).
15079 (variably_modified_type_p): Prevent endless recursion for Ada
15080 cyclic pointer types.
15081 * lto-streamer-in.c: Include debug.h.
15082 (dref_queue): New global.
15083 (lto_read_tree_1): Stream in DIE references.
15084 (lto_input_tree): Register DIE references.
15085 (input_function): Stream DECL_DEBUG_ARGS.
15086 * lto-streamer-out.c: Include debug.h.
15087 (lto_write_tree_1): Output DIE references.
15088 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15089 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15090 (output_function): Stream DECL_DEBUG_ARGS.
15091 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15092 Stream DECL_ABSTRACT_ORIGIN.
15093 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15094 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15095 DECL_CONTEXT for file-scope decls.
15096 * lto-streamer.h (struct dref_entry): Declare.
15097 (dref_queue): Likewise.
15098 * cfgexpand.c (pass_expand::execute): Do not call the
15099 outlining_inline_function hook here.
15100 * lto-wrapper.c (debug_obj): New global.
15101 (tool_cleanup): Unlink it if required.
15102 (debug_objcopy): New function.
15103 (run_gcc): Handle early debug sections in the IL files by
15104 extracting them to separate files, partially linkin them and
15105 feeding the result back as result to the linker.
15106 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15107 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15108 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15109 sections into a separate segment.
15110 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15112 (darwin_asm_dwarf_section): Likewise.
15113 (darwin_asm_output_dwarf_offset): Likewise.
15114 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15116 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15117 Alan Hayward <alan.hayward@arm.com>
15118 David Sherwood <david.sherwood@arm.com>
15120 * read-md.h (md_reader::record_potential_iterator_use): Replace
15121 pointer argument with an rtx and an index.
15122 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15123 (apply_mode_iterator): Likewise.
15124 (apply_code_iterator): Likewise.
15125 (apply_int_iterator): Likewise.
15126 (apply_subst_iterator): Likewise.
15127 (record_iterator_use): Likewise.
15128 (record_attribute_use): Likewise.
15129 (md_reader::record_potential_iterator_use): Likewise. Update calls
15130 to record_iterator_use and apply_iterator.
15131 (iterator_use): Replace ptr with x and index.
15132 (attribute_use): Likewise.
15133 (apply_attribute_uses): Update calls to apply_iterator.
15134 (apply_iterators): Likewise. Update initialization of iterator_use.
15135 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
15136 and record_potential_iterator_use.
15137 (rtx_reader::read_rtx_operand): Likewise.
15139 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15140 Alan Hayward <alan.hayward@arm.com>
15141 David Sherwood <david.sherwood@arm.com>
15143 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
15146 2017-08-21 Richard Biener <rguenther@suse.de>
15148 PR middle-end/81884
15149 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
15150 at struct end conservatively when comparing common bases.
15152 2017-08-21 Richard Biener <rguenther@suse.de>
15154 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
15155 (mem_ref_in_stmt): Remove.
15156 (determine_max_movement): Use ref index to get at the reference.
15157 (invariantness_dom_walker::before_dom_children): Deal with
15158 lim data already initialized.
15159 (gather_mem_refs_stmt): Initialize lim data and record ref index.
15161 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15163 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
15164 (TARGET_ISA_ROUND): Ditto.
15165 (TARGET_ROUND): Ditto.
15166 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
15167 * config/i386/i386.md: Ditto.
15168 * config/i386/sse.md: Ditto.
15169 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
15170 with OPTION_MASK_ISA_SSE4_1.
15172 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15175 * doc/extend.texi (x86 Built-in Functions): Correct the name of
15176 __builtin_ia32_lzcnt_u16.
15178 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
15181 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
15182 (rs6000_set_current_function): Rewrite function to use it.
15184 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
15187 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
15188 and TYPE_WARN_IF_NOT_ALIGN.
15189 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
15190 (handle_warn_if_not_align): New.
15191 (place_union_field): Call handle_warn_if_not_align.
15192 (place_field): Call handle_warn_if_not_align.
15193 Copy TYPE_WARN_IF_NOT_ALIGN.
15194 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
15195 (layout_type): Likewise.
15196 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
15198 (tree_decl_common): Add warn_if_not_align.
15199 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
15200 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
15201 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
15202 (DECL_WARN_IF_NOT_ALIGN): Likewise.
15203 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
15204 * doc/extend.texi: Document warn_if_not_aligned attribute.
15205 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
15207 2017-08-17 Martin Liska <mliska@suse.cz>
15210 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
15211 (get_data_dependence): Use it as pointer type.
15212 (distribute_loop): Likewise.
15214 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15216 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
15217 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
15218 (p8_vmrgow_<mode>_direct): New define_insn.
15219 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
15220 handle endianness for vmrgew and vmrgow permute patterns.
15222 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15224 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
15225 * config/rs6000/rs6000-cpus.def: Remove comment.
15226 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
15227 (POWERPC_MASKS): Likewise.
15228 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
15229 use of TARGET_VSX_TIMODE.
15230 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
15231 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
15232 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
15233 (rs6000_option_override_internal): Remove dead code.
15234 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15235 (rs6000_legitimize_reload_address): Likewise.
15236 (rs6000_legitimate_address_p): Likewise.
15237 (rs6000_opt_masks): Delete "vsx-timode".
15238 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15239 from function comment.
15240 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15241 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15242 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15244 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15245 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15246 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15247 (VSX_TI): Likewise.
15249 (define_peephole2): Likewise.
15251 2017-08-17 Martin Sebor <msebor@redhat.com>
15254 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15255 past the end of an array.
15256 (test_pp_format): Add test cases.
15258 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15260 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15261 missing ECF_NOTHROW flags.
15263 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15266 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15267 operands residing in integer registers.
15268 (*vsx_le_perm_load_<mode>): Likewise.
15269 (*vsx_le_perm_store_<mode>): Likewise.
15270 (define_peephole2): Add peepholes to optimize the above.
15272 2017-08-17 Marek Polacek <polacek@redhat.com>
15274 PR middle-end/81814
15275 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15276 to mimic what shorten_compare did. Change the return type to bool.
15277 (fold_cond_expr_with_comparison): Update call to
15278 operand_equal_for_comparison_p.
15279 (fold_ternary_loc): Likewise.
15281 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15283 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15285 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15287 2017-08-17 Richard Biener <rguenther@suse.de>
15289 * tree-ssa-structalias.c (solve_graph): When propagating
15290 to successors update the graphs succ edges and avoid duplicate work.
15292 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15295 * config/i386/i386.c (ix86_option_override_internal): Save target
15296 specific options after ix86_stack_protector_guard_reg was changed.
15298 2017-08-17 Richard Biener <rguenther@suse.de>
15300 PR tree-optimization/81827
15301 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15302 (new_var_info): Initialize it conservatively.
15303 (get_call_vi): Mark register vars.
15304 (new_scalar_tmp_constraint_exp): Likewise.
15305 (handle_rhs_call): Likewise.
15306 (handle_const_call): Likewise.
15307 (create_function_info_for): Likewise.
15308 (solve_constraints): Sort varinfos to separate register from
15309 non-register vars to pack points-to solution bitmaps during
15312 2017-08-17 Marek Polacek <polacek@redhat.com>
15314 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15316 2017-08-17 Richard Biener <rguenther@suse.de>
15318 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15319 to true when overflow is undefined and we saturated the result.
15321 2017-08-17 Alan Modra <amodra@gmail.com>
15324 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15325 Don't use store multiple if only one reg needs saving.
15326 (interesting_frame_related_regno): New function.
15327 (rs6000_frame_related): Don't emit frame info for regs that
15329 (rs6000_emit_epilogue): Likewise.
15331 2017-08-16 Nathan Sidwell <nathan@acm.org>
15333 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15334 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15335 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15336 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15337 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15338 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15339 (verify_type): Adjust for TYPE_BINFO move.
15340 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15341 process TYPE_BINFO directly.
15342 (hash_tree): Likewise.
15343 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15345 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15348 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15350 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15352 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15355 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15356 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15357 Add ix86_binary_operator_ok to insn constraint.
15358 (*iordi_1_bts): Ditto.
15359 (*xordi_1_btc): Ditto.
15360 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15361 Update corresponding peephole2 pattern.
15365 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15367 PR tree-optimization/81832
15368 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15369 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15371 2017-08-16 Marek Polacek <polacek@redhat.com>
15374 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15375 regarding size_zero_node.
15377 2017-08-16 Martin Liska <mliska@suse.cz>
15380 * config.gcc: Respect previously set extra_objs in case
15383 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15385 PR tree-optimization/81835
15386 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15387 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15388 not depend on the phi.
15390 2017-08-16 Alan Modra <amodra@gmail.com>
15392 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15395 2017-08-16 Alan Modra <amodra@gmail.com>
15397 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15398 (save_reg_p): ..into this. Update all callers.
15399 (first_reg_to_save): Simplify.
15401 2017-08-16 Alan Modra <amodra@gmail.com>
15403 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15406 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15410 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15411 constant count if that count is less than 32.
15413 2017-08-15 Nathan Sidwell <nathan@acm.org>
15415 * gcc.c (execute): Emit friendlier message if inferior is killed
15416 by an external cause.
15418 2017-08-15 Richard Biener <rguenther@suse.de>
15420 PR tree-optimization/81790
15421 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15422 CONSTRUCTORs from simplifying and VN.
15424 2017-08-14 Martin Sebor <msebor@redhat.com>
15426 * builtin-attrs.def: Add comments.
15428 2017-08-14 Martin Sebor <msebor@redhat.com>
15431 * doc/extend.texi (attribute nonstring): Document new attribute.
15433 2017-08-14 Martin Sebor <msebor@redhat.com>
15436 * tree-diagnostic.c (default_tree_printer): Handle %G.
15437 * gimple-pretty-print.h (percent_G_format): Declare new function.
15438 * gimple-pretty-print.c (percent_G_format): Define.
15439 * tree-pretty-print.c (percent_K_format): Add argument.
15441 2017-08-14 Martin Sebor <msebor@redhat.com>
15443 PR translation/79998
15444 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
15445 Remove a stray space.
15447 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
15450 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
15451 (*iordi_1_bts): Ditto.
15452 (*xordi_1_btc): Ditto.
15454 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15457 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
15458 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15460 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
15461 quoted strings, and make more translator-friendly.
15462 (darwin_rs6000_override_options): Likewise.
15463 (rs6000_option_override_internal): Likewise.
15464 (rs6000_return_in_memory): Fix overlong line.
15465 (init_cmulative_args): Use quoted strings, and make more
15466 translator-friendly.
15467 (rs6000_pass_by_reference): Fix overlong line.
15468 (def_builtin): Use quoted strings.
15469 (altivec_expand_predicate_builtin): Use quoted strings, and make
15470 more translator-friendly.
15471 (htm_expand_builtin): Use quoted strings.
15472 (cpu_expand_builtin): Use quoted strings, and make more
15473 translator-friendly.
15474 (altivec_expand_builtin): Likewise.
15475 (paired_expand_predicate_builtin): Likewise.
15476 (rs6000_invalid_builtin): Likewise.
15477 (builtin_function_type): Use quoted strings.
15478 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
15479 more translator-friendly.
15480 (rs6000_trampoline_init): Likewise.
15481 (rs6000_handle_altivec_attribute): Likewise.
15482 (rs6000_inner_target_options): Use quoted strings.
15483 (rs6000_disable_incompatible_switches): Likewise.
15484 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
15485 strings, and make more translator-friendly.
15486 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
15488 2017-08-14 Bin Cheng <bin.cheng@arm.com>
15490 PR tree-optimization/81799
15491 * tree-loop-distribution.c (version_loop_by_alias_check): Force
15492 cond_expr to simple gimple operand.
15494 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
15496 PR middle-end/46932
15497 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
15499 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
15503 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
15504 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
15505 TARGET_GASISR_PROLOGUES.
15506 * config/avr/avr.c (avr_option_override): Same.
15507 (avr_pass_pre_proep::execute): Same.
15509 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
15512 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
15513 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
15515 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
15517 * config/i386/i386.md (*load_tp_<mode>): Redefine as
15518 define_insn_and_split. Split to a memory load from 0 in
15519 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
15520 using PTR mode iterator.
15521 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
15522 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
15523 (*add_tp_<mode>): Redefine as define_insn_and_split.
15524 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
15525 address space. Merge with *add_tp_x32 using PTR mode iterator.
15526 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
15527 Split to an add with a memory load from 0 in
15528 DEFAULT_TLS_SEG_REG address space.
15530 2017-08-12 Andrew Pinski <apinski@cavium.com>
15532 * config/aarch64/aarch64-option-extensions.def (rdma):
15533 Fix feature string to what Linux prints out in /proc/cpuinfo.
15535 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
15538 * dwarf2out.c (modified_type_die): For C typedef types that have
15539 an ultimate origin, process the ultimate origin instead of the
15541 (gen_typedef_die): Assert that input DECLs have no ultimate
15543 (gen_type_die_with_usage): For typedef variants that have an
15544 ultimate origin, just call gen_decl_die on the original DECL.
15545 (process_scope_var): Avoid creating DIEs for local typedefs and
15546 concrete static variables.
15548 2017-08-12 Alan Modra <amodra@gmail.com>
15552 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
15553 match gnu-user.h startfile.
15554 (ENDFILE_LINUX_SPEC): Similarly.
15556 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
15559 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
15561 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
15563 2017-08-11 Tamar Christina <tamar.christina@arm.com>
15564 * config/aarch64/aarch64.md (mov<mode>): Change.
15565 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15566 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
15567 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
15569 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
15571 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
15572 for storage order barriers.
15574 2017-08-11 Martin Liska <mliska@suse.cz>
15577 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15578 variables of void type.
15580 2017-08-11 Martin Liska <mliska@suse.cz>
15582 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15583 TARGET_SUPPORTS_ALIASES.
15584 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15585 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15586 (optimize_weakref): Likewise.
15587 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15588 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15589 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15591 2017-08-11 Martin Liska <mliska@suse.cz>
15594 * config/i386/i386.c (make_resolver_func): Do complete
15595 refactoring of the function.
15597 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15600 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15601 * config/i386/i386.c (ix86_stack_protect_guard): Use
15602 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15603 * doc/invoke.texi (x86 Options): Document
15604 -mstack-protector-guard-symbol= option.
15606 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15608 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15609 * config/i386/i386.c (ix86_split_stack_guard): New function.
15610 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15611 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15612 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15613 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15614 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15615 (split_stack_space_check): Call ix86_split_stack_guard.
15617 2017-08-10 Martin Sebor <msebor@redhat.com>
15619 * print-tree.c (print_node): Print location using the established
15621 Replace spaces with colons.
15622 (debug_raw, debug): Ditto.
15624 2017-08-10 Martin Sebor <msebor@redhat.com>
15627 * pretty-print.c (pp_format): Correct the handling of %s precision.
15629 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15632 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15634 (ix86_finalize_stack_frame_flags): This. Also clear
15635 frame_pointer_needed if -fno-omit-frame-pointer is used without
15637 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15638 with ix86_finalize_stack_frame_flags.
15639 (ix86_expand_epilogue): Likewise.
15640 (ix86_expand_split_stack_prologue): Likewise.
15641 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15643 2017-08-10 Martin Liska <mliska@suse.cz>
15646 * c-attribs.c (handle_target_attribute):
15647 Report warning for an empty string argument of target attribute.
15649 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15652 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15654 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15655 or DECL_NONLOCAL labels.
15656 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15657 or DECL_NONLOCAL labels here.
15659 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15661 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15662 to indicate when early gimple folding has been disabled.
15663 (rs6000_gimple_fold_builtin): Add debug content.
15664 (rs6000_invalid_builtin): Fix whitespace.
15665 (rs6000_expand_builtin): Fix whitespace.
15666 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15668 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15671 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15672 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15674 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15676 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15677 (qdf24xx): Likewise.
15678 * config/aarch64/aarch64.md: Include falkor.md.
15679 * config/aarch64/falkor.md: New.
15681 2017-08-09 Marek Polacek <polacek@redhat.com>
15684 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15685 * diagnostic.c (emit_diagnostic): Add a comment.
15686 (emit_diagnostic_valist): New function.
15688 2017-08-09 Marek Polacek <polacek@redhat.com>
15691 * input.c (make_location): New overload.
15692 * input.h (make_location): Declare.
15694 2017-08-08 Alan Modra <amodra@gmail.com>
15695 H.J. Lu <hongjiu.lu@intel.com>
15698 * gcc.c (NO_PIE_SPEC): Delete.
15699 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15701 (LINK_PIE_SPEC): ..to here.
15702 (LINK_COMMAND_SPEC): Support -no-pie.
15703 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15704 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15706 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15707 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15708 (ENDFILE_CRTEND_SPEC): Similarly.
15710 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15713 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15714 (mstack-protector-guard-offset=): Ditto.
15715 * config/i386/i386.c (ix86_option_override): Handle
15716 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15718 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15719 ix86_stack_protect_guard_offset variables.
15720 (TARGET_STACK_PROTECT_GUARD): Always define.
15721 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15722 and -mstack-protector-guard-offset= options.
15724 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15726 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15727 boundary case for the last candidate.
15729 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15731 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15732 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15734 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15736 PR middle-end/19706
15737 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15738 * config/aarch64/aarch64-builtins.c
15739 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15740 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15741 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15743 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15744 Andrew Pinski <pinskia@gmail.com>
15746 PR middle-end/19706
15747 * internal-fn.def (XORSIGN): New.
15748 * optabs.def (xorsign_optab): New.
15749 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15750 (convert_expand_mult_copysign): New.
15751 (pass_optimize_widening_mul::execute): Call
15752 convert_expand_mult_copysign.
15754 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15756 PR tree-optimization/81354
15757 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15758 Insert on edges rather than explicitly creating landing pads.
15759 (analyze_candidates_and_replace): Commit edge inserts.
15761 2017-08-08 Richard Biener <rguenther@suse.de>
15763 PR middle-end/81719
15764 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15765 (expand_simple_operations): Also look through ADDR_EXPRs with
15766 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15768 2017-08-08 Richard Biener <rguenther@suse.de>
15770 PR tree-optimization/81723
15771 * tree-vect-slp.c (struct bst_traits): New hash traits.
15772 (bst_fail): New global.
15773 (vect_build_slp_tree_2): New worker, split out from ...
15774 (vect_build_slp_tree): ... this now wrapping it with using
15775 bst_fail set to cache SLP tree build fails. Properly handle
15777 (vect_analyze_slp_instance): Allocate and free bst_fail.
15779 2017-08-08 Martin Liska <mliska@suse.cz>
15782 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15783 LABEL_DECLs that can be from a different function.
15785 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15787 PR tree-optimization/81744
15788 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15789 loop's number of iterations.
15791 2017-08-08 Martin Liska <mliska@suse.cz>
15793 * asan.c: Include header files.
15794 * attribs.c (build_decl_attribute_variant): New function moved
15796 (build_type_attribute_qual_variant): Likewise.
15797 (cmp_attrib_identifiers): Likewise.
15798 (simple_cst_list_equal): Likewise.
15799 (omp_declare_simd_clauses_equal): Likewise.
15800 (attribute_value_equal): Likewise.
15801 (comp_type_attributes): Likewise.
15802 (build_type_attribute_variant): Likewise.
15803 (lookup_ident_attribute): Likewise.
15804 (remove_attribute): Likewise.
15805 (merge_attributes): Likewise.
15806 (merge_type_attributes): Likewise.
15807 (merge_decl_attributes): Likewise.
15808 (merge_dllimport_decl_attributes): Likewise.
15809 (handle_dll_attribute): Likewise.
15810 (attribute_list_equal): Likewise.
15811 (attribute_list_contained): Likewise.
15812 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15813 (lookup_attribute_by_prefix): Likewise.
15814 * bb-reorder.c: Include header files.
15815 * builtins.c: Likewise.
15816 * calls.c: Likewise.
15817 * cfgexpand.c: Likewise.
15818 * cgraph.c: Likewise.
15819 * cgraphunit.c: Likewise.
15820 * convert.c: Likewise.
15821 * dwarf2out.c: Likewise.
15822 * final.c: Likewise.
15823 * fold-const.c: Likewise.
15824 * function.c: Likewise.
15825 * gimple-expr.c: Likewise.
15826 * gimple-fold.c: Likewise.
15827 * gimple-pretty-print.c: Likewise.
15828 * gimple.c: Likewise.
15829 * gimplify.c: Likewise.
15830 * hsa-common.c: Likewise.
15831 * hsa-gen.c: Likewise.
15832 * internal-fn.c: Likewise.
15833 * ipa-chkp.c: Likewise.
15834 * ipa-cp.c: Likewise.
15835 * ipa-devirt.c: Likewise.
15836 * ipa-fnsummary.c: Likewise.
15837 * ipa-inline.c: Likewise.
15838 * ipa-visibility.c: Likewise.
15840 * lto-cgraph.c: Likewise.
15841 * omp-expand.c: Likewise.
15842 * omp-general.c: Likewise.
15843 * omp-low.c: Likewise.
15844 * omp-offload.c: Likewise.
15845 * omp-simd-clone.c: Likewise.
15846 * opts-global.c: Likewise.
15847 * passes.c: Likewise.
15848 * predict.c: Likewise.
15849 * sancov.c: Likewise.
15850 * sanopt.c: Likewise.
15851 * symtab.c: Likewise.
15852 * toplev.c: Likewise.
15853 * trans-mem.c: Likewise.
15854 * tree-chkp.c: Likewise.
15855 * tree-eh.c: Likewise.
15856 * tree-into-ssa.c: Likewise.
15857 * tree-object-size.c: Likewise.
15858 * tree-parloops.c: Likewise.
15859 * tree-profile.c: Likewise.
15860 * tree-ssa-ccp.c: Likewise.
15861 * tree-ssa-live.c: Likewise.
15862 * tree-ssa-loop.c: Likewise.
15863 * tree-ssa-sccvn.c: Likewise.
15864 * tree-ssa-structalias.c: Likewise.
15865 * tree-ssa.c: Likewise.
15866 * tree-streamer-in.c: Likewise.
15867 * tree-vectorizer.c: Likewise.
15868 * tree-vrp.c: Likewise.
15869 * tsan.c: Likewise.
15870 * ubsan.c: Likewise.
15871 * varasm.c: Likewise.
15872 * varpool.c: Likewise.
15873 * tree.c: Remove functions moved to attribs.[ch].
15874 * tree.h: Likewise.
15875 * config/aarch64/aarch64.c: Add attrs.h header file.
15876 * config/alpha/alpha.c: Likewise.
15877 * config/arc/arc.c: Likewise.
15878 * config/arm/arm.c: Likewise.
15879 * config/avr/avr.c: Likewise.
15880 * config/bfin/bfin.c: Likewise.
15881 * config/c6x/c6x.c: Likewise.
15882 * config/cr16/cr16.c: Likewise.
15883 * config/cris/cris.c: Likewise.
15884 * config/darwin.c: Likewise.
15885 * config/epiphany/epiphany.c: Likewise.
15886 * config/fr30/fr30.c: Likewise.
15887 * config/frv/frv.c: Likewise.
15888 * config/ft32/ft32.c: Likewise.
15889 * config/h8300/h8300.c: Likewise.
15890 * config/i386/winnt.c: Likewise.
15891 * config/ia64/ia64.c: Likewise.
15892 * config/iq2000/iq2000.c: Likewise.
15893 * config/lm32/lm32.c: Likewise.
15894 * config/m32c/m32c.c: Likewise.
15895 * config/m32r/m32r.c: Likewise.
15896 * config/m68k/m68k.c: Likewise.
15897 * config/mcore/mcore.c: Likewise.
15898 * config/microblaze/microblaze.c: Likewise.
15899 * config/mips/mips.c: Likewise.
15900 * config/mmix/mmix.c: Likewise.
15901 * config/mn10300/mn10300.c: Likewise.
15902 * config/moxie/moxie.c: Likewise.
15903 * config/msp430/msp430.c: Likewise.
15904 * config/nds32/nds32-isr.c: Likewise.
15905 * config/nds32/nds32.c: Likewise.
15906 * config/nios2/nios2.c: Likewise.
15907 * config/nvptx/nvptx.c: Likewise.
15908 * config/pa/pa.c: Likewise.
15909 * config/pdp11/pdp11.c: Likewise.
15910 * config/powerpcspe/powerpcspe.c: Likewise.
15911 * config/riscv/riscv.c: Likewise.
15912 * config/rl78/rl78.c: Likewise.
15913 * config/rx/rx.c: Likewise.
15914 * config/s390/s390.c: Likewise.
15915 * config/sh/sh.c: Likewise.
15916 * config/sol2.c: Likewise.
15917 * config/sparc/sparc.c: Likewise.
15918 * config/spu/spu.c: Likewise.
15919 * config/stormy16/stormy16.c: Likewise.
15920 * config/tilegx/tilegx.c: Likewise.
15921 * config/tilepro/tilepro.c: Likewise.
15922 * config/v850/v850.c: Likewise.
15923 * config/vax/vax.c: Likewise.
15924 * config/visium/visium.c: Likewise.
15925 * config/xtensa/xtensa.c: Likewise.
15927 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15930 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15931 constraints since the -mupper-regs-* switches have been
15933 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15934 into a vector from a double word element that was extracted from
15935 another vector, and eliminate extra XXPERMDI instructions.
15936 (vsx_concat_<mode>_2): Likewise.
15937 (vsx_concat_<mode>_3): Likewise.
15938 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15939 concat to allow optimizing inserts from previous extracts.
15941 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15943 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15944 memory reference to a SSP offset in TLS address space.
15945 (ix86_print_operand) <case '@'>: Remove.
15946 (ix86_print_operand_punct_valid_p): Remove '@' code.
15947 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15948 UNSPEC_SP_TLS_TEST.
15949 (stack_tls_protect_set_<mode>): Remove.
15950 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15951 (stack_tls_protect_test_<mode>): Remove.
15952 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15954 2017-08-07 Olivier Hainque <hainque@adacore.com>
15957 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15959 2017-08-07 Douglas Rupp <rupp@adacore.com>
15961 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15962 variable was referenced as multidir in command.
15964 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15967 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15970 2017-08-07 Martin Liska <mliska@suse.cz>
15972 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15973 * config/rl78/rl78.c: Add include of attribs.h.
15974 * config/sh/sh.c: Likewise.
15975 * config/v850/v850.c: Likewise.
15977 2017-08-07 Tom de Vries <tom@codesourcery.com>
15979 PR middle-end/78266
15980 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15982 2017-08-07 Martin Liska <mliska@suse.cz>
15984 * config/mips/mips.c: Include attribs.h.
15986 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15989 * doc/invoke.texi: Document change in behvaior for -Ofast for
15992 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15994 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15996 (aarch64_pop_regs): Likewise.
15997 (aarch64_gen_load_pair): Likewise.
15998 (aarch64_save_callee_saves): Likewise.
15999 (aarch64_restore_callee_saves): Likewise.
16001 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16003 * config/i386/i386.c: Revert the last change.
16005 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16008 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16010 (ix86_finalize_stack_frame_flags): This. Also clear
16011 frame_pointer_needed if -fno-omit-frame-pointer is used without
16013 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16014 with ix86_finalize_stack_frame_flags.
16015 (ix86_expand_epilogue): Likewise.
16016 (ix86_expand_split_stack_prologue): Likewise.
16018 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16021 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16022 to P_AES for Westmere.
16024 2017-08-07 Jonathan Yong <10walls@gmail.com>
16026 * config/i386/mingw.opt (fset-stack-executable): Removed.
16027 * config/i386/cygming.opt (fset-stack-executable): Moved
16029 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16031 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16033 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16035 2017-08-07 Marek Polacek <polacek@redhat.com>
16037 PR middle-end/81737
16038 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16040 2017-08-07 Martin Liska <mliska@suse.cz>
16042 * attribs.h (canonicalize_attr_name): New function.
16043 (cmp_attribs): Move from c-format.c and adjusted.
16044 (is_attribute_p): Moved from tree.h.
16045 * tree-inline.c: Add new includes.
16046 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16047 (private_is_attribute_p): Remove.
16048 (private_lookup_attribute): Likewise.
16049 (private_lookup_attribute_by_prefix): Simplify.
16050 (remove_attribute): Use is_attribute_p.
16051 * tree.h: Remove removed declarations.
16053 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16055 PR middle-end/81698
16056 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16057 instead of computing it in the function. Formatting fix.
16058 (expand_case): Don't rely on default_edge being the first edge,
16059 clear it if removing it, pass default_edge to
16060 emit_case_dispatch_table.
16061 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16064 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16066 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16067 insn in the function, emit NOP after the insn.
16069 2017-08-06 Tom de Vries <tom@codesourcery.com>
16071 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16074 2017-08-06 Tom de Vries <tom@codesourcery.com>
16076 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16079 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16081 PR tree-optimization/57371
16082 * match.pd: New pattern.
16084 2017-08-04 Marek Polacek <polacek@redhat.com>
16086 PR middle-end/81695
16087 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16088 perform the computation in offset_int.
16090 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16092 PR tree-optimization/81136
16093 * tree-vectorizer.h: Include tree-hash-traits.h.
16094 (vec_base_alignments): New typedef.
16095 (vec_info): Add a base_alignments field.
16096 (vect_record_base_alignments): Declare.
16097 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16099 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16100 (create_data_ref): Add an is_conditional_in_stmt argument.
16101 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16102 the is_conditional_in_stmt field.
16103 (data_ref_loc): Add an is_conditional_in_stmt field.
16104 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16105 (find_data_references_in_stmt): Update call to create_data_ref.
16106 (graphite_find_data_references_in_stmt): Likewise.
16107 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16108 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16109 (vect_record_base_alignment): New function.
16110 (vect_record_base_alignments): Likewise.
16111 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16112 for nested statements even if we fail to compute a misalignment.
16113 Use pooled base alignments for unconditional references.
16114 (vect_find_same_alignment_drs): Compare base addresses instead
16116 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16117 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16119 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16121 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16122 Add an explicit name for the enum. Use auto_vec for slp_instances
16123 and grouped_stores.
16124 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16126 (_bb_vec_info): Add a constructor and destructor.
16127 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16128 (destroy_loop_vec_info): Delete.
16129 (vect_destroy_datarefs): Likewise.
16130 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16131 (vec_info::vec_info): New function.
16132 (vec_info::~vec_info): Likewise.
16133 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16134 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
16135 destroy_loop_vec_info.
16136 * tree-vect-loop.c (new_loop_vec_info): Replace with...
16137 (_loop_vec_info::_loop_vec_info): ...this.
16138 (destroy_loop_vec_info): Replace with...
16139 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
16140 the stmt_vec_infos. Leave handling of vec_info information to its
16141 destructor. Remove explicit vector releases.
16142 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
16143 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
16144 * tree-vect-slp.c (new_bb_vec_info): Replace with...
16145 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
16146 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
16147 (destroy_bb_vec_info): Replace with...
16148 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
16149 information to its destructor.
16150 (vect_slp_analyze_bb_1): Use new and delete instead of
16151 new_bb_vec_info and destroy_bb_vec_info.
16152 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
16155 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16157 * tree-data-ref.h (subscript): Add access_fn field.
16158 (data_dependence_relation): Add could_be_independent_p.
16159 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
16160 (same_access_functions): Move to tree-data-ref.c.
16161 * tree-data-ref.c (ref_contains_union_access_p): New function.
16162 (access_fn_component_p): Likewise.
16163 (access_fn_components_comparable_p): Likewise.
16164 (dr_analyze_indices): Add a reference to access_fn_component_p.
16165 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
16167 (constant_access_functions): Likewise.
16168 (add_other_self_distances): Likewise.
16169 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
16170 (initialize_data_dependence_relation): Use XCNEW and remove
16171 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
16172 of access functions that have the same type. Allow the
16173 subsequence to end with different bases in some circumstances.
16174 Record the chosen access functions in SUB_ACCESS_FN.
16175 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
16176 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
16177 (subscript_dependence_tester_1): Likewise dra and drb.
16178 (build_classic_dist_vector): Update calls accordingly.
16179 (subscript_dependence_tester): Likewise.
16180 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
16181 DDR_COULD_BE_INDEPENDENT_P.
16182 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
16183 comp_alias_ddrs instead of may_alias_ddrs.
16184 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16186 (vect_analyze_data_ref_dependence): Use it if
16187 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
16188 distance vectors if that fails.
16189 (dependence_distance_ge_vf): New function.
16190 (vect_prune_runtime_alias_test_list): Use it. Don't clear
16191 LOOP_VINFO_MAY_ALIAS_DDRS.
16193 2017-08-04 Richard Biener <rguenther@suse.de>
16195 PR middle-end/81705
16196 * fold-const.c (fold_binary_loc): Properly restrict
16197 minus_var0 && minus_var1 case when associating undefined overflow
16200 2017-08-04 Tom de Vries <tom@codesourcery.com>
16202 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
16204 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16206 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16207 Don't start diagnostic messages with a capital letter.
16208 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16210 (rs6000_invalid_builtin): Likewise.
16211 (rs6000_trampoline_init): Likewise.
16213 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16216 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
16217 after setting changeable df flags.
16219 2017-08-03 Richard Biener <rguenther@suse.de>
16221 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
16222 up if the use is in USE - X.
16224 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
16226 * toplev.c (dumpfile.h): New include.
16227 (internal_error_reentered): New static function. Use it...
16228 (internal_error_function): ...here to handle reentered internal_error.
16230 2017-08-03 Richard Biener <rguenther@suse.de>
16232 PR middle-end/81148
16233 * fold-const.c (split_tree): Add minus_var and minus_con
16234 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16235 here but always use minus_*.
16236 (associate_trees): Assert we never associate with MINUS_EXPR
16237 and NULL first operand. Do not recurse for PLUS_EXPR operands
16238 when associating as MINUS_EXPR either.
16239 (fold_binary_loc): Track minus_var and minus_con.
16241 2017-08-03 Tom de Vries <tom@codesourcery.com>
16244 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16245 ACCEL_COMPILER, apply finish_options on
16246 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16248 2017-08-03 Tom de Vries <tom@codesourcery.com>
16251 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16252 function_entry_patch_area_size > 0.
16254 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16257 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16258 instead of 10??LU, perform unit multiplication in wide_int,
16259 don't change alloc_object_size_limit if the limit is larger
16262 PR tree-optimization/81655
16263 PR tree-optimization/81588
16264 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16265 the case when ranges[i].low and high are 1 for unsigned type with
16268 PR middle-end/81052
16269 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16270 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16272 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16274 * tree-vrp.h: Add include guard.
16276 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16279 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16280 (ud2): New insn pattern.
16281 * config/i386/i386.c (ix86_expand_epilogue):
16282 For naked functions, generate ud2 instead of trap insn.
16284 2017-08-02 Marek Polacek <polacek@redhat.com>
16287 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16289 2017-08-02 Tom de Vries <tom@codesourcery.com>
16290 Cesar Philippidis <cesar@codesourcery.com>
16292 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16293 Add missing edge probabilities.
16295 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16297 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16298 Correct endianness.
16300 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16302 PR middle-end/79499
16303 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16304 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16305 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16306 prologue_seq sequences - if any.
16308 2017-08-02 Richard Biener <rguenther@suse.de>
16310 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16311 via vectors if supported, integer extracts via punning if supported
16312 or otherwise vector extracts.
16314 2017-08-02 Richard Biener <rguenther@suse.de>
16316 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16318 (bitmap_insert_into_set): ... this.
16320 2017-08-02 Richard Biener <rguenther@suse.de>
16322 PR tree-optimization/81633
16324 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16326 PR tree-optimization/71752
16327 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16329 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16331 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16332 (machine_function::call_ms2sysv_pad_out): Remove field.
16333 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16334 (ix86_compute_frame_layout): Likewise.
16336 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16339 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16340 attribute with interrupt attribute.
16342 2017-08-01 Andrew Pinski <apinski@cavium.com>
16344 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16345 BIT_INSERT_EXPR's operand 1
16346 to see if the types precision matches.
16348 2017-08-01 Martin Liska <mliska@suse.cz>
16350 PR middle-end/70140
16351 * builtins.c (expand_builtin_memcpy_args): Remove.
16352 (expand_builtin_memcpy): Call newly added function
16353 expand_builtin_memory_copy_args.
16354 (expand_builtin_memcpy_with_bounds): Likewise.
16355 (expand_builtin_mempcpy): Remove last argument.
16356 (expand_builtin_mempcpy_with_bounds): Likewise.
16357 (expand_builtin_memory_copy_args): New function created from
16358 expand_builtin_mempcpy_args with small modifications.
16359 (expand_builtin_mempcpy_args): Remove.
16360 (expand_builtin_stpcpy): Remove unused argument.
16361 (expand_builtin): Likewise.
16362 (expand_builtin_with_bounds): Likewise.
16364 2017-08-01 Martin Liska <mliska@suse.cz>
16367 Make mempcpy more optimal (PR middle-end/70140).
16369 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16372 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16373 __builtin_vec_cmpne verify both arguments are compatible vectors
16374 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16375 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16376 move computation of aligned to after checking the argument types.
16380 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16381 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16384 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16385 Doug Rupp <rupp@adacore.com>
16386 Olivier Hainque <hainque@adacore.com>
16388 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16389 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16391 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16392 for TARGET_OS_CPP_BUILTIN.
16393 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16394 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16396 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16397 passing required abi options to the assembler for EABI configurations.
16398 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16399 of .text.hot and .text.unlikely sections for kernel modules when
16400 using ARM style exceptions.
16401 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16402 options. Add EXTRA_CC1_SPEC.
16403 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16405 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16407 (ARM_TARGET2_DWARF_FORMAT): Define.
16408 * config/arm/t-vxworks: Adjust multilib control to removal of the
16409 Diab command line options.
16411 2017-08-01 Martin Liska <mliska@suse.cz>
16413 PR gcov-profile/81561
16414 * gcov.c (unblock): Make unblocking safe as we need to preserve
16415 index correspondence of blocks and block_lists.
16417 2017-08-01 Richard Biener <rguenther@suse.de>
16419 PR tree-optimization/81181
16420 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16421 (compute_antic): ... end of iteration here.
16423 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16425 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16426 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16427 (ftree-slp-vectorize): Likewise.
16428 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16429 can no longer be set independent of flag_tree_loop_vectorize.
16430 * omp-general.c (emp_max_vf): Likewise.
16431 * opts.c (enable_fdo_optimizations): Remove references to
16432 flag_tree_vectorize, these are now implicit.
16433 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16434 and leave it for the options machinery.
16436 2017-08-01 Martin Liska <mliska@suse.cz>
16438 PR middle-end/70140
16439 * builtins.c (expand_builtin_memcpy_args): Remove.
16440 (expand_builtin_memcpy): Call newly added function
16441 expand_builtin_memory_copy_args.
16442 (expand_builtin_memcpy_with_bounds): Likewise.
16443 (expand_builtin_mempcpy): Remove last argument.
16444 (expand_builtin_mempcpy_with_bounds): Likewise.
16445 (expand_builtin_memory_copy_args): New function created from
16446 expand_builtin_mempcpy_args with small modifications.
16447 (expand_builtin_mempcpy_args): Remove.
16448 (expand_builtin_stpcpy): Remove unused argument.
16449 (expand_builtin): Likewise.
16450 (expand_builtin_with_bounds): Likewise.
16452 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16455 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
16456 print "ds:" only for immediates in generic address space.
16458 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16461 * config/i386/i386.c (ix86_funciton_naked): New prototype.
16462 (ix86_function_ok_for_sibcall): Return false for naked functions.
16464 2017-08-01 Richard Biener <rguenther@suse.de>
16466 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
16467 (compute_antic): Seed worklist with exit block predecessors.
16468 * cfganal.c (dfs_find_deadend): For a cycle return the source
16469 of the edge closing it.
16471 2017-08-01 Tamar Christina <tamar.christina@arm.com>
16473 * config/aarch64/aarch64.c
16474 (aarch64_can_const_movi_rtx_p): Move 0 check.
16476 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16478 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
16479 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
16481 * match.pd: Ditto in address comparison pattern.
16483 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16485 PR tree-optimization/81627
16486 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
16487 closed ssa form for store-store chain.
16489 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16491 PR tree-optimization/81620
16492 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
16493 for store-store chain.
16495 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16497 PR tree-optimization/81588
16498 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
16499 ranges[i].in_p, invert comparison code ccode. For >/>=,
16500 swap rhs1 and rhs2 and comparison code unconditionally,
16501 for </<= don't do that. Don't swap rhs1/rhs2 again if
16502 ranges[i].in_p, instead invert comparison code ccode if
16503 opcode or oe->rank is BIT_IOR_EXPR.
16506 * optabs.def (vec_extract_optab, vec_init_optab): Change from
16507 a direct optab to conversion optab.
16508 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
16509 with GET_MODE_INNER as last argument instead of optab_handler.
16510 * expmed.c (extract_bit_field_1): Likewise. Use vector from
16511 vector extraction if possible and optab is available.
16512 * expr.c (store_constructor): Use convert_optab_handler instead
16513 of optab_handler. Use vector initialization from smaller
16514 vectors if possible and optab is available.
16515 * tree-vect-stmts.c (vectorizable_load): Likewise.
16516 * doc/md.texi (vec_extract, vec_init): Document that the optabs
16517 now have two modes.
16518 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
16519 of vec_init from half-sized vectors with the same element mode.
16520 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
16521 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
16522 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
16523 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
16524 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
16525 after mode in gen_vec_extract* calls.
16526 (vec_extract<mode>): Renamed to ...
16527 (vec_extract<mode><ssescalarmodelower>): ... this.
16528 (vec_extract<mode><ssehalfvecmodelower>): New expander.
16529 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
16530 element mode after mode in gen_vec_init* calls.
16531 (VEC_INIT_HALF_MODE): New mode iterator.
16532 (vec_init<mode>): Renamed to ...
16533 (vec_init<mode><ssescalarmodelower>): ... this.
16534 (vec_init<mode><ssehalfvecmodelower>): New expander.
16535 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
16536 (vec_extractv2sfsf): ... this.
16537 (vec_initv2sf): Renamed to ...
16538 (vec_initv2sfsf): ... this.
16539 (vec_extractv2si): Renamed to ...
16540 (vec_extractv2sisi): ... this.
16541 (vec_initv2si): Renamed to ...
16542 (vec_initv2sisi): ... this.
16543 (vec_extractv4hi): Renamed to ...
16544 (vec_extractv4hihi): ... this.
16545 (vec_initv4hi): Renamed to ...
16546 (vec_initv4hihi): ... this.
16547 (vec_extractv8qi): Renamed to ...
16548 (vec_extractv8qiqi): ... this.
16549 (vec_initv8qi): Renamed to ...
16550 (vec_initv8qiqi): ... this.
16551 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
16552 (vec_init<mode>): Renamed to ...
16553 (vec_init<mode><VEC_base_l>): ... this.
16554 (vec_extract<mode>): Renamed to ...
16555 (vec_extract<mode><VEC_base_l>): ... this.
16556 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
16557 (vec_initv2sfsf): ... this.
16558 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
16559 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16560 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
16561 element mode after mode in gen_vec_init* calls.
16562 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
16563 (vec_init<mode><Vel>): ... this.
16564 (vec_extract<mode>): Renamed to ...
16565 (vec_extract<mode><Vel>): ... this.
16566 * config/aarch64/iterators.md (Vel): New mode attribute.
16567 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
16568 Add element mode after mode in gen_vec_extract* calls.
16569 * config/s390/vector.md (non_vec_l): New mode attribute.
16570 (vec_extract<mode>): Renamed to ...
16571 (vec_extract<mode><non_vec_l>): ... this.
16572 (vec_init<mode>): Renamed to ...
16573 (vec_init<mode><non_vec_l>): ... this.
16574 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16575 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16577 * config/arm/iterators.md (V_elem_l): New mode attribute.
16578 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16579 (vec_extract<mode><V_elem_l>): ... this.
16580 (vec_extractv2di): Renamed to ...
16581 (vec_extractv2didi): ... this.
16582 (vec_init<mode>): Renamed to ...
16583 (vec_init<mode><V_elem_l>): ... this.
16584 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16585 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16586 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16587 Add element mode after gen_vec_extract* calls.
16588 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16589 (vec_init<mode><unitmode>): ... this.
16590 (vec_extract<mode>): Renamed to ...
16591 (vec_extract<mode><unitmode>): ... this.
16592 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16593 (vec_init<mode><unitmode>): ... this.
16594 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16595 (vec_initv2sfsf): ... this.
16596 (vec_extractv2sf): Renamed to ...
16597 (vec_extractv2sfsf): ... this.
16598 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16599 Add element mode after gen_vec_extract* calls.
16600 * config/mips/mips.md (unitmode): New mode iterator.
16601 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16602 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16603 * config/spu/spu.md (inner_l): New mode attribute.
16604 (vec_init<mode>): Renamed to ...
16605 (vec_init<mode><inner_l>): ... this.
16606 (vec_extract<mode>): Renamed to ...
16607 (vec_extract<mode><inner_l>): ... this.
16608 * config/sparc/sparc.md (veltmode): New mode iterator.
16609 (vec_init<VMALL:mode>): Renamed to ...
16610 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16611 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16612 (vec_initv2sisi): ... this.
16613 (vec_initv2sf): Renamed to ...
16614 (vec_initv2sfsf): ... this.
16615 (vec_extractv2sf): Renamed to ...
16616 (vec_extractv2sfsf): ... this.
16617 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16618 (vec_init<mode>): Renamed to ...
16619 (vec_init<mode><VEC_base_l>): ... this.
16620 (vec_extract<mode>): Renamed to ...
16621 (vec_extract<mode><VEC_base_l>): ... this.
16622 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16623 (vec_initv2sfsf): ... this.
16624 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16625 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16626 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16627 gen_vec_init* calls.
16629 2017-08-01 Richard Biener <rguenther@suse.de>
16631 PR tree-optimization/81297
16632 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16633 TREE_OVERFLOW from INTEGER_CSTs.
16635 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16637 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16639 2017-07-31 Carl Love <cel@us.ibm.com>
16641 * config/rs6000/rs6000-c: Add support for built-in functions
16642 vector signed char vec_xl_be (signed long long, signed char *);
16643 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16644 vector signed int vec_xl_be (signed long long, signed int *);
16645 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16646 vector signed long long vec_xl_be (signed long long, signed long long *);
16647 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16648 vector signed short vec_xl_be (signed long long, signed short *);
16649 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16650 vector double vec_xl_be (signed long long, double *);
16651 vector float vec_xl_be (signed long long, float *);
16652 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16653 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16654 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16656 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16657 (altivec_expand_builtin): Add switch statement to call
16658 altivec_expand_xl_be for each builtin.
16659 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16660 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16661 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16662 __builtin_vsx_le_be_v16qi.
16663 * doc/extend.texi: Update the built-in documentation file for the
16664 new built-in functions.
16666 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16669 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16671 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16673 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16675 * config.gcc: Add z14.
16676 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16677 CPU model numbers for z13s and z14.
16678 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16680 * config/s390/s390-opts.h (enum processor_type): Rename
16681 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16682 * config/s390/s390.c (processor_table): Add field for CPU name to
16683 be passed to Binutils.
16684 (s390_asm_output_machine_for_arch): Use the new field in
16685 processor_table for Binutils.
16686 (s390_expand_builtin): Replace arch12 with z14.
16687 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16688 (s390_get_sched_attrmask): Likewise.
16689 (s390_get_unit_mask): Likewise.
16690 * config/s390/s390.opt: Add z14 to processor_type enum.
16692 2017-07-31 Martin Jambor <mjambor@suse.cz>
16695 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16696 regardless of optimization level.
16698 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16699 Martin Liska <mliska@suse.cz>
16701 * predict.def: Remove old comment and adjust probability.
16702 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16703 PREDICT statements.
16705 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16708 * config/i386/i386.c (ix86_function_naked): New function.
16709 (ix86_can_use_return_insn_p): Return false for naked functions.
16710 (ix86_expand_prologue): Skip prologue for naked functions.
16711 (ix86_expand_epilogue): Skip epilogue for naked functions
16712 and emit trap instruction.
16713 (ix86_warn_func_return): New function.
16714 (ix86_attribute_table): Add "naked" attribute specification.
16715 (TARGET_WARN_FUNC_RETURN): Define.
16716 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16718 2017-07-31 Martin Liska <mliska@suse.cz>
16720 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16721 (dump_gimple_bb_header): Always dump BB info.
16722 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16724 2017-07-31 Martin Liska <mliska@suse.cz>
16727 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16728 also with current_function_decl non-null equality.
16730 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16733 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16734 change type to the element type, instead add eltype variable and
16735 use it where we are interested in the element type.
16737 PR tree-optimization/81603
16738 * ipa-polymorphic-call.c
16739 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16740 offset arithmetic in offset_int, bail out if the resulting bit offset
16741 doesn't fit into shwi.
16743 2017-07-31 Martin Liska <mliska@suse.cz>
16745 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16746 (gimplify_save_expr): Fix comment.
16748 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16751 * config/i386/i386.c (ix86_function_arg): Update arguments for
16753 (ix86_compute_frame_layout): Set the initial stack offset to
16754 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16755 INCOMING_FRAME_SP_OFFSET.
16756 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16757 stack before exception handler returns.
16758 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16759 the 'ERROR_CODE' for exception handler.
16761 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16763 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16764 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16765 (ASM_OUTPUT_REG_POP): Ditto.
16766 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16767 instead of asm_fprintf to output pure string.
16769 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16771 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16772 to imported_module_or_decl hook.
16773 (debug_nothing_tree_tree_tree_bool): Remove.
16774 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16775 * debug.c (do_nothing_debug_hooks): Use
16776 debug_nothing_tree_tree_tree_bool_bool instead of
16777 debug_nothing_tree_tree_tree_bool.
16778 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16779 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16780 * sdbout.c (sdb_debug_hooks): Likewise.
16781 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16782 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16784 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16785 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16786 attribute, don't add anything.
16788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16790 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16791 (fold_build2_stat_loc): Likewise.
16792 (fold_build3_stat_loc): Likewise.
16793 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16794 (fold_build1_loc): Remove macro.
16795 (fold_build2_loc): Likewise.
16796 (fold_build3_loc): Likewise.
16798 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16800 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16801 (gimple_build_debug_bind_source_stat): Likewise.
16802 * gimple.h (gimple_build_debug_bind): Remove macro.
16803 (gimple_build_debug_bind_source): Likewise.
16805 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16807 * bitmap.c (bitmap_alloc): Adjust.
16808 (bitmap_gc_alloc): Likewise.
16809 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16811 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16813 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16814 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16815 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16816 (bitmap_gc_alloc_stat): Likewise.
16817 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16819 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16821 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16822 * rtl.h (shallow_copy_rtx): Remove macro.
16824 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16826 * emit-rtl.c (gen_raw_REG): Adjust.
16827 * gengenrtl.c (gendef): Likewise.
16828 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16829 * rtl.h (rtx_alloc): Remove macro.
16831 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16833 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16834 (build_tree_list_stat): Likewise.
16835 * tree.h (build_tree_list): Remove macro.
16836 (build_tree_list_vec): Likewise.
16838 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16840 * tree.c (make_vector_stat): Remove _stat from name.
16841 (build_vector_stat): Likewise.
16842 * tree.h (make_vector_stat): Remove macro.
16843 (build_vector_stat): Likewise.
16845 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16847 * tree.h (build_var_debug_value): Remove prototype.
16849 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16851 * tree.c (tree_cons_stat): Remove _stat from name.
16852 * tree.h (tree_cons): Remove macro.
16854 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16856 * tree.c (build_vl_exp_stat): Remove _stat from name.
16857 * tree.h (build_vl_exp): Remove macro.
16859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16861 * tree.c (build_decl_stat): Remove _stat from name.
16862 * tree.h (build_decl): Remove macro.
16864 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16866 * gimple.c (gimple_build_with_ops_stat): Adjust.
16867 (gimple_alloc_stat): Remove _stat from name.
16868 * gimple.h (gimple_alloc): Remove macro.
16870 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16872 * tree.c (make_tree_vec_stat): Remove _stat from name.
16873 (grow_tree_vec_stat): Likewise.
16874 * tree.h (make_tree_vec_stat): Adjust prototype.
16875 (grow_tree_vec_stat): Likewise.
16876 (make_tree_vec): Remove macro.
16877 (grow_tree_vec): Likewise.
16879 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16881 * fold-const.c (fold_build1_stat_loc): Adjust.
16882 (fold_build2_stat_loc): Likewise.
16883 (fold_build3_stat_loc): Likewise.
16884 * tree.c (build0_stat): Remove _stat from name.
16885 (build1_stat): Likewise.
16886 (build2_stat): Likewise.
16887 (build3_stat): Likewise.
16888 (build4_stat): Likewise.
16889 (build5_stat): Likewise.
16890 * tree.h (build1_loc): Remove macro, and rename _stat function
16892 (build2_loc): Likewise.
16893 (build3_loc): Likewise.
16894 (build4_loc): Likewise.
16895 (build5_loc): Likewise.
16897 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16899 * tree.c (make_int_cst_stat): Remove _stat from name.
16900 * tree.h (make_int_cst_stat): Adjust prototype.
16901 (make_int_cst): Remove macro.
16903 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16905 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16906 * tree.h (make_tree_binfo_stat): Adjust prototype.
16907 (make_tree_binfo): Remove.
16909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16911 * tree.c (copy_node_stat): Rename to copy_node.
16912 (build_distinct_type_copy): Adjust.
16913 * tree.h (copy_node_stat): Adjust prototype.
16914 (copy_node): Remove macro.
16916 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16918 * tree.c (make_node_stat): rename to make_node.
16919 (build_tree_list_stat): Adjust.
16920 (build0_stat): Likewise.
16921 (build2_stat): Likewise.
16922 (build3_stat): Likewise.
16923 (build4_stat): Likewise.
16924 (build5_stat): Likewise.
16925 (build_decl_stat): Likewise.
16926 * tree.h (make_node_stat): Adjust prototype.
16927 (make_node): remove macro.
16929 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16931 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16932 (PPC_FEATURE2_SCV): Likewise.
16933 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16935 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16937 * config/aarch64/aarch64.c
16938 (aarch64_internal_mov_immediate): Add new special pattern.
16939 * config/aarch64/aarch64.md (*movdi_aarch64):
16940 Add reg/32bit const mov case.
16942 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16943 Richard Sandiford <richard.sandiford@linaro.org>
16945 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16946 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16947 Add integer and movi cases.
16948 (movi-split-hf-df-sf split, fp16): New.
16949 (enabled): Added TARGET_FP_F16INST.
16950 * config/aarch64/iterators.md (GPF_HF): New.
16951 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16953 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16955 * config/aarch64/aarch64.c
16956 (aarch64_simd_container_mode): Add prototype.
16957 (aarch64_expand_mov_immediate): Add HI support.
16958 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16959 (aarch64_can_const_movi_rtx_p): New.
16960 (aarch64_preferred_reload_class):
16961 Remove restrictions of using FP registers for certain SIMD operations.
16962 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16963 (aarch64_valid_floating_const): Add integer move validation.
16964 (aarch64_simd_imm_scalar_p): Remove.
16965 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16966 (aarch64_legitimate_constant_p): Expand list of supported cases.
16967 * config/aarch64/aarch64-protos.h
16968 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16969 (aarch64_reinterpret_float_as_int): New.
16970 (aarch64_simd_imm_scalar_p): Remove.
16971 * config/aarch64/constraints.md (Uvi): New.
16972 (Dd): Split into Ds and new Dd.
16973 * config/aarch64/aarch64.md (*movsi_aarch64):
16975 (*movdi_aarch64): Add SIMD mov case.
16977 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16979 * tree-predcom.c: (struct chain): Handle store-store chain in which
16980 stores for elimination only store loop invariant values.
16981 (execute_pred_commoning_chain): Ditto.
16982 (prepare_initializers_chain_store_elim): Ditto.
16983 (prepare_finalizers): Ditto.
16984 (is_inv_store_elimination_chain): New function.
16985 (initialize_root_vars_store_elim_1): New function.
16987 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16989 * tree-predcom.c: Revise general description of the pass.
16990 (enum chain_type): New enum type for store elimination.
16991 (struct chain): New field supporting store elimination.
16992 (struct component): Ditto.
16993 (dump_chain): Dump store-stores chain.
16994 (release_chain): Release resources.
16995 (split_data_refs_to_components): Compute and create component
16996 contains only stores for elimination.
16997 (get_chain_last_ref_at): New function.
16998 (make_invariant_chain): Initialization.
16999 (make_rooted_chain): Specify chain type in parameter and record it.
17000 (add_looparound_copies): Skip for store-stores chain.
17001 (determine_roots_comp): Compute type of chain and pass it to
17003 (initialize_root_vars_store_elim_2): New function.
17004 (finalize_eliminated_stores): New function.
17005 (remove_stmt): Handle store for elimination.
17006 (execute_pred_commoning_chain): Execute predictive commoning on
17007 store-store chains.
17008 (determine_unroll_factor): Skip unroll for store-stores chain.
17009 (prepare_initializers_chain_store_elim): New function.
17010 (prepare_initializers_chain): Hanlde store-store chain.
17011 (prepare_finalizers_chain, prepare_finalizers): New function.
17012 (tree_predictive_commoning_loop): Return integer value indicating
17013 if loop is unrolled or lcssa form is corrupted.
17014 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17016 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17018 * tree-predcom.c (initialize_root): Delete.
17019 (execute_pred_commoning_chain): Initialize root vars and replace
17020 reference of non-combined chain directly, rather than call above
17023 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17025 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17026 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17028 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17030 * tree-predcom.c (struct chain): New field init_seq.
17031 (release_chain): Release init_seq.
17032 (prepare_initializers_chain): Record intialization stmts in above
17034 (insert_init_seqs): New function.
17035 (tree_predictive_commoning_loop): Call insert_init_seqs.
17037 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17039 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17041 2017-07-28 Richard Biener <rguenther@suse.de>
17043 * match.pd: Remove superfluous :c.
17044 * genmatch.c (simplify::id): Add member.
17045 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17047 (current_id): New global.
17048 (dt_node::parent): Move from ...
17049 (dt_operand::parent): ... here. Add for_id member.
17050 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17051 (decision_tree::find_node): Relax order requirement when
17052 merging DT_TRUE nodes to ones inbetween the current simplify
17053 and the one we try to merge with. Add diagnostic whenever
17054 we need to enforce pattern order by not merging.
17055 (decision_tree::insert): Set current_id.
17056 (decision_tree::print_node): Dump parent node and for_id.
17057 (parser::last_id): Add member.
17058 (parser::push_simplify): Assign unique id.
17059 (parser::parser): Initialize last_id.
17061 2017-07-28 Martin Liska <mliska@suse.cz>
17064 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17065 gimple_build_debug_bind.
17067 2017-07-28 Richard Biener <rguenther@suse.de>
17069 PR tree-optimization/81502
17070 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17072 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17074 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17075 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17076 for BIT_FIELD_REF args.
17077 * fold-const.c (make_bit_field_ref): Likewise.
17078 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17080 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17083 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17084 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17085 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17086 Or it into SANITIZER_UNDEFINED.
17087 * ubsan.c: Include gimple-fold.h and varasm.h.
17088 (ubsan_expand_ptr_ifn): New function.
17089 (instrument_pointer_overflow): New function.
17090 (maybe_instrument_pointer_overflow): New function.
17091 (instrument_object_size): Formatting fix.
17092 (pass_ubsan::execute): Call instrument_pointer_overflow
17093 and maybe_instrument_pointer_overflow.
17094 * internal-fn.c (expand_UBSAN_PTR): New function.
17095 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17096 * sanitizer.def (__ubsan_handle_pointer_overflow,
17097 __ubsan_handle_pointer_overflow_abort): New builtins.
17098 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17099 * internal-fn.def (UBSAN_PTR): New internal function.
17100 * opts.c (sanitizer_opts): Add pointer-overflow.
17101 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17102 * fold-const.c (build_range_check): Compute pointer range check in
17103 integral type if pointer arithmetics would be needed. Formatting
17106 2017-07-28 Martin Liska <mliska@suse.cz>
17109 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17110 parameters that are of a variable-length.
17112 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17114 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17116 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17117 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17118 (CRT_CALL_STATIC_FUNCTION): Likewise.
17119 (ASM_DEFAULT_SPEC): New define.
17120 (ASM_SPEC32): Likewise.
17121 (ASM_SPEC64): Likewise.
17122 (ASM_SPEC_COMMON): Likewise.
17123 (ASM_SPEC): Likewise.
17124 (INVALID_64BIT): Likewise.
17125 (LINK_OS_DEFAULT_SPEC): Likewise.
17126 (LINK_OS_SPEC32): Likewise.
17127 (LINK_OS_SPEC64): Likewise.
17128 (POWERPC_LINUX): Likewise.
17129 (PTRDIFF_TYPE): Likewise.
17130 (RESTORE_FP_PREFIX): Likewise.
17131 (RESTORE_FP_SUFFIX): Likewise.
17132 (SAVE_FP_PREFIX): Likewise.
17133 (SAVE_FP_SUFFIX): Likewise.
17134 (SIZE_TYPE): Likewise.
17135 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
17136 (TARGET_64BIT): Likewise.
17137 (TARGET_64BIT): Likewise.
17138 (TARGET_AIX): Likewise.
17139 (WCHAR_TYPE_SIZE): Likewise.
17140 (WCHAR_TYPE): Undefine.
17141 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
17142 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
17143 (CPP_OS_RTEMS_SPEC): Delete.
17144 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
17145 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
17147 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
17149 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17151 PR tree-optimization/81578
17152 * tree-parloops.c (build_new_reduction): Bail out if
17153 reduction_code isn't one of the standard OpenMP reductions.
17154 Move the details printing after that decision.
17156 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17158 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
17159 related to reload_in_progress.
17160 (splat_input_operand): Likewise.
17161 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
17163 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
17165 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
17166 (TARGET_INSTANTIATE_DECLS): Likewise.
17167 (legitimate_indexed_address_p): Delete reload_in_progress code.
17168 (rs6000_debug_legitimate_address_p): Likewise.
17169 (rs6000_eliminate_indexed_memrefs): Likewise.
17170 (rs6000_emit_le_vsx_store): Likewise.
17171 (rs6000_emit_move_si_sf_subreg): Likewise.
17172 (rs6000_emit_move): Likewise.
17173 (register_to_reg_type): Likewise.
17174 (rs6000_pre_atomic_barrier): Likewise.
17175 (rs6000_machopic_legitimize_pic_address): Likewise.
17176 (rs6000_allocate_stack_temp): Likewise.
17177 (rs6000_address_for_fpconvert): Likewise.
17178 (rs6000_address_for_altivec): Likewise.
17179 (rs6000_secondary_memory_needed_rtx): Delete function.
17180 (rs6000_check_sdmode): Likewise.
17181 (rs6000_alloc_sdmode_stack_slot): Likewise.
17182 (rs6000_instantiate_decls): Likewise.
17183 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
17184 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
17185 Delete reload_in_progress.
17186 (*vec_reload_and_plus_<mptrsize>): Likewise.
17187 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
17188 (vsx_div_v2di): Likewise.
17189 (vsx_udiv_v2di): Likewise.
17191 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17193 * config/rs6000/rs6000.opt (mlra): Replace with stub.
17194 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
17195 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
17196 (rs6000_debug_reg_global): Delete print of LRA status.
17197 (rs6000_option_override_internal): Delete dead LRA related code.
17198 (rs6000_lra_p): Delete function.
17199 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
17201 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
17203 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
17204 * config/riscv/rtems.h: New file.
17206 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17207 Sudakshina Das <sudi.das@arm.com>
17209 * config/aarch64/aarch64.md
17210 (define_split for and<mode>3nr_compare): Move
17211 non aarch64_logical_operand to a register.
17212 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
17213 register immediate operand to a register.
17214 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
17216 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17218 PR middle-end/81564
17219 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
17221 2017-07-27 Richard Biener <rguenther@suse.de>
17223 PR tree-optimization/81573
17224 PR tree-optimization/81494
17225 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
17226 multi defuse cycle case.
17228 2017-07-27 Richard Biener <rguenther@suse.de>
17230 PR tree-optimization/81571
17231 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17234 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17236 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17237 earlier and only if MASK_FPU is set. Adjust formatting.
17239 2017-07-27 Martin Liska <mliska@suse.cz>
17241 * opt-functions.awk: Add validation of value of Init.
17242 * optc-gen.awk: Pass new argument.
17244 2017-07-27 Martin Liska <mliska@suse.cz>
17246 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17247 Fix wrong condition.
17249 2017-07-27 Martin Liska <mliska@suse.cz>
17251 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17252 BBs and edges seen by autoFDO.
17254 2017-07-27 Richard Biener <rguenther@suse.de>
17256 PR tree-optimization/81502
17257 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17258 with incompatible but same sized type.
17259 (execute_update_addresses_taken): Likewise.
17261 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17263 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17264 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17266 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17269 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17270 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17271 Change s_operand to memory_operand.
17273 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17275 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17276 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17277 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17278 Emit instructions rather than returning an expression. Handle TFmode
17279 and KFmode by casting to TImode.
17280 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17281 (rs6000_emit_le_vsx_store): Likewise.
17282 * config/rs6000/vsx.md (VSX_TI): New iterator.
17283 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17284 (*vsx_le_undo_permute_<mode>): Likewise.
17285 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17286 emit the split sequence.
17287 (*vsx_le_perm_store_<mode>): Likewise.
17289 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17291 PR tree-optimization/81555
17292 PR tree-optimization/81556
17293 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17294 if true, force CHANGED for the recursive invocation.
17295 (reassociate_bb): Remember original length of ops array, pass
17296 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17298 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17299 attributes for noipa attribute. For naked attribute use
17300 lookup_attribute first before lookup_attribute_spec.
17301 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17303 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17304 for functions with noipa attribute.
17305 (cgraph_externally_visible_p): Return true for functions with noipa
17307 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17308 for functions with noipa attribute.
17309 * doc/extend.texi: Document noipa function attribute.
17310 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17311 also for functions with noipa attribute.
17312 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17314 2017-07-26 Andrew Pinski <apinski@cavium.com>
17316 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17317 vec_unalign_load_cost and vec_unalign_store_cost.
17319 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17321 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17322 -mvsx-small-integer option.
17323 (ISA_3_0_MASKS_IEEE): Likewise.
17324 (OTHER_VSX_VECTOR_MASKS): Likewise.
17325 (POWERPC_MASKS): Likewise.
17326 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17327 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17328 code, only testing for DImode being allowed in non-VSX floating
17330 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17331 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17333 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17334 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17335 TARGET_P8_VECTOR test.
17336 (rs6000_secondary_reload_simple_move): Likewise.
17337 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17338 since TARGET_P9_VECTOR was already tested.
17339 (rs6000_opt_masks): Remove -mvsx-small-integer.
17340 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17341 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17343 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17344 test for TARGET_VEXTRACTUB was used, and that uses
17346 (p9 extract splitter): Likewise.
17347 (vsx_extract_<mode>_di_p9): Likewise.
17348 (vsx_extract_<mode>_store_p9): Likewise.
17349 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17350 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17351 the elimination of TARGET_VSX_SMALL_INTEGER.
17352 (vsx_extract_<mode>_p8): Likewise.
17353 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17354 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17355 (vsx_set_<mode>_p9): Likewise.
17356 (vsx_set_v4sf_p9): Likewise.
17357 (vsx_set_v4sf_p9_zero): Likewise.
17358 (vsx_insert_extract_v4sf_p9): Likewise.
17359 (vsx_insert_extract_v4sf_p9_2): Likewise.
17360 * config/rs6000/rs6000.md (sign extend splitter): Change
17361 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17362 (floatsi<mode>2_lfiwax_mem): Likewise.
17363 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17364 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17365 since a test for TARGET_P9_VECTOR was used.
17366 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17367 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17368 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17369 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17370 TARGET_P8_VECTOR test.
17371 (fix_trunc<mode>si2_stfiwx): Likewise.
17372 (fix_trunc<mode>si2_internal): Likewise.
17373 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17374 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17376 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17377 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17378 TARGET_P8_VECTOR test.
17379 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17380 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17381 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17383 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17384 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17385 since a test for TARGET_P9_VECTOR was used.
17386 (splitter for loading small constants): Likewise.
17388 2017-07-26 Andrew Pinski <apinski@cavium.com>
17390 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17393 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17396 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17397 (fp_valid_at): Likewise.
17399 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17401 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17402 (qdf24xx_addrcost_table): Likewise.
17403 (cortexa57_tunings): Update to use generic_branch_cost.
17404 (cortexa72_tunings): Likewise.
17405 (cortexa73_tunings): Likewise.
17406 (qdf24xx_tunings): Likewise.
17408 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17410 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17411 (thunderx2t99_branch_cost): Likewise.
17412 (cortexa35_tunings): Update to use generic_branch_cost.
17413 (cortexa53_tunings): Likewise.
17414 (cortexa57_tunings): Likewise.
17415 (cortexa72_tunings): Likewise.
17416 (cortexa73_tunings): Likewise.
17417 (thunderx2t99_tunings): Likewise.
17419 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17421 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17422 (sparc_option_override): Honour MASK_FSMULD.
17423 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17424 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17425 * config/sparc/sparc.opt (mfsmuld): New option.
17426 * doc/invoke.texi (mfsmuld): Document option.
17428 2017-07-26 Marek Polacek <polacek@redhat.com>
17430 PR middle-end/70992
17431 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17433 2017-07-26 Richard Biener <rguenther@suse.de>
17435 * gimple-match-head.c (do_valueize): Return OP if valueize
17437 (get_def): New helper to get at the def stmt of a SSA name
17438 if valueize allows.
17439 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
17440 do_valueize to get at the def stmt.
17441 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
17443 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
17445 PR middle-end/46932
17446 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
17448 2017-07-26 Martin Liska <mliska@suse.cz>
17451 * function.c (expand_function_start): Make expansion of
17452 nonlocal_goto_save_area after parm_birth_insn.
17454 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17456 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
17457 from all CPU target flags enable members.
17459 2017-07-26 Richard Biener <rguenther@suse.de>
17461 * genmatch.c (dt_simplify::gen): Make iterator vars const.
17462 (decision_tree::gen): Make 'type' const.
17463 (write_predicate): Likewise.
17465 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17467 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
17468 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
17469 (rs6000_option_override_internal): Likewise.
17470 (rs6000_expand_vector_set): Likewise.
17471 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
17472 (TARGET_UPPER_REGS_SF): Likewise.
17473 (TARGET_UPPER_REGS_DI): Likewise.
17474 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
17475 (TARGET_DIRECT_MOVE_64BIT): Likewise.
17476 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
17477 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17478 (Splitters for DI constants in Altivec registers): Likewise.
17479 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
17480 (vsx_set_v4sf_p9): Likewise.
17481 (vsx_set_v4sf_p9_zero): Likewise.
17482 (vsx_insert_extract_v4sf_p9): Likewise.
17483 (vsx_insert_extract_v4sf_p9_2): Likewise.
17485 2017-07-25 Carl Love <cel@us.ibm.com>
17487 * doc/extend.texi: Update the built-in documentation file for the
17488 existing built-in functions
17489 vector signed char vec_cnttz (vector signed char);
17490 vector unsigned char vec_cnttz (vector unsigned char);
17491 vector signed short vec_cnttz (vector signed short);
17492 vector unsigned short vec_cnttz (vector unsigned short);
17493 vector signed int vec_cnttz (vector signed int);
17494 vector unsigned int vec_cnttz (vector unsigned int);
17495 vector signed long long vec_cnttz (vector signed long long);
17496 vector unsigned long long vec_cnttz (vector unsigned long long);
17498 2017-07-25 Andrew Pinski <apinski@cavium.com>
17500 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
17501 accesses where the use is for the first operand of a BIT_INSERT.
17503 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17506 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17507 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17509 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17511 * config/i386/gstabs.h: Delete.
17512 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
17514 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
17516 * config/i386/i386.c (ix86_decompose_address): Do not check for
17517 register RTX when looking at index_reg or base_reg.
17518 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
17520 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
17522 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
17523 to update EH info here.
17525 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17527 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
17529 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17531 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
17533 2017-07-25 Torsten Duwe <duwe@suse.de>
17535 * common.opt: Introduce -fpatchable-function-entry
17536 command line option, and its variables function_entry_patch_area_size
17537 and function_entry_patch_area_start.
17538 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
17539 including a two-value parser.
17540 * target.def (print_patchable_function_entry): New target hook.
17541 * targhooks.h (default_print_patchable_function_entry): New function.
17542 * targhooks.c (default_print_patchable_function_entry): Likewise.
17543 * toplev.c (process_options): Switch off IPA-RA if
17544 patchable function entries are being generated.
17545 * varasm.c (assemble_start_function): Look at the
17546 patchable-function-entry command line switch and current
17547 function attributes and maybe generate NOP instructions by
17548 calling the print_patchable_function_entry hook.
17549 * doc/extend.texi: Document patchable_function_entry attribute.
17550 * doc/invoke.texi: Document -fpatchable_function_entry
17551 command line option.
17552 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
17554 * doc/tm.texi: Re-generate.
17556 2017-07-25 Jakub Jelinek <jakub@redhat.com>
17559 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
17560 TARGET_AVX512DQ rather than TARGET_AVX512BW.
17562 2017-07-25 Tamar Christina <tamar.christina@arm.com>
17564 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
17566 2017-07-25 Richard Biener <rguenther@suse.de>
17568 PR tree-optimization/81455
17569 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
17570 not walk in cycles when looking for guards.
17572 2017-07-25 Richard Biener <rguenther@suse.de>
17574 PR tree-optimization/81529
17575 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17576 when optimizing backedge uses.
17578 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17580 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17582 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17583 to dl_section_ref. On AIX, append an expression to subtract
17584 the size of the section length to dl_section_ref.
17586 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17588 * configure.ac: If any of the config.* scripts fail, exit 1.
17589 * configure: Regenerate.
17591 2017-07-25 Richard Biener <rguenther@suse.de>
17593 PR middle-end/81546
17594 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17595 of immediate uses, be more verbose on errors.
17597 2017-07-25 Richard Biener <rguenther@suse.de>
17599 PR tree-optimization/81510
17600 * tree-vect-loop.c (vect_is_simple_reduction): When the
17601 reduction stmt is not inside the loop bail out.
17603 2017-07-25 Richard Biener <rguenther@suse.de>
17605 PR tree-optimization/81303
17606 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17607 profitability check against LOOP_VINFO_NITERSM1.
17609 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17611 * domwalk.c (cmp_bb_postorder): Simplify.
17612 (sort_bbs_postorder): New function. Use it...
17613 (dom_walker::walk): ...here to optimize common cases.
17615 2017-07-25 Martin Liska <mliska@suse.cz>
17618 * ipa-visibility.c (function_and_variable_visibility): Make the
17619 redirection just on target that supports aliasing.
17620 Fix GNU coding style.
17622 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17625 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17627 * config/arm/rtems.h: Likewise.
17628 * config/bfin/rtems.h: Likewise.
17629 * config/i386/rtemself.h: Likewise.
17630 * config/lm32/rtems.h: Likewise.
17631 * config/m32c/rtems.h: Likewise.
17632 * config/m68k/rtemself.h: Likewise.
17633 * config/microblaze/rtems.h: Likewise.
17634 * config/mips/rtems.h: Likewise.
17635 * config/moxie/rtems.h: Likewise.
17636 * config/nios2/rtems.h: Likewise.
17637 * config/powerpcspe/rtems.h: Likewise.
17638 * config/rs6000/rtems.h: Likewise.
17639 * config/rtems.h: Likewise.
17640 * config/sh/rtems.h: Likewise.
17641 * config/sh/rtemself.h: Likewise.
17642 * config/sparc/rtemself.h: Likewise.
17644 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17647 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17648 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17649 * tree-ssa-structalias.c (alias_get_name): Same.
17651 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17654 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17655 instructions if no du chain is found.
17657 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17659 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17661 2017-07-25 Richard Biener <rguenther@suse.de>
17663 PR middle-end/81505
17664 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17667 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17669 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17670 upper-regs options.
17671 (ISA_2_7_MASKS_SERVER): Likewise.
17672 (ISA_3_0_MASKS_IEEE): Likewise.
17673 (OTHER_P8_VECTOR_MASKS): Likewise.
17674 (OTHER_VSX_VECTOR_MASKS): Likewise.
17675 (POWERPC_MASKS): Likewise.
17676 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17677 duplicate list of options.
17678 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17679 explicit -mupper-regs options.
17680 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17681 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17682 alias for -mupper-regs-df.
17683 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17684 (rs6000_init_hard_regno_mode_ok): Likewise.
17685 (rs6000_option_override_internal): Likewise.
17686 (rs6000_opt_masks): Likewise.
17687 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17688 options in terms of whether -mvsx or -mpower8-vector was used.
17689 (TARGET_UPPER_REGS_DI): Likewise.
17690 (TARGET_UPPER_REGS_SF): Likewise.
17691 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17692 -mupper-regs-* options.
17694 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17696 * passes.c (emergency_dump_function): Print some empty lines and a
17697 header before the RTL dump.
17699 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17701 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17703 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17706 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17707 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17709 2017-07-24 Carl Love <cel@us.ibm.com>
17711 * config/rs6000/rs6000-c.c: Add support for built-in functions
17712 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17713 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17714 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17715 vec_extract_fp_from_shortl): Add defines for the two builtins.
17716 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17717 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17719 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17720 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17721 * doc/extend.texi: Update the built-in documentation file for the
17722 new built-in function.
17724 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17727 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17729 * doc/generic.texi: Likewise.
17730 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17731 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17733 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17735 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17736 (aarch64_mls_elt_merge<mode>): Likewise.
17738 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17740 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17741 having __cxa_atexit.
17743 2017-07-23 Michael Collison <michael.collison@arm.com>
17745 * config/arm/arm.c (arm_option_override): Deprecate
17746 use of -mstructure-size-boundary.
17747 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17748 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17750 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17753 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17754 Reduce cost estimate for direct moves.
17756 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17759 * config/i386/i386.c (ix86_option_override_internal): Disable
17760 BMI, BMI2 and TBM instructions for -m16.
17762 2017-07-21 Carl Love <cel@us.ibm.com>
17764 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17765 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17766 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17767 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17768 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17769 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17770 VMULOSW): New enum "unspec" values.
17771 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17772 altivec_vmulosw): New patterns.
17773 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17774 VMULOSW): Add definitions.
17776 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17778 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17779 (qdf24xx): Likewise.
17780 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17781 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17782 (AARCH64_FL_V8_1): Renumber.
17783 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17784 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17785 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17786 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17787 rdma to feature modifiers list.
17789 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17791 PR middle-end/56727
17792 * ipa-visibility (function_and_variable_visibility): Convert
17793 recursive PLT call to direct call if appropriate.
17795 2017-07-21 Andrew Pinski <apinski@cavium.com>
17797 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17798 operand 1 to see if the types precision matches.
17799 * fold-const.c (operand_equal_p): Likewise.
17801 2017-07-21 Richard Biener <rguenther@suse.de>
17803 PR tree-optimization/81303
17804 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17805 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17806 (vect_peeling_hash_get_lowest_cost): Adjust.
17807 (vect_enhance_data_refs_alignment): Likewise. Use
17808 vect_get_peeling_costs_all_drs to compute the penalty for no
17809 peeling to match up costs.
17811 2017-07-21 Richard Biener <rguenther@suse.de>
17813 PR tree-optimization/81500
17814 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17815 we didn't identify a reduction path.
17817 2017-07-21 Tom de Vries <tom@codesourcery.com>
17818 Cesar Philippidis <cesar@codesourcery.com>
17820 PR gcov-profile/81442
17821 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17824 2017-07-21 Tom de Vries <tom@codesourcery.com>
17827 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17829 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17830 nvptx_override_options_after_change.
17832 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17834 * dwarf2out.c (output_file_names): Avoid double testing for
17835 dwarf_version >= 5.
17837 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17839 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17841 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17843 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17845 (try_crossjump_to_edge): Do not punt on partitioned functions.
17847 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17849 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17850 Put all BBs reachable only via paths crossing cold region to cold
17852 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17854 2016-07-21 Richard Biener <rguenther@suse.de>
17856 PR tree-optimization/81303
17857 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17858 into account prologue and epilogue iterations when raising
17859 min_profitable_iters to sth at least covering one vector iteration.
17861 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17863 * config/arm/arm.c (arm_test_cpu_arch_dat):
17866 2017-07-20 Nathan Sidwell <nathan@acm.org>
17868 Remove TYPE_METHODS.
17869 * tree.h (TYPE_METHODS): Delete.
17870 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17871 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17872 (dbxout_type_methods): Scan TYPE_FIELDS.
17873 (dbxout_type): Don't check TYPE_METHODS here.
17874 * function.c (use_register_for_decl): Always ignore register for
17875 class types when not optimizing.
17876 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17877 * tree.c (free_lang_data_in_type): Stitch out member functions and
17878 templates from TYPE_FIELDS.
17879 (build_distinct_type_copy, verify_type_variant,
17880 verify_type): Member fns are on TYPE_FIELDS.
17881 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17882 * tree-pretty-print.c (dump_generic_node): Likewise.
17884 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17887 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17888 V2TImode and V4TImode.
17889 (ix86_expand_vector_extract): Likewise.
17890 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17891 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17892 (ssescalarmode): Handle V4TImode and V2TImode.
17893 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17894 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17895 (VEXTRACTI128_MODE): New mode iterator.
17896 (splitter for *vec_extractv?ti first element): New.
17897 (VEC_INIT_MODE): New mode iterator.
17898 (vec_init<mode>): Consolidate 3 expanders into one using
17899 VEC_INIT_MODE mode iterator.
17901 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17903 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17904 non_spilled_static_chain_regno_p.
17906 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17908 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17910 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17912 * bb-reorder.c (connect_traces): Allow copying of blocks within
17915 2017-07-20 Richard Biener <rguenther@suse.de>
17917 * gimple.h (gimple_phi_result): Add gphi * overload.
17918 (gimple_phi_result_ptr): Likewise.
17919 (gimple_phi_arg): Likewise. Adjust index assert to only
17920 allow actual argument accesses rather than all slots available
17922 (gimple_phi_arg_def): Add gphi * overload.
17923 * tree-phinodes.c (make_phi_node): Initialize only actual
17925 (resize_phi_node): Clear memory not covered by old node,
17926 do not initialize excess argument slots.
17927 (reserve_phi_args_for_new_edge): Initialize new argument slot
17930 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17932 PR tree-optimization/81388
17934 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17936 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17937 by removing computation of may_be_zero.
17939 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17940 Tom de Vries <tom@codesourcery.com>
17942 PR middle-end/81030
17943 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17944 when gimple level profile disagrees with what RTL expander did.
17946 2017-07-20 Richard Biener <rguenther@suse.de>
17948 PR tree-optimization/61171
17949 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17950 (vect_analyze_stmt): Add slp instance parameter.
17951 (vectorizable_reduction): Likewise.
17952 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17953 (vect_is_simple_reduction): Deal with chains not detected
17954 as SLP reduction chain, specifically not properly associated
17955 chains containing a mix of plus/minus.
17956 (get_reduction_op): Remove.
17957 (get_initial_defs_for_reduction): Simplify, pass in whether
17958 this is a reduction chain, pass in the SLP node for the PHIs.
17959 (vect_create_epilog_for_reduction): Get the SLP instance as
17961 (vectorizable_reduction): Get the SLP instance as arg.
17962 During analysis remember the SLP node with the PHIs in the
17963 instance. Simplify getting at the vectorized reduction PHIs.
17964 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17965 through SLP instance.
17966 (vect_slp_analyze_operations): Likewise.
17967 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17968 (vect_transform_stmt): Likewise.
17970 2017-07-20 Tom de Vries <tom@codesourcery.com>
17972 PR tree-optimization/81489
17973 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17974 read of phi arg location to before loop that modifies phi.
17976 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17978 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17981 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17983 PR middle-end/81331
17984 * except.c (execute): Fix ordering issue.
17986 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17988 PR rtl-optimization/81423
17989 * combine.c (make_compound_operation_int): Don't try to optimize
17990 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17992 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17994 PR rtl-optimization/81423
17995 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17996 with a constant that is -1 in the truncated to mode.
17998 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18000 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18001 (determine_unlikely_bbs): ... here.
18002 * predict.h (propagate_unlikely_bbs_forward): Declare.
18003 * cfgexpand.c (pass_expand::execute): Use it.
18004 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18006 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18007 propagate_unlikely_bbs_forward.
18009 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18011 PR middle-end/81331
18012 * except.c (maybe_add_nop_after_section_switch): New function.
18015 2017-07-19 Tom de Vries <tom@codesourcery.com>
18017 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18019 2017-07-19 Tom de Vries <tom@codesourcery.com>
18021 * gimple.h (gimple_phi_arg): Make assert more strict.
18023 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18025 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18026 * config/rs6000/mmintrin.h: New file.
18027 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18029 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18031 PR tree-optimization/81346
18032 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18034 2017-07-19 Tom de Vries <tom@codesourcery.com>
18036 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18038 2017-07-19 Tom de Vries <tom@codesourcery.com>
18040 * config/nvptx/nvptx-modes.def: Add V2DImode.
18041 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18042 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18043 (nvptx_output_mov_insn): Handle lack of mov.b128.
18044 (nvptx_print_operand): Handle 'H' and 'L' codes.
18045 (nvptx_vector_mode_supported): Allow V2DImode.
18046 (nvptx_preferred_simd_mode): New function.
18047 (nvptx_data_alignment): New function.
18048 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18049 nvptx_preferred_simd_mode.
18050 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18052 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18054 2017-07-19 Tom de Vries <tom@codesourcery.com>
18056 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18057 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18058 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18059 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18060 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18061 (mov<VECIM>_insn): New define_insn.
18062 (define_expand "mov<VECIM>): New define_expand.
18064 2017-07-19 Tom de Vries <tom@codesourcery.com>
18066 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18068 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18070 PR tree-optimization/81346
18071 * fold-const.h (fold_div_compare, range_check_type): Declare.
18072 * fold-const.c (range_check_type): New function.
18073 (build_range_check): Use range_check_type.
18074 (fold_div_compare): No longer static, rewritten into
18075 a match.pd helper function.
18076 (fold_comparison): Don't call fold_div_compare here.
18077 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18078 as helper function.
18080 2017-07-19 Nathan Sidwell <nathan@acm.org>
18082 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18083 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18084 * tree.c (find_decls_types_r, verify_type): Use
18085 TYPE_{MIN,MAX}_VALUE_RAW.
18086 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18087 (hash_tree): Likewise.
18088 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18090 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18093 2017-07-18 Tom de Vries <tom@codesourcery.com>
18095 PR middle-end/81464
18096 * omp-expand.c (expand_omp_for_static_chunk): Handle
18097 equal-argument loop exit phi.
18099 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18102 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18103 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18104 operand 2 predicate.
18105 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18106 operand 2 predicate.
18107 (ror,rol -> rorx splitters): Use const_int_operand as
18108 operand 2 predicate.
18110 2017-06-18 Richard Biener <rguenther@suse.de>
18112 PR tree-optimization/81410
18113 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18114 the gap in the ! slp_perm SLP case after each group.
18116 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18118 PR middle-end/81463
18119 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18122 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18124 PR middle-end/81462
18125 * predict.c (set_even_probabilities): Cleanup; do not affect
18126 probabilities that are already known.
18127 (combine_predictions_for_bb): Call even when count is set.
18129 2017-07-18 Nathan Sidwell <nathan@acm.org>
18131 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18134 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18137 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
18138 optimization for loop niter analysis.
18140 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
18143 * config/avr/avr.c (avr_optimize_casesi): Don't use
18144 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
18146 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
18148 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
18149 body_cost_vec from _vect_peel_extended_info.
18150 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
18151 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
18154 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18156 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
18158 2017-07-18 Richard Biener <rguenther@suse.de>
18160 PR tree-optimization/80620
18161 PR tree-optimization/81403
18162 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
18163 info when re-using a VN table entry.
18165 2017-07-18 Richard Biener <rguenther@suse.de>
18167 PR tree-optimization/81418
18168 * tree-vect-loop.c (vectorizable_reduction): Properly compute
18169 vectype_in. Verify that with lane-reducing reduction operations
18170 we have a single def-use cycle.
18172 2017-07-17 Carl Love <cel@us.ibm.com>
18174 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
18176 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18177 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18178 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18179 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18180 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18181 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18182 VMULOSW): New enum "unspec" values.
18183 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18184 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18185 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18186 altivec_vmulosw): New patterns.
18187 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18188 VMULOSW): Add definitions.
18189 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
18191 * config/alpha/alpha.c: Include predict.h.
18193 2017-07-17 Yury Gribov <tetra2005@gmail.com>
18195 * tree-vrp.c (compare_assert_loc): Fix comparison function
18196 to return predictable results.
18198 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18200 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
18202 (subdi3): Likewise.
18203 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
18204 * doc/invoke.texi (mexpand-adddi): Update text.
18206 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18208 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
18209 that also clobbers the CC register. The old expand code is moved
18211 (*arc_clzsi2): ... here.
18212 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
18213 the CC register. The old expand code is moved to ...
18214 (arc_ctzsi2): ... here.
18216 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18218 * config/arc/arc.opt (mindexed-loads): Use initial value
18219 TARGET_INDEXED_LOADS_DEFAULT.
18220 (mauto-modify-reg): Use initial value
18221 TARGET_AUTO_MODIFY_REG_DEFAULT.
18222 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18223 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18224 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18225 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18227 2017-07-17 Martin Liska <mliska@suse.cz>
18230 * opts.c (finish_options): Do not allow -fgnu-tm
18231 w/ -fsanitize={kernel-,}address. Say sorry.
18233 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18236 * tree-loop-distribution.c (classify_partition): Only assert on
18237 numer of iterations.
18238 (merge_dep_scc_partitions): Delete prameter. Update function call.
18239 (distribute_loop): Remove code handling loop with unknown niters.
18240 (pass_loop_distribution::execute): Skip loop with unknown niters.
18242 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18245 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18246 function sort_partitions_by_post_order.
18248 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18250 PR tree-optimization/81374
18251 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18252 the max index of basic blocks, rather than number of basic blocks.
18254 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18256 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18258 (arc_legitimate_pic_operand_p): Likewise.
18259 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18261 (arc_needs_pcl_p): Likewise.
18262 (arc_legitimate_pc_offset_p): Likewise.
18263 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18264 function is also used in constrains.md.
18265 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18266 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18267 PLUS. Only return true/false in known cases, otherwise assert.
18268 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18269 is already called in arc_legitimate_constant_p.
18270 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18272 (LEGITIMATE_PIC_OPERAND_P): Use
18273 arc_raw_symbolic_reference_mentioned_p function.
18274 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18279 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18280 Andrew Burgess <andrew.burgess@embecosm.com>
18282 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18283 (arc_return_address_register): New function.
18284 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18285 (arc_handle_fndecl_attribute): Add naked attribute.
18286 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18287 (TARGET_WARN_FUNC_RETURN): Likewise.
18288 (arc_allocate_stack_slots_for_args): New function.
18289 (arc_warn_func_return): Likewise.
18290 (machine_function): Change type fn_type.
18291 (arc_compute_function_type): Consider new naked function type,
18292 change function return type.
18293 (arc_must_save_register): Adapt to handle new
18294 arc_compute_function_type's return type.
18295 (arc_expand_prologue): Likewise.
18296 (arc_expand_epilogue): Likewise.
18297 (arc_return_address_regs): Delete.
18298 (arc_return_address_register): New function.
18299 (arc_epilogue_uses): Use above function.
18300 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18301 (arc_function_type): Change encoding, add naked type.
18302 (ARC_INTERRUPT_P): Change to handle the new encoding.
18303 (ARC_FAST_INTERRUPT_P): Likewise.
18304 (ARC_NORMAL_P): Define.
18305 (ARC_NAKED_P): Likewise.
18306 (arc_compute_function_type): Delete prototype.
18307 * config/arc/arc.md (in_ret_delay_slot): Use
18308 arc_return_address_register function.
18309 (simple_return): Likewise.
18310 (p_return_i): Likewise.
18312 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18314 PR tree-optimization/81428
18315 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18316 can't be built for those types.
18318 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18320 Remove stuff dead since r239246.
18322 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18323 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18324 (avr_inform_devices): Remove dead stuff.
18326 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18328 * config/arm/arm_neon.h: Fix softp typo.
18330 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18332 PR tree-optimization/81365
18333 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18334 aggregate moves onto bb predecessor edges, make sure there are no
18335 loads that could alias the lhs in between the start of bb and the
18338 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18341 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18342 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18343 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18345 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18347 PR tree-optimization/81396
18348 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18349 (init_symbolic_number): Initialize it to 1.
18350 (perform_symbolic_merge): Add n_ops from both operands into the new
18352 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18353 without base_addr as useless if they need more than one operation.
18354 (bswap_replace): Handle !bswap case for NULL base_addr.
18356 2017-07-17 Tom de Vries <tom@codesourcery.com>
18359 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18362 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18364 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18365 conditional builtin define __FIX_LEON3FT_B2BST.
18367 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18369 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18370 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18373 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18375 PR rtl-optimization/81424
18376 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18377 to remove potential trapping from operands if -fnon-call-exceptions.
18379 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18381 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18382 profile_proability for scalling.
18383 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18385 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18387 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18389 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18391 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18392 fixpoint arithmetics.
18394 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18396 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18397 fixpoint arithmetics.
18399 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18401 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18402 fixpoint arithmetics.
18404 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18406 * profile-count.h (profile_probability::from_reg_br_prob_note,
18407 profile_probability::to_reg_br_prob_note): New functions.
18408 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18409 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18410 * predict.c (probability_reliable_p): Update.
18411 (edge_probability_reliable_p): Update.
18412 (br_prob_note_reliable_p): Update.
18413 (invert_br_probabilities): Update.
18414 (add_reg_br_prob_note): New function.
18415 (combine_predictions_for_insn): Update.
18416 * asan.c (asan_clear_shadow): Update.
18417 * cfgbuild.c (compute_outgoing_frequencies): Update.
18418 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18419 (update_br_prob_note): Update.
18420 (rtl_verify_edges): Update.
18421 (purge_dead_edges): Update.
18422 (fixup_reorder_chain): Update.
18423 * emit-rtl.c (try_split): Update.
18424 * ifcvt.c (cond_exec_process_insns): Update.
18425 (cond_exec_process_if_block): Update.
18426 (dead_or_predicable): Update.
18427 * internal-fn.c (expand_addsub_overflow): Update.
18428 (expand_neg_overflow): Update.
18429 (expand_mul_overflow): Update.
18430 * loop-doloop.c (doloop_modify): Update.
18431 * loop-unroll.c (compare_and_jump_seq): Update.
18432 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18433 * predict.h: Update.
18434 * reorg.c (mostly_true_jump): Update.
18436 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18437 * config/alpha/alpha.c (emit_unlikely_jump): Update.
18438 * config/arc/arc.c: (emit_unlikely_jump): Update.
18439 * config/arm/arm.c: (emit_unlikely_jump): Update.
18440 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
18441 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
18442 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
18443 (ix86_print_operand): Update.
18444 (ix86_split_fp_branch): Update.
18445 (predict_jump): Update.
18446 * config/ia64/ia64.c (ia64_print_operand): Update.
18447 * config/mmix/mmix.c (mmix_print_operand): Update.
18448 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
18449 (rs6000_expand_split_stack_prologue): Update.
18450 * config/rs6000/rs6000.c: Update.
18451 * config/s390/s390.c (s390_expand_vec_strlen): Update.
18452 (s390_expand_vec_movstr): Update.
18453 (s390_expand_cs_tdsi): Update.
18454 (s390_expand_split_stack_prologue): Update.
18455 * config/sh/sh.c (sh_print_operand): Update.
18456 (expand_cbranchsi4): Update.
18457 (expand_cbranchdi4): Update.
18458 * config/sparc/sparc.c (output_v9branch): Update.
18459 * config/spu/spu.c (get_branch_target): Update.
18460 (ea_load_store_inline): Update.
18461 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
18462 * config/tilepro/tilepro.c: Update.
18464 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18466 * gimplify.c (mostly_copy_tree_r): Revert latest change.
18467 (gimplify_save_expr): Likewise.
18469 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18471 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
18473 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18475 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
18477 * timevar.def (TV_IPA_FNSUMMARY): Define.
18479 2017-07-16 Daniel Cederman <cederman@gaisler.com>
18481 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
18482 to back store errata sensitive sequence from being generated.
18483 (sqrtdf2_fix): Likewise.
18485 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18487 * tree-ssa-threadupdate.c (compute_path_counts,
18488 update_joiner_offpath_counts): Use profile_probability.
18490 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
18493 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18495 * config/arm/arm-c.c (arm_cpu_builtins): Define
18496 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18498 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18500 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18501 array entries to represent __ieee128 versions of the
18502 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
18503 scalar_extract_sig, and scalar_insert_exp built-in functions.
18504 (altivec_resolve_overloaded_builtin): Add special case handling
18505 for the __builtin_scalar_insert_exp function, as represented by
18506 the P9V_BUILTIN_VEC_VSIEDP constant.
18507 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
18508 exponent support for __ieee128 argument.
18509 (VSESQP): Add scalar extract signature support for __ieee128
18511 (VSTDCNQP): Add scalar test negative support for __ieee128
18513 (VSIEQP): Add scalar insert exponent support for __int128 argument
18514 with __ieee128 result.
18515 (VSIEQPF): Add scalar insert exponent support for __ieee128
18516 argument with __ieee128 result.
18517 (VSTDCQP): Add scalar test data class support for __ieee128
18519 (VSTDCNQP): Add overload support for scalar test negative with
18520 __ieee128 argument.
18521 (VSTDCQP): Add overload support for scalar test data class
18522 __ieee128 argument.
18523 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
18524 UNSPEC_VSX_SXSIGDP.
18525 (UNSPEC_VSX_SIEXPQP): New constant.
18526 (xsxexpqp): New insn for VSX scalar extract exponent quad
18528 (xsxsigqp): New insn for VSX scalar extract significand quad
18530 (xsiexpqpf): New insn for VSX scalar insert exponent quad
18531 precision with floating point argument.
18532 (xststdcqp): New expand for VSX scalar test data class quad
18534 (xststdcnegqp): New expand for VSX scalar test negative quad
18536 (xststdcqp): New insn to match expansions for VSX scalar test data
18537 class quad precision and VSX scalar test negative quad precision.
18538 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
18539 special case operand checking to enforce that second operand of
18540 VSX scalar test data class with quad precision argument is a 7-bit
18542 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18543 prototypes and descriptions of __ieee128 versions of
18544 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
18545 scalar_test_data_class, and scalar_test_neg built-in functions.
18547 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18549 PR tree-optimization/81162
18550 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
18551 replace a negate with an add.
18553 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
18555 * doc/invoke.texi (arm/-mcpu): Document +crypto.
18557 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18559 * config/arm/arm-c.c (arm_cpu_builtins): Define
18560 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18562 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18564 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
18565 (armv8-r): Set ARM Cortex-R52 as default CPU.
18566 * config/arm/arm-tables.opt: Regenerate.
18567 * config/arm/arm-tune.md: Regenerate.
18568 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18570 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
18571 extension for -mcpu=cortex-r52.
18573 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18575 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18576 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18577 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18578 (fp-armv8): Define it as FP_ARMv8 only.
18579 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18580 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18582 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18583 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18584 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18585 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18586 than TARGET_FPU_ARMV8.
18587 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18588 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18589 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18591 * config/arm/neon.md (neon_vrint): Likewise.
18592 (neon_vcvt): Likewise.
18593 (neon_<fmaxmin_op><mode>): Likewise.
18594 (<fmaxmin><mode>3): Likewise.
18595 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18596 * config/arm/predicates.md (arm_cond_move_operator): Check against
18597 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18599 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18601 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18602 to top of function.
18604 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18606 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18607 loop in comment with memset.
18609 2017-07-14 Martin Liska <mliska@suse.cz>
18611 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18612 * dwarf2out.c (is_java): Remove the function.
18613 (output_pubname): Remove usage of the function.
18614 (lower_bound_default): Remove usage of DW_LANG_Java.
18615 (gen_compile_unit_die): Likewise.
18616 * gcc.c: Remove compiler defaults for .java and .zip files.
18617 * gimple-expr.c (remove_suffix): Change as there's no longer
18618 extension than 4-letter one.
18619 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18620 (gimplify_save_expr): Likewise.
18621 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18622 as it's possible even for other languages than Java.
18623 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18624 * lto-opts.c (lto_write_options): Remove reference to Java.
18625 * opts.c (strip_off_ending): Update file extension handling.
18626 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18627 * tree-eh.c (lower_resx): Likewise.
18628 * tree.c (free_lang_data_in_type): Remove dead code.
18629 (find_decls_types_r): Likewise.
18630 (build_common_builtin_nodes): Remove Java from a comment.
18631 (verify_type): Remove dead code.
18632 * varasm.c (assemble_external): Remove Java from a comment.
18634 2017-07-14 Martin Liska <mliska@suse.cz>
18636 * opts.c (finish_options): Add quotes.
18637 (common_handle_option): Likewise.
18639 2017-07-14 Martin Liska <mliska@suse.cz>
18641 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18642 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18643 Remove N_SO_PASCAL.
18644 * dwarf2out.c (lower_bound_default): Do not handle
18646 (gen_compile_unit_die): Likewise.
18647 * gcc.c: Remove default extension binding for GNU Pascal.
18648 * stmt.c: Remove Pascal language from a comment.
18649 * xcoffout.c: Likewise.
18651 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18654 * diagnostic-show-locus.c (fixit_cmp): New function.
18655 (layout::layout): Sort m_fixit_hints.
18656 (column_range::column_range): Assert that the values are valid.
18657 (struct char_span): New struct.
18658 (correction::overwrite): New method.
18659 (struct source_line): New struct.
18660 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18661 calls in terms of classes source_line and char_span, and
18662 correction::overwrite.
18663 (selftest::test_overlapped_fixit_printing_2): New function.
18664 (selftest::diagnostic_show_locus_c_tests): Call it.
18666 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18668 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18669 early if there is no lhs.
18671 2017-07-13 Martin Liska <mliska@suse.cz>
18673 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18674 (gen_reference_type_die): Likewise.
18675 * stor-layout.c: Remove Pascal-related comment.
18677 2017-07-13 Martin Liska <mliska@suse.cz>
18679 * opts.c (finish_options): Add quotes to error messages.
18680 (parse_sanitizer_options): Likewise.
18682 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18684 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18686 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18688 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18690 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18692 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18694 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18696 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18699 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18700 provides the hardware capability bits, define the macro
18701 __BUILTIN_CPU_SUPPORTS__.
18702 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18703 if GLIBC does not provide the hardware capability bits. Add a
18704 gcc_unreachable call if the built-in cpu function is neither
18705 __builtin_cpu_is nor __builtin_cpu_supports.
18706 (rs6000_get_function_versions_dispatcher): Change the warning
18707 that an old GLIBC is used which does not export the capability
18708 bits to be an error.
18709 * doc/extend.texi (target_clones attribute): Document the
18710 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18711 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18712 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18713 the macros defined by GCC if the newer GLIBC is available.
18715 2017-07-12 Jeff Law <law@redhat.com>
18717 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18718 remaining includes slightly.
18719 * config/riscv/riscv-builtins.c: Include profile-count.h.
18721 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18724 * config/avr/avr.c (avr_set_current_function): In diagnostic
18725 messages: Quote keywords and (parts of) identifiers.
18726 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18729 2017-07-12 Carl Love <cel@us.ibm.com>
18731 * config/rs6000/rs6000-c.c: Add support for built-in functions
18732 vector bool char vec_revb (vector bool char);
18733 vector bool short vec_revb (vector short char);
18734 vector bool int vec_revb (vector bool int);
18735 vector bool long long vec_revb (vector bool long long);
18736 * doc/extend.texi: Update the built-in documentation file for the
18737 new built-in functions.
18739 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18741 * config/s390/s390.md: Remove movcc splitter.
18743 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18745 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18746 load/store on condition.
18748 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18751 * config/avr/avr.c (avr_encode_section_info)
18752 [progmem && !TREE_READONLY]: Error if progmem object needs
18755 2017-07-11 Michael Collison <michael.collison@arm.com>
18757 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18760 2017-07-11 Carl Love <cel@us.ibm.com>
18762 * config/rs6000/rs6000-c.c: Add support for builtins
18763 vector unsigned int vec_parity_lsbb (vector signed int);
18764 vector unsigned int vec_parity_lsbb (vector unsigned int);
18765 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18766 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18767 vector unsigned long long vec_parity_lsbb (vector signed long long);
18768 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18769 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18770 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18771 * doc/extend.texi: Update the built-in documentation file for the
18772 new built-in functions.
18774 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18776 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18777 (layout::m_primary_loc): New field.
18778 (layout::layout): Initialize new field. Move location filtering
18779 logic from here to...
18780 (layout::maybe_add_location_range): ...this new method. Add
18781 support for filtering to just the lines already specified by other
18783 (layout::will_show_line_p): New method.
18784 (gcc_rich_location::add_location_if_nearby): New method.
18785 (selftest::test_add_location_if_nearby): New test function.
18786 (selftest::diagnostic_show_locus_c_tests): Call it.
18787 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18790 2017-07-11 Tom de Vries <tom@codesourcery.com>
18792 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18793 (bb_first_real_insn): New function.
18794 (nvptx_single): Add extra initialization of broadcasted condition
18797 2017-07-11 Nathan Sidwell <nathan@acm.org>
18799 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18801 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18803 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18804 Binutils doc as TEXI will mess them up.
18805 * doc/invoke.texi (AVR Options): Same here.
18807 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18809 * config/sparc/sparc.opt (mfix-ut700): New option.
18810 (mfix-gr712rc): Likewise.
18811 (sparc_fix_b2bst): New variable.
18812 * doc/invoke.texi (SPARC options): Document them.
18813 (ARM options): Fix warnings.
18814 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18815 instructions to prevent sequences that can trigger the store-store
18816 errata for certain LEON3FT processors.
18817 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18818 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18819 * config/sparc/sparc.md (fix_b2bst): New attribute.
18820 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18822 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18825 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18827 (*rsqrtsf2_sse): Ditto.
18829 (div<mode>3): Macroize insn from divdf3 and divsf3
18830 using MODEF mode iterator.
18832 2017-07-10 Martin Sebor <msebor@redhat.com>
18834 PR tree-optimization/80397
18835 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18836 instead of testing for equality to INTEGER_TYPE.
18838 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18840 * config.gcc: Remove uclibc from arc target spec.
18842 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18844 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18846 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18849 * lto-wrapper.c (remove_option): New function.
18850 (merge_and_complain): Merge PIC/PIE options more realistically.
18852 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18854 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18858 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18859 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18860 * config.in: Regenerate.
18861 * configure: Regenerate.
18862 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18863 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18864 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18865 (TARGET_GASISR_PROLOGUES): ...target mask.
18866 * common/config/avr/avr-common.c
18867 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18868 Set -mgas-isr-prologues.
18869 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18870 INSERT_PASS_BEFORE for it.
18871 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18872 * config/avr/avr.c (avr_option_override)
18873 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18874 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18875 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18876 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18877 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18879 (make_avr_pass_pre_proep): New function.
18880 (emit_push_sfr) <treg>: Add argument to function and use it
18881 instead of TMP_REG.
18882 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18883 and set machine->gasisr.yes.
18884 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18885 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18886 __gcc_isr.n_pushed to .L__stack_usage.
18887 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18888 (avr_asm_final_postscan_insn): ...this new static function.
18889 * config/avr/avr.h (machine_function)
18890 <is_no_gccisr, use_L__stack_usage>: New fields.
18891 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18892 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18893 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18894 (gasisr, *gasisr): New expander and insn.
18895 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18896 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18897 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18899 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18901 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18904 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18906 Move jump-tables out of .text again.
18909 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18910 (ASM_OUTPUT_ADDR_VEC): New function.
18911 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18912 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18913 INSN_ADDRESSes as asm comment.
18914 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18915 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18916 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18917 * config/avr/avr.md (*tablejump): Adjust comment.
18918 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18919 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18921 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18922 (avr_output_addr_vec): New proto.
18923 (avr_log_t) <insn_addresses>: New field.
18925 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18928 * config/i386/i386.c (ix86_function_arg_advance): Set
18929 outgoing_args_on_stack to true if there are outgoing arguments
18931 (ix86_function_arg): Likewise.
18932 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18933 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18934 * config/i386/i386.h (machine_function): Add
18935 outgoing_args_on_stack.
18937 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18939 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18940 supporting pthreds.
18941 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18943 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18945 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18946 (REAL_H): Remove $(MACHMODE_H).
18947 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18949 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18950 $(MACHMODE_H) and double-int.h.
18951 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18953 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18956 2017-07-07 Andrew Pinski <apinski@cavium.com>
18958 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18959 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18961 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18963 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18964 Add warning if GCC was not configured to link against a GLIBC that
18965 exports the hardware capability bits.
18966 (make_resolver_func): Make resolver function private and not a
18967 COMDAT function. Create the name with clone_function_name instead
18968 of make_unique_name.
18971 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18972 correct operand in doing the split.
18974 2017-07-07 Carl Love <cel@us.ibm.com>
18976 * config/rs6000/rs6000-c: Add support for built-in function
18977 vector unsigned short vec_pack_to_short_fp32 (vector float,
18979 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18980 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18981 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18982 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18983 (convert_4f32_8i16): Add define_expand.
18984 * doc/extend.texi: Update the built-in documentation file for the
18985 new built-in function.
18987 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18989 * config/sparc/m8.md: New file.
18990 * config/sparc/sparc.md: Include m8.md.
18992 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18994 * config/sparc/sparc.opt: New option -mvis4b.
18995 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18996 (sparc_option_override): Handle VIS4B.
18997 (enum sparc_builtins): Define
18998 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18999 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19000 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19001 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19002 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19003 (check_constant_argument): New function.
19004 (sparc_vis_init_builtins): Define builtins
19005 __builtin_vis_dictunpack{8,16,32},
19006 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19007 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19008 __builtin_vis_fpcmpde{8,16,32}shl and
19009 __builtin_vis_fpcmpur{8,16,32}shl.
19010 (sparc_expand_builtin): Check that the constant operands to
19011 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19012 constant and in range.
19013 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19015 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19016 (SPARC_IMM5_P): Likewise.
19017 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19018 (enabled): Handle vis4b.
19019 (UNSPEC_DICTUNPACK): New unspec.
19020 (UNSPEC_FPCMPSHL): Likewise.
19021 (UNSPEC_FPUCMPSHL): Likewise.
19022 (UNSPEC_FPCMPDESHL): Likewise.
19023 (UNSPEC_FPCMPURSHL): Likewise.
19024 (cpu_feature): New CPU feature `vis4b'.
19025 (dictunpack{8,16,32}): New insns.
19026 (FPCSMODE): New mode iterator.
19027 (fpcscond): New code iterator.
19028 (fpcsucond): Likewise.
19029 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19030 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19031 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19032 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19033 * config/sparc/constraints.md: Define constraints `q' for unsigned
19034 2-bit integer constants and `t' for unsigned 5-bit integer
19036 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19038 (imm5_operand_dictunpack16): Likewise.
19039 (imm5_operand_dictunpack32): Likewise.
19040 (imm2_operand): Likewise.
19041 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19042 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19043 ditunpack* and fpcmp*shl builtins.
19045 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19047 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19048 * config.in: Add HAVE_AS_SPARC6 define.
19049 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19051 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19053 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19054 (CPP_CPU_SPEC): Handle m8.
19055 (ASM_CPU_SPEC): Likewise.
19056 * config/sparc/sparc-opts.h (enum processor_type): Add
19058 * config/sparc/sparc.c (m8_costs): New struct.
19059 (sparc_option_override): Handle TARGET_CPU_m8.
19060 (sparc32_initialize_trampoline): Likewise.
19061 (sparc64_initialize_trampoline): Likewise.
19062 (sparc_issue_rate): Likewise.
19063 (sparc_register_move_cost): Likewise.
19064 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19065 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19066 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19067 (CPP_CPU_SPEC): Handle M8.
19068 (ASM_CPU_SPEC): Likewise.
19069 (AS_M8_FLAG): Define.
19070 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19071 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19072 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19074 * configure: Regenerate.
19075 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19078 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19080 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19082 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19083 ("*movdi_insn_sp32"): Do not set v3pipe.
19084 ("*movsi_insn"): Likewise.
19085 ("*movdi_insn_sp64"): Likewise.
19086 ("*movsf_insn"): Likewise.
19087 ("*movdf_insn_sp32"): Likewise.
19088 ("*movdf_insn_sp64"): Likewise.
19089 ("*zero_extendsidi2_insn_sp64"): Likewise.
19090 ("*sign_extendsidi2_insn"): Likewise.
19091 ("*mov<VM32:mode>_insn"): Likewise.
19092 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19093 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19094 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19095 ("<vlop:code><VL:mode>3"): Likewise.
19096 ("*not_<vlop:code><VL:mode>3"): Likewise.
19097 ("*nand<VL:mode>_vis"): Likewise.
19098 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19099 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19100 ("one_cmpl<VL:mode>2"): Likewise.
19101 ("faligndata<VM64:mode>_vis"): Likewise.
19102 ("alignaddrsi_vis"): Likewise.
19103 ("alignaddrdi_vis"): Likweise.
19104 ("alignaddrlsi_vis"): Likewise.
19105 ("alignaddrldi_vis"): Likewise.
19106 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19107 ("bmaskdi_vis"): Likewise.
19108 ("bmasksi_vis"): Likewise.
19109 ("bshuffle<VM64:mode>_vis"): Likewise.
19110 ("cmask8<P:mode>_vis"): Likewise.
19111 ("cmask16<P:mode>_vis"): Likewise.
19112 ("cmask32<P:mode>_vis"): Likewise.
19113 ("pdistn<P:mode>_vis"): Likewise.
19114 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19116 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19118 * config/sparc/sparc.md ("subtype"): New insn attribute.
19119 ("*wrgsr_sp64"): Set insn subtype.
19120 ("*rdgsr_sp64"): Likewise.
19121 ("alignaddrsi_vis"): Likewise.
19122 ("alignaddrdi_vis"): Likewise.
19123 ("alignaddrlsi_vis"): Likewise.
19124 ("alignaddrldi_vis"): Likewise.
19125 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19126 ("fexpand_vis"): Likewise.
19127 ("fpmerge_vis"): Likewise.
19128 ("faligndata<VM64:mode>_vis"): Likewise.
19129 ("bshuffle<VM64:mode>_vis"): Likewise.
19130 ("cmask8<P:mode>_vis"): Likewise.
19131 ("cmask16<P:mode>_vis"): Likewise.
19132 ("cmask32<P:mode>_vis"): Likewise.
19133 ("fchksm16_vis"): Likewise.
19134 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
19135 ("fmean16_vis"): Likewise.
19136 ("fp<plusminus_insn>64_vis"): Likewise.
19137 ("<plusminus_insn>v8qi3"): Likewise.
19138 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19139 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
19140 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
19141 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
19142 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
19143 ("*movqi_insn"): Likewise.
19144 ("*movhi_insn"): Likewise.
19145 ("*movsi_insn"): Likewise.
19146 ("movsi_pic_gotdata_op"): Likewise.
19147 ("*movdi_insn_sp32"): Likewise.
19148 ("*movdi_insn_sp64"): Likewise.
19149 ("movdi_pic_gotdata_op"): Likewise.
19150 ("*movsf_insn"): Likewise.
19151 ("*movdf_insn_sp32"): Likewise.
19152 ("*movdf_insn_sp64"): Likewise.
19153 ("*zero_extendhisi2_insn"): Likewise.
19154 ("*zero_extendqihi2_insn"): Likewise.
19155 ("*zero_extendqisi2_insn"): Likewise.
19156 ("*zero_extendqidi2_insn"): Likewise.
19157 ("*zero_extendhidi2_insn"): Likewise.
19158 ("*zero_extendsidi2_insn_sp64"): Likewise.
19159 ("ldfsr"): Likewise.
19160 ("prefetch_64"): Likewise.
19161 ("prefetch_32"): Likewise.
19162 ("tie_ld32"): Likewise.
19163 ("tie_ld64"): Likewise.
19164 ("*tldo_ldub_sp32"): Likewise.
19165 ("*tldo_ldub1_sp32"): Likewise.
19166 ("*tldo_ldub2_sp32"): Likewise.
19167 ("*tldo_ldub_sp64"): Likewise.
19168 ("*tldo_ldub1_sp64"): Likewise.
19169 ("*tldo_ldub2_sp64"): Likewise.
19170 ("*tldo_ldub3_sp64"): Likewise.
19171 ("*tldo_lduh_sp32"): Likewise.
19172 ("*tldo_lduh1_sp32"): Likewise.
19173 ("*tldo_lduh_sp64"): Likewise.
19174 ("*tldo_lduh1_sp64"): Likewise.
19175 ("*tldo_lduh2_sp64"): Likewise.
19176 ("*tldo_lduw_sp32"): Likewise.
19177 ("*tldo_lduw_sp64"): Likewise.
19178 ("*tldo_lduw1_sp64"): Likewise.
19179 ("*tldo_ldx_sp64"): Likewise.
19180 ("*mov<VM32:mode>_insn"): Likewise.
19181 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19182 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19184 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19186 * config/sparc/sparc.md ("type"): New insn type viscmp.
19187 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
19189 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
19190 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
19191 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19192 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
19194 ("n7_vis_logical_11cycle"): Likewise.
19195 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
19196 * config/sparc/niagara2.md ("niag3_vis": Likewise.
19197 * config/sparc/niagara.md ("niag_vis"): Likewise.
19198 * config/sparc/ultra3.md ("us3_fga"): Likewise.
19199 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
19201 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19203 * config/sparc/sparc.md: New instruction type `bmask'.
19204 (bmaskdi_vis): Use the `bmask' type.
19205 (bmasksi_vis): Likewise.
19206 * config/sparc/ultra3.md (us3_array): Likewise.
19207 * config/sparc/niagara7.md (n7_array): Likewise.
19208 * config/sparc/niagara4.md (n4_array): Likewise.
19209 * config/sparc/niagara2.md (niag2_vis): Likewise.
19210 (niag3_vis): Likewise.
19211 * config/sparc/niagara.md (niag_vis): Likewise.
19213 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19215 * ipa-comdats.c: Remove optimize check from gate.
19216 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
19217 for functions not optimized.
19218 (ipa_fn_summary_read): Skip optimize check.
19219 (ipa_fn_summary_write): Likewise.
19220 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
19222 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
19224 (can_inline_edge_p): Check flag_pcc_struct_return for match.
19225 (check_callers): Give up on caller which is not optimized.
19226 (inline_small_functions): Likewise.
19227 (ipa_inline): Do not give up when not optimizing.
19228 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
19229 away unoptimizes cdtors.
19230 (whole_program_function_and_variable_visibility): Do
19231 ipa_discover_readonly_nonaddressable_vars in LTO mode.
19232 * ipa.c (process_references): Do not check optimize.
19233 (symbol_table::remove_unreachable_nodes): Update optimize check.
19234 (set_writeonly_bit): Update optimize check.
19235 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19236 (pass_ipa_single_use::gate): Remove.
19238 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19240 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19241 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19242 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19243 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19244 permute_load, permute_store, adjust_extract, adjust_splat,
19245 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19246 replace_swap_with_copy, dump_swap_insn_table,
19247 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19248 recombine_lvx_pattern, recombine_stvx_pattern,
19249 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19250 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19251 to file rs6000-p8swap.c.
19252 * config/rs6000/rs6000-p8swap.c: New file.
19253 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19254 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19255 and rs6000*-*-* targets.
19257 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19259 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19261 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19263 * lto-wrapper.c (merge_and_complain): Do not merge
19264 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19265 fsigned_zeros, ftrapping_math, fwrapv.
19266 (append_compiler_options): Do not track these options.
19267 (append_linker_options): Likewie
19269 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19271 * cgraphunit.c (cgraph_node::finalize_function): When
19272 !flag_toplevel_reorde set no_reorder flag.
19273 (varpool_node::finalize_decl): Likewise.
19274 (symbol_table::compile): Drop no toplevel reorder path.
19276 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19278 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19279 edges; zero probability is not better than uninitialized.
19281 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19283 * asan.h (asan_sanitize_allocas_p): Declare.
19284 * asan.c (asan_sanitize_allocas_p): New function.
19285 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19286 (handle_builtin_alloca): Likewise.
19287 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19288 if !asan_sanitize_allocas_p.
19289 * params.def (asan-instrument-allocas): Add new option.
19290 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19291 * opts.c (common_handle_option): Disable allocas sanitization for
19294 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19296 * asan.c: Include gimple-fold.h.
19297 (get_last_alloca_addr): New function.
19298 (handle_builtin_stackrestore): Likewise.
19299 (handle_builtin_alloca): Likewise.
19300 (asan_emit_allocas_unpoison): Likewise.
19301 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19302 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19303 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19304 (instrument_builtin_call): Pass gimple iterator to
19305 get_mem_refs_of_builtin_call.
19306 (last_alloca_addr): New global.
19307 * asan.h (asan_emit_allocas_unpoison): Declare.
19308 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19309 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19310 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19311 if function calls alloca.
19312 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19313 * gimple-fold.h (replace_call_with_value): Declare.
19314 * internal-fn.c: Include asan.h.
19315 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19316 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19318 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19320 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19321 (C_SELFTEST_FLAGS): New.
19322 (CPP_SELFTEST_FLAGS): New.
19323 (SELFTEST_DEPS): New, from deps of s-selftest.
19324 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19325 (CPP_SELFTEST_DEPS): New.
19326 (selftest): Add dependency on s-selftest-c++.
19327 (s-selftest): Rename to...
19328 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19329 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19330 than SELFTEST_FLAGS.
19331 (selftest-gdb): Rename to...
19332 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19334 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19335 (selftest-valgrind): Rename to...
19336 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19338 (selftest-valgrind): Reintroduce as an alias for
19339 selftest-c-valgrind.
19340 (s-selftest-c++): New.
19341 (selftest-c++-gdb): New.
19342 (selftest-c++-valgrind): New.
19344 2017-07-06 Olivier Hainque <hainque@adacore.com>
19346 * gcc.c (process_command): When deciding if undefined variables
19347 should be ignored when processing specs, accept "gcc -v" as well.
19349 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19351 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19352 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19354 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19356 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19357 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19358 * config/arm/arm-tables.opt: Regenerate.
19359 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19361 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19363 2017-07-06 Carl Love <cel@us.ibm.com>
19365 * ChangeLog: Clean up from mid air collision
19367 2017-07-06 Carl Love <cel@us.ibm.com>
19369 * config/rs6000/rs6000-c.c: Add support for built-in functions
19370 vector signed int vec_subc (vector signed int, vector signed int);
19371 vector signed __int128 vec_subc (vector signed __int128,
19372 vector signed __int128);
19373 vector unsigned __int128 vec_subc (vector unsigned __int128,
19374 vector unsigned __int128);
19375 vector signed int vec_sube (vector signed int, vector signed int,
19376 vector signed int);
19377 vector unsigned int vec_sube (vector unsigned int,
19378 vector unsigned int,
19379 vector unsigned int);
19380 vector signed __int128 vec_sube (vector signed __int128,
19381 vector signed __int128,
19382 vector signed__int128);
19383 vector unsigned __int128 vec_sube (vector unsigned __int128,
19384 vector unsigned __int128,
19385 vector unsigned __int128);
19386 vector signed int vec_subec (vector signed int, vector signed int,
19387 vector signed int);
19388 vector unsigned int vec_subec (vector unsigned int,
19389 vector unsigned int,
19390 vector unsigned int);
19391 vector signed __int128 vec_subec (vector signed __int128,
19392 vector signed __int128,
19393 vector signed__int128);
19394 vector unsigned __int128 vec_subec (vector unsigned __int128,
19395 vector unsigned __int128,
19396 vector unsigned __int128);
19397 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19398 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19399 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19400 BU_ALTIVEC_OVERLOAD_X definitions.
19401 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19402 * doc/extend.texi: Update the built-in documentation file for the new
19403 built-in functions.
19405 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19408 * diagnostic-show-locus.c (layout::layout): Use start and finish
19409 spelling location for the start and finish of each range.
19410 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19411 Add unused aspect param.
19412 * input.c (expand_location_1): Add "aspect" param, and use it
19413 to access the correct part of the location.
19414 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19416 (expand_location_to_spelling_point): Likewise.
19417 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19418 param, and pass it to expand_location_1.
19420 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19422 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19423 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19424 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19425 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19426 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19427 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19428 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19429 _mm_maskz_getmant_ss): New intrinsics.
19430 (__builtin_ia32_getexpss128_mask): Changed to ...
19431 __builtin_ia32_getexpss128_round ... this.
19432 (__builtin_ia32_getexpsd128_mask): Changed to ...
19433 __builtin_ia32_getexpsd128_round ... this.
19434 * config/i386/i386-builtin-types.def
19435 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19436 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19437 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
19438 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
19439 __builtin_ia32_getmantss_mask_round): New builtins.
19440 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
19441 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
19442 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
19443 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
19444 * config/i386/sse.md
19445 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
19446 avx512f_sgetexp<mode><mask_scalar_name>
19447 <round_saeonly_scalar_name> ... this.
19448 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
19449 %0, %1, %2<round_saeonly_op3>}): Changed to ...
19450 vgetexp<ssescalarmodesuffix>
19451 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19452 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
19453 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
19454 avx512f_vgetmant<mode><mask_scalar_name>
19455 <round_saeonly_scalar_name> ... this.
19456 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19457 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
19458 vgetmant<ssescalarmodesuffix>
19459 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
19460 %0<mask_scalar_operand4>, %1, %2
19461 <round_saeonly_scalar_mask_op4>, %3} ... this.
19462 * config/i386/subst.md (mask_scalar_operand4,
19463 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
19464 round_saeonly_scalar_nimm_predicate): New subst attributes.
19466 2017-07-06 Julia Koval <julia.koval@intel.com>
19468 * config/i386/i386.c (ix86_erase_embedded_rounding):
19469 Remove code for old rounding pattern.
19471 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
19473 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
19475 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
19477 * doc/sourcebuild.texi (Test Directives, Variants of
19478 dg-require-support): Add documentation for dg-require-stack-check.
19480 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
19482 * config/i386/subst.md (mask_scalar, round_scalar,
19483 round_saeonly_scalar): New meta-templates.
19484 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
19485 round_scalar_mask_operand3, round_scalar_mask_op3,
19486 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
19487 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
19488 round_saeonly_scalar_constraint,
19489 round_saeonly_scalar_prefix): New subst attribute.
19490 * config/i386/sse.md
19491 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
19492 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
19493 <round_scalar_name> ... this.
19494 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
19495 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
19496 <round_scalar_name> ... this.
19497 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
19498 <sse>_vm<code><mode>3<mask_scalar_name>
19499 <round_saeonly_scalar_name> ... this.
19500 (v<plusminus_mnemonic><ssescalarmodesuffix>
19501 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19502 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19503 v<plusminus_mnemonic><ssescalarmodesuffix>
19504 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19505 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19506 (v<multdiv_mnemonic><ssescalarmodesuffix>
19507 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19508 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19509 v<multdiv_mnemonic><ssescalarmodesuffix>
19510 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19511 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19512 (v<maxmin_float><ssescalarmodesuffix>
19513 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
19514 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
19515 v<maxmin_float><ssescalarmodesuffix>
19516 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19517 %0<mask_scalar_operand3>, %1, %<iptr>2
19518 <round_saeonly_scalar_mask_op3>} ... this.
19520 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
19522 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
19523 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
19525 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
19526 Alan Hayward <alan.hayward@arm.com>
19527 David Sherwood <david.sherwood@arm.com>
19529 * combine.c (simplify_if_then_else): Remove "enum" before
19531 * compare-elim.c (can_eliminate_compare): Likewise.
19532 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19534 (aarch64_lookup_simd_builtin_type): Likewise.
19535 (aarch64_simd_builtin_type): Likewise.
19536 (aarch64_init_simd_builtin_types): Likewise.
19537 (aarch64_simd_expand_args): Likewise.
19538 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
19540 (aarch64_reverse_mask): Likewise.
19541 (aarch64_simd_emit_reg_reg_move): Likewise.
19542 (aarch64_gen_adjusted_ldpstp): Likewise.
19543 (aarch64_ccmp_mode_to_code): Likewise.
19544 (aarch64_operands_ok_for_ldpstp): Likewise.
19545 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19546 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19548 (aarch64_min_divisions_for_recip_mul): Likewise.
19549 (aarch64_reassociation_width): Likewise.
19550 (aarch64_get_condition_code_1): Likewise.
19551 (aarch64_simd_emit_reg_reg_move): Likewise.
19552 (aarch64_simd_attr_length_rglist): Likewise.
19553 (aarch64_reverse_mask): Likewise.
19554 (aarch64_operands_ok_for_ldpstp): Likewise.
19555 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19556 (aarch64_gen_adjusted_ldpstp): Likewise.
19557 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
19559 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
19560 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19561 (arm_lookup_simd_builtin_type): Likewise.
19562 (arm_simd_builtin_type): Likewise.
19563 (arm_init_simd_builtin_types): Likewise.
19564 (arm_expand_builtin_args): Likewise.
19565 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
19566 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
19567 (ft32_setup_incoming_varargs): Likewise.
19568 (ft32_function_arg): Likewise.
19569 (ft32_function_arg_advance): Likewise.
19570 (ft32_pass_by_reference): Likewise.
19571 (ft32_arg_partial_bytes): Likewise.
19572 (ft32_valid_pointer_mode): Likewise.
19573 (ft32_addr_space_pointer_mode): Likewise.
19574 (ft32_addr_space_legitimate_address_p): Likewise.
19575 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19577 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19578 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19579 (iamcu_alignment): Likewise.
19580 (canonicalize_vector_int_perm): Likewise.
19581 (ix86_noce_conversion_profitable_p): Likewise.
19582 (ix86_mpx_bound_mode): Likewise.
19583 (ix86_operands_ok_for_move_multiple): Likewise.
19584 * config/microblaze/microblaze-protos.h
19585 (microblaze_expand_conditional_branch_reg): Likewise.
19586 * config/microblaze/microblaze.c
19587 (microblaze_expand_conditional_branch_reg): Likewise.
19588 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19590 (rs6000_reassociation_width): Likewise.
19591 (rs6000_invalid_binary_op): Likewise.
19592 (fusion_p9_p): Likewise.
19593 (emit_fusion_p9_load): Likewise.
19594 (emit_fusion_p9_store): Likewise.
19595 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19597 (riscv_hard_regno_mode_ok_p): Likewise.
19598 (riscv_address_insns): Likewise.
19599 (riscv_split_symbol): Likewise.
19600 (riscv_legitimize_move): Likewise.
19601 (riscv_function_value): Likewise.
19602 (riscv_hard_regno_nregs): Likewise.
19603 (riscv_expand_builtin): Likewise.
19604 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19605 (riscv_build_integer): Likewise.
19606 (riscv_split_integer): Likewise.
19607 (riscv_legitimate_constant_p): Likewise.
19608 (riscv_cannot_force_const_mem): Likewise.
19609 (riscv_regno_mode_ok_for_base_p): Likewise.
19610 (riscv_valid_base_register_p): Likewise.
19611 (riscv_valid_offset_p): Likewise.
19612 (riscv_valid_lo_sum_p): Likewise.
19613 (riscv_classify_address): Likewise.
19614 (riscv_legitimate_address_p): Likewise.
19615 (riscv_address_insns): Likewise.
19616 (riscv_load_store_insns): Likewise.
19617 (riscv_force_binary): Likewise.
19618 (riscv_split_symbol): Likewise.
19619 (riscv_force_address): Likewise.
19620 (riscv_legitimize_address): Likewise.
19621 (riscv_move_integer): Likewise.
19622 (riscv_legitimize_const_move): Likewise.
19623 (riscv_legitimize_move): Likewise.
19624 (riscv_address_cost): Likewise.
19625 (riscv_subword): Likewise.
19626 (riscv_output_move): Likewise.
19627 (riscv_canonicalize_int_order_test): Likewise.
19628 (riscv_emit_int_order_test): Likewise.
19629 (riscv_function_arg_boundary): Likewise.
19630 (riscv_pass_mode_in_fpr_p): Likewise.
19631 (riscv_pass_fpr_single): Likewise.
19632 (riscv_pass_fpr_pair): Likewise.
19633 (riscv_get_arg_info): Likewise.
19634 (riscv_function_arg): Likewise.
19635 (riscv_function_arg_advance): Likewise.
19636 (riscv_arg_partial_bytes): Likewise.
19637 (riscv_function_value): Likewise.
19638 (riscv_pass_by_reference): Likewise.
19639 (riscv_setup_incoming_varargs): Likewise.
19640 (riscv_print_operand): Likewise.
19641 (riscv_elf_select_rtx_section): Likewise.
19642 (riscv_save_restore_reg): Likewise.
19643 (riscv_for_each_saved_reg): Likewise.
19644 (riscv_register_move_cost): Likewise.
19645 (riscv_hard_regno_mode_ok_p): Likewise.
19646 (riscv_hard_regno_nregs): Likewise.
19647 (riscv_class_max_nregs): Likewise.
19648 (riscv_memory_move_cost): Likewise.
19649 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19650 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19651 (rl78_addr_space_address_mode): Likewise.
19652 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19654 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19655 (rs6000_reassociation_width): Likewise.
19656 (rs6000_invalid_binary_op): Likewise.
19657 (fusion_p9_p): Likewise.
19658 (emit_fusion_p9_load): Likewise.
19659 (emit_fusion_p9_store): Likewise.
19660 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19661 (ok_for_simple_move_operands): Likewise.
19662 (ok_for_simple_move_strict_operands): Likewise.
19663 (ok_for_simple_arith_logic_operands): Likewise.
19664 (visium_legitimize_reload_address): Likewise.
19665 (visium_select_cc_mode): Likewise.
19666 (output_cbranch): Likewise.
19667 (visium_split_double_move): Likewise.
19668 (visium_expand_copysign): Likewise.
19669 (visium_expand_int_cstore): Likewise.
19670 (visium_expand_fp_cstore): Likewise.
19671 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19672 (visium_function_arg): Likewise.
19673 (visium_function_arg_advance): Likewise.
19674 (visium_libcall_value): Likewise.
19675 (visium_setup_incoming_varargs): Likewise.
19676 (visium_legitimate_constant_p): Likewise.
19677 (visium_legitimate_address_p): Likewise.
19678 (visium_legitimize_address): Likewise.
19679 (visium_secondary_reload): Likewise.
19680 (visium_register_move_cost): Likewise.
19681 (visium_memory_move_cost): Likewise.
19682 (prepare_move_operands): Likewise.
19683 (ok_for_simple_move_operands): Likewise.
19684 (ok_for_simple_move_strict_operands): Likewise.
19685 (ok_for_simple_arith_logic_operands): Likewise.
19686 (visium_function_value_1): Likewise.
19687 (rtx_ok_for_offset_p): Likewise.
19688 (visium_legitimize_reload_address): Likewise.
19689 (visium_split_double_move): Likewise.
19690 (visium_expand_copysign): Likewise.
19691 (visium_expand_int_cstore): Likewise.
19692 (visium_expand_fp_cstore): Likewise.
19693 (visium_split_cstore): Likewise.
19694 (visium_select_cc_mode): Likewise.
19695 (visium_split_cbranch): Likewise.
19696 (output_cbranch): Likewise.
19697 (visium_print_operand_address): Likewise.
19698 * expmed.c (flip_storage_order): Likewise.
19699 * expmed.h (emit_cstore): Likewise.
19700 (flip_storage_order): Likewise.
19701 * genrecog.c (validate_pattern): Likewise.
19702 * hsa-gen.c (gen_hsa_addr): Likewise.
19703 * internal-fn.c (expand_arith_overflow): Likewise.
19704 * ira-color.c (allocno_copy_cost_saving): Likewise.
19705 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19706 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19707 (process_invariant_for_inheritance): Likewise.
19708 * lra-eliminations.c (move_plus_up): Likewise.
19709 * omp-low.c (lower_oacc_reductions): Likewise.
19710 * simplify-rtx.c (simplify_subreg): Likewise.
19711 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19712 (TARGET_CHKP_BOUND_MODE): Likewise..
19713 * targhooks.c (default_chkp_bound_mode): Likewise.
19714 (default_setup_incoming_vararg_bounds): Likewise.
19715 * targhooks.h (default_chkp_bound_mode): Likewise.
19716 (default_setup_incoming_vararg_bounds): Likewise.
19717 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19718 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19719 (have_whole_vector_shift): Likewise.
19720 * tree-vect-stmts.c (vectorizable_load): Likewise.
19721 * doc/tm.texi: Regenerate.
19723 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19725 Graceful degrade if Binutils PR21472 is not available.
19728 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19729 .rodata in flash test fails.
19730 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19731 * confgure: Regenerate.
19732 * config.in: Regenerate.
19733 * config/avr/avr.c (avr_asm_named_section)
19734 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19735 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19736 (avr_asm_init_sections): Same.
19738 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19740 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19741 (fma<VH:mode>4_intrinsic): Likewise.
19742 (*fmsub<VCVTF:mode>4): Likewise.
19743 (*fmsub<VH:mode>4_intrinsic): Likewise.
19745 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19748 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19749 Don't depend on "optimize > 0".
19750 (out_movhi_r_mr, out_movqi_mr_r): Same.
19751 (out_movhi_mr_r, out_movqi_r_mr): Same.
19752 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19753 io_address_operand on "optimize > 0".
19755 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19757 * tree-loop-distribution.c: Add general explanantion on the pass.
19758 (generate_loops_for_partition): Mark distributed loop.
19759 (pg_add_dependence_edges): New parameter. Handle alias data
19760 dependence specially and record it in the parameter if asked.
19761 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19762 (init_partition_graph_vertices, add_partition_graph_edge): New.
19763 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19764 (free_partition_graph_vdata, build_partition_graph): New.
19765 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19766 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19767 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19768 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19769 (version_for_distribution_p, finalize_partitions): New.
19770 (distribute_loop): Handle alias data dependence specially. Factor
19771 out loop fusion code as functions and call these functions.
19773 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19775 * tree-loop-distribution.c (classify_partition): New parameter and
19776 better handle reduction statement.
19777 (rdg_build_partitions): Revise comment.
19778 (distribute_loop): Compute statements in all partitions and pass it
19779 to classify_partition.
19781 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19783 * tree-loop-distribution.c (enum partition_type): New.
19784 (struct partition): New field type.
19785 (partition_merge_into): Add parameter. Update partition type.
19786 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19787 (build_rdg_partition_for_vertex): Compute partition type.
19788 (rdg_build_partitions): Dump partition type.
19789 (distribute_loop): Update calls to partition_merge_into.
19791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19793 * tree-loop-distribution.c (struct ddr_hasher): New.
19794 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19796 (classify_partition): Call get_data_dependence.
19797 (pg_add_dependence_edges): Ditto.
19798 (distribute_loop): Release data dependence hash table.
19800 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19802 * tree-loop-distribution.c (ref_base_address): Delete.
19803 (similar_memory_accesses): Rename ...
19804 (share_memory_accesses): ... to this. Check if partitions access
19805 the same memory reference.
19806 (distribute_loop): Call share_memory_accesses.
19808 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19810 * tree-loop-distribution.c (struct partition): New field recording
19811 its data reference.
19812 (partition_alloc, partition_free): Init and release data refs.
19813 (partition_merge_into): Merge data refs.
19814 (build_rdg_partition_for_vertex): Collect data refs for partition.
19815 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19817 (distribute_loop): Remve data refs from vertice data of partition
19820 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19822 * tree-loop-distribution.c (params.h): Include header file.
19823 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19824 (datarefs_vec): New global var.
19825 (create_rdg_vertices): Use datarefs_vec directly.
19826 (free_rdg): Don't free data references.
19827 (build_rdg): Update use. Don't free data references.
19828 (distribute_loop): Compute global variable for data references.
19829 Bail out if there are too many data references.
19831 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19833 * tree-loop-distribution.c (loop_nest): New global var.
19834 (build_rdg): Use loop directly, rather than loop nest.
19835 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19837 (distribute_loop): Compute global variable loop nest. Update use.
19839 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19841 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19842 (partition_merge_into): New parameter. Dump reason for fusion.
19843 (distribute_loop): Update use of partition_merge_into.
19845 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19847 * tree-loop-distribution.c (bb_top_order_index): New.
19848 (bb_top_order_index_size, bb_top_order_cmp): New.
19849 (stmts_from_loop): Use topological order.
19850 (pass_loop_distribution::execute): Compute and release topological
19851 order for basic blocks.
19853 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19855 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19858 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19860 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19861 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19862 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19863 * internal-fn.def (LOOP_DIST_ALIAS): New.
19864 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19865 (fold_loop_internal_call): ... this.
19866 (vect_loop_dist_alias_call): New function.
19867 (set_uid_loop_bbs): Call fold_loop_internal_call.
19868 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19871 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19874 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19875 Require dead FLAGS_REG at the beginning of a peephole.
19877 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19880 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19881 arguments in the call to __builtin_ia32_sbb_u32.
19882 (_subborrow_u64): Swap _X and _Y arguments in the call to
19883 __builtin_ia32_sbb_u64.
19885 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19888 * tree-vrp.c (compare_assert_loc): Turn into a function template
19889 with stable template parameter. Only test if a->e is NULL,
19890 !a->e == !b->e has been verified already. Use e == NULL or
19891 e != NULL instead of e or ! e tests. If stable is true, don't use
19892 iterative_hash_expr, on the other side allow a or b or both NULL
19893 and sort the NULLs last.
19894 (process_assert_insertions): Sort using compare_assert_loc<false>
19895 instead of compare_assert_loc, later sort using
19896 compare_assert_loc<true> before calling process_assert_insertions_for
19897 in a loop. Use break instead of continue once seen NULL pointer.
19899 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19901 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19902 Cortex-R7 and Cortex-R8 processors.
19904 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19906 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19907 uninitialized while src is not.
19909 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19911 * common/config/arm/arm-common.c: Adjust include path for
19913 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19914 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19915 (arm-cpu-data.h): Likewise.
19916 (arm-cpu-cdata.h): Likewise.
19917 * config/arm/arm-cpu.h: Delete.
19918 * config/arm/arm-cpu-cdata.h: Delete.
19919 * config/arm/arm-cpu-data.h: Delete.
19921 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19923 * config/arm/arm-cpus.in (cortex-a55): New.
19924 (cortex-a75): Likewise.
19925 (cortex-a75.cortex-a55): Likewise.
19926 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19928 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19929 * config/arm/arm-cpu-cdata.h: Regenerate.
19930 * config/arm/arm-cpu-data.h: Regenerate.
19931 * config/arm/arm-cpu.h: Regenerate.
19932 * config/arm/arm-tables.opt: Regenerate.
19933 * config/arm/arm-tune.md: Regenerate.
19935 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19937 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19939 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19941 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19944 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19946 PR tree-optimization/81292
19947 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19948 full_string_p, also call adjust_related_strinfos if the adjustment
19949 is simple, otherwise invalidate related strinfos.
19951 2017-07-04 Martin Liska <mliska@suse.cz>
19954 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19955 newly created variable as DECL_IGNORED_P.
19957 2017-07-04 Martin Liska <mliska@suse.cz>
19960 * ipa-inline.c (inline_small_functions):
19961 Use xstrdup_for_dump.
19963 2017-07-04 Tom de Vries <tom@codesourcery.com>
19965 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19967 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19970 * config/darwin.c (darwin_function_switched_text_sections):
19973 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19975 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19977 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19979 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19981 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19983 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19984 min_profitable_iters, and th as inclusive lower bounds.
19985 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19986 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19987 for min_profitable_iters and min_profitable_estimate.
19988 (vect_transform_loop): Treat th as an inclusive lower bound.
19989 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19991 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19994 * config/darwin.c (darwin_function_switched_text_sections):
19995 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19996 in two pieces, and suppress the use of buf.
19998 2017-07-03 Nathan Sidwell <nathan@acm.org>
20000 * hash-table.h (hash_table_mod1): Fix indentation.
20002 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20004 PR middle-end/81290
20005 * predict.c (force_edge_cold): Be more careful about propagation
20007 * profile-count.h (profile_probability::guessed,
20008 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20010 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20012 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20014 * doc/invoke.texi (rcpc architecture extension): Document it.
20016 2017-07-03 Richard Biener <rguenther@suse.de>
20018 PR tree-optimization/60510
20019 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20020 the scalar reduction PHI and use it.
20021 (vectorizable_reduction): Properly guard the single_defuse_cycle
20022 path for non-SLP reduction chains where we cannot use it.
20023 Rework reduc_def/index and vector type deduction. Rework
20024 vector operand gathering during reduction op code-gen.
20025 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20026 chains dissolve the chain and leave it to non-SLP reduction
20029 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20031 * tree-data-ref.h (dr_alignment): Declare.
20032 * tree-data-ref.c (dr_alignment): New function.
20033 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20034 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20036 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20038 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20040 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20041 and base_misalignment fields.
20042 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20043 * tree-data-ref.c: Include builtins.h.
20044 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20045 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20046 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20047 * tree-vect-data-refs.c: Include tree-cfg.h.
20048 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20049 fields instead of calculating an alignment here.
20050 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20051 innermost_loop_behavior fields.
20053 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20055 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20057 (DR_STEP_ALIGNMENT): New macro.
20058 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20059 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20060 (create_data_ref): Print it.
20061 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20062 to tell whether the step preserves vector (mis)alignment.
20063 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20064 Move the check for an integer step and generalise to all INTEGER_CST.
20065 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20066 Print the outer step alignment.
20068 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20070 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20071 with offset_alignment.
20072 (DR_ALIGNED_TO): Delete.
20073 (DR_OFFSET_ALIGNMENT): New macro.
20074 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20075 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20076 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20077 (create_data_ref): Likewise.
20078 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20079 (vect_analyze_data_refs): Likewise.
20080 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20081 creating dummy innermost behavior.
20083 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20085 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20086 with a "innermost_loop_behavior *" and refeence tree.
20087 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20088 (create_data_ref): Update call accordingly.
20089 * tree-predcom.c (find_looparound_phi): Likewise.
20091 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20093 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20094 fields with dr_wrt_vec_loop.
20095 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20096 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20097 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20098 (vect_dr_behavior): New function.
20099 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20100 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20101 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20102 track whether the step preserves the misalignment.
20103 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20104 Use vect_dr_behavior.
20105 (vect_setup_realignment): Update call accordingly.
20106 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20107 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20108 call to vect_create_addr_base_for_vector_ref.
20109 (vect_create_cond_for_align_checks): Likewise.
20110 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20111 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20112 (vect_recog_mask_conversion_pattern): Likewise.
20113 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20114 (new_stmt_vec_info): Remove redundant zeroing.
20116 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20118 * common/config/arm/arm-common.c (arm_be8_option): New function.
20119 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20120 (ISA_ARMv6): Add isa_bit_be8.
20121 * config/arm/arm.h (arm_be8_option): Add prototype.
20122 (BE8_SPEC_FUNCTION): New define.
20123 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20124 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20125 (mlittle-endian): Similarly.
20126 (mbe8, mbe32): New options.
20127 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20128 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20130 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20132 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20134 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20136 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
20137 (cleanup_tree_cfg_bb): Use it.
20138 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
20140 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
20142 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20145 * loop-doloop.c (add_test): Update profile.
20147 2017-07-03 Martin Liska <mliska@suse.cz>
20150 * sanopt.c (rewrite_usage_of_param): New function.
20151 (sanitize_rewrite_addressable_params): Likewise.
20152 (pass_sanopt::execute): Call rewrite_usage_of_param.
20154 2017-07-03 Richard Biener <rguenther@suse.de>
20156 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
20157 back to using VIEW_CONVERT_EXPR.
20159 2017-07-03 Martin Liska <mliska@suse.cz>
20162 * doc/extend.texi: Document when a resolver function is
20163 generated for target_clones.
20165 2017-07-03 Martin Liska <mliska@suse.cz>
20167 * asan.c (asan_emit_stack_protection): Unpoison just red zones
20168 and shadow memory of auto variables which are subject of
20169 use-after-scope sanitization.
20170 (asan_expand_mark_ifn): Add do set only when is_poison.
20172 2016-07-03 Richard Biener <rguenther@suse.de>
20174 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
20176 (vect_force_simple_reduction): Record reduction def -> phi mapping.
20177 (vectorizable_reduction): Perform reduction PHI creation when
20178 visiting a reduction PHI and adjust and simplify code generation
20179 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
20180 (vect_transform_loop): Visit reduction PHIs.
20181 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
20182 defs into the SLP tree.
20183 (vect_build_slp_tree): Reduction defs terminate the recursion.
20184 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
20186 (vect_get_vec_defs_for_stmt_copy): Export.
20187 (vect_get_vec_defs): Likewise.
20188 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
20190 (vect_get_vec_defs_for_stmt_copy): Declare.
20191 (vect_get_vec_defs): Likewise.
20193 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20195 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
20196 parameter with a "loop" parameter and use it instead of the
20197 loop containing DR_STMT. Don't check simple_iv when doing
20198 BB analysis. Describe the two analysis modes in the comment.
20200 2017-07-03 Tom de Vries <tom@codesourcery.com>
20202 PR tree-optimization/69468
20203 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
20204 (find_same_succ_bb): Handle ignore_edge_flags.
20206 2017-07-03 Tom de Vries <tom@codesourcery.com>
20208 PR tree-optimization/81192
20209 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
20211 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
20213 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
20215 2017-07-03 Tom de Vries <tom@codesourcery.com>
20217 PR tree-optimization/81192
20218 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
20219 BB_SAME_SUCC (bb) == NULL.
20221 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20223 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
20226 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20228 * dumpfile.c: Include profile-count.h
20229 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
20231 (insert_cond_bb): Update profile.
20232 * tree-cfg.h (insert_cond_bb): Update prototype.
20233 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20234 * tree-dump.c: Do not include tree-cfg.
20236 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20238 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20240 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20242 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20245 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20247 * tree-complex.c (expand_complex_div_wide): update profile.
20249 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20250 Alan Hayward <alan.hayward@arm.com>
20251 David Sherwood <david.sherwood@arm.com>
20253 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20254 (CORETYPES_H): New define.
20255 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20256 (insn-modes-inline.h, s-modes-inline-h): New rules.
20257 (generated_files): Add insn-modes-inline.h.
20258 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20259 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20260 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20261 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20262 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20263 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20264 (build/gencodes.o, build/genconditions.o): Likewise.
20265 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20266 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20267 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20268 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20269 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20270 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20271 * coretypes.h: Include everything up to real.h for generators.
20272 Include insn-modes.h first. Include wide-int-print.h after
20273 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20274 * machmode.h: Don't include insn-modes.h here.
20275 * function-tests.c: Remove includes of signop.h, machmode.h,
20276 double-int.h and wide-int.h.
20278 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20280 * optc-save-gen.awk: Likewise.
20281 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20282 * godump.c: Remove include of wide-int-print.h.
20283 * pretty-print.h: Likewise.
20284 * wide-int-print.cc: Likewise.
20285 * wide-int.cc: Likewise.
20286 * hash-map-tests.c: Remove include of signop.h.
20287 * hash-set-tests.c: Likewise.
20288 * rtl-tests.c: Likewise.
20289 * mkconfig.sh: Remove include of machmode.h.
20290 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20292 (emit_insn_modes_inline_h): ...this new function. Emit the code
20293 into an insn-modes-inline.h header file, adding appropriate
20294 include guards and end comments.
20295 (emit_insn_modes_c_header): Remove include of machmode.h.
20296 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20298 (main): Handle -i flag and call emit_insn_modes_inline_h when
20301 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20303 * tree-ssa-strlen.c (strinfo): Rename the length field to
20304 nonzero_chars. Add a full_string_p field.
20305 (compare_nonzero_chars, zero_length_string_p): New functions.
20306 (get_addr_stridx): Add an offset_out parameter.
20307 Use compare_nonzero_chars.
20308 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20309 (new_strinfo): Update after above changes to strinfo.
20310 (set_endptr_and_length): Set full_string_p.
20311 (get_string_length): Update after above changes to strinfo.
20312 (unshare_strinfo): Update call to new_strinfo.
20313 (maybe_invalidate): Likewise.
20314 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20315 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20316 as a uhwi instead of an shwi. Update after above changes to
20317 strinfo and new_strinfo.
20318 (zero_length_string): Assert that chainsi contains full strings.
20319 Use zero_length_string_p. Update call to new_strinfo.
20320 (adjust_related_strinfos): Update after above changes to strinfo.
20321 Copy full_string_p from origsi.
20322 (adjust_last_stmt): Use zero_length_string_p.
20323 (handle_builtin_strlen): Update after above changes to strinfo and
20324 new_strinfo. Install the lhs as the string length if the previous
20325 entry didn't describe a full string.
20326 (handle_builtin_strchr): Update after above changes to strinfo
20328 (handle_builtin_strcpy): Likewise.
20329 (handle_builtin_strcat): Likewise.
20330 (handle_builtin_malloc): Likewise.
20331 (handle_pointer_plus): Likewise.
20332 (handle_builtin_memcpy): Likewise. Track nonzero characters
20333 that aren't necessarily followed by a nul terminator.
20334 (handle_char_store): Likewise.
20336 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20338 PR tree-optimization/80769
20339 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20340 for malloc and calloc. Document the new invariant that all related
20341 strinfos have delayed lengths or none do.
20342 (verify_related_strinfos): Move earlier in file.
20343 (set_endptr_and_length): New function, split out from...
20344 (get_string_length): ...here. Also set the lengths of related
20346 (zero_length_string): Assert that chainsi has known (rather than
20348 (adjust_related_strinfos): Likewise.
20350 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20352 PR tree-optimization/81136
20353 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20354 assert that two references with the same misalignment have the same
20355 compile-time misalignment if those compile-time misalignments
20358 2017-07-01 Andi Kleen <ak@linux.intel.com>
20360 * print-tree.c (print_node): Print all attributes.
20362 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20364 * cfg.c (scale_bbs_frequencies): New function.
20365 * cfg.h (scale_bbs_frequencies): Declare it.
20366 * cfgloopanal.c (single_likely_exit): Cleanup.
20367 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20369 (scale_loop_profile): Likewise.
20370 (loop_version): Likewise.
20371 (create_empty_loop_on_edge): Update.
20372 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20373 scale_loop_frequencies, scale_loop_profile, loopify,
20374 loop_version): Update prototypes.
20375 * modulo-sched.c (sms_schedule): Update.
20376 * predict.c (unlikely_executed_edge_p): Also check probability.
20377 (probably_never_executed_edge_p): Fix typo.
20378 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20379 * tree-parloops.c (gen_parallel_loop): Update.
20380 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20381 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20382 * tree-ssa-loop-split.c (split_loop): Update.
20383 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20384 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20385 (vect_loop_versioning): Update.
20386 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20388 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20390 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20392 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20394 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20395 to keep profile consistent.
20397 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20399 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20400 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20401 * profile-count.h (max_safe_multiplier): Make unsigned.
20402 (profile_count::guessed_zero): New.
20404 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20406 * bb-reorder.c (fix_up_crossing_landing_pad,
20407 fix_crossing_conditional_branches): Use make_single_succ_edge
20408 to keep profile consistent.
20410 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20412 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20415 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20418 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20419 the right scopes, make sure cond_jump isn't preserved between multiple
20420 iterations. Search for fallthru edge whenever there are 3+ edges and
20421 use find_fallthru_edge for it.
20423 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20425 Patch by Alexander Monakov <amonakov@ispras.ru>
20426 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20427 probabilities consistently.
20429 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20431 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20432 emit_cmp_and_jump_insns.
20434 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20437 * tree-inline.c (expand_call_inline): Combine profile statuses.
20439 2017-06-30 Andrew Pinski <apinski@cavium.com>
20441 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
20442 fold_stmt returned true.
20444 2017-06-30 Nathan Sidwell <nathan@acm.org>
20446 * ggc.h (empty_string): Delete.
20447 * cfgexpand.c (expand_asm_stmt): Use plain "".
20448 * optabs.c (expand_asm_memory_barrier): Likewise.
20449 * stringpool.c (empty_string): Delete.
20450 (digit_vector, digit_string): Delete.
20451 (ggc_alloc_string): Use plain "", don't optimize single digit
20452 strings. Use ggc_alloc_atomic.
20454 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
20456 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
20457 comparison set and one other set, use the cost of the non-comparison
20460 2017-06-30 Nathan Sidwell <nathan@acm.org>
20462 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
20465 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
20467 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
20468 loops. Remove now unneeded calls to gimple_switch_set_label() that
20469 just set removed labels to NULL_TREE.
20471 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
20473 * tree-ssanames.c (set_range_info_raw): Abstract from ...
20474 (set_range_info): ...here. Only call set_range_info_raw if domain
20476 (set_nonzero_bits): Call set_range_info_raw.
20477 * tree-ssanames.h (set_range_info_raw): New.
20479 2017-06-30 Jakub Jelinek <jakub@redhat.com>
20482 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
20483 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
20484 of nonimmediate_operand and <store_mask_constraint> instead of m
20485 for the input operand. For V8FI iterator, always split if input
20486 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
20487 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
20488 <store_mask_predicate> instead of register_operand and
20489 <store_mask_constraint> instead of v for the input operand. Make
20490 sure both operands aren't MEMs for if not <mask_applied>.
20492 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
20494 * lto-wrapper.c (copy_file) Close both file descriptors before
20497 2017-06-30 Martin Liska <mliska@suse.cz>
20500 * multiple_target.c (create_dispatcher_calls): Make ifunc
20501 also for function that don't have calls or are not referenced.
20503 2017-06-30 Richard Biener <rguenther@suse.de>
20505 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
20506 analyze the first scalar stmt. Move vector type computation
20507 for the BB case here from ...
20508 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
20509 live operation processing in the SLP case properly.
20511 2017-06-30 Richard Biener <rguenther@suse.de>
20513 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
20515 2017-06-30 Martin Liska <mliska@suse.cz>
20518 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
20519 before BUILT_IN_UNWIND_RESUME when ASAN is used.
20521 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
20523 * doc/invoke.texi (AArch64): Add missing options and remove redundant
20526 2017-06-30 Richard Biener <rguenther@suse.de>
20528 PR tree-optimization/81249
20529 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
20530 condition reduction result to original scalar type.
20532 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20534 * profile-count.h (enum profile_quality): Fix typos and whitespace
20537 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20539 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
20540 type for branch probabilities.
20542 2017-06-29 Julian Brown <julian@codesourcery.com>
20543 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20545 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
20546 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
20547 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
20548 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
20550 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20552 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
20553 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
20554 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
20555 CC usage from generic code to here.
20556 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
20557 CC usage into the target macros.
20559 2017-06-29 Maya Rashish <coypu@sdf.org>
20561 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
20564 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20566 * arm/arm-builtins.c: Include profile-count.h
20567 * except.c (sjlj_emit_function_enter): Use
20568 profile_probability::unlikely.
20570 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20572 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20573 and tocrel_offset be pointer args rather than implicitly using
20575 (legitimate_constant_pool_address_p, rs6000_emit_move,
20576 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20577 tocrel_offset and use in toc_relative_expr_p call.
20578 (print_operand, print_operand_address): Use static tocrel_base_oac
20579 and tocrel_offset_oac.
20580 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20583 2017-06-29 Maya Rashish <coypu@sdf.org>
20585 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20587 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20589 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20590 objects, take into account only the alignment of 'op0' and 'mode1' if
20593 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20595 * ccmp.c (ccmp_tree_comparison_p): New function.
20596 (ccmp_candidate_p): Update to use above function.
20597 (get_compare_parts): New function.
20598 (expand_ccmp_next): Update to use new functions.
20599 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20601 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20602 take mode as argument.
20603 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20604 * expr.c (expand_expr_real_1): Pass mode as argument.
20606 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20608 * combine.c (combine_instructions): Print insns to dump_file, together
20611 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20613 * asan.c (asan_emit_stack_protection): Update.
20614 (create_cond_insert_point): Update.
20615 * auto-profile.c (afdo_propagate_circuit): Update.
20616 * basic-block.h (struct edge_def): Turn probability to
20617 profile_probability.
20618 (EDGE_FREQUENCY): Update.
20619 * bb-reorder.c (find_traces_1_round): Update.
20620 (better_edge_p): Update.
20621 (sanitize_hot_paths): Update.
20622 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20623 (make_single_succ_edge): Update.
20624 (check_bb_profile): Update.
20625 (dump_edge_info): Update.
20626 (update_bb_profile_for_threading): Update.
20627 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20628 probabilitycount to 0.
20629 * cfgbuild.c (compute_outgoing_frequencies): Update.
20630 * cfgcleanup.c (try_forward_edges): Update.
20631 (outgoing_edges_match): Update.
20632 (try_crossjump_to_edge): Update.
20633 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20634 (expand_gimple_tailcall): Update.
20635 (construct_init_block): Use make_single_succ_edge.
20636 (construct_exit_block): Use make_single_succ_edge.
20637 * cfghooks.c (verify_flow_info): Update.
20638 (redirect_edge_succ_nodup): Update.
20639 (split_edge): Update.
20640 (account_profile_record): Update.
20641 * cfgloopanal.c (single_likely_exit): Update.
20642 * cfgloopmanip.c (scale_loop_profile): Update.
20643 (set_zero_probability): Remove.
20644 (duplicate_loop_to_header_edge): Update.
20645 * cfgloopmanip.h (loop_version): Update prototype.
20646 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20647 (force_nonfallthru_and_redirect): Update.
20648 (update_br_prob_note): Update.
20649 (rtl_verify_edges): Update.
20650 (purge_dead_edges): Update.
20651 (rtl_lv_add_condition_to_bb): Update.
20652 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20653 * cgraphunit.c (init_lowered_empty_function): Update.
20654 (cgraph_node::expand_thunk): Update.
20655 * cilk-common.c: Include profile-count.h
20656 * dojump.c (inv): Remove.
20657 (jumpifnot): Update.
20658 (jumpifnot_1): Update.
20659 (do_jump_1): Update.
20661 (do_jump_by_parts_greater_rtx): Update.
20662 (do_compare_rtx_and_jump): Update.
20663 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20664 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20665 * dwarf2cfi.c: Include profile-count.h
20666 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20667 (sjlj_emit_dispatch_table): Likewise.
20668 * explow.c: Include profile-count.h
20669 * expmed.c (emit_store_flag_force): Update.
20670 (do_cmp_and_jump): Update.
20671 * expr.c (compare_by_pieces_d::generate): Update.
20672 (compare_by_pieces_d::finish_mode): Update.
20673 (emit_block_move_via_loop): Update.
20674 (store_expr_with_bounds): Update.
20675 (store_constructor): Update.
20676 (expand_expr_real_2): Update.
20677 (expand_expr_real_1): Update.
20678 * expr.h (try_casesi, try_tablejump): Update prototypes.
20679 * gimple-pretty-print.c (dump_probability): Update.
20680 (dump_profile): New.
20681 (dump_gimple_label): Update.
20682 (dump_gimple_bb_header): Update.
20683 * graph.c (draw_cfg_node_succ_edges): Update.
20684 * hsa-gen.c (convert_switch_statements): Update.
20685 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20686 (find_if_case_1): Update.
20687 (find_if_case_2): Update.
20688 * internal-fn.c (expand_arith_overflow_result_store): Update.
20689 (expand_addsub_overflow): Update.
20690 (expand_neg_overflow): Update.
20691 (expand_mul_overflow): Update.
20692 (expand_vector_ubsan_overflow): Update.
20693 * ipa-cp.c (good_cloning_opportunity_p): Update.
20694 * ipa-split.c (split_function): Use make_single_succ_edge.
20695 * ipa-utils.c (ipa_merge_profiles): Update.
20696 * loop-doloop.c (add_test): Update.
20697 (doloop_modify): Update.
20698 * loop-unroll.c (compare_and_jump_seq): Update.
20699 (unroll_loop_runtime_iterations): Update.
20700 * lra-constraints.c (lra_inheritance): Update.
20701 * lto-streamer-in.c (input_cfg): Update.
20702 * lto-streamer-out.c (output_cfg): Update.
20703 * mcf.c (adjust_cfg_counts): Update.
20704 * modulo-sched.c (sms_schedule): Update.
20705 * omp-expand.c (expand_omp_for_init_counts): Update.
20706 (extract_omp_for_update_vars): Update.
20707 (expand_omp_ordered_sink): Update.
20708 (expand_omp_for_ordered_loops): Update.
20709 (expand_omp_for_generic): Update.
20710 (expand_omp_for_static_nochunk): Update.
20711 (expand_omp_for_static_chunk): Update.
20712 (expand_cilk_for): Update.
20713 (expand_omp_simd): Update.
20714 (expand_omp_taskloop_for_outer): Update.
20715 (expand_omp_taskloop_for_inner): Update.
20716 * omp-simd-clone.c (simd_clone_adjust): Update.
20717 * optabs.c (expand_doubleword_shift): Update.
20718 (expand_abs): Update.
20719 (emit_cmp_and_jump_insn_1): Update.
20720 (expand_compare_and_swap_loop): Update.
20721 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20722 * predict.c (predictable_edge_p): Update.
20723 (edge_probability_reliable_p): Update.
20724 (set_even_probabilities): Update.
20725 (combine_predictions_for_insn): Update.
20726 (combine_predictions_for_bb): Update.
20727 (propagate_freq): Update.
20728 (estimate_bb_frequencies): Update.
20729 (force_edge_cold): Update.
20730 * profile-count.c (profile_count::dump): Add missing space into dump.
20731 (profile_count::debug): Add newline.
20732 (profile_count::differs_from_p): Explicitly convert to unsigned.
20733 (profile_count::stream_in): Update.
20734 (profile_probability::dump): New member function.
20735 (profile_probability::debug): New member function.
20736 (profile_probability::differs_from_p): New member function.
20737 (profile_probability::differs_lot_from_p): New member function.
20738 (profile_probability::stream_in): New member function.
20739 (profile_probability::stream_out): New member function.
20740 * profile-count.h (profile_count_quality): Rename to ...
20741 (profile_quality): ... this one.
20742 (profile_probability): New.
20743 (profile_count): Update.
20744 * profile.c (compute_branch_probabilities): Update.
20745 * recog.c (peep2_attempt): Update.
20746 * sched-ebb.c (schedule_ebbs): Update.
20747 * sched-rgn.c (find_single_block_region): Update.
20748 (compute_dom_prob_ps): Update.
20749 (schedule_region): Update.
20750 * sel-sched-ir.c (compute_succs_info): Update.
20751 * stmt.c (struct case_node): Update.
20752 (do_jump_if_equal): Update.
20753 (get_outgoing_edge_probs): Update.
20754 (conditional_probability): Update.
20755 (emit_case_dispatch_table): Update.
20756 (expand_case): Update.
20757 (expand_sjlj_dispatch_table): Update.
20758 (emit_case_nodes): Update.
20759 * targhooks.c: Update.
20760 * tracer.c (better_p): Update.
20761 (find_best_successor): Update.
20762 * trans-mem.c (expand_transaction): Update.
20763 * tree-call-cdce.c: Update.
20764 * tree-cfg.c (gimple_split_edge): Upate.
20765 (move_sese_region_to_fn): Upate.
20766 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20767 * tree-eh.c (lower_resx): Upate.
20768 (cleanup_empty_eh_move_lp): Upate.
20769 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20770 * tree-inline.c (copy_edges_for_bb): Update.
20771 (copy_cfg_body): Update.
20772 * tree-parloops.c (gen_parallel_loop): Update.
20773 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20774 (gimple_gen_time_profiler): Update.
20775 * tree-ssa-dce.c (remove_dead_stmt): Update.
20776 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20777 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20778 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20779 (unloop_loops): Update.
20780 (try_peel_loop): Update.
20781 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20782 * tree-ssa-loop-split.c (connect_loops): Update.
20783 (split_loop): Update.
20784 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20785 (hoist_guard): Update.
20786 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20787 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20788 (value_replacement): Update.
20789 * tree-ssa-reassoc.c (branch_fixup): Update.
20790 * tree-ssa-tail-merge.c (replace_block_by): Update.
20791 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20792 (create_edge_and_update_destination_phis): Update.
20793 (compute_path_counts): Update.
20794 (recompute_probabilities): Update.
20795 (update_joiner_offpath_counts): Update.
20796 (freqs_to_counts_path): Update.
20797 (duplicate_thread_path): Update.
20798 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20799 (struct switch_conv_info): Update.
20800 (gen_inbound_check): Update.
20801 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20802 (vect_do_peeling): Update.
20803 (vect_loop_versioning): Update.
20804 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20805 (optimize_mask_stores): Update.
20806 * ubsan.c (ubsan_expand_null_ifn): Update.
20807 * value-prof.c (gimple_divmod_fixed_value): Update.
20808 (gimple_divmod_fixed_value_transform): Update.
20809 (gimple_mod_pow2): Update.
20810 (gimple_mod_pow2_value_transform): Update.
20811 (gimple_mod_subtract): Update.
20812 (gimple_mod_subtract_transform): Update.
20813 (gimple_ic): Update.
20814 (gimple_stringop_fixed_value): Update.
20815 (gimple_stringops_transform): Update.
20816 * value-prof.h: Update.
20818 2017-06-29 Carl Love <cel@us.ibm.com>
20820 * config/rs6000/rs6000-c.c: Add support for built-in functions
20821 vector signed int vec_signed (vector float);
20822 vector signed long long vec_signed (vector double);
20823 vector signed int vec_signed2 (vector double, vector double);
20824 vector signed int vec_signede (vector double);
20825 vector signed int vec_signedo (vector double);
20826 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20827 instruction generator.
20828 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20829 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20830 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20832 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20833 vunsignede_v2df): Add define_expands.
20834 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20835 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20836 VEC_UNSIGNEDO): Add definitions.
20837 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20838 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20839 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20840 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20841 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20842 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20843 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20844 vec_unsignede, vec_unsignedo): Add builtin defines.
20845 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20847 * doc/extend.texi: Update the built-in documentation file for the
20848 new built-in functions.
20850 2017-06-29 Richard Biener <rguenther@suse.de>
20852 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20853 reduction chains to LOOP_VINFO_REDUCTIONS.
20854 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20855 SLP reductions after processing reduction chains.
20857 2017-06-29 Nathan Sidwell <nathan@acm.org>
20859 * builtins.c (fold_builtin_FUNCTION): Use
20860 lang_hooks.decl_printable_name.
20862 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20864 PR middle-end/81194
20865 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20866 with only one label.
20867 * stmt.c (expand_case): Assert NCASES is greater than one.
20869 2017-06-29 Richard Biener <rguenther@suse.de>
20871 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20873 (group_case_labels): Likewise.
20874 (find_taken_edge): Push sanity checking on val to workers...
20875 (find_taken_edge_cond_expr): ... here
20876 (find_taken_edge_switch_expr): ... and here, handle cases
20877 with just a default label.
20878 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20879 (group_case_labels): Likewise.
20880 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20881 group_case_labels does anything cleanup the CFG again.
20883 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20885 PR tree-optimization/81196
20886 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20887 exit condition comparing two IVs.
20889 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20891 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20892 profile to the dummy entry at the end of the list of architectures.
20893 * config/arm/arm-cpu-cdata.h: Regenerated.
20895 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20896 Michael Collison <michael.collison@arm.com>
20899 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20901 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20902 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20903 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20904 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20905 of shift when the shift amount is masked with constant equal to
20906 the size of the mode.
20907 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20910 2017-06-29 Martin Liska <mliska@suse.cz>
20912 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20914 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20917 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20920 (edge_growth_cache_entry::edge_growth_cache_entry): New
20922 (reset_edge_growth_cache): Update to use constructor.
20924 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20926 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20927 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20928 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20930 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20932 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20933 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20935 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20937 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20938 (*-linux-uclibc*): Add t-uclibc tmake_file.
20939 * config/t-musl: New.
20940 * config/t-uclibc: New.
20942 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20944 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20946 (gen_comm_data): Emit architectural setting of arch_prof.
20947 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20949 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20950 (armv8-m.base, armv8-m.main): Likewise.
20951 * arm-protos.h (arm_build_target): Add profile field.
20952 (arch_option): Likewise.
20953 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20955 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20956 arm_active_target.profile.
20958 2017-06-28 Richard Biener <rguenther@suse.de>
20960 PR middle-end/81227
20961 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20962 TYPE_OVERFLOW_WRAPS.
20963 * match.pd (negate_expr_p): Likewise.
20964 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20965 fold_build2, not fold_binary.
20967 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20969 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20970 Convert memory address to Pmode.
20971 (aarch64_print_operand): Assert MEM operands are always Pmode.
20973 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20976 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20977 Remove redundant if.
20978 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20979 * config/arm/aarch-common-protos.h
20980 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20981 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20983 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20986 * multiple_target.c (create_dispatcher_calls): Set the default
20987 clone to be static, not public.
20989 2017-06-28 Richard Biener <rguenther@suse.de>
20991 * tree-vect-loop.c (vectorizable_reduction): Move special
20992 cond reduction IV var creation ...
20993 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20994 parameter. Use STMT_VINFO_VECTYPE.
20995 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20998 2017-06-28 Martin Liska <mliska@suse.cz>
21001 * ipa-visibility.c (non_local_p): Handle visibility.
21003 2017-06-28 Martin Liska <mliska@suse.cz>
21006 * common.opt: Add IntegerRange to various options.
21007 * opt-functions.awk (integer_range_info): New function.
21008 * optc-gen.awk: Add integer_range_info to cl_options struct.
21009 * opts-common.c (decode_cmdline_option): Handle
21010 CL_ERR_INT_RANGE_ARG.
21011 (cmdline_handle_error): Likewise.
21012 * opts.c (print_filtered_help): Show valid interval in
21013 when --help is provided.
21014 * opts.h (struct cl_option): Add range_min and range_max fields.
21015 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21017 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21019 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21020 (x * C EQ/NE y * C): New transformation.
21022 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21024 * genmultilib (combination_space): Accept '+' in option names.
21026 2017-06-28 Martin Liska <mliska@suse.cz>
21029 * asan.c (instrument_derefs): Bail out inner references
21030 that are hard register variables.
21032 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21035 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21036 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21038 2017-06-28 Richard Biener <rguenther@suse.de>
21040 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21041 (vect_get_slp_defs): Adjust.
21042 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21044 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21046 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21047 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21048 (vectorizable_reduction): Adjust.
21049 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21051 (vect_get_slp_defs): Likewise.
21052 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21053 (vectorizable_bswap): Adjust.
21054 (vectorizable_call): Likewise.
21055 (vectorizable_conversion): Likewise.
21056 (vectorizable_assignment): Likewise.
21057 (vectorizable_shift): Likewise.
21058 (vectorizable_operation): Likewise.
21059 (vectorizable_store): Likewise.
21060 (vectorizable_condition): Likewise.
21061 (vectorizable_comparison): Likewise.
21063 2017-06-28 Michael Collison <michael.collison@arm.com>
21066 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21068 (arm_gen_movmemqi): Removed unused variable 'i'.
21069 Convert 'for' loop into 'while' loop.
21070 (arm_expand_prologue): Remove last unnecessary set of insn.
21071 (thumb_pop): Remove unused variable 'pushed_words'.
21072 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21074 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21076 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21077 * config/s390/s390-protos.h: Add prototype of
21078 s390_rel_address_ok_p.
21079 * config/s390/s390.c (s390_got_symbol): New function.
21080 (s390_rel_address_ok_p): New function.
21081 (legitimize_pic_address): Use s390_rel_address_ok_p.
21082 (s390_load_got): Use s390_got_symbol.
21083 (s390_option_override): Issue error if
21084 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21085 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21087 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21089 2017-06-27 Andrew Pinski <apinski@cavium.com>
21091 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21092 (X * copysign (1.0, X)): New pattern.
21093 (X * copysign (1.0, -X)): New pattern.
21094 (copysign (-1.0, CST)): New pattern.
21096 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21098 * genmultilib (combination_space): Remove variable.
21099 Validate reuse rules against regular expression for any sequence
21100 of multilib options in any order.
21102 2017-06-27 Michael Collison <michael.collison@arm.com>
21104 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21105 call aarch64_split_simd_combine.
21106 * (aarch64_combine_internal<mode>): Delete pattern.
21107 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21108 Allow register and subreg operands.
21110 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21112 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21113 specific need, just fallback on defaults.
21114 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21116 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21117 Olivier Hainque <hainque@adacore.com>
21119 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21121 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21122 targets. Pick a default if no particular attempt applied.
21123 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21126 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21128 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21129 (x86_64-wrs-vxworks7): Likewise.
21131 2017-06-27 Marek Polacek <polacek@redhat.com>
21134 * ubsan.c (instrument_null): Check get_base_address's result for null.
21136 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21138 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
21140 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21142 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
21143 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
21144 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
21145 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
21146 New function types.
21147 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
21148 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
21149 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
21150 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
21151 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
21152 BUILT_IN_FEUPDATEENV): New builtins.
21153 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
21154 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
21155 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
21156 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
21158 (builtin_structptr_types): Adjust size.
21159 * tree.c (builtin_structptr_types): Add four entries.
21161 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21162 Olivier Hainque <hainque@adacore.com>
21164 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
21165 (TLS_SYM): New local macro, forcing reference to __tls__ on
21166 link command lines for VxWorks 7 RTPs, triggering initialization
21168 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
21169 OS features TLS support, true for RTPs on VxWorks 7.
21170 * config/vxworks.c (vxworks_override_options): Setup emutls
21173 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21175 * predict.c (test_prediction_value_range): Use -1U instead of -1
21176 to avoid narrowing conversion warning.
21177 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
21178 to avoid narrowing conversion warning.
21179 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
21181 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
21183 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21185 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
21186 64bit configurations.
21187 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
21188 (SIZE_TYPE): Likewise.
21189 * config/vxworks.c (vxworks_emutls_var_fields): Use
21190 long_unsigned_type_node instead of unsigned_type_node as the offset
21191 field type, which is "pointer" mode in emutls.c.
21193 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21196 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
21198 PR middle-end/81207
21199 * gimple-fold.c (replace_call_with_call_and_fold): Handle
21200 gimple_vuse copying separately from gimple_vdef copying.
21202 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21204 * value-prof.c (free_hist): Remove call to memset and the enclosing if
21207 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21208 Olivier Hainque <hainque@adacore.com>
21210 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
21211 for all vxworks7 targets.
21212 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
21213 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
21214 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
21215 variations for VX6/VX7 and 32/64bits later on in ...
21216 (VXWORKS_LIB_SPEC): Leverage new macros.
21217 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
21218 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
21220 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21222 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
21223 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
21225 2017-06-26 Carl Love <cel@us.ibm.com>
21227 * config/rs6000/rs6000-c.c: Add support for built-in functions
21228 vector bool char vec_reve (vector bool char);
21229 vector signed char vec_reve (vector signed char);
21230 vector unsigned char vec_reve (vector unsigned char);
21231 vector bool int vec_reve (vector bool int);
21232 vector signed int vec_reve (vector signed int);
21233 vector unsigned int vec_reve (vector unsigned int);
21234 vector bool long long vec_reve (vector bool long long);
21235 vector signed long long vec_reve (vector signed long long);
21236 vector unsigned long long vec_reve (vector unsigned long long);
21237 vector bool short vec_reve (vector bool short);
21238 vector signed short vec_reve (vector signed short);
21239 vector double vec_reve (vector double);
21240 vector float vec_reve (vector float);
21241 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21242 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21243 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21244 (altivec_vreve): New pattern.
21245 * config/rs6000/altivec.h (vec_reve): New define.
21246 * doc/extend.texi (vec_rev): Update the built-in documentation file
21247 for the new built-in functions.
21249 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21251 PR tree-optimization/71815
21252 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21254 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21256 (slsr_process_phi): Likewise.
21257 (replace_uncond_cands_and_profitable_phis): Don't replace a
21258 multiply candidate with a stride of 1 (copy or cast).
21259 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21261 (lowest_cost_path): Likewise.
21262 (total_savings): Likewise.
21264 2017-06-26 Richard Biener <rguenther@suse.de>
21267 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21268 Use def_builtin_pure for all gather builtins.
21270 2017-06-26 Richard Biener <rguenther@suse.de>
21272 PR tree-optimization/81203
21273 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21274 non-dominating BBs.
21276 2017-06-26 Marek Polacek <polacek@redhat.com>
21279 * doc/invoke.texi: Document -Wmultistatement-macros.
21281 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21283 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21284 arm_neon_ok_no_float_abi effective target.
21286 2017-06-26 Richard Biener <rguenther@suse.de>
21288 PR tree-optimization/80928
21289 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21290 (copy_bbs): Set BB_DUPLICATED flag early.
21291 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21293 (execute_on_shrinking_pred): Likewise.
21294 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21295 BB_DUPLICATED blocks.
21296 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21297 iterate over all PHIs considering removal of *gsi.
21299 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21301 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21304 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21306 * config/rs6000/rs6000-string.c: (expand_block_clear,
21307 do_load_for_compare, select_block_compare_mode,
21308 compute_current_alignment, expand_block_compare,
21309 expand_strncmp_align_check, expand_strn_compare,
21310 expand_block_move, rs6000_output_load_multiple)
21311 Move functions related to string/block move/compare
21312 to a separate file.
21313 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21314 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21315 for this function which is now used in two files.
21316 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21317 * config.gcc: Add rs6000-string.o to extra_objs for
21318 targets powerpc*-*-* and rs6000*-*-*.
21320 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21323 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21324 32-bit, since indexed is not valid for DImode.
21325 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21326 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21327 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21328 (define_peephole2 for Altivec d-form store): Likewise.
21331 * multiple_target.c (create_dispatcher_calls): Only create the
21332 dispatcher call if the function is the default clone of a
21333 versioned function.
21335 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21337 PR middle-end/80902
21338 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21339 a call, force the call to not be a tail call.
21341 2017-06-23 Jeff Law <law@redhat.com>
21343 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21346 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21348 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21349 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21350 handling for early expansion of vector shifts (sl,sr,sra,rl).
21351 (builtin_function_type): Add vector shift right instructions
21352 to the unsigned argument list.
21354 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21356 rtl-optimizatoin/79286
21357 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21358 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21359 trap. PIC register plus a const unspec without offset can never trap.
21361 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21363 * tree.h (builtin_structptr_type): New type.
21364 (builtin_structptr_types): Declare new array.
21365 * tree.c (builtin_structptr_types): New array.
21366 (free_lang_data, build_common_tree_nodes): Use it.
21368 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21371 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21374 2017-06-22 Matt Turner <mattst88@gmail.com>
21376 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21377 Lake models to skylake case. Assume skylake for unknown
21378 models with clflushopt.
21380 2017-06-22 Jeff Law <law@redhat.com>
21382 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21383 frame sizes that do not satisfy aarch64_uimm12_shift.
21385 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21387 * profile-count.h (apply_probability,
21388 apply_scale, probability_in): Fix checks for zero.
21390 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21392 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21393 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21395 2016-06-22 Richard Biener <rguenther@suse.de>
21397 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21398 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21399 REDUC_MAX_EXPR support.
21400 (vectorizable_reduction): Likewise.
21401 (vect_create_epilog_for_reduction): Likewise.
21403 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21405 * match.pd (A / (1 << B) -> A >> B): New.
21406 * generic-match-head.c: Include optabs-tree.h.
21407 * gimple-match-head.c: Likewise.
21408 * optabs-tree.h (target_supports_op_p): New.
21409 * optabs-tree.c (target_supports_op_p): New.
21411 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21413 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21414 $gcc_cv_ld --help output.
21415 (gcc_cv_ld_demangle): Likewise.
21416 (gcc_cv_ld_eh_frame_hdr): Likewise.
21417 (gcc_cv_ld_pie): Likewise.
21418 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21419 (gcc_cv_ld_buildid): Likewise.
21420 (gcc_cv_ld_sysroot): Likewise.
21421 (ld_bndplt_support): Likewise.
21422 (ld_pushpopstate_support): Likewise.
21423 * configure: Regenerate.
21424 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21426 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21429 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21430 operands indexes to avoid gap between operands and match_dups.
21432 2017-06-21 Andrew Pinski <apinski@cavium.com>
21434 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21435 Increment Arith_shift and Arith_shift_reg by 1.
21436 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
21438 * config/aarch64/aarch64.c (thunderx_tunings): Enable
21439 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
21440 (aarch64_strip_extend): Add new argument and test for it.
21441 (aarch64_cheap_mult_shift_p): New function.
21442 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
21443 add a cost if it is true.
21444 Update calls to aarch64_strip_extend.
21445 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
21447 2017-06-21 Andrew Pinski <apinski@cavium.com>
21449 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
21451 (thunderxt88): Likewise.
21452 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
21453 (thunderx_prefetch_tune): New variable.
21454 (thunderx2t99_prefetch_tune): Update for the correct values.
21455 (thunderxt88_tunings): New variable.
21456 (thunderx_tunings): Use thunderx_prefetch_tune instead of
21457 generic_prefetch_tune.
21458 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
21460 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21462 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
21463 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
21464 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
21465 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
21466 (aarch64_atomic_cas<mode>, GPI): Likewise.
21468 2017-06-21 Martin Liska <mliska@suse.cz>
21470 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
21471 statements on cold and hot labels.
21472 * predict.c (tree_estimate_probability_bb): Remove the
21473 prediction from this place.
21475 2017-06-21 Martin Liska <mliska@suse.cz>
21477 PR tree-optimization/79489
21478 * gimplify.c (maybe_add_early_return_predict_stmt): New
21480 (gimplify_return_expr): Call the function.
21481 * predict.c (tree_estimate_probability_bb): Remove handling
21483 * predict.def: Update comment about early return predictor.
21484 * gimple-predict.h (is_gimple_predict): New function.
21485 * predict.def: Change default value of early return to 66.
21486 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
21488 * passes.def: Put pass_strip_predict_hints to the beginning of
21491 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
21493 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
21494 FUNCTION_DECL declarations.
21495 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
21497 (dwaf2out_decl): Likewise.
21498 * godump.c (go_early_global_decl): Skip call to the real debug hook
21499 for FUNCTION_DECL declarations.
21500 * passes.c (rest_of_decl_compilation): Skip call to the
21501 early_global_decl debug hook for FUNCTION_DECL declarations, unless
21502 -fdump-go-spec is passed.
21504 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21506 * config/i386/i386.c (struct builtin_isa): New field pure_p.
21507 Reorder for compactness.
21508 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
21509 (def_builtin_pure, def_builtin_pure2): New functions.
21510 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
21512 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21514 * match.pd (nop_convert): New predicate.
21515 ((A +- CST1) +- CST2): Allow some NOP conversions.
21517 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21520 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
21521 with ctors/dtors if GOVD_SHARED is set.
21523 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21525 * config/aarch64/aarch64.md (movti_aarch64):
21526 Emit mov rather than orr.
21527 (movtf_aarch64): Likewise.
21528 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
21529 Emit mov rather than orr.
21531 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21533 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
21534 Swap alternatives, make integer dup more expensive.
21536 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21538 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21539 Return true for non-tls symbols.
21541 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
21543 * config/aarch64/aarch64-cores.def (cortex-a55): New.
21544 (cortex-a75): Likewise.
21545 (cortex-a75.cortex-a55): Likewise.
21546 * config/aarch64/aarch64-tune.md: Regenerate.
21547 * doc/invoke.texi (-mtune): Document new values for -mtune.
21549 2017-06-21 Tom de Vries <tom@codesourcery.com>
21551 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
21552 stack_size feature.
21553 (Effective-Target Keywords, Other attributes): Suggest using
21554 dg-add-options stack_size feature to get stack limit in stack_size
21555 effective target documentation.
21557 2017-06-21 Julian Brown <julian@codesourcery.com>
21558 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21560 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
21561 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
21562 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
21564 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
21565 attribute type list for neon_multiply.
21566 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
21567 attribute type list for neon_multiply.
21568 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
21569 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
21570 attribute type list for neon_multiply.
21571 * config/arm/types.md (crypto_pmull): Add.
21572 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21573 attribute type list.
21575 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21577 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21580 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21582 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21583 to make sure not to dereference a NULL cost_classes_ptr pointer.
21585 2017-06-20 Carl Love <cel@us.ibm.com>
21587 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21588 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21589 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21590 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21591 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21592 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21593 VMULOSW): New enum "unspec" values.
21594 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21595 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21596 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21597 altivec_vmulosw): New patterns.
21598 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21599 VMULOSW): Add definitions.
21601 2017-06-20 Julia Koval <julia.koval@intel.com>
21603 * config/i386/i386.c: Fix rounding expand for new pattern.
21604 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21606 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21608 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21609 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21611 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21613 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21616 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21618 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21619 architecture, then by implementer ID.
21620 * config/aarch64/aarch64-tune.md: Regenerate.
21622 2017-06-20 Richard Biener <rguenther@suse.de>
21624 PR middle-end/81097
21625 * fold-const.c (split_tree): Fold to type before negating.
21627 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21629 * diagnostic-show-locus.c
21630 (selftest::test_fixit_deletion_affecting_newline): New function.
21631 (selftest::diagnostic_show_locus_c_tests): Call it.
21633 2017-06-20 Andreas Schwab <schwab@suse.de>
21636 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21639 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21641 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21642 __ARM_FEATURE_COPROC according to support.
21644 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21646 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21647 Rewritten to avoid overflow for > 32-bit pointers.
21650 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21651 by removing enum keyword.
21652 (ubsan_type_descriptor): Likewise. Formatting fix.
21655 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21656 splitter): Require TARGET_SSE2 in the condition.
21658 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21661 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21662 for doing vector set of SFmode on ISA 3.0.
21663 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21664 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21666 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21667 SFmode value into a V4SF variable that was extracted from another
21668 V4SF variable without converting the element to double precision
21669 and back to single precision vector format.
21670 (vsx_insert_extract_v4sf_p9_2): Likewise.
21672 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21674 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21675 in UWHI to avoid undefined overflow.
21678 * ubsan.h (enum ubsan_encode_value_phase): New.
21679 (ubsan_encode_value): Change second argument to
21680 enum ubsan_encode_value_phase with default value of
21681 UBSAN_ENCODE_VALUE_GENERIC.
21682 * ubsan.c (ubsan_encode_value): Change second argument to
21683 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21684 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21685 create_tmp_var_raw instead of create_tmp_var and use a
21687 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21688 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21689 ubsan_encode_value callers.
21692 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21693 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21694 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21696 2017-06-19 Richard Biener <rguenther@suse.de>
21698 PR middle-end/81118
21699 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21700 estimates if we changed anything.
21702 2017-06-19 Richard Biener <rguenther@suse.de>
21704 PR tree-optimization/80887
21705 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21706 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21707 simplified lookups, then reset mprts_hook.
21708 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21710 (try_to_simplify): Likewise.
21712 2017-06-19 Martin Liska <mliska@suse.cz>
21715 * gimplify.c (gimplify_switch_expr):
21716 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21718 2017-06-19 Martin Liska <mliska@suse.cz>
21720 * doc/install.texi: Document that PGO runs in 4 stages.
21722 2017-06-19 Martin Liska <mliska@suse.cz>
21725 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21726 to dispatcher function name.
21727 * multiple_target.c (replace_function_decl): New function.
21728 (create_dispatcher_calls): Redirect both edges and references.
21730 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21732 * profile-count.c (profile_count::dump): Dump quality.
21733 (profile_count::differs_from_p): Update for unsigned val.
21734 * profile-count.h (profile_count_quality): New enum.
21735 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21737 2017-06-19 Richard Biener <rguenther@suse.de>
21739 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21740 struct function as arg.
21741 (estimate_numbers_of_iterations): Export overload with loop arg.
21742 (free_numbers_of_iterations_estimates_loop): Use an overload of
21743 free_numbers_of_iterations_estimates instead.
21744 * tree-cfg.c (remove_bb): Adjust.
21745 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21746 * tree-parloops.c (gen_parallel_loop): Likewise.
21747 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21749 (tree_unroll_loops_completely): Likewise.
21750 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21751 Use an overload instead and export.
21752 (estimated_loop_iterations): Adjust.
21753 (max_loop_iterations): Likewise.
21754 (likely_max_loop_iterations): Likewise.
21755 (estimate_numbers_of_iterations): Take struct function as arg
21757 (loop_exits_before_overflow): Adjust.
21758 (free_numbers_of_iterations_estimates_loop): Use an overload.
21759 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21760 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21762 2017-06-19 Richard Biener <rguenther@suse.de>
21765 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21766 RANGE_EXPR conservatively.
21768 2017-06-16 Carl Love <cel@us.ibm.com>
21770 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21771 definitions for vec_float, vec_float2, vec_floato,
21772 vec_floate built-ins.
21773 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21774 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21776 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21777 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21778 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21779 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21780 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21781 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21782 vec_floato): Add builtin defines.
21783 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21784 Update the built-in documentation file for the new built-in
21787 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21789 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21790 (mthumb): Mark as the negative of -marm.
21792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21794 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21796 (ARM Options, -mtune): Document that this accepts the same
21797 extension options as -mcpu.
21798 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21802 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21803 permitted extensions.
21805 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21807 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21808 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21809 (armv8-m.main): Add option +nodsp.
21810 * config/arm/arm-cpu-cdata.h: Regenerated.
21812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21814 * config/arm/t-fuchsia: New file.
21815 * config.gcc (arm*-*-fuchsia*): Use it.
21817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21819 * config/arm/t-symbian: Rewrite for new option infrastructure.
21821 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21823 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21824 (MULTILIB_REQUIRED): Likewise.
21826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21828 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21829 (MULTILIB_RESUE): Likewise.
21830 (MULTILIB_MATCHES): Likewise.
21831 (MULTLIB_REQUIRED): Likewise.
21833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21835 * config/arm/t-rtems: Rewrite for new option framework.
21837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21839 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21840 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21841 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21842 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21843 * config/arm/t-multilib: ... here.
21844 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21845 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21846 armv7-a and armv8*-a when A-profile libraries have not been built.
21847 * config/arm/t-rmprofile: Rewrite.
21849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21851 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21852 with a backslash. Remove the backslash after substituting unescaped
21854 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21858 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21859 the driver to pass through the default setting of -mfloat-abi.
21860 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21862 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21863 (all_feat_combs): New rule.
21864 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21866 * config/arm/t-aprofile: Rewrite.
21868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21870 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21871 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21872 fpu is not specified by the user/command-line.
21873 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21874 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21875 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21876 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21877 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21878 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21880 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21882 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21883 * config/arm/t-arm-elf: Rewritten.
21885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21887 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21888 have some floating-point instructions.
21889 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21890 (TARGET_MAYBE_HARD_FLOAT): New macro.
21891 * config/arm/arm-builtins.c (arm_init_builtins): Use
21892 TARGET_MAYBE_HARD_FLOAT.
21893 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21895 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21897 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21898 (configargs.h): Include it.
21899 (arm_print_hint_for_fpu_option): New function.
21900 (arm_parse_fpu_option): New function.
21901 (candidate_extension): New class.
21902 (arm_canon_for_multilib): New function.
21903 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21904 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21905 (ARCH_CANONICAL_SPECS): New macro.
21906 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21908 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21910 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21911 are set after handling multilib fragments. Set target_cpu_default2
21914 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21916 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21918 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21919 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21921 (check_arch): Likewise.
21922 * config/arm/arm.c (arm_configure_build_target): Handle
21923 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21924 options in the default.
21926 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21928 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21929 when an option is an alias of another.
21930 * config/arm/parsecpu.awk (optalias): New parser token.
21931 (gen_comm_data): Mark non-alias options as such. Emit entries
21932 for extension aliases.
21933 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21934 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21935 (armv6kz, armv6zk, armv6t2): Likewise.
21936 (armv7): Make vfpv3-d16 an alias.
21937 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21939 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21940 Sort in canonical order.
21941 (armv8-a): Sort in canonical order.
21942 (armv8.1-a, armv8.2-a): Likewise.
21943 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21945 (cortex-a9): Sort in canonical order.
21946 * config/arm/arm.c (selftests.h): Include it.
21947 (arm_test_cpu_arch_data): New function.
21948 (arm_run_self_tests): New function.
21949 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21950 (targetm): Move declaration to the end of the file.
21951 * arm-cpu-cdata.h: Regenerated.
21953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21955 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21956 call to target_mode_check describing the type of option passed.
21957 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21958 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21959 arm_parse_cpu_option_name to match parameters against list of
21961 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21962 arm_arch_core_flags data structure.
21963 * config/arm/arm-cpu_cdata.h: Regenerated.
21965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21967 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21969 (arm_print_hint_for_cpu_option): Likewise.
21970 (arm_print_hint_for_arch_option): Likewise.
21971 (arm_parse_cpu_option_name): Likewise.
21972 (arm_parse_arch_option_name): Likewise.
21973 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21974 of entries in the all_fpus list.
21975 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21976 (arm_parse_cpu_option_name): Declare.
21977 (arm_parse_arch_option_name): Declare.
21978 (arm_parse_option_features): Declare.
21979 (arm_intialize_isa): Declare.
21980 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21982 (gen_comm_data): ... here. Make definitions non-static.
21983 * config/arm/arm-cpu-data.h: Regenerated.
21984 * config/arm/arm-cpu-cdata.h: Regenerated.
21986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21988 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21989 (cpu_arch_extension): New structure.
21990 (cpu_arch_option, arch_option, cpu_option): New structures.
21991 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21992 architecture types.
21993 (gen_data): Generate new format data tables.
21994 * config/arm/arm.c (cpu_tune): New structure.
21995 (cpu_option, processors): Delete.
21996 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21997 (arm_print_hint_for_cpu_option): ... this and ...
21998 (arm_print_hint_for_arch_option): ... this.
21999 (arm_parse_arch_cpu_name): Delete. Replace with ...
22000 (arm_parse_cpu_option_name): ... this and ...
22001 (arm_parse_arch_option_name): ... this.
22002 (arm_unrecognized_feature): Change type of target parameter to
22004 (arm_parse_arch_cpu_features): Delete. Replace with ...
22005 (arm_parse_option_features): ... this.
22006 (arm_configure_build_target): Rework to use new configuration data
22008 (arm_print_tune_info): Rework for new configuration data tables.
22009 * config/arm/arm-cpu-data.h: Regenerated.
22010 * config/arm/arm-cpu.h: Regenerated.
22012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22014 * Makefile.in (OBJS): Move sbitmap.o from here ...
22015 (OBJS-libcommon): ... to here.
22017 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22019 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22020 (ISA_ALL_CRYPTO): New macro.
22021 (ISA_ALL_SIMD): New macro
22022 (ISA_ALL_FP): New macro.
22023 * config/arm/arm.c (fpu_bitlist): Update initializer.
22024 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22026 (arm9e): Add fpu. Add option for nofp
22027 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22028 (arm926ej-s, arm1026ej-s): Likewise.
22029 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22030 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22031 neon-fp16, neon-vfpv4, nofp and nosimd.
22032 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22033 (cortex-a8): Add fpu. Add option for nofp.
22034 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22035 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22036 (cortex-r4f): Add fpu.
22037 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22038 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22040 (cortex-r8): Likewise.
22041 (cortex-m4): Add fpu. Add option for nofp.
22042 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22043 (cortex-a17.cortex-a7): Likewise.
22044 (cortex-a32): Add fpu. Add options for crypto and nofp.
22045 (cortex-a35, cortex-a53): Likewise.
22046 (cortex-a57): Add fpu. Add option for crypto.
22047 (cortex-a72, cortex-a73): Likewise.
22048 (exynos-m1): Likewise.
22049 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22050 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22051 (cortex-m33): Add fpu. Add option for nofp.
22052 * config/arm/arm-cpu-cdata.h: Regenerated
22053 * config/arm/arm-cpu-data.h: Regenerated.
22055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22057 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22058 (armv5te, armv5tej): Likewise.
22059 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22060 (armv7): Add options fp and vfpv3-d16.
22061 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22062 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22064 (armv7ve): Likewise.
22065 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22066 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22067 (armv8-a): Add nocrypto option.
22068 (armv8.1-a, armv8.2-a): Likewise.
22069 (armv8-m.main): add options fp, fp.dp and nofp.
22071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22073 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22075 (armv8-a+crc): Delete.
22076 (armv8.1-a): Add options simd, crypto and nofp.
22077 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22078 (armv8.2-a+fp16): Delete.
22079 (armv8-m.main): Add option dsp.
22080 (armv8-m.main+dsp): Delete.
22081 (cortex-a8): Add fpu. Add nofp option.
22082 (cortex-a9): Add fpu. Add nofp and nosimd options.
22083 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22084 link to main cpu and architecture data structures.
22085 (gen_comm_data): Only put isa attributes from the main architecture
22087 (option): New statement for architecture and CPU entries.
22088 * arm.c (struct cpu_option): New structure.
22089 (struct processors): Add entry for options.
22090 (arm_unrecognized_feature): New function.
22091 (arm_parse_arch_cpu_name): Ignore any characters after the first
22093 (arm_parse_arch_cpu_feature): New function.
22094 (arm_configure_build_target): Separate out any CPU and architecture
22095 features and parse separately. Don't error out if -mfpu=auto is
22096 used with only an architecture string.
22097 (arm_print_asm_arch_directives): New function.
22098 (arm_file_start): Call it.
22099 * config/arm/arm-cpu-cdata.h: Regenerated.
22100 * config/arm/arm-cpu-data.h: Likewise.
22101 * config/arm/arm-tables.opt: Likewise.
22103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22105 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22106 assembler when it is not -mfpu=auto.
22108 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22110 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22111 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22112 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22113 (ASM_CPU_SPEC): Rewrite.
22114 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22115 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22116 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22117 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22118 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22119 copied string is NUL-terminated. Also strip any characters prefixed
22121 (arm_rewrite_selected_arch): New function.
22122 (arm_rewrite_march): New function.
22124 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22126 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22127 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22128 (march, mcpu, mtune): Convert to string-based options.
22129 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22130 (arm_parse_arch_cpu_name): New function.
22131 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22132 identify selected architecture or CPU.
22133 (arm_option_save): New function.
22134 (TARGET_OPTION_SAVE): Redefine.
22135 (arm_option_restore): Restore string options.
22136 (arm_option_print): Print string options.
22138 2017-06-16 Martin Sebor <msebor@redhat.com>
22140 PR tree-optimization/80933
22141 PR tree-optimization/80934
22142 * builtins.c (fold_builtin_3): Do not handle bcmp here.
22143 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
22144 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
22145 (gimple_fold_builtin): Call them.
22147 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22149 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
22150 as unlikely; update profile.
22152 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22154 * predict.c (force_edge_cold): Handle declaring edges impossible
22157 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22159 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
22161 (try_unroll_loop_completely): Fix reporting.
22163 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22165 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
22167 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
22170 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
22171 if given a non-constant argument for an intrinsic which requires a
22174 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22176 * profile.c (compare_freqs): New function.
22177 (branch_prob): Sort edge list.
22178 (find_spanning_tree): Assume that the list is priority sorted.
22180 2017-06-16 Richard Biener <rguenther@suse.de>
22182 PR tree-optimization/81090
22183 * passes.def (pass_record_bounds): Remove.
22184 * tree-pass.h (make_pass_record_bounds): Likewise.
22185 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
22186 make_pass_record_bounds): Likewise.
22187 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
22188 not free niter estimates at the beginning but at the end.
22189 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
22191 2017-06-16 Richard Biener <rguenther@suse.de>
22193 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
22194 initializer to workaround ICE in host GCC 4.8.
22196 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22198 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
22200 (clone_inlined_nodes): Update.
22202 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22204 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
22205 prefetch settings, and enable prefetching by default at -O3.
22207 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22209 * config/aarch64/aarch64.c (aarch64_override_options_internal):
22210 Set flag_prefetch_loop_arrays according to tuning data.
22212 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22214 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
22215 New tune structure.
22216 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
22217 [Unrelated to main purpose of the patch] Place the pointer field last
22218 to enable type checking errors when tune structure are wrongly merged.
22219 * config/aarch64/aarch64.c (generic_prefetch_tune,)
22220 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
22221 (thunderx2t99_prefetch_tune): New tune constants.
22222 (tune_params *_tunings): Update all tunings (no functional change).
22223 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
22224 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
22225 from tunings structures.
22227 2017-06-16 Jakub Jelinek <jakub@redhat.com>
22230 * ubsan.c (instrument_null): Add T argument, use it instead
22231 of computing it based on IS_LHS.
22232 (instrument_object_size): Likewise.
22233 (pass_ubsan::execute): Adjust instrument_null and
22234 instrument_object_size callers to pass gimple_get_lhs or
22235 gimple_assign_rhs1 result to it. Use instrument_null instead of
22236 calling get_base_address and instrument_mem_ref. Handle
22237 aggregate call arguments for object-size sanitization.
22239 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22241 PR tree-optimization/81089
22242 * tree-vrp.c (is_masked_range_test): Validate operands of
22245 2017-06-15 Martin Sebor <msebor@redhat.com>
22248 * dumpfile.c (dump_register): Avoid calling memset to initialize
22249 a class with a default ctor.
22250 * gcc.c (struct compiler): Remove const qualification.
22251 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22252 * hash-table.h: Ditto.
22253 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22255 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22256 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22258 * params.h (struct param_info): Make struct members non-const.
22259 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22260 with default initialization.
22261 * vec.h (vec_copy_construct, vec_default_construct): New helper
22263 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22264 with vec_copy_construct.
22265 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22266 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22267 * doc/invoke.texi (-Wclass-memaccess): Document.
22269 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22271 * emit-rtl.h (is_leaf): Update comment about local
22272 register allocator.
22274 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22277 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22278 for a variable to have a section before checking if the section has a
22280 Set section to.persistent if persistent attribute is set.
22281 Warn if .persistent attribute is used on an automatic variable.
22283 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22285 PR rtl-optimization/80474
22286 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22288 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22290 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22293 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22295 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22296 (casesi): Emit bounds checking as RTL.
22297 (casesi_internal_mips16_<mode>): Remove bounds checking.
22299 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22301 * config/xtensa/xtensa.c (xtensa_option_override): Append
22302 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22303 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22304 xtensa_doloop_hooks): Define unconditionally.
22305 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22307 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22308 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22309 for it in xtensa_option_override.
22310 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22311 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22313 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22315 * doc/cppopts.texi: Document '-' special value to -MF.
22317 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22319 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22320 (cortex_a53_fconst): Likewise.
22321 (cortex_a53_fpmul): Likewise.
22322 (cortex_a53_f_load_64): Likewise.
22323 (cortex_a53_f_load_many): Likewise.
22324 (cortex_a53_advsimd_alu): Likewise.
22325 (cortex_a53_advsimd_alu_q): Likewise.
22326 (cortex_a53_advsimd_mul): Likewise.
22327 (cortex_a53_advsimd_mul_q): Likewise.
22328 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22329 Add missing fmul, r2f_cvt and fconst cases.
22331 2017-06-14 Richard Biener <rguenther@suse.de>
22333 PR middle-end/81088
22334 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22336 (fold_binary_loc): When associating do not treat pre-existing
22337 TREE_OVERFLOW on literal constants as a reason to allow
22338 TREE_OVERFLOW on associated literal constants.
22340 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22342 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22343 (MASK_FEATURES): New macro.
22344 * config/sparc/sparc.c (sparc_option_override): Remove the special
22345 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22347 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22349 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22350 a division of 0 if non-call exceptions are enabled.
22352 2017-06-14 Andrew Pinski <apinski@cavium.com>
22353 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22356 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22357 Improve vector initialization code gen for only variable case.
22359 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22361 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22363 2017-06-14 Richard Biener <rguenther@suse.de>
22365 PR tree-optimization/81083
22366 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22369 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22371 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22372 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22373 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22374 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22375 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22376 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22378 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22380 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22381 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22383 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22385 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22387 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22389 * config/rs6000/t-rtems: Don't handle SPE.
22391 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22393 * config/rs6000/t-linux: Don't handle SPE.
22395 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22397 * config/rs6000/eabispe.h: Delete file.
22399 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22401 * config/rs6000/t-spe: Delete file.
22403 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22405 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22406 (rs6000_legitimate_offset_address_p): Return false for anything in
22407 V2SImode or V2SFmode.
22409 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22411 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22412 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22413 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22415 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22416 (rs6000_legitimate_offset_address_p): Ditto.
22417 (rs6000_emit_move): Ditto.
22418 (rs6000_init_builtins): Remove V4HI_type_node.
22420 2017-06-13 Martin Liska <mliska@suse.cz>
22423 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22424 (gate_asan): Likewise.
22425 * asan.h (asan_no_sanitize_address_p): Remove the function.
22426 (sanitize_flags_p): New function.
22427 * builtins.def: Fix coding style.
22428 * common.opt: Use renamed enum value.
22429 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22430 * doc/extend.texi: Document no_sanitize attribute.
22431 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22432 to SANITIZE_UNDEFINED_NONDEFAULT.
22433 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22434 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22435 Use sanitize_flags_p.
22436 * gimplify.c (gimplify_function_tree): Likewise.
22437 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
22438 * opts.c (parse_no_sanitize_attribute): New function.
22439 (common_handle_option): Use renamed enum value.
22440 * opts.h (parse_no_sanitize_attribute): Declare.
22441 * tree.c (sanitize_flags_p): New function.
22442 * tree.h: Declared here.
22443 * tsan.c: Use sanitize_flags_p.
22444 * ubsan.c (ubsan_expand_null_ifn): Likewise.
22445 (instrument_mem_ref): Likewise.
22446 (instrument_bool_enum_load): Likewise.
22447 (do_ubsan_in_current_function): Remove the function.
22448 (pass_ubsan::execute): Use sanitize_flags_p.
22449 * ubsan.h: Remove do_ubsan_in_current_function
22450 * tree-cfg.c (print_no_sanitize_attr_value): New function.
22451 (dump_function_to_file): Use it here.
22453 2017-06-13 Martin Jambor <mjambor@suse.cz>
22455 PR tree-optimization/80803
22456 PR tree-optimization/81063
22457 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
22458 (propagate_subaccesses_across_link): Enqueue subtree whenever
22459 necessary instead of relying on the caller.
22461 2017-06-13 Martin Jambor <mjambor@suse.cz>
22463 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
22464 that have a first_link.
22465 (sort_and_splice_var_accesses): Do not check first_link before
22467 (subtree_mark_written_and_enqueue): Likewise.
22468 (propagate_all_subaccesses): Likewise and do not stop at first
22469 parent with a first_link.
22471 2017-06-13 Martin Jambor <mjambor@suse.cz>
22473 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
22476 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22478 * match.pd: New pattern.
22480 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22482 * tree-vrp.c (is_masked_range_test): New function.
22483 (register_edge_assert_for): Determine ranges for
22486 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22488 PR tree-optimization/67328
22489 * fold-const.c (maskable_range_p): New function.
22490 (build_range_check): Generate bittests if possible.
22492 2017-06-13 Martin Liska <mliska@suse.cz>
22494 * gimple-pretty-print.c (dump_probability): Add new argument.
22495 (dump_edge_probability): Dump both probability and count.
22496 (dump_gimple_label): Likewise.
22497 (dump_gimple_bb_header): Likewise.
22499 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
22502 * config/avr/avr-devices.c: Fix indentation.
22503 * config/avr/gen-avr-mmcu-specs.c: Dito.
22505 2017-06-13 Richard Biener <rguenther@suse.de>
22507 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
22508 instead get vector type from stmt_info.
22509 (vectorizable_reduction): Adjust. Remove dead code.
22511 2017-06-13 Richard Biener <rguenther@suse.de>
22513 PR middle-end/81065
22514 * fold-const.c (extract_muldiv_1): Remove bogus distribution
22515 case of C * (x * C2 + C3).
22516 (fold_addr_of_array_ref_difference): Properly fold index difference.
22518 2017-06-12 David S. Miller <davem@davemloft.net>
22521 * config/sparc/sparc.md (return expander): Emit frame blockage if
22522 function uses alloca.
22524 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
22526 * combine.c (make_field_assignment): Check len rather than the mode
22527 precision when calling force_to_mode.
22529 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
22531 Support multilibs and devices that see flash in RAM address range.
22534 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
22535 (avr_mcu_t) <flash_pm_offset>: New field.
22536 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
22537 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
22538 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
22539 (AVR_TINY_PM_OFFSET): Remove macro.
22540 * config/avr/avr.opt (-mshort-calls): New option.
22541 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22542 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
22543 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
22544 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
22545 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
22546 instead of avr_arch->have_jmp_call.
22547 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
22548 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
22549 avr_arch->flash_pm_offset to define.
22550 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
22551 new field flash_pm_offset. Add entry for avrxmega3.
22552 (avr_texinfo): Add entry for avrxmega3.
22553 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
22554 attiny212, attiny214,
22555 attiny412, attiny414, attiny416, attiny417,
22556 attiny814, attiny816, attiny817,
22557 attiny1614, attiny1616, attiny1617,
22558 attiny3214, attiny3216, attiny3217.
22559 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
22560 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
22561 (avr_print_operand_address) [AVR_TINY]: Same.
22562 (avr_asm_init_sections) <readonly_data_section>: Only patch
22563 callback if avr_arch->flash_pm_offset = 0.
22564 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
22565 for rodata if avr_arch->flash_pm_offset != 0.
22566 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
22567 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
22568 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
22569 (m_options): Append opt_rcall.
22570 (m_dirnames): Append dir_rcall.
22571 * config/avr/t-multilib: Regenerate.
22573 * configure.ac [target=avr]: Check whether avrxmega3 default
22574 linker description file works as needed.
22575 * configure: Regenerate.
22576 * doc/avr-mmcu.texi: Regenerate.
22577 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22578 <__AVR_ARCH__>: Document avrxmega3 and 103.
22579 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22580 <__AVR_SHORT_CALLS__>: Document it.
22581 <__AVR_PM_BASE_ADDRESS__>: Document it.
22582 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22583 (AVR Variable Attributes) <progmem>: Document this is
22584 not needed for avrxmega3.
22585 (AVR Named Address Spaces) <__flash>: Dito.
22587 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22589 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22591 2017-06-12 Doug Rupp <rupp@adacore.com>
22593 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22594 Append vxworks-stdint.h to the tm_file list.
22595 * config/vxworks-stdint.h: New file.
22597 2017-06-12 Martin Liska <mliska@suse.cz>
22599 PR tree-optimization/81041
22600 * tree-profile.c (gimple_gen_ic_func_profiler):
22601 Create an extra BB in profile-generate
22602 (gimple_gen_time_profiler): Likewise.
22604 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22606 PR tree-optimization/81003
22607 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22608 (update_range_test): Use it instead of force_gimple_operand_gsi.
22610 2017-06-12 Richard Biener <rguenther@suse.de>
22612 PR tree-optimization/81053
22613 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22614 with backedge value not defined in loop. Simplify def stmt
22617 2017-06-11 Tom de Vries <tom@codesourcery.com>
22620 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22622 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22623 nvptx_cannot_force_const_mem.
22625 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22627 * opts.c (finish_options): Move test for flag_split_stack after
22628 it has been initialized.
22630 2017-06-11 Jason Merrill <jason@redhat.com>
22632 * tree.h (id_equal): New.
22633 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22634 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22635 instead of strcmp of IDENTIFIER_POINTER.
22637 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22639 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22640 (mark_all_inlined_calls_cdtor): Fix formating.
22641 (inline_transform): Rescale profile before inlining.
22643 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22645 * cgraph.h (cgraph_edge::clone): Update prototype.
22646 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22647 (cgraph_node::create_clone): Update.
22648 (cgraph_node::create_version_clone): Update.
22649 * tree-inline.c (copy_bb): Update.
22650 (expand_call_inline): Update.
22652 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22654 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22655 factored out from ...
22656 (rs6000_emit_prologue): ... here.
22658 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22660 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22661 factored out from ...
22662 (rs6000_emit_prologue): ... here.
22664 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22666 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22668 (handle_missing_profiles): Fix computation of tp_first_run.
22669 (counts_to_freqs): Do not touch freqs when count is 0.
22671 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22673 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22676 2017-06-10 Tom de Vries <tom@codesourcery.com>
22678 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22679 attributes): Document signal effective target.
22681 2017-06-10 Tom de Vries <tom@codesourcery.com>
22683 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22684 Document effective target stack_size.
22686 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22688 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22689 to the edit_context if they can be auto-applied.
22691 2017-06-9 Ian Lance Taylor <iant@golang.org>
22693 * opts.c (finish_options): If -fsplit-stack, disable implicit
22694 -forder-blocks-and-partition.
22695 * doc/invoke.texi (Optimize Options): Document that when using
22696 -fsplit-stack -forder-blocks-and-partition is not implicitly
22699 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22701 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22702 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22703 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22704 * builtins.def (abort, trap, unreachable): Declare cold.
22705 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22706 * tree-core.h (ECF_COLD): New.
22707 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22708 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22710 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22712 * predict.c (unlikely_executed_stmt_p): Cleanup.
22714 2017-06-09 Richard Biener <rguenther@suse.de>
22716 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22717 model if the ref is always written to.
22719 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22721 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22723 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22725 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22728 2017-06-09 Tom de Vries <tom@codesourcery.com>
22731 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22732 "target cannot support label values" when encountering LABEL_REF.
22734 2017-06-09 Martin Liska <mliska@suse.cz>
22736 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22737 (gimple_gen_ic_func_profiler): Emit direct comparison
22738 of __gcov_indirect_call_callee with NULL.
22739 (gimple_gen_time_profiler): Change probability from
22740 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22742 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22744 * profile.c (edge_gcov_counts): Turn to pointer.
22745 (compute_branch_probabilities, compute_branch_probabilities): Update.
22746 (branch_prob): Do not clear edge_gcov_count.
22747 * profile.h (edge_gcov_counts): Turn to pointer.
22748 (edge_gcov_count): Update.
22750 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22752 * gimple.h (gimple_check_failed): Mark cold.
22754 2017-06-09 Richard Biener <rguenther@suse.de>
22756 PR tree-optimization/66623
22757 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22758 refactor check_reduction into two parts, properly computing
22759 whether we have to check reduction validity for outer loop
22762 2017-06-09 Richard Biener <rguenther@suse.de>
22764 PR tree-optimization/79483
22765 * graphite-scop-detection.c (order): New global.
22766 (get_order): Compute bb to order mapping that satisfies code
22767 generation constraints.
22768 (cmp_pbbs): New helper.
22769 (build_scops): Start domwalk at entry block, sort generated
22772 2017-06-09 Richard Biener <rguenther@suse.de>
22774 PR middle-end/81007
22775 * ipa-polymorphic-call.c
22776 (ipa_polymorphic_call_context::restrict_to_inner_class):
22777 Skip FIELD_DECLs with error_mark_node type.
22778 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22781 2017-06-09 Martin Liska <mliska@suse.cz>
22783 * predict.c (struct branch_predictor): New struct.
22784 (test_prediction_value_range): New test.
22785 (predict_c_tests): New function.
22786 * selftest-run-tests.c (selftest::run_tests): Run the function.
22787 * selftest.h: Declare new tests.
22789 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22792 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22793 gen_add3_insn did not fail.
22794 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22795 r0, construct that number in a temporary reg and add that reg to r0.
22796 If asked to put the result in r0 as well, fail.
22798 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22800 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22801 for early expansion of vec_eqv.
22803 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22805 PR middle-end/81005
22806 * ubsan.c (instrument_null): Avoid pointless code temporary.
22807 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22809 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22813 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22816 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22819 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22821 * predict.c (unlikely_executed_edge_p): Move ahead.
22822 (probably_never_executed_edge_p): Use it.
22824 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22826 PR middle-end/79988
22827 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22828 gimple_call_builtin_p call.
22830 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22832 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22833 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22834 rtl_check_failed_type2, rtl_check_failed_code1,
22835 rtl_check_failed_code2, rtl_check_failed_code_mode,
22836 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22837 rtvec_check_failed_bounds, rtl_check_failed_flag,
22838 _fatal_insn_not_found, _fatal_insn): Likewise.
22839 * tree.h (tree_contains_struct_check_failed,
22840 tree_check_failed, tree_not_check_failed,
22841 tree_class_check_failed, tree_range_check_failed,
22842 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22843 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22844 tree_operand_check_failed, omp_clause_check_failed,
22845 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22848 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22850 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22851 flag_branch_probabilities.
22852 * ipa-inline.c (edge_badness): Likewise.
22853 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22854 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22855 * predict.c (maybe_hot_frequency_p): Likewise.
22856 (probably_never_executed): Likewise.
22857 * sched-ebb.c (schedule_ebbs): Likewise.
22858 * sched-rgn.c (find_single_block_region): Likewise.
22859 * tracer.c (tail_duplicate): Likewise.
22861 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22863 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22864 longer requires x_flag_profile_use.
22866 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22868 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22869 instead of flag_reorder_blocks_and_partition.
22870 * dbxout.c (dbxout_function_end): Likewise.
22871 * dwarf2out.c (gen_subprogram_die): Likewise.
22872 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22873 * hw-doloop.c (reorg_loops): Likewise.
22874 * varasm.c (assemble_start_function,
22875 assemble_end_function): Likewise.
22876 (decide_function_section): Do not check for
22877 flag_reorder_blocks_and_partition.
22879 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22881 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22883 (chkp_get_hard_register_fake_addr_expr): Ditto.
22884 (chkp_build_addr_expr): Add check for hard reg case.
22885 (chkp_parse_array_and_component_ref): Ditto.
22886 (chkp_find_bounds_1): Ditto.
22887 (chkp_process_stmt): Don't generate bounds store for
22890 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22892 * predict.c (maybe_hot_bb_p): Do not check profile status.
22893 (maybe_hot_edge_p): Likewise.
22894 (probably_never_executed): Check for zero counts even if profile
22896 (unlikely_executed_edge_p): New function.
22897 (unlikely_executed_stmt_p): New function.
22898 (unlikely_executed_bb_p): New function.
22899 (set_even_probabilities): Use unlikely predicates.
22900 (combine_predictions_for_bb): Likewise.
22901 (predict_paths_for_bb): Likewise.
22902 (predict_paths_leading_to_edge): Likewise.
22903 (determine_unlikely_bbs): New function.
22904 (estimate_bb_frequencies): Use it.
22905 (compute_function_frequency): Use zero counts even if profile is
22907 * profile-count.h: Fix typo.
22909 2017-08-08 Julia Koval <julia.koval@intel.com>
22911 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22912 _mm512_mask_cvtsepi16_storeu_epi8,
22913 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22914 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22915 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22916 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22917 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22918 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22919 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22920 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22921 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22922 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22923 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22924 __builtin_ia32_pmovuswb256mem_mask,
22925 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22926 __builtin_ia32_pmovwb512mem_mask): New builtins.
22928 2017-08-08 Julia Koval <julia.koval@intel.com>
22930 PR target/73350,80862
22931 * config/i386/subst.md (round): Fix round pattern.
22932 * config/i386/i386.c (ix86_erase_embedded_rounding):
22933 Fix erasing rounding for the fixed pattern.
22935 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22937 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22939 2017-06-08 Martin Liska <mliska@suse.cz>
22941 PR gcov-profile/80911
22942 * gcov.c (block_info::block_info): New constructor.
22944 2017-06-07 Carl Love <cel@us.ibm.com>
22946 * config/rs6000/rs6000-c: The return type of the following
22947 built-in functions was implemented as int not long long. Fix sign
22948 of return value for the unsigned version of vec_mulo and vec_mule.
22949 vector unsigned long long vec_bperm (vector unsigned long long,
22950 vector unsigned char)
22951 vector signed long long vec_mule (vector signed int,
22953 vector unsigned long long vec_mule (vector unsigned int,
22954 vector unsigned int)
22955 vector signed long long vec_mulo (vector signed int,
22957 vector unsigned long long vec_mulo (vector unsigned int,
22958 vector unsigned int)
22959 * doc/extend.texi: Fix the documentation for the built-in
22962 2017-06-07 Carl Love <cel@us.ibm.com>
22965 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22968 2017-06-07 Carl Love <cel@us.ibm.com>
22970 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22971 support, Generate doublehv for signed int/float for BE case only.
22973 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22975 * doc/invoke.texi (mcx16): Rewrite.
22977 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22979 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22980 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22981 *mov<mode>_softfloat, and an anonymous splitter): Use
22982 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22984 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22986 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22988 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22989 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22990 (rs6000_debug_reg_global): Adjust.
22991 (rs6000_init_hard_regno_mode_ok): Adjust.
22992 (rs6000_dbx_register_number): Adjust.
22993 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22994 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22995 Remove SPE_ACC and SPEFSCR.
22996 (REG_ALLOC_ORDER): Ditto.
22997 (FRAME_POINTER_REGNUM): Change to 111.
22998 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22999 (REG_CLASS_NAMES): Ditto.
23000 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23001 (REGISTER_NAMES): Ditto.
23002 (ADDITIONAL_REG_NAMES): Ditto.
23003 (rs6000_reg_names): Ditto.
23004 * config/rs6000/rs6000.md: Renumber some register number
23007 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23009 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23011 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23012 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23014 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23015 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23016 Delete the SPE high registers.
23017 (REG_ALLOC_ORDER): Ditto.
23018 (enum reg_class): Remove SPE_HIGH_REGS.
23019 (REG_CLASS_NAMES): Ditto.
23020 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23021 (REGISTER_NAMES): Ditto.
23022 (rs6000_reg_names): Ditto.
23023 * doc/tm.texi.in: Remove SPE as example.
23024 * doc/tm.texi: Regenerate.
23026 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23028 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23029 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23030 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23031 * config/rs6000/rs6000.md (type): Remove "brinc".
23033 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23035 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23036 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23037 * config/rs6000/linuxspe.h: Delete file.
23038 * config/rs6000/rs6000.md: Don't include spe.md.
23039 * config/rs6000/spe.h: Delete file.
23040 * config/rs6000/spe.md: Delete file.
23041 * config/rs6000/t-rs6000: Remove spe.md.
23043 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23045 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23046 (reg_or_none500mem_operand): Delete.
23047 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23048 instead of reg_or_none500mem_operand.
23050 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23052 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23053 handling of SPE flags.
23054 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23056 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23058 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23060 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23061 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23062 paired_divv2sf3): Similar.
23063 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23064 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23065 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23067 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23068 Rename the paired_* instruction patterns.
23069 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23071 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23072 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23073 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23074 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23075 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23076 PAIRED_VECTOR_MODE.
23077 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23078 (spe_func_has_64bit_regs_p): Delete.
23079 (spe_expand_predicate_builtin): Delete.
23080 (spe_expand_evsel_builtin): Delete.
23081 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23082 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23083 (invalid_e500_subreg): Delete.
23084 (rs6000_legitimize_address): Always force_reg op2 as well, for
23085 paired single memory accesses.
23086 (rs6000_member_type_forces_blk): Delete.
23087 (rs6000_spe_function_arg): Delete.
23088 (rs6000_expand_unop_builtin): Delete SPE handling.
23089 (rs6000_expand_binop_builtin): Ditto.
23090 (spe_expand_stv_builtin): Delete.
23091 (bdesc_2arg_spe): Delete.
23092 (spe_expand_builtin): Delete.
23093 (spe_expand_predicate_builtin): Delete.
23094 (spe_expand_evsel_builtin): Delete.
23095 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23096 (spe_init_builtins): Delete.
23097 (spe_func_has_64bit_regs_p): Delete.
23098 (savres_routine_name): Delete "info" parameter. Adjust callers.
23099 (rs6000_emit_stack_reset): Ditto.
23100 (rs6000_dwarf_register_span): Delete.
23101 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23102 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23103 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23105 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23107 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23108 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23109 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23110 mulv2sf3, divv2sf3): Delete expanders.
23112 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23114 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23118 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23119 * config/rs6000/rs6000.c: Ditto.
23121 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23123 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23124 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23125 comparison_operator.
23127 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23129 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23130 * config/rs6000/rs6000.opt: Ditto.
23131 * config/rs6000/t-rtems: Ditto.
23133 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23135 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
23136 TARGET_E500_SINGLE by 0, simplify.
23137 * config/rs6000/rs6000.c: Ditto.
23138 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
23139 (spe_build_register_parallel): Delete.
23140 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
23141 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
23142 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
23143 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
23144 (E500_CONVERT): Delete.
23145 * config/rs6000/spe.md: Remove many patterns and all define_constants.
23147 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23149 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
23150 * config/rs6000/dfp.md: Ditto.
23151 (negdd2, *negdd2_fpr): Merge.
23152 (absdd2, *absdd2_fpr): Merge.
23153 (negtd2, *negtd2_fpr): Merge.
23154 (abstd2, *abstd2_fpr): Merge.
23155 * config/rs6000/e500.h: Delete file.
23156 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
23157 TARGET_FPRS by 1 and simplify.
23158 * config/rs6000/rs6000-c.c: Ditto.
23159 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
23160 TARGET_DF_SPE by 0.
23161 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
23163 * config/rs6000/rs6000.md: Ditto.
23164 (floatdidf2, *floatdidf2_fpr): Merge.
23165 (move_from_CR_gt_bit): Delete.
23166 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
23167 (E500_CR_IOR_COMPARE): Delete.
23168 (All patterns that require !TARGET_FPRS): Delete.
23169 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
23171 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23173 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
23175 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23177 * graphds.c (add_edge): Intitialize edge's attached data.
23178 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
23179 pointer parameter. Call pointed function on each edge during
23180 graph traversing. Skip traversing the edge when the function
23182 (graphds_dfs, graphds_scc): Ditto.
23183 (for_each_edge): New parameter. Pass the new parameter to callback
23185 * graphds.h (skip_edge_callback): New function pointer type.
23186 (graphds_dfs, graphds_scc): New function pointer parameter.
23187 (graphds_edge_callback, for_each_edge): New parameter.
23189 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23191 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
23192 out code checking if runtime alias check is possible to below ...
23193 Call the new function.
23194 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
23195 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
23197 2017-06-07 Marek Polacek <polacek@redhat.com>
23200 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
23201 TYPE_OVERFLOW_WRAPS check.
23203 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23205 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
23206 if versioning is required.
23207 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
23208 peeling with the check for versioning.
23210 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23212 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
23213 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
23214 Set true to new parameter if new ssa variable is defined.
23215 (vect_gen_vector_loop_niters): Refactor. Set range information
23216 for the new vector loop bound variable.
23217 (vect_do_peeling): Ditto.
23219 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23221 * tree-affine.c (ssa.h): Include header file.
23222 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
23223 has wrapping overflow behavior.
23225 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23227 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
23229 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23231 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
23232 (tree_to_aff_combination): ... here.
23234 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23236 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23237 reg_pressure model function.
23238 (ivopts_global_cost_for_size): Delete.
23239 (determine_set_costs, iv_ca_recount_cost): Call new model function
23240 ivopts_estimate_reg_pressure.
23242 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23244 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23245 expensive than udiv. Remove floating point cases from mod.
23247 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23249 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23250 Increase idiv cost.
23252 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23254 * config/aarch64/aarch64.md
23255 (copysignsf3): Fix mask generation.
23257 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23259 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23261 (class dump_manager): Add register_dumps method.
23262 * dumpfile.c: Include langhooks.h.
23263 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23264 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23265 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23266 (dump_manager::dump_register): Start with 512 entries instead of 32.
23267 (dump_manager::register_dumps): New method.
23268 * toplev.c (general_init): Instead of invoking register_dumps
23269 langhook, invoke register_dumps method on the dump manager.
23270 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23273 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23275 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23276 Say that # requires an associated define_split to exist, and that
23277 the define_split must be suitable for use after register allocation.
23279 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23281 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23282 (compute_outgoing_frequencies): Also initialize zero counts.
23283 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23284 around loops; preserve more of profile when nothing changes.
23286 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23288 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23290 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23291 * config/arm/arm-cpu-cdata.h: Regenerate.
23292 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23293 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23294 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23295 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23296 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23298 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23300 * config/arm/t-rmprofile: Likewise.
23301 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23303 2017-06-06 David S. Miller <davem@davemloft.net>
23306 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23307 blockage if function uses alloca.
23309 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23311 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23312 New "uid" fields to hold pretty-print IDs of group and ref.
23313 Memory references are now identified as <group_id>:<ref_id>
23314 instead of using [random] addresses.
23315 (dump_mem_details): Simplify, no functional change.
23316 (dump_mem_ref): Simplify and make output more concise.
23317 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23318 (find_or_create_group): Initialize group uid.
23319 (record_ref): Initialize ref uid. Improve debug output.
23320 (prune_group_by_reuse, should_issue_prefetch_p,)
23321 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23322 (mark_nontemporal_store, determine_loop_nest_reuse):
23323 Improve debug output.
23325 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23327 * dbgcnt.def (prefetch): New debug counter.
23328 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23329 (schedule_prefetches): Stop issueing prefetches if debug counter
23332 2017-06-06 Tom de Vries <tom@codesourcery.com>
23334 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23335 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23337 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23339 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23340 Use aarch64_reg_or_zero predicate for operand 4.
23341 (aarch64_compare_and_swap<mode> define_insn_and_split):
23342 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23343 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23345 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23347 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23348 (arm_compute_save_core_reg_mask): This.
23349 (thumb1_compute_save_reg_mask): Rename into ...
23350 (thumb1_compute_save_core_reg_mask): This.
23351 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23352 (arm_compute_frame_layout): Likewise.
23354 2017-06-06 Richard Biener <rguenther@suse.de>
23356 PR tree-optimization/80974
23357 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23358 keep or clear leaders SSA info.
23360 2017-06-06 Tom de Vries <tom@codesourcery.com>
23362 * config/nvptx/nvptx.c (split_mode_p): New function.
23363 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23365 2017-06-06 Tom de Vries <tom@codesourcery.com>
23367 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23369 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23372 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23375 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23377 * shrink-wrap.c (handle_simple_exit): Update profile.
23378 (try_shrink_wrapping): Upate profile.
23380 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23382 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23383 (tree_guess_outgoing_edge_probabilities): New.
23384 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23385 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23387 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23389 * ipa-split.c (split_function): Initialize return bb profile.
23391 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23393 * profile.c (compute_branch_probabilities): Also initialize
23394 EXIT_BLOCK profile.
23396 2017-06-06 Richard Biener <rguenther@suse.de>
23398 PR tree-optimization/80928
23399 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23400 (vect_analyze_loop_operations): Properly guard analysis for
23402 (vect_transform_loop): Likewise.
23403 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23404 (vect_model_induction_cost): Do not cost for pure SLP.
23405 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23406 of induction in inner loop vectorization.
23407 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23408 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23409 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23411 (vect_analyze_slp_cost_1): Cost induction.
23412 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23413 (vect_get_slp_vect_defs): Likewise.
23414 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23415 (vect_transform_stmt): Handle SLP reductions.
23416 * tree-vectorizer.h (vectorizable_induction): Adjust.
23418 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23420 * config/rs6000/rs6000.c (make_resolver_func): Update
23421 init_lowered_empty_function call.
23423 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23425 * doc/invoke.texi: Document the -fprofile-abs-path option.
23426 * common.opt (fprofile-abs-path): New option.
23427 * gcov-io.h (gcov_write_filename): Declare.
23428 * gcov-io.c (gcov_write_filename): New function.
23429 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23430 * profile.c (output_location): Likewise.
23432 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23434 * shring-wrap.c: Revert accidental commit.
23436 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
23438 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
23440 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23442 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
23444 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
23445 profile in callgraph edge.
23446 * profile-count.h (apply_probability): If THIS is 0, then result is 0
23447 (apply_scale): Likewise.
23448 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
23449 Also scale profile when inlining function with zero profile.
23450 (initialize_cfun): Update exit block profile even when it is zero.
23451 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
23452 when profile is read.
23454 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23456 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
23457 (CLONE_*): New constants to define the processors we can generate
23458 code for with the target_clone attribute.
23459 (rs6000_clone_map): New array to identify which clone processors
23460 the current program is running on.
23461 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
23462 target_clone attribute.
23463 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
23464 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
23465 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
23466 (cpu_expand_builtin): Add support for target_clone attribute.
23467 (rs6000_valid_attribute_p): Allow "default" attribute.
23468 (get_decl_name): New debug function to simplify printing the
23469 current function name in debugging statements.
23470 (rs6000_clone_priority): New functions to support the target_clone
23471 attribute, and be able to generate code to switch between ISA 2.05
23472 through ISA 3.0 (power6 through power9).
23473 (rs6000_compare_version_priority): Likewise.
23474 (rs6000_get_function_versions_dispatcher): Likewise.
23475 (make_resolver_func): Likewise.
23476 (add_condition_to_bb): Likewise.
23477 (dispatch_function_versions): Likewise.
23478 (rs6000_generate_version_dispatcher_body): Likewise.
23479 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
23480 (fusion_gpr_load_p): Fix a spacing issue.
23481 * doc/extend.texi (Common Function Attributes): Document that the
23482 PowerPC supports the target_clone attribute.
23484 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23486 * config/arm/arm.h: explain F symbol found in description of ARM
23487 register allocation in its legend.
23489 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23491 * config/mips/frame-header-opt.c: Include profile-count.h.
23492 * config/riscv/riscv.c: Include profile-count.h
23494 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23496 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
23498 (sm_set_flag_if_changed): Add bbs field.
23499 (execute_sm_if_changed_flag_set): Pass BBS.
23500 (execute_sm): Update.
23502 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23504 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
23507 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23509 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
23510 (peephole2): New peephole2 to emit the above.
23511 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
23513 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23515 * config/aarch64/aarch64.c (define_peephole2 above
23516 *sub_<shift>_<mode>): New peephole.
23518 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23520 * config/i386/i386.c (make_resolver_func): Update.
23521 * Makefile.in: Add profile-count.h and profile-count.o
23522 * auto-profile.c (afdo_indirect_call): Update to new API.
23523 (afdo_set_bb_count): Update.
23524 (afdo_propagate_edge): Update.
23525 (afdo_propagate_circuit): Update.
23526 (afdo_calculate_branch_prob): Update.
23527 (afdo_annotate_cfg): Update.
23528 * basic-block.h: Include profile-count.h
23529 (struct edge_def): Turn count to profile_count.
23530 (struct basic_block_def): Likewie.
23531 (REG_BR_PROB_BASE): Move to profile-count.h
23532 (RDIV): Move to profile-count.h
23533 * bb-reorder.c (max_entry_count): Turn to profile_count.
23534 (find_traces): Update.
23535 (rotate_loop):Update.
23536 (connect_traces):Update.
23537 (sanitize_hot_paths):Update.
23538 * bt-load.c (migrate_btr_defs): Update.
23539 * cfg.c (RDIV): Remove.
23540 (init_flow): Use alloc_block.
23541 (alloc_block): Uninitialize count.
23542 (unchecked_make_edge): Uninitialize count.
23543 (check_bb_profile): Update.
23544 (dump_edge_info): Update.
23545 (dump_bb_info): Update.
23546 (update_bb_profile_for_threading): Update.
23547 (scale_bbs_frequencies_int): Update.
23548 (scale_bbs_frequencies_gcov_type): Update.
23549 (scale_bbs_frequencies_profile_count): New.
23550 * cfg.h (update_bb_profile_for_threading): Update.
23551 (scale_bbs_frequencies_profile_count): Declare.
23552 * cfgbuild.c (compute_outgoing_frequencies): Update.
23553 (find_many_sub_basic_blocks): Update.
23554 * cfgcleanup.c (try_forward_edges): Update.
23555 (try_crossjump_to_edge): Update.
23556 * cfgexpand.c (expand_gimple_tailcall): Update.
23557 (construct_exit_block): Update.
23558 * cfghooks.c (verify_flow_info): Update.
23559 (dump_bb_for_graph): Update.
23560 (split_edge): Update.
23561 (make_forwarder_block): Update.
23562 (duplicate_block): Update.
23563 (account_profile_record): Update.
23564 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
23565 (get_estimated_loop_iterations): Update.
23566 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
23567 (single_likely_exit): Update.
23568 * cfgloopmanip.c (scale_loop_profile): Update.
23570 (set_zero_probability): Update.
23571 (lv_adjust_loop_entry_edge): Update.
23572 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23573 (purge_dead_edges): Update.
23574 (rtl_account_profile_record): Update.
23575 * cgraph.c (cgraph_node::create): Uninitialize count.
23576 (symbol_table::create_edge): Uninitialize count.
23577 (cgraph_update_edges_for_call_stmt_node): Update.
23578 (cgraph_edge::dump_edge_flags): Update.
23579 (cgraph_node::dump): Update.
23580 (cgraph_edge::maybe_hot_p): Update.
23581 * cgraph.h: Include profile-count.h
23582 (create_clone), create_edge, create_indirect_edge): Update.
23583 (cgraph_node): Turn count to profile_count.
23584 (cgraph_edge0: Likewise.
23585 (make_speculative, clone): Update.
23586 (create_edge): Update.
23587 (init_lowered_empty_function): Update.
23588 * cgraphclones.c (cgraph_edge::clone): Update.
23589 (duplicate_thunk_for_node): Update.
23590 (cgraph_node::create_clone): Update.
23591 * cgraphunit.c (cgraph_node::analyze): Update.
23592 (cgraph_node::expand_thunk): Update.
23593 * final.c (dump_basic_block_info): Update.
23594 * gimple-streamer-in.c (input_bb): Update.
23595 * gimple-streamer-out.c (output_bb): Update.
23596 * graphite.c (print_global_statistics): Update.
23597 (print_graphite_scop_statistics): Update.
23598 * hsa-brig.c: Include basic-block.h.
23599 * hsa-dump.c: Include basic-block.h.
23600 * hsa-gen.c (T sum_slice): Update.
23601 (convert_switch_statements):Update.
23602 * hsa-regalloc.c: Include basic-block.h.
23603 * ipa-chkp.c (chkp_produce_thunks): Update.
23604 * ipa-cp.c (struct caller_statistics): Update.
23605 (init_caller_stats): Update.
23606 (gather_caller_stats): Update.
23607 (ipcp_cloning_candidate_p): Update.
23608 (good_cloning_opportunity_p): Update.
23609 (get_info_about_necessary_edges): Update.
23610 (dump_profile_updates): Update.
23611 (update_profiling_info): Update.
23612 (update_specialized_profile): Update.
23613 (perhaps_add_new_callers): Update.
23614 (decide_about_value): Update.
23615 (ipa_cp_c_finalize): Update.
23616 * ipa-devirt.c (struct odr_type_warn_count): Update.
23617 (struct decl_warn_count): Update.
23618 (struct final_warning_record): Update.
23619 (possible_polymorphic_call_targets): Update.
23620 (ipa_devirt): Update.
23621 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23622 * ipa-icf.c (sem_function::merge): Update.
23623 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23624 * ipa-inline.c (compute_uninlined_call_time): Update.
23625 (compute_inlined_call_time): Update.
23626 (want_inline_small_function_p): Update.
23627 (want_inline_self_recursive_call_p): Update.
23628 (edge_badness): Update.
23629 (lookup_recursive_calls): Update.
23630 (recursive_inlining): Update.
23631 (inline_small_functions): Update.
23632 (dump_overall_stats): Update.
23633 (dump_inline_stats): Update.
23634 * ipa-profile.c (ipa_profile_generate_summary): Update.
23635 (ipa_propagate_frequency): Update.
23636 (ipa_profile): Update.
23637 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23638 * ipa-utils.c (ipa_merge_profiles): Update.
23639 * loop-doloop.c (doloop_modify): Update.
23640 * loop-unroll.c (report_unroll): Update.
23641 (unroll_loop_runtime_iterations): Update.
23642 * lto-cgraph.c (lto_output_edge): Update.
23643 (lto_output_node): Update.
23644 (input_node): Update.
23645 (input_edge): Update.
23646 (merge_profile_summaries): Update.
23647 * lto-streamer-in.c (input_cfg): Update.
23648 * lto-streamer-out.c (output_cfg): Update.
23649 * mcf.c (create_fixup_graph): Update.
23650 (adjust_cfg_counts): Update.
23651 (sum_edge_counts): Update.
23652 * modulo-sched.c (sms_schedule): Update.
23653 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23654 * predict.c (maybe_hot_count_p): Update.
23655 (probably_never_executed): Update.
23656 (dump_prediction): Update.
23657 (combine_predictions_for_bb): Update.
23658 (propagate_freq): Update.
23659 (handle_missing_profiles): Update.
23660 (counts_to_freqs): Update.
23661 (rebuild_frequencies): Update.
23662 (force_edge_cold): Update.
23663 * predict.h: Include profile-count.h
23664 (maybe_hot_count_p, counts_to_freqs): UPdate.
23665 * print-rtl-function.c: Do not include cfg.h
23666 * print-rtl.c: Include basic-block.h
23667 * profile-count.c: New file.
23668 * profile-count.h: New file.
23669 * profile.c (is_edge_inconsistent): Update.
23670 (correct_negative_edge_counts): Update.
23671 (is_inconsistent): Update.
23672 (set_bb_counts): Update.
23673 (read_profile_edge_counts): Update.
23674 (compute_frequency_overlap): Update.
23675 (compute_branch_probabilities): Update; Initialize and deinitialize
23677 (branch_prob): Update.
23678 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23679 (edge_gcov_count): New.
23680 (bb_gcov_count): New.
23681 * shrink-wrap.c (try_shrink_wrapping): Update.
23682 * tracer.c (better_p): Update.
23683 * trans-mem.c (expand_transaction): Update.
23684 (ipa_tm_insert_irr_call): Update.
23685 (ipa_tm_insert_gettmclone_call): Update.
23686 * tree-call-cdce.c: Update.
23687 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23688 (gimple_duplicate_sese_tail): Update.
23689 (gimple_account_profile_record): Update.
23690 (execute_fixup_cfg): Update.
23691 * tree-inline.c (copy_bb): Update.
23692 (copy_edges_for_bb): Update.
23693 (initialize_cfun): Update.
23694 (freqs_to_counts): Update.
23695 (copy_cfg_body): Update.
23696 (expand_call_inline): Update.
23697 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23698 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23699 (try_unroll_loop_completely): Update.
23700 (try_peel_loop): Update.
23701 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23702 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23703 * tree-ssa-loop-split.c (connect_loops): Update.
23704 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23705 * tree-ssa-reassoc.c (branch_fixup): Update.
23706 * tree-ssa-tail-merge.c (replace_block_by): Update.
23707 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23708 (compute_path_counts): Update.
23709 (update_profile): Update.
23710 (recompute_probabilities): Update.
23711 (update_joiner_offpath_counts): Update.
23712 (estimated_freqs_path): Update.
23713 (freqs_to_counts_path): Update.
23714 (clear_counts_path): Update.
23715 (ssa_fix_duplicate_block_edges): Update.
23716 (duplicate_thread_path): Update.
23717 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23718 (struct switch_conv_info): Update.
23719 * tree-tailcall.c (decrease_profile): Update.
23720 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23721 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23722 * value-prof.c (check_counter): Update.
23723 (gimple_divmod_fixed_value): Update.
23724 (gimple_mod_pow2): Update.
23725 (gimple_mod_subtract): Update.
23726 (gimple_ic_transform): Update.
23727 (gimple_stringop_fixed_value): Update.
23728 * value-prof.h (gimple_ic): Update.
23730 2017-06-02 Carl Love <cel@us.ibm.com>
23732 * config/rs6000/rs6000-c: Add support for built-in functions
23733 vector double vec_doublee (vector signed int);
23734 vector double vec_doublee (vector unsigned int);
23735 vector double vec_doublee (vector float);
23736 vector double vec_doubleh (vector signed int);
23737 vector double vec_doubleh (vector unsigned int);
23738 vector double vec_doubleh (vector float);
23739 vector double vec_doublel (vector signed int);
23740 vector double vec_doublel (vector unsigned int);
23741 vector double vec_doublel (vector float);
23742 vector double vec_doubleo (vector signed int);
23743 vector double vec_doubleo (vector unsigned int);
23744 vector double vec_doubleo (vector float);.
23745 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23746 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23748 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23749 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23750 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23752 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23753 vec_doublel, vec_doubleh.
23754 * doc/extend.texi: Update the built-in documentation file for the
23755 new built-in functions.
23757 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23760 * ipa-inline-analysis.c (free_growth_caches): Set
23761 edge_removal_hook_holder to NULL after removing it.
23763 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23765 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23766 comparision with zero.
23768 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23769 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23770 for early expansion of vec_min and vec_max builtins.
23771 (builtin_function_type): Add min/max unsigned variants to those
23772 identified as having unsigned arguments.
23774 2017-06-02 Olivier Hainque <hainque@adacore.com>
23776 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23778 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23780 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23781 Use VALL_F16 iterator rather than VALL.
23783 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23785 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23786 Emit CBNZ inside loop when doing a strong exchange and comparing
23787 against zero. Generate the CC flags after the loop.
23789 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23791 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23792 (dl_section_ref): New.
23793 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23794 On AIX, append an expression to subtract the size of the
23795 section length to dl_section_ref.
23797 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23799 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23800 for early expansion of vector absolute builtins.
23802 2017-06-02 Richard Biener <rguenther@suse.de>
23804 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23805 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23807 2017-06-02 Richard Biener <rguenther@suse.de>
23809 PR tree-optimization/80948
23810 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23811 stmt order as well.
23813 2017-06-02 Richard Biener <rguenther@suse.de>
23815 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23817 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23818 for inductions as relevant.
23820 2017-06-02 Richard Biener <rguenther@suse.de>
23822 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23823 (vectorizable_induction): ... this. Remove dead code.
23825 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23827 * builtins. (expand_builtin_alloca): Remove second parameter and
23828 infer its value from the first parameter instead.
23829 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23831 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23833 PR rtl-optimization/80903
23834 * loop-doloop.c (add_test): Unshare sequence.
23836 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23838 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23840 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23842 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23844 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23845 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23846 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23847 (xlogue_layout::get_stub_rtx): Make static.
23848 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23849 (xlogue_layout::compute_stub_managed_regs): Rename to...
23850 (xlogue_layout::count_stub_managed_regs): ...this.
23851 (xlogue_layout::is_stub_managed_reg): New function.
23852 (xlogue_layout::m_stub_names): Rename to...
23853 (xlogue_layout::s_stub_names): ...this, make static.
23854 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23855 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23856 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23857 xlogue_layout::s_stub_names): Instantiate statics.
23858 (stub_managed_regs): Remove.
23859 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23860 (disable_call_ms2sysv_xlogues): Rename to...
23861 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23862 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23864 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23865 change after reload_completed.
23866 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23868 (ix86_expand_prologue): Likewise.
23869 (ix86_expand_epilogue): Likewise.
23870 (ix86_expand_split_stack_prologue): Likewise.
23871 (ix86_compute_frame_layout): Remove frame parameter ...
23872 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23873 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23875 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23876 (ix86_frame): Move from here ...
23877 * config/i386/i386.h (ix86_frame): ... to here.
23878 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23879 complete ix86_frame data structure instead. Remove some_ld_name.
23881 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23883 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23884 symbols that hold a DECL_VALUE_EXPR.
23886 2017-06-01 Martin Jambor <mjambor@suse.cz>
23888 PR tree-optimization/80898
23889 * tree-sra.c (process_subtree_disqualification): Removed.
23890 (disqualify_candidate): Do not acll
23891 process_subtree_disqualification.
23892 (subtree_mark_written_and_enqueue): New function.
23893 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23894 RHS has been disqualified and re-queue LHS if necessary. Apart
23895 from that, ignore disqualified RHS.
23897 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23899 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23900 address fetch for z10 or later.
23902 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23904 * config/arc/arc.md (tst_movb): Add guard when splitting.
23906 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23908 * config/arc/arc.c (arc_can_eliminate): Test against
23909 arc_frame_pointer_needed.
23911 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23913 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23914 to prevent store reordering.
23915 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23916 (type): Add block type.
23917 (stack_tie): Define special instruction to be used in
23920 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23922 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23923 constraint. It is not valid for the pattern.
23924 (noncommutative_binary_comparison): Likewise.
23926 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23928 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23931 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23933 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23934 be used by the reg-alloc.
23936 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23938 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23939 reg-alloc when having mul64 or mul32x16 instructions.
23940 (mulsidi3): Likewise.
23941 (umulsidi3): Likewise.
23942 (mulsi32x16): New pattern.
23943 (mulsi64): Likewise.
23944 (mulsidi64): Likewise.
23945 (umulsidi64): Likewise.
23946 (MUL32x16_REG): Define.
23947 (mul64_600): Use MUL32x16_REG.
23948 (mac64_600): Likewise.
23949 (umul64_600): Likewise.
23950 (umac64_600): Likewise.
23952 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23954 * config/arc/arc.md (mulsi3_700): Make it commutative.
23956 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23958 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23960 (*sign_extendsidi2_insn): Likewise for movstosw.
23962 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23964 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23965 the type of the input discriminant value. Convert the
23966 discriminant value of signedness vary.
23968 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23970 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23971 Add to -Wall section.
23973 2017-06-01 Richard Biener <rguenther@suse.de>
23975 PR middle-end/66313
23976 * fold-const.c (fold_plusminus_mult_expr): If the factored
23977 factor may be zero use a wrapping type for the inner operation.
23978 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23979 and handle moved defs.
23980 (process_assignment): Properly guard the unary op case. Return a
23981 tri-state indicating that moving the stmt before the call may allow
23982 to continue. Pass through to_move.
23983 (find_tail_calls): Handle moving unrelated defs before
23986 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23989 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23990 splitter result in the canonical way.
23992 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23994 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23995 also for 32bit target. Update insn attributes.
23996 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23998 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24000 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24001 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24002 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24003 (_mm_maskz_min_ss): New intrinsics.
24005 2017-05-31 Martin Liska <mliska@suse.cz>
24007 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24008 Change comment style to one we normally use.
24009 (vectorizable_reduction): Likewise.
24010 (vectorizable_induction): Likewise.
24011 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24012 (vectorizable_call): Likewise.
24013 (vectorizable_simd_clone_call): Likewise.
24014 (vectorizable_conversion): Likewise.
24015 (vectorizable_assignment): Likewise.
24016 (vectorizable_shift): Likewise.
24017 (vectorizable_operation): Likewise.
24018 (vectorizable_store): Likewise.
24019 (vectorizable_load): Likewise.
24020 * tree-vectorizer.h: Likewise.
24022 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24024 * passes.c (emergency_dump_function): New.
24025 * tree-pass.h (emergency_dump_function): Declare.
24026 * plugin.c (plugins_internal_error_function): Remove.
24027 * plugin.h (plugins_internal_error_function): Remove declaration.
24028 * toplev.c (internal_error_function): New static function. Use it...
24029 (general_init): ...here.
24031 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24033 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24034 (arc_rtx_costs): Add costs for new patterns.
24035 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24036 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24038 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24040 * tree-ssa-strlen.c (get_next_strinfo): New function.
24041 (get_stridx_plus_constant): Use it.
24042 (zero_length_string): Likewise.
24043 (adjust_related_strinfos): Likewise.
24044 (adjust_last_stmt): Likewise.
24046 2017-05-31 Richard Biener <rguenther@suse.de>
24049 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24050 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24052 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24054 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24055 loop_vinfo argument and use of dependence distance vectors.
24056 Check instead whether the two references differ only in their
24057 initial value and assume that they have the same alignment if the
24058 difference is a multiple of the vector alignment.
24059 (vect_analyze_data_refs_alignment): Update call accordingly.
24061 2017-05-31 Martin Liska <mliska@suse.cz>
24064 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24066 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24068 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24069 (create_intersect_range_checks): Move from ...
24070 * tree-data-ref.c (create_intersect_range_checks_index)
24071 (create_intersect_range_checks): ... to here.
24072 (create_runtime_alias_checks): New function factored from ...
24073 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24074 here. Call above function.
24075 * tree-data-ref.h (create_runtime_alias_checks): New function.
24077 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24079 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24080 segment length for dr_b and compute it in wide_int.
24082 2017-05-31 Richard Biener <rguenther@suse.de>
24084 PR tree-optimization/80906
24085 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24086 and pass through iv_map.
24087 (copy_bb_and_scalar_dependences): Adjust.
24088 (translate_pending_phi_nodes): Likewise.
24089 (copy_loop_close_phi_args): Handle code-generating IVs instead
24092 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24094 * diagnostic-color.c (color_dict): Add "type-diff".
24095 (parse_gcc_colors): Update comment.
24096 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24097 -fdiagnostics-show-template-tree and -fno-elide-type.
24098 (GCC_COLORS): Add type-diff to example.
24100 (-fdiagnostics-show-template-tree): New.
24101 (-fno-elide-type): New.
24102 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24103 the pp_format_decoder callback. Call any m_format_postprocessor's
24105 (pretty_printer::pretty_printer): Initialize
24106 m_format_postprocessor.
24107 (pretty_printer::~pretty_printer): Delete any
24108 m_format_postprocessor.
24109 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24110 (class format_postprocessor): New class.
24111 (struct pretty_printer::format_decoder): Document the new parameters.
24112 (struct pretty_printer::m_format_postprocessor): New field.
24113 * tree-diagnostic.c (default_tree_printer): Update for new
24114 bool and const char ** params.
24115 * tree-diagnostic.h (default_tree_printer): Likewise.
24117 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24119 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24120 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24122 rs6000_gen_cell_microcode code.
24123 (rs6000_final_prescan_insn): Delete.
24124 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24125 "warn-cell-microcode" entries.
24126 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24127 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24128 throughout. Change cc_reg_not_micro_cr0_operand to
24129 cc_reg_not_cr0_operand throughout.
24130 (*extendhi<mode>2_noload): Delete.
24131 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24132 (mwarn-cell-microcode): Delete.
24133 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24134 -mgen-cell-microcode and -mwarn-cell-microcode.
24136 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
24139 * config/i386/constraints.md (Yd): New constraint.
24141 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
24142 and (?Yd, r) alternatives. Update insn attributes.
24143 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
24144 and (?*Yd, r) alternatives. Update insn attributes.
24145 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
24147 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
24149 * gimplify.c (gimplify_modify_expr): Don't create a
24150 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
24153 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
24155 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
24157 2017-05-30 Richard Biener <rguenther@suse.de>
24159 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
24160 and reduc_def fields.
24161 (STMT_VINFO_REDUC_TYPE): New define.
24162 (STMT_VINFO_REDUC_DEF): Likewise.
24163 (vect_force_simple_reduction): Adjust prototype.
24164 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
24165 (vect_is_simple_reduction): Remove check_reduction argument.
24166 (vect_force_simple_reduction): Adjust and set
24167 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24168 (vectorizable_reduction): Do not re-do reduction analysis
24169 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24170 * tree-parloops.c (gather_scalar_reductions): Adjust.
24172 2017-05-30 Richard Biener <rguenther@suse.de>
24174 PR middle-end/80901
24175 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
24178 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
24180 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
24181 Introduce unknown_misalignment parameter and remove vf.
24182 (vect_peeling_hash_get_lowest_cost):
24183 Pass unknown_misalignment parameter.
24184 (vect_enhance_data_refs_alignment):
24185 Fix unsupportable data ref treatment.
24187 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24189 * tree-vect-data-refs.c (vect_get_data_access_cost):
24190 Workaround for SLP handling.
24191 (vect_enhance_data_refs_alignment):
24192 Compute costs for doing no peeling at all, compare to the best
24193 peeling costs so far and avoid peeling if cheaper.
24195 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24197 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
24198 Return peeling info and set costs to zero for unlimited cost
24200 (vect_enhance_data_refs_alignment): Also inspect all datarefs
24201 with unknown misalignment. Compute and costs for unknown
24202 misalignment, compare them to the costs for known misalignment
24203 and choose the cheapest for peeling.
24205 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24207 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
24208 (vect_get_peeling_costs_all_drs): Create function.
24209 (vect_peeling_hash_get_lowest_cost):
24210 Use vect_get_peeling_costs_all_drs.
24211 (vect_peeling_supportable): Create function.
24212 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
24214 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24216 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
24217 DR_HAS_NEGATIVE_STEP.
24218 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
24219 (vect_enhance_data_refs_alignment): Use.
24220 (vect_duplicate_ssa_name_ptr_info): Use.
24221 * tree-vectorizer.h (dr_misalignment): Use.
24222 (known_alignment_for_access_p): Use.
24224 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24227 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
24229 (has_section_name): New function.
24231 2017-05-30 Martin Liska <mliska@suse.cz>
24234 * auto-profile.c (get_function_decl_from_block): Fix
24237 2017-05-30 Richard Biener <rguenther@suse.de>
24239 PR middle-end/80876
24240 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24242 2017-05-30 Martin Liska <mliska@suse.cz>
24244 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24245 * dumpfile.h (struct dump_file_info): Remove ctors.
24247 2017-05-30 Martin Liska <mliska@suse.cz>
24249 * predict.def: Fix GNU coding style.
24251 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24253 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24254 Mark 'to' argument with ATTRIBUTE_UNUSED.
24256 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24258 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24259 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24260 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24263 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24265 * doc/install.texi (Options specification): Restore entry of
24266 --enable-sjlj-exceptions.
24268 2017-05-27 Michael Eager <eager@eagercon.com>
24271 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24273 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24275 * config/microblaze/microblaze.h
24276 (FIXED_REGISTERS): Update in macro.
24277 (CALL_USED_REGISTERS): Update in macro.
24279 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24281 * doc/install.texi: Add links to macOS binary distributions.
24283 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24287 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24289 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24291 2017-05-26 Martin Liska <mliska@suse.cz>
24293 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24295 2017-05-26 Martin Liska <mliska@suse.cz>
24297 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24298 always leading ';; '.
24299 (dump_bb_info): Likewise.
24300 (brief_dump_cfg): Likewise.
24301 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24302 * dumpfile.c: Remove usage of TDF_VERBOSE.
24303 * dumpfile.h (enum dump_kind): Likewise.
24304 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24305 * print-tree.c (debug_verbose): Remove.
24306 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24307 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24308 * tree-diagnostic.c (default_tree_printer): Replace
24309 TDF_DIAGNOSTIC with TDF_SLIM.
24311 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24313 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24314 in parameter loop, rather than loop_vinfo.
24315 (create_intersect_range_checks): Ditto.
24316 (vect_create_cond_for_alias_checks): Update call to above functions.
24318 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24320 PR tree-optimization/80815
24321 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24322 for merging runtime alias checks. Handle negative DR_STEPs.
24324 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24326 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24328 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24329 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24330 out code pruning runtime alias checks.
24331 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24332 factored out from above.
24333 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24335 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24337 (prune_runtime_alias_test_list): New decalaration.
24339 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24341 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24342 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24343 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24344 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24345 (operator==, comp_dr_with_seg_len_pair): Ditto.
24346 (vect_prune_runtime_alias_test_list): Ditto.
24348 2017-05-26 Martin Liska <mliska@suse.cz>
24351 * params.def: Bound partial-inlining-entry-probability param.
24353 2017-05-26 Marek Polacek <polacek@redhat.com>
24356 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24359 2017-05-26 Richard Biener <rguenther@suse.de>
24361 PR tree-optimization/80842
24362 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24365 2017-05-26 Richard Biener <rguenther@suse.de>
24367 PR tree-optimization/80844
24368 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24370 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24372 * doc/md.texi (Machine Constraints): Update x86 family
24373 machine constraints section to match 'config/i386/constraints.md'.
24375 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24377 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24379 2017-05-25 Nathan Sidwell <nathan@acm.org>
24381 * doc/invoke.texi (--enable-languages): Update documentation.
24383 2017-05-25 Martin Liska <mliska@suse.cz>
24385 * dumpfile.c: Add TDF_FOLDING.
24386 * dumpfile.h (enum dump_kind): Likewise.
24387 * genmatch.c (dt_simplify::gen_1): Use it.
24389 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24391 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24393 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24395 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24396 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24398 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24400 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24401 * match.pd (X == C): Rewrite it here.
24402 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24403 with_certain_nonzero_bits2): New predicates.
24404 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24406 2017-05-24 Nathan Sidwell <nathan@acm.org>
24408 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24411 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24414 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24416 * config/powerpcspe: New port. Files are copied from the rs6000
24417 port, with "rs6000" in filenames replaced by "powerpcspe".
24419 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24421 PR rtl-optimization/80754
24422 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24424 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24426 * config/sparc/sparc.md (length): Return the correct value for -mflat
24427 sibcalls to match output_sibcall.
24429 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24433 * config/rs6000/rs6000.c (struct machine_function): Add new field
24435 (rs6000_get_separate_components): Init that field, use it.
24436 (rs6000_components_for_bb): Use the field.
24438 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
24440 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
24442 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
24444 PR middle-end/80823
24445 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
24447 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24450 * config/s390/s390.c (s390_check_qrst_address): Check incoming
24451 address against address_operand predicate.
24452 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
24454 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
24456 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
24457 parameters passed indirectly.
24459 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
24461 * config/i386/i386.md (*movdi_internal): Remove SSE4
24462 alternative 18 (?r, *v). Update insn attributes.
24463 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
24464 Update insn attributes.
24465 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
24466 Update insn attributes.
24467 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
24468 alternative 1 (r, v). Remove isa attribute.
24469 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
24470 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
24471 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
24473 2017-05-23 Tom de Vries <tom@codesourcery.com>
24475 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
24478 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24480 * cgraphunit.c (symbol_table::process_new_functions): Update.
24481 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
24482 (inline_generate_summary): Rename to ...
24483 (ipa_fn_summary_generate): ... this one.
24484 (inline_read_summary): Rename to ...
24485 (ipa_fn_summary_read): ... this one.
24486 (inline_write_summary): Rename to ...
24487 (ipa_fn_summary_write): ... this one.
24488 (inline_free_summary): Rename to ...
24489 (ipa_free_fn_summary): ... this one.
24490 (pass_data_local_fn_summary, pass_local_fn_summary,
24491 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
24492 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
24493 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
24494 make_pass_ipa_fn_summary): New.
24495 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
24496 inline_write_summary, inline_free_summary): Remove.
24497 (ipa_free_fn_summary) : New.
24498 * ipa-inline.c (ipa_inline): Update.
24499 (pass_ipa_inline): Do not generate summaries.
24500 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
24502 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
24503 and add pass_ipa_fn_summary.
24504 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
24506 (make_pass_inline_parameters): Remove.
24508 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
24510 * omp-low.c (struct omp_context): Remove "default_kind" member.
24513 * omp-offload.c (execute_oacc_device_lower): Remove the
24514 parallelism dimensions function attributes for unparallelized
24515 OpenACC kernels constructs.
24517 2017-05-23 Martin Liska <mliska@suse.cz>
24519 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
24521 (cgraph_edge::make_speculative): Likewise.
24522 (cgraph_edge::resolve_speculation): Likewise.
24523 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24524 (cgraph_node::dump): Likewise.
24525 * cgraph.h: Likewise.
24526 * cgraphunit.c (analyze_functions): Likewise.
24527 (symbol_table::compile): Likewise.
24528 * ipa-cp.c (print_all_lattices): Likewise.
24529 (determine_versionability): Likewise.
24530 (initialize_node_lattices): Likewise.
24531 (ipcp_verify_propagated_values): Likewise.
24532 (estimate_local_effects): Likewise.
24533 (update_profiling_info): Likewise.
24534 (create_specialized_node): Likewise.
24535 (perhaps_add_new_callers): Likewise.
24536 (decide_about_value): Likewise.
24537 (decide_whether_version_node): Likewise.
24538 (identify_dead_nodes): Likewise.
24539 (ipcp_store_bits_results): Likewise.
24540 * ipa-devirt.c (dump_targets): Likewise.
24541 (ipa_devirt): Likewise.
24542 * ipa-icf.c (sem_item::dump): Likewise.
24543 (sem_function::equals): Likewise.
24544 (sem_variable::equals): Likewise.
24545 (sem_item_optimizer::read_section): Likewise.
24546 (sem_item_optimizer::execute): Likewise.
24547 (congruence_class::dump): Likewise.
24548 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
24549 (dump_inline_summary): Likewise.
24550 (estimate_node_size_and_time): Likewise.
24551 (inline_analyze_function): Likewise.
24552 * ipa-inline-transform.c (inline_call): Likewise.
24553 * ipa-inline.c (report_inline_failed_reason): Likewise.
24554 (want_early_inline_function_p): Likewise.
24555 (edge_badness): Likewise.
24556 (update_edge_key): Likewise.
24557 (inline_small_functions): Likewise.
24558 * ipa-profile.c (ipa_profile): Likewise.
24559 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
24560 (ipa_make_edge_direct_to_target): Likewise.
24561 (remove_described_reference): Likewise.
24562 (ipa_impossible_devirt_target): Likewise.
24563 (propagate_controlled_uses): Likewise.
24564 (ipa_print_node_params): Likewise.
24565 (ipcp_transform_function): Likewise.
24566 * ipa-pure-const.c (pure_const_read_summary): Likewise.
24567 (propagate_pure_const): Likewise.
24568 * ipa-reference.c (generate_summary): Likewise.
24569 (read_write_all_from_decl): Likewise.
24570 (propagate): Likewise.
24571 (ipa_reference_read_optimization_summary): Likewise.
24572 * ipa-utils.c (ipa_merge_profiles): Likewise.
24573 * ipa.c (walk_polymorphic_call_targets): Likewise.
24574 (symbol_table::remove_unreachable_nodes): Likewise.
24575 (ipa_single_use): Likewise.
24576 * passes.c (execute_todo): Likewise.
24577 * predict.c (drop_profile): Likewise.
24578 * symtab.c (symtab_node::get_dump_name): New function.
24579 (symtab_node::dump_name): Likewise.
24580 (symtab_node::dump_asm_name): Likewise.
24581 (symtab_node::dump_references): Likewise.
24582 (symtab_node::dump_referring): Likewise.
24583 (symtab_node::dump_base): Likewise.
24584 (symtab_node::debug_symtab): Likewise.
24585 * tree-sra.c (convert_callers_for_node): Likewise.
24586 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24587 * value-prof.c (init_node_map): Likewise.
24589 2017-05-23 Martin Liska <mliska@suse.cz>
24591 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24592 and symtab_node::debug_symtab to symbol_table::debug.
24593 * cgraphunit.c (analyze_functions): Use the renamed function.
24594 (symbol_table::compile): Likewise.
24595 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24596 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24597 * passes.c (execute_todo): Likewise.
24598 * symtab.c (symbol_table::dump): New function.
24599 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24601 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24603 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24604 that nonconst implies exec.
24606 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24608 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24609 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24610 (inline_edge_summary_vec): Turn into ...
24611 (ipa_call_summaries): ... this one.
24612 (redirect_to_unreachable, edge_set_predicate,
24613 evaluate_properties_for_edge, inline_summary_alloc,
24614 reset_ipa_call_summary, reset_inline_summary,
24615 inline_summary_t::duplicate): Update.
24616 (inline_edge_duplication_hook): Turn to ...
24617 (ipa_call_summary_t::duplicate): ... this one.
24618 (inline_edge_removal_hook): Turn to ...
24619 (ipa_call_summary_t::remove): ... this one.
24620 (dump_inline_edge_summary): Turn to ...
24621 (dump_ipa_call_summary): ... this one.
24622 (estimate_function_body_sizes): Update.
24623 (inline_update_callee_summaries): Update.
24624 (remap_edge_change_prob): Update.
24625 (remap_edge_summaries): Update.
24626 (inline_merge_summary): Update.
24627 (do_estimate_edge_time): Update.
24628 (inline_generate_summary): Update.
24629 (inline_read_section): Update.
24630 (inline_read_summary): Update.
24631 (inline_free_summary): Update.
24632 * ipa-inline.c (can_inline_edge_p): Update.
24633 (compute_inlined_call_time): Update.
24634 (want_inline_small_function_p): Update.
24635 (edge_badness): Update.
24636 (early_inliner): Update.
24637 * ipa-inline.h (inline_edge_summary): Turn to ...
24638 (ipa_call_summary): ... this one.
24639 (ipa_call_summary_t): New class.
24640 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24641 (ipa_call_summaries): New.
24642 (inline_edge_summary): Remove.
24643 (estimate_edge_growth): Update.
24644 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24645 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24646 * ipa-split.c (execute_split_functions): Update.
24647 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24649 2017-05-23 Tom de Vries <tom@codesourcery.com>
24651 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24652 attributes): Document rdrand effective target.
24654 2017-05-23 Tom de Vries <tom@codesourcery.com>
24656 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24657 attributes): Sort alphabetically.
24659 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24661 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24663 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24666 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24667 V2DF/V2DI splat into two separate patterns, one that handles
24668 registers, and the other that only handles memory. Drop support
24669 for splatting from a GPR on ISA 2.07 and then splitting the
24670 splat into direct move and splat.
24671 (vsx_splat_<mode>_reg): Likewise.
24672 (vsx_splat_<mode>_mem): Likewise.
24674 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24676 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24678 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24680 PR middle-end/80809
24681 * omp-low.c (finish_taskreg_remap): New function.
24682 (finish_taskreg_scan): If unit size of ctx->record_type
24683 is non-constant, unshare the size expression and replace
24684 decls in it with possible outer var refs.
24686 PR middle-end/80809
24687 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24688 GOVD_SHARED rather than GOVD_PRIVATE with it.
24689 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24690 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24692 PR middle-end/80853
24693 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24694 as last argument to build_outer_var_ref for pointer bases of array
24695 section reductions.
24697 2017-05-19 Martin Sebor <msebor@redhat.com>
24699 * print-tree.c (print_node): Print DECL_READ_P flag.
24701 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24703 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24704 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24705 * cgraph.c: Likewise.
24706 * cgraphunit.c: Likewise.
24707 * gengtype.c: Likewise.
24708 * ipa-cp.c: Likewise.
24709 * ipa-devirt.c: Likewise.
24710 * ipa-icf.c: Likewise.
24711 * ipa-predicate.c: Likewise.
24712 * ipa-profile.c: Likewise.
24713 * ipa-prop.c: Likewise.
24714 * ipa-split.c: Likewise.
24716 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24717 edge_predicate_pool, dump_inline_hints,
24718 inline_summary::account_size_time, redirect_to_unreachable,
24719 edge_set_predicate, set_hint_predicate,
24720 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24721 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24722 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24723 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24724 ipa_call_summary_t::remove, initialize_growth_caches,
24725 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24726 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24727 mark_modified, unmodified_parm_1, unmodified_parm,
24728 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24729 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24730 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24731 will_be_nonconstant_predicate, record_modified_bb_info,
24732 get_minimal_bb, record_modified, param_change_prob,
24733 phi_result_unknown_predicate, predicate_for_phi_result,
24734 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24735 estimate_function_body_sizes, compute_inline_parameters,
24736 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24737 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24738 inline_update_callee_summaries, remap_edge_change_prob,
24739 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24740 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24741 inline_analyze_function, inline_summary_t::insert,
24742 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24743 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24744 inline_free_summary): Move to ipa-fnsummary.h
24745 (predicate_t): Remove.
24746 * ipa-fnsummary.c: New file.
24747 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24748 (enum inline_hints_vals, inline_hints, agg_position_info,
24749 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24750 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24751 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24752 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24753 inline_read_summary, inline_write_summary, inline_free_summary,
24754 inline_analyze_function, initialize_inline_failed,
24755 inline_merge_summary, inline_update_overall_summary,
24756 compute_inline_parameters): Move to ipa-fnsummary.h
24757 * ipa-fnsummary.h: New file.
24758 * ipa-inline-transform.h: Include ipa-inline.h.
24759 * ipa-inline.c: LIkewise.
24761 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24763 * ipa-inline.c (edge_badness): Use inlined_time instead of
24764 inline_summaries->get.
24766 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24768 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24770 2017-05-22 Nathan Sidwell <nathan@acm.org>
24772 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24773 (fdump-lang): Document 'raw' option.
24774 * dumpfile.h (TDI_tu): Delete.
24775 * dumpfile.c (dump_files): Remove translation-unit.
24776 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24778 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24780 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24781 command option from $(AWK) call.
24782 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24783 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24784 [FORMAT]: Remove handling of variable.
24785 * config/avr/t-multilib: Regenerate.
24787 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24789 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24791 (dump_inline_summary): Do not print self_time.
24792 (estimate_function_body_sizes): Do not set self_time.
24793 (compute_inline_parameters): Likewise.
24794 (inline_read_section, inline_write_summary): Do not stream self_time.
24795 * ipa-inline.h (inline_summary): Drop self_time.
24797 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24799 * ipa-inline-analysis.c (account_size_time): Rename to ...
24800 (inline_summary::account_size_time): ... this one.
24801 (reset_ipa_call_summary): Turn to ...
24802 (ipa_call_summary::reset): ... this one.
24803 (reset_inline_summary): Turn to ...
24804 (inline_summary::reset): ... this one.
24805 (inline_summary_t::remove): Update.
24806 (inline_summary_t::duplicate): Update.
24807 (ipa_call_summary_t::remove): Update.
24808 (dump_inline_summary): Update.
24809 (estimate_function_body_sizes): Update.
24810 (compute_inline_parameters): Update.
24811 (estimate_node_size_and_time): Update.
24812 (inline_merge_summary): Update.
24813 (inline_update_overall_summary): Update.
24814 (inline_read_section): Update.
24815 (inline_write_summary): Update.
24816 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24817 add account_size_time and reset member functions.
24818 (ipa_call_summary): Add reset function.
24819 * ipa-predicate.h (predicate::operator &): Constify.
24821 2017-05-22 Richard Biener <rguenther@suse.de>
24823 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24825 2017-05-19 Jason Merrill <jason@redhat.com>
24827 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24829 2017-05-19 Marek Polacek <polacek@redhat.com>
24832 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24833 TYPE_OVERFLOW_WRAPS checks.
24835 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24837 * tree-core.h (enum omp_clause_default_kind): Add
24838 "OMP_CLAUSE_DEFAULT_PRESENT".
24839 * tree-pretty-print.c (dump_omp_clause): Handle it.
24840 * gimplify.c (enum gimplify_omp_var_data): Add
24841 "GOVD_MAP_FORCE_PRESENT".
24842 (gimplify_adjust_omp_clauses_1): Map it to
24843 "GOMP_MAP_FORCE_PRESENT".
24844 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24846 * gimplify.c (oacc_default_clause): Clarify.
24848 2017-05-19 Nathan Sidwell <nathan@acm.org>
24850 LANG_HOOK_REGISTER_DUMPS
24851 * toplev.c (general_init): Call register dump lang hook.
24852 * doc/invoke.texi: Document -fdump-lang option family.
24853 * dumpfile.c (dump_files): Remove class dump here.
24854 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24855 * dumpfile.h (tree_dump_index): Remove TDI_class.
24856 * langhooks-def.h (lhd_register_dumps): Declare.
24857 (LANG_HOOKS_REGISTER_DUMPS): Define.
24858 (LANG_HOOKS_INITIALIZER): Add it.
24859 * langhooks.c (lhd_register_dumps): Define.
24860 * langhooks.h (struct lang_hooks): Add register_dumps.
24862 2017-05-19 Nathan Sidwell <nathan@acm.org>
24864 * context.h (context::set_passes): New.
24865 * context.c (context::context): Do not create pass manager.
24866 * toplev.c (general_init): Create pass manager here.
24868 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24870 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24871 use this splitter if two add or or instructions would also work for
24872 the constant we want to generate.
24874 2017-05-19 Richard Biener <rguenther@suse.de>
24877 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24878 predicate evaluation.
24880 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24882 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24884 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24887 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24889 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24890 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24891 (inline_edge_summary_vec): Turn into ...
24892 (ipa_call_summaries): ... this one.
24893 (redirect_to_unreachable, edge_set_predicate,
24894 evaluate_properties_for_edge, inline_summary_alloc,
24895 reset_ipa_call_summary, reset_inline_summary,
24896 inline_summary_t::duplicate): Update.
24897 (inline_edge_duplication_hook): Turn to ...
24898 (ipa_call_summary_t::duplicate): ... this one.
24899 (inline_edge_removal_hook): Turn to ...
24900 (ipa_call_summary_t::remove): ... this one.
24901 (dump_inline_edge_summary): Turn to ...
24902 (dump_ipa_call_summary): ... this one.
24903 (estimate_function_body_sizes): Update.
24904 (inline_update_callee_summaries): Update.
24905 (remap_edge_change_prob): Update.
24906 (remap_edge_summaries): Update.
24907 (inline_merge_summary): Update.
24908 (do_estimate_edge_time): Update.
24909 (inline_generate_summary): Update.
24910 (inline_read_section): Update.
24911 (inline_read_summary): Update.
24912 (inline_free_summary): Update.
24913 * ipa-inline.c (can_inline_edge_p): Update.
24914 (compute_inlined_call_time): Update.
24915 (want_inline_small_function_p): Update.
24916 (edge_badness): Update.
24917 (early_inliner): Update.
24918 * ipa-inline.h (inline_edge_summary): Turn to ...
24919 (ipa_call_summary): ... this one.
24920 (ipa_call_summary_t): New class.
24921 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24922 (ipa_call_summaries): New.
24923 (inline_edge_summary): Remove.
24924 (estimate_edge_growth): Update.
24925 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24926 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24927 * ipa-split.c (execute_split_functions): Update.
24928 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24930 2017-05-19 Richard Biener <rguenther@suse.de>
24932 PR middle-end/80764
24933 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24935 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24937 * config/rs6000/rs6000.c (struct machine_function): Add field
24938 fpr_is_wrapped_separately.
24939 (rs6000_get_separate_components): Use 64 components. Handle the
24940 new FPR components.
24941 (rs6000_components_for_bb): Handle the FPR components.
24942 (rs6000_emit_prologue_components): Handle the FPR components.
24943 (rs6000_emit_epilogue_components): Handle the FPR components.
24944 (rs6000_set_handled_components): Handle the FPR components.
24945 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24946 that are already separately shrink-wrapped.
24947 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24948 that are already separately shrink-wrapped.
24950 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24953 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24956 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24957 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24958 cases where the register allocator uses a move and an offsettable
24959 memory operation to/from a FPR register on ISA 2.06/2.07.
24960 (define_peephole2 for Altivec d-form store): Likewise.
24962 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24965 * config/i386/mmx.md (*mov<mode>_internal): Enable
24966 alternatives 11, 12, 13 and 14 also for 32bit targets.
24967 Remove alternatives 15, 16, 17 and 18.
24968 * config/i386/sse.md (vec_concatv2di): Change
24969 alternative (!x, *y) to (x, ?!*Yn).
24971 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24973 * dumpfile.h (enum dump_kind): Remove stray comma.
24975 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24977 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24978 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24979 predicate::num_conditions
24980 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24981 (CHANGED): turn into predicate::changed.
24982 (agg_position_info): Move to ipa-predicate.h
24983 (add_condition, predicate::add_clause, predicate::operator &=,
24984 predicate::or_with, predicate::evaluate, predicate::probability,
24985 dump_condition, dump_clause, predicate::dump,
24986 predicate::remap_after_duplication, predicate::remap_after_inlining,
24987 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24988 (evaluate_conditions_for_known_args): Update.
24989 (set_cond_stmt_execution_predicate): Update.
24990 * ipa-inline.h: Include ipa-predicate.h
24991 (condition, inline_param_summary, conditions, agg_position_info,
24992 predicate): Move to ipa-predicate.h
24993 * ipa-predicate.c: New file.
24994 * ipa-predicate.h: New file.
24996 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24998 * final.c (leaf_function_p): Check we are not in a sequence.
25000 2017-05-18 Martin Liska <mliska@suse.cz>
25002 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25003 * dumpfile.c (dump_register): Use new enum dump_kind.
25004 (get_dump_file_name): Likewise.
25005 (dump_enable_all): Likewise.
25006 (dump_switch_p_1): Likewise.
25007 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25008 * dumpfile.h (enum dump_kind): New enum type.
25009 (struct dump_file_info): Create constructor and
25010 format fields and comments.
25011 * passes.c (pass_manager::register_one_dump_file):
25013 * statistics.c (statistics_early_init): Likewise.
25014 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25015 TDF_TREE with TDF_SLIM.
25016 (gather_memory_references_ref): Likewise.
25018 2017-05-18 Martin Liska <mliska@suse.cz>
25020 * vec.h (struct vnull): Use it.
25022 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25024 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25025 (true_predicate, false_predicate, true_predicate_p,
25026 false_predicate_p): Remove.
25027 (single_cond_predicate, not_inlined_predicate): Turn to member function
25029 (add_condition): Update.
25030 (add_clause): Turn to...
25031 (predicate::add_clause): ... this one; update; allow passing NULL
25033 (and_predicates): Turn to ...
25034 (predicate::operator &=): ... this one.
25035 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25036 (or_predicates): Turn to ...
25037 (predicate::or_with): ... this one.
25038 (evaluate_predicate): Turn to ...
25039 (predicate::evaluate): ... this one.
25040 (predicate_probability): Turn to ...
25041 (predicate::probability): ... this one.
25042 (dump_condition): Update.
25043 (dump_predicate): Turn to ...
25044 (predicate::dump): ... this one.
25045 (account_size_time): Update.
25046 (edge_set_predicate): Update.
25047 (set_hint_predicate): UPdate.
25048 (evaluate_conditions_for_known_args): Update.
25049 (evaluate_properties_for_edge): Update.
25050 (remap_predicate_after_duplication): Turn to...
25051 (predicate::remap_after_duplication): ... this one.
25052 (remap_hint_predicate_after_duplication): Update.
25053 (inline_summary_t::duplicate): UPdate.
25054 (dump_inline_edge_summary): Update.
25055 (dump_inline_summary): Update.
25056 (set_cond_stmt_execution_predicate): Update.
25057 (set_switch_stmt_execution_predicate): Update.
25058 (compute_bb_predicates): Update.
25059 (will_be_nonconstant_expr_predicate): Update.
25060 (will_be_nonconstant_predicate): Update.
25061 (phi_result_unknown_predicate): Update.
25062 (predicate_for_phi_result): Update.
25063 (array_index_predicate): Update.
25064 (estimate_function_body_sizes): Update.
25065 (estimate_node_size_and_time): Update.
25066 (estimate_ipcp_clone_size_and_time): Update.
25067 (remap_predicate): Rename to ...
25068 (predicate::remap_after_inlining): ... this one.
25069 (remap_hint_predicate): Update.
25070 (inline_merge_summary): Update.
25071 (inline_update_overall_summary): Update.
25072 (estimate_size_after_inlining): Update.
25073 (read_predicate): Rename to ...
25074 (predicate::stream_in): ... this one.
25075 (read_inline_edge_summary): Update.
25076 (write_predicate): Rename to ...
25077 (predicate::stream_out): ... this one.
25078 (write_inline_edge_summary): Update.
25079 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25080 (clause_t): Turn to uint32_t
25081 (predicate): Turn to class; implement constructor and operators
25083 (size_time_entry): Update.
25084 (inline_summary): Update.
25085 (inline_edge_summary): Update.
25087 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25089 * fold-const.c (fold_binary_loc): Move transformation...
25090 * match.pd (C - X CMP X): ... here.
25092 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25094 * config/sparc/sparc.c (sparc_option_override): Set function
25095 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25096 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25098 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25100 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25102 2017-05-18 Marek Polacek <polacek@redhat.com>
25105 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25106 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25107 gimple_assign_load_p.
25109 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25111 PR middle-end/80692
25112 * real.c (do_compare): Give decimal_do_compare preference over
25113 comparing just the signs.
25115 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25117 * doc/md.texi (Canonicalization of Instructions): Describe the
25118 canonical form of instructions that inherently set a condition
25121 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25123 PR middle-end/80775
25124 * tree-cfg.c: Move deletion of unreachable case statements to after
25125 the merging of consecutive case labels.
25127 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25129 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25130 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25131 restoring of callee-saved registers.
25133 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
25135 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
25136 * config/visium/visium.c (single_set_and_flags): Likewise.
25137 * config/visium/visium.md (Substitutions): Likewise.
25139 2017-05-17 Martin Liska <mliska@suse.cz>
25141 * cfg.c: Introduce dump_flags_t type and
25142 use it instead of int type.
25144 * cfghooks.c: Likewise.
25145 * cfghooks.h (struct cfg_hooks): Likewise.
25146 * cfgrtl.c: Likewise.
25147 * cfgrtl.h: Likewise.
25148 * cgraph.c (cgraph_node::get_body): Likewise.
25149 * coretypes.h: Likewise.
25150 * domwalk.c: Likewise.
25151 * domwalk.h: Likewise.
25152 * dumpfile.c (struct dump_option_value_info): Likewise.
25153 (dump_enable_all): Likewise.
25154 (dump_switch_p_1): Likewise.
25155 (opt_info_switch_p): Likewise.
25156 * dumpfile.h (enum tree_dump_index): Likewise.
25157 (struct dump_file_info): Likewise.
25158 * genemit.c: Likewise.
25159 * generic-match-head.c: Likewise.
25160 * gengtype.c (open_base_files): Likewise.
25161 * gimple-pretty-print.c: Likewise.
25162 * gimple-pretty-print.h: Likewise.
25163 * graph.c (print_graph_cfg): Likewise.
25164 * graphite-scop-detection.c (dot_all_sese): Likewise.
25165 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
25166 * loop-unroll.c (report_unroll): Likewise.
25167 * passes.c (pass_manager::register_one_dump_file): Likewise.
25168 * print-tree.c: Likewise.
25169 * statistics.c: Likewise.
25170 * tree-cfg.c: Likewise.
25171 * tree-cfg.h: Likewise.
25172 * tree-dfa.c: Likewise.
25173 * tree-dfa.h: Likewise.
25174 * tree-dump.c (dump_function): Likewise.
25175 * tree-dump.h (struct dump_info): Likewise.
25176 * tree-pretty-print.c: Likewise.
25177 * tree-pretty-print.h: Likewise.
25178 * tree-ssa-live.c: Likewise.
25179 * tree-ssa-live.h: Likewise.
25180 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
25181 * tree-vect-loop.c: Likewise.
25182 * tree-vect-slp.c: Likewise.
25184 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
25185 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25187 PR tree-optimization/80457
25188 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
25189 of all arguments to a statement as scalar_to_vec operations.
25190 (vectorizable_call): Adjust call to vect_model_simple_cost for
25192 (vectorizable_conversion): Likewise.
25193 (vectorizable_assignment): Likewise.
25194 (vectorizable_shift): Likewise.
25195 (vectorizable_operation): Likewise.
25196 (vectorizable_comparison): Likewise.
25197 (vect_is_simple_cond): Record the def types for operands.
25198 (vectorizable_condition): Likewise, call vect_model_simple_cost.
25199 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
25200 for statement argument count.
25202 2017-05-16 Carl Love <cel@us.ibm.com>
25204 * config/rs6000/rs6000-c: Add support for built-in functions
25205 vector unsigned long long vec_bperm (vector unsigned long long,
25206 vector unsigned char)
25207 vector signed long long vec_mule (vector signed int,
25209 vector unsigned long long vec_mule (vector unsigned int,
25210 vector unsigned int)
25211 vector signed long long vec_mulo (vector signed int,
25213 vector unsigned long long vec_mulo (vector unsigned int,
25214 vector unsigned int)
25215 vector signed char vec_sldw (vector signed char,
25216 vector signed char,
25218 vector unsigned char vec_sldw (vector unsigned char,
25219 vector unsigned char,
25221 vector signed short vec_sldw (vector signed short,
25222 vector signed short,
25224 vector unsigned short vec_sldw (vector unsigned short,
25225 vector unsigned short,
25227 vector signed int vec_sldw (vector signed int,
25230 vector unsigned int vec_sldw (vector unsigned int,
25231 vector unsigned int,
25233 vector signed long long vec_sldw (vector signed long long,
25234 vector signed long long,
25236 vector unsigned long long vec_sldw (vector unsigned long long,
25237 vector unsigned long long,
25239 * config/rs6000/rs6000-c: Add support for built-in functions
25240 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25241 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25242 * doc/extend.texi: Update the built-in documentation for the
25243 new built-in functions.
25245 2017-05-16 Marek Polacek <polacek@redhat.com>
25249 * tree.c (save_expr): Don't fold the expression.
25251 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25253 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25254 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25255 and (?*y,m). Update insn attributes.
25257 2017-05-16 Martin Liska <mliska@suse.cz>
25259 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25260 flags argument of print_gimple_stmt, print_gimple_expr,
25261 print_generic_stmt and print_generic_expr.
25262 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25263 * coretypes.h: Likewise.
25264 * except.c (dump_eh_tree): Likewise.
25265 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25266 * gimple-pretty-print.h: Likewise.
25267 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25268 (backprop::push_to_worklist): Likewise.
25269 (backprop::pop_from_worklist): Likewise.
25270 (backprop::process_use): Likewise.
25271 (backprop::intersect_uses): Likewise.
25272 (note_replacement): Likewise.
25273 * gimple-ssa-store-merging.c
25274 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25275 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25276 (pass_store_merging::execute): Likewise.
25277 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25278 (ssa_base_cand_dump_callback): Likewise.
25279 (dump_incr_vec): Likewise.
25280 (replace_refs): Likewise.
25281 (replace_mult_candidate): Likewise.
25282 (create_add_on_incoming_edge): Likewise.
25283 (create_phi_basis): Likewise.
25284 (insert_initializers): Likewise.
25285 (all_phi_incrs_profitable): Likewise.
25286 (introduce_cast_before_cand): Likewise.
25287 (replace_one_candidate): Likewise.
25288 * gimplify.c (gimplify_expr): Likewise.
25289 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25290 (set_rename): Likewise.
25291 (rename_uses): Likewise.
25292 (copy_loop_phi_nodes): Likewise.
25293 (add_close_phis_to_merge_points): Likewise.
25294 (copy_loop_close_phi_args): Likewise.
25295 (copy_cond_phi_args): Likewise.
25296 (graphite_copy_stmts_from_block): Likewise.
25297 (translate_pending_phi_nodes): Likewise.
25298 * graphite-poly.c (print_pdr): Likewise.
25299 (dump_gbb_cases): Likewise.
25300 (dump_gbb_conditions): Likewise.
25301 (print_scop_params): Likewise.
25302 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25303 (build_cross_bb_scalars_use): Likewise.
25304 (gather_bbs::before_dom_children): Likewise.
25305 * hsa-dump.c (dump_hsa_immed): Likewise.
25306 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25307 (get_replacement_map): Likewise.
25308 * ipa-inline-analysis.c (dump_condition): Likewise.
25309 (estimate_function_body_sizes): Likewise.
25310 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25311 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25312 * ipa-prop.c (ipa_dump_param): Likewise.
25313 (ipa_print_node_jump_functions_for_edge): Likewise.
25314 (ipa_modify_call_arguments): Likewise.
25315 (ipa_modify_expr): Likewise.
25316 (ipa_dump_param_adjustments): Likewise.
25317 (ipa_dump_agg_replacement_values): Likewise.
25318 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25319 * ipa-pure-const.c (check_stmt): Likewise.
25320 (pass_nothrow::execute): Likewise.
25321 * ipa-split.c (execute_split_functions): Likewise.
25322 * omp-offload.c (dump_oacc_loop_part): Likewise.
25323 (dump_oacc_loop): Likewise.
25324 * trans-mem.c (tm_log_emit): Likewise.
25325 (tm_memopt_accumulate_memops): Likewise.
25326 (dump_tm_memopt_set): Likewise.
25327 (dump_tm_memopt_transform): Likewise.
25328 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25329 (print_loop): Likewise.
25330 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25331 (chkp_gather_checks_info): Likewise.
25332 (chkp_get_check_result): Likewise.
25333 (chkp_remove_check_if_pass): Likewise.
25334 (chkp_use_outer_bounds_if_possible): Likewise.
25335 (chkp_reduce_bounds_lifetime): Likewise.
25336 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25337 (chkp_mark_completed_bounds): Likewise.
25338 (chkp_register_incomplete_bounds): Likewise.
25339 (chkp_mark_invalid_bounds): Likewise.
25340 (chkp_maybe_copy_and_register_bounds): Likewise.
25341 (chkp_build_returned_bound): Likewise.
25342 (chkp_get_bound_for_parm): Likewise.
25343 (chkp_build_bndldx): Likewise.
25344 (chkp_get_bounds_by_definition): Likewise.
25345 (chkp_generate_extern_var_bounds): Likewise.
25346 (chkp_get_bounds_for_decl_addr): Likewise.
25347 * tree-chrec.c (chrec_apply): Likewise.
25348 * tree-data-ref.c (dump_data_reference): Likewise.
25349 (dump_subscript): Likewise.
25350 (dump_data_dependence_relation): Likewise.
25351 (analyze_overlapping_iterations): Likewise.
25352 * tree-inline.c (expand_call_inline): Likewise.
25353 (tree_function_versioning): Likewise.
25354 * tree-into-ssa.c (dump_defs_stack): Likewise.
25355 (dump_currdefs): Likewise.
25356 (dump_names_replaced_by): Likewise.
25357 (dump_update_ssa): Likewise.
25358 (update_ssa): Likewise.
25359 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25360 * tree-parloops.c (build_new_reduction): Likewise.
25361 (try_create_reduction_list): Likewise.
25362 (ref_conflicts_with_region): Likewise.
25363 (oacc_entry_exit_ok_1): Likewise.
25364 (oacc_entry_exit_single_gang): Likewise.
25365 * tree-pretty-print.h: Likewise.
25366 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25367 (get_scalar_evolution): Likewise.
25368 (add_to_evolution): Likewise.
25369 (get_loop_exit_condition): Likewise.
25370 (analyze_evolution_in_loop): Likewise.
25371 (analyze_initial_condition): Likewise.
25372 (analyze_scalar_evolution): Likewise.
25373 (instantiate_scev): Likewise.
25374 (number_of_latch_executions): Likewise.
25375 (gather_chrec_stats): Likewise.
25376 (final_value_replacement_loop): Likewise.
25377 (scev_const_prop): Likewise.
25378 * tree-sra.c (dump_access): Likewise.
25379 (disqualify_candidate): Likewise.
25380 (create_access): Likewise.
25381 (reject): Likewise.
25382 (maybe_add_sra_candidate): Likewise.
25383 (create_access_replacement): Likewise.
25384 (analyze_access_subtree): Likewise.
25385 (analyze_all_variable_accesses): Likewise.
25386 (sra_modify_assign): Likewise.
25387 (initialize_constant_pool_replacements): Likewise.
25388 (find_param_candidates): Likewise.
25389 (decide_one_param_reduction): Likewise.
25390 (replace_removed_params_ssa_names): Likewise.
25391 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25392 * tree-ssa-copy.c (dump_copy_of): Likewise.
25393 (copy_prop_visit_cond_stmt): Likewise.
25394 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25395 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25396 (record_equivalences_from_stmt): Likewise.
25397 * tree-ssa-dse.c (compute_trims): Likewise.
25398 (delete_dead_call): Likewise.
25399 (delete_dead_assignment): Likewise.
25400 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25401 (forward_propagate_into_cond): Likewise.
25402 (pass_forwprop::execute): Likewise.
25403 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25404 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25406 (move_computations_worker): Likewise.
25407 (execute_sm): Likewise.
25408 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25409 (remove_exits_and_undefined_stmts): Likewise.
25410 (remove_redundant_iv_tests): Likewise.
25411 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25412 (adjust_iv_update_pos): Likewise.
25413 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25414 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25415 (value_replacement): Likewise.
25416 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25417 * tree-ssa-pre.c (print_pre_expr): Likewise.
25418 (get_representative_for): Likewise.
25419 (create_expression_by_pieces): Likewise.
25420 (insert_into_preds_of_block): Likewise.
25421 (eliminate_insert): Likewise.
25422 (eliminate_dom_walker::before_dom_children): Likewise.
25423 (eliminate): Likewise.
25424 (remove_dead_inserted_code): Likewise.
25425 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25426 * tree-ssa-reassoc.c (get_rank): Likewise.
25427 (eliminate_duplicate_pair): Likewise.
25428 (eliminate_plus_minus_pair): Likewise.
25429 (eliminate_not_pairs): Likewise.
25430 (undistribute_ops_list): Likewise.
25431 (eliminate_redundant_comparison): Likewise.
25432 (update_range_test): Likewise.
25433 (optimize_range_tests_var_bound): Likewise.
25434 (optimize_vec_cond_expr): Likewise.
25435 (rewrite_expr_tree): Likewise.
25436 (rewrite_expr_tree_parallel): Likewise.
25437 (linearize_expr): Likewise.
25438 (break_up_subtract): Likewise.
25439 (linearize_expr_tree): Likewise.
25440 (attempt_builtin_powi): Likewise.
25441 (attempt_builtin_copysign): Likewise.
25442 (transform_stmt_to_copy): Likewise.
25443 (transform_stmt_to_multiply): Likewise.
25444 (dump_ops_vector): Likewise.
25445 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
25446 (print_scc): Likewise.
25447 (set_ssa_val_to): Likewise.
25448 (visit_reference_op_store): Likewise.
25449 (visit_use): Likewise.
25450 (sccvn_dom_walker::before_dom_children): Likewise.
25451 (run_scc_vn): Likewise.
25452 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
25454 (expr_hash_elt::print): Likewise.
25455 (const_and_copies::pop_to_marker): Likewise.
25456 (const_and_copies::record_const_or_copy_raw): Likewise.
25457 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
25458 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
25459 (dump_predicates): Likewise.
25460 (find_uninit_use): Likewise.
25461 (warn_uninitialized_phi): Likewise.
25462 (pass_late_warn_uninitialized::execute): Likewise.
25463 * tree-ssa.c (verify_vssa): Likewise.
25464 (verify_ssa): Likewise.
25465 (maybe_optimize_var): Likewise.
25466 * tree-vrp.c (dump_value_range): Likewise.
25467 (dump_all_value_ranges): Likewise.
25468 (dump_asserts_for): Likewise.
25469 (register_edge_assert_for_2): Likewise.
25470 (vrp_visit_cond_stmt): Likewise.
25471 (vrp_visit_switch_stmt): Likewise.
25472 (vrp_visit_stmt): Likewise.
25473 (vrp_visit_phi_node): Likewise.
25474 (simplify_cond_using_ranges_1): Likewise.
25475 (fold_predicate_in): Likewise.
25476 (evrp_dom_walker::before_dom_children): Likewise.
25477 (evrp_dom_walker::push_value_range): Likewise.
25478 (evrp_dom_walker::pop_value_range): Likewise.
25479 (execute_early_vrp): Likewise.
25481 2017-05-16 Richard Biener <rguenther@suse.de>
25483 * dwarf2out.c (loc_list_from_tree_1): Do not create
25484 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
25486 2017-05-16 Richard Biener <rguenther@suse.de>
25488 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
25490 (note_variable_value_in_expr): If we resolved the decl ref
25491 do not push to the stack.
25493 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
25495 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
25496 operations in fast-math mode.
25497 (vaddq_f16): Likewise.
25498 (vmul_f16): Likewise.
25499 (vmulq_f16): Likewise.
25500 (vsub_f16): Likewise.
25501 (vsubq_f16): Likewise.
25502 * config/arm/neon.md (add<mode>3): New.
25504 (fma:<VH:mode>3): New. Also remove outdated comment.
25507 2017-05-16 Martin Liska <mliska@suse.cz>
25511 * ipa-devirt.c (warn_types_mismatch): Fix typo.
25512 (odr_types_equivalent_p): Likewise.
25514 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
25516 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
25518 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
25521 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
25522 non-interunit SSE move alternatives with '?'.
25523 (zero-extendsidi peephole2): New peephole to skip intermediate
25524 general register in SSE zero-extend sequence.
25526 2017-05-15 Jeff Law <law@redhat.com>
25528 * reorg.c (relax_delay_slots): Create a new variable to hold
25529 the temporary target rather than clobbering TARGET_LABEL.
25531 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
25532 missing argument to extract_bit_field call.
25533 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
25535 2017-05-15 Martin Liska <mliska@suse.cz>
25538 * gcc.c (process_command): Do not allow empty argument of -o option.
25540 2017-05-15 Renlin Li <renlin.li@arm.com>
25542 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
25543 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
25544 * config/aarch64/constraints.md (Usf): Add long call check.
25545 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
25546 (call_value): Likewise.
25547 (sibcall): Likewise.
25548 (sibcall_value): Likewise.
25550 (call_value_insn): New.
25551 (sibcall_insn): Update rtx pattern.
25552 (sibcall_value_insn): Likewise.
25553 (call_internal): Remove.
25554 (call_value_internal): Likewise.
25555 (sibcall_internal): Likewise.
25556 (sibcall_value_internal): Likewise.
25557 (call_reg): Likewise.
25558 (call_symbol): Likewise.
25559 (call_value_reg): Likewise.
25560 (call_value_symbol): Likewise.
25562 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
25565 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
25567 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
25569 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
25570 compatible with CCGOCmode and with CCZmode.
25572 2017-05-14 Martin Sebor <msebor@redhat.com>
25574 PR middle-end/77671
25575 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25576 (gimple_fold_builtin_snprintf): Same.
25577 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25578 (gimple_fold_builtin_snprintf): Same.
25579 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25580 of character types.
25581 (is_call_safe): New function.
25582 (try_substitute_return_value): Call it.
25583 (try_simplify_call): New function.
25584 (pass_sprintf_length::handle_gimple_call): Call it.
25586 2017-05-14 Martin Sebor <msebor@redhat.com>
25588 PR middle-end/80669
25589 * builtins.c (expand_builtin_stpncpy): Simplify.
25591 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25593 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25594 * config/i386/i386.h
25595 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25596 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25597 (struct machine_function): Add new members call_ms2sysv,
25598 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25599 (struct machine_frame_state): New fields sp_realigned and
25600 sp_realigned_offset.
25601 * config/i386/i386.c
25602 (enum xlogue_stub): New enum.
25603 (enum xlogue_stub_sets): New enum.
25604 (class xlogue_layout): New class.
25605 (struct ix86_frame): New fields stack_realign_allocate_offset,
25606 stack_realign_offset and outlined_save_offset. Modify comments to
25607 detail stack layout when using out-of-line stubs.
25608 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25609 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25610 -mcall-ms2sysv-xlogues.
25611 (stub_managed_regs): New static variable.
25612 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25613 registers managed by out-of-line stub.
25614 (disable_call_ms2sysv_xlogues): New function.
25615 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25616 m->call_ms2sysv when appropriate and compute frame layout for
25618 (sp_valid_at, fp_valid_at): New inline functions.
25619 (choose_basereg): New function.
25620 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25622 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25623 Use align parameter of choose_baseaddr to generated aligned SSE movs
25625 (pro_epilogue_adjust_stack): Modify to track
25626 machine_frame_state::sp_realigned.
25627 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25628 (ix86_nsaved_sseregs): Likewise.
25629 (ix86_emit_save_regs): Likewise.
25630 (ix86_emit_save_regs_using_mov): Likewise.
25631 (ix86_emit_save_sse_regs_using_mov): Likewise.
25632 (get_scratch_register_on_entry): Likewise.
25633 (gen_frame_set): New function.
25634 (gen_frame_load): Likewise.
25635 (gen_frame_store): Likewise.
25636 (emit_outlined_ms2sysv_save): Likewise.
25637 (emit_outlined_ms2sysv_restore): Likewise.
25638 (ix86_expand_prologue): Modify stack re-alignment code and call
25639 emit_outlined_ms2sysv_save when appropriate.
25640 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25641 parameter rtx_insn *insn, which allows the function to be used to only
25642 generate the notes.
25643 (ix86_expand_epilogue): Modify validity checks of frame and stack
25644 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25645 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25646 * config/i386/predicates.md
25647 (save_multiple): New predicate.
25648 (restore_multiple): Likewise.
25649 * config/i386/sse.md
25650 (save_multiple<mode>): New pattern.
25651 (save_multiple_realign<mode>): Likewise.
25652 (restore_multiple<mode>): Likewise.
25653 (restore_multiple_and_return<mode>): Likewise.
25654 (restore_multiple_leave_return<mode>): Likewise.
25655 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25657 2017-05-14 Julia Koval <julia.koval@intel.com>
25659 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25660 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25661 (__builtin_ia32_xsetbv): New builtins.
25662 * config/i386/i386.c (ix86_expand_special_args_builtin):
25664 (ix86_expand_builtin): Special expand for new intrinsics.
25665 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25666 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25667 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25669 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25671 * cfganal.c (inverted_post_order_compute): Change argument type
25673 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25674 * df-core.c (rest_of_handle_df_initialize): Adjust.
25675 (rest_of_handle_df_finish): Likewise.
25676 (df_analyze_1): Likewise.
25677 (df_analyze): Likewise.
25678 (loop_inverted_post_order_compute): Change argument to be a vec *.
25679 (df_analyze_loop): Adjust.
25680 (df_get_n_blocks): Likewise.
25681 (df_get_postorder): Likewise.
25682 * df.h (struct df_d): Change field to be a vec.
25683 * lcm.c (compute_laterin): Adjust.
25684 (compute_available): Likewise.
25685 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25686 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25687 * tree-ssa-pre.c (compute_antic): Likewise.
25689 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25691 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25692 (depth_first_search::depth_first_search): Change structure init
25693 function to this constructor.
25694 (depth_first_search::add_bb): Rename function to this member.
25695 (depth_first_search::execute): Likewise.
25696 (flow_dfs_compute_reverse_finish): Adjust.
25698 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25700 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25701 (longest_simple_path): Likewise.
25702 * shrink-wrap.c (spread_components): Likewise.
25703 (disqualify_problematic_components): Likewise.
25704 (emit_common_heads_for_components): Likewise.
25705 (emit_common_tails_for_components): Likewise.
25706 (insert_prologue_epilogue_for_components): Likewise.
25708 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25710 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25713 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25715 * df-core.c (df_set_blocks): Start using auto_bitmap.
25716 (df_compact_blocks): Likewise.
25717 * df-problems.c (df_rd_confluence_n): Likewise.
25718 * df-scan.c (df_insn_rescan_all): Likewise.
25719 (df_process_deferred_rescans): Likewise.
25720 (df_update_entry_block_defs): Likewise.
25721 (df_update_exit_block_uses): Likewise.
25722 (df_entry_block_bitmap_verify): Likewise.
25723 (df_exit_block_bitmap_verify): Likewise.
25724 (df_scan_verify): Likewise.
25725 * lra-constraints.c (lra_constraints): Likewise.
25726 (undo_optional_reloads): Likewise.
25727 (lra_undo_inheritance): Likewise.
25728 * lra-remat.c (calculate_gen_cands): Likewise.
25729 (do_remat): Likewise.
25730 * lra-spills.c (assign_spill_hard_regs): Likewise.
25731 (spill_pseudos): Likewise.
25732 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25733 (bitmap_set_subtract_values): Likewise.
25735 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25737 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25738 management with auto_bitmap.
25739 (fix_inter_tick): Likewise.
25740 (fix_recovery_deps): Likewise.
25741 * ira.c (add_store_equivs): Likewise.
25742 (find_moveable_pseudos): Likewise.
25743 (split_live_ranges_for_shrink_wrap): Likewise.
25744 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25745 (rtx_reuse_manager::seen_def_p): Likewise.
25746 (rtx_reuse_manager::set_seen_def): Likewise.
25747 * print-rtl.h (class rtx_reuse_manager): Likewise.
25749 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25751 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25753 (migrate_btr_def): Likewise.
25754 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25755 * df-core.c (loop_post_order_compute): Likewise.
25756 (loop_inverted_post_order_compute): Likewise.
25757 * hsa-common.h: Likewise.
25758 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25759 * init-regs.c (initialize_uninitialized_regs): Likewise.
25760 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25761 (inline_small_functions): Likewise.
25762 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25763 * ira.c (combine_and_move_insns): Likewise.
25764 (build_insn_chain): Likewise.
25765 * loop-invariant.c (find_invariants): Likewise.
25766 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25767 * predict.c (tree_predict_by_opcode): Likewise.
25768 (predict_paths_leading_to): Likewise.
25769 (predict_paths_leading_to_edge): Likewise.
25770 (estimate_loops_at_level): Likewise.
25771 (estimate_loops): Likewise.
25772 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25773 (spread_components): Likewise.
25774 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25775 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25776 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25777 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25778 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25779 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25780 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25781 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25782 (mark_threaded_blocks): Likewise.
25783 (thread_through_all_blocks): Likewise.
25784 * tree-ssa.c (verify_ssa): Likewise.
25785 (execute_update_addresses_taken): Likewise.
25786 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25788 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25790 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25792 (post_order_compute): Likewise.
25793 (inverted_post_order_compute): Likewise.
25794 (pre_and_rev_post_order_compute_fn): Likewise.
25796 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25798 * genrecog.c (int_set::int_set): Explicitly construct our
25799 auto_vec base class.
25800 * vec.h (auto_vec::auto_vec): New constructor.
25802 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25804 * bitmap.h (class auto_bitmap): New constructor taking
25805 bitmap_obstack * argument.
25807 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25809 * bitmap.h (class auto_bitmap): Change type of m_bits to
25810 bitmap_head, and adjust ctor / dtor and member operators.
25812 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25814 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25815 when returned register mode doesn't match original mode.
25817 2017-05-12 Jeff Law <law@redhat.com>
25818 Jakub Jelinek <jakub@redhat.com>
25820 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25821 we look for cc setter after the compare-elim changes.
25822 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25823 within the vector to match what compare-elim now expects.
25824 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25825 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25827 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25828 after the compare-elim changes.
25829 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25830 the vector to match what compare-elim now expects.
25831 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25832 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25833 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25834 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25835 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25837 * config/visium/visium.c (single_set_and_flags): Fix where
25838 we look for cc setter after the compare-elim changes.
25839 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25840 with the vector to match what compare-elim now expects.
25841 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25842 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25843 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25844 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25845 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25846 (neg<mode>2_insn_set_overflow): Likewise.
25848 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25850 PR middle-end/79794
25851 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25852 maybe_expand_insn call, set ops[0].target. If still set after call,
25853 set alt_rtl. Add extra arg to recursive calls.
25854 (extract_bit_field): Add alt_rtl argument. Pass to
25856 * expmed.h (extract_bit_field): Fix prototype.
25857 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25858 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25859 to extract_bit_field_calls.
25860 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25861 Pass alt_rtl to extract_bit_field calls.
25862 * calls.c (store_unaligned_arguments_into_psuedos)
25863 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25864 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25866 * optabs.h (struct expand_operand): Add target bitfield.
25868 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25870 * compare-elim.c (try_eliminate_compare): Canonicalize
25871 operation with embedded compare to
25872 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25873 (set (reg) (operation)].
25875 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25877 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25880 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25881 cost of adding a carry flag for ADC instruction.
25882 [case MINUS]: Ignore the cost of subtracting a carry flag
25883 for SBB instruction.
25885 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25887 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25889 * config/rs6000/bmiintrin.h: New file.
25890 * config/rs6000/bmi2intrin.h: New file.
25891 * config/rs6000/x86intrin.h: New file.
25893 2017-05-12 Jeff Law <law@redhat.com>
25895 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25898 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25900 PR middle-end/80707
25901 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25903 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25905 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25906 early expansion of vector divide builtins.
25907 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25908 builtins identified as having unsigned arguments.
25910 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25912 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25913 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25914 expansion of vector logical operations (and, andc, or, xor,
25917 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25919 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25920 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25922 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25924 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25925 early GIMPLE expansion of vector multiplies.
25927 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25929 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25930 TARGET_HAVE_MOVT conditional.
25931 (movt splitter): Likewise.
25933 2017-05-12 Richard Biener <rguenther@suse.de>
25935 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25936 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25937 Fold all stmts not inplace.
25939 2017-05-12 Richard Biener <rguenther@suse.de>
25941 PR tree-optimization/80713
25942 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25943 inserted_exprs bit for not removed stmts.
25945 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25947 PR middle-end/69921
25948 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25949 parallelized" attribute for parallelized OpenACC kernels.
25950 * omp-offload.c (execute_oacc_device_lower): Use it.
25952 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25953 Set "oacc kernels" attribute.
25954 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25955 parameter. Adjust all users.
25956 (oacc_fn_attrib_kernels_p): Remove function.
25957 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25958 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25959 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25960 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25961 assert "oacc kernels" attribute is set.
25963 2017-05-11 Carl Love <cel@us.ibm.com>
25965 * config/rs6000/rs6000-c: Add support for built-in functions
25966 vector unsigned char vec_popcnt (vector signed char)
25967 vector unsigned char vec_popcnt (vector unsigned char)
25968 vector unsigned short vec_popcnt (vector signed short)
25969 vector unsigned short vec_popcnt (vector unsigned short)
25970 vector unsigned int vec_popcnt (vector signed int)
25971 vector unsigned int vec_popcnt (vector unsigned int)
25972 vector unsigned long long vec_popcnt (vector signed long long)
25973 vector unsigned long long vec_popcnt (vector unsigned long long)
25974 vector signed long long vec_slo (vector signed long long,
25975 vector signed char)
25976 vector signed long long vec_slo (vector signed long long,
25977 vector unsigned char)
25978 vector unsigned long long vec_slo (vector unsigned long long,
25979 vector signed char)
25980 vector unsigned long long vec_slo (vector unsigned long long,
25981 vector unsigned char)
25982 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25983 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25984 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25985 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25986 * doc/extend.texi: Update the built-in documentation file for the
25987 new built-in functions.
25989 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25991 * attribs.h (sorted_attr_string): Move machine independent
25992 functions for target clone support from the i386 port to common
25993 code. Rename ix86_function_versions to common_function_versions.
25994 Rename make_name to make_unique_name.
25995 (common_function_versions): Likewise.
25996 (make_unique_name): Likewise.
25997 (make_dispatcher_decl): Likewise.
25998 (is_function_default_version): Likewise.
25999 * attribs.c (attr_strcmp): Likewise.
26000 (sorted_attr_string): Likewise.
26001 (common_function_versions): Likewise.
26002 (make_unique_name): Likewise.
26003 (make_dispatcher_decl): Likewise.
26004 (is_function_default_version): Likewise.
26005 * config/i386/i386.c (attr_strcmp): Likewise.
26006 (sorted_attr_string): Likewise.
26007 (ix86_function_versions): Likewise.
26008 (make_name): Likewise.
26009 (make_dispatcher_decl): Likewise.
26010 (is_function_default_version): Likewise.
26011 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26013 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26016 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26017 Account for direct move costs for vec_construct of integer
26020 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26023 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26024 (UNSPEC_STX_ATOMIC): Ditto.
26025 (loaddi_via_sse): New insn.
26026 (storedi_via_sse): Ditto.
26027 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26028 Update corresponding peephole2 patterns.
26029 (atomic_storedi_fpu): Ditto.
26031 2017-05-11 Julia Koval <julia.koval@intel.com>
26033 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26034 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26036 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26037 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26038 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26040 2017-05-11 Nathan Sidwell <nathan@acm.org>
26042 * graphite-poly.c: Include dumpfile.h.
26044 * dumpfle.h (dump_function): Declare here ...
26045 * tree-dump.h (dump_function): ... not here.
26046 * dumpfile.c: #include tree-cfg.h.
26047 (dump_function): Move here from ...
26048 * tree-dump.c (dump_function): ... here.
26049 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26050 * graphite-poly.c: Don't include tree-dump.h.
26051 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26052 * print-tree.c: Likewise.
26053 * stor-layout.c: Likewise.
26054 * tree-nested.c: Likewise.
26056 * dumpfile.c (dump_start): Use TDF_FLAGS.
26057 (dump_enable_all): Fix TDF_KIND check thinko.
26059 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26061 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26062 array entries to represent two legal parameterizations of the
26063 overloaded __builtin_cmpb function, as represented by the
26064 P6_OV_BUILTIN_CMPB constant.
26065 (altivec_resolve_overloaded_builtin): Add special case handling
26066 for the __builtin_cmpb function, as represented by the
26067 P6_OV_BUILTIN_CMPB constant.
26068 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26069 (BU_P6_64BIT_2): New macro.
26070 (BU_P6_OVERLOAD_2): New macro
26071 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26072 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26073 (CMPB): Add overload support to represent both 32-bit and 64-bit
26074 compare-bytes function.
26075 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26076 support for TARGET_CMPB.
26077 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26078 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26079 documentation of the __builtin_cmpb overloaded built-in function.
26081 2017-05-11 Richard Biener <rguenther@suse.de>
26083 PR tree-optimization/80705
26084 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26085 bases are not vectorizable.
26087 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26089 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26090 when counting register pressure.
26092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26094 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26095 (iv_ca_more_deps): Renamed to ...
26096 (iv_ca_compare_deps): ... this.
26097 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26099 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26101 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26103 (determine_group_iv_costs): ... here.
26104 (find_inv_vars_cb): Record inv var if it's not recorded before.
26106 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26108 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26109 (get_shiftadd_cost): Ditto.
26111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26113 * tree-ssa-address.c: Include header file.
26114 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26116 (add_to_parts): Refactor.
26117 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26118 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26121 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26123 PR tree-optimization/53090
26124 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26126 (extract_cond_operands): Detect condition with IV on both sides
26127 and return COMP_IV_EXPR_2.
26128 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26129 (rewrite_use_compare): Simplify by removing call to function
26130 extract_cond_operands.
26132 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26134 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
26135 (extract_cond_operands): Detect condition comparing against non-
26136 invariant bound and return appropriate enum value.
26137 (find_interesting_uses_cond): Update use of extract_cond_operands.
26138 Handle its return value accordingly.
26139 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
26141 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26143 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
26144 nonlinear iv_use computation in loop invariant sensitive way.
26146 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26148 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
26149 (find_iv_candidates): Call relate_compare_use_with_all_cands.
26151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26153 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
26154 (dump_cand): Support iv_cand.inv_exprs.
26155 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
26157 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
26160 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26162 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
26164 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
26165 as local function. Include necessary header files.
26166 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
26168 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26170 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
26172 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26174 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
26175 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
26176 RSHIFT_EXPR and BIT_NOT_EXPR.
26178 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26180 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
26181 (adjust_setup_cost): New parameter supporting round up adjustment.
26182 (struct address_cost_data): Delete.
26183 (force_expr_to_var_cost): Don't bound cost with spill_cost.
26184 (split_address_cost, ptr_difference_cost): Delete.
26185 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
26186 (struct ainc_cost_data): New struct.
26187 (get_address_cost_ainc): New function.
26188 (get_address_cost, get_computation_cost): Reimplement.
26189 (determine_group_iv_cost_address): Record inv_expr for all uses of
26191 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
26192 (iv_ca_has_deps): Reimplemented to ...
26193 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
26195 (iv_ca_extend): Call iv_ca_more_deps.
26197 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26199 * tree-ssa-address.c (struct mem_address): Move to header file.
26200 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
26201 * tree-ssa-address.h (struct mem_address): Move from C file.
26202 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
26204 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26206 * tree-affine.h (aff_combination_type): New interface.
26207 (aff_combination_zero_p): Remove static.
26208 (aff_combination_const_p): New interface.
26209 (aff_combination_singleton_var_p): New interfaces.
26211 2017-05-11 Richard Biener <rguenther@suse.de>
26213 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26214 Skip unreachable blocks and destinations.
26215 (eliminate): Move stmt removal and fixup ...
26216 (fini_eliminate): ... here. Skip inserted exprs.
26217 (pass_pre::execute): Move fini_pre after fini_eliminate.
26218 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
26219 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
26220 PRE to get rid of dead code that has invalid SSA form and
26221 split critical edges again.
26223 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26225 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26227 2017-05-11 Richard Biener <rguenther@suse.de>
26229 * passes.c (execute_function_todo): Verify loops if they are
26230 said to be up-to-date.
26231 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
26232 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26234 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26237 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26238 handle calling assemble_external ourself.
26241 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26242 modes with zero size. Enhance comment.
26244 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26246 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26247 built-ins for vec_xl and vec_xst with short and char pointer
26250 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26252 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26253 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26254 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26255 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26256 (_mm_maskz_min_round_ss): New intrinsics.
26257 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26258 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26259 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26260 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26261 (__builtin_ia32_minss_mask_round): New builtins.
26262 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26263 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26264 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26266 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26267 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26269 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26272 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26274 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26275 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26276 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26277 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26278 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26279 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26280 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26281 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26282 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26283 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26284 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26285 (__builtin_ia32_mulss_mask_round): New builtins.
26286 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26287 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26288 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26290 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26291 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26293 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26296 2017-05-10 Julia Koval <julia.koval@intel.com>
26298 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26299 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26300 (_mm256_setr_m128i): New intrinsics.
26302 2017-05-10 Julia Koval <julia.koval@intel.com>
26304 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26305 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26306 (_mm_maskz_rcp14_ss): New intrinsics.
26307 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26308 (__builtin_ia32_rcp14ss_mask): New builtins.
26309 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26311 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26313 PR tree-optimization/51513
26314 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26315 (assert_unreachable_fallthru_edge_p): Use it.
26316 (group_case_labels_stmt): Likewise.
26317 * tree-cfg.h: Prototype it.
26318 * stmt.c: Include cfghooks.h and tree-cfg.h.
26319 (emit_case_dispatch_table) <gap_label>: New local variable.
26320 Use it to fill dispatch table gaps.
26321 Test for default_label before updating probabilities.
26322 (expand_case) <default_label>: Remove unneeded initialization.
26323 Test for unreachable default case statement and remove its edge.
26324 Set default_label accordingly.
26325 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26327 2017-05-10 Carl Love <cel@us.ibm.com>
26329 * config/rs6000/rs6000-c: Add support for built-in functions
26330 vector signed char vec_neg (vector signed char)
26331 vector signed short int vec_neg (vector short int)
26332 vector signed int vec_neg (vector signed int)
26333 vector signed long long vec_neg (vector signed long long)
26334 vector float vec_neg (vector float)
26335 vector double vec_neg (vector double)
26336 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26338 * config/rs6000/altivec.h: Add define for vec_neg
26339 * doc/extend.texi: Update the built-in documentation for the
26340 new built-in functions.
26342 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26344 PR tree-optimization/77644
26345 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26347 2017-05-10 Nathan Sidwell <nathan@acm.org>
26349 * dumpfile.h (TDI_lang_all): New.
26350 (TDF_KIND): New. Renumber others
26351 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26353 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26355 (get_dump_file_name): Adjust suffix generation.
26356 (dump_enable_all): Use TDF_KIND.
26357 * doc/invoke.texi (-fdump-lang-all): Document.
26359 * dumpfile.h: Tabify.
26361 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26364 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26365 Move member access before delete.
26367 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26369 * tree-inline.c (expand_call_inline): Split block at stmt
26372 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26375 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26376 are now unused after splitting mov{sf,sd}_hardfloat.
26377 (f32_lr2): Likewise.
26378 (f32_lm): Likewise.
26379 (f32_lm2): Likewise.
26380 (f32_li): Likewise.
26381 (f32_li2): Likewise.
26382 (f32_lv): Likewise.
26383 (f32_sr): Likewise.
26384 (f32_sr2): Likewise.
26385 (f32_sm): Likewise.
26386 (f32_sm2): Likewise.
26387 (f32_si): Likewise.
26388 (f32_si2): Likewise.
26389 (f32_sv): Likewise.
26390 (f32_dm): Likewise.
26391 (f32_vsx): Likewise.
26392 (f32_av): Likewise.
26393 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26394 For movsf, order stores so the VSX stores occur before the GPR
26395 store which encourages the register allocator to use a traditional
26396 FPR instead of a GPR. For movsd, order the stores so that the GPR
26397 store comes before the VSX stores to allow the power6 to work.
26398 This is due to the power6 not having a 32-bit integer store
26399 instruction from a FPR.
26400 (movsf_hardfloat): Likewise.
26401 (movsd_hardfloat): Likewise.
26403 2017-05-09 Martin Sebor <msebor@redhat.com>
26405 PR translation/80280
26406 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26409 PR translation/80280
26410 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26411 data member added in r247778.
26412 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26414 2017-05-09 Nathan Sidwell <nathan@acm.org>
26416 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26418 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26421 2017-05-09 Marek Polacek <polacek@redhat.com>
26423 * doc/invoke.texi: Fix typo.
26425 2017-05-09 Richard Biener <rguenther@suse.de>
26427 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26428 (vrp_val_is_min): Likewise.
26429 (set_value_range_to_value): Likewise.
26430 (set_value_range_to_nonnegative): Likewise.
26431 (gimple_assign_nonzero_p): Likewise.
26432 (gimple_stmt_nonzero_p): Likewise.
26433 (vrp_int_const_binop): Likewise. Remove unreachable case.
26434 (adjust_range_with_scev): Adjust comments.
26435 (compare_range_with_value): Likewise.
26436 (extract_range_from_phi_node): Likewise.
26437 (test_for_singularity): Likewise.
26439 2017-05-09 Richard Biener <rguenther@suse.de>
26441 * tree-vrp.c (get_single_symbol): Add assert that we don't
26442 get overflowed constants as invariant part.
26443 (compare_values_warnv): Add comment before the TREE_NO_WARNING
26444 checks. Use wi::cmp instead of recursing for integer constants.
26445 (compare_values): Just ignore whether we assumed undefined
26446 overflow instead of failing the compare.
26447 (extract_range_for_var_from_comparison_expr): Add comment before the
26448 TREE_NO_WARNING sets.
26449 (test_for_singularity): Likewise.
26450 (extract_range_from_comparison): Do not disable optimization
26451 when we assumed undefined overflow.
26452 (extract_range_basic): Remove init of unused var.
26454 2017-05-09 Richard Biener <rguenther@suse.de>
26456 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
26457 (extract_range_from_multiplicative_op_1): Adjust.
26458 (extract_range_from_binary_expr_1): Use int_const_binop.
26460 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
26463 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
26464 rs6000_store_data_bypass_p in seven define_bypass directives and
26465 in several comments.
26466 * config/rs6000/rs6000-protos.h: Add prototype for
26467 rs6000_store_data_bypass_p function.
26468 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
26469 function implements slightly different (rs6000-specific) semantics
26470 than store_data_bypass_p, returning false rather than aborting
26471 with assertion error when arguments do not satisfy the
26472 requirements of store data bypass.
26473 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
26474 rs6000_store_data_bypass_p.
26476 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
26478 * config/xtensa/xtensa-protos.h
26479 (xtensa_initial_elimination_offset): New declaration.
26480 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26481 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
26482 macro definition, add case for FRAME_POINTER_REGNUM when
26483 FRAME_GROWS_DOWNWARD.
26484 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
26485 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
26486 xtensa_initial_elimination_offset.
26488 2017-05-08 Nathan Sidwell <nathan@acm.org>
26490 * doc/invoke.texi: Alphabetize -fdump options.
26492 2017-05-08 Martin Sebor <msebor@redhat.com>
26494 PR translation/80280
26495 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
26497 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
26499 * target.def (compute_frame_layout): New optional target hook.
26500 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
26501 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
26502 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
26504 * reload1.c (verify_initial_elim_offsets): Likewise.
26505 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
26506 (use_simple_return_p): Call arm_compute_frame_layout if needed.
26507 (arm_get_frame_offsets): Split up into this ...
26508 (arm_compute_frame_layout): ... and this function.
26510 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
26512 * config/aarch64/constraints.md (Usa): New constraint.
26513 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
26515 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
26517 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
26518 with_multilib_list after it has been checked.
26520 2017-05-08 Richard Biener <rguenther@suse.de>
26522 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
26523 (bitmap_set_subtract_values): Likewise.
26525 2017-05-08 Richard Biener <rguenther@suse.de>
26527 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
26528 (gimple_assign_nonzero): ... this and remove strict_overflow_p
26530 (gimple_stmt_nonzero_warnv_p): Rename to ...
26531 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
26533 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
26534 (extract_range_basic): Adjust, do not disable propagation on
26535 strict overflow sensitive simplification.
26536 (vrp_visit_cond_stmt): Likewise.
26538 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
26540 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
26541 body size unconditionally.
26543 2017-05-07 Jeff Law <law@redhat.com>
26546 2017-05-06 Jeff Law <law@redhat.com>
26547 PR tree-optimization/78496
26548 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26551 PR tree-optimization/78496
26552 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26553 (simplify_stmt_using_ranges): Call it.
26554 (vrp_dom_walker::before_dom_children): Extract equivalences
26555 from an ASSERT_EXPR with an equality comparison against a
26558 2017-05-06 Jeff Law <law@redhat.com>
26560 PR tree-optimization/78496
26561 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26564 PR tree-optimization/78496
26565 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26566 (simplify_stmt_using_ranges): Call it.
26567 (vrp_dom_walker::before_dom_children): Extract equivalences
26568 from an ASSERT_EXPR with an equality comparison against a
26571 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26573 * lra-constraints.c (lra_copy_reg_equiv): New function.
26574 (split_reg): Use it to copy equivalence information from the
26575 original register to the spill register.
26577 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26579 PR rtl-optimization/75964
26580 * simplify-rtx.c (simplify_const_relational_operation): Remove
26581 invalid handling of comparisons of integer ABS.
26583 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26585 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26586 initialize to zero.
26587 (init_regs): Remove declaration.
26588 (function_arg_advance_32): Initialize error_p as boolean variable.
26590 2017-05-05 Nathan Sidwell <nathan@acm.org>
26592 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26593 lines. Use for (;;).
26595 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26597 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26598 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26599 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26600 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26601 VF=2 that require versioning.
26603 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26605 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26608 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26610 * diagnostic.h (diagnostic_override_option_index): Convert from
26611 macro to inline function.
26613 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26615 * diagnostic.c (last_module_changed_p): New function.
26616 (set_last_module): New function.
26617 (diagnostic_report_current_module): Convert macro usage to
26618 the above functions.
26619 * diagnostic.h (diagnostic_context::last_module): Strengthen
26620 from const line_map * to const line_map_ordinary *.
26621 (diagnostic_last_module_changed): Delete macro.
26622 (diagnostic_set_last_module): Delete macro.
26624 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26626 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26627 with diagnostic_report_diagnostic.
26628 (diagnostic_n_impl_richloc): Likewise.
26629 * diagnostic.h (report_diagnostic): Delete macro.
26630 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26631 with diagnostic_report_diagnostic.
26632 * substring-locations.c (format_warning_va): Likewise.
26634 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26636 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26637 save/restor of format_spec. Move option-printing code to...
26638 (print_option_information): ...this new function, and
26639 reimplement by simply printing to the pretty_printer,
26640 rather than appending to the format string.
26642 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26644 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26645 handling logic into...
26646 (update_effective_level_from_pragmas): ...this new function.
26648 2017-05-04 Andrew Waterman <andrew@sifive.com>
26650 * config/riscv/riscv.opt (mstrict-align): New option.
26651 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26652 (SLOW_UNALIGNED_ACCESS): Define.
26653 (riscv_slow_unaligned_access): Declare.
26654 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26656 (riscv_slow_unaligned_access): New variable.
26657 (rocket_tune_info): Set slow_unaligned_access to true.
26658 (optimize_size_tune_info): Set slow_unaligned_access to false.
26659 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26660 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26661 (riscv_option_override): Set riscv_slow_unaligned_access.
26662 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26664 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26666 * config/riscv/riscv.md: Unify indentation.
26668 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26673 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26675 (extendsi<mode>2): Add support for doing sign extension via
26676 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26677 don't have ISA 3.0 instructions.
26678 (extendsi<mode>2 splitter): Likewise.
26679 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26680 generate the normal insns since SImode can now go in vector
26681 registers. Disallow the special UNSPECs needed for previous
26682 machines to hide SImode being used. Add new insns
26683 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26684 (fix_trunc<mode>si2_stfiwx): Likewise.
26685 (fix_trunc<mode>si2_internal): Likewise.
26686 (fixuns_trunc<mode>si2): Likewise.
26687 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26688 (fctiw<u>z_<mode>_smallint): Likewise.
26689 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26690 of floating point to 32-bit integer from doing a direct move to
26691 the GPR registers to do a store.
26692 (fctiwz_<mode>): Break long line.
26694 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26696 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26697 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26698 (addr_list, addr_offset_valid_p): New.
26699 (split_address_groups): Check offset validity with above function.
26700 (gt-tree-ssa-loop-ivopts.h): Include header file.
26702 2017-05-05 Nathan Sidwell <nathan@acm.org>
26704 * config.gcc (arm*-*-*): Add missing 'fi'.
26706 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26708 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26709 included in -fopt-info does not matter.
26710 * doc/optinfo.texi (-fopt-info): Fix description of default
26711 behavour. Explicitly say order of options included in -fopt-info
26714 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26716 * config.gcc: Allow combinations of aprofile and rmprofile values for
26717 --with-multilib-list.
26718 * config/arm/t-multilib: New file.
26719 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26720 variables. Remove setting of ISA and floating-point ABI in
26721 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26722 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26723 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26724 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26726 * config/arm/t-rmprofile: Likewise except for the matches changes.
26727 * doc/install.texi (--with-multilib-list): Document the combination of
26728 aprofile and rmprofile values and warn about pitfalls in doing that.
26730 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26732 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26733 (movdi_aarch64): Likewise.
26735 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26737 PR tree-optimization/80632
26738 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26740 (build_arrays): Initialize it for virtual phis.
26741 (fix_phi_nodes): Use it for virtual phis.
26743 PR tree-optimization/80558
26744 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26745 [x, y] op z into [x op, y op z] for op & or | if conditions
26748 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26749 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26752 * config/arm/arm.md (use_literal_pool): Remove.
26753 (64-bit immediate split): No longer takes cost into consideration
26754 if arm_disable_literal_pool is enabled.
26755 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26756 used when arm_disable_literal_pool is enabled.
26757 (arm_max_const_double_inline_cost): Remove use of
26758 arm_disable_literal_pool.
26759 (push_minipool_fix): Add assert.
26760 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26761 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26762 (no_literal_pool_sf_immediate): New.
26764 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26766 PR tree-optimization/80613
26767 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26768 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26770 2017-05-05 Richard Biener <rguenther@suse.de>
26772 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26774 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26776 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26777 of this flag from insn conditions due to removal from r247495.
26779 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26781 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26783 (arm_early_store_addr_dep_ptr): Likewise.
26784 * config/arm/aarch-common-protos.h
26785 (arm_early_load_addr_dep_ptr): Add prototype.
26786 (arm_early_store_addr_dep_ptr): Likewise.
26787 * config/arm/cortex-a53.md: Add new bypasses.
26789 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26791 * tree.c (next_type_uid): Change type to unsigned.
26792 (type_hash_canon): Decrement back next_type_uid if
26793 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26794 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26797 2017-05-04 Martin Sebor <msebor@redhat.com>
26799 * builtins.c: Fix a trivial typo in a comment.
26801 PR middle-end/79234
26802 * builtins.c (check_sizes): Adjust to handle reading past the end.
26803 Avoid printing excessive upper bound of ranges. Use %E to print
26804 tree nodes instead of converting them to %wu.
26805 (expand_builtin_memchr): New function.
26806 (compute_dest_size): Rename...
26807 (compute_objsize): ...to this.
26808 (expand_builtin_memcpy): Adjust.
26809 (expand_builtin_mempcpy): Adjust.
26810 (expand_builtin_strcat): Adjust.
26811 (expand_builtin_strcpy): Adjust.
26812 (check_strncat_sizes): Adjust.
26813 (expand_builtin_strncat): Adjust.
26814 (expand_builtin_strncpy): Adjust and simplify.
26815 (expand_builtin_memset): Adjust.
26816 (expand_builtin_bzero): Adjust.
26817 (expand_builtin_memcmp): Adjust.
26818 (expand_builtin): Handle memcmp.
26819 (maybe_emit_chk_warning): Check strncat just once.
26821 2017-05-04 Martin Sebor <msebor@redhat.com>
26823 PR preprocessor/79214
26824 PR middle-end/79222
26825 PR middle-end/79223
26826 * builtins.c (check_sizes): Add inlining context and issue
26827 warnings even when -Wno-system-headers is set.
26828 (check_strncat_sizes): Same.
26829 (expand_builtin_strncat): Same.
26830 (expand_builtin_memmove): New function.
26831 (expand_builtin_stpncpy): Same.
26832 (expand_builtin): Handle memmove and stpncpy.
26834 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26836 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26837 which is not used any more.
26839 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26841 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26843 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26845 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26846 (cortexa53_tunings): Likewise.
26847 (cortexa57_tunings): Likewise.
26848 (cortexa72_tunings): Likewise.
26849 (cortexa73_tunings): Likewise.
26851 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26853 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26854 Set loop alignment to 8.
26856 2017-05-04 Martin Sebor <msebor@redhat.com>
26858 PR translation/80280
26859 * builtins.c (expand_builtin_object_size): Add missing quoting to
26860 %D and like directives.
26861 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26862 (hsa_type_for_tree_type): Same.
26863 (verify_function_arguments): Same.
26864 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26865 * varasm.c (get_section): Same.
26868 2017-05-04 Martin Sebor <msebor@redhat.com>
26870 PR translation/80280
26871 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26873 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26875 * config/aarch64/aarch64.c (generic_addrcost_table):
26876 Change HI/TI mode setting.
26878 2017-05-04 Martin Jambor <mjambor@suse.cz>
26880 PR tree-optimization/80622
26881 * tree-sra.c (comes_initialized_p): New function.
26882 (build_accesses_from_assign): Only set write lazily when
26883 comes_initialized_p is false.
26884 (analyze_access_subtree): Use comes_initialized_p.
26885 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26886 instead of testing for PARM_DECL.
26888 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26890 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26891 constraint on operand 0 to allow more general addressing modes.
26892 Adjust output template.
26893 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26895 * config/aarch64/aarch64-protos.h
26896 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26897 * config/aarch64/constraints.md (Dp): New address constraint.
26898 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26901 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26903 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26904 update use of estimate_ipcp_clone_size_and_time.
26905 (estimate_local_effects): Update use of
26906 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26907 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26908 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26909 Return nonspecialized time.
26911 2017-05-04 Richard Biener <rguenther@suse.de>
26913 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26914 for the last VUSE which def dominates the PHI. Directly call
26916 (get_continuation_for_phi_1): Remove.
26918 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26920 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26921 to explain the use of truncating division. Cap the number of
26922 iterations to the maximum given by nb_iterations_upper_bound,
26925 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26927 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26928 * configure: Regenerate.
26929 * config.in: Regenerate.
26930 * config/i386/driver-mingw32.c: new file.
26931 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26932 * config.host: Link driver-mingw32.o on MinGW host.
26933 * doc/install.texi: Document new --enable-mingw-wildcard configure
26936 2017-05-04 Marek Polacek <polacek@redhat.com>
26938 PR tree-optimization/80612
26939 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26941 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26942 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26944 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26945 (movt splitter): Likewise.
26946 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26947 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26948 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26949 block for Thumb-1 with MOVT.
26950 (thumb2_legitimate_address_p): Move code block ...
26951 (can_avoid_literal_pool_for_label_p): ... into this new function.
26952 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26954 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26955 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26956 "M-profile targets with the MOVT instruction".
26958 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26960 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26961 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26962 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26964 2017-05-04 Martin Liska <mliska@suse.cz>
26966 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26967 variable cond_code.
26969 2017-05-04 Richard Biener <rguenther@suse.de>
26971 * tree.c (array_at_struct_end_p): Handle arrays at struct
26972 end with flexarrays more conservatively. Refactor and treat
26973 arrays of arrays or aggregates more strict. Fix
26974 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26975 * tree.c (array_at_struct_end_p): Adjust prototype.
26976 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26977 * gimple-fold.c (get_range_strlen): Likewise.
26978 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26980 2017-05-04 Richard Biener <rguenther@suse.de>
26982 PR tree-optimization/31130
26983 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26985 (supports_overflow_infinity): Likewise.
26986 (is_negative_overflow_infinity): Likewise.
26987 (is_positive_overflow_infinity): Likewise.
26988 (is_overflow_infinity): Likewise.
26989 (stmt_overflow_infinity): Likewise.
26990 (overflow_infinity_range_p): Likewise.
26991 (usable_range_p): Remove as always returning true.
26992 (make_overflow_infinity): Remove.
26993 (negative_overflow_infinity): Likewise.
26994 (positive_overflow_infinity): Likewise.
26995 (avoid_overflow_infinity): Likewise.
26996 (set_value_range): Adjust accordingly.
26997 (set_value_range_to_nonnegative): Likewise, remove now unused
26998 overflow_infinity arg.
26999 (vrp_operand_equal_p): Adjust.
27000 (update_value_range): Likewise.
27001 (range_int_cst_singleton_p): Likewise.
27002 (operand_less_p): Likewise.
27003 (compare_values_warnv): Likewise.
27004 (extract_range_for_var_from_comparison_expr): Likewise.
27005 (vrp_int_const_binop): Likewise.
27006 (zero_nonzero_bits_from_vr): Likewise.
27007 (extract_range_from_multiplicative_op_1): Likewise.
27008 (extract_range_from_binary_expr_1): Likewise.
27009 (extract_range_from_unary_expr): Likewise.
27010 (extract_range_from_comparison): Likewise.
27011 (extract_range_basic): Likewise.
27012 (adjust_range_with_scev): Likewise.
27013 (compare_ranges): Likewise.
27014 (compare_range_with_value): Likewise.
27015 (dump_value_range): Likewise.
27016 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27018 (simplify_cond_using_ranges): Adjust.
27020 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27022 * brig-builtins.def: Added a builtin for class_f64.
27023 * builtin-types.def: Added a builtin type needed by class_f64.
27025 2017-05-03 Jason Merrill <jason@redhat.com>
27027 * timevar.def: Add TV_CONSTEXPR.
27029 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27031 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27033 2017-05-03 Martin Jambor <mjambor@suse.cz>
27035 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27036 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27037 * ipa-cp.c (ipcp_propagate_stage): Do not call
27038 ipa_update_after_lto_read.
27039 * ipa-inline.c (ipa_inline): Likewise.
27041 2017-05-03 Martin Jambor <mjambor@suse.cz>
27043 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27044 tag. Added a default constructor and a destructor.
27045 (ipa_edge_args_sum_t): New class;
27046 (ipa_edge_args_sum): Declare.
27047 (ipa_edge_args_vector): Remove declaration.
27048 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27049 (ipa_free_edge_args_substructures): Remove declaration.
27050 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27051 (ipa_edge_args_info_available_for_edge_p): Likewise.
27052 * ipa-prop.c (ipa_edge_args_vector): Removed.
27053 (edge_removal_hook_holder): Likewise.
27054 (edge_duplication_hook_holder): Likewise.
27055 (ipa_edge_args_sum): New variable.
27056 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27057 ipa_edge_args_vector.
27058 (ipa_free_edge_args_substructures): Likewise.
27059 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27060 ipa_edge_args_vector.
27061 (ipa_edge_removal_hook): Turned into method
27062 ipa_edge_args_sum_t::remove.
27063 (ipa_edge_duplication_hook): Turned into method
27064 ipa_edge_args_sum_t::duplicate.
27065 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27066 registering edge hooks.
27067 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27068 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27069 ipa_edge_args_sum instead of ipa_edge_args_vector.
27070 * ipa-profile.c (ipa_profile): Likewise.
27072 2017-05-03 Martin Jambor <mjambor@suse.cz>
27074 * symbol-summary.h (function_summary): New method exists.
27075 (function_summary::symtab_removal): Deallocate through release.
27076 (call_summary): New class.
27077 (gt_ggc_mx): New overload.
27078 (gt_pch_nx): Likewise.
27079 (gt_pch_nx): Likewise.
27081 2017-05-03 Jeff Law <law@redhat.com>
27083 PR tree-optimization/78496
27084 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27085 from simplify_cond_using_ranges. Split off code to walk
27086 backwards through casts into ...
27087 (simplify_cond_using_ranges_2): New function.
27088 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27089 (execute_vrp): After identifying jump threads, call
27090 simplify_cond_using_ranges_2.
27092 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27095 * ipa-inline.h (inline_summary): Add ctor.
27096 (create_ggc): Do not use ggc_cleared_alloc.
27098 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27099 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27101 * gcc.c (handle_braces): Support escaping in switch matching
27103 * doc/invoke.texi (Spec Files): Document it.
27104 Remove superfluous @code markup in items.
27106 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27108 * diagnostic-show-locus.c (struct column_range): New struct.
27109 (get_affected_columns): New function.
27110 (get_printed_columns): New function.
27111 (struct correction): New struct.
27112 (correction::ensure_capacity): New function.
27113 (correction::ensure_terminated): New function.
27114 (struct line_corrections): New struct.
27115 (line_corrections::~line_corrections): New dtor.
27116 (line_corrections::add_hint): New function.
27117 (layout::print_trailing_fixits): Reimplement in terms of the new
27119 (selftest::test_overlapped_fixit_printing): New function.
27120 (selftest::diagnostic_show_locus_c_tests): Call it.
27122 2017-05-03 Nathan Sidwell <nathan@acm.org>
27124 Canonicalize canonical type hashing
27125 * tree.h (type_hash_canon_hash): Declare.
27126 * tree.c (type_hash_list, attribute_hash_list): Move into
27127 type_hash_canon_hash.
27128 (build_type_attribute_qual_variant): Break out hash code calc into
27129 type_hash_canon_hash.
27130 (type_hash_canon_hash): New. Generic type hash computation.
27131 (build_range_type_1, build_array_type_1, build_function_type,
27132 build_method_type_directly, build_offset_type, build_complex_type,
27133 make_vector_type): Call it.
27135 2017-05-03 Richard Biener <rguenther@suse.de>
27137 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
27138 When all DRs have unknown misaligned do not always peel
27139 when there is a store but apply the same costing model as if
27140 there were only loads.
27142 2017-05-03 Richard Biener <rguenther@suse.de>
27145 PR tree-optimization/80492
27146 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27147 compare_base_decls returning dont-know properly.
27149 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
27151 * config/arm/iterators.md (CCSI): New mode iterator.
27152 (arch): New mode attribute.
27153 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
27154 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
27155 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
27156 code iterator for success result mode.
27157 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
27158 the corresponding new insn generators.
27160 2017-05-03 Bin Cheng <bin.cheng@arm.com>
27163 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27164 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27166 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
27168 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
27169 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
27170 (DDR_A): Wrap DDR argument in brackets.
27171 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
27172 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
27173 (DDR_REVERSED_P): Likewise.
27175 2017-05-03 Jakub Jelinek <jakub@redhat.com>
27177 PR tree-optimization/79472
27178 * tree-switch-conversion.c (struct switch_conv_info): Add
27179 contiguous_range and default_case_nonstandard fields.
27180 (collect_switch_conv_info): Compute contiguous_range and
27181 default_case_nonstandard fields, don't clear final_bb if
27182 contiguous_range and only the default case doesn't have the required
27184 (check_all_empty_except_final): Set default_case_nonstandard instead
27185 of failing if contiguous_range and the default case doesn't have empty
27187 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
27188 and only the default case doesn't have the required constants. Skip
27190 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
27191 if default_case_nonstandard.
27192 (build_constructors): Build constant 1 just once. Assert that default
27193 values aren't inserted in between cases if contiguous_range. Skip
27195 (build_arrays): Skip virtual phis.
27196 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
27197 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
27198 Handle virtual phis.
27199 (gen_inbound_check): Handle default_case_nonstandard case.
27200 (process_switch): Adjust check_final_bb caller. Call
27201 gather_default_values with the first non-default case instead of
27202 default case if default_case_nonstandard.
27204 2017-05-02 Nathan Sidwell <nathan@acm.org>
27206 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
27207 check. Fix formatting.
27209 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
27211 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
27212 errors when comparing specialized and unspecialized times.
27214 2017-05-02 David Malcolm <dmalcolm@redhat.com>
27216 * diagnostic-show-locus.c
27217 (layout::should_print_annotation_line_p): Make private.
27218 (layout::print_annotation_line): Make private.
27219 (layout::annotation_line_showed_range_p): Make private.
27220 (layout::show_ruler): Make private.
27221 (layout::print_source_line): Make private. Pass in line and
27222 line_width, rather than calling location_get_source_line. Drop
27224 (layout::print_leading_fixits): New method.
27225 (layout::print_any_fixits): Rename to...
27226 (layout::print_trailing_fixits): ...this, and make private.
27227 Don't print newline fixits.
27228 (diagnostic_show_locus): Move logic for printing one row into...
27229 (layout::print_line): ...this new function. Move the
27230 location_get_source_line call and error-handling from
27231 print_source_line to here. Call print_leading_fixits, and rename
27232 print_any_fixits to print_trailing_fixits.
27233 (selftest::test_fixit_insert_containing_newline): Update now that
27234 newlines are partially supported.
27235 (selftest::test_fixit_insert_containing_newline_2): New test.
27236 (selftest::test_fixit_replace_containing_newline): Update comments.
27237 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27238 * edit-context.c (class added_line): New class.
27239 (class edited_line): Describe newline handling in comment.
27240 (edited_line::actually_edited_p): New method.
27241 (edited_line::print_content): Delete redundant decl.
27242 (edited_line::m_predecessors): New field.
27243 (edited_file::print_content): Call edited_line::print_content.
27244 (edited_file::print_diff): Update to support newlines.
27245 (edited_file::print_diff_hunk): Likewise.
27246 (edited_file::print_run_of_changed_lines): New function.
27247 (edited_file::print_diff_line): Convert to...
27248 (print_diff_line): ...this.
27249 (edited_file::get_effective_line_count): New function.
27250 (edited_line::edited_line): Initialize new field m_predecessors.
27251 (edited_line::~edited_line): Clean up m_predecessors.
27252 (edited_line::apply_fixit): Handle newlines.
27253 (edited_line::get_effective_line_count): New function.
27254 (edited_line::print_content): New function.
27255 (edited_line::print_diff_lines): New function.
27256 (selftest::test_applying_fixits_insert_containing_newline): New
27258 (selftest::test_applying_fixits_replace_containing_newline): New
27260 (selftest::insert_line): New function.
27261 (selftest::test_applying_fixits_multiple_lines): Add example of
27263 (selftest::edit_context_c_tests): Call the new tests.
27265 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27267 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27268 parameter cand. Update dump information.
27269 (get_computation_cost): Update uses.
27271 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27273 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27274 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27275 (get_computation_at, rewrite_use_address): Update use of
27276 get_computation_aff.
27278 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27280 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27281 (get_computation): Delete.
27282 (get_computation_cost): Implement like get_computation_cost_at.
27283 Use get_computation_at.
27284 (get_computation_cost_at): Delete.
27285 (rewrite_use_nonlinear_expr): Use get_computation_at.
27286 (rewrite_use_compare, remove_unused_ivs): Ditto.
27288 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27290 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27292 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27294 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27295 (ivopts_global_cost_for_size): Rename parameter and update uses.
27296 (iv_ca_recount_cost): Update uses.
27297 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27298 candidates seperately in n_invs and n_cands.
27299 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27301 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27303 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27304 (find_inv_vars_cb): New.
27305 (find_depends): Renamed to ...
27306 (find_inv_vars): ... this.
27307 (add_candidate_1, force_var_cost): Call find_inv_vars.
27308 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27310 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27312 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27313 inv_vars. Add inv_exprs.
27314 (struct iv_cand): Rename depends_on to inv_vars.
27315 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27316 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27317 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27318 (dump_cand): Dump inv_vars.
27319 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27320 (record_invariant, find_depends, add_candidate_1): Ditto.
27321 (set_group_iv_cost, force_var_cost): Ditto.
27322 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27323 (get_computation_cost_at, get_computation_cost): Ditto.
27324 (determine_group_iv_cost_generic): Ditto.
27325 (determine_group_iv_cost_address): Ditto.
27326 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27327 (determine_group_iv_costs): Ditto.
27328 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27329 (iv_ca_set_remove_invariants): Renamed to ...
27330 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27331 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27332 (iv_ca_set_add_invariants): Renamed to ...
27333 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27334 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27335 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27336 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27337 (create_new_ivs): Remove useless dump.
27339 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27341 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27343 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27344 (iv_ca_set_no_cp, create_new_iv): Ditto.
27346 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27348 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27350 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27352 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27353 function tree_check2.
27355 2017-05-02 Martin Liska <mliska@suse.cz>
27357 * doc/gcov.texi: Add missing preposition.
27358 * gcov.c (function_info::function_info): Properly fill up
27359 all member variables.
27361 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27363 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27365 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27367 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27369 2017-05-02 Martin Liska <mliska@suse.cz>
27372 * lto-streamer-in.c (lto_read_tree_1): Remove
27373 LTO_STREAMER_DEBUG.
27374 * lto-streamer.c (struct tree_hash_entry): Likewise.
27375 (struct tree_entry_hasher): Likewise.
27376 (tree_entry_hasher::hash): Likewise.
27377 (tree_entry_hasher::equal): Likewise.
27378 (lto_streamer_init): Likewise.
27379 (lto_orig_address_map): Likewise.
27380 (lto_orig_address_get): Likewise.
27381 (lto_orig_address_remove): Likewise.
27382 * lto-streamer.h: Likewise.
27383 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27384 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27386 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27388 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27389 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27390 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27391 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27392 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27393 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27394 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27395 (mm_maskz_sub_ss): New intrinsics.
27396 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27397 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27398 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27399 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27400 (__builtin_ia32_subss_mask_round): New builtins.
27401 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27402 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27403 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27405 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27406 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27408 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27411 2017-05-02 Martin Jambor <mjambor@suse.cz>
27413 PR tree-optimization/78687
27414 * tree-sra.c (access): New field parent.
27415 (process_subtree_disqualification): New function.
27416 (disqualify_candidate): Call it.
27417 (build_accesses_from_assign): Reset write flag if creating an
27419 (build_access_subtree): Fill in parent field and also prpagate
27420 down grp_write flag.
27421 (create_artificial_child_access): New parameter set_grp_write, set
27422 grp_write to its value.
27423 (propagate_subaccesses_across_link): Also propagate grp_write flag
27425 (propagate_all_subaccesses): Push the closest parent back to work
27426 queue if add_access_to_work_queue returned true.
27428 2017-05-02 Richard Biener <rguenther@suse.de>
27430 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27431 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27432 -fstrict-overflow documentation.
27433 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27434 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27435 flag_strict_overflow.
27436 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27437 * lto-opts.c (lto_write_options): Do not stream it.
27438 * lto-wrapper.c (merge_and_complain): Do not handle it.
27439 * opts.c (default_options_table): Do not set -fstrict-overflow.
27440 (finish_options): Likewise do not clear it when sanitizing.
27441 * simplify-rtx.c (simplify_const_relational_operation): Do not
27442 test flag_strict_overflow.
27444 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27446 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
27447 using enabled attribute.
27448 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
27449 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
27450 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
27451 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
27452 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
27453 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
27454 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
27455 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
27456 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
27457 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
27459 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27461 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
27463 2017-05-02 Richard Biener <rguenther@suse.de>
27465 PR tree-optimization/80591
27467 2017-04-10 Richard Biener <rguenther@suse.de>
27469 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27472 2017-05-02 Richard Biener <rguenther@suse.de>
27474 PR tree-optimization/80549
27475 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
27476 (cleanup_tree_cfg_noloop): Create forwarders to known loop
27477 headers if they do not have a preheader.
27479 2017-05-02 Martin Liska <mliska@suse.cz>
27482 * common.opt: Fix typo.
27483 * doc/invoke.texi: Likewise.
27485 2017-05-01 Jan Beulich <jbeulich@suse.com>
27487 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
27488 swapping, add (x,x,m,x,n) alternative.
27490 2017-05-01 Nathan Sidwell <nathan@acm.org>
27492 * calls.c (combine_pending_stack_adjustment_and_call): Remove
27493 unnecessary unadjusted_alignment check.
27495 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27498 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
27500 * gimplify.c (gimplify_cilk_detach): New function.
27501 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
27502 * tree-core.h: Document EXPR_CILK_SPAWN.
27503 * tree.h (EXPR_CILK_SPAWN): Define.
27505 2017-05-01 David Malcolm <dmalcolm@redhat.com>
27507 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
27508 to use new fixit_hint representation, using the "replace" logic.
27509 (get_line_span_for_fixit_hint): Likewise.
27510 (layout::print_any_fixits): Likewise.
27511 (selftest::test_one_liner_many_fixits): Rename to...
27512 (selftest::test_one_liner_many_fixits_1): ...this, and update
27513 comment and expected output to reflect that the multiple fix-it
27514 hints are now consolidated into one insertion.
27515 (selftest::test_one_liner_many_fixits_2): New test.
27516 (selftest::test_diagnostic_show_locus_one_liner): Update for
27518 (selftest::test_fixit_consolidation): Update for fix-it API
27520 * diagnostic.c (print_parseable_fixits): Likewise.
27521 * edit-context.c (edited_line::m_line_events): Convert from
27522 auto_vec <line_event *> to auto_vec <line_event>.
27523 (class line_event): Convert from abstract base class to a concrete
27524 class, taking over the role of replace_event.
27525 (class insert_event): Delete.
27526 (class replace_event): Rename to class line_event. Convert to
27528 (edit_context::add_fixits): Reimplement.
27529 (edit_context::apply_insert): Delete.
27530 (edit_context::apply_replace): Rename to...
27531 (edit_context::apply_fixit): ...this. Convert to half-open range.
27532 (edited_file::apply_insert): Delete.
27533 (edited_file::apply_replace): Rename to...
27534 (edited_file::apply_fixit): ...this.
27535 (edited_line::~edited_line): Drop deletion of events.
27536 (edited_line::apply_insert): Delete.
27537 (edited_line::apply_replace): Rename to...
27538 (edited_line::apply_fixit): ...this. Convert to half-open range.
27539 Update for change to type of m_line_events.
27540 * edit-context.h (edit_context::apply_insert): Delete.
27541 (edit_context::apply_replace): Rename to...
27542 (edit_context::apply_fixit): ...this.
27544 2017-05-01 Martin Sebor <msebor@redhat.com>
27546 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
27549 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
27552 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
27553 __get_cpuid_max returns 0.
27554 (__get_cpuid_count): Ditto.
27556 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
27558 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
27559 replacement expression is another instance of one of its arguments.
27561 2017-05-01 Jakub Jelinek <jakub@redhat.com>
27564 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
27565 check for stack push/pop autoinc.
27566 * config/i386/i386.c (ix86_agi_dependent): Return false
27567 if the only reason why modified_in_p returned true is that
27568 addr is SP based and set_insn is a push or pop.
27570 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27572 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27575 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27578 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27579 (account_size_time): Use two predicates - exec_pred and
27581 (evaluate_conditions_for_known_args): Compute both clause and
27583 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27584 (inline_summary_t::duplicate): Update.
27585 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27587 (compute_inline_parameters): Likewise.
27588 (estimate_edge_size_and_time): Update caluclation of time.
27589 (estimate_node_size_and_time): Compute both time and nonspecialized
27591 (estimate_ipcp_clone_size_and_time): Update.
27592 (inline_merge_summary): Update.
27593 (do_estimate_edge_time): Update.
27594 (do_estimate_edge_size): Update.
27595 (do_estimate_edge_hints): Update.
27596 (inline_read_section, inline_write_summary): Stream both new predicates.
27597 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27599 (compute_inlined_call_time): Cleanup.
27600 (big_speedup_p): Update.
27601 (edge_badness): Update.
27602 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27603 (size_time_entry): Replace predicate by exec_predicate and
27604 nonconst_predicate.
27605 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27606 (estimate_edge_time): Return also nonspec_time.
27607 (reset_edge_growth_cache): Update.
27609 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27611 PR rtl-optimization/80491
27612 * ifcvt.c (noce_process_if_block): When looking for x setter
27613 with missing else_bb, don't check only the insn right before
27614 cond_earliest, but look for the last insn that x is modified in
27615 within the same bb.
27617 PR rtl-optimization/80491
27618 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27620 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27622 PR tree-optimization/80487
27623 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27625 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27627 PR tree-optimization/79697
27628 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27629 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27630 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27632 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27633 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27635 2017-04-28 Martin Sebor <msebor@redhat.com>
27637 PR tree-optimization/80523
27638 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27639 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27641 (maybe_warn, format_directive, parse_directive): Use new functions.
27642 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27644 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27646 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27648 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27650 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27651 target_header_dir): Set correctly.
27652 * configure: Regenerated.
27653 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27654 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27655 instead of SYSTEM_HEADER_DIR.
27657 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27659 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27660 (estimate_local_effects): Likewise.
27661 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27662 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27663 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27664 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27665 do_estimate_edge_time, estimate_edge_time): Likewise.
27666 * ipa-inline-analysis.c (estimate_node_size_and_time,
27667 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27668 (estimate_time_after_inlining): Remove.
27670 2017-04-28 Martin Liska <mliska@suse.cz>
27672 * doc/gcov.texi: Enhance documentation of gcov.
27674 2017-04-28 Martin Liska <mliska@suse.cz>
27676 * doc/gcov.texi: Sort options in alphabetic order.
27677 * doc/gcov-dump.texi: Likewise.
27678 * doc/gcov-tool.texi: Likewise.
27679 * gcov.c (print_usage): Likewise.
27680 * gcov-dump.c (print_usage): Likewise.
27681 * gcov-tool.c (print_merge_usage_message): Likewise.
27682 (print_rewrite_usage_message): Likewise.
27683 (print_overlap_usage_message): Likewise.
27685 2017-04-28 Martin Liska <mliska@suse.cz>
27687 PR gcov-profile/53915
27688 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27690 2017-04-28 Martin Liska <mliska@suse.cz>
27692 PR gcov-profile/79891
27693 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27694 is marked by compiler as living on a line.
27695 (get_cycles_count): Remove usage of the union.
27696 (output_intermediate_file): Likewise.
27697 (find_source): Fix GNU coding style.
27698 (accumulate_line_counts): Remove old non-all block mode.
27699 (output_lines): Remove usage of the union.
27700 * profile.c (output_location): Include all BBs, even if
27701 belonging to a same line (and file) as a previous BB.
27703 2017-04-28 Martin Liska <mliska@suse.cz>
27705 * gcov.c (process_args): Handle new argument 'w'.
27706 (read_graph_file): Assign ID to BBs.
27707 (output_branch_count): Display BB # if verbose flag is set.
27708 (output_lines): Likewise for arcs.
27709 (print_usage): Add '--verbose' option help.
27710 * doc/gcov.texi: Document --verbose (-w) option.
27712 2017-04-28 Martin Liska <mliska@suse.cz>
27714 * gcov.c (struct block_location_info): New struct.
27715 (process_file): Fill up the new structure.
27716 (read_graph_file): Replace usage of encoding by the newly added
27718 (add_line_counts): Likewise.
27719 (accumulate_line_counts): Remove usage of the union.
27720 (function_info::function_info): New function.
27721 (function_info::~function_info): Likewise.
27722 (process_file): Call delete instead of release_function.
27723 (release_function): Release the function.
27724 (release_structures): Call delete instead of release_function.
27725 (solve_flow_graph): Replace usage of num_blocks.
27726 (find_exception_blocks): Likewise.
27727 (output_lines): Fix GNU coding style.
27729 2017-04-28 Martin Liska <mliska@suse.cz>
27732 * coverage.c (coverage_remove_note_file): New function.
27733 * coverage.h: Declare the function.
27734 * toplev.c (finalize): Clean if an error has been seen.
27736 2017-04-28 Martin Liska <mliska@suse.cz>
27738 PR gcov-profile/80031
27739 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27740 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27741 * gcov.c (read_graph_file): Read just number of blocks.
27742 * profile.c (branch_prob): Do not stream 0 flags per a basic
27745 2017-04-28 Martin Liska <mliska@suse.cz>
27747 * gcov-dump.c (tag_*): Add new argument to declarations.
27748 (dump_gcov_file): Likewise.
27749 (tag_blocks): Add and use new argument depth.
27750 (tag_arcs): Likewise.
27751 (tag_lines): Likewise.
27752 (tag_counters): Likewise.
27753 (tag_summary): Likewise.
27754 (dump_working_sets): Use depth to do a proper indentation.
27756 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27759 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27760 * symtab.c (symtab_node::debug_symtab): Move definition here.
27762 2017-04-28 Richard Biener <rguenther@suse.de>
27764 * lto-streamer.h (LTO_major_version): Bump to 7.
27766 2017-04-28 Richard Biener <rguenther@suse.de>
27768 * tree-vrp.c (assert_info): New struct.
27769 (add_assert_info): New helper.
27770 (register_edge_assert_for_2): Refactor to add asserts to a vector
27772 (register_edge_assert_for_1): Likewise.
27773 (register_edge_assert_for): Likewise.
27774 (finish_register_edge_assert_for): New helper actually registering
27775 asserts where live on edge.
27776 (find_conditional_asserts): Adjust.
27777 (find_switch_asserts): Likewise.
27778 (evrp_dom_walker::try_find_new_range): Generalize.
27779 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27781 2017-04-27 Marek Polacek <polacek@redhat.com>
27784 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27785 arg10 and arg11 to itype.
27787 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27789 * doc/extend.texi (Object Size Checking): Improve grammar.
27791 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27794 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27795 that the logic for permitting reciprocal estimates matches that
27798 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27801 * tree.c (type_cache_hasher::equal): Only compare
27802 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27803 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27804 non-aggregate element types.
27805 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27806 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27808 2017-04-27 Richard Biener <rguenther@suse.de>
27810 PR middle-end/80533
27811 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27812 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27813 keeping a reference to a trailing array.
27815 2017-04-27 Richard Biener <rguenther@suse.de>
27817 PR middle-end/80539
27818 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27819 being in loop-closed SSA form conservatively.
27820 (chrec_fold_multiply_poly_poly): Likewise.
27822 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27824 PR middle-end/79665
27825 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27826 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27828 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27831 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27832 (aarch64_function_arg_alignment): Return unsigned int again, but still
27833 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27834 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27835 Don't emit -Wpsabi note.
27836 (aarch64_function_arg_boundary): Likewise.
27837 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27840 2017-04-26 Nathan Sidwell <nathan@acm.org>
27842 * tree.h (crc32_unsigned_n): Declare.
27843 (crc32_unsigned, crc32_unsigned): Make inline.
27844 * tree.c (crc32_unsigned_bits): Replace with ...
27845 (crc32_unsigned_n): ... this.
27846 (crc32_unsigned, crc32_byte): Remove.
27847 (crc32_string): Remove unnecessary braces.
27849 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27851 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27852 * ipa-inline-analysis.c (MAX_TIME): Remove.
27853 (account_size_time): Use sreal for time.
27854 (dump_inline_summary): Update.
27855 (estimate_function_body_sizes): Update.
27856 (estimate_edge_size_and_time): Update.
27857 (estimate_calls_size_and_time): Update.
27858 (estimate_node_size_and_time): Update.
27859 (inline_merge_summary): Update.
27860 (inline_update_overall_summary): Update.
27861 (estimate_time_after_inlining): Update.
27862 (inline_read_section): Update.
27863 (inline_write_summary): Update.
27864 * ipa-inline.c (compute_uninlined_call_time): Update.
27865 (compute_inlined_call_time): Update.
27866 (recursive_inlining): Update.
27867 (inline_small_functions): Update.
27868 (dump_overall_stats): Update.
27869 * ipa-inline.h: Include sreal.h.
27870 (size_time_entry): Turn time to sreal.
27871 (inline_summary): Turn self_time nad time to sreal.
27873 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27875 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27877 (sreal::stream_out, sreal::stream_in): New.
27878 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27880 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27882 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27885 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27888 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27890 (dimode_scalar_chain::compute_convert_gain): Ditto.
27891 (dimode_scalar_chain::make_vector_copies): Ditto.
27892 (dimode_scalar_chain::convert_reg): Ditto.
27893 (dimode_scalar_chain::convert_insn): Ditto.
27894 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27895 (VI248_AVX512BW_1): New mode iterator.
27896 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27897 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27900 2017-04-25 Martin Sebor <msebor@redhat.com>
27902 PR tree-optimization/80497
27903 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27904 constants are representable in HOST_WIDE_INT.
27905 (parse_directive): Ditto.
27907 2017-04-25 Martin Sebor <msebor@redhat.com>
27910 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27911 (new_zero_array): Adjust signature.
27912 (dom_info::dom_init): Used unsigned rather that size_t.
27913 (dom_info::dom_info): Same.
27915 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27916 Jakub Jelinek <jakub@redhat.com>
27919 * config/arm/arm.c: Include gimple.h.
27920 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27921 returns negative, increment ncrn only if it returned positive.
27922 (arm_needs_doubleword_align): Return int instead of bool,
27923 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27924 members, but if there is any such non-FIELD_DECL
27925 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27926 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27927 returns negative, increment nregs only if it returned positive.
27928 (arm_setup_incoming_varargs): Likewise.
27929 (arm_function_arg_boundary): Emit -Wpsabi note if
27930 arm_needs_doubleword_align returns negative, return
27931 DOUBLEWORD_ALIGNMENT only if it returned positive.
27933 2017-04-25 Marek Polacek <polacek@redhat.com>
27936 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27937 first argument to type.
27939 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27942 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27943 type checks to test for compatibility instead of equality.
27945 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27946 Jakub Jelinek <jakub@redhat.com>
27949 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27951 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27952 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27953 the alignment computation, but return their maximum in warn_alignment.
27954 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27955 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27957 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27958 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27961 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27963 * config/arc/simdext.md (dmpyh): Fix typo.
27965 2017-04-25 Richard Biener <rguenther@suse.de>
27967 PR tree-optimization/80492
27968 * alias.c (compare_base_decls): Handle registers with asm
27969 specification conservatively.
27970 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27971 compare_base_decls returning dont-know properly.
27973 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27975 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27976 (legitimate_offset_address_p): New function.
27977 (arc_legitimate_address_p): Use above function.
27979 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27981 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27983 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27985 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27986 ACCH registers whenever they are available.
27988 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27990 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27991 double regs fix when not used.
27993 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27995 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27997 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27998 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28000 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28002 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28003 when emitting short ADD instructions.
28005 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28007 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28009 (cmpsi_cc_c_insn): Likewise.
28010 (cbranchsi4_scratch): Compute proper instruction length using
28011 compact_hreg_operand.
28012 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28014 2017-04-25 Richard Biener <rguenther@suse.de>
28016 PR middle-end/80509
28017 * passes.c (pass_manager::pass_manager): Initialize
28018 m_name_to_pass_map.
28020 2017-04-25 Richard Biener <rguenther@suse.de>
28022 PR tree-optimization/79201
28023 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28025 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28028 * config/s390/vector.md: Split MEM->GPR vector moves for
28029 non-s_operand addresses.
28031 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28034 * config/s390/predicates.md (reload_const_wide_int_operand): New
28036 * config/s390/s390.md ("movti"): Remove d/P alternative.
28037 ("movti_bigconst"): New pattern definition.
28039 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28042 * s390-protos.h (s390_expand_cs_hqi): Removed.
28043 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28044 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28045 modes as well as CCZ1mode and CCZmode.
28046 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28047 signature of s390_emit_compare_and_swap.
28048 (s390_expand_cs_hqi): Likewise, make static.
28049 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28050 compare-and-swap, in some cases.
28051 (s390_expand_cs): Wrapper function.
28052 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28054 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28055 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28056 patterns for small and large integers. Forbid symref memory operands.
28057 Move expander to s390.c. Require cc register.
28058 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28059 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28060 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28061 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28062 symref memory operands. Remove CC mode and call s390_match_ccmode
28064 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28066 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28068 * config/s390/s390.md (define_peephole2): New peephole to help
28069 combining the load-and-test pattern with volatile memory.
28071 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28073 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28074 with CCZmode for TARGET_Z196.
28076 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28078 PR rtl-optimization/80501
28079 * combine.c (make_compound_operation_int): Set subreg_code to SET
28080 even for AND with mask of the sign bit of mode.
28082 PR rtl-optimization/80500
28083 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28084 sum's initial value.
28086 2017-04-25 Julian Brown <julian@codesourcery.com>
28087 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28089 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28091 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28093 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28095 2017-04-25 Julian Brown <julian@codesourcery.com>
28096 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28098 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28099 (thunderx2t99_sha): New Reservation.
28101 2017-04-25 Julian Brown <julian@codesourcery.com>
28102 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28104 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28105 type for 1-element load.
28107 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28109 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28111 2017-04-24 Martin Jambor <mjambor@suse.cz>
28113 PR tree-optimization/80293
28114 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28115 char arrays not totally scalarizable if it is false.
28116 (analyze_all_variable_accesses): Pass correct value in the new
28117 parameter. Add a statistics counter.
28119 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28121 PR middle-end/79931
28122 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28124 2017-04-24 Richard Biener <rguenther@suse.de>
28126 PR tree-optimization/80494
28127 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28128 out for complex types.
28130 2017-04-24 Richard Biener <rguenther@suse.de>
28132 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28133 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28134 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
28135 (DFS): Adjust and never fail.
28136 (sccvn_dom_walker::fail): Remove.
28137 (sccvn_dom_walker::before_dom_children): Adjust.
28138 (run_scc_vn): Likewise and never fail.
28139 * tree-ssa-pre.c (pass_pre::execute): Adjust.
28140 (pass_fre::execute): Likewise.
28142 2017-04-24 Richard Biener <rguenther@suse.de>
28144 PR tree-optimization/79725
28145 * tree-ssa-sink.c (statement_sink_location): Return whether
28146 failure reason was zero uses. Move that check later.
28147 (sink_code_in_bb): Deal with zero uses by removing the stmt
28150 2017-04-24 Richard Biener <rguenther@suse.de>
28153 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
28154 pointer-based references.
28156 2017-04-24 Richard Biener <rguenther@suse.de>
28159 * pass_manager.h (pass_manager::operator new): Remove.
28160 (pass_manager::operator delete): Likewise.
28161 * passes.c (pass_manager::operator new): Remove.
28162 (pass_manager::operator delete): Likewise.
28163 (pass_manager::pass_manager): Zero individual pass members.
28165 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
28168 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
28169 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
28170 Check "XEXP (src, 1)" operand here.
28171 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
28172 Check "XEXP (src, 1)" operand here.
28173 (dimode_scalar_chain::make_vector_copies): Detect count register
28174 of a shift instruction. Zero extend count register from QImode
28175 to DImode to satisfy vector shift pattern count operand predicate.
28176 Substitute vector shift count operand with a DImode copy.
28177 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
28180 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
28182 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
28183 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28184 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
28185 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28186 (UNSPEC_NOREX_MEM): Remove definition.
28188 2017-04-21 Richard Biener <rguenther@suse.de>
28190 PR tree-optimization/79547
28191 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28192 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
28193 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
28194 without any constraints.
28196 2017-04-21 Richard Biener <rguenther@suse.de>
28198 PR tree-optimization/78847
28199 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
28201 2017-04-21 Richard Biener <rguenther@suse.de>
28203 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
28204 (build_distinct_type_copy): Likewise.
28205 (build_variant_type_copy): Likewise.
28206 * tree.c (build_qualified_type): Pass down mem-stat info.
28207 (build_distinct_type_copy): Likewise.
28208 (build_variant_type_copy): Likewise.
28210 2017-04-21 Richard Biener <rguenther@suse.de>
28212 PR tree-optimization/80237
28213 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
28215 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
28216 for a simplified result.
28218 2016-04-21 Richard Biener <rguenther@suse.de>
28220 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
28221 sth as strict as a simple_iv but a chrec without symbols and an
28222 operand defined in the loop we are peeling (and not some subloop).
28223 (propagate_constants_for_unrolling): Propagate all constants.
28225 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28228 * config/i386/i386.c (print_reg): Remove assert for disalowed
28229 regno values, call output_operand_lossage instead.
28231 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28234 * config/i386/constraints.md (Yc): New register constraint.
28235 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28236 Use Yc constraint for alternative 2 of operand 0. Remove
28237 preferred_for_speed attribute.
28239 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28241 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28242 lastprivate clauses in SIMT case.
28244 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28246 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28248 2017-04-20 Richard Biener <rguenther@suse.de>
28250 PR tree-optimization/57796
28251 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28252 as N scalar stores.
28253 (vect_model_load_cost): Cost gathers as N scalar loads.
28255 2017-04-20 Richard Biener <rguenther@suse.de>
28257 * ggc-page.c (ggc_allocated_p): Rename to ...
28258 (safe_lookup_page_table_entry): ... this and return the lookup
28260 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28262 2017-04-20 Richard Biener <rguenther@suse.de>
28264 PR tree-optimization/80453
28265 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28266 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28267 from the conditions.
28268 (vn_phi_eq): Pass them down.
28269 (vn_phi_lookup): Record them.
28270 (vn_phi_insert): Likewise.
28272 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28274 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28275 uninitialized variable warning to avoid buffer overrun.
28277 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28280 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28281 is suppressed for '{ 0 }' in C.
28283 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28285 * BASE-VER: Set to 8.0.0.
28287 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28289 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28290 priority .init_array and .fini_array section with SECTION_NOTYPE
28293 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28295 PR middle-end/80423
28296 * tree.h (build_array_type): Add typeless_storage default argument.
28297 * tree.c (type_cache_hasher::equal): Also compare
28298 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28299 (build_array_type): Add typeless_storage argument, set
28300 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28302 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28303 (build_array_type): Likewise. Add typeless_storage argument.
28305 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28306 Jakub Jelinek <jakub@redhat.com>
28308 PR tree-optimization/80426
28309 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28310 operation on symbolic operands, also compute the overflow for the
28311 invariant part when the operation degenerates into a negation.
28313 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28316 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28317 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28320 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28322 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28325 * config/avr/avr.c (tree.h): Include it.
28326 (cgraph.h): Include it.
28327 (avr_encode_section_info): Don't warn for uninitialized progmem
28328 variable if it's just an alias.
28330 2017-04-19 Richard Biener <rguenther@suse.de>
28333 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28334 when needed by AutoPGO.
28336 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28339 * doc/lto.texi: Remove an extra 'that'.
28341 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28343 PR rtl-optimization/80429
28344 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28345 are only used in debug insns.
28347 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28348 Vladimir Makarov <vmakarov@redhat.com>
28350 * config/sparc/predicates.md (input_operand): Add comment. Return
28351 true for any memory operand when LRA is in progress.
28352 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28354 2017-04-18 Jeff Law <law@redhat.com>
28357 * mips.md ({return,simple_return}_internal): Do not overwrite
28360 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28362 PR tree-optimization/80443
28363 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28364 instead of adding 1, subtract -1 and similarly instead of subtracting
28367 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28369 PR rtl-optimization/80357
28370 * haifa-sched.c (tmp_bitmap): New variable.
28371 (model_recompute): Handle duplicate use records.
28372 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28373 (free_global_sched_pressure_data): Free it.
28375 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28378 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28379 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28380 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28381 instead of SYSTEM_HEADER_DIR.
28383 2017-04-18 Jeff Law <law@redhat.com>
28385 PR middle-end/80422
28386 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28387 predecessors after walking up the insn chain.
28389 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28392 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28393 sizetype type into debug info.
28395 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28398 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28399 unneeded test for TARGET_UPPER_REGS_SF.
28400 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28402 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28405 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28406 instead of gsi_after_labels.
28408 2017-04-18 Jeff Law <law@redhat.com>
28410 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28414 2017-04-13 Jeff Law <law@redhat.com>
28415 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28416 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28418 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28421 * config/avr/avr.c (intl.h): Include it.
28422 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28424 2017-04-18 Martin Liska <mliska@suse.cz>
28426 PR gcov-profile/78783
28427 * gcov-tool.c (gcov_output_files): Validate that destination
28428 file is either removed by the tool or by a user.
28430 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28431 Guy Benyei <guybe@mellanox.com>
28433 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28434 block, and do not negate it, the stored id is already negative.
28436 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28438 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
28440 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28443 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
28444 masks of options that should be turned off if the VSX vector
28445 options are turned off.
28446 (OTHER_P8_VECTOR_MASKS): Likewise.
28447 (OTHER_VSX_VECTOR_MASKS): Likewise.
28448 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
28449 rs6000_disable_incompatible_switches to validate no type switches
28451 (rs6000_incompatible_switch): New function to disallow turning on
28452 other vector options if -mno-vsx, -mno-power8-vector, or
28453 -mno-power9-vector are specified.
28455 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28457 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
28459 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28461 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
28462 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
28463 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
28464 (ARG_POINTER_CFA_OFFSET): Likewise.
28466 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28468 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
28469 conditions to take advantage of various optimizations.
28471 2017-04-13 Jeff Law <law@redhat.com>
28473 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28474 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28475 (zero_extendsidi2_dext): Likewise.
28477 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28480 * fold-const.c (fold_ternary_loc): Revert
28481 use op0 instead of fold_convert_loc (loc, type, arg0) part of
28484 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
28486 PR rtl-optimization/80343
28487 * lra-remat.c (update_scratch_ops): Assign original hard reg to
28488 new scratch pseudo.
28490 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
28493 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
28494 to ubsan_encode_value.
28496 2017-04-13 Jeff Law <law@redhat.com>
28498 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
28499 appearing in DEBUG_INSNs.
28501 2017-04-13 Martin Liska <mliska@suse.cz>
28503 PR gcov-profile/80413
28504 * gcov-io.c (gcov_write_string): Copy to buffer just when
28505 allocated size is greater than zero.
28507 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28510 * dwarf2out.c (decls_for_scope): Ignore declarations of
28511 current_function_decl in BLOCK_NONLOCALIZED_VARS.
28513 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
28516 * ipa-visibility.c (non_local_p): Fix typos.
28517 (localize_node): When localizing symbol in same comdat group,
28518 dissolve the group only when we know external symbols are going
28520 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
28522 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28524 PR tree-optimization/79390
28525 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
28526 order does not result in usable sequence, retry with reversed operand
28532 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
28533 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
28534 op0 instead of fold_convert_loc (loc, type, arg0).
28536 2017-04-12 Jeff Law <law@redhat.com>
28538 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
28539 has a delay slot in the generated code.
28541 * config/cris/cris.md (cris_preferred_reload_class): Return
28542 GENNONACR_REGS rather than GENERAL_REGS.
28544 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28547 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
28548 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
28549 signedness of the result type.
28551 2017-04-12 Richard Biener <rguenther@suse.de>
28552 Jeff Law <law@redhat.com>
28554 PR tree-optimization/80359
28555 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
28556 trim stores to TARGET_MEM_REFs.
28558 2017-04-12 Richard Biener <rguenther@suse.de>
28560 PR tree-optimization/79390
28561 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
28562 threading case even more.
28564 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
28567 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
28568 for quad_address_p for TImode, instead of just not indexed_address.
28570 2017-04-12 Richard Biener <rguenther@suse.de>
28571 Bernd Edlinger <bernd.edlinger@hotmail.de>
28573 PR middle-end/79671
28574 * alias.c (component_uses_parent_alias_set_from): Handle
28575 TYPE_TYPELESS_STORAGE.
28576 (get_alias_set): Likewise.
28577 * tree-core.h (tree_type_common): Add typeless_storage flag.
28578 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28579 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28580 for types containing members with TYPE_TYPELESS_STORAGE.
28581 (place_field): Likewise.
28582 (layout_type): Likewise for ARRAY_TYPE.
28583 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28584 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28585 TYPE_TYPELESS_STORAGE.
28586 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28588 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28591 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28592 first argument to type.
28594 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28598 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28599 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28600 (rs6000_expand_binop_builtin): Likewise.
28601 (rs6000_expand_ternop_builtin): Likewise; also add missing
28602 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28603 vshasigma built-ins.
28604 * doc/extend.texi: Document that vec_xxpermdi's third argument
28605 must be a constant.
28607 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28609 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28610 Use shift_const cost parameter when calculating gain of STV shifts.
28612 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28614 PR rtl-optimization/70478
28615 * lra-constraints.c (process_alt_operands): Check memory for
28616 disfavoring memory insn operand.
28618 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28620 PR middle-end/80100
28621 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28622 left shift in unsigned HOST_WIDE_INT type.
28624 PR rtl-optimization/80385
28625 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28626 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28629 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28630 if they have any depend clauses.
28632 2017-04-11 Martin Liska <mliska@suse.cz>
28635 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28636 * ipa-split.c (split_function): Create a local comdat symbol
28637 if caller is in a comdat group.
28639 2017-04-11 Martin Liska <mliska@suse.cz>
28642 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28645 2017-04-11 Martin Sebor <msebor@redhat.com>
28647 PR middle-end/80364
28648 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28649 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28651 (directive::set_width, directive::set_precision, format_character):
28653 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28656 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28659 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28660 conflict, set target->arch_name instead of target->cpu_name.
28662 2017-04-11 Richard Biener <rguenther@suse.de>
28664 PR tree-optimization/80374
28665 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28666 build_zero_cst, remove fold_convertible_p check again.
28668 2017-04-11 Martin Liska <mliska@suse.cz>
28671 * ubsan.c (instrument_object_size): Do not instrument register
28674 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28677 * config/i386/i386-builtin-types.def
28678 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28679 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28680 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28681 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28682 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28683 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28684 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28685 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28686 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28687 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28688 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28689 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28690 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28691 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28692 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28693 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28694 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28695 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28696 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28697 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28698 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28699 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28700 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28701 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28702 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28703 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28704 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28705 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28706 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28707 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28708 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28709 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28710 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28711 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28712 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28713 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28714 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28715 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28716 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28717 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28718 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28719 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28720 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28721 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28722 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28723 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28725 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28726 flag to second_arg_count, handle 4 argument function type _COUNT
28727 aliases, handle second_arg_count on second argument rather than last.
28729 2017-04-10 Jeff Law <law@redhat.com>
28731 PR tree-optimization/80374
28732 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28733 record anything if we can not convert integer_zero_node to the
28736 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28739 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28740 Enhance special handling given to the TARGET_P9_MINMAX option in
28741 relation to certain other options.
28743 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28745 PR tree-optimization/80153
28746 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28747 remove POINTER_PLUS_EXPR's base part directly, rather than through
28750 2017-04-10 Richard Biener <rguenther@suse.de>
28751 Bin Cheng <bin.cheng@arm.com>
28753 PR tree-optimization/80153
28754 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28755 the first element of pointer type aff_tree. Build result expr in
28757 (add_elt_to_tree): Convert to type unconditionally. Remove other
28758 fold_convert calls.
28759 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28760 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28762 2017-04-10 Richard Biener <rguenther@suse.de>
28764 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28767 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28769 PR rtl-optimization/70478
28770 * lra-constraints.c (curr_small_class_check): New.
28771 (update_and_check_small_class_inputs): New.
28772 (process_alt_operands): Update curr_small_class_check. Disfavor
28773 alternative insn memory operands. Check available regs for small
28776 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28779 * config/mips/mips.opt (-mvirt): Update description.
28780 * doc/invoke.texi (-mvirt): Likewise.
28782 2017-04-10 Richard Biener <rguenther@suse.de>
28784 PR middle-end/80362
28785 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28786 looking for NEGATE_EXPR in -A / -B to A / B folding.
28788 2017-04-10 Martin Liska <mliska@suse.cz>
28790 PR gcov-profile/80224
28791 * gcov.c (print_usage): Fix usage string.
28792 (get_gcov_intermediate_filename): Remove.
28793 (output_gcov_file): Use both for normal and intermediate format.
28794 (generate_results): Do not initialize special file for
28795 intermediate format.
28797 2017-04-10 Richard Biener <rguenther@suse.de>
28799 PR tree-optimization/80304
28800 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28803 2017-04-10 Nathan Sidwell <nathan@acm.org>
28806 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28807 (rs6000_init_builtins): Use it.
28809 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28811 * config/arm/arm.md (<mrc>): Add mode to SET source.
28812 (<mrrc>): Likewise.
28814 2017-04-10 Richard Biener <rguenther@suse.de>
28816 PR middle-end/80344
28817 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28819 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28822 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28823 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28824 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28825 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28826 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28827 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28828 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28829 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28830 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28831 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28832 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28833 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28834 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28835 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28836 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28837 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28838 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28839 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28840 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28841 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28842 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28843 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28844 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28846 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28848 PR rtl-optimization/70478
28849 * lra-constraints.c: Reverse the last patch.
28851 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28853 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28854 Add comment for WCHAR_T.
28856 2017-04-08 Martin Liska <mliska@suse.cz>
28859 2017-04-07 Martin Liska <mliska@suse.cz>
28862 * ipa-split.c (split_function): Add function part to a same comdat
28865 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28868 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28870 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28872 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28873 (vec_pairing): ...to this.
28874 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28875 (rs6000_sched_init): Adjust for name change.
28876 (struct rs6000_sched_context): Likewise.
28877 (rs6000_init_sched_context): Likewise.
28878 (rs6000_set_sched_context): Likewise.
28880 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28886 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28888 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28889 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28890 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28892 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28894 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28896 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28898 PR rtl-optimization/70703
28899 * ira-color.c (update_conflict_hard_regno_costs): Use
28900 int64_t instead of HOST_WIDE_INT.
28902 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28904 PR rtl-optimization/70478
28905 * lra-constraints.c (process_alt_operands): Disfavor alternative
28906 insn memory operands.
28908 2017-04-07 Jeff Law <law@redhat.com>
28910 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28911 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28913 2017-04-07 Martin Liska <mliska@suse.cz>
28916 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28917 Show error message instead of an ICE.
28919 2017-04-07 Martin Liska <mliska@suse.cz>
28922 * ipa-split.c (split_function): Add function part to a same comdat
28925 2017-04-07 Richard Biener <rguenther@suse.de>
28927 PR middle-end/80341
28928 * tree.c (get_unwidened): Also handle ! for_type case for
28930 * convert.c (do_narrow): Split out from ...
28931 (convert_to_integer_1): ... here. Do not pass final truncation
28932 type to get_unwidened for TRUNC_DIV_EXPR.
28934 2017-04-07 Richard Biener <rguenther@suse.de>
28936 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28938 (aff_combination_const): Adjust.
28939 (aff_combination_scale): Likewise.
28940 (aff_combination_add_elt): Likewise.
28941 (aff_combination_add_cst): Likewise.
28942 (aff_combination_convert): Likewise.
28943 (add_elt_to_tree): Likewise. Remove unused argument.
28944 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28946 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28948 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28950 * config/arm/arm.c (arm_default_short_enums): Use
28951 ARM_DEFAULT_SHORT_ENUMS.
28952 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28954 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28957 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28958 members with redundant out-of-class redeclaration.
28960 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28963 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28964 * config/i386/i386.md (*zero_extendsidi2):
28965 Add (?*x,*x) and (?*v,*v) alternatives.
28967 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28970 * config/i386/i386.c (ix86_expand_builtin)
28971 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28972 mode from insn data. Convert operands to insn operand mode.
28973 Copy operands that don't satisfy insn predicate to a register.
28975 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28977 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28980 2017-04-06 Richard Biener <rguenther@suse.de>
28982 PR tree-optimization/80334
28983 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28984 preserve alignment of accesses.
28986 2017-04-06 Richard Biener <rguenther@suse.de>
28988 PR tree-optimization/80262
28989 * tree-sra.c (build_ref_for_offset): Preserve address-space
28991 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28992 Drop useless address-space information on MEM_REF offsets.
28994 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28996 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28998 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29000 PR rtl-optimization/70703
29001 * ira-color.c (update_conflict_hard_regno_costs): Use
29002 HOST_WIDE_INT instead of long.
29004 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29007 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29008 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29010 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29011 for x86_64 target. Handle -m3dnowa option.
29013 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29015 PR rtl-optimization/70703
29016 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29017 (update_conflict_hard_regno_costs): Use long instead of unsigned
29018 arithmetic for cost calculation.
29020 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29021 Bernd Edlinger <bernd.edlinger@hotmail.de>
29024 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29027 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29030 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29031 ptr_mode with Pmode throughout.
29032 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29033 into probe_stack_range and use DImode.
29035 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29038 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29039 call_eh_return is true.
29041 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29043 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29044 Initialize last_match_fntype_index.
29046 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29049 * tree-nvr.c: Include internal-fn.h.
29050 (pass_return_slot::execute): Ignore internal calls without
29053 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29054 Richard Biener <rguenther@suse.de>
29057 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29058 captures used multiple times, except for the last use.
29059 * generic-match-head.c: Include gimplify.h.
29061 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29063 PR tree-optimization/79390
29064 * target.h (struct noce_if_info): Declare.
29065 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29066 * target.def (noce_conversion_profitable_p): New target hook.
29067 * ifcvt.h (struct noce_if_info): New type, moved from ...
29068 * ifcvt.c (struct noce_if_info): ... here.
29069 (noce_conversion_profitable_p): Renamed to ...
29070 (default_noce_conversion_profitable_p): ... this. No longer
29072 (noce_try_store_flag_constants, noce_try_addcc,
29073 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29074 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29075 instead of noce_conversion_profitable_p.
29076 * config/i386/i386.c: Include ifcvt.h.
29077 (ix86_option_override_internal): Don't override
29078 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29079 (ix86_noce_conversion_profitable_p): New function.
29080 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29081 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29082 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29083 * doc/tm.texi: Regenerated.
29085 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29087 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29090 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29093 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29094 instructions for small multiply cores.
29096 2017-04-04 Jeff Law <law@redhat.com>
29098 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29100 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29101 that are not set by the loop over the elements.
29103 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29106 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29107 int mode, convert_modes it to mode as unsigned, otherwise use
29108 lowpart_subreg to mode rather than SImode.
29109 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29110 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29111 Use DImode instead of SImode for the shift count operand.
29112 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29115 2017-04-04 Richard Biener <rguenther@suse.de>
29117 PR middle-end/80281
29118 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29119 arithmetic done for the negate or the plus. Simplify.
29120 (A - (-B) -> A + B): Likewise.
29121 * fold-const.c (split_tree): Make sure to not negate pointers.
29123 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29125 PR rtl-optimization/60818
29126 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29127 a compare of comparisons with the thing compared if this results
29128 in a different machine mode.
29130 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29132 * alias.c (base_alias_check): Fix typo in comment.
29133 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29134 * cgraphunit.c (symbol_table::compile): Likewise.
29135 * collect2.c (maybe_run_lto_and_relink): Likewise.
29136 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
29137 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
29138 * config/avr/avr.c (avr_map_op_t): Likewise.
29139 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
29140 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
29141 * config/epiphany/epiphany.md (movcc): Likewise.
29142 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
29143 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
29145 * config/mips/mips.c (mips_save_restore_reg): Likewise.
29146 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
29147 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
29148 * config/sh/sh.c (sh_rtx_costs): Likewise.
29149 * fold-const.c (fold_truth_andor): Likewise.
29150 * genautomata.c (collapse_flag): Likewise.
29151 * gengtype.h (struct type::u::s): Likewise.
29152 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
29153 * input.c (FORMAT_AMOUNT): Likewise.
29154 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
29155 (known_aggs_to_agg_replacement_list): Likewise.
29156 * ipa-inline-analysis.c: Likewise.
29157 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
29158 * ipa-polymorphic-call.c
29159 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
29160 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
29161 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
29163 * modulo-sched.c (apply_reg_moves): Likewise.
29164 * omp-expand.c (build_omp_regions_1): Likewise.
29165 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
29166 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
29167 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
29168 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29169 * value-prof.c: Likewise.
29170 * var-tracking.c (val_reset): Likewise.
29172 2017-04-03 Richard Biener <rguenther@suse.de>
29174 PR tree-optimization/80275
29175 * fold-const.c (split_address_to_core_and_offset): Handle
29178 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
29180 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
29181 descriptors is at least equal to that of functions.
29183 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29185 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
29187 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29190 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
29191 (mov<IMOD4:mode>): New expander.
29192 (*mov<IMOD4:mode>_internal): New insn and split pattern.
29194 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
29196 PR rtl-optimization/79405
29197 * fwprop.c (propagations_left): New variable.
29198 (forward_propagate_into): Decrement it.
29199 (fwprop_init): Initialize it.
29200 (fw_prop): If the variable has reached zero, stop propagating.
29201 (fwprop_addr): Ditto.
29203 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29206 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
29207 a FUNCTION_DECL, pass it as decl instead of origin to
29210 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
29212 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
29215 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
29218 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
29219 TARGET_VSX_SMALL_INTEGER.
29221 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29223 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29224 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
29226 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29228 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
29229 extraction from odd-numbered MSA register.
29231 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29233 PR middle-end/80173
29234 * expmed.c (store_bit_field_1): Don't attempt to create
29235 a word subreg out of hard registers wider than word if they
29236 have HARD_REGNO_NREGS of 1 for their mode.
29238 PR middle-end/80163
29239 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29240 conversions to integer types wider than word and pointer.
29243 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29244 (rtx_equal_for_cselib_p): Pass 0 to it.
29245 * cselib.c (cselib_hasher::equal): Likewise.
29246 (rtx_equal_for_cselib_1): Add depth argument. If depth
29247 is 128, don't look up VALUE locs and punt. Increment
29248 depth in recursive calls when walking VALUE locs.
29250 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29252 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29253 (make_gcov_file_name): Use the canonical path name for generating
29255 (read_line): Fix handling of files with ascii null bytes.
29257 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29259 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29260 to initialise a vector register instead
29261 of using a const_int.
29263 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29265 PR translation/80189
29266 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29267 diagnostic messages.
29269 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29272 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29273 (dfp_diex_<mode>): Update mode of operand 1.
29274 * doc/extend.texi (dxex, dxexq): Document change to return type.
29275 (diex, diexq): Document change to argument type.
29277 2017-03-30 Martin Jambor <mjambor@suse.cz>
29280 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29281 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29282 it reflects the signature changes performed at the callee side.
29283 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29284 to cgraph_build_function_type_skip_args.
29285 (build_function_decl_skip_args): Adjust call to the above function.
29287 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29290 * config/i386/sse.md
29291 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29292 register as dest whenever it is a MEM not rtx_equal_p to the
29293 corresponding dup operand, and when forcing into reg move the
29294 reg into the memory afterwards.
29295 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29296 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29297 for the force_reg mode.
29298 (avx512vl_vextractf128<mode>): Use register as dest either
29299 always when a MEM, or when it is a MEM not rtx_equal_p to the
29300 corresponding dup operand, or even not when it is a CONST_VECTOR
29301 depending on the mode and lo vs. hi.
29302 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29304 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29305 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29306 Likewise. Require that operands[2] is even.
29307 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29308 Remove extraneous parens. Require that operands[2] is a multiple
29310 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29311 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29312 disallow memory then.
29314 2017-03-30 Richard Biener <rguenther@suse.de>
29316 PR tree-optimization/77498
29317 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29318 to non-constants over backedges.
29320 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29322 PR rtl-optimization/80233
29323 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29324 as last_combined_insn. Do not test for BARRIER_P separately.
29326 2017-03-29 Andreas Schwab <schwab@suse.de>
29329 * calls.c (prepare_call_address): Convert funexp to Pmode before
29330 copying to temp reg.
29332 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29334 PR tree-optimization/80158
29335 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29336 Handle possible future case of more than one alternate
29338 (replace_rhs_if_not_dup): Likewise.
29339 (replace_one_candidate): Likewise.
29341 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29343 PR rtl-optimization/80193
29344 * ira.c (ira): Do not check allocation for LRA.
29346 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29348 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29349 (nvptx_output_simt_exit): Declare.
29350 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29351 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29352 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29353 (nvptx_declare_function_name): ...here. Emit declaration of local
29354 memory space buffer for omp_simt_enter insn.
29355 (nvptx_output_unisimt_switch): New.
29356 (nvptx_output_softstack_switch): New.
29357 (nvptx_output_simt_enter): New.
29358 (nvptx_output_simt_exit): New.
29359 * config/nvptx/nvptx.h (struct machine_function): New fields
29360 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29361 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29362 (UNSPECV_SIMT_EXIT): Ditto.
29363 (omp_simt_enter_insn): New insn.
29364 (omp_simt_enter): New expansion.
29365 (omp_simt_exit): New insn.
29366 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29368 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29369 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29370 (expand_GOMP_SIMT_EXIT): New.
29371 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29372 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29373 (GOMP_SIMT_EXIT): Ditto.
29374 * target-insns.def (omp_simt_enter): New insn.
29375 (omp_simt_exit): Ditto.
29376 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29378 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29379 (lower_rec_input_clauses): Likewise.
29380 (lower_lastprivate_clauses): Handle SIMT privatization.
29382 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29383 (ompdevlow_adjust_simt_enter): New.
29384 (find_simtpriv_var_op): New.
29385 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29386 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29388 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29389 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29390 (copy_decl_for_dup_finish): Ditto.
29392 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29394 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29397 * config/i386/i386.c (ix86_option_override_internal): Always
29398 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29400 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29402 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29404 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29406 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29407 mark new edge's irreducible flag accordign to it.
29408 (vect_do_peeling): Check loop preheader edge's irreducible flag
29409 and pass it to function slpeel_add_loop_guard.
29411 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29413 PR tree-optimization/80218
29414 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29415 Update block frequencies and counts.
29417 2017-03-28 Richard Biener <rguenther@suse.de>
29419 PR tree-optimization/78644
29420 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29421 of a simplification result we may not use it at all.
29423 2017-03-28 Richard Biener <rguenther@suse.de>
29426 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29427 without arguments, generate default definition of a SSA name.
29429 2017-03-28 Richard Biener <rguenther@suse.de>
29431 PR middle-end/80222
29432 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29433 TYPE_REF_CAN_ALIAS_ALL references.
29434 * fold-const.c (fold_indirect_ref_1): Likewise.
29436 2017-03-28 Martin Liska <mliska@suse.cz>
29439 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
29440 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
29442 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29443 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
29445 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
29446 (EXTRA_SPECS): Define.
29447 (SUBTARGET_EXTRA_SPECS): Likewise.
29448 (SUBTARGET_CPP_SPEC): Likewise.
29449 * config/arc/elf.h (EXTRA_SPECS): Renamed to
29450 SUBTARGET_EXTRA_SPECS.
29451 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
29453 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29455 * config/arc/simdext.md (vst64_insn): Update pattern.
29456 (vld32wh_insn): Likewise.
29457 (vld32wl_insn): Likewise.
29458 (vld64_insn): Likewise.
29459 (vld32_insn): Likewise.
29461 2017-03-28 Marek Polacek <polacek@redhat.com>
29464 * fold-const.c (fold_comparison): Use protected_set_expr_location
29465 instead of SET_EXPR_LOCATION.
29467 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
29469 * tree.c (add_expr): Avoid name lookup warning.
29471 2017-03-27 Jeff Law <law@redhat.com>
29473 PR tree-optimization/80216
29474 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
29475 function name. Limit recursion depth.
29476 (record_temporary_equivalences): Corresponding changes.
29478 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
29480 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
29483 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29486 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
29488 * cfgcleanup.c (reg_note_cfa_p): New array.
29489 (insns_have_identical_cfa_notes): New function.
29490 (old_insns_match_p): Don't cross-jump in between /f
29491 and non-/f instructions. If both i1 and i2 are frame related,
29492 verify all CFA notes, their order and content.
29494 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
29497 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
29498 HImode and SImode with zero extend to DImode to one insn.
29499 (bswap<mode>2_extenddi): Likewise.
29500 (bswapsi2_extenddi): Likewise.
29501 (bswaphi2_extendsi): Likewise.
29502 (bswaphi2): Combine bswap HImode and SImode into one insn.
29503 Separate memory insns from swapping register.
29504 (bswapsi2): Likewise.
29505 (bswap<mode>2): Likewise.
29506 (bswaphi2_internal): Delete, no longer used.
29507 (bswapsi2_internal): Likewise.
29508 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
29509 store, and gpr<-gpr swap insns.
29510 (bswap<mode>2_store): Likewise.
29511 (bswaphi2_reg): Register only splitter, combine with the splitter.
29512 (bswaphi2 splitter): Likewise.
29513 (bswapsi2_reg): Likewise.
29514 (bswapsi2 splitter): Likewise.
29515 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
29516 the insns into load, store, and register/register insns.
29517 (bswapdi2_ldbrx): Likewise.
29518 (bswapdi2_load): Likewise.
29519 (bswapdi2_store): Likewise.
29520 (bswapdi2_reg): Likewise.
29522 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
29524 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
29525 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
29527 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
29530 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
29532 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29533 special handling for target option conflicts between dform
29534 options (-mpower9-dform, -mpower9-dform-vector,
29535 -mpower9-dform-scalar) and -mno-direct-move.
29537 2017-03-27 Richard Biener <rguenther@suse.de>
29539 PR tree-optimization/80181
29540 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
29542 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29544 * config/arc/predicates.md (move_double_src_operand): Replace the
29545 call to move_double_src_operand with a call to address_operand.
29547 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29549 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
29550 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
29551 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
29553 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29555 * config/arc/predicates.md (long_immediate_loadstore_operand):
29556 Consider scaled addresses cases.
29558 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29560 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
29561 restored when in interrupt.
29562 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
29563 doesn't have delay slot.
29565 2017-03-27 Richard Biener <rguenther@suse.de>
29568 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
29569 inlined thunk clones.
29571 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29574 * asan.c (instrument_derefs): Copy over last operand from
29575 original COMPONENT_REF to the new COMPONENT_REF with
29576 DECL_BIT_FIELD_REPRESENTATIVE.
29577 * ubsan.c (instrument_object_size): Likewise.
29579 2017-03-27 Richard Biener <rguenther@suse.de>
29581 PR tree-optimization/80170
29582 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29583 sure DR/SCEV didnt fold in constants we do not see when looking
29584 at the reference base alignment.
29586 2017-03-27 Richard Biener <rguenther@suse.de>
29588 PR middle-end/80171
29589 * gimple-fold.c (fold_ctor_reference): Properly guard against
29590 NULL return value from canonicalize_constructor_val.
29592 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29595 * config/i386/i386.c (ix86_expand_builtin)
29596 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29597 flags reg setting and flags reg using instructions.
29598 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29599 clobbering instructions to zero extend op2.
29601 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29603 * doc/install.texi (Configuration) <--with-aix-soname>:
29604 Update link to AIX ld.
29606 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29608 PR rtl-optimization/80160
29609 PR rtl-optimization/80159
29610 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29611 reg_alternate_class into account.
29613 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29616 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29617 to consider in curr_insn_transform.
29619 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29621 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29622 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29623 and emit_mode_inner.
29625 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29627 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29628 argument to the overloaded builtin variants. Use the new flag to
29629 deprecate certain builtin variants.
29630 * config/s390/s390-builtin-types.def: Add new builtin types.
29631 * config/s390/s390-builtins.h: Support new flags field for
29632 overloaded builtins.
29633 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29634 (s390_macro_to_expand): Enable vector float data type.
29635 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29636 builtins by incrementing the __VEC__ version number.
29637 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29639 (s390_resolve_overloaded_builtin): Emit error messages depending
29640 on the builtin flags.
29641 * config/s390/s390.c (s390_expand_builtin): Support additional
29642 flags argument. Change error message to match the messages
29643 emitted in s390-c.c.
29644 * config/s390/s390.md: New UNSPEC_* constants.
29645 (op_type): Add new instruction types.
29646 * config/s390/vecintrin.h: Add new builtins and test data class
29648 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29649 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29650 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29651 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29652 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29653 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29655 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29656 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29657 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29658 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29660 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29661 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29662 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29663 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29665 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29666 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29667 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29668 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29669 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29670 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29671 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29673 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29674 ("vec_scatter_element<V_HW_4:mode>_DI")
29675 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29676 ("vec_fpint<mode>", "vflls")
29677 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29678 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29679 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29680 ("*vec_cmphe<mode>_cc"): ... these.
29682 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29683 mode constant instead of magic value.
29685 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29687 * config/s390/s390.c (s390_expand_vec_compare): Support other
29688 vector floating point modes than just V2DF.
29689 (s390_expand_vcond): Likewise.
29690 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29691 (s390_cannot_change_mode_class): Prevent mode changes between TF
29692 and V1TF in vector registers.
29693 * config/s390/s390.md (DF, SF): New mode attributes.
29694 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29695 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29696 SFmode support for VRs.
29697 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29699 (VFT, VF_HW): New mode iterators.
29700 (vw, sdx): New mode attributes.
29701 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29702 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29703 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29704 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29705 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29706 also the new vector floating point modes. Renaming to ...
29708 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29709 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29710 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29711 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29712 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29713 ("vec_unordered<mode>"): ... these.
29715 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29716 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29717 ("*vec_extendv2df"): New insn definitions.
29719 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29721 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29722 ("mulditi3_2", "*muldi3_sign"): New patterns.
29723 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29724 rename the pattern definition.
29726 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29728 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29730 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29732 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29734 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29735 instruction if possible.
29736 * config/s390/vector.md (vec_halfnumelts): New mode
29738 ("*vec_vllezlf<mode>"): New pattern.
29740 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29742 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29743 ("popcountv4si2", "popcountv2di2"): Rename to ...
29744 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29745 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29747 ("popcount<mode>2_vxe"): New pattern.
29749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29751 * common/config/s390/s390-common.c (processor_flags_table): Add
29753 * config.gcc: Add arch12.
29754 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29755 Default to arch12 for unknown CPU model numbers.
29756 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29757 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29758 PROCESSOR_max sanity check.
29759 * config/s390/s390-opts.h (enum processor_type): Add
29761 * config/s390/s390.c (processor_table): Add arch12.
29762 (s390_expand_builtin): Add check for B_VXE flag.
29763 (s390_issue_rate): Add PROCESSOR_ARCH12.
29764 (s390_get_sched_attrmask): Likewise.
29765 (s390_get_unit_mask): Likewise.
29766 (s390_sched_score): Enable z13 scheduling for arch12.
29767 (s390_sched_reorder): Likewise.
29768 (s390_sched_variable_issue): Likewise.
29769 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29771 (s390_tune_attr): Use z13 scheduling also for arch12.
29772 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29773 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29774 (TARGET_VXE_P): New macros.
29775 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29776 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29777 * config/s390/s390.opt: Add arch12 as processor_type.
29779 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29781 * config/s390/s390.md
29782 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29783 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29784 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29786 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29787 Rename expanders to ...
29789 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29790 ("fixuns_truncdddi2_emu"): ... these.
29792 ("fixuns_trunc<mode>si2_emu"): New expander.
29794 ("*fixuns_truncdfdi2_z13"): Rename to ...
29795 ("*fixuns_truncdfdi2_vx"): ... this.
29797 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29799 * config/s390/2964.md: Remove the single element vector compare
29800 instructions which are no longer used.
29801 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29803 (s390_canonicalize_comparison): Remove handling of DFmode
29805 (s390_expand_vec_compare_scalar): Remove function.
29806 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29807 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29809 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29811 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29813 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29815 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29816 will anyway by matched by mov<mode>_64dfp.
29818 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29820 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29821 vlef/vstef. Add missing operand to vleif.
29823 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29825 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29826 pair for all vector types with 64 bit elements.
29827 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29828 * config/s390/vector.md (V_HW_64): ... here.
29829 (V_128_NOSINGLE): New mode iterator.
29830 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29831 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29832 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29833 ("*vec_load_pairv2di"): Change to ...
29834 ("*vec_load_pair<mode>"): ... this one.
29836 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29838 * config/s390/constraints.md: Add comments.
29839 (jKK): Reject element sizes > 8 bytes.
29840 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29842 * config/s390/s390.md: Add the s_operand checks formerly in
29843 s390_split_ok_p to various splitters where they are still
29845 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29846 for 128 bit vectors. Plus two splitters.
29848 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29850 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29856 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29857 error if the boundary argument is not constant.
29859 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29861 PR rtl-optimization/80112
29862 * loop-doloop.c (doloop_condition_get): Don't check condition
29863 if cmp isn't SET with IF_THEN_ELSE src.
29865 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29867 PR tree-optimization/80158
29868 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29869 replacing a candidate statement, also replace it for the
29870 candidate's alternate interpretation.
29871 (replace_rhs_if_not_dup): Likewise.
29872 (replace_one_candidate): Likewise.
29874 2017-03-24 Richard Biener <rguenther@suse.de>
29876 PR tree-optimization/80167
29877 * graphite-isl-ast-to-gimple.c
29878 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29880 (translate_isl_ast_to_gimple::get_rename): Likewise.
29882 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29884 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29885 handling of certain combinations of target options, including the
29886 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29887 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29889 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29892 * config/arm/arm.md (*load_multiple): Add reload_completed to
29893 matching condition.
29895 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29896 Richard Biener <rguenth@suse.de>
29898 PR tree-optimization/79908
29899 PR tree-optimization/80136
29900 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29901 been cast away, gimplify_and_add suffices.
29903 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29905 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29907 2017-03-23 Richard Biener <rguenther@suse.de>
29909 PR tree-optimization/80032
29910 * gimplify.c (gimple_push_cleanup): Forced unconditional
29911 cleanups still have to go to the conditional_cleanups
29914 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29916 PR tree-optimization/80072
29917 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29919 (next_operand_entry_id): Change type to unsigned int.
29920 (sort_by_operand_rank): Make sure to return the right return value
29921 even if unsigned fields are bigger than INT_MAX.
29922 (struct oecount): Change cnt and id type to unsigned int.
29923 (oecount_hasher::equal): Formatting fix.
29924 (oecount_cmp): Make sure to return the right return value
29925 even if unsigned fields are bigger than INT_MAX.
29926 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29929 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29930 TREE_READONLY on result if writing it more than once.
29933 * doc/invoke.texi (-fsanitize=thread): Document that with
29934 -fnon-call-exceptions atomics are not able to throw
29938 * tsan.c: Include tree-eh.h.
29939 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29940 maybe_clean_or_replace_eh_stmt where needed.
29941 (instrument_memory_accesses): Add cfg_changed argument.
29942 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29943 if it returned true.
29944 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29946 PR rtl-optimization/63191
29947 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29948 wrapper function, moved the whole old content into ...
29949 (ix86_delegitimize_address_1): ... this. New inline function.
29950 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29951 true as last argument instead of ix86_delegitimize_address.
29953 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29955 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29956 cortexa57_branch_cost.
29958 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29960 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29962 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29965 * doc/md.texi (Constraints): Document wA constraint.
29966 * config/rs6000/constraints.md (wA): New.
29967 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29968 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29969 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29970 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29972 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29975 * gimplify.c (is_oacc_declared): New function.
29976 (oacc_default_clause): Use it to set default flags for acc declared
29977 variables inside parallel regions.
29978 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29979 declared variables.
29980 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29981 declare attribute to any decl as necessary.
29983 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29986 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29987 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29988 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29989 (arm_arch_lpae): This.
29990 * config/arm/arm.c (arm_arch7ve): Rename into ...
29991 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29992 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29995 2017-03-22 Martin Liska <mliska@suse.cz>
29998 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29999 error message instead of an ICE.
30001 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30003 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30005 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30007 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30009 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30012 2017-03-21 Martin Sebor <msebor@redhat.com>
30014 * doc/extend.texi: Use "cannot" instead of "can't."
30015 * doc/hostconfig.texi: Same.
30016 * doc/install.texi: Same.
30017 * doc/invoke.texi: Same.
30018 * doc/loop.texi: Same.
30019 * doc/md.texi: Same.
30020 * doc/objc.texi: Same.
30021 * doc/rtl.texi: Same.
30022 * doc/tm.texi: Same.
30023 * doc/tm.texi.in: Same.
30024 * doc/trouble.texi: Same.
30026 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30029 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30030 (collect_checksum_attributes): Set it.
30031 (die_checksum_ordered): Use it.
30033 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30035 PR tree-optimization/79908
30036 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30037 change: For a VA_ARG whose LHS has been cast away, use
30038 force_gimple_operand to construct the side effects.
30040 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30042 PR translation/80001
30043 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30044 more amenable to translation.
30045 (oacc_loop_auto_partitions): Likewise.
30047 2017-03-21 Marek Polacek <polacek@redhat.com>
30048 Martin Sebor <msebor@redhat.com>
30050 PR tree-optimization/80109
30051 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30052 on INTEGRAL_TYPE_P.
30054 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30055 Segher Boessenkool <segher@kernel.crashing.org>
30058 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30059 check reg_used_between_p between insn and one of succ or succ2
30060 depending on if succ is artificial insn not inserted into insn
30063 2017-03-21 Martin Liska <mliska@suse.cz>
30065 PR gcov-profile/80081
30066 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30067 * doc/gcc.texi: Include gcov-dump stuff.
30068 * doc/gcov-dump.texi: New file.
30070 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30072 PR rtl-optimization/79150
30073 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30074 conditional jump, if the jump is the last insn of the loop.
30076 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30077 Richard Biener <rguenth@suse.de>
30079 PR tree-optimization/79908
30080 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30081 been cast away, use force_gimple_operand to construct the side
30084 2017-03-21 Martin Liska <mliska@suse.cz>
30086 PR libfortran/79956
30087 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30090 2017-03-21 Brad Spengler <spender@grsecurity.net>
30093 * plugin.c (htab_hash_plugin): New function.
30094 (add_new_plugin): Use it and adjust.
30095 (parse_plugin_arg_opt): Adjust.
30096 (init_one_plugin): Likewise.
30098 2017-03-21 Richard Biener <rguenther@suse.de>
30100 PR tree-optimization/80032
30101 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30102 if set force the cleanup to happen unconditionally.
30103 (gimplify_target_expr): Push inserted clobbers with force_uncond
30104 to avoid them being removed by control-dependent DCE.
30106 2017-03-21 Richard Biener <rguenther@suse.de>
30108 PR tree-optimization/80122
30109 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30110 va_arg_pack_len when the inlined call stmt requires pack
30112 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30114 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30117 * tsan.c (instrument_builtin_call): If the memory model argument
30118 is not a constant, assume it is valid.
30121 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30124 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30126 PR rtl-optimization/79910
30127 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30128 if its dest is used by an insn before I2 (other than the combined
30129 insns themselves, which are properly handled already).
30131 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30134 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30136 * combine.c (record_used_regs): New static function.
30137 (try_combine): Handle situations where there is an additional
30138 instruction between I2 and I3 which needs to have a LOG_LINK
30142 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30144 * combine.c (try_combine): Delete redundant i1 test. Call
30145 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30147 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30150 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
30151 alternatives 13/14.
30153 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30155 PR tree-optimization/80054
30156 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
30157 the optimization if a PHI or any of its arguments is not dominated
30158 by the candidate's basis. Use gphi* rather than gimple* as
30160 (replace_profitable_candidates): Clean up a gimple* variable that
30161 should be a gphi* variable.
30163 2017-03-20 Martin Sebor <msebor@redhat.com>
30166 * doc/extend.texi (attribute constructor): Document present limitation.
30168 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
30171 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
30172 __POWER9_VECTOR__ #ifdef control, change template definition to
30173 use Power9-specific built-in function.
30174 (vec_any_eq): Likewise.
30175 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
30176 to control outcomes from this test.
30177 (vector_ae_<mode>p): For VEC_F modes, likewise.
30179 2017-03-20 Ian Lance Taylor <iant@google.com>
30181 * config/i386/i386.c (ix86_function_regparm): Save an extra
30182 register for -fsplit-stack with DECL_STATIC_CHAIN.
30184 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30187 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
30188 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
30190 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30192 * config/riscv/riscv.c (riscv_print_operand): Use "fence
30194 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
30197 2017-03-20 Marek Polacek <polacek@redhat.com>
30200 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
30202 2017-03-20 Richard Biener <rguenther@suse.de>
30204 PR tree-optimization/80113
30205 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
30206 allocate extra SSA name for PHI def.
30207 (add_close_phis_to_outer_loops): Likewise.
30208 (add_close_phis_to_merge_points): Likewise.
30209 (copy_loop_close_phi_args): Likewise.
30210 (copy_cond_phi_nodes): Likewise.
30212 2017-03-20 Martin Liska <mliska@suse.cz>
30214 PR middle-end/79753
30215 * tree-chkp.c (chkp_build_returned_bound): Do not build
30216 returned bounds for a LHS that's not a BOUNDED_P type.
30218 2017-03-20 Martin Liska <mliska@suse.cz>
30222 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
30223 COMPLEX_CST and VECTOR_CST.
30225 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30228 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
30229 target operand. A new splitter adds the clobber statement in case
30230 the target operand is dead anyway.
30232 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30234 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30235 to age-old versions of binutils and glibc.
30237 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30239 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30241 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30243 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30245 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30247 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30248 requirement for binutils 2.13.
30250 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30252 * combine.c (try_combine): Delete redundant i1 test. Call
30253 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30255 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30257 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30258 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30260 <riscv64-*-elf>: Re-arrange section
30261 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30262 <riscv32-*-linux>: Likewise.
30263 <riscv64-*-elf>: Likewise
30264 <riscv64-*-linux>: Likewise.
30266 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30269 * aarch64.opt(verbose-cost-dump): Fix typo.
30271 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30274 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30275 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30277 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30279 * reload.c (find_reloads): When reloading a nonoffsettable address,
30280 use RELOAD_OTHER for it and its address reloads.
30282 PR rtl-optimization/79910
30283 * combine.c (record_used_regs): New static function.
30284 (try_combine): Handle situations where there is an additional
30285 instruction between I2 and I3 which needs to have a LOG_LINK
30288 2017-03-17 Jeff Law <law@redhat.com>
30290 PR tree-optimization/71437
30291 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30292 conditional in the hash table first.
30293 (vrp_dom_walker::before_dom_children): Extract condition from
30294 ASSERT_EXPR. Record condition, its inverion and any implied
30295 conditions as well.
30297 2017-03-17 Marek Polacek <polacek@redhat.com>
30298 Markus Trippelsdorf <markus@trippelsdorf.de>
30300 PR tree-optimization/80079
30301 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30304 2017-03-17 Richard Biener <rguenther@suse.de>
30306 PR middle-end/80075
30307 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30308 Properly verify the LHS before the RHS possibly claims to be
30310 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30313 2017-03-17 Martin Jambor <mjambor@suse.cz>
30315 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30316 (List of -O2 options): Likewise.
30317 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30320 2017-03-17 Tom de Vries <tom@codesourcery.com>
30322 * gcov-dump.c (print_usage): Print bug_report_url.
30324 2017-03-17 Richard Biener <rguenther@suse.de>
30326 PR middle-end/80050
30327 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30328 (parser::peek): Likewise.
30330 2017-03-17 Richard Biener <rguenther@suse.de>
30332 PR tree-optimization/80048
30333 * sese.c (free_sese_info): Properly release rename_map and
30334 copied_bb_map elements.
30336 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30338 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30339 Add linked-list forward and backlinks. Insert on
30340 construction, remove on destruction.
30341 (class pass_store_merging): Add m_stores_head field.
30342 (pass_store_merging::terminate_and_process_all_chains):
30343 Iterate over m_stores_head list.
30344 (pass_store_merging::terminate_all_aliasing_chains):
30346 (pass_store_merging::execute): Check for debug stmts first.
30347 Push new chains onto the m_stores_head stack.
30349 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30352 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30353 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30354 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30356 2017-03-16 Jeff Law <law@redhat.com>
30358 PR tree-optimization/71437
30359 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30360 member function. Implementation moved into after_dom_children
30361 member function and into the threader's thread_outgoing_edges
30363 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30364 some code into new thread_outgoing_edges.
30365 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30366 definition. Simplify marker handling (do it here). Assume we always
30367 have the available expression and the const/copies tables.
30368 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30370 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30371 * tree-vrp.c (equiv_stack): No longer file scoped.
30372 (vrp_dom_walker): New class.
30373 (vrp_dom_walker::before_dom_children): New member function.
30374 (vrp_dom_walker::after_dom_children): Likewise.
30375 (identify_jump_threads): Setup domwalker. Use it rather than
30376 walking edges in a random order by hand. Simplify setup/finalization.
30377 (finalize_jump_threads): Remove.
30378 (vrp_finalize): Do not call identify_jump_threads here.
30379 (execute_vrp): Do it here instead and call thread_through_all_blocks
30382 PR tree-optimization/71437
30383 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30385 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30387 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30388 (dom_opt_dom_walker::thread_across_edge): Remove
30389 handle_dominating_asserts argument. All callers changed.
30390 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30391 changes. Remove calls to lhs_of_dominating_assert. Other
30392 uses of handle_dominating_asserts turn into unconditional code
30393 (simplify_control_stmt_condition_1): Likewise.
30394 (simplify_control_stmt_condition): Likewise.
30395 (thread_through_normal_block, thread_across_edge): Likewise.
30396 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30397 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30398 object if it is not an SSA_NAME.
30399 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30400 before calling into the VRP specific simplifiers.
30401 (identify_jump_threads): Remove handle_dominating_asserts
30404 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30407 * tree-diagnostic.c (default_tree_printer): No longer static.
30408 * tree-diagnostic.h (default_tree_printer): New prototype.
30410 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30412 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30413 Change ins into fmov.
30415 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30417 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30418 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30419 Use h_con constraint for operand 1.
30420 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30421 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30423 2017-03-15 Jeff Law <law@redhat.com>
30425 PR tree-optimization/71437
30426 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30427 (record_temporary_equivalences): Use it.
30429 PR tree-optimization/71437
30430 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30431 tree-ssa-scopedtables.
30432 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30433 (record_conditions, record_cond, vuse_eq): Likewise.
30434 (record_edge_info): Adjust to API tweak of record_conditions.
30435 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30436 (record_temporary_equivalences, optimize_stmt): Likewise.
30437 (eliminate_redundant_computations): Likewise.
30438 (record_equivalences_from_stmt): Likewise.
30439 * tree-ssa-scopedtables.c: Include options.h and params.h.
30440 (vuse_eq): New function, moved from tree-ssa-dom.c
30441 (build_and_record_new_cond): Likewise.
30442 (record_conditions): Likewise. Accept vector of conditions rather
30443 than edge_equivalence structure for first argument.
30444 for the first argument.
30445 (avail_exprs_stack::lookup_avail_expr): New member function, moved
30446 from tree-ssa-dom.c.
30447 (avail_exprs_stack::record_cond): Likewise.
30448 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
30449 from tree-ssa-dom.c.
30450 (avail_exprs_stack): Add new member functions lookup_avail_expr
30452 (record_conditions): Declare.
30454 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
30457 * lra-constraints.c (process_alt_operands): Increase reject for
30458 reloading an input/output operand.
30460 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
30463 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
30464 insns to convert from signed/unsigned char/short to IEEE 128-bit
30466 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
30468 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
30471 * config/i386/i386.c (ix86_vector_duplicate_value): Create
30472 subreg of inner mode for values already in registers.
30474 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
30476 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
30477 iteration reg is used after the loop.
30479 2017-03-14 Martin Sebor <msebor@redhat.com>
30481 PR tree-optimization/79800
30482 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
30483 precision in negative-positive range.
30484 (format_floating): Call non-const overload with adjusted precision.
30486 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30489 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
30492 2017-03-14 Martin Sebor <msebor@redhat.com>
30494 PR middle-end/80020
30495 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
30496 * builtins.def (aligned_alloc): Use it.
30499 * Makefile.in (GTFILES): Add calls.c.
30500 * calls.c: Include "gt-calls.h".
30502 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
30504 PR rtl-optimization/79728
30505 * regs.h (struct target_regs): New field
30506 x_contains_allocatable_regs_of_mode.
30507 (contains_allocatable_regs_of_mode): New macro.
30508 * reginfo.c (init_reg_sets_1): Initialize it, and change
30509 contains_reg_of_mode so it includes global regs as well.
30510 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
30511 rather than contains_regs_of_mode.
30513 2017-03-14 Martin Liska <mliska@suse.cz>
30515 * doc/invoke.texi: Document options that can't be combined with
30516 -fcheck-pointer-bounds.
30518 2017-03-14 Martin Liska <mliska@suse.cz>
30520 PR middle-end/79831
30521 * doc/invoke.texi (-Wchkp): Document the option.
30523 2017-03-14 Martin Liska <mliska@suse.cz>
30525 * Makefile.in: Install gcov-dump.
30527 2017-03-14 Martin Liska <mliska@suse.cz>
30529 * multiple_target.c (expand_target_clones): Bail out for
30530 an invalid attribute.
30532 2017-03-14 Richard Biener <rguenther@suse.de>
30534 * alias.c (struct alias_set_entry): Pack properly.
30535 * cfgloop.h (struct loop): Likewise.
30536 * cse.c (struct set): Likewise.
30537 * ipa-utils.c (struct searchc_env): Likewise.
30538 * loop-invariant.c (struct invariant): Likewise.
30539 * lra-remat.c (struct cand): Likewise.
30540 * recog.c (struct change_t): Likewise.
30541 * rtl.h (struct address_info): Likewise.
30542 * symbol-summary.h (function_summary): Likewise.
30543 * tree-loop-distribution.c (struct partition): Likewise.
30544 * tree-object-size.c (struct object_size_info): Likewise.
30545 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
30546 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
30547 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
30548 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
30549 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
30550 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
30551 (struct _stmt_vec_info): Likewise.
30553 2017-03-14 Martin Liska <mliska@suse.cz>
30556 * multiple_target.c (create_dispatcher_calls): Check that
30557 a target can create a function dispatcher.
30559 2017-03-14 Martin Liska <mliska@suse.cz>
30562 * multiple_target.c (expand_target_clones): Drop local.local
30563 flag for default implementation.
30565 2017-03-14 Richard Biener <rguenther@suse.de>
30567 PR tree-optimization/80030
30568 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
30570 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
30572 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30573 gcc_fallthrough() instead of __attribute__((fallthrough));
30575 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30577 * doc/gcc.texi: Remove "up" link to (DIR).
30578 * doc/gccint.texi: Ditto.
30580 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30582 * doc/install.texi (Specific) <avr>: Remove reference to
30585 2017-03-13 Jeff Law <law@redhat.com>
30587 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30588 attribute rather than comments.
30590 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30591 match_scratch operand is highest.
30593 2017-03-13 Martin Liska <mliska@suse.cz>
30595 PR middle-end/78339
30596 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30597 is a CHKP clone, use original declaration.
30599 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30601 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30602 (arc_conditional_register_usage): Use a different allocation order
30603 when optimizing for size.
30604 * common/config/arc/arc-common.c (arc_option_optimization_table):
30605 Section anchors default on when optimizing for size.
30607 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30609 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30611 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30613 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30614 * config/arc/arc.md (cpu_facility): Add cd variant.
30615 (*movqi_insn): Add code density variant.
30616 (*movhi_insn): Likewise.
30617 (*movqi_insn): Likewise.
30618 (*addsi3_mixed): Likewise.
30619 (subsi3_insn): Likewise.
30621 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30623 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30625 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30627 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30628 expressions with MINUS and UNARY ops.
30630 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30633 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30635 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30636 between vec_select and vector argument.
30637 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30638 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30639 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30640 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30641 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30642 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30644 2017-03-13 Richard Biener <rguenther@suse.de>
30647 * params.def (vect-max-peeling-for-alignment): Fix typo.
30649 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30651 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30652 issue that only occurred with binutils below 2.18.
30654 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30656 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30657 refer to binutils 2.11/2.12 minimum.
30659 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30661 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30662 ftp.kernel.org and simplify binutils requirement.
30664 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30666 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30668 (Optimize Options): Ditto. Also remove redundancy.
30670 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30672 PR translation/79848
30673 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30675 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30676 to G_ to avoid double translation.
30678 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30680 PR translation/79923
30681 * auto-profile.c (get_combined_location): Convert leading
30682 character of diagnostics to lower case and remove trailing period.
30683 (read_profile): Likewise for various diagnostics.
30684 * config/arm/arm.c (arm_option_override): Remove trailing period
30685 from various diagnostics.
30686 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30687 (msp430_expand_delay_cycles): Likewise.
30689 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30692 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30693 full command-line argument, rather than just "str".
30694 (aarch64_validate_march): Likewise.
30695 (aarch64_validate_mtune): Likewise.
30697 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30699 PR rtl-optimization/78911
30700 * lra-assigns.c (must_not_spill_p): New function.
30701 (spill_for): Use it.
30703 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30705 PR tree-optimization/79981
30706 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30707 ATOMIC_COMPARE_EXCHANGE ifn result.
30708 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30709 IFN_ATOMIC_COMPARE_EXCHANGE.
30711 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30714 * opts.c (parse_sanitizer_options): Add missing question mark to
30715 "did you mean" message.
30717 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30719 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30721 (VMULEUH_UNS): Likewise.
30722 (VMULOUB_UNS): Likewise.
30723 (VMULOUH_UNS): Likewise.
30724 * config/rs6000/rs6000.c (builtin_function_type): Remove
30725 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30727 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30730 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30731 x with result of extra_parsing_for_operand_code_0.
30732 (function_reader::extra_parsing_for_operand_code_0): Convert
30733 return type from void to rtx, returning x. When reading
30734 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30735 larger size containing struct block_symbol.
30737 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30739 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30740 -mfloat128-hardware without -m64.
30742 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30745 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30746 entries to the case statement that marks unsigned arguments to
30747 overloaded functions.
30749 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30751 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30752 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30754 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30757 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30758 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30760 2017-03-10 Martin Liska <mliska@suse.cz>
30764 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30765 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30768 2017-03-10 Olivier Hainque <hainque@adacore.com>
30770 * tree-switch-conversion (array_value_type): Start by resetting
30771 candidate type to it's main variant.
30773 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30775 PR rtl-optimization/79909
30776 * combine.c (try_combine): Use simplify_replace_rtx on individual
30777 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30778 of the whole CALL_INSN_FUNCTION_USAGE.
30780 PR tree-optimization/79972
30781 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30782 get_range_info on SSA_NAMEs. Formatting fixes.
30784 2017-03-10 Richard Biener <rguenther@suse.de>
30785 Jakub Jelinek <jakub@redhat.com>
30787 PR tree-optimization/77975
30788 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30789 edge to be constant.
30790 (get_val_for): For constant x return it. Formatting fix.
30791 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30792 would use the same bases as the current one.
30794 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30796 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30797 instead of vec_select for V1TImode.
30798 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30800 (VSX_LE_128): Add V1TI to this mode iterator.
30801 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30802 (*vsx_le_perm_store_<mode>): Likewise.
30803 (pre-reload splitter for VSX stores): Likewise.
30804 (post-reload splitter for VSX stores): Likewise.
30805 (*vsx_xxpermdi2_le_<mode>): Likewise.
30806 (*vsx_lxvd2x2_le_<mode>): Likewise.
30807 (*vsx_stxvd2x2_le_<mode>): Likewise.
30809 2017-03-09 Michael Eager <eager@eagercon.com>
30811 Correct failures with --enable-checking=yes,rtl.
30813 * config/microblaze/microblaze.c (microblaze_expand_shift):
30814 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30815 test for const0_rtx.
30816 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30817 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30819 2017-03-09 Richard Biener <rguenther@suse.de>
30821 PR tree-optimization/79977
30822 * graphite-scop-detection.c (scop_detection::merge_sese):
30823 Handle the case of extra exits to blocks dominating the entry.
30825 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30827 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30830 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30832 PR rtl-optimization/79949
30833 * lra-constraints.c (process_alt_operands): Check memory when
30834 trying to predict a cycle. Print about the overall increase.
30836 2017-03-09 Richard Biener <rguenther@suse.de>
30838 PR middle-end/79971
30839 * gimple-expr.c (useless_type_conversion_p): Preserve
30840 TYPE_SATURATING for fixed-point types.
30842 2017-03-09 Richard Biener <rguenther@suse.de>
30845 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30846 alignment of BLKmode params.
30848 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30851 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30852 (VALL_NO_V2Q): Likewise.
30854 * config/aarch64/aarch64-simd.md
30855 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30857 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30858 VALL_NO_V2Q mode iterator.
30859 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30861 2017-03-09 Martin Liska <mliska@suse.cz>
30863 PR tree-optimization/79631
30864 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30865 tree_int_cst_sign_bit just for INTEGER constants.
30867 2017-03-09 Martin Liska <mliska@suse.cz>
30871 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30874 2017-03-09 Marek Polacek <polacek@redhat.com>
30877 * tree.c (inchash::add_expr): Handle TREE_VEC.
30879 2017-03-09 Martin Liska <mliska@suse.cz>
30882 (chkp_narrow_size_and_offset): New function.
30883 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30884 (void chkp_parse_bit_field_ref): New function.
30885 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30886 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30888 2017-03-09 Martin Liska <mliska@suse.cz>
30891 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30892 (chkp_find_bounds_1): Remove gcc_unreachable.
30894 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30897 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30898 BUILT_IN_SYNC*, determine the access type from the size suffix and
30899 always build a MEM_REF with that type. Handle forgotten
30900 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30903 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30904 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30905 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30906 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30907 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30908 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30909 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30910 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30911 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30912 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30913 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30914 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30915 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30916 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30917 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30918 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30919 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30920 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30921 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30922 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30923 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30924 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30925 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30926 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30927 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30928 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30929 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30930 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30931 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30932 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30933 definitions outside of __OPTIMIZE__ guarded section.
30936 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30937 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30938 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30939 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30942 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30944 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30945 ("vfenez<mode>"): Add missing constraints.
30947 2017-03-08 Martin Sebor <msebor@redhat.com>
30950 * config/nds32/nds32.c (nds32_option_override):
30951 Fix misspelled diagnostic.
30953 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30956 * gimplify.c (gimplify_omp_for): Replace index var in outer
30957 taskloop statement with an artificial variable and add
30958 OMP_CLAUSE_PRIVATE clause for it.
30960 2017-03-08 Richard Biener <rguenther@suse.de>
30962 PR tree-optimization/79955
30963 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30964 for accesses that are completely outside of the variable.
30966 2017-03-08 Andrew Haley <aph@redhat.com>
30968 PR tree-optimization/79943
30969 * tree-ssa-loop-split.c (compute_new_first_bound): When
30970 calculating the new upper bound, (END-BEG) should be added, not
30973 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30975 * config/avr/avr.md (setmemhi): Make sure match_dup
30976 operand number comes before match_scratch.
30978 2017-03-08 Richard Biener <rguenther@suse.de>
30980 PR tree-optimization/79920
30981 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30982 with ncopies == 1 to ...
30983 (vect_transform_slp_perm_load): ... here. Properly compute
30984 all element loads by iterating VF times over the group. Do
30985 not handle ncopies (computed in a broken way) in
30986 vect_create_mask_and_perm.
30988 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30991 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30992 is a uniform vector, use uniform_vector_p return value instead of
30993 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30995 2017-03-07 Marek Polacek <polacek@redhat.com>
30997 PR middle-end/79809
30998 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30999 (alloca_call_type): Likewise.
31001 2017-03-07 Martin Liska <mliska@suse.cz>
31003 * gcov.c (process_args): Put comment to correct location.
31005 2017-03-07 Martin Liska <mliska@suse.cz>
31007 PR middle-end/68270
31008 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31009 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31010 (chkp_narrow_bounds_for_field): Likewise.
31011 (chkp_parse_array_and_component_ref): Pass one more argument to
31014 2017-03-07 Richard Biener <rguenther@suse.de>
31016 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31019 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31021 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31022 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31024 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31027 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31028 to end of description.
31029 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31031 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31033 PR rtl-optimization/79901
31034 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31036 (*avx512f_<code><mode>3<mask_name>): ... this.
31037 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31038 iterator instead of VI8_AVX2_AVX512BW.
31040 PR rtl-optimization/79901
31041 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31042 min/max expander, expand it using expand_vec_cond_expr.
31045 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31048 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31051 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31052 to void * for PCH reasons.
31053 * dwarf2out.c (output_loc_operands, output_die): Cast
31054 v.val_vec.array to unsigned char *.
31056 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31059 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31062 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31064 PR rtl-optimization/79571
31065 * lra-constraints.c (process_alt_operands): Calculate static
31066 reject and subtract it from overall when only addresses will be
31069 2017-03-06 Julia Koval <julia.koval@intel.com>
31072 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31073 incoming stack boundary to 128 for 64-bit targets.
31075 2017-03-06 Richard Biener <rguenther@suse.de>
31077 PR tree-optimization/79894
31078 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31079 to NULL after folding it.
31081 2017-03-06 Richard Biener <rguenther@suse.de>
31083 PR tree-optimization/79824
31084 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31085 check disabling peeling for gaps.
31087 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31089 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31090 attributes): Document gettimeofday.
31092 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31094 * config/s390/s390.c (s390_option_override_internal): Set
31095 PARAM_MIN_VECT_LOOP_BOUND
31097 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31099 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31100 * config/s390/s390.md: Likewise.
31102 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31105 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31106 (<avx2_avx512>_perm<mode>): Rename to ...
31107 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31109 (<avx512>_perm<mode>_mask): Rename to ...
31110 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31112 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31113 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31114 instead of VI8F_256_512.
31115 (avx512f_perm<mode>): New define_expand.
31116 (avx512f_perm<mode>_mask): Likewise.
31117 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31118 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31120 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31122 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31123 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31125 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31127 2017-03-06 Martin Liska <mliska@suse.cz>
31130 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31131 when having a SSA NAME w/o VAR_DECL assigned to it.
31133 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31135 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
31136 msa_dpsub_<su>_d): Fix MODE for vec_select.
31138 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31140 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
31142 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
31144 2017-03-06 Richard Biener <rguenther@suse.de>
31146 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
31147 * plugin.c (register_plugin_info): Likewise.
31148 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
31150 2017-03-05 Jakub Jelinek <jakub@redhat.com>
31152 * config/i386/sse.md (sse_storehps, sse_storelps,
31153 avx_<castmode><avxsizesuffix>_<castmode>,
31154 avx512f_<castmode><avxsizesuffix>_<castmode>,
31155 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
31156 in condition that at least one operand is not a MEM.
31158 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31160 PR middle-end/79805
31161 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
31162 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
31164 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
31165 gimple_call_nothrow_p flag based on whether original builtin can throw.
31166 If it can, emit following stmts on the fallthrough edge.
31167 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
31168 don't create new bb if inserting just debug stmts on the edge, try to
31169 insert them on the fallthru bb or just reset debug stmts.
31171 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
31174 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
31175 restore recog_data (including the operand rtxes inside it) around
31176 the call to get_insn_template.
31178 2017-03-03 Martin Sebor <msebor@redhat.com>
31180 PR tree-optimization/79699
31181 * context.c (context::~context): Free MPFR caches to avoid
31182 a memory leak on program exit.
31184 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31186 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
31187 Use wide_int::ulow () instead of .elt (0).
31189 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31191 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
31192 (*pushxf): Limit oF constraint to 32bit targets and add oC
31193 constraint for 64bit targets.
31194 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
31195 (*pushdf): Change rmF constraint to rmC.
31197 2017-03-03 Martin Liska <mliska@suse.cz>
31199 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
31200 Remove unused variable.
31202 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31205 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
31206 is a memory operand, increase num_memory.
31207 (ix86_expand_args_builtin): Likewise.
31209 2017-03-03 Jan Hubicka <jh@suse.cz>
31212 * ipa-devirt.c (maybe_record_node): Properly handle
31213 __cxa_pure_virtual visibility.
31215 2017-03-03 Martin Liska <mliska@suse.cz>
31217 PR tree-optimization/79803
31218 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
31220 (pass_loop_prefetch::execute): Disabled optimization if an
31221 assumption about L1 cache size is not met.
31223 2017-03-03 Martin Liska <mliska@suse.cz>
31225 PR rtl-optimization/79574
31226 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
31227 (hash_scan_set): Likewise.
31228 (dump_hash_table): Likewise.
31229 (hoist_code): Likewise.
31231 2017-03-03 Richard Biener <rguenther@suse.de>
31233 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31234 in place of uhigh ().
31235 (fixed_convert_from_real): Likewise.
31237 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31240 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31242 2017-03-03 Richard Biener <rguenther@suse.de>
31244 PR middle-end/79818
31245 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31246 TYPE_OVERFLOW_UNDEFINED check.
31248 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31250 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31252 (vector_ae_<mode>_p): Likewise.
31253 (vector_nez_<mode>_p): Likewise.
31254 (vector_ne_v2di_p): Likewise.
31255 (vector_ae_v2di_p): Likewise.
31256 (vector_ne_<mode>_p): Likewise.
31257 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31259 (vsx_tsqrt<mode>2_fe): Likewise.
31261 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31264 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31266 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31268 PR rtl-optimization/79780
31269 * cprop.c (one_cprop_pass): When second and further conditional trap
31270 in a single basic block is turned into an unconditional trap, turn it
31271 into a deleted note to avoid RTL verification failures.
31273 2017-03-02 Richard Biener <rguenther@suse.de>
31275 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31277 2017-03-02 Richard Biener <rguenther@suse.de>
31279 PR tree-optimization/79345
31281 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31282 param and abort the walk, returning -1 if it is hit.
31283 (walk_aliased_vdefs): Take a limit param and pass it on.
31284 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31285 defaulting to 0 and return a signed int.
31286 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31287 (check_defs): New helper.
31288 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31289 about uninitialized memory.
31290 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31291 bogus uninitialized warning.
31292 (fixed_convert_from_real): Likewise.
31294 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31296 PR tree-optimization/66768
31297 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31298 iv_use if base object can't be determined.
31300 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31302 PR tree-optimization/79345
31303 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31304 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31305 (get_pattern_stats): Initialize it.
31306 * genemit.c (gen_expand): Verify match_scratch numbers come after
31307 match_operand/match_dup numbers.
31308 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31309 match_scratch numbers.
31310 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31312 * config/s390/s390.md (trunctdsd2): Likewise.
31314 2017-03-02 Richard Biener <rguenther@suse.de>
31316 * wide-int.h (wide_int_storage::operator=): Implement in terms
31319 2017-03-02 Richard Biener <rguenther@suse.de>
31321 PR tree-optimization/79777
31322 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31323 the to insert expression to sth existing.
31325 2017-03-01 Martin Sebor <msebor@redhat.com>
31327 PR middle-end/79692
31328 * gimple-ssa-sprintf.c
31329 (directive::known_width_and_precision): New function.
31330 (format_integer): Use it.
31331 (get_mpfr_format_length): Consider the full range of precision
31332 when computing %g output with the # flag. Set the likely byte
31333 count to 3 rather than 1 when precision is indeterminate.
31334 (format_floating): Correct the lower bound of precision.
31336 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31338 * doc/invoke.texi: Document default code model for 64-bit Linux.
31340 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31343 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31344 udiv rather than div since input pattern is unsigned.
31346 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31348 * config/i386/i386.c (print_reg): Warn for values of
31349 unsupported size in integer register.
31351 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31354 * config/rs6000/predicates.md (current_file_function_operand): Do
31355 not allow self calls to be local if the function is replaceable.
31357 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31360 * config/rs6000/altivec.h (vec_ctz and others): Change the
31361 preprocessor macro that controls conditional compilation from
31362 _ARCH_PWR9 to __POWER9_VECTOR__.
31363 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31364 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31365 control (instead of _ARCH_PWR9 control) so that template
31366 definition uses power9-specific function.
31367 (vec_any_eq): Likewise.
31368 (vec_all_ne): Change macro definition to use a power9-specific
31369 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31370 _ARCH_PWR9 control).
31371 (vec_any_eq) Likewise.
31372 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31373 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31374 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31375 support for xvcmpnedp instruction.
31376 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31377 macro expansion so that Power9 implementation of vec_all_ne does
31378 not use the AltiVec predicate framework.
31379 (VCMPNEH_P): Likewise.
31380 (VCMPNEW_P): Likewise.
31381 (VCMPNED_P): Likewise.
31382 (VCMPNEFP_P): Likewise.
31383 (VCMPNEDP_P): Likewise.
31384 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31385 implementation of vec_any_eq to not use AltiVec predicate
31387 (VCMPAEH_P): Likewise.
31388 (VCMPAEW_P): Likewise.
31389 (VCMPAED_P): Likewise.
31390 (VCMPAEFP_P): Likewise.
31391 (VCMPAEDP_P): Likewise.
31392 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31393 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31394 not use the AltiVec predicate framework.
31395 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31396 of vec_any_eq to not use AltiVec predicate framework.
31397 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31398 support for predefined __POWER9_VECTOR__ macro to indicate that
31399 Power9 instruction selection is enabled.
31400 (altivec_overloaded_builtins): Remove extraneous
31401 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31402 function argument types RS6000_BTI_bool_V16QI and
31403 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31404 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31405 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31406 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31407 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31408 Power9 for implementations of vec_cmpne. Change the signature for
31409 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31410 (representing vec_all_ne) to remove the previously described first
31411 argument of type RS6000_BTI_INTSI, as this was an artifact of
31412 reliance on the AltiVec predicate framework, which is no longer
31413 used in the implementation of these functions. Add
31414 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31415 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31416 since, unlike the AltiVec predicate framework implementation, we
31417 do not share function descriptors between vec_alle and vec_anyeq.
31418 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31419 set of modes that receive special treatment even when
31420 TARGET_P9_VECTOR is true. The special treatment emits code that
31421 does not depend on Power9 instructions.
31422 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31423 define_expand to not rely on AltiVec predicate framework.
31424 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31426 (vector_ne_v2di_p): Change this define_expand to not rely on
31427 AltiVec predicate framework.
31428 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31430 (vector_ne_<mode>_p): Change this define_expand to not rely on
31431 AltiVec predicate framework.
31432 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31434 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31435 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31436 define_insn pattern.
31437 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
31438 define_insn pattern because the xvcmpne<VSs>. instruction is not
31440 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
31441 instruction is not supported.
31443 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31445 * config/nvptx/nvptx.c: Include intl.h.
31447 2017-03-01 Martin Jambor <mjambor@suse.cz>
31450 * ipa-prop.h (ipa_bits): Removed field known.
31451 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
31452 to pointers. Adjusted their comments to warn about their sharing.
31453 (ipcp_transformation_summary): Change bits to a vector of pointers.
31454 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
31455 (ipa_get_ipa_bits_for_value): Declare.
31456 * tree-vrp.h (value_range): Mark as GTY((for_user)).
31457 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
31458 (ipa_bits_hash_table): Likewise.
31459 (ipa_vr_ggc_hash_traits): Likewise.
31460 (ipa_vr_hash_table): Likewise.
31461 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
31462 being pointers and vr_known being removed.
31463 (ipa_set_jf_unknown): Likewise.
31464 (ipa_get_ipa_bits_for_value): New function.
31465 (ipa_set_jfunc_bits): Likewise.
31466 (ipa_get_value_range): New overloaded functions.
31467 (ipa_set_jfunc_vr): Likewise.
31468 (ipa_compute_jump_functions_for_edge): Use the above functions to
31469 construct bits and vr parts of jump functions.
31470 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
31471 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31473 (ipcp_grow_transformations_if_necessary): Also allocate
31474 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31476 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
31477 them. Fix too long lines.
31478 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
31479 vr_known being removed.
31480 (ipa_read_jump_function): Use new setter functions to construct bits
31481 and vr parts of jump functions or set them to NULL.
31482 (write_ipcp_transformation_info): Adjust for bits being pointers.
31483 (read_ipcp_transformation_info): Likewise.
31484 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
31486 Include gt-ipa-prop.h.
31487 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
31489 (ipcp_store_bits_results): Likewise.
31490 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
31491 Do not write to existing jump functions but use a temporary instead.
31493 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31496 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
31497 attempt to use its first operand as BIT_FIELD_REF base.
31499 2017-03-01 Richard Biener <rguenther@suse.de>
31501 PR middle-end/79721
31502 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
31503 interpolating formula in wrapping arithmetic.
31504 (chrec_apply): Convert chrec_evaluate return value to wanted type.
31506 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31508 PR tree-optimization/79734
31509 * tree-vect-generic.c (expand_vector_condition): Optimize
31510 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
31511 Handle VEC_COND_EXPR where comparison has different inner width from
31512 type's inner width.
31514 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
31516 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
31517 markup, and similar issues. Remove @opindex entries for things
31518 that aren't options. Add missing -mmpy-option entries.
31520 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31522 PR tree-optimization/79737
31523 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
31524 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
31525 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
31526 instead of byte_size. Formatting fix.
31527 (shift_bytes_in_array_right): Formatting fix.
31529 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
31532 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
31533 condition on optimize for the leaf function test.
31535 2017-02-28 Martin Liska <mliska@suse.cz>
31538 * read-rtl-function.c (function_reader::handle_unknown_directive):
31539 Bail out when one uses -flto.
31541 2017-02-28 Martin Liska <mliska@suse.cz>
31543 * common.opt: Replace space with tabular for options of <number>
31545 * config/i386/i386.opt: Show <number> value for
31546 -mlarge-data-threshold.
31547 * opts.c (print_filtered_help): Do not display number in hexadecimal
31550 2017-02-28 Martin Liska <mliska@suse.cz>
31552 * common.opt: Fix --help=option -Q for options which are of
31555 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
31557 * config/i386/i386.c (print_reg): Error out for values
31558 of 8-bit size in invalid integer register.
31560 2017-02-28 Martin Sebor <msebor@redhat.com>
31562 PR tree-optimization/79691
31563 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
31565 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31568 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
31569 gcc_unreachable with output_operand_lossage.
31571 2017-02-28 Richard Biener <rguenther@suse.de>
31573 PR tree-optimization/79740
31574 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31576 (visit_nary_op): Insert the nary into the hashtable if we
31577 pattern-matched sth.
31578 * tree-ssa-pre.c (eliminate_insert): Robustify.
31580 2017-02-28 Richard Biener <rguenther@suse.de>
31582 PR middle-end/79731
31583 * fold-const.c (decode_field_reference): Reject out-of-bound
31586 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31588 * config/i386/i386.c: Include intl.h.
31589 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31590 instead of just cond ? "..." : "...".
31591 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31592 * coverage.c (read_counts_file): Likewise.
31593 * omp-offload.c: Include intl.h.
31594 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31595 of just cond ? "..." : "...".
31596 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31597 of just cond ? "..." : "...".
31599 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31602 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31604 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31605 'tune for' CPU name.
31606 * config/arm/arm-cpu-data.h: Regenerated.
31608 2017-02-28 Richard Biener <rguenther@suse.de>
31610 PR tree-optimization/79732
31611 * tree-inline.c (expand_call_inline): Do not shadow var.
31613 2017-02-28 Richard Biener <rguenther@suse.de>
31615 PR tree-optimization/79723
31616 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31617 address-space properly.
31619 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31621 * doc/optinfo.texi (Optimization groups): Fix option used for
31623 * doc/invoke.texi (-fopt-info): Document "omp".
31624 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31625 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31626 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31627 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31628 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31630 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31632 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31633 OPTGROUP_OMP with "omp".
31635 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31638 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31639 for arithmetic shift of unsigned V2DI.
31641 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31643 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31644 arc/linux.h headers.
31645 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31646 (LINK_SPEC): Likewise.
31647 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31648 (EXTRA_SPECS): Likewise.
31649 (STARTFILE_SPEC): Likewise.
31650 (ENDFILE_SPEC): Likewise.
31651 (LIB_SPEC): Likewise.
31652 (TARGET_SDATA_DEFAULT): Likewise.
31653 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31654 (MULTILIB_DEFAULTS): Likewise.
31655 (DWARF2_UNWIND_INFO): Likewise.
31656 * config/arc/big.h: New file.
31657 * config/arc/elf.h: Likewise.
31658 * config/arc/linux.h: Likewise.
31659 * config/arc/t-uClibc: Remove.
31661 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31663 PR tree-optimization/77536
31664 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31665 (tree_transform_and_unroll_loop): Use above function to compute the
31666 estimated niter of unrolled loop and use it when scaling profile.
31667 Also use count info rather than frequency if it's non-zero.
31668 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31669 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31670 (vect_transform_loop): Call above function.
31672 2017-02-27 Richard Biener <rguenther@suse.de>
31674 PR tree-optimization/45397
31675 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31676 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31677 (visit_nary_op): Add pattern matching for CSEing sign-changed
31678 or truncated operations with wider ones.
31680 2017-02-27 Richard Biener <rguenther@suse.de>
31682 PR tree-optimization/79690
31683 * tree-vect-stmts.c (vectorizable_store): Use vector type
31684 built from the DR with address-space.
31686 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31688 * doc/invoke.texi (Optimize Options): Refine the description
31689 of asan-use-after-return.
31691 2017-02-25 Alan Modra <amodra@gmail.com>
31693 PR rtl-optimization/79584
31694 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31695 base, not ad->base_term, the reg within base. Remove assertion
31696 that ad->base == ad->base_term. Replace gen_int_mode using
31697 bogus mode with const0_rtx.
31699 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31701 PR middle-end/79396
31702 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31703 FMA_EXPR like tcc_binary or tcc_unary.
31705 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31708 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31710 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31711 (output_loc_operands): Handle DW_OP_call_ref and
31712 DW_OP_GNU_variable_value.
31713 (struct variable_value_struct): New type.
31714 (struct variable_value_hasher): Likewise.
31715 (variable_value_hash): New variable.
31716 (string_types): Remove.
31717 (copy_loc_descr): New function.
31718 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31719 (prepend_loc_descr_to_each): New function.
31720 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31721 instead of add_loc_descr_to_each if the first argument is single
31722 location list and the second has multiple.
31723 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31724 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31725 when looking for variable value which doesn't have other location info.
31726 (loc_list_from_tree): Formatting fix.
31727 (gen_array_type_die): Simplify DW_AT_string_length handling.
31728 (adjust_string_types): Remove.
31729 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31730 string_types. Call resolve_variable_values.
31731 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31732 (resolve_addr_in_expr): Likewise. Add A argument.
31733 (copy_deref_exprloc): Remove deref argument. Adjust for the
31734 original expression being DW_OP_GNU_variable_value with optionally
31735 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31736 optionally after it.
31737 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31738 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31739 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31740 (variable_value_hasher::hash, variable_value_hasher::equal): New
31742 (resolve_variable_value_in_expr, resolve_variable_value,
31743 resolve_variable_values, note_variable_value_in_expr,
31744 note_variable_value): New functions.
31745 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31748 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31751 * opts.h (handle_generated_option): Add GENERATED_P argument.
31752 * opts-common.c (handle_option): Adjust function comment.
31753 (handle_generated_option): Add GENERATED_P argument, pass it to
31755 (control_warning_option): Pass false to handle_generated_option
31757 * opts.c (maybe_default_option): Pass true to handle_generated_option
31759 * optc-gen.awk: Likewise.
31761 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31763 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31764 a REG, look at the REG it is a SUBREG of.
31765 (splitter for cmpeqsi_t): Ditto.
31767 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31769 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31770 the special USEs with the pattern of the insn, not the insn itself.
31772 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31775 * doc/invoke.texi: Document -mload-store-pairs.
31777 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31778 Sandra Loosemore <sandra@codesourcery.com>
31780 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31781 argument isn't a CONST_INT.
31782 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31783 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31784 (nios2_validate_compare): Bypass alternate compare logic if *op2
31785 is not a CONST_INT.
31786 (ldstwm_operation_p): Return false if first_base is not a REG or
31787 if first_offset is not a CONST_INT.
31789 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31791 * config/cris/cris.md: Use correct operand in a define_peephole2.
31793 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31795 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31797 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31799 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31800 this_insn if it is an INSN or JUMP_INSN.
31801 (force_offsettable): Look at base, not at addr.
31802 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31803 on things that aren't necessarily CONST_INTs.
31805 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31807 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31808 -mfpmath=sse is the default also for x86-32 targets with SSE2
31809 instruction set when @option{-ffast-math} is enabled
31811 2017-02-24 Jeff Law <law@redhat.com>
31813 PR rtl-optimizatoin/79286
31814 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31817 2017-02-24 Richard Biener <rguenther@suse.de>
31819 PR tree-optimization/79389
31820 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31823 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31825 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31826 function comment to reflect reality.
31827 (loop_exits_before_overflow): Fix typo in function description.
31829 2017-02-24 Richard Biener <rguenther@suse.de>
31831 PR tree-optimization/79389
31832 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31833 properly that a threading opportunity exists. Detect conditional
31834 copy/constant propagation opportunities.
31836 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31838 * config/visium/visium.md (type): Add trap.
31839 (b): New mode attribute.
31840 (*btst): Rename into...
31841 (*btst<mode>): ...this and adjust.
31842 (*cbranchsi4_btst_insn): Rename into...
31843 (*cbranch<mode>4_btst_insn): ...this and adjust.
31844 (trap): New define_insn.
31846 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31848 PR tree-optimization/79389
31849 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31850 (noce_reversed_cond_code): New function.
31851 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31852 reversed_comparison_code. Formatting fix.
31853 (noce_try_store_flag): Test rev_cond != NULL in addition to
31854 reversed_comparison_code.
31855 (noce_try_store_flag_constants): Likewise.
31856 (noce_try_store_flag_mask): Likewise.
31857 (noce_try_addcc): Use rev_cond if non-NULL instead of
31858 reversed_comparison_code.
31859 (noce_try_cmove_arith): Likewise. Formatting fixes.
31860 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31861 (noce_find_if_block): Initialize rev_cond.
31862 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31863 instead of false as last argument never attempt to reverse it
31866 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31868 PR tree-optimization/79663
31869 * tree-predcom.c (combine_chains): Process refs in reverse order
31870 only for ZERO length chains, and add explaining comment.
31872 2017-02-23 Jeff Law <law@redhat.com>
31874 PR tree-optimization/79578
31875 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31876 in call to operand_equal_p.
31878 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31881 * config/i386/cpuid.h: Fix another undefined behavior.
31883 2017-02-23 Richard Biener <rguenther@suse.de>
31885 PR tree-optimization/79683
31886 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31887 vector types for data-refs.
31889 2017-02-23 Martin Liska <mliska@suse.cz>
31891 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31893 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31895 PR middle-end/79665
31896 * internal-fn.c (get_range_pos_neg): Moved to ...
31897 * tree.c (get_range_pos_neg): ... here. No longer static.
31898 * tree.h (get_range_pos_neg): New prototype.
31899 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31900 are known to be in between 0 and signed maximum inclusive, try to
31901 expand both unsigned and signed divmod and use the cheaper one from
31904 2017-02-22 Jeff Law <law@redhat.com>
31906 PR tree-optimization/79578
31907 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31908 to compare base operands.
31910 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31913 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31914 gpc_reg_operand instead of fpr_reg_operand.
31916 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31918 * config/mips/mips.c (mips_return_in_memory): Force FP
31919 vector types to be returned in memory for o32 ABI.
31921 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31923 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31924 instead of DW_TAG_member for static data member declarations and don't
31925 set no_linkage_name for static inline data members.
31926 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31929 2017-02-22 Martin Liska <mliska@suse.cz>
31931 * doc/invoke.texi: Replace inequality signs with square brackets
31934 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31937 * lra-constraints.c (simplify_operand_subreg): Handle
31938 WORD_REGISTER_OPERATIONS targets.
31940 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31943 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31944 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31945 elimination by swapping fld*.
31947 2017-02-22 Richard Biener <rguenther@suse.de>
31949 PR tree-optimization/79673
31950 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31951 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31952 irrelevant address-space qualifiers and avoiding a
31953 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31955 2017-02-22 Richard Biener <rguenther@suse.de>
31957 PR tree-optimization/79666
31958 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31959 to not symbolically negate if that may introduce undefined
31962 2017-02-22 Martin Liska <mliska@suse.cz>
31965 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31966 * data-streamer-out.c (streamer_write_gcov_count_stream):
31968 * value-prof.c (stream_out_histogram_value): Make assert more
31969 precise based on type of counter.
31971 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31974 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31975 Use nonimmediate_operand instead of memory_operand for operand 1.
31976 (float-extend standard_x87sse_constant_load splitter): Ditto.
31978 2017-02-21 Jeff Law <law@redhat.com>
31980 PR tree-optimization/79621
31981 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31982 blocks with edges to themselves.
31984 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31987 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31988 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31989 Use gimple_call_builtin_p.
31992 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31993 on temporarily removed DEBUG_INSNs.
31995 PR tree-optimization/79649
31996 * tree-loop-distribution.c (classify_partition): Give up on
31997 non-generic address space loads/stores.
31999 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32001 * doc/loop.texi (Loop manipulation): Remove nonexistent
32002 tree_ssa_loop_version from the documentation.
32003 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32005 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32008 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32009 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32010 * config/rs6000/rs6000.c: Include except.h.
32011 (rs6000_expand_split_stack_prologue): Call
32012 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32014 2017-02-21 Martin Jambor <mjambor@suse.cz>
32017 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32018 have been analyzed.
32020 2017-02-21 Martin Jambor <mjambor@suse.cz>
32022 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32023 for backward compatibility only.
32024 * doc/invoke.texi (Option Summary): Remove all references to
32025 -fipa-cp-alignment.
32027 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32031 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32033 * lra-constraints.c (curr_insn_transform): Handle
32034 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32036 2017-02-21 Martin Liska <mliska@suse.cz>
32038 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32040 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32042 PR translation/79638
32043 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32045 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32048 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32049 (arm_function_ok_for_sibcall): Return false for an indirect call by
32050 descriptor if all the argument registers are used.
32051 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32052 alignment of the function.
32054 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32056 PR tree-optimization/61441
32057 * simplify-rtx.c (simplify_const_unary_operation): For
32058 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32059 the sNaN unmodified.
32061 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32063 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32064 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32065 instead of SYSTEM_HEADER_DIR.
32067 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32068 Martin Liška <mliska@suse.cz>
32070 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32071 Fix typos and grammar, use active voice, and clarify.
32073 2017-02-20 Marek Polacek <polacek@redhat.com>
32075 PR middle-end/79537
32076 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32079 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32081 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32084 * config/i386/i386.c (ix86_expand_builtin): Handle
32085 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32086 ix86_builtins_isa[fcode].isa as a requirement of those
32087 flags and any other flag in the bitmask.
32088 (ix86_init_mmx_sse_builtins): Use 0 instead of
32089 ~OPTION_MASK_ISA_64BIT as mask.
32090 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32091 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32092 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32093 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32095 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32098 * lra-constraints.c (split_reg): Check requested split mode
32099 is supported by the register.
32101 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32103 * lra-constraints.c (simplify_operand_subreg): Remove early
32106 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32109 * lra-constraints.c (curr_insn_transform): Tighten condition
32110 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32112 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32115 * lra-constraints.c (curr_insn_transform): Handle
32116 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32118 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32121 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32123 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32125 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32128 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32131 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32133 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
32136 2017-02-18 Jakub Jelinek <jakub@redhat.com>
32138 * final.c (last_columnnum, override_columnnum): New variables.
32139 (final_start_function): Set last_columnnum, pass it to begin_prologue
32140 hook and pass 0 to dwarf2out_begin_prologue.
32141 (final_scan_insn): Update override_columnnum. Pass last_columnnum
32142 to source_line debug hook.
32143 (notice_source_line): Compute last_columnnum and for debug_column_info
32144 return true on column changes.
32145 * debug.h (struct gcc_debug_hooks): Add column argument to
32146 source_line and begin_prologue hooks.
32147 (debug_nothing_int_charstar_int_bool): Remove prototype.
32148 (debug_nothing_int_int_charstar,
32149 debug_nothing_int_int_charstar_int_bool): New prototypes.
32150 (dwarf2out_begin_prologue): Add column argument.
32151 * debug.c (do_nothing_debug_hooks): Adjust source_line and
32152 begin_prologue hooks.
32153 (debug_nothing_int_charstar_int_bool): Remove.
32154 (debug_nothing_int_int_charstar,
32155 debug_nothing_int_int_charstar_int_bool): New functions.
32156 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
32157 through to dwarf2out_source_line.
32158 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
32159 (dwarf2out_source_line): Add column argument, emit it if requested.
32160 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
32162 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32163 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32164 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
32165 through to dwarf2out_begin_prologue.
32166 (vmsdbgout_source_line): Add column argument, pass it through to
32167 dwarf2out_source_line.
32168 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
32169 dbxout_source_line caller.
32170 (dbxout_source_line): Add column argument.
32172 * common.opt (gno-column-info, gcolumn-info): New options.
32173 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
32174 (check_die): Also test for multiple DW_AT_decl_column attributes.
32175 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
32176 DW_AT_decl_column if requested.
32177 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
32179 (gen_variable_die): Likewise.
32180 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
32181 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
32184 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
32185 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
32186 (ix86_handle_option): Handle OPT_m3dnowa.
32187 * doc/invoke.texi (-m3dnowa): Document.
32188 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
32189 -m3dnowa instead of -m3dnow -march=athlon.
32192 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
32193 instead of gcc_assert for K, r and R code checks. Formatting fixes.
32195 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32198 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
32199 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
32200 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
32201 generator for vsx_xxpermdi_<mode>_be.
32202 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
32203 force big-endian semantics.
32204 (vsx_xxpermdi_<mode>_be): New define_expand with same
32205 implementation as previous version of vsx_xxpermdi_<mode>.
32207 2017-02-17 Jakub Jelinek <jakub@redhat.com>
32209 PR tree-optimization/79327
32210 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
32211 variable, its initialization and use.
32213 2017-02-17 Julia Koval <julia.koval@intel.com>
32215 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
32216 (OPTION_MASK_ISA_PKU_UNSET): New.
32217 (ix86_handle_option): Handle -mrdpid.
32218 * config/i386/cpuid.h (bit_RDPID): New.
32219 * config/i386/driver-i386.c (host_detect_local_cpu):
32220 Detect RDPID feature.
32221 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
32222 * config/i386/i386-c.c (ix86_target_macros_internal):
32224 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
32225 (ix86_valid_target_attribute_inner_p): Add "rdpid".
32226 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
32227 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
32228 * config/i386/i386.md (define_insn "rdpid"): New.
32229 * config/i386/i386.opt Add -mrdpid.
32230 * config/i386/immintrin.h (_rdpid_u32): New.
32232 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32234 PR rtl-optimization/79541
32235 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32236 instead of transforming it into USE.
32238 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32240 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32241 If HONOR_SNANS (SFmode) force the input to a register.
32242 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32243 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32244 an frsp or similar insn.
32246 2017-02-17 Martin Liska <mliska@suse.cz>
32248 PR rtl-optimization/79577
32249 * params.def (selsched-max-sched-times): Increase minimum to 1.
32251 2017-02-17 Martin Liska <mliska@suse.cz>
32253 PR rtl-optimization/79574
32254 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32256 2017-02-17 Martin Liska <mliska@suse.cz>
32258 PR tree-optimization/79529
32259 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32260 ssa_defined_default_def_p to handle cases which are implicitly
32262 * tree-ssa.c (ssa_defined_default_def_p): New function.
32263 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32264 which are implicitly defined.
32265 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32267 2017-02-17 Richard Biener <rguenther@suse.de>
32269 PR middle-end/79576
32270 * params.def (max-ssa-name-query-depth): Limit to 10.
32272 2017-02-17 Richard Biener <rguenther@suse.de>
32274 PR tree-optimization/79552
32275 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32278 2017-02-17 Richard Biener <rguenther@suse.de>
32281 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32283 2017-02-17 Marek Polacek <polacek@redhat.com>
32285 PR middle-end/79536
32286 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32287 (fold_negate_expr): New wrapper.
32289 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32291 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32292 Correct terminology and de-emphasize pre-standard behavior.
32294 2017-02-16 Alan Modra <amodra@gmail.com>
32296 PR rtl-optimization/79286
32297 * ira.c (def_dominates_uses): New function.
32298 (update_equiv_regs): Don't create an equivalence for insns that
32299 may trap where the register def does not dominate the use.
32301 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32303 PR rtl-optimization/78127
32304 * lra.c (lra): Call lra_eliminate before finish the loop after
32307 2017-02-16 Richard Biener <rguenther@suse.de>
32309 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32311 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32312 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32313 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32314 (isl_val_int_from_wi): New function.
32315 (extract_affine_gmp): Rename to ...
32316 (extract_affine_wi): ... this, take a widest_int.
32317 (extract_affine_int): Just wrap extract_affine_wi.
32318 (add_param_constraints): Use isl_val_int_from_wi.
32319 (add_loop_constraints): Likewise, and extract_affine_wi.
32321 2017-02-15 Jeff Law <law@redhat.com>
32323 PR middle-end/79521
32324 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32325 ira_init_register_move_cost_if_necessary.
32327 2017-02-15 Martin Sebor <msebor@redhat.com>
32329 PR middle-end/32003
32330 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32331 removed in a prior commit.
32333 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32335 PR tree-optimization/79347
32336 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32337 counters during peeling.
32339 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32341 * Makefile.in (site.exp): Remove "set ISLVER".
32343 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32346 * real.c (real_from_integer): Call real_convert even for decimal.
32348 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32351 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32353 2017-02-14 Andrew Pinski <apinski@cavium.com>
32355 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32356 cores and change the partno/implementer to be correct.
32357 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32358 the 'B" as the implementer.
32359 * config/aarch64/aarch64-tune.md: Regenerate.
32361 2017-02-14 Carl Love <cel@us.ibm.com>
32363 * config/rs6000/rs6000.c: Add case statement entry to make the
32364 xvcvuxdsp built-in argument unsigned.
32365 * config/rs6000/vsx.md: Fix the source and return operand types so they
32366 match the instruction definitions from the ISA document. Fix typo
32367 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32370 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32373 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32374 member early_clobber_alts.
32375 * lra-lives.c (reg_early_clobber_p): New.
32376 (process_bb_lives): Use it.
32377 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32378 (debug_operand_data): Initialize early_clobber_alts.
32379 (setup_operand_alternative): Set up early_clobber_alts.
32380 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32381 alternatives to new_insn_reg.
32382 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32384 (lra_update_insn_regno_info): Pass the new arg.
32386 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32388 PR middle-end/79505
32389 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32390 (new_oacc_loop_raw): Don't clear already cleared fields.
32393 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32394 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32395 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32397 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32400 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32402 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32405 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32406 the extra instruction to the right place to store 128-bit constant
32409 2017-02-14 Martin Sebor <msebor@redhat.com>
32411 PR middle-end/79448
32412 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32413 warning for strings of unknown length.
32415 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32417 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32419 2017-02-14 Jeff Law <law@redhat.com>
32422 * ira-costs.c (scan_one_insn): Initialize register move costs
32423 for pseudos seen in USE/CLOBBER insns.
32425 PR tree-optimization/79095
32426 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32427 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32428 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32429 if the operands are known to be not equal, then the resulting range
32431 (intersect_ranges): If the new range is ~[0,0] and the old range is
32432 wide, then prefer ~[0,0].
32433 * tree-vrp.c (overflow_comparison_p_1): New function.
32434 (overflow_comparison_p): New function.
32435 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32436 if NAME is used in an overflow test.
32437 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
32438 overflow check that can be expressed as an equality test, then adjust
32439 ops to be that equality test.
32441 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32443 * config/s390/s390-builtin-types.def: Remove flags argument.
32444 * config/s390/s390.c (s390_init_builtins): Likewise.
32446 2017-02-14 Martin Liska <mliska@suse.cz>
32448 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
32449 vector. Fix trailing white spaces.
32451 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
32453 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
32456 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32458 PR rtl-optimization/68664
32459 * config/arm/arm.c (arm_sched_can_speculate_insn):
32460 New function. Declare prototype.
32461 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32463 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32465 PR rtl-optimization/68664
32466 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
32468 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32470 2017-02-14 Amit Pawar <amit.pawar@amd.com>
32472 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
32473 max skip bytes for function, loop and jump.
32475 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32477 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
32478 ABS_EXPR for gimple dump.
32480 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32483 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
32485 PR tree-optimization/79408
32486 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
32487 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
32488 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
32489 also if rhs1 is INTEGER_CST.
32491 2017-02-14 Richard Biener <rguenther@suse.de>
32493 PR middle-end/79432
32494 * tree-into-ssa.c (insert_phi_nodes): When the function can
32495 have abnormal edges rewrite SSA names with broken use-def
32496 dominance out of SSA and register them for PHI insertion.
32498 2017-02-13 Martin Sebor <msebor@redhat.com>
32500 PR middle-end/79496
32501 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
32502 clearing info.nowrite flag when snprintf size argument is a range.
32504 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32506 * cprop.c (cprop_jump): Add missing space in string literal.
32507 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
32508 (get_constraint_for_component_ref): Likewise.
32509 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
32510 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
32511 * lra-constraints.c (process_alt_operands): Likewise.
32512 * ipa-inline.c (inline_small_functions): Likewise.
32513 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
32514 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
32515 * trans-mem.c (diagnose_tm_1_op): Likewise.
32516 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
32517 (grid_parallel_clauses_gridifiable): Likewise.
32519 * config/nvptx/mkoffload.c (process): Add space in between
32522 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
32523 "MOD4_SSE_REGS" and "ALL_REGS".
32525 * spellcheck.c (test_data): Add , in between "foo" and "food".
32527 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32530 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
32531 boundary crossing check and subsequent code generation agree.
32533 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32535 * config/aarch64/aarch64.c (has_memory_op): Delete.
32536 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
32539 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32541 PR rtl-optimization/79388
32542 PR rtl-optimization/79450
32543 * combine.c (distribute_notes): When removing TEM_INSN for which
32544 corresponding dest has last value recorded, invalidate that last
32547 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32549 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
32550 of explicit '@'. Add missing assembly comment marker on branch costs
32553 2017-02-13 Nathan Sidwell <nathan@acm.org>
32555 * gengtype-lex.l (<in_struct>): Add '/'.
32557 2017-02-13 Martin Liska <mliska@suse.cz>
32560 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
32562 2017-02-13 Richard Biener <rguenther@suse.de>
32564 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
32566 * configure: Re-generate.
32567 * config.in: Likewise.
32568 * graphite-dependences.c: Simplify as if
32569 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
32570 * graphite-isl-ast-to-gimple.c: Likewise.
32571 * graphite-optimize-isl.c: Likewise.
32572 * graphite-poly.c: Likewise.
32573 * graphite-sese-to-poly.c: Likewise.
32574 * graphite.h: Likewise.
32575 * toplev.c: Include isl/version.h and use isl_version () for
32576 printing the ISL version.
32577 * doc/install.texi: Update ISL requirement.
32579 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32581 * doc/standards.texi (Standards): Update reference to
32584 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32586 * doc/extend.texi (Named Address Spaces): sourceware.org now
32588 * doc/install.texi (Binaries): Ditto.
32591 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32593 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32594 standard terminology "stringize"/"stringizing" throughout.
32595 * doc/cppinternals.texi: Likewise.
32597 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32599 * doc/extend.texi: Fix some spelling mistakes and typos.
32600 * doc/invoke.texi: Likewise.
32602 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32605 * params.def (inline-min-speedup) Change from 10 to 8.
32607 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32609 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32612 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32615 * ipa-inline-analysis.c (get_minimal_bb): New function.
32616 (record_modified): Use it.
32617 (remap_edge_change_prob): Handle also ancestor functions.
32619 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32621 * doc/contrib.texi (Contributors): Remove broken link into
32622 the Mauve CVS repository.
32624 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32626 PR middle-end/79454
32627 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32628 result computation whenever lhs doesn't have vector mode, not
32629 just when it has BLKmode.
32631 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32633 * doc/makefile.texi (profiledbootstrap): Refer to the
32634 installation instructions only in textual form.
32636 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32639 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32641 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32643 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32644 (Specific): Update mingw-w64 reference.
32646 (Specific): Remove broken link to Renesas RX processor.
32648 2017-02-10 Richard Biener <rguenther@suse.de>
32650 * toplev.c (process_options): Do not mention obsolete graphite
32651 options when printing sorry message about missing graphite support.
32652 Mention -floop-nest-optimize.
32654 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32656 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32657 (vtst_p16): Likewise.
32658 (vtstq_p8): Likewise.
32659 (vtstq_p16): Likewise.
32661 (vtstq_p64): Likewise.
32662 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32663 (vset_lane_p64): New.
32664 (vsetq_lane_p64): New.
32666 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32668 PR tree-optimization/79411
32669 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32670 stmt operands are SSA_NAMEs used in abnormal phis.
32671 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32674 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32677 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32680 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32682 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32684 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32686 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32690 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32691 not arbitrary TREE_CONSTANT.
32694 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32695 "omp declare target link" attribute unless is_global_var.
32696 * omp-offload.c (find_link_var_op): Likewise.
32698 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32699 Chung-Lin Tang <cltang@codesourcery.com>
32701 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32703 (gimplify_adjust_omp_clauses): Don't delete TILE.
32704 (gimplify_omp_for): Deal with TILE.
32705 * internal-fn.c (expand_GOACC_TILE): New function.
32706 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32708 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32709 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32711 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32712 avoid DIV for outermost collapse var.
32713 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32714 Remove out of date comments, fix whitespace.
32715 * omp-general.c (omp_extract_for_data): Deal with tiling.
32716 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32717 adjust OLF_DIM_BASE value.
32718 (struct omp_for_data): Add tiling field.
32719 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32720 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32721 for auto loops. Remove default auto determining, moved to
32722 oacc_loop_fixed_partitions.
32723 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32724 stmts, add e_mask field.
32725 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32726 (oacc_thread_numbers): Use oacc_dim_call.
32727 (oacc_xform_tile): New.
32728 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32729 (finish_oacc_loop): Adjust for ifns vector.
32730 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32731 add case for GOACC_TILE fns.
32732 (oacc_loop_xform_loop): Delete.
32733 (oacc_loop_process): Iterate over call list directly, and add
32734 handling for GOACC_TILE fns.
32735 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32737 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32738 vector partitioning to outer loops. Assign 2 partitions to loops
32739 when available. Add TILE handling.
32740 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32741 (execite_oacc_device_lower): Process GOACC_TILE fns,
32742 ignore unknown specs.
32743 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32744 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32745 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32747 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32749 * configure.ac (ACX_BUGURL): Update.
32750 * configure: Regenerate.
32752 2017-02-09 Richard Biener <rguenther@suse.de>
32754 PR tree-optimization/69823
32755 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32756 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32758 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32760 * config/arc/arc-c.def: Add __NPS400__ definition.
32761 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32762 (TARGET_NPS400): Define.
32764 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32766 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32768 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32769 pointer, arch_info.
32770 (arc_cpu_types): Fill the arch_info field with a pointer into the
32771 arc_arch_types table.
32772 (arc_selected_cpu): Declare.
32773 * config/arc/arc.c (arc_selected_cpu): Make global.
32774 (arc_selected_arch): Delete.
32775 (arc_base_cpu): Delete.
32776 (arc_override_options): Remove references to deleted variables,
32777 update access to arch information.
32778 (ARC_OPT): Update access to arch information.
32779 (ARC_OPTX): Likewise.
32780 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32781 (TARGET_ARC600): Update access to arch information.
32782 (TARGET_ARC601): Likewise.
32783 (TARGET_ARC700): Likewise.
32784 (TARGET_EM): Likewise.
32785 (TARGET_HS): Likewise.
32786 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32789 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32792 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32793 condition/operands for integer GE/LE/GEU/LEU operations.
32795 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32797 PR translation/79397
32798 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32801 2017-02-08 Martin Jambor <mjambor@suse.cz>
32804 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32805 whether allocation happened.
32806 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32807 nothing was allocated.
32809 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32811 PR tree-optimization/79408
32812 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32813 constant, but SSA_NAME with a known integer range, use the minimum
32814 of that range instead of op1 to determine if modulo can be replaced
32815 with its first operand.
32817 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32819 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32821 2017-02-08 Richard Biener <rguenther@suse.de>
32823 PR tree-optimization/71824
32824 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32825 Check all loops contained in the merged region.
32827 2017-02-07 Andrew Pinski <apinski@cavium.com>
32829 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32831 2017-02-07 Andrew Pinski <apinski@cavium.com>
32833 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32834 (thunderxt88): Likewise.
32835 (thunderxt81): Disable LSE and change v8.1 to v8.
32836 (thunderxt83): Likewise.
32838 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32839 Richard Biener <rguenther@suse.de>
32841 PR middle-end/79399
32842 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32843 type from int to size_t.
32844 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32846 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32848 PR rtl-optimization/79386
32849 * cprop.c (bypass_conditional_jumps): Initialize
32850 bypass_last_basic_block already before splitting bbs after
32851 unconditional traps...
32852 (bypass_conditional_jumps): ... rather than here.
32855 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32856 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32857 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32858 fix -masm=intel patterns.
32860 2017-02-07 Richard Biener <rguenther@suse.de>
32862 PR tree-optimization/79256
32863 PR middle-end/79278
32864 * builtins.c (get_object_alignment_2): Use min_align_of_type
32865 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32866 and ADJUST_FIELD_ALIGN.
32868 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32870 * doc/tm.texi: Regenerate.
32871 * stor-layout.c (layout_decl): Adjust.
32872 (update_alignment_for_field): Likewise.
32873 (place_field): Likewise.
32874 (min_align_of_type): Likewise.
32875 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32876 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32877 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32878 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32879 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32880 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32881 * config/i386/i386.c (x86_field_alignment): Likewise.
32882 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32883 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32884 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32885 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32886 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32887 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32891 2017-01-30 Richard Biener <rguenther@suse.de>
32893 PR tree-optimization/79256
32894 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32895 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32898 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32900 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32901 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32902 builtins to SImode and emit a zero-extend, if necessary.
32904 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32906 * docs/invoke.texi (RISC-V Options): Alphabetize.
32908 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32910 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32913 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32915 * config/riscv/riscv.c: New file.
32916 * common/config/riscv/riscv-common.c: Likewise.
32917 * config.gcc: Likewise.
32918 * config/riscv/constraints.md: Likewise.
32919 * config/riscv/elf.h: Likewise.
32920 * config/riscv/generic.md: Likewise.
32921 * config/riscv/linux.h: Likewise.
32922 * config/riscv/multilib-generator: Likewise.
32923 * config/riscv/peephole.md: Likewise.
32924 * config/riscv/pic.md: Likewise.
32925 * config/riscv/predicates.md: Likewise.
32926 * config/riscv/riscv-builtins.c: Likewise.
32927 * config/riscv/riscv-c.c: Likewise.
32928 * config/riscv/riscv-ftypes.def: Likewise.
32929 * config/riscv/riscv-modes.def: Likewise.
32930 * config/riscv/riscv-opts.h: Likewise.
32931 * config/riscv/riscv-protos.h: Likewise.
32932 * config/riscv/riscv.h: Likewise.
32933 * config/riscv/riscv.md: Likewise.
32934 * config/riscv/riscv.opt: Likewise.
32935 * config/riscv/sync.md: Likewise.
32936 * config/riscv/t-elf-multilib: Likewise.
32937 * config/riscv/t-linux: Likewise.
32938 * config/riscv/t-linux-multilib: Likewise.
32939 * config/riscv/t-riscv: Likewise.
32940 * configure.ac: Likewise.
32941 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32942 Waterman as RISC-V maintainers.
32943 * doc/install.texi: Add RISC-V entries.
32944 * doc/invoke.texi: Add RISC-V options section.
32945 * doc/md.texi: Add RISC-V constraints section.
32946 * configure: Regenerated.
32948 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32951 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32952 false values to be constant vectors with all 0 or all 1 bits set.
32953 (vcondu<mode><mode>): Likewise.
32954 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32956 (fpmask_comparison_operator): Update comment.
32957 (vecint_comparison_operator): New predicate.
32958 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32959 vector conditionals when the true and false values are constant
32960 vectors with all 0 bits or all 1 bits set.
32962 2017-02-06 Martin Sebor <msebor@redhat.com>
32964 PR tree-optimization/79376
32965 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32967 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32969 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32970 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32971 to simplify split condition.
32973 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32975 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32976 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32979 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32981 PR rtl-optimization/68664
32982 * target.def (can_speculate_insn): New hook.
32983 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32984 * doc/tm.texi: Regenerate.
32985 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32986 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32987 (rs6000_sched_can_speculate_insn): New function.
32989 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32991 PR tree-optimization/79284
32992 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32993 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32994 vectorizable_mask_load_store, vectorizable_operation,
32995 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32996 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32997 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32998 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32999 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33000 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33001 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33002 is_gimple_assign (stmt). Replace another such test with
33003 is_gimple_assign (stmt).
33005 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33008 * config/avr/avr.c (rtl-iter.h): Include it.
33009 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33010 (avr_legitimate_combined_insn): ...and implementation.
33012 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33014 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33015 * config/s390/s390.c (s390_const_operand_ok)
33016 (s390_canonicalize_comparison, s390_extract_part)
33017 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33018 (s390_contiguous_bitmask_p, s390_rtx_costs)
33019 (legitimize_pic_address): Likewise.
33020 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33021 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33022 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33023 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33024 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33026 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33028 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33029 REGNO($0) == REGNO($1).
33031 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33033 * config/s390/linux.h(SIZE_TYPE): Add comment.
33035 2017-02-06 Julian Brown <julian@codesourcery.com>
33036 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33037 Virendra Pathak <virendra.pathak@broadcom.com>
33039 * config/aarch64/aarch64-cores.def: Change the scheduler
33041 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33042 * config/aarch64/thunderx2t99.md: New file.
33044 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33046 * doc/standards.texi (Go Language): Update link to language
33049 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33051 * tree-eh.c (lower_resx): Sanitize profile.
33052 (cleanup_empty_eh_move_lp): Likewise.
33054 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33057 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33059 * cfgloopmanip.h (loop_version): Update prototype.
33060 * modulo-sched.c (sms_schedule): Update call of loop_version.
33061 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33062 * tree-parloops.c (gen_parallel_loop): Likewise.
33063 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33064 * tree-ssa-loop-split.c (split_loop): Likewise.
33065 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33066 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33068 2017-02-05 Martin Liska <mliska@suse.cz>
33071 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33073 (print_operand_address): Initialize a struct to zero.
33075 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33077 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33078 garbage collector only in textual form.
33080 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33082 * doc/extend.texi (x86 specific memory model extensions for
33083 transactional memory): Simplify a phrase.
33085 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33088 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33089 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33090 (atomic_storedi_1): Likewise.
33092 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33094 PR tree-optimization/79338
33095 * tree-parloops.c (gather_scalar_reductions): Don't call
33096 vect_analyze_loop_form for loop->inner before destroying loop's
33099 2017-02-03 Martin Sebor <msebor@redhat.com>
33101 PR tree-optimization/79327
33102 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33103 when precision has resulted in leading zeros.
33104 (format_integer): Adjust the likely counter to assume an unknown
33105 argument that may be zero is non-zero.
33107 2017-02-03 Jason Merrill <jason@redhat.com>
33110 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33111 avoid copying non-taken branch.
33113 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33115 PR tree-optimization/79340
33116 * tree-vect-loop.c (vectorizable_reduction): Release
33117 vec_defs elements after safe_splicing them into other vectors.
33120 PR tree-optimization/79327
33121 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33122 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33124 (format_integer): Use wide_int_to_tree instead of build_int_cst
33125 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33126 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33127 of shortest and longest sequence.
33129 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33131 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33132 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33134 2017-02-03 Walter Lee <walt@tilera.com>
33137 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
33138 after initial stackframe link reg save.
33139 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
33141 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33144 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
33145 wu for stxssp alternative.
33147 2017-02-03 Martin Sebor <msebor@redhat.com>
33149 PR tree-optimization/79352
33150 * gimple-fold.c (get_range_strlen): Add argument.
33151 (get_range_strlen): Change return type to bool.
33152 (get_maxval_strlen): Pass in a dummy argument.
33153 * gimple-fold.h (get_range_strlen): Change return type to bool.
33154 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
33155 * tree.h (array_at_struct_end_p): Add argument.
33156 * tree.c (array_at_struct_end_p): Handle it.
33158 2017-02-03 Martin Liska <mliska@suse.cz>
33161 * multiple_target.c (create_dispatcher_calls): Redirect edge
33162 from a caller of a dispatcher.
33163 (expand_target_clones): Make the clones local.
33164 (ipa_target_clone): Do both target clones and resolvers.
33165 (ipa_dispatcher_calls): Remove the pass.
33166 (pass_dispatcher_calls::gate): Likewise.
33167 (make_pass_dispatcher_calls): Likewise.
33168 * passes.def (pass_target_clone): Put as very first IPA early
33171 2017-02-03 Martin Liska <mliska@suse.cz>
33173 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
33174 in case of a function with ifunc attribute.
33176 2017-02-03 Martin Liska <mliska@suse.cz>
33178 * cgraph.c (cgraph_node::dump): Dump function version info.
33179 * symtab.c (symtab_node::dump_base): Add missing new line.
33181 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33183 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
33184 (ifcombine_ifandif): Use it.
33186 2017-02-03 Martin Liska <mliska@suse.cz>
33188 * doc/invoke.texi: Document default value for
33189 use-after-scope-direct-emission-threshold.
33191 2017-02-03 Martin Liska <mliska@suse.cz>
33193 PR tree-optimization/79339
33194 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
33195 (format_floating): Likewise.
33197 2017-02-03 Martin Liska <mliska@suse.cz>
33200 * ipa-prop.c (ipa_node_params_t::insert): Remove current
33202 (ipa_node_params_t::remove): Likewise.
33203 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
33204 initialization from removed ipa_node_params_t::insert.
33205 (ipa_node_params::~ipa_node_params): Move from removed
33206 ipa_node_params_t::release.
33207 * symbol-summary.h (symbol_summary::m_released): New member.
33208 Do not release a summary twice. Do not allow to call finalizer
33209 for types of a summary that live in GGC memory.
33211 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33213 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
33216 2017-02-02 Martin Sebor <msebor@redhat.com>
33218 PR middle-end/79275
33219 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
33220 (format_string): Tighten up the range of output for non-constant
33221 strings and correct the expected range for wide non-constant strings.
33223 2017-02-02 Martin Sebor <msebor@redhat.com>
33225 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
33227 PR middle-end/32003
33228 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
33230 (-fdump-tree-@var): Add to index and document how to come up
33231 with pass-specific option and dump file names.
33232 (-fdump-passes): Clarify where to look for output.
33234 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33236 PR middle-end/77445
33237 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33238 statistics of the analyzed path; allow threading for speed when
33239 any of BBs along the path are optimized for speed.
33241 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33243 PR middle-end/78468
33244 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33245 settings of the virtual registers.
33248 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33250 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33251 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33254 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33256 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33257 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33259 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33261 * config/s390/s390.md: Add missing comments with the expanded
33263 * config/s390/vector.md: Likewise.
33264 * config/s390/vx-builtins.md: Likewise.
33266 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33269 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33270 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33271 conditions on a single line.
33273 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33275 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33276 __S390_VX__ to __VX__.
33278 2017-02-01 Andrew Pinski <apinski@cavium.com>
33280 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33281 stmt_info to record_stmt_cost.
33282 (vect_get_known_peeling_cost): Pass stmt_info if known to
33284 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33285 cpu_vector_cost field into
33286 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33287 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33288 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33289 splitting of scalar_stmt_cost and vec_stmt_cost.
33290 (thunderx_vector_cost): Likewise.
33291 (cortexa57_vector_cost): LIkewise.
33292 (exynosm1_vector_cost): Likewise.
33293 (xgene1_vector_cost): Likewise.
33294 (thunderx2t99_vector_cost): Improve after the splitting of the two
33296 (aarch64_builtin_vectorization_cost): Update for the splitting of
33297 scalar_stmt_cost and vec_stmt_cost.
33299 2017-02-01 Torvald Riegel <triegel@redhat.com>
33300 Richard Henderson <rth@redhat.com>
33302 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33303 conditional on existance of a fast atomic load.
33304 * optabs-query.c (can_atomic_load_p): New function.
33305 * optabs-query.h (can_atomic_load_p): Declare it.
33306 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33307 no fast atomic load is available for the particular size of access.
33308 (expand_atomic_compare_and_swap): Likewise.
33309 (expand_atomic_load): Likewise.
33310 (expand_atomic_store): Likewise.
33311 (expand_atomic_fetch_op): Likewise.
33312 * testsuite/lib/target-supports.exp
33313 (check_effective_target_sync_int_128): Remove x86 because it provides
33314 no fast atomic load.
33315 (check_effective_target_sync_int_128_runtime): Likewise.
33317 2017-02-01 Richard Biener <rguenther@suse.de>
33319 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33320 (graphite_transform_loops): Provide opt-info for optimized nests.
33321 * tree-parloop.c (parallelize_loops): Provide opt-info for
33322 parallelized loops.
33324 2017-02-01 Richard Biener <rguenther@suse.de>
33326 PR middle-end/79315
33327 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33328 was not set before.
33330 2017-02-01 Richard Biener <rguenther@suse.de>
33332 PR tree-optimization/71824
33333 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33334 Verify the loops are valid in the merged SESE region.
33335 (scop_detection::can_represent_loop_1): Check analyzing the
33336 evolution of the number of iterations in the region succeeds.
33338 2017-01-31 Ian Lance Taylor <iant@golang.org>
33340 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33341 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33343 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33345 PR preprocessor/79210
33346 * input.c (get_substring_ranges_for_loc): Replace line_width
33347 assertion with error-handling.
33349 2017-01-31 Richard Biener <rguenther@suse.de>
33351 PR tree-optimization/77318
33352 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33353 (create_pw_aff_from_tree): Take loop parameter.
33354 (add_condition_to_pbb): Pass loop of the condition to
33355 create_pw_aff_from_tree.
33357 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33359 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33360 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33362 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33366 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33368 (convert_int_to_float128): Likewise.
33369 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33370 (convert_int_to_float128): Likewise.
33371 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33372 (UNSPEC_IEEE128_CONVERT): Likewise.
33373 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33374 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33375 Use local variables for IBM extended format.
33376 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33377 (fix_trunc<mode>si2_fprs): Likewise.
33378 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33379 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33380 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33381 to know that we can now have integers of all sizes in vector
33383 (fix<uns>_<mode>di2_hw): Likewise.
33384 (float<uns>_<mode>si2_hw): Likewise.
33385 (fix_<mode>si2_hw): Likewise.
33386 (fixuns_<mode>si2_hw): Likewise.
33387 (float<uns>_<mode>di2_hw): Likewise.
33388 (float_<mode>di2_hw): Likewise.
33389 (float_<mode>si2_hw): Likewise.
33390 (floatuns_<mode>di2_hw): Likewise.
33391 (floatuns_<mode>si2_hw): Likewise.
33392 (xscvqp<su>wz_<mode>): Delete, no longer used.
33393 (xscvqp<su>dz_<mode>): Likewise.
33394 (xscv<su>dqp_<mode>): Likewise.
33395 (ieee128_mfvsrd_64bit): Likewise.
33396 (ieee128_mfvsrd_32bit): Likewise.
33397 (ieee128_mfvsrwz): Likewise.
33398 (ieee128_mtvsrw): Likewise.
33399 (ieee128_mtvsrd_64bit): Likewise.
33400 (ieee128_mtvsrd_32bit): Likewise.
33402 2017-01-31 Martin Liska <mliska@suse.cz>
33405 * ipa-prop.c (ipa_free_all_node_params): Call release method
33406 instead of ~sumbol_summary to not to trigger double times
33409 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33411 PR tree-optimization/71691
33412 * bitmap.h (class auto_bitmap): New.
33413 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33414 is_maybe_undefined instead of ssa_undefined_value_p.
33416 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33418 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33419 __S390_ARCH_LEVEL__ to __ARCH__.
33421 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33423 PR tree-optimization/79267
33424 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33425 if should_remove_lhs_p is true.
33427 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33430 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33431 (add_alignment_attribute): New.
33432 (base_type_die): Add alignment attribute.
33433 (subrange_type_die): Likewise.
33434 (modified_type_die): Likewise.
33435 (gen_array_type_die): Likewise.
33436 (gen_descr_array_type_die: Likewise.
33437 (gen_enumeration_type_die): Likewise.
33438 (gen_subprogram_die): Likewise.
33439 (gen_variable_die): Likewise.
33440 (gen_field_die): Likewise.
33441 (gen_ptr_to_mbr_type_die): Likewise.
33442 (gen_struct_or_union_type_die): Likewise.
33443 (gen_subroutine_type_die): Likewise.
33444 (gen_typedef_die): Likewise.
33445 (base_type_cmp): Compare alignment attribute.
33447 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33450 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
33451 (setb_unsigned) New pattern for setb with CCUNS.
33452 * config/rs6000/rs6000.c (expand_block_compare): Use a different
33453 subfc./subfe sequence to avoid overflow problems. Generate a
33454 shorter sequence with cmpld/setb for power9.
33455 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
33456 for generating subfc. instruction.
33457 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
33458 now uses this instruction.
33460 2017-01-30 Ian Lance Taylor <iant@google.com>
33463 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
33464 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
33466 2017-01-30 Martin Sebor <msebor@redhat.com>
33468 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
33469 Move constant to the right of a relational operator.
33470 (get_mpfr_format_length, format_character, format_string): Ditto.
33471 (should_warn_p, maybe_warn): Same.
33473 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
33475 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
33478 * asan.c (get_translation_unit_decl): Remove function.
33479 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
33481 2017-01-30 Martin Liska <mliska@suse.cz>
33483 PR gcov-profile/79259
33484 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
33485 -fprofile-generate.
33487 2017-01-30 Martin Liska <mliska@suse.cz>
33490 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
33491 Initialize variables with NULL value.
33493 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
33496 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
33498 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
33500 2017-01-30 Richard Biener <rguenther@suse.de>
33502 PR tree-optimization/79276
33503 * tree-vrp.c (process_assert_insertions): Properly adjust common
33504 when removing a duplicate.
33506 2017-01-30 Richard Biener <rguenther@suse.de>
33508 PR tree-optimization/79256
33509 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33510 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33512 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
33514 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
33517 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
33518 ("*r<noxa>sbg_<mode>_sll_bitmask")
33519 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
33520 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
33521 Use contiguous_bitmask_nowrap_operand.
33523 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33526 * config/rs6000/altivec.h (vec_xl): Revise #define.
33527 (vec_xst): Likewise.
33529 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
33531 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
33533 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
33535 PR rtl-optimization/79194
33536 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
33537 traps before call to bypass_conditional_jumps.
33539 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33541 PR tree-optimization/71374
33542 * lra-constraints.c (check_conflict_input_operands): New.
33543 (match_reload): Use it.
33545 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33548 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
33549 account to calculate conflict_set.
33551 2017-01-27 Bin Cheng <bin.cheng@arm.com>
33553 PR rtl-optimization/78559
33554 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
33555 other_insn in combine.
33557 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33559 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
33560 uint16_type_node for BT_UINT16.
33562 2017-01-27 David Malcolm <dmalcolm@redhat.com>
33564 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
33565 "RTL Tests" to menu.
33566 (GIMPLE Tests): New node.
33567 (RTL Tests): New node.
33569 2017-01-27 Richard Biener <rguenther@suse.de>
33571 PR tree-optimization/79245
33572 * tree-loop-distribution.c (distribute_loop): Apply cost
33573 modeling also to detected patterns.
33575 2017-01-27 Richard Biener <rguenther@suse.de>
33577 PR tree-optimization/71433
33578 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33579 (compare_assert_loc): New function.
33580 (process_assert_insertions): Sort and optimize assert locations
33581 to remove duplicates and push down identical assertions on
33582 edges to their destination block.
33584 2017-01-27 Richard Biener <rguenther@suse.de>
33586 PR tree-optimization/79244
33587 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33588 out SSA names even if abnormal.
33590 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33592 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33593 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33594 instead of MPFR_RNDN.
33596 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33599 * arm.c (arm_option_override): Don't call build_target_option_node
33600 until after doing all option overrides.
33601 (arm_valid_target_attribute_tree): Likewise.
33603 2017-01-27 Martin Liska <mliska@suse.cz>
33605 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33606 for {cd}tors and C++ {cd}tors.
33608 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33610 * config/s390/s390.md ("*setmem_long_and")
33611 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33613 2017-01-26 Martin Sebor <msebor@redhat.com>
33615 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33618 2017-01-26 Martin Sebor <msebor@redhat.com>
33620 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33621 HAVE_DFmode before using XFmode or DFmode.
33622 (parse_directive): Avoid using the z length modifier to avoid
33623 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33625 PR middle-end/78703
33626 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33627 to accept adjustment as an array.
33628 (get_int_range): New function.
33629 (struct directive): Make width and prec arrays.
33630 (directive::set_width, directive::set_precision): Call get_int_range.
33631 (format_integer, format_floating): Handle width and precision ranges.
33632 (format_string, parse_directive): Same.
33634 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33637 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33638 comdat_type_p set, just clone them, but keep the children in the
33642 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33643 which have direct callers with -fvar-tracking-assignments enabled
33645 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33646 inside of type units.
33648 2017-01-26 Martin Sebor <msebor@redhat.com>
33650 PR middle-end/78703
33651 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33653 (struct format_result): Replace number_chars, number_chars_min,
33654 and number_chars_max with a single member of struct result_range.
33656 (format_result::operator+=): Adjust.
33657 (struct fmtresult): Remove bounded. Handle likely and unlikely
33659 (fmtresult::adjust_for_width_or_precision): New function.
33660 (fmtresult:type_max_digits): New function.
33661 (bytes_remaining): Handle likely and unlikely counters.
33662 (min_bytes_remaining): Remove.
33663 (format_percent): Simplify.
33664 (format_integer, format_floating): Set likely and unlikely counters.
33665 (get_string_length, format_character, format_string): Same.
33666 (format_plain, should_warn_p): New function.
33667 (maybe_warn): Call should_warn_p. Update diagnostic messages
33668 and handle those for all directives, including plain strings.
33669 (format_directive): Handle likely and unlikely counters.
33670 Remove unnecessary quoting from diagnostics. Add an informational
33672 (add_bytes): Remove.
33673 (pass_sprintf_length::compute_format_length): Simplify.
33674 (try_substitute_return_value): Handle likely and unlikely counters.
33676 2017-01-26 Carl Love <cel@us.ibm.com>
33678 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33679 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33681 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33684 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33685 endianess for subregs into account.
33686 * lra-constraints.c (lra_constraints): Do risky transformations
33687 always on the first iteration.
33688 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33689 last_call_used_reg_set.
33690 (process_bb_lives): Define and use last_call_used_reg_set.
33691 * lra.c (lra): Always continue after lra_constraints on the first
33694 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33696 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33698 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33700 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33702 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33703 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33704 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33705 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33706 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33707 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33708 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33709 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33710 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33712 2017-01-26 Marek Polacek <polacek@redhat.com>
33715 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33716 for the third operand.
33718 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33720 PR middle-end/79236
33721 * omp-low.c (struct omp_context): Add simt_stmt field.
33722 (scan_omp_for): Return omp_context *.
33723 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33724 context to the _simt_ SIMD stmt.
33725 (lower_omp_for): For combined SIMD with sibling _simt_
33726 SIMD, make sure to use the same decls in _looptemp_
33727 clauses as in the sibling.
33729 2017-01-26 David Sherwood <david.sherwood@arm.com>
33731 PR middle-end/79212
33732 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33735 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33738 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33739 emit fld b; fld a; if possible.
33741 * brig-builtins.def: Update copyright years.
33742 * config/arm/arm_acle_builtins.def: Update copyright years.
33744 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33747 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33748 constraint instead of o for the stxsd instruction.
33750 2017-01-25 Carl Love <cel@us.ibm.com>
33752 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33753 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33755 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33757 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33759 2017-01-25 Richard Biener <rguenther@suse.de>
33761 PR tree-optimization/69264
33762 * target.def (vector_alignment_reachable): Improve documentation.
33763 * doc/tm.texi: Regenerate.
33764 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33766 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33767 earlier changes with respect to TYPE_USER_ALIGN.
33768 (vector_alignment_reachable_p): Likewise. Improve dumping.
33770 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33773 * config/arm/arm.md (xordi3): Force constant operand into a register
33776 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33778 * doc/invoke.texi (-fstore-merging): Correct default optimization
33779 levels at which it is enabled.
33780 (-O): Move -fstore-merging from list to...
33783 2017-01-25 Richard Biener <rguenther@suse.de>
33786 * omp-expand.c: Include debug.h.
33787 (expand_omp_taskreg): Make sure to generate early debug before
33788 outlining anything from a function.
33789 (expand_omp_target): Likewise.
33790 (grid_expand_target_grid_body): Likewise.
33792 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33795 * asan.c (get_translation_unit_decl): New function.
33796 (asan_add_global): Extract modules file name from globals
33797 TRANSLATION_UNIT_DECL name.
33799 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33802 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33803 for long calls with APCS frame and VFP.
33805 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33807 * cfg.c (original_copy_tables_initialized_p): New function.
33808 * cfg.h (original_copy_tables_initialized_p): New decl.
33809 * cfgrtl.c (relink_block_chain): Guard the call to
33810 free_original_copy_tables with a call to
33811 original_copy_tables_initialized_p.
33812 * cgraph.h (symtab_node::native_rtl_p): New decl.
33813 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33814 (symtab_node::needed_p): Don't assert for early assembly output
33815 for __RTL functions.
33816 (cgraph_node::finalize_function): Set "force_output" for __RTL
33818 (cgraph_node::analyze): Bail out early for __RTL functions.
33819 (analyze_functions): Update assertion to support __RTL functions.
33820 (cgraph_node::expand): Bail out early for __RTL functions.
33821 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33823 * function.h (struct function): Update comment for field
33825 * gimple-expr.c: Include "tree-pass.h".
33826 (gimple_has_body_p): Return false for __RTL functions.
33827 * Makefile.in (OBJS): Add run-rtl-passes.o.
33828 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33830 (gcc::pass_manager::get_clean_slate): New accessor.
33831 * passes.c: Include "insn-addr.h".
33832 (should_skip_pass_p): Add logging. Update logic for running
33833 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33834 property-provider override so it is only done for gimple passes.
33836 (skip_pass): New function.
33837 (execute_one_pass): Call skip_pass when skipping passes.
33838 * read-md.c (md_reader::read_char): Support filtering
33839 the input to a subset of line numbers.
33840 (md_reader::md_reader): Initialize fields
33841 m_first_line and m_last_line.
33842 (md_reader::read_file_fragment): New function.
33843 * read-md.h (md_reader::read_file_fragment): New decl.
33844 (md_reader::m_first_line): New field.
33845 (md_reader::m_last_line): New field.
33846 * read-rtl-function.c (function_reader::create_function): Only
33847 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33848 curr_properties. Set DECL_INITIAL to a dummy block.
33849 (read_rtl_function_body_from_file_range): New function.
33850 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33852 * run-rtl-passes.c: New file.
33853 * run-rtl-passes.h: New file.
33855 2017-01-24 Jeff Law <law@redhat.com>
33857 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33860 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33862 PR tree-optimization/79159
33863 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33864 (record_nonwrapping_iv): Improve boundary using above function if no
33865 value range information.
33867 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33868 Martin Jambor <mjambor@suse.cz>
33870 * brig-builtins.def: New file.
33871 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33872 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33873 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33874 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33875 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33876 * builtin-types.def (BT_INT8): New.
33877 (BT_INT16): Likewise.
33878 (BT_UINT8): Likewise.
33879 (BT_UINT16): Likewise.
33880 (BT_FN_ULONG): Likewise.
33881 (BT_FN_UINT_INT): Likewise.
33882 (BT_FN_UINT_ULONG): Likewise.
33883 (BT_FN_UINT_LONG): Likewise.
33884 (BT_FN_UINT_PTR): Likewise.
33885 (BT_FN_ULONG_PTR): Likewise.
33886 (BT_FN_INT8_FLOAT): Likewise.
33887 (BT_FN_INT16_FLOAT): Likewise.
33888 (BT_FN_UINT32_FLOAT): Likewise.
33889 (BT_FN_UINT16_FLOAT): Likewise.
33890 (BT_FN_UINT8_FLOAT): Likewise.
33891 (BT_FN_UINT64_FLOAT): Likewise.
33892 (BT_FN_UINT16_UINT32): Likewise.
33893 (BT_FN_UINT32_UINT16): Likewise.
33894 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33895 (BT_FN_INT_PTR_INT): Likewise.
33896 (BT_FN_UINT_PTR_UINT): Likewise.
33897 (BT_FN_LONG_PTR_LONG): Likewise.
33898 (BT_FN_ULONG_PTR_ULONG): Likewise.
33899 (BT_FN_VOID_UINT64_UINT64): Likewise.
33900 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33901 (BT_FN_INT8_INT8_INT8): Likewise.
33902 (BT_FN_INT16_INT16_INT16): Likewise.
33903 (BT_FN_INT_INT_INT): Likewise.
33904 (BT_FN_UINT_FLOAT_UINT): Likewise.
33905 (BT_FN_FLOAT_UINT_UINT): Likewise.
33906 (BT_FN_ULONG_UINT_UINT): Likewise.
33907 (BT_FN_ULONG_UINT_PTR): Likewise.
33908 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33909 (BT_FN_UINT_UINT_UINT): Likewise.
33910 (BT_FN_VOID_UINT_PTR): Likewise.
33911 (BT_FN_UINT_UINT_PTR: Likewise.
33912 (BT_FN_UINT32_UINT64_PTR): Likewise.
33913 (BT_FN_INT_INT_UINT_UINT): Likewise.
33914 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33915 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33916 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33917 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33918 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33919 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33920 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33921 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33922 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33923 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33924 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33925 * doc/frontends.texi: List BRIG FE.
33926 * doc/install.texi (Testing): Add BRIG tesring requirements.
33927 * doc/invoke.texi (Overall Options): Mention BRIG.
33928 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33930 2017-01-24 Richard Biener <rguenther@suse.de>
33932 PR translation/79208
33933 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33935 2017-01-24 Martin Jambor <mjambor@suse.cz>
33938 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33939 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33940 and known_contexts.
33942 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33944 PR middle-end/79123
33945 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33946 casts from signed to unsigned really don't have a range.
33948 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33950 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33951 GMP_RNDx for compatiblity.
33953 2017-01-24 Martin Liska <mliska@suse.cz>
33956 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33957 that would prevent us to call alloca with -1 as argument.
33959 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33961 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33962 Avoid -Wformat-security warning.
33964 2017-01-23 Andrew Pinski <apinski@cavium.com>
33966 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33969 2017-01-23 Martin Sebor <msebor@redhat.com>
33971 PR middle-end/78703
33972 * gimple-ssa-sprintf.c (warn_level): New global.
33973 (format_integer): Use it here and throughout the rest of the file.
33974 Use the same switch to compute sign as base.
33975 (maybe_warn): New function.
33976 (format_directive): Factor out warnings into maybe_warn.
33977 Add debugging output. Use warn_level.
33978 (add_bytes): Use warn_level.
33979 (pass_sprintf_length::compute_format_length): Add debugging output.
33980 (try_substitute_return_value): Same.
33981 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33983 PR middle-end/78703
33984 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33985 (struct fmtresult, format_integer, format_floating): Adjust.
33986 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33987 (get_string_length, format_string,format_directive): Same.
33988 (pass_sprintf_length::compute_format_length): Same.
33989 (try_substitute_return_value): Simplify slightly.
33991 PR middle-end/78703
33992 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33993 (fmtresult::operator+=): Outlined.
33994 (struct fmtresult): Add ctors.
33995 (struct conversion_spec): Rename...
33996 (struct directive): ...to this. Add and remove data members.
33997 (directive::set_width, directive::set_precision): New functions.
33998 (format_percent): Use fmtresult ctor.
33999 (get_width_and_precision): Remove.
34000 (format_integer): Make naming changes. Avoid computing width and
34002 (format_floating): Same. Adjust indentation.
34003 (format_character, format_none): New functions.
34004 (format_string): Moved character handling to format_character.
34005 (format_directive): Remove arguments, change return type.
34006 (parse_directive): New function.
34007 (pass_sprintf_length::compute_format_length): Move directive
34008 parsing to parse_directive.
34010 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34012 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34013 (assign_assembler_name_if_needed): ... this.
34014 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34015 (assign_assembler_name_if_needed): ... this.
34016 (free_lang_data_in_cgraph): Adjust callers.
34017 * cgraphunit.c (cgraph_node::analyze): Likewise.
34018 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34021 2017-01-23 Richard Biener <rguenther@suse.de>
34023 PR tree-optimization/79088
34024 PR tree-optimization/79188
34025 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34026 resetting loop bounds after last path deletion. Reset loop
34027 bounds of the target loop, make code match the comments.
34028 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34029 Make sure loops need no fixups.
34031 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34033 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34034 exponent support with double type for first argument.
34035 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34036 type returned by __builtin_vec_extract_sig,
34037 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34038 functions from "vector int" to "vector unsigned int" or from
34039 "vector long long int" to "vector unsigned long long int".
34040 Changed type returned by __builtin_vec_extract_exp,
34041 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34042 functions from "vector int" to "vector unsigned int" or from
34043 "vector long long int" to "vector unsigned long long int".
34044 Changed return type of __builtin_vec_test_data_class,
34045 __builtin_vec_test_data_class_sp, and
34046 __builtin_vec_test_data_class_dp from "vector int" to
34047 "vector bool int" or from "vector long long int" to "vector bool
34048 long long int" and changed second argument type from "unsigned
34049 int" to "int". Added new overloaded function forms "vector float
34050 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34051 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34052 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34053 double, vector unsigned long long int)" and "vector double
34054 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34055 long int)". Changed return type of
34056 __builtin_scalar_test_data_class and
34057 __builtin_scalar_test_data_class_sp and
34058 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34059 int" and changed second argument from "unsigned int" to "int".
34060 Changed type returned by __builtin_scalar_test_neg,
34061 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34062 from "int" to "bool int". Added new overloaded function form
34063 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34064 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34065 exponent double-precision with floating point first argument.
34066 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34067 documentation of scalar_test_data_class, scalar_test_neg,
34068 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34069 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34070 vec_test_data_class built-in functions to reflect refinements in
34071 their type signatures.
34073 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34075 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34077 (aarch64_elf_asm_destructor): Likewise.
34079 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34081 PR rtl-optimization/78634
34082 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34083 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34084 * ifcvt.c (noce_try_cmove): Add missing cost check.
34086 PR rtl-optimization/71724
34087 * combine.c (if_then_else_cond): Look for situations where it is
34088 beneficial to undo the work of one of the recursive calls.
34090 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34092 PR tree-optimization/70754
34093 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34094 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34095 combined stmt before it if not NULL.
34096 (combine_chains): Process refs reversely and compute dominance point
34099 2017-01-23 Martin Liska <mliska@suse.cz>
34101 PR tree-optimization/79196
34102 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34103 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34105 (strlen_optimize_stmt): Call the renamed function.
34107 2017-01-23 Michael Matz <matz@suse.de>
34109 PR tree-optimization/78384
34110 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34112 2017-01-23 Richard Biener <rguenther@suse.de>
34114 PR tree-optimization/79186
34115 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34116 both incoming edges before moving an assert.
34118 2017-01-23 Martin Jambor <mjambor@suse.cz>
34120 * ipa-prop.c (load_from_param_1): Removed.
34121 (load_from_unmodified_param): Bits from load_from_param_1 put back
34123 (load_from_param): Removed.
34124 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34125 with stmt. Reverted back to use of load_from_unmodified_param.
34127 2017-01-23 Martin Jambor <mjambor@suse.cz>
34130 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34131 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34132 field a pointer to garbage collected vector, mark lattices and
34133 ipcp_orig_node with GTY((skip)).
34134 (ipa_get_param_count): Adjust to descriptors being a pointer.
34135 (ipa_get_param): Likewise.
34136 (ipa_get_type): Likewise.
34137 (ipa_get_param_move_cost): Likewise.
34138 (ipa_set_param_used): Likewise.
34139 (ipa_get_controlled_uses): Likewise.
34140 (ipa_set_controlled_uses): Likewise.
34141 (ipa_is_param_used): Likewise.
34142 (ipa_node_params_t): Move into garbage collector. New methods insert
34144 (ipa_node_params_sum): Annotate wth GTY(()).
34145 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
34147 (ipa_load_from_parm_agg): Adjust declaration.
34148 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
34149 * ipa-profile.c (ipa_profile): Likewise.
34150 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
34151 (ipa_populate_param_decls): Make descriptors parameter garbage
34153 (ipa_dump_param): Adjust to descriptors being a pointer.
34154 (ipa_alloc_node_params): Likewise.
34155 (ipa_initialize_node_params): Likewise.
34156 (load_from_param_1): Make descriptors parameter garbage collected.
34157 (load_from_unmodified_param): Likewise.
34158 (load_from_param): Likewise.
34159 (ipa_load_from_parm_agg): Likewise.
34160 (ipa_node_params::~ipa_node_params): Removed.
34161 (ipa_free_all_node_params): Remove call to delete operator.
34162 (ipa_node_params_t::insert): New.
34163 (ipa_node_params_t::remove): Likewise.
34164 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
34165 copy known_csts and known_contexts vectors.
34166 (ipa_read_node_info): Adjust to descriptors being a pointer.
34167 (ipcp_modif_dom_walker): Make m_descriptors field garbage
34169 (ipcp_transform_function): Make descriptors variable garbage
34172 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
34174 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
34175 * config/i386/avx512dqintrin.h: Ditto.
34176 * config/i386/avx512fintrin.h: Ditto.
34177 * config/i386/i386.c: Handle new builtins.
34178 * config/i386/i386-builtin.def: Add new builtins.
34179 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
34180 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
34182 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34183 Martin Liska <mliska@suse.cz>
34185 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
34186 * asan.c (asan_expand_poison_ifn): Support stores and use
34187 appropriate ASAN report function.
34188 * internal-fn.c (expand_ASAN_POISON_USE): New function.
34189 * internal-fn.def (ASAN_POISON_USE): Declare.
34190 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
34191 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
34192 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
34193 ASAN_POISON calls w/o LHS.
34194 * tree-ssa.c (execute_update_addresses_taken): Create clobber
34195 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
34196 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
34197 * gimplify.c (asan_poison_variables): Add attribute
34198 use_after_scope_memory to variables that really needs to live
34200 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
34201 having the attribute.
34203 2017-01-23 Martin Liska <mliska@suse.cz>
34205 * asan.c (create_asan_shadow_var): New function.
34206 (asan_expand_poison_ifn): Likewise.
34207 * asan.h (asan_expand_poison_ifn): New declaration.
34208 * internal-fn.c (expand_ASAN_POISON): Likewise.
34209 * internal-fn.def (ASAN_POISON): New builtin.
34210 * sanopt.c (pass_sanopt::execute): Expand
34211 asan_expand_poison_ifn.
34212 * tree-inline.c (copy_decl_for_dup_finish): Make function
34214 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
34215 * tree-ssa.c (is_asan_mark_p): New function.
34216 (execute_update_addresses_taken): Rewrite local variables
34217 (identified just by use-after-scope as addressable) into SSA.
34219 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34221 * doc/install.texi (Specific): opensource.apple.com uses https
34222 now. Remove trailing slash.
34224 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34226 * README.Portability: Remove note on an Irix compatibility issue.
34228 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
34230 * gcov.c (INCLUDE_ALGORITHM): Define.
34231 (INCLUDE_VECTOR): Define.
34232 No longer include <vector> and <algorithm> directly.
34234 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34236 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34238 * doc/invoke.texi (Code Gen Options): Ditto.
34240 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34243 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34245 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34247 rtl-optimization/79125
34248 * cprop.c (local_cprop_pass): Handle cases where we make an
34249 unconditional trap.
34251 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34255 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34256 read from, for big endian.
34258 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34260 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34261 register pauth builtins for LP64 only.
34263 2017-01-20 Marek Polacek <polacek@redhat.com>
34266 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34269 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34271 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34273 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34274 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34275 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34277 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34280 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34281 in big-endian mode when they are not a single duplicated value.
34283 2017-01-20 Richard Biener <rguenther@suse.de>
34285 * BASE-VER: Bump to 7.0.1.
34287 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34289 * omp-low.c (omplow_simd_context): New struct. Use it...
34290 (lower_rec_simd_input_clauses): ...here and...
34291 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34292 references to idx, lane, max_vf, is_simt.
34294 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34296 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34299 2017-01-20 Martin Jambor <mjambor@suse.cz>
34301 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34302 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34304 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34305 (GTFILES): Rename hsa.c to hsa-common.c.
34306 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34307 * hsa-dump.c: Likewise.
34308 * hsa-gen.c: Likewise.
34309 * hsa-regalloc.c: Likewise.
34310 * ipa-hsa.c: Likewise.
34311 * omp-expand.c: Likewise.
34312 * omp-low.c: Likewise.
34313 * toplev.c: Likewise.
34315 2017-01-20 Marek Polacek <polacek@redhat.com>
34318 * doc/invoke.texi: Document -Wduplicated-branches.
34319 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34320 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34321 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34322 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34323 return 0 only when not OEP_LEXICOGRAPHIC.
34324 (fold_build_cleanup_point_expr): Use the expression
34325 location when building CLEANUP_POINT_EXPR.
34326 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34327 * tree.c (add_expr): Handle error_mark_node.
34329 2017-01-20 Martin Liska <mliska@suse.cz>
34332 * tree-profile.c (init_ic_make_global_vars): Do not call
34334 (gimple_init_gcov_profiler): Likewise.
34336 2017-01-20 Martin Liska <mliska@suse.cz>
34339 * cgraph.h (maybe_create_reference): Remove argument and
34341 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34343 * ipa-cp.c (create_specialized_node): Likewise.
34344 * symtab.c (symtab_node::maybe_create_reference): Handle
34345 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34347 2017-01-20 Martin Liska <mliska@suse.cz>
34349 * read-rtl-function.c (function_reader::create_function): Use
34350 build_decl instread of build_decl_stat.
34352 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34354 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34355 * config/i386/avx512dqintrin.h: Ditto.
34356 * config/i386/avx512fintrin.h: Ditto.
34357 * config/i386/i386-builtin-types.def: Add new types.
34358 * config/i386/i386.c: Handle new types.
34359 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34360 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34361 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34362 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34363 (__builtin_ia32_kshiftridi): New.
34364 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34366 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34370 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34371 define to rs6000_init_stack_protect_guard.
34372 (rs6000_init_stack_protect_guard): New function.
34374 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34375 Yunqiang Su <yunqiang.su@imgtec.com>
34377 * config.gcc (supported_defaults): Add madd4.
34378 (with_madd4): Add validation.
34379 (all_defaults): Add madd4.
34380 * config/mips/mips.opt (mmadd4): New option.
34381 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34383 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34385 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34386 (ISA_HAS_FUSED_MADD4): Likewise.
34387 * doc/invoke.texi (-mmadd4): Document the new option.
34388 * doc/install.texi (--with-madd4): Document the new option.
34390 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34392 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34393 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34394 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34395 (aarch64_init_pauth_hint_builtins): New.
34396 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34397 (aarch64_expand_builtin): Expand new builtins.
34399 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34401 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34402 * combine-stack-adj.c (no_unhandled_cfa): Handle
34403 REG_CFA_TOGGLE_RA_MANGLE.
34404 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34405 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34406 info for return address signing.
34407 (aarch64_expand_epilogue): Likewise.
34409 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34411 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34412 * config/aarch64/aarch64-protos.h
34413 (aarch64_return_address_signing_enabled): New declaration.
34414 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34416 (aarch64_expand_prologue): Sign return address before it's pushed onto
34418 (aarch64_expand_epilogue): Authenticate return address fetched from
34420 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34421 (aarch64_attributes): New function attributes for "sign-return-address".
34422 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34423 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34424 ("*do_return"): Generate combined instructions according to key index.
34425 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34426 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34428 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34429 * config/aarch64/aarch64.opt (msign-return-address=): New.
34430 * doc/extend.texi (AArch64 Function Attributes): Documents
34431 "sign-return-address=".
34432 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34434 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34436 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34437 overall option summary.
34439 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34441 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
34442 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
34443 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
34444 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
34446 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
34448 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
34449 -mpower9-minmax by default for -mcpu=power9.
34450 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
34451 128-bit floating point.
34453 2017-01-20 Alan Modra <amodra@gmail.com>
34455 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
34456 optimizing for size.
34458 2017-01-20 Alan Modra <amodra@gmail.com>
34461 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
34462 for strcmp and strncmp from corresponding builtin decl.
34464 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34466 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
34467 instead of i386/rtems-64.h.
34468 * config/i386/rtems-64.h: Remove.
34470 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34474 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
34476 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
34478 2017-01-19 Tamar Christina <tamar.christina@arm.com>
34480 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
34481 Change int to HOST_WIDE_INT.
34482 * config/aarch64/aarch64-protos.h
34483 (aarch64_simd_gen_const_vector_dup): Likewise.
34484 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
34486 2017-01-19 David Malcolm <dmalcolm@redhat.com>
34488 * langhooks-def.h (lhd_type_for_size): New decl.
34489 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
34490 * langhooks.c (lhd_type_for_size): New function, taken from
34493 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
34495 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
34496 define_bypass for CR latency.
34497 (power9-cracked-alu): Update bypass latency and remove power9-branch.
34498 (power9-alu2): Add define_bypass for CR latency.
34500 (power9-mul): Update insn latency.
34501 (power9-mul-compare): Update insn latency, bypass latency and remove
34504 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34506 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
34508 * config/aarch64/aarch64.md
34509 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
34510 aarch64_nopcrelative_literal_loads.
34511 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
34513 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
34515 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
34516 TARGET_LOONGSON_3A.
34517 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
34519 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
34522 * config.gcc (supported_defaults): Add lxc1-sxc1.
34523 (with_lxc1_sxc1): Add validation.
34524 (all_defaults): Add lxc1-sxc1.
34525 * config/mips/mips.opt (mlxc1-sxc1): New option.
34526 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34528 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34529 __mips_no_lxc1_sxc1.
34530 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
34531 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
34532 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
34534 2017-01-19 Richard Biener <rguenther@suse.de>
34536 PR tree-optimization/72488
34537 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
34538 sure to restore SSA info.
34539 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
34541 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
34543 PR rtl-optimization/79121
34544 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
34545 of the inner type when shifting an extended value.
34547 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34550 * symtab.c (symtab_node::equal_address_to): Fix comparing of
34551 interposable aliases.
34553 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
34556 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
34557 Use the evmergelohi instruction.
34558 (mov_si<mode>_e500_subreg4_2_le): Likewise.
34559 (mov_sitf_e500_subreg8_2_be): Likewise.
34560 (mov_sitf_e500_subreg12_2_le): Likewise.
34561 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
34562 (mov_si<mode>_e500_subreg4_2_be): Likewise.
34563 (mov_sitf_e500_subreg8_2_le): Likewise.
34564 (mov_sitf_e500_subreg12_2_be): Likewise.
34566 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34568 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
34569 attribute from vecsimple to vecperm.
34570 (altivec_vbpermq2): Likewise.
34572 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34575 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34577 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34578 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34579 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34580 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34581 case where N arg is SIZE_MAX.
34582 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34583 (cmpstrsi): Add pattern.
34585 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34587 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34588 __builtin_vec_revb builtins.
34589 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34590 built-in functions to support generation of the ISA 3.0 XXBR<x>
34591 vector byte reverse instructions.
34592 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34593 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34594 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34595 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34596 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34597 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34598 (P9V_BUILTIN_VEC_REVB): Likewise.
34599 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34600 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34601 (p9_xxbrq_v16qi): Likewise.
34602 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34603 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34604 (p9_xxbrh_v8hi): Likewise.
34605 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34606 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34607 vec_revb built-in functions.
34609 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34611 PR rtl-optimization/78952
34612 * config/i386/i386.md (any_extract): New code iterator.
34613 (*insvqi_2): Use any_extract for source operand.
34614 (*insvqi_3): Use any_shiftrt for source operand.
34616 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34618 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34620 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34622 2017-01-18 Matthias Klose <doko@ubuntu.com>
34624 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34626 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34628 * config/rs6000/altivec.h (vec_bperm): Change #define.
34629 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34630 (altivec_vbpermq2): New define_insn.
34631 (altivec_vbpermd): Likewise.
34632 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34633 function interface.
34634 (VBPERMD): Likewise.
34635 (VBPERM): New polymorphic function interface.
34636 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34637 Add entries for P9V_BUILTIN_VEC_VBPERM.
34638 * doc/extend.texi: Add interfaces for vec_bperm.
34640 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34642 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34643 first letter of error messages.
34644 (s390_resolve_overloaded_builtin): Likewise.
34645 * config/s390/s390.c (s390_expand_builtin): Likewise.
34646 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34647 (s390_valid_target_attribute_inner_p): Likewise.
34648 * config/s390/s390.md ("tabort"): Likewise.
34650 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34652 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34653 (ISA_AVOID_DIV_HILO): New macro.
34654 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34655 (ISA_HAS_DDIV): Likewise.
34657 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34659 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34661 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34663 * doc/invoke.texi (fabi-version): Spelling fix.
34665 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34668 * doc/invoke.texi (fabi-version): Mention mangling fix for
34671 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34674 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34676 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34679 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34680 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34682 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34684 (stack_protect_test): Ditto.
34685 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34686 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34688 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34689 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34690 -mstack-protector-guard-offset=.
34691 (RS/6000 and PowerPC Options): Ditto.
34693 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34695 * config/i386/i386.h (MASK_CLASS_P): New define.
34696 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34697 there are no registers from different register sets also when
34698 mask registers are used. Update function comment.
34699 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34700 to (*k/*r) and (*k/*km) alternatives.
34702 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34704 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34705 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34706 (EH_RETURN_HANDLER_RTX): New define.
34707 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34708 Force frame pointer in EH return functions.
34709 (aarch64_expand_epilogue): Add barrier for eh_return.
34710 (aarch64_final_eh_return_addr): Remove.
34711 (aarch64_eh_return_handler_rtx): New function.
34712 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34714 (aarch64_eh_return_handler_rtx): New prototype.
34716 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34718 * config/rs6000/altivec.h (vec_rlmi): New #define.
34719 (vec_vrlnm): Likewise.
34720 (vec_rlnm): Likewise.
34721 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34722 (UNSPEC_VRLNM): Likewise.
34723 (VIlong): New mode iterator.
34724 (altivec_vrl<VI_char>mi): New define_insn.
34725 (altivec_vrl<VI_char>nm): Likewise.
34726 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34728 (VRLDNM): Likewise.
34729 (RLNM): New polymorphic function entry.
34730 (VRLWMI): New monomorphic function entry.
34731 (VRLDMI): Likewise.
34732 (RLMI): New polymorphic function entry.
34733 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34734 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34735 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34738 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34741 * dwarf2out.c (field_byte_offset): Restore the
34742 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34743 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34744 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34747 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34750 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34752 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34755 * dwarf2out.c (add_data_member_location_attribute): For constant
34756 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34757 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34758 DW_AT_byte_size attributes.
34760 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34762 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34763 after forcing to constant memory when the code model is medium.
34765 2017-01-17 Julia Koval <julia.koval@intel.com>
34768 * config/i386/avx512fintrin.h
34769 (_mm512_i32gather_ps): Change __addr type to void const*.
34770 (_mm512_mask_i32gather_ps): Ditto.
34771 (_mm512_i32gather_pd): Ditto.
34772 (_mm512_mask_i32gather_pd): Ditto.
34773 (_mm512_i64gather_ps): Ditto.
34774 (_mm512_mask_i64gather_ps): Ditto.
34775 (_mm512_i64gather_pd): Ditto.
34776 (_mm512_mask_i64gather_pd): Ditto.
34777 (_mm512_i32gather_epi32): Ditto.
34778 (_mm512_mask_i32gather_epi32): Ditto.
34779 (_mm512_i32gather_epi64): Ditto.
34780 (_mm512_mask_i32gather_epi64): Ditto.
34781 (_mm512_i64gather_epi32): Ditto.
34782 (_mm512_mask_i64gather_epi32): Ditto.
34783 (_mm512_i64gather_epi64): Ditto.
34784 (_mm512_mask_i64gather_epi64): Ditto.
34785 (_mm512_i32scatter_ps): Change __addr type to void*.
34786 (_mm512_mask_i32scatter_ps): Ditto.
34787 (_mm512_i32scatter_pd): Ditto.
34788 (_mm512_mask_i32scatter_pd): Ditto.
34789 (_mm512_i64scatter_ps): Ditto.
34790 (_mm512_mask_i64scatter_ps): Ditto.
34791 (_mm512_i64scatter_pd): Ditto.
34792 (_mm512_mask_i64scatter_pd): Ditto.
34793 (_mm512_i32scatter_epi32): Ditto.
34794 (_mm512_mask_i32scatter_epi32): Ditto.
34795 (_mm512_i32scatter_epi64): Ditto.
34796 (_mm512_mask_i32scatter_epi64): Ditto.
34797 (_mm512_i64scatter_epi32): Ditto.
34798 (_mm512_mask_i64scatter_epi32): Ditto.
34799 (_mm512_i64scatter_epi64): Ditto.
34800 (_mm512_mask_i64scatter_epi64): Ditto.
34801 * config/i386/avx512pfintrin.h
34802 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34803 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34804 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34805 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34806 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34807 (_mm512_prefetch_i32scatter_ps): Ditto.
34808 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34809 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34810 (_mm512_prefetch_i64scatter_pd): Ditto.
34811 (_mm512_prefetch_i64scatter_ps): Ditto.
34812 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34813 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34814 * config/i386/avx512vlintrin.h
34815 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34816 (_mm_mmask_i32gather_ps): Ditto.
34817 (_mm256_mmask_i32gather_pd): Ditto.
34818 (_mm_mmask_i32gather_pd): Ditto.
34819 (_mm256_mmask_i64gather_ps): Ditto.
34820 (_mm_mmask_i64gather_ps): Ditto.
34821 (_mm256_mmask_i64gather_pd): Ditto.
34822 (_mm_mmask_i64gather_pd): Ditto.
34823 (_mm256_mmask_i32gather_epi32): Ditto.
34824 (_mm_mmask_i32gather_epi32): Ditto.
34825 (_mm256_mmask_i32gather_epi64): Ditto.
34826 (_mm_mmask_i32gather_epi64): Ditto.
34827 (_mm256_mmask_i64gather_epi32): Ditto.
34828 (_mm_mmask_i64gather_epi32): Ditto.
34829 (_mm256_mmask_i64gather_epi64): Ditto.
34830 (_mm_mmask_i64gather_epi64): Ditto.
34831 (_mm256_i32scatter_ps): Change __addr type to void*.
34832 (_mm256_mask_i32scatter_ps): Ditto.
34833 (_mm_i32scatter_ps): Ditto.
34834 (_mm_mask_i32scatter_ps): Ditto.
34835 (_mm256_i32scatter_pd): Ditto.
34836 (_mm256_mask_i32scatter_pd): Ditto.
34837 (_mm_i32scatter_pd): Ditto.
34838 (_mm_mask_i32scatter_pd): Ditto.
34839 (_mm256_i64scatter_ps): Ditto.
34840 (_mm256_mask_i64scatter_ps): Ditto.
34841 (_mm_i64scatter_ps): Ditto.
34842 (_mm_mask_i64scatter_ps): Ditto.
34843 (_mm256_i64scatter_pd): Ditto.
34844 (_mm256_mask_i64scatter_pd): Ditto.
34845 (_mm_i64scatter_pd): Ditto.
34846 (_mm_mask_i64scatter_pd): Ditto.
34847 (_mm256_i32scatter_epi32): Ditto.
34848 (_mm256_mask_i32scatter_epi32): Ditto.
34849 (_mm_i32scatter_epi32): Ditto.
34850 (_mm_mask_i32scatter_epi32): Ditto.
34851 (_mm256_i32scatter_epi64): Ditto.
34852 (_mm256_mask_i32scatter_epi64): Ditto.
34853 (_mm_i32scatter_epi64): Ditto.
34854 (_mm_mask_i32scatter_epi64): Ditto.
34855 (_mm256_i64scatter_epi32): Ditto.
34856 (_mm256_mask_i64scatter_epi32): Ditto.
34857 (_mm_i64scatter_epi32): Ditto.
34858 (_mm_mask_i64scatter_epi32): Ditto.
34859 (_mm256_i64scatter_epi64): Ditto.
34860 (_mm256_mask_i64scatter_epi64): Ditto.
34861 (_mm_i64scatter_epi64): Ditto.
34862 (_mm_mask_i64scatter_epi64): Ditto.
34863 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34864 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34865 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34866 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34867 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34868 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34869 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34870 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34871 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34872 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34873 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34874 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34875 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34876 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34877 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34878 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34879 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34880 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34881 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34882 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34883 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34884 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34885 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34886 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34887 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34888 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34889 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34890 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34891 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34892 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34893 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34894 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34895 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34896 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34897 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34898 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34899 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34900 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34901 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34902 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34903 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34904 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34905 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34906 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34907 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34908 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34909 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34910 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34911 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34912 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34913 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34914 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34915 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34916 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34917 definitions accordingly.
34919 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34920 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34923 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34926 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34929 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34930 hard regs for inner regno.
34932 2017-01-17 Martin Liska <mliska@suse.cz>
34935 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34936 assumption and add comment.
34938 2017-01-17 Nathan Sidwell <nathan@acm.org>
34940 * ipa-visibility.c (localize_node): New function, broken out of ...
34941 (function_and_variable_visibility): ... here. Call it.
34943 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34945 PR middle-end/77445
34946 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34947 correctly set frequency of oudgoing edge.
34948 (duplicate_thread_path): Fix profile updating.
34950 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34953 * configure.ac: Add GCC_BASE_VER.
34954 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34955 version from BASE-VER file.
34956 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34957 (gcc.o): Depend on $(BASEVER).
34958 * common.opt (dumpfullversion): New option.
34959 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34960 * doc/invoke.texi: Document -dumpfullversion.
34961 * doc/install.texi: Document --with-gcc-major-version-only.
34962 * configure: Regenerated.
34964 2017-01-17 Richard Biener <rguenther@suse.de>
34966 PR tree-optimization/71433
34967 * tree-vrp.c (register_new_assert_for): Merge same asserts
34968 on all incoming edges.
34969 (process_assert_insertions_for): Handle insertions at the
34972 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34974 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34975 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34977 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34980 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34983 2017-01-17 Alan Modra <amodra@gmail.com>
34986 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34987 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34988 symbolic stack limit when pic.
34990 2017-01-16 Martin Sebor <msebor@redhat.com>
34992 PR tree-optimization/78608
34993 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34995 2017-01-16 Jeff Law <law@redhat.com>
34998 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34999 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35000 for several include directories that may be relative to sysroot.
35001 * config/i386/x-mingw32 (gplus_includedir): Define.
35002 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35003 (native_system_includedir): Likewise.
35004 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35005 override if TARGET_SYSTEM_ROOT is defined.
35006 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35008 PR tree-optimization/79090
35009 PR tree-optimization/33562
35010 PR tree-optimization/61912
35011 PR tree-optimization/77485
35012 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35013 and computed trims into the dump file.
35015 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35017 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35019 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35022 * gimplify.c (gimplify_init_constructor): If want_value and
35023 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35027 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35028 sequence. Formatting fixes.
35029 (doloop_optimize): Formatting fixes.
35032 * gcc.c (debug_level_greater_than_spec_func): New function.
35033 (static_spec_functions): Add debug-level-gt spec function.
35034 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35036 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35037 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35038 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35039 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35040 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35041 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35043 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35045 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35046 QImode fixups to general and mask registers only.
35048 2017-01-16 Carl Love <cel@us.ibm.com>
35050 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35051 for built-in functions
35052 vector signed char vec_nabs (vector signed char)
35053 vector signed short vec_nabs (vector signed short)
35054 vector signed int vec_nabs (vector signed int)
35055 vector signed long long vec_nabs (vector signed long long)
35056 vector float vec_nabs (vector float)
35057 vector double vec_nabs (vector double)
35058 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35060 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35061 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35062 * doc/extend.texi: Update the documentation file for the new built-in
35065 2017-01-16 Martin Sebor <msebor@redhat.com>
35067 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35070 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35072 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35073 UNSPEC_VSX__XXSPLTD to require special splat handling.
35075 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35078 * system.h: Poison strndup.
35080 2017-01-16 Alan Modra <amodra@gmail.com>
35083 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35086 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35088 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35090 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35092 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35093 call recog here. Assert that INSN_CODE (insn) is non-negative.
35095 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35098 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35100 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35101 in the currently scheduled RTL fragment.
35103 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35105 PR rtl-optimization/78751
35106 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35109 2017-01-14 Jeff Law <law@redhat.com>
35111 PR tree-optimization/79090
35112 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35113 variable length stores.
35114 (compute_trims): Delete dead assignment to *trim_tail.
35115 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35118 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35120 PR rtl-optimization/78626
35121 PR rtl-optimization/78727
35122 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35123 of a block, and split such blocks after everything else is finished.
35125 2017-01-14 Alan Modra <amodra@gmail.com>
35128 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35129 target legitimate_combined_insn.
35130 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35131 (rs6000_legitimate_combined_insn): New function.
35132 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35134 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
35135 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
35136 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
35138 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
35140 * doc/frontends.texi (G++ and GCC): Remove references to Java.
35142 2017-01-13 Jeff Law <law@redhat.com>
35144 PR tree-optimization/33562
35145 PR tree-optimization/61912
35146 PR tree-optimization/77485
35147 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
35149 (delete_dead_assignment): Likewise.
35150 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
35151 statement to delete_dead_call and delete_dead_assignment.
35153 2017-01-13 David Malcolm <dmalcolm@redhat.com>
35156 * substring-locations.c (format_warning_va): Strengthen case 1 so
35157 that both endpoints of the substring must be within the format
35158 range for just the substring to be printed.
35160 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
35162 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
35163 * config/i386/i386.c (ix86_target_string): Add missing options
35164 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
35165 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
35166 flags_other and ix86_target_other to flags2_other. Display unknown
35168 (ix86_valid_target_attribute_inner_p): Add missing options and
35169 reorder options by implied ISAs, as in ix86_target_string.
35171 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35173 * hash-table.h (hash_table::too_empty_p): New function.
35174 (hash_table::expand): Use it.
35175 (hash_table::traverse): Likewise.
35176 (hash_table::empty_slot): Use sizeof (value_type) instead of
35177 sizeof (PTR) to convert bytes to elements. Shrink the table
35178 if the current size is excessive for the current number of
35181 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35183 * ira-costs.c (record_reg_classes): Break from the inner loop
35184 early once alt_fail is known to be true. Update outer loop
35185 handling accordingly.
35187 2017-01-13 Jeff Law <law@redhat.com>
35189 * tree-ssa-dse.c (decrement_count): New function.
35190 (increment_start_addr, maybe_trim_memstar_call): Likewise.
35191 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
35192 when we know the partially dead statement is a mem* function.
35194 PR tree-optimization/61912
35195 PR tree-optimization/77485
35196 * tree-ssa-dse.c: Include expr.h.
35197 (maybe_trim_constructor_store): New function.
35198 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
35200 PR tree-optimization/33562
35201 PR tree-optimization/61912
35202 PR tree-optimization/77485
35203 * doc/invoke.texi: Document new dse-max-object-size param.
35204 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
35205 * tree-ssa-dse.c: Include params.h.
35206 (dse_store_status): New enum.
35207 (initialize_ao_ref_for_dse): New, partially extracted from
35209 (valid_ao_ref_for_dse, normalize_ref): New.
35210 (setup_live_bytes_from_ref, compute_trims): Likewise.
35211 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
35212 (maybe_trim_partially_dead_store): Likewise.
35213 (maybe_trim_complex_store): Likewise.
35214 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
35215 Track what bytes live from the original store. Return tri-state
35216 for dead, partially dead or live.
35217 (dse_dom_walker): Add constructor, destructor and new private members.
35218 (delete_dead_call, delete_dead_assignment): New extracted from
35220 (dse_optimize_stmt): Make a member of dse_dom_walker.
35221 Use initialize_ao_ref_for_dse.
35223 PR tree-optimization/33562
35224 PR tree-optimization/61912
35225 PR tree-optimization/77485
35226 * sbitmap.h (bitmap_count_bits): Prototype.
35227 (bitmap_clear_range, bitmap_set_range): Likewise.
35228 * sbitmap.c (bitmap_clear_range): New function.
35229 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
35231 2017-01-13 Martin Liska <mliska@suse.cz>
35234 * function.c (set_cfun): Add new argument force.
35235 * function.h (set_cfun): Likewise.
35236 * ipa-inline-transform.c (inline_call): Use the function when
35237 strict alising from is dropped for function we inline to.
35239 2017-01-13 Richard Biener <rguenther@suse.de>
35241 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35242 for dumping GIMPLE INTEGER_CSTs.
35244 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35246 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35247 to 201112L since C++17.
35249 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35252 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35253 if -fsanitize=kernel-address is present.
35255 2017-01-13 Richard Biener <rguenther@suse.de>
35257 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35258 as _Literal ( type ) number in case usual suffixes do not
35259 preserve all information.
35261 2017-01-13 Richard Biener <rguenther@suse.de>
35263 PR tree-optimization/77283
35264 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35265 and ssa-iterators.h.
35266 (is_feasible_trace): Implement a cost model based on joiner
35269 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35272 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35273 char or short to __float128/_Float128 directly.
35275 2017-01-12 Martin Sebor <msebor@redhat.com>
35277 to -Wformat-overflow.
35278 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35279 (min_bytes_remaining): Same.
35280 (get_string_length): Same.
35281 (format_string): Same.
35282 (format_directive): Same.
35284 (pass_sprintf_length::handle_gimple_call): Same.
35286 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35288 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35289 info.nowrite calls with no lhs that can't throw. Return bool
35290 whether gsi_remove has been called or not.
35291 (pass_sprintf_length::handle_gimple_call): Return bool whether
35292 try_substitute_return_value called gsi_remove. Formatting fix.
35293 (pass_sprintf_length::execute): Don't use gsi_remove if
35294 handle_gimple_call returned true.
35297 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35298 be removed due to side-effects, don't remove following barrier nor
35299 turn the successor edge into fallthru edge.
35301 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35304 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35305 element-reversing loads and stores as not swappable.
35307 2017-01-12 Nathan Sidwell <nathan@acm.org>
35308 Nicolai Stange <nicstange@gmail.com>
35310 * combine.c (try_combine): Don't ignore result of overlap checking
35311 loop. Combine overlap & asm check into single loop.
35313 2017-01-12 Richard Biener <rguenther@suse.de>
35315 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35316 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35318 2017-01-12 Richard Biener <rguenther@suse.de>
35320 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35321 and TS_TARGET_OPTION directly derive from TS_BASE.
35322 * tree-core.h (tree_optimization_option): Derive from tree_base.
35323 (tree_target_option): Likewise.
35325 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35327 * config/i386/i386.c (memory_address_length): Increase len
35328 only when rip_relative_addr_p returns false.
35330 2017-01-11 Julia Koval <julia.koval@intel.com>
35332 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35333 (OPTION_MASK_ISA_SGX_SET): New.
35334 (ix86_handle_option): Handle OPT_msgx.
35335 * config.gcc: Added sgxintrin.h.
35336 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35337 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35338 * config/i386/i386.c (ix86_target_string): Add -msgx.
35340 (ix86_option_override_internal): Handle new options.
35341 (ix86_valid_target_attribute_inner_p): Add sgx.
35342 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35343 * config/i386/i386.opt: Add msgx.
35344 * config/i386/sgxintrin.h: New file.
35345 * config/i386/x86intrin.h: Add sgxintrin.h.
35347 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35350 * fold-const.c (maybe_nonzero_address): Return 1 for function
35352 (tree_single_nonzero_warnv_p): Don't handle function local objects
35356 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35359 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35362 * opts.c: Include "spellcheck.h"
35363 (struct string_fragment): New struct.
35364 (struct edit_distance_traits<const string_fragment &>): New
35366 (get_closest_sanitizer_option): New function.
35367 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35369 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35371 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35373 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35374 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35375 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35376 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35377 for initial die_offset if dwarf_split_debug_info.
35378 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35379 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35381 (output_skeleton_debug_sections): Formatting fix. Use
35382 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35383 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35385 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35387 * config/arm/cortex-a53.md: Add bypasses for
35388 cortex_a53_r2f_cvt.
35389 (cortex_a53_r2f): Only use for transfers.
35390 (cortex_a53_f2r): Likewise.
35391 (cortex_a53_r2f_cvt): Add reservation for conversions.
35392 (cortex_a53_f2r_cvt): Likewise.
35394 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35396 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35397 to all inlined functions, change static to extern.
35399 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35402 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35404 (arm_assemble_integer): Likewise.
35406 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35408 * config.gcc: Use new awk script to check CPU, FPU and architecture
35409 parameters for --with-... options.
35410 * config/arm/parsecpu.awk: New file
35411 * config/arm/arm-cpus.in: New file.
35412 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35414 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35416 * config/arm/t-arm: Update dependency rules.
35417 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35418 of processing .def files.
35419 * config/arm/genopt.sh: Deleted.
35420 * config/arm/gentune.sh: Deleted.
35421 * config/arm/arm-cores.def: Deleted.
35422 * config/arm/arm-arches.def: Deleted.
35423 * config/arm/arm-fpus.def: Deleted.
35424 * config/arm/arm-tune.md: Regenerated.
35425 * config/arm/arm-tables.opt: Regenerated.
35426 * config/arm/arm-cpu.h: New generated file.
35427 * config/arm/arm-cpu-data.h: New generated file.
35428 * config/arm/arm-cpu-cdata.h: New generated file.
35430 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35433 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35435 (input_varpool_node): Unpack dynamically_initialized bit.
35437 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
35439 PR rtl-optimization/79032
35440 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
35441 the alignment of the adjusted memory reference against that of MODE,
35442 instead of the alignment of the original memory reference.
35444 2017-01-11 Martin Jambor <mjambor@suse.cz>
35446 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
35448 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
35449 decorated functions.
35451 2017-01-11 Richard Biener <rguenther@suse.de>
35453 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
35454 set range/nonnull info for PHI results. Do not set it on
35455 stmts marked for removal.
35457 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
35459 * expr.c (store_field): In the bitfield case, fetch the return value
35460 from the registers before applying a single big-endian adjustment.
35461 Always do a final load for a BLKmode value not larger than a word.
35463 2017-01-10 David Malcolm <dmalcolm@redhat.com>
35466 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35467 that we correctly handle column numbers greater than
35468 LINE_MAP_MAX_COLUMN_NUMBER.
35470 2017-01-10 Martin Sebor <msebor@redhat.com>
35472 PR middle-end/78245
35473 * gimple-ssa-sprintf.c (get_destination_size): Call
35474 {init,fini}object_sizes.
35475 * tree-object-size.c (addr_object_size): Adjust.
35476 (pass_through_call): Adjust.
35477 (pass_object_sizes::execute): Adjust.
35478 * tree-object-size.h (fini_object_sizes): Declare.
35480 2017-01-10 Martin Sebor <msebor@redhat.com>
35482 PR tree-optimization/78775
35483 * builtins.c (get_size_range): Move...
35484 * calls.c: ...to here.
35485 (alloc_max_size): Accept zero argument.
35486 (operand_signed_p): Remove.
35487 (maybe_warn_alloc_args_overflow): Call get_size_range.
35488 * calls.h (get_size_range): Declare.
35490 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
35492 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
35493 from TI's devices.csv file as of September 2016.
35494 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
35496 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
35498 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
35499 * doc/invoke.texi: Likewise.
35500 * doc/md.texi: Likewise.
35501 * doc/objc.texi: Likewise.
35503 2017-01-10 Joshua Conner <joshconner@google.com>
35505 * config/arm/fuchsia-elf.h: New file.
35506 * config/fuchsia.h: New file.
35507 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
35508 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
35510 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
35512 2016-01-10 Richard Biener <rguenther@suse.de>
35514 PR tree-optimization/79034
35515 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35516 Propagate out degenerate PHIs in the joiner.
35518 2017-01-10 Martin Liska <mliska@suse.cz>
35520 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
35521 (sort_congruence_classes_by_decl_uid): Likewise.
35522 (sort_congruence_class_groups_by_decl_uid): Likewise.
35523 (sem_item_optimizer::merge_classes): Sort class, groups in these
35524 classes and members in the groups by DECL_UID of declarations.
35525 This would make merge operations stable.
35527 2017-01-10 Martin Liska <mliska@suse.cz>
35529 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
35530 usage of m_classes_vec.
35531 (sem_item_optimizer::~sem_item_optimizer): Likewise.
35532 (sem_item_optimizer::get_group_by_hash): Likewise.
35533 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
35534 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
35535 (sem_item_optimizer::verify_classes): Likewise.
35536 (sem_item_optimizer::process_cong_reduction): Likewise.
35537 (sem_item_optimizer::dump_cong_classes): Likewise.
35538 (sem_item_optimizer::merge_classes): Likewise.
35539 * ipa-icf.h (congruence_class_hash): Rename from
35540 congruence_class_group_hash. Remove declaration of m_classes_vec.
35542 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
35544 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
35545 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
35546 * config.gcc: Add avx512vpopcntdqintrin.h.
35547 * config/i386/avx512vpopcntdqintrin.h: New.
35548 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
35549 * config/i386/i386-builtin-types.def: Add new types.
35550 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
35551 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
35552 __builtin_ia32_vpopcountq_v8di_mask): New.
35553 * config/i386/i386-c.c (ix86_target_macros_internal): Define
35554 __AVX512VPOPCNTDQ__.
35555 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
35556 (PTA_AVX512VPOPCNTDQ): Define.
35557 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
35558 TARGET_AVX512VPOPCNTDQ_P): Define.
35559 * config/i386/i386.opt: Add mavx512vpopcntdq.
35560 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
35561 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
35563 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35565 PR middle-end/77484
35566 * predict.def (PRED_CALL): Set to 67.
35568 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
35570 * expr.c (store_field): In the bitfield case, if the value comes from
35571 a function call and is of an aggregate type returned in registers, do
35572 not modify the field mode; extract the value in all cases if the mode
35573 is BLKmode and the size is not larger than a word.
35575 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35578 * config/i386/cpuid.h: Fix undefined behavior.
35580 2017-01-04 Jeff Law <law@redhat.com>
35582 PR tree-optimization/79007
35583 PR tree-optimization/67955
35584 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35585 conservative for pt.null when flag_non_call_exceptions is on.
35587 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35589 PR translation/79019
35590 PR translation/79020
35591 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35592 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35593 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35595 * config/avr/avr.opt (maccumulate-args): Likewise.
35596 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35597 * common.opt (freport-bug): Likewise.
35598 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35599 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35600 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35601 translatable string.
35602 * config/i386/i386.c (function_value_32): Likewise.
35603 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35604 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35606 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35607 * common/config/msp430/msp430-common.c (msp430_handle_option):
35609 * symtab.c (symtab_node::verify_base): Likewise.
35610 * opts.c (set_debug_level): Likewise.
35611 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35612 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35613 missing whitespace to translatable strings.
35614 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35615 * config/sh/superh.h: Likewise.
35616 * config/i386/xopintrin.h: Likewise.
35617 * config/i386/znver1.md: Likewise.
35618 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35619 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35620 * double-int.h (struct double_int): Likewise.
35621 * double-int.c (div_and_round_double): Likewise.
35622 * wide-int.cc: Likewise.
35623 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35624 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35625 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35626 (crossjumps_occurred): ... this.
35627 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35630 PR tree-optimization/78899
35631 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35632 returning bool return struct loop *, NULL for failure and the new
35634 (versionable_outer_loop_p): Don't version outer loop if it has
35635 dont_vectorized bit set.
35636 (tree_if_conversion): When versioning outer loop, ensure
35637 tree_if_conversion is performed also on the inner loop of the
35638 non-vectorizable outer loop copy.
35639 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35640 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35641 prevent vectorization of it.
35642 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35643 the outer loop vectorization of the non-scalar version is attempted
35644 before vectorization of the inner loop in scalar version. If
35645 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35646 vectorization of its inner loop.
35647 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35648 has 2 inner loops, rename also on edges from bb whose single pred
35649 is outer_loop->header. Fix typo in function comment.
35651 2017-01-09 Martin Sebor <msebor@redhat.com>
35654 * asan.c (asan_emit_stack_protection): Increase local buffer size
35655 to avoid snprintf truncation warning.
35657 2017-01-09 Andrew Pinski <apinski@cavium.com>
35659 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35660 to reference thunderx2t99 for the tuning structure
35661 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35663 (thunderx2t99_extra_costs): This.
35664 * config/aarch64/aarch64-tune.md: Regenerate.
35665 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35666 (vulcan_addrcost_table): This.
35667 (vulcan_regmove_cost): Rename to ...
35668 (thunderx2t99_regmove_cost): This.
35669 (vulcan_vector_cost): Rename to ...
35670 (thunderx2t99_vector_cost): this.
35671 (vulcan_branch_cost): Rename to ...
35672 (thunderx2t99_branch_cost): This.
35673 (vulcan_tunings): Rename to ...
35674 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35675 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35677 2017-01-09 Martin Jambor <mjambor@suse.cz>
35681 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35682 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35683 (propagate_vr_accross_jump_function): Use the above function for all
35684 value range computations for pass-through jump functions and type
35685 converasion from explicit value range values.
35686 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35687 parameters from TYPE_ARG_TYPES.
35688 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35689 (ipa_write_node_info): Stream type of the actual argument.
35690 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35692 2017-01-09 Martin Liska <mliska@suse.cz>
35695 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35696 (lookup_compiler): Do not show error message with have_E.
35698 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35700 PR tree-optimization/78938
35701 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35702 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35703 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35704 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35707 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35709 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35712 2017-01-09 Richard Biener <rguenther@suse.de>
35714 PR tree-optimization/78997
35715 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35716 name condition properly.
35718 2017-01-09 Richard Biener <rguenther@suse.de>
35721 * dwarf2out.c (is_cxx): New overload with context.
35722 (is_naming_typedef_decl): Use it.
35724 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35726 * invoke.texi (Option Summary): Correct spacing in option lists
35727 and add line breaks to fix over-long lines.
35729 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35731 PR middle-end/17660
35733 * extend.texi (Common Variable Attributes): Add xref to GCC
35734 Internals manual to explain mode attribute keywords.
35736 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35739 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35740 and Preprocessor Options.
35741 (Options for Linking): Document -pthread here....
35742 (RS/6000 and PowerPC Options): ...not here.
35743 (Solaris 2 Options): ...or here.
35744 * doc/cppopts.texi: Document -pthread.
35746 2017-01-08 Martin Sebor <msebor@redhat.com>
35748 PR middle-end/77708
35749 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35750 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35751 New member functions.
35752 (format_directive): Used them.
35754 (pass_sprintf_length::handle_gimple_call): Same.
35755 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35756 to avoid truncation for any argument.
35757 (extract_affine_mul): Same.
35758 * tree.c (get_file_function_name): Same.
35760 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35762 PR middle-end/77484
35763 * predict.def (PRED_INDIR_CALL): Set to 86.
35765 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35767 PR preprocessor/54124
35768 * doc/cppopts.texi: Reformat -d subtable to list the full name
35769 of the options. Add cross-reference to the docs for the general
35770 compiler -d options.
35771 * doc/invoke.texi (Developer Options): Add cross-reference to the
35772 preprocessor-specific -d option documentation.
35774 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35776 PR preprocessor/13498
35777 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35778 redudant material, and reflect new command-line options.
35779 (System Headers): Likewise.
35781 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35783 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35784 -isystem, and -idirafter. Copy-edit.
35785 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35786 default for -ftrack-macro-expansion. Delete obsolete and
35787 badly-formatted implementation details about -fdebug-cpp output.
35788 * doc/cppwarnopts.texi: Copy-edit.
35790 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35793 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35794 that the transition from a max line width >= 1<<10 to narrower
35795 lines works correctly.
35797 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35799 * doc/options.texi (PerFunction): New.
35800 * opt-functions.awk (switch_flags): Map both Optimization and
35801 PerFunction to CL_OPTIMIZATION.
35802 * opth-gen.awk: Test for PerFunction flag along with
35804 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35805 it only when the latter is present. Skip those that don't in
35806 the hash function generator.
35807 * common.opt (fvar-tracking): Mark as PerFunction instead of
35809 (fvar-tracking-assignments): Likewise.
35810 (fvar-tracking-assignments-toggle): Likewise.
35811 (fvar-tracking-uninit): Likewise.
35813 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35815 PR translation/79018
35816 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35819 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35822 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35823 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35824 TARGET_LONG_JUMP_TABLE_OFFSETS.
35825 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35826 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35827 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35828 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35829 * config/m68k/m68k.md (tablejump expander): Likewise.
35830 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35831 TARGET_LONG_JUMP_TABLE_OFFSETS.
35832 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35833 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35835 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35836 David Holsgrove <david.holsgrove@xilinx.com>
35838 * common/config/microblaze/microblaze-common.c
35839 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35840 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35842 * config/microblaze/microblaze.c (microblaze_must_save_register)
35843 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35845 (microblaze_eh_return): New function.
35846 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35847 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35848 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35849 * config/microblaze/microblaze.md (eh_return): New pattern.
35851 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35853 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35854 GCC_DIAGNOSTIC_STRINGIFY): Define.
35856 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35858 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35860 * config/arm/arm.md (<mcrr>): New.
35862 * config/arm/arm.c (arm_arch5te): New.
35863 (arm_option_override): Set arm_arch5te.
35864 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35866 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35867 (arm_mcrr_qualifiers): ... this. New.
35868 (MRRC_QUALIFIERS): Define to...
35869 (arm_mrrc_qualifiers): ... this. New.
35870 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35872 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35873 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35874 (MRRCI, mrrc, MRRC): New.
35875 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35876 VUNSPEC_MRRC2): New.
35878 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35880 * config/arm/arm.md (<mcr>): New.
35882 * config/arm/arm.c (arm_coproc_builtin_available): Add
35883 support for mcr, mrc, mcr2 and mrc2.
35884 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35885 (arm_mcr_qualifiers): ... this. New.
35886 (MRC_QUALIFIERS): Define to ...
35887 (arm_mrc_qualifiers): ... this. New.
35888 (MCR_QUALIFIERS): Define to ...
35889 (arm_mcr_qualifiers): ... this. New.
35890 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35892 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35893 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35894 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35895 VUNSPEC_MRC2): New.
35897 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35899 * config/arm/arm.md (*ldc): New.
35903 * config/arm/arm.c (arm_coproc_builtin_available): Add
35904 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35905 (arm_coproc_ldc_stc_legitimate_address): New.
35906 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35907 'qualifier_const_pointer'.
35908 (LDC_QUALIFIERS): Define to...
35909 (arm_ldc_qualifiers): ... this. New.
35910 (STC_QUALIFIERS): Define to...
35911 (arm_stc_qualifiers): ... this. New.
35912 * config/arm/arm-protos.h
35913 (arm_coproc_ldc_stc_legitimate_address): New.
35914 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35915 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35916 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35917 stc2, stcl, stc2l): New.
35918 * config/arm/constraints.md (Uz): New.
35919 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35920 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35921 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35922 VUNSPEC_STC2L): New.
35924 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35926 * config/arm/arm.md (<cdp>): New.
35927 * config/arm/arm.c (neon_const_bounds): Rename this ...
35928 (arm_const_bounds): ... this.
35929 (arm_coproc_builtin_available): New.
35930 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35931 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35932 (CDP_QUALIFIERS): Define to...
35933 (arm_cdp_qualifiers): ... this. New.
35935 (arm_expand_builtin_args): Add case for 6 arguments.
35936 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35937 (arm_const_bounds): ... this.
35938 (arm_coproc_builtin_available): New.
35939 * config/arm/arm_acle.h (__arm_cdp): New.
35941 * config/arm/arm_acle_builtins.def (cdp): New.
35943 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35944 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35945 'arm_const_bounds'.
35946 * config/arm/types.md (coproc): New.
35947 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35948 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35949 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35950 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35952 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35954 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35955 (UBINOP_QUALIFIERS): New.
35957 (acle_builtin_data): New. Change comment.
35958 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35959 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35960 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35961 arm_acle_builtins.def.
35962 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35963 (arm_init_acle_builtins): New.
35964 (CRC32_BUILTIN): Remove.
35965 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35966 crc32cb, crc32ch and crc32cw.
35967 (arm_init_crc32_builtins): Remove.
35968 (arm_init_builtins): Use arm_init_acle_builtins rather
35969 than arm_init_crc32_builtins.
35970 (arm_expand_acle_builtin): New.
35971 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35972 * config/arm/arm_acle_builtins.def: New.
35974 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35976 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35977 (arm_builtin_datum): ... this.
35978 (arm_init_neon_builtin): Rename to ...
35979 (arm_init_builtin): ... this. Add a new parameters PREFIX
35980 and USE_SIG_IN_NAME.
35981 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35982 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35983 'arm_builtin_datum'.
35984 (arm_init_vfp_builtins): Likewise.
35985 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35986 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35987 (arm_expand_neon_args): Rename to ...
35988 (arm_expand_builtin_args): ... this. Rename builtin_arg
35989 enum values and differentiate between ARG_BUILTIN_MEMORY
35990 and ARG_BUILTIN_NEON_MEMORY.
35991 (arm_expand_neon_builtin_1): Rename to ...
35992 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35993 values, arm_expand_builtin_args and add bool parameter NEON.
35994 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35995 (arm_expand_vfp_builtin): Likewise.
35996 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35998 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36000 PR middle-end/77484
36001 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36002 * predict.c (tree_estimate_probability_bb): Reverse direction of
36003 polymorphic call predictor.
36005 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36007 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36008 (determine_pass_name_match): ...this new function and...
36009 (should_skip_pass_p): ...this new function.
36011 2017-01-06 Nathan Sidwell <nathan@acm.org>
36013 * ipa-visibility.c (function_and_variable_visibility): Reformat
36014 comments and long lines. Remove extrneous if.
36015 * symtab.c (symtab_node::make_decl_local): Fix code format.
36016 (symtab_node::set_section_for_node): Fix comment typo.
36018 2017-01-06 Martin Liska <mliska@suse.cz>
36021 * lra-constraints.c: Rename invariant to lra_invariant.
36022 * predict.c (set_even_probabilities): Initialize e to NULL.
36024 2017-01-05 Martin Sebor <msebor@redhat.com>
36026 PR tree-optimization/78910
36027 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36028 (format_integer): Correct off-by-one error in the handling
36029 of precision with negative numbers in signed conversions..
36031 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36033 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36035 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36037 PR tree-optimization/71016
36038 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36039 factor_out_conditional_conversion. Formatting fix.
36040 (factor_out_conditional_conversion): Add cond_stmt argument.
36041 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36042 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36045 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36047 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36048 read-rtl-function.o, and selftest-rtl.o.
36049 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36050 (selftest::aarch64_test_loading_full_dump): New function.
36051 (selftest::aarch64_run_selftests): New function.
36052 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36053 selftest::aarch64_run_selftests.
36054 * config/i386/i386.c
36055 (selftest::ix86_test_loading_dump_fragment_1): New function.
36056 (selftest::ix86_test_loading_call_insn): New function.
36057 (selftest::ix86_test_loading_full_dump): New function.
36058 (selftest::ix86_test_loading_unspec): New function.
36059 (selftest::ix86_run_selftests): Call the new functions.
36060 * emit-rtl.c (maybe_set_max_label_num): New function.
36061 * emit-rtl.h (maybe_set_max_label_num): New decl.
36062 * function.c (instantiate_decls): Guard call to
36063 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36064 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36066 * gensupport.c (gen_reader::gen_reader): Pass "false"
36067 for new "compact" param of rtx_reader.
36068 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36069 rather than an empty string for NULL strings.
36070 * read-md.c: Potentially include config.h rather than bconfig.h.
36071 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36072 (have_error): New global, copied from errors.c.
36073 (md_reader::read_name): Rename to...
36074 (md_reader::read_name_1): ...this, adding "out_loc" param,
36075 and converting "missing name or number" to returning false, rather
36077 (md_reader::read_name): Reimplement in terms of read_name_1.
36078 (md_reader::read_name_or_nil): New function.
36079 (md_reader::read_string): Handle "(nil)" by returning NULL.
36080 (md_reader::md_reader): Add new param "compact".
36081 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36082 (md_reader::read_file): New method.
36083 * read-md.h (md_reader::md_reader): Add new param "compact".
36084 (md_reader::read_file): New method.
36085 (md_reader::is_compact): New accessor.
36086 (md_reader::read_name): Convert return type from void to file_location.
36087 (md_reader::read_name_or_nil): New decl.
36088 (md_reader::read_name_1): New decl.
36089 (md_reader::m_compact): New field.
36090 (noop_reader::noop_reader): Pass "false" for new "compact" param
36092 (rtx_reader::rtx_reader): Add new "compact" param.
36093 (rtx_reader::read_rtx_operand): Make virtual and convert return
36094 type from void to rtx.
36095 (rtx_reader::read_until): New decl.
36096 (rtx_reader::handle_any_trailing_information): New virtual function.
36097 (rtx_reader::postprocess): New virtual function.
36098 (rtx_reader::finalize_string): New virtual function.
36099 (rtx_reader::m_in_call_function_usage): New field.
36100 (rtx_reader::m_reuse_rtx_by_id): New field.
36101 * read-rtl-function.c: New file.
36102 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36103 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36104 (selftest::verify_three_block_rtl_cfg): New decl.
36105 * read-rtl-function.h: New file.
36106 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36107 For host, include function.h, memmodel.h, and emit-rtl.h.
36108 (one_time_initialization): New function.
36109 (struct compact_insn_name): New struct.
36110 (compact_insn_names): New array.
36111 (find_code): Handle insn codes in compact dumps.
36112 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36113 (bind_subst_iter_and_attr): Likewise.
36114 (add_condition_to_string): Likewise.
36115 (add_condition_to_rtx): Likewise.
36116 (apply_attribute_uses): Likewise.
36117 (add_current_iterators): Likewise.
36118 (apply_iterators): Likewise.
36119 (initialize_iterators): Guard usage of apply_subst_iterator with
36120 #ifdef GENERATOR_FILE.
36121 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36122 (md_reader::read_mapping): Likewise.
36123 (add_define_attr_for_define_subst): Likewise.
36124 (add_define_subst_attr): Likewise.
36125 (read_subst_mapping): Likewise.
36126 (check_code_iterator): Likewise.
36127 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36129 (one_time_initialization): New function.
36130 (rtx_reader::read_until): New method.
36131 (read_flags): New function.
36132 (parse_reg_note_name): New function.
36133 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36134 Handle reuse_rtx ids.
36135 Wrap iterator lookup within #ifdef GENERATOR_FILE.
36136 Add parsing support for RTL dumps, mirroring the special-cases in
36137 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
36138 values, and calling handle_any_trailing_information.
36139 (rtx_reader::read_rtx_operand): Convert return type from void
36140 to rtx, returning return_rtx. Handle case 'e'. Call
36141 finalize_string on XSTR and XTMPL fields.
36142 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
36143 "(nil)" values were omitted. Call the postprocess vfunc on the
36145 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
36146 class ctor. Initialize m_in_call_function_usage. Call
36147 one_time_initialization.
36148 * rtl-tests.c (selftest::test_uncond_jump): Call
36149 set_new_first_and_last_insn.
36150 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
36151 * selftest-rtl.c: New file.
36152 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
36153 (selftest::get_insn_by_uid): New decl.
36154 * selftest-run-tests.c (selftest::run_tests): Call
36155 read_rtl_function_c_tests.
36156 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
36157 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
36160 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
36162 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
36163 operands in a special way. Assert that pos+len <= mode precision.
36165 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36167 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
36168 3 argument Alias with unlimited for the negative form.
36169 (fno-vect-cost-model): Removed.
36171 2017-01-05 Martin Liska <mliska@suse.cz>
36173 * hsa-gen.c (gen_hsa_divmod): New function.
36174 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
36176 2017-01-05 Martin Liska <mliska@suse.cz>
36179 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
36182 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36184 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
36185 small constant length operands.
36187 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36189 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
36190 between loop iterations.
36192 2017-01-05 Martin Liska <mliska@suse.cz>
36195 * gimplify.c (gimplify_decl_expr): Compare to
36196 asan_poisoned_variables instread of checking flags.
36197 (gimplify_target_expr): Likewise.
36198 (gimplify_expr): Likewise.
36199 (gimplify_function_tree): Conditionally initialize
36200 asan_poisoned_variables.
36202 2017-01-04 Jeff Law <law@redhat.com>
36204 PR tree-optimizatin/78812
36205 * rtl.h (contains_mem_rtx_p): Prototype.
36206 * ifcvt.c (containts_mem_rtx_p): Move from here to...
36207 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
36208 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
36209 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
36210 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
36212 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36214 * input.c (assert_char_at_range): Default-initialize actual_range.
36216 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36218 * df-scan.c (df_ref_create_structure): Make regno unsigned,
36219 to match the caller.
36221 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36223 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
36224 insns after final jump in test to emit dummy move.
36226 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36228 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
36229 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
36231 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36233 * multiple_target.c (create_dispatcher_calls): Init e_next.
36234 * tree-ssa-loop-split.c (split_loop): Init border.
36235 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36238 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36243 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36244 (altivec_register_operand): Do not return true if the operand
36245 contains a SUBREG mixing SImode and SFmode.
36246 (vsx_register_operand): Likewise.
36247 (vsx_reg_sfsubreg_ok): New predicate.
36248 (vfloat_operand): Do not return true if the operand contains a
36249 SUBREG mixing SImode and SFmode.
36250 (vint_operand): Likewise.
36251 (vlogical_operand): Likewise.
36252 (gpc_reg_operand): Likewise.
36253 (int_reg_operand): Likewise.
36254 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36255 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36256 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36258 (rs6000_emit_move_si_sf_subreg): New helper function.
36259 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36260 fixup SUBREGs involving SImode and SFmode.
36261 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36262 numbers for the new peephole2 optimization.
36263 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36264 the GLIBC math library that do AND/IOR/XOR operations on single
36265 precision floating point.
36266 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36267 target macros to say whether we need to avoid SUBREGs mixing
36269 (TARGET_ALLOW_SF_SUBREG): Likewise.
36270 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36271 (UNSPEC_SI_FROM_SF): Likewise.
36272 (iorxor): Change spacing.
36273 (and_ior_xor): New iterator for AND, IOR, and XOR.
36274 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36275 (movdi_from_sf_zero_ext): Likewise.
36276 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36277 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36278 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36279 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36280 (fms<mode>4): Likewise.
36281 (fnma<mode>4): Likewise.
36282 (fnms<mode>4): Likewise.
36283 (nfma<mode>4): Likewise.
36284 (nfms<mode>4): Likewise.
36286 2017-01-04 Marek Polacek <polacek@redhat.com>
36289 * doc/invoke.texi: Document -Wpointer-compare.
36291 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36293 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36296 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36297 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36300 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36303 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36304 documentation of the powerpc_popcntb_ok attribute.
36305 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36306 code to issue warning messages if a requested CPU configuration is
36307 not supported by the binary (assembler and loader) toolchain.
36308 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36309 made to define a built-in function that has been disabled.
36310 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36311 made to define a built-in function that has been disabled.
36312 (altivec_init_builtins): Add comment explaining why definition
36313 of the DST built-in functions is not preceded by an assertion
36314 check. Add assertions to prevent ICE if attempts are made to
36315 define an altivec predicate or an abs* built-in function that has
36317 (htm_init_builtins): Add comment explaining why definition of the
36318 htm built-in functions is not preceded by an assertion check.
36320 2017-01-04 Jeff Law <law@redhat.com>
36322 PR tree-optimizatin/67955
36323 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36324 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36325 the points-to solution does not include pt_null. Use DECL_PT_UID
36328 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36330 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36331 Use gen_int_mode instead of gen_lopwart for const_int operands.
36333 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36335 PR tree-optimization/71563
36336 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36337 out of range value - has low bits known to be zero.
36339 2017-01-04 Alan Modra <amodra@gmail.com>
36341 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36342 * configure: Regenerate.
36343 * config.in: Regenerate.
36345 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36348 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36349 a substring of the message, but strcmp with the whole message. Ifdef
36350 ENABLE_NLS, translate the message first using dgettext.
36352 2017-01-03 Jeff Law <law@redhat.com>
36354 PR tree-optimizatin/78856
36355 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36356 (mark_threaded_blocks): Remove code to truncate thread paths that
36357 cross multiple loop headers. Instead invalidate the cached loop
36358 iteration information and handle case of a thread path walking
36359 into an irreducible region.
36361 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36364 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36365 assertions. Add support for doing the signbit if the IEEE 128-bit
36366 floating point value is in a GPR.
36367 * config/rs6000/rs6000.md (Fsignbit): Delete.
36368 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36369 Update the length attribute if the value is in a GPR.
36370 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36371 the sign or zero extension instruction, since the value is always 0/1.
36372 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36375 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36376 extracting SImode to a GPR register so that we can generate a
36377 store, limit the vector to be in a traditional Altivec register
36378 for the vextuwrx instruction.
36380 2017-01-03 Ian Lance Taylor <iant@google.com>
36382 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36384 2017-01-03 Martin Sebor <msebor@redhat.com>
36386 PR tree-optimization/78696
36387 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36388 precision. Use MPFR for %f for greater fidelity. Correct handling
36390 (pass_sprintf_length::compute_format_length): Set width and precision
36391 specified by asrerisk to void_node for vararg functions.
36392 (try_substitute_return_value): Adjust dump output.
36394 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36396 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36398 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36400 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36401 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36402 -mlra/-mno-lra was passed to the compiler.
36404 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36406 PR rtl-optimization/65618
36407 * emit-rtl.c (try_split): Move initialization of "before" and
36408 "after" to just before the call to emit_insn_after_setloc.
36410 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36412 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36414 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36416 * dwarf2out.c (gen_enumeration_type_die): When
36417 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36419 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36421 PR tree-optimization/78965
36422 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36423 Change first argument from const call_info & to call_info &. For %n
36424 set info.nowrite to false.
36426 PR middle-end/78901
36427 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36428 possibly throwing calls.
36430 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36431 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36432 and fns handling, rather than in a separate case SSA_NAME.
36434 2017-01-02 Jeff Law <law@redhat.com>
36436 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36437 fixes for first_period and second_period variables.
36439 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
36442 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
36443 (*insvqi_1): New insn pattern.
36444 (*insvqi_1_mem_rex64): Ditto.
36445 (*insvqi_2): Ditto.
36446 (*insvqi_3): Rename from *insvqi.
36448 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
36450 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
36452 * doc/cfg.texi (Edges): Remove reference to Java.
36453 (Maintaining the CFG): Ditto.
36455 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36457 PR middle-end/77674
36458 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
36459 transparent aliases.
36461 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36463 PR middle-end/77484
36464 * predict.def (PRED_CALL): Update hitrate.
36465 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
36466 * predict.c (tree_estimate_probability_bb): Split CALL predictor
36467 into direct/indirect/polymorphic variants.
36469 2017-01-01 Jakub Jelinek <jakub@redhat.com>
36471 Update copyright years.
36473 * gcc.c (process_command): Update copyright notice dates.
36474 * gcov-dump.c (print_version): Ditto.
36475 * gcov.c (print_version): Ditto.
36476 * gcov-tool.c (print_version): Ditto.
36477 * gengtype.c (create_file): Ditto.
36478 * doc/cpp.texi: Bump @copying's copyright year.
36479 * doc/cppinternals.texi: Ditto.
36480 * doc/gcc.texi: Ditto.
36481 * doc/gccint.texi: Ditto.
36482 * doc/gcov.texi: Ditto.
36483 * doc/install.texi: Ditto.
36484 * doc/invoke.texi: Ditto.
36486 Copyright (C) 2017 Free Software Foundation, Inc.
36488 Copying and distribution of this file, with or without modification,
36489 are permitted in any medium without royalty provided the copyright
36490 notice and this notice are preserved.