1 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4 (aarch64_frame): Add emit_frame_chain boolean.
5 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6 Move eh_return case to aarch64_layout_frame.
7 (aarch64_layout_frame): Initialize emit_frame_chain.
8 (aarch64_expand_prologue): Use emit_frame_chain.
10 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
12 * config/aarch64/aarch64.c (aarch64_layout_frame):
13 Ensure LR is always stored at the bottom of the callee-saves.
14 Remove rarely used frame layout which saves callee-saves at top of
15 frame, so the store of LR can be used as a valid probe in all cases.
17 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
19 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
20 Improve unaligned TImode/TFmode base/offset split.
22 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
23 Alan Hayward <alan.hayward@arm.com>
24 David Sherwood <david.sherwood@arm.com>
26 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
27 * combine.c (find_single_use_1): Likewise.
28 (expand_field_assignment): Likewise.
29 (move_deaths): Likewise.
30 * lra-constraints.c (simplify_operand_subreg): Likewise.
31 (curr_insn_transform): Likewise.
32 * lra.c (collect_non_operand_hard_regs): Likewise.
33 (add_regs_to_insn_regno_info): Likewise.
34 * rtlanal.c (reg_referenced_p): Likewise.
35 (covers_regno_no_parallel_p): Likewise.
37 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
39 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
40 Don't print any bits outside the precision of the value.
41 * wide-int.cc (test_printing): Add some new tests.
43 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
45 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
46 supports -xbrace_comment option.
47 * configure: Regenerate.
48 * config.in: Regenerate.
49 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
50 (ASM_CPU_SPEC): Use it.
52 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
54 * target.def (static_rtx_alignment): New hook.
55 * targhooks.h (default_static_rtx_alignment): Declare.
56 * targhooks.c (default_static_rtx_alignment): New function.
57 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
58 * doc/tm.texi: Regenerate.
59 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
60 instead of targetm.constant_alignment. Remove call to
62 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
63 (cris_preferred_mininum_alignment): New function, split out from...
64 (cris_constant_alignment): ...here.
65 (cris_static_rtx_alignment): New function.
66 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
68 (ix86_constant_alignment): ...here.
69 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
70 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
71 (mmix_static_rtx_alignment): New function.
72 * config/spu/spu.c (spu_static_rtx_alignment): New function.
73 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
75 2017-10-26 Tamar Christina <tamar.christina@arm.com>
78 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): Add flag_trapping_math
79 and flag_fp_int_builtin_inexact.
81 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
83 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
84 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
85 mark as a sign-extending load.
86 (local_pic_load_u): Define.
88 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
91 * fold-const.c (operand_equal_for_comparison_p): Also return true
92 if ARG0 is a simple variant of ARG1 with narrower precision.
93 (fold_ternary_loc): Always pass unstripped operands to the predicate.
95 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
97 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
99 * i386.h (processor_costs): Add gather_static, gather_per_elt,
100 scatter_static, scatter_per_elt.
101 * x86-tune-costs.h: Add new cost entries.
103 2017-10-25 Richard Biener <rguenther@suse.de>
105 * tree-ssa-sccvn.h (vn_eliminate): Declare.
106 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
107 class pass_fre): Move to ...
108 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
109 class pass_fre): ... here and adjust for statistics.
111 2017-10-25 Jakub Jelinek <jakub@redhat.com>
114 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
115 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
116 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
117 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
120 2017-10-25 Richard Biener <rguenther@suse.de>
122 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
123 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
124 eliminate_push_avail, eliminate_insert): Move inside...
125 (class eliminate_dom_walker): ... this class in preparation
127 (fini_eliminate): Remove by merging with ...
128 (eliminate): ... this function. Adjust for class changes.
129 (pass_pre::execute): Remove fini_eliminate call.
130 (pass_fre::execute): Likewise.
132 2017-10-24 Jakub Jelinek <jakub@redhat.com>
135 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
136 (VPERMI2, VPERMI2I): New mode iterators.
137 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
138 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
140 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
141 mode iterator. Remove 3 old define_insn patterns.
142 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
143 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
144 VPERMI2 mode iterator, remove the other two expanders.
145 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
146 to use VPERMI2 mode iterator, add another alternative for vpermi2*
147 instructions, remove the other two patterns.
148 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
149 mode iterator, remove the other two patterns.
150 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
151 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
152 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
153 and adjust argument order accordingly.
154 (ix86_expand_vec_perm): Adjust caller.
155 (expand_vec_perm_1): Likewise.
156 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
157 (expand_vec_perm_vpermt2_vpshub2): ... this.
158 (ix86_expand_vec_perm_const_1): Adjust caller.
159 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
162 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
163 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
164 (vec_shl_<mode>): Remove unused expander.
165 (avx512bw_<shift_insn><mode>3): New define_insn.
166 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
167 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
169 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
172 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
175 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
177 PR rtl-optimization/82396
178 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
179 (autopref_multipass_init): Simplify initialization.
180 (autopref_rank_data): Simplify sort order.
181 * gcc/sched-int.h (autopref_multipass_data_): Remove
182 multi_mem_insn_p, min_offset and max_offset.
184 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
187 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
188 Check special value of flag_omit_frame_pointer.
189 (aarch64_can_eliminate): Likewise.
190 (aarch64_override_options_after_change_1): Simplify handling of
191 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
193 2017-10-24 Richard Biener <rguenther@suse.de>
195 PR tree-optimization/82697
196 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
197 zero for conditional load and unconditional store.
199 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
201 * doc/install.texi: Document bootstrap-cet.
203 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
206 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
207 ENDBR instruction at function entrance if function is only
210 2017-10-24 Jakub Jelinek <jakub@redhat.com>
213 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
214 patterns to better describe from which operation the CF is computed.
215 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
216 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
217 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
218 is 0, use _0 suffixed expanders instead of emitting a comparison
221 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
223 * config/i386/i386.md(*movsf_internal, *movdf_internal):
224 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
226 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
229 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
230 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
231 * loop-iv.c (iv_get_reaching_def): Likewise.
232 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
233 variable is promoted and the partition contains undefined values.
235 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
237 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
239 (nios2_address_cost): Define.
240 (nios2_legitimize_address): Recognize (exp + constant) directly.
241 (TARGET_ADDRESS_COST): Define.
243 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
245 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
246 (nios2_symbolic_memory_operand_p): Declare.
247 (nios2_split_large_constant): Declare.
248 (nios2_split_symbolic_memory_operand): Declare.
249 * config/nios2/nios2.c: Adjust includes.
250 (nios2_symbolic_constant_allowed): New.
251 (nios2_symbolic_constant_p): New.
252 (nios2_plus_symbolic_constant_p): New.
253 (nios2_valid_addr_expr_p): Recognize addresses involving
255 (nios2_legitimate_address_p): Likewise, also LO_SUM.
256 (nios2_symbolic_memory_operand_p): New.
257 (nios2_large_constant_p): New.
258 (nios2_split_large_constant): New.
259 (nios2_split_plus_large_constant): New.
260 (nios2_split_symbolic_memory_operand): New.
261 (nios2_legitimize_address): Code refactoring. Handle addresses
262 involving symbolic constants.
263 (nios2_emit_move_sequence): Likewise.
264 (nios2_print_operand): Improve error output.
265 (nios2_print_operand_address): Handle LO_SUM.
266 (nios2_cdx_narrow_form_p): Likewise.
267 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
268 operands involving symbolic constants.
269 (movhi_internal, movsi_internal): Likewise.
270 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
271 (extendhisi2, extendqi<mode>2): Likewise.
273 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
275 * tree-pass.h (PROP_rtl_split_insns): Define.
276 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
278 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
280 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
282 2017-10-23 Jakub Jelinek <jakub@redhat.com>
285 * target.def (const_not_ok_for_debug_p): Default to
286 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
287 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
288 * targhooks.c (default_const_not_ok_for_debug_p): New function.
289 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
290 which targetm.const_not_ok_for_debug_p returned true.
291 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
293 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
295 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
296 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
297 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
299 (ix86_const_not_ok_for_debug_p): New function.
300 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
301 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
303 2017-10-23 David Malcolm <dmalcolm@redhat.com>
306 * system.h: Conditionally include "unique-ptr.h" if
307 INCLUDE_UNIQUE_PTR is defined.
308 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
309 of defining INCLUDE_UNIQUE_PTR before including "system.h".
311 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
313 * config/rl78/rl78.md: New define_expand "subdi3".
315 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
318 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
319 DF_REF_INSN if DF_REF_INSN_INFO is false.
321 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
323 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
324 xmm_move instead of sse_move.
325 (sse_store_index): New function.
326 (ix86_register_move_cost): Be more sensible about mismatch stall;
327 model AVX moves correctly; make difference between sse->integer and
329 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
330 moves; make difference between SSE and AVX.
331 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
332 and zmm_move. Increase size of sse load and store tables;
333 add unaligned load and store tables; add ssemmx_to_integer.
334 * x86-tune-costs.h: Update all entries according to real
335 move latencies from Agner Fog's manual and chip documentation.
337 2017-10-23 Jakub Jelinek <jakub@redhat.com>
340 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
341 * config/i386/constraints.md (Wf): New constraint.
342 * config/i386/i386.md (UNSPEC_SBB): New unspec.
343 (cmp<dwi>_doubleword): Removed.
344 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
345 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
346 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
347 expand with cmp<dwi>_doubleword. For LTU and GEU use
348 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
350 * common.opt (gcolumn-info): Enable by default.
351 * doc/invoke.texi (gcolumn-info): Document new default.
353 2017-10-23 Richard Biener <rguenther@suse.de>
355 PR tree-optimization/82672
356 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
357 Fold the stmt if we propagated into it.
359 2017-10-23 Richard Biener <rguenther@suse.de>
361 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
362 (bitmap_remove_expr_from_set): ... this. All callers call this
363 for non-constant values.
364 (bitmap_set_subtract): Rename to...
365 (bitmap_set_subtract_expressions): ... this. Adjust and
367 (bitmap_set_contains_value): Remove superfluous check.
368 (bitmap_set_replace_value): Inline into single caller ...
369 (bitmap_value_replace_in_set): ... here and simplify.
370 (dependent_clean): Merge into ...
371 (clean): ... this using an overload. Adjust.
372 (prune_clobbered_mems): Adjust.
373 (compute_antic_aux): Likewise.
374 (compute_partial_antic_aux): Likewise.
376 2017-10-23 Richard Biener <rguenther@suse.de>
378 PR tree-optimization/82129
380 2017-08-01 Richard Biener <rguenther@suse.de>
382 PR tree-optimization/81181
383 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
384 (compute_antic): ... end of iteration here.
386 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
388 * target.def (starting_frame_offset): New hook.
389 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
390 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
391 * doc/tm.texi.in: Regenerate.
392 * hooks.h (hook_hwi_void_0): Declare.
393 * hooks.c (hook_hwi_void_0): New function.
394 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
395 STARTING_FRAME_OFFSET.
396 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
397 * reload1.c (reload): Likewise.
398 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
399 instead of STARTING_FRAME_OFFSET.
400 * function.c (try_fit_stack_local): Likewise.
401 (assign_stack_local_1): Likewise
402 (instantiate_virtual_regs): Likewise.
403 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
404 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
405 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
406 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
407 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
408 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
409 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
410 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
411 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
412 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
413 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
414 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
415 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
416 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
417 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
418 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
419 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
420 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
421 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
422 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
423 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
424 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
425 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
426 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
427 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
428 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
429 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
430 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
431 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
432 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
433 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
434 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
435 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
436 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
437 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
438 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
439 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
440 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
441 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
442 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
443 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
444 * config/avr/avr.c (avr_starting_frame_offset): Make static and
445 return a HOST_WIDE_INT.
446 (avr_builtin_setjmp_frame_value): Use it instead of
447 STARTING_FRAME_OFFSET.
448 (TARGET_STARTING_FRAME_OFFSET): Redefine.
449 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
450 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
452 (TARGET_STARTING_FRAME_OFFSET): Redefine.
453 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
454 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
455 (TARGET_CONSTANT_ALIGNMENT): Redefine.
456 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
457 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
458 (TARGET_STARTING_FRAME_OFFSET): Redefine.
459 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
460 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
461 (TARGET_STARTING_FRAME_OFFSET): Redefine.
462 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
463 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
465 (TARGET_STARTING_FRAME_OFFSET): Redefine.
466 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
467 * config/mips/mips.c (mips_compute_frame_info): Refer to
468 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
469 (mips_starting_frame_offset): New function.
470 (TARGET_STARTING_FRAME_OFFSET): Redefine.
471 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
472 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
473 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
474 and return a HOST_WIDE_INT.
475 (TARGET_STARTING_FRAME_OFFSET): Redefine.
476 (mmix_initial_elimination_offset): Refer to
477 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
478 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
479 * config/pa/pa.c (pa_starting_frame_offset): New function.
480 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
481 (pa_expand_prologue): Likewise.
482 (TARGET_STARTING_FRAME_OFFSET): Redefine.
483 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
484 !FRAME_GROWS_DOWNWARD handling to...
485 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
486 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
487 !FRAME_GROWS_DOWNWARD handling to...
488 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
489 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
490 !FRAME_GROWS_DOWNWARD handling to...
491 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
492 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
494 (rs6000_starting_frame_offset): New function.
495 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
496 !FRAME_GROWS_DOWNWARD handling to...
497 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
498 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
499 !FRAME_GROWS_DOWNWARD handling to...
500 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
501 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
502 !FRAME_GROWS_DOWNWARD handling to...
503 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
504 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
505 (rs6000_starting_frame_offset): New function.
506 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
507 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
508 * config/vax/vax.c (vax_starting_frame_offset): New function.
509 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
510 (TARGET_STARTING_FRAME_OFFSET): Redefine.
511 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
512 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
513 (TARGET_STARTING_FRAME_OFFSET): Redefine.
514 * system.h (STARTING_FRAME_OFFSET): Poison.
516 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
518 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
519 SCALAR_TYPE_MODE instead of TYPE_MODE.
521 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
522 Alan Hayward <alan.hayward@arm.com>
523 David Sherwood <david.sherwood@arm.com>
525 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
527 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
528 Alan Hayward <alan.hayward@arm.com>
529 David Sherwood <david.sherwood@arm.com>
531 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
533 2017-10-23 Richard Biener <rguenther@suse.de>
535 PR tree-optimization/82129
536 * tree-ssa-pre.c (bitmap_set_and): Remove.
537 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
538 canonicalizing expressions in the set to those with lowest
539 ID rather than taking that from the first edge.
541 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
543 * combine.c (rtx_equal_for_field_assignment_p): Use
546 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
547 Alan Hayward <alan.hayward@arm.com>
548 David Sherwood <david.sherwood@arm.com>
550 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
551 to a SUBREG_PROMOTED_VAR.
553 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
554 Alan Hayward <alan.hayward@arm.com>
555 David Sherwood <david.sherwood@arm.com>
557 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
558 (expand_debug_source_expr): Likewise.
559 * combine.c (combine_simplify_rtx): Likewise.
560 * cse.c (fold_rtx): Likewise.
561 * optabs.c (expand_float): Likewise.
562 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
563 (simplify_binary_operation_1): Likewise.
565 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
566 Alan Hayward <alan.hayward@arm.com>
567 David Sherwood <david.sherwood@arm.com>
569 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
570 (record_promoted_value): Likewise.
571 * expr.c (expand_expr_real_2): Likewise.
572 * ree.c (update_reg_equal_equiv_notes): Likewise.
573 (combine_set_extension): Likewise.
574 * rtlanal.c (low_bitmask_len): Likewise.
575 * simplify-rtx.c (neg_const_int): Likewise.
576 (simplify_binary_operation_1): Likewise.
578 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
579 Alan Hayward <alan.hayward@arm.com>
580 David Sherwood <david.sherwood@arm.com>
582 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
583 * regcprop.c (maybe_mode_change): Likewise.
584 * reload1.c (alter_reg): Likewise.
586 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
588 * inchash.h (inchash::hash::add_wide_int): New function.
589 * lto-streamer-out.c (hash_tree): Use it.
591 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
593 * inchash.h (inchash::hash::add_wide_int): Rename to...
594 (inchash::hash::add_hwi): ...this.
595 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
596 (polymorphic_call_target_hasher::hash): Likewise.
597 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
598 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
599 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
600 * lto-streamer-out.c (hash_tree): Likewise.
601 * optc-save-gen.awk: Likewise.
602 * tree.c (add_expr): Likewise.
604 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
607 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
608 for ordered inequality comparisons even with TARGET_IEEE_FP.
610 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
613 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
614 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
615 Expand with cmp<dwi>_doubleword.
617 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
619 * extend.texi: Add x86 specific to 'nocf_check' attribute.
621 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
622 specific to -fcf-protection option.
624 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
626 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
627 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
628 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
629 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
630 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
631 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
632 (extra_objs): Add cet.o for Linux/x86 targets.
633 (tmake_file): Add i386/t-cet for Linux/x86 targets.
634 * config/i386/cet.c: New file.
635 * config/i386/cetintrin.h: Likewise.
636 * config/i386/t-cet: Likewise.
637 * config/i386/cpuid.h (bit_SHSTK): New.
639 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
640 pass IBT and SHSTK bits.
641 * config/i386/i386-builtin-types.def
642 (VOID_FTYPE_UNSIGNED_PVOID): New.
643 (VOID_FTYPE_UINT64_PVOID): Likewise.
644 * config/i386/i386-builtin.def: Add CET intrinsics.
645 * config/i386/i386-c.c (ix86_target_macros_internal): Add
646 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
647 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
648 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
650 * config/i386/i386.c (rest_of_insert_endbranch): New.
651 (pass_data_insert_endbranch): Likewise.
652 (pass_insert_endbranch): Likewise.
653 (make_pass_insert_endbranch): Likewise.
654 (ix86_notrack_prefixed_insn_p): Likewise.
655 (ix86_target_string): Add -mibt, -mshstk flags.
656 (ix86_option_override_internal): Add flag_cf_protection
658 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
659 (ix86_print_operand): Add 'notrack' prefix output.
660 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
661 (ix86_expand_builtin): Expand CET intrinsics.
662 (x86_output_mi_thunk): Add 'endbranch' instruction.
663 * config/i386/i386.h (TARGET_IBT): New.
664 (TARGET_IBT_P): Likewise.
665 (TARGET_SHSTK): Likewise.
666 (TARGET_SHSTK_P): Likewise.
667 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
668 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
669 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
670 (builtin_setjmp_setup): New pattern.
671 (builtin_longjmp): Likewise.
672 (rdssp<mode>): Likewise.
673 (incssp<mode>): Likewise.
674 (saveprevssp): Likewise.
675 (rstorssp): Likewise.
676 (wrss<mode>): Likewise.
677 (wruss<mode>): Likewise.
678 (setssbsy): Likewise.
679 (clrssbsy): Likewise.
680 (nop_endbr): Likewise.
681 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
683 * config/i386/immintrin.h: Include <cetintrin.h>.
684 * config/i386/linux-common.h
685 (file_end_indicate_exec_stack_and_cet): New prototype.
686 (TARGET_ASM_FILE_END): New.
688 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
690 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
691 latencies instead of having separate table; make difference between
692 integer and float costs.
693 * i386.h (processor_costs): Remove scalar_stmt_cost,
694 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
695 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
697 * x86-tune-costs.h: Remove entries which has been removed in
698 procesor_costs from all tables; make cond_taken_branch_cost
699 and cond_not_taken_branch_cost COST_N_INSNS based.
701 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
703 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
705 2017-10-20 Jakub Jelinek <jakub@redhat.com>
707 * config/i386/i386.md (isa): Remove fma_avx512f.
708 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
709 <avx512>_fmadd_<mode>_mask3<round_name>,
710 <avx512>_fmsub_<mode>_mask<round_name>,
711 <avx512>_fmsub_<mode>_mask3<round_name>,
712 <avx512>_fnmadd_<mode>_mask<round_name>,
713 <avx512>_fnmadd_<mode>_mask3<round_name>,
714 <avx512>_fnmsub_<mode>_mask<round_name>,
715 <avx512>_fnmsub_<mode>_mask3<round_name>,
716 <avx512>_fmaddsub_<mode>_mask<round_name>,
717 <avx512>_fmaddsub_<mode>_mask3<round_name>,
718 <avx512>_fmsubadd_<mode>_mask<round_name>,
719 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
720 (*vec_widen_umult_even_v16si<mask_name>,
721 *vec_widen_smult_even_v16si<mask_name>): Likewise.
722 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
724 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
726 * extend.texi: Add 'nocf_check' documentation.
727 * gimple.texi: Add second parameter to
728 gimple_build_call_from_tree.
729 * invoke.texi: Add -fcf-protection documentation.
730 * rtl.texi: Add REG_CALL_NOTRACK documenation.
732 2017-10-20 Richard Biener <rguenther@suse.de>
734 PR tree-optimization/82473
735 * tree-vect-loop.c (vectorizable_reduction): Properly get at
736 the largest input type.
738 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
740 * c-attribs.c (handle_nocf_check_attribute): New function.
741 (c_common_attribute_table): Add 'nocf_check' handling.
742 * gimple-parser.c: Add second argument NULL to
743 gimple_build_call_from_tree.
744 * attrib.c (comp_type_attributes): Check nocf_check attribute.
745 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
747 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
748 * common.opt: Add fcf-protection flag.
749 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
750 * flag-types.h: Add enum cf_protection_level.
751 * gimple.c (gimple_build_call_from_tree): Add second parameter.
752 Add 'nocf_check' attribute propagation to gimple call.
753 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
754 (gimple_build_call_from_tree): Update prototype.
755 (gimple_call_nocf_check_p): New function.
756 (gimple_call_set_nocf_check): Likewise.
757 * gimplify.c: Add second argument to gimple_build_call_from_tree.
758 * ipa-icf.c: Add nocf_check attribute in statement hash.
759 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
760 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
761 * toplev.c (process_options): Add flag_cf_protection handling.
763 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
765 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
767 2017-10-20 Richard Biener <rguenther@suse.de>
769 PR tree-optimization/82603
770 * tree-if-conv.c (predicate_mem_writes): Make sure to only
771 remove false predicated stores.
773 2017-10-20 Richard Biener <rguenther@suse.de>
775 * graphite-isl-ast-to-gimple.c
776 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
777 Remove return value and simplify, dump copied stmt after lhs
779 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
780 Reduce dump verbosity.
781 (gsi_insert_earliest): Likewise.
782 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
783 * graphite.c (print_global_statistics): Adjust dumping.
784 (print_graphite_scop_statistics): Likewise.
785 (print_graphite_statistics): Do not dump loops here.
786 (graphite_transform_loops): But here.
788 2017-10-20 Nicolas Roche <roche@adacore.com>
790 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
791 * configure: Regenerate.
793 2017-10-20 Jakub Jelinek <jakub@redhat.com>
796 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
797 functions when optimizing replace GIMPLE_RETURN stmts with
798 calls to __builtin_unreachable ().
801 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
802 for -fsanitize=thread link of executables.
803 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
807 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
809 (<shift_insn><mode>3<mask_name>): Change the last of the 3
810 define_insns for logical vector shifts to use VI248_AVX512BW
811 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
812 condition, useless isa and prefix attributes. Change the first
813 2 of these define_insns to ...
814 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
815 define_insn for avx512vl.
816 (<shift_insn><mode>3): ... and this, new define_insn without
817 masking for non-avx512vl.
820 * config/i386/sse.md (*andnot<mode>3,
821 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
822 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
823 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
824 not applied use empty suffix even for TARGET_AVX512VL.
825 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
826 is applied, supply evex,evex or evex,evex,evex instead of just
829 2017-10-20 Julia Koval <julia.koval@intel.com>
831 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
832 (OPTION_MASK_ISA_GFNI_UNSET): New.
833 (ix86_handle_option): Handle OPT_mgfni.
834 * config/i386/cpuid.h (bit_GFNI): New.
835 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
836 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
837 * config/i386/i386.c (ix86_target_string): Add -mgfni.
838 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
839 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
840 * config/i386/i386.opt: Add mgfni.
842 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
844 * config/msp430/msp430.c (msp430_option_override): Disable
845 -fdelete-null-pointer-checks.
846 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
848 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
850 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
851 of x87 and SSE instructions.
853 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
855 * asan.c (create_cond_insert_point): Do not update edge count.
856 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
857 (afdo_propagate_circuit): Likewise.
858 (afdo_calculate_branch_prob): Likewise.
859 (afdo_annotate_cfg): Likewise.
860 * basic-block.h (struct edge_def): Remove count.
861 (edge_def::count): New accessor.
862 * bb-reorder.c (rotate_loop): Update.
863 (find_traces_1_round): Update.
864 (connect_traces): Update.
865 (sanitize_hot_paths): Update.
866 * cfg.c (unchecked_make_edge): Update.
867 (make_single_succ_edge): Update.
868 (check_bb_profile): Update.
869 (dump_edge_info): Update.
870 (update_bb_profile_for_threading): Update.
871 (scale_bbs_frequencies_int): Update.
872 (scale_bbs_frequencies_gcov_type): Update.
873 (scale_bbs_frequencies_profile_count): Update.
874 (scale_bbs_frequencies): Update.
875 * cfganal.c (connect_infinite_loops_to_exit): Update.
876 * cfgbuild.c (compute_outgoing_frequencies): Update.
877 (find_many_sub_basic_blocks): Update.
878 * cfgcleanup.c (try_forward_edges): Update.
879 (try_crossjump_to_edge): Update
880 * cfgexpand.c (expand_gimple_cond): Update
881 (expand_gimple_tailcall): Update
882 (construct_exit_block): Update
883 * cfghooks.c (verify_flow_info): Update
884 (redirect_edge_succ_nodup): Update
886 (make_forwarder_block): Update
887 (duplicate_block): Update
888 (account_profile_record): Update
889 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
890 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
891 * cfgloopmanip.c (scale_loop_profile): Update.
893 (lv_adjust_loop_entry_edge): Update.
894 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
895 (force_nonfallthru_and_redirect): Update.
896 (purge_dead_edges): Update.
897 (rtl_flow_call_edges_add): Update.
898 * cgraphunit.c (init_lowered_empty_function): Update.
899 (cgraph_node::expand_thunk): Update.
900 * gimple-pretty-print.c (dump_probability): Update.
901 (dump_edge_probability): Update.
902 * gimple-ssa-isolate-paths.c (isolate_path): Update.
903 * haifa-sched.c (sched_create_recovery_edges): Update.
904 * hsa-gen.c (convert_switch_statements): Update.
905 * ifcvt.c (dead_or_predicable): Update.
906 * ipa-inline-transform.c (inline_transform): Update.
907 * ipa-split.c (split_function): Update.
908 * ipa-utils.c (ipa_merge_profiles): Update.
909 * loop-doloop.c (add_test): Update.
910 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
911 * lto-streamer-in.c (input_cfg): Update.
912 (input_function): Update.
913 * lto-streamer-out.c (output_cfg): Update.
914 * modulo-sched.c (sms_schedule): Update.
915 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
916 * predict.c (maybe_hot_edge_p): Update.
917 (unlikely_executed_edge_p): Update.
918 (probably_never_executed_edge_p): Update.
919 (dump_prediction): Update.
920 (drop_profile): Update.
921 (propagate_unlikely_bbs_forward): Update.
922 (determine_unlikely_bbs): Update.
923 (force_edge_cold): Update.
924 * profile.c (compute_branch_probabilities): Update.
925 * reg-stack.c (better_edge): Update.
926 * shrink-wrap.c (handle_simple_exit): Update.
927 * tracer.c (better_p): Update.
928 * trans-mem.c (expand_transaction): Update.
929 (split_bb_make_tm_edge): Update.
930 * tree-call-cdce.c: Update.
931 * tree-cfg.c (gimple_find_sub_bbs): Update.
932 (gimple_split_edge): Update.
933 (gimple_duplicate_sese_region): Update.
934 (gimple_duplicate_sese_tail): Update.
935 (gimple_flow_call_edges_add): Update.
936 (insert_cond_bb): Update.
937 (execute_fixup_cfg): Update.
938 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
939 * tree-complex.c (expand_complex_div_wide): Update.
940 * tree-eh.c (lower_resx): Update.
941 (unsplit_eh): Update.
942 (cleanup_empty_eh_move_lp): Update.
943 * tree-inline.c (copy_edges_for_bb): Update.
944 (freqs_to_counts): Update.
945 (copy_cfg_body): Update.
946 * tree-ssa-dce.c (remove_dead_stmt): Update.
947 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
948 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
949 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
950 (unloop_loops): Update.
951 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
952 * tree-ssa-loop-split.c (connect_loops): Update.
953 (split_loop): Update.
954 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
955 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
956 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
957 * tree-ssa-reassoc.c (branch_fixup): Update.
958 * tree-ssa-tail-merge.c (replace_block_by): Update.
959 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
960 (compute_path_counts): Update.
961 (update_profile): Update.
962 (recompute_probabilities): Update.
963 (update_joiner_offpath_counts): Update.
964 (estimated_freqs_path): Update.
965 (freqs_to_counts_path): Update.
966 (clear_counts_path): Update.
967 (ssa_fix_duplicate_block_edges): Update.
968 (duplicate_thread_path): Update.
969 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
970 (case_bit_test_cmp): Update.
971 (collect_switch_conv_info): Update.
972 (gen_inbound_check): Update.
973 (do_jump_if_equal): Update.
974 (emit_cmp_and_jump_insns): Update.
975 * tree-tailcall.c (decrease_profile): Update.
976 (eliminate_tail_call): Update.
977 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
978 (vect_do_peeling): Update.
979 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
980 * ubsan.c (ubsan_expand_null_ifn): Update.
981 (ubsan_expand_ptr_ifn): Update.
982 * value-prof.c (gimple_divmod_fixed_value): Update.
983 (gimple_mod_pow2): Update.
984 (gimple_mod_subtract): Update.
986 (gimple_stringop_fixed_value): Update.
988 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
991 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
993 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
995 PR rtl-optimization/82395
996 * ira-color.c (allocno_priority_compare_func): Fix comparison step
997 based on non_spilled_static_chain_regno_p.
999 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1001 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1002 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1003 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1005 2017-10-19 Martin Sebor <msebor@redhat.com>
1007 PR tree-optimization/82596
1008 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1010 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1012 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1013 (get_mem_refs_of_builtin_call): Likewise.
1014 * builtins.c (expand_builtin_apply): Adjust call to
1015 allocate_dynamic_stack_space.
1016 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1017 the third argument to allocate_dynamic_stack_space, otherwise -1.
1018 (expand_builtin): Deal with all alloca variants.
1019 (is_inexpensive_builtin): Likewise.
1020 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1021 * calls.c (special_function_p): Deal with all alloca variants.
1022 (initialize_argument_information): Adjust call to
1023 allocate_dynamic_stack_space.
1024 (expand_call): Likewise.
1025 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1026 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1027 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1028 use it for the stack usage computation.
1029 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1030 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1031 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1032 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1033 (in_loop_p): Remove first argument and useless check.
1034 (pass_walloca::execute): Remove useless test and adjust call to above.
1035 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1036 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1037 (gimplify_call_expr): Deal with all alloca variants.
1038 * hsa-gen.c (gen_hsa_alloca): Likewise.
1039 (gen_hsa_insns_for_call): Likewise.
1040 * ipa-pure-const.c (special_builtin_state): Likewise.
1041 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1042 * tree-object-size.c (alloc_object_size): Likewise.
1043 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1044 (call_may_clobber_ref_p_1): Likewise.
1045 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1046 (ccp_fold_stmt): Likewise.
1047 (optimize_stack_restore): Likewise.
1048 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1049 (mark_all_reaching_defs_necessary_1): Likewise.
1050 (propagate_necessity): Likewise.
1051 (eliminate_unnecessary_stmts): Likewise.
1052 * tree.c (build_common_builtin_nodes): Build
1053 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1054 (build_alloca_call_expr): New function.
1055 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1056 (CASE_BUILT_IN_ALLOCA): Likewise.
1057 (build_alloca_call_expr): Declare.
1058 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1060 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1063 * dwarf2out.c (new_die_raw): New static inline function.
1064 (new_die): Use it to create the DIE.
1065 (add_AT_external_die_ref): Likewise.
1066 (clone_die): Likewise.
1067 (clone_as_declaration): Likewise.
1068 (dwarf2out_vms_debug_main_pointer): Likewise.
1069 (base_type_die): Likewise. Remove early return for corner cases.
1070 Do not call add_pubtype on the DIE here.
1071 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1072 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1073 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1074 native order exists for base types, attach the DIE manually and call
1075 add_pubtype on it. Do not equate a reverse order DIE to the type.
1077 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1079 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1080 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1082 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1084 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1086 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1087 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1088 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1089 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1090 * opts.c (sanitizer_opts): Add builtin.
1091 * ubsan.c (instrument_builtin): New function.
1092 (pass_ubsan::execute): Call it.
1093 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1094 * doc/invoke.texi: Document -fsanitize=builtin.
1096 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1097 builtins, store max (log2 (align), 0) into uchar field instead of
1098 align into uptr field.
1099 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1100 store uchar 0 field instead of uptr 0 field.
1101 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1102 instead of passing one address of struct with 2 locations pass
1103 two addresses of structs with 1 location each.
1104 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1105 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1106 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1107 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1108 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1109 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1110 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1111 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1113 2017-10-19 Martin Liska <mliska@suse.cz>
1116 * file-find.c (remove_prefix): Remove.
1117 * file-find.h (remove_prefix): Likewise.
1118 * gcc-ar.c: Remove smartness of lookup.
1120 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1122 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1123 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1124 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1126 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1129 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1130 (setcc + and to xor + setcc): New peephole2.
1132 2017-10-19 Tom de Vries <tom@codesourcery.com>
1134 * doc/sourcebuild.texi (Test Directives, Variants of
1135 dg-require-support): Add dg-require-stack-size.
1137 2017-10-19 Martin Liska <mliska@suse.cz>
1140 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1141 that have a large alignment.
1142 (gimplify_target_expr): Likewise.
1144 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1146 PR rtl-optimization/82602
1147 * ira.c (rtx_moveable_p): Return false for volatile asm.
1149 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1152 * config/i386/i386-modes.def (CCGZ): New CC mode.
1153 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1154 * config/i386/predicates.md (ix86_comparison_operator):
1156 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1157 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1158 with double-word subtraction.
1159 (put_condition_code): Handle CCGZmode.
1161 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1163 * wide-int.cc (debug (const wide_int &)): New.
1164 (debug (const wide_int *)): New.
1165 (debug (const widest_int &)): New.
1166 (debug (const widest_int *)): New.
1168 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1171 * lra-constraints.c (curr_insn_transform): Use non-input operand
1172 instead of output one for matched reload.
1174 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1176 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1177 (tree-ssa-loop-ivopts.h): New header file.
1178 (struct builtin_info): New fields.
1179 (classify_builtin_1): Compute and record base and offset parts for
1180 memset builtin partition by calling strip_offset.
1181 (offset_cmp, fuse_memset_builtins): New functions.
1182 (finalize_partitions): Fuse adjacent memset partitions by calling
1184 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1185 Expose the interface.
1186 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1188 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1190 PR tree-optimization/82574
1191 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1192 that data reference must be executed exactly once per iteration
1193 against the outermost loop in nest.
1194 (classify_partition): Update call to above function.
1196 2017-10-18 Richard Biener <rguenther@suse.de>
1198 PR tree-optimization/82591
1199 * graphite.c (graphite_transform_loops): Move code gen message
1201 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1202 Here. Handle scop_to_isl_ast failing.
1203 (scop_to_isl_ast): Limit the number of ISL operations.
1205 2017-10-18 Richard Biener <rguenther@suse.de>
1207 * graphite-isl-ast-to-gimple.c
1208 (translate_isl_ast_to_gimple::set_rename): Simplify.
1209 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1210 (graphite_copy_stmts_from_block): ... here.
1211 (copy_bb_and_scalar_dependences): Simplify.
1212 (add_parameters_to_ivs_params): Canonicalize.
1213 (generate_entry_out_of_ssa_copies): Simplify.
1214 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1215 by passing in ISL dimension.
1216 (parameter_index_in_region_1): Rename to ...
1217 (parameter_index_in_region): ... this.
1218 (extract_affine): Adjust assert, pass down parameter index.
1219 (add_param_constraints): Use range-info when available.
1220 (build_scop_context): Adjust.
1221 * sese.c (new_sese_info): Adjust.
1222 (free_sese_info): Likewise.
1223 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1224 Remove unused typedefs.
1225 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1227 2017-10-18 Martin Liska <mliska@suse.cz>
1229 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1231 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1233 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1234 (s390_sched_init): Do not reset s390_sched_state if we entered the
1235 current basic block via a fallthru edge and all others are unlikely.
1237 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1239 * config/s390/s390.c (NUM_SIDES): New variable.
1240 (LONGRUNNING_THRESHOLD): New variable.
1241 (LATENCY_FACTOR): New variable.
1242 (s390_sched_score): Decrease score for long-running instructions on
1244 (s390_sched_variable_issue): Perform bookkeeping for long-running
1247 2017-10-18 Richard Biener <rguenther@suse.de>
1249 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1250 Simplify with removal of the parameter rename map.
1251 (set_rename): Likewise.
1252 (should_copy_to_new_region): Likewise.
1253 (graphite_copy_stmts_from_block): Likewise.
1254 (copy_bb_and_scalar_dependences): Remove initialization of
1255 unused copied_bb_map.
1257 (copy_internal_parameters): Likewise.
1258 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1259 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1260 Use INTEGRAL_TYPE_P.
1261 (parameter_index_in_region_1): Rename to ...
1262 (assign_parameter_index_in_region): ... this. Assert we have
1263 a parameter we handle.
1264 (scan_tree_for_params): Adjust.
1265 * sese.h (parameter_rename_map_t): Remove.
1266 (struct sese_info_t): Remove unused parameter_rename_map and
1267 copied_bb_map members.
1268 * sese.c (new_sese_info): Adjust.
1269 (free_sese_info): Likewise.
1271 2017-10-18 Martin Liska <mliska@suse.cz>
1274 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1275 on an abnormal edge.
1277 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1279 * doc/invoke.texi (ffunction-sections and fdata-sections):
1282 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1284 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1285 the use statement can throw internally.
1287 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1289 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1290 any RTX present on the RHS of a SET.
1291 * compare-elim.c (try_eliminate_compare): Restore comment.
1293 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1295 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1296 may be also called on tcc_type nodes.
1297 * langhooks.c (lhd_tree_size): Likewise.
1299 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1301 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1302 format_warning_at_substring.
1303 (maybe_warn): Convert source_range * param to a location_t. Pass
1304 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1305 (format_directive): Remove code to extract source_ranges and
1306 source_range * in favor of just a location_t.
1307 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1309 * substring-locations.c (format_warning_va): Convert
1310 source_range * param to a location_t.
1311 (format_warning_at_substring): Likewise.
1312 * substring-locations.h (format_warning_va): Likewise.
1313 (format_warning_at_substring): Likewise.
1315 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1317 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1319 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1320 and scatter/gather ops.
1322 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1323 vec_gather_load and vec_scatter_store.
1324 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1325 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1326 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1327 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1328 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1329 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1331 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1333 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1334 Detect FTST instruction and handle its register pops. Only pop
1335 second operand if can_pop_second_op is true.
1336 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1337 set can_pop_second_op to false in the compare_for_stack_reg call.
1339 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1340 output_fp_compare for stack register operands.
1341 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1342 instructions here. Do not emit stack register pops here. Assert
1343 that FCOMPP pops next to top stack register. Rewrite function.
1345 2017-10-17 Nathan Sidwell <nathan@acm.org>
1348 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1349 use DECL_ASSEMBLER_NAME_RAW.
1352 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1355 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1356 Wilco Dijkstra <wilco.dijkstra@arm.com>
1358 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1359 converstion to Pmode from the buf_addr.
1361 2017-10-17 Richard Biener <rguenther@suse.de>
1363 * graphite-dependences.c (scop_get_reads_and_writes): Change
1364 output parameters to references.
1366 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1368 PR 71026/tree-optimization
1369 * fold-const.c (distribute_real_division): Removed.
1370 (fold_binary_loc): Remove calls to distribute_real_divison.
1372 2017-10-17 Richard Biener <rguenther@suse.de>
1374 * graphite-scop-detection.c
1375 (scop_detection::stmt_has_simple_data_refs_p): Always use
1376 the full nest as region.
1377 (try_generate_gimple_bb): Likewise.
1378 * sese.c (scalar_evolution_in_region): Simplify now that
1379 SCEV can handle instantiation in regions.
1380 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1381 in the non-loop part of a function if requested.
1383 2017-10-17 Richard Biener <rguenther@suse.de>
1385 PR tree-optimization/82563
1386 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1388 (graphite_regenerate_ast_isl): Call it.
1389 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1391 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1393 PR tree-optimization/82549
1394 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1395 Formatting fixes. Instead of calling make_bit_field_ref with negative
1398 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1400 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1401 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1402 _mm_maskz_reduce_ss): New.
1403 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1404 __builtin_ia32_reducess_mask): Ditto..
1405 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1406 * config/i386/sse.md (reduces<mode>): Renamed to ...
1407 (reduces<mode><mask_scalar_name>): ... this.
1408 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
1410 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1411 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1413 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1415 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1416 * selftest-run-tests.c (selftest::run_tests): Call
1417 selftest::unique_ptr_tests_cc_tests.
1418 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1419 * unique-ptr-tests.cc: New file.
1421 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1424 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1426 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1427 (make_hard_regno_born, make_hard_regno_dead): Update
1428 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1430 2017-10-16 Jeff Law <law@redhat.com>
1432 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1434 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1436 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1438 2017-10-16 Olivier Hainque <hainque@adacore.com>
1440 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1441 with_cpu if we were configured for an e500v2 target cpu name.
1443 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1445 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1446 * doc/invoke.texi: Document +nodsp as a valid extension for
1449 2017-10-16 Martin Liska <mliska@suse.cz>
1451 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1452 (test_set_range): Likewise.
1453 (test_range_functions): Rename to ...
1454 (test_bit_in_range): ... this.
1455 (sbitmap_c_tests): Add new test.
1457 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1459 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
1460 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1461 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1463 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1465 * config/aarch64/aarch64-builtins.c
1466 (aarch64_types_quadopu_lane_qualifiers): New.
1467 (TYPES_QUADOPU_LANE): New.
1468 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1469 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1470 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1471 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1472 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1473 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1474 (Vdottype, DOTPROD): New.
1475 (sur): Add SDOT and UDOT.
1477 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1479 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1480 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1481 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
1482 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1483 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
1484 (cortex-a75.cortex-a55): Likewise.
1485 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1487 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1489 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1490 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
1491 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
1492 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1493 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1494 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1495 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1496 * config/arm/types.md (neon_dot, neon_dot_q): New.
1497 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1499 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1501 * config/arm/arm.h (TARGET_DOTPROD): New.
1502 * config/arm/arm.c (arm_arch_dotprod): New.
1503 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1504 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1505 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1506 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1507 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1508 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1509 * doc/invoke.texi (armv8.2-a): Document dotprod
1511 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1513 * i386.c (ix86_vec_cost): New function.
1514 (ix86_rtx_costs): Handle vector operations better.
1515 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1516 * x86-tune-costs.h: Add new costs to all tables.
1518 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1520 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
1522 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
1523 divsd, sqrtss and sqrtsd
1524 * x86-tune-costs.h: Add new entries to all costs.
1525 (znver1_cost): Fix to match real instruction latencies.
1527 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1528 Michael Collison <michael.collison@arm.com>
1530 * compare-elim.c: Include emit-rtl.h.
1531 (can_merge_compare_into_arith): New function.
1532 (try_validate_parallel): Likewise.
1533 (try_merge_compare): Likewise.
1534 (try_eliminate_compare): Call the above when no previous clobber
1536 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
1539 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1543 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
1544 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
1548 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
1549 to be any operand_equal_p operands. For & (B - 1) require
1550 B to be power of 2. Recognize
1551 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
1553 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1556 * optabs.c (expand_memory_blockage): Fix call of
1557 targetm.have_memory_blockage.
1559 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1562 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
1563 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
1564 objects to extra_objs instead of overwriting it.
1566 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1568 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
1569 Use any_fp_register_operand as operand[3] predicate. Simplify
1570 equality test for operands[2] and operands[4] memory location.
1571 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
1572 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
1573 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
1574 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
1575 any_fp_register_operand as operand[1] predicate. Simplify
1576 equality test for operands[0] and operands[3] memory location.
1577 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
1578 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
1579 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
1581 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1583 * target-insns.def: Add memory_blockage.
1584 * optabs.c (expand_memory_blockage): New function.
1585 (expand_asm_memory_barrier): Rename ...
1586 (expand_asm_memory_blockage): ... to this.
1587 (expand_mem_thread_fence): Call expand_memory_blockage
1588 instead of expand_asm_memory_barrier.
1589 (expand_mem_singnal_fence): Ditto.
1590 (expand_atomic_load): Ditto.
1591 (expand_atomic_store): Ditto.
1592 * doc/md.texi (Standard Pattern Names For Generation):
1593 Document memory_blockage instruction pattern.
1595 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
1597 * config/rl78/rl78.c (rl78_emit_libcall): New function.
1598 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
1599 * config/rl78/rl78.md: New define_expand "adddi3".
1601 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1603 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1606 2017-10-13 Jeff Law <law@redhat.com>
1608 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1610 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1613 * internal-fn.c (expand_mul_overflow): If both operands have
1614 the same highpart of -1 or 0 and the topmost bit of lowpart
1615 is different, overflow is if res <= 0 rather than res < 0.
1617 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1619 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1620 TARGET_P9_VECTOR code for unaligned_load case.
1622 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1624 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1626 2017-10-13 Nathan Sidwell <nathan@acm.org>
1628 * tree-core.h (tree_contains_struct): Make bool.
1629 * tree.c (tree_contains_struct): Likewise.
1630 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1631 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1632 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1633 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1634 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1636 2017-10-13 Richard Biener <rguenther@suse.de>
1638 * graphite-isl-ast-to-gimple.c
1639 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1640 parameters and dominance check.
1641 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1642 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1643 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1644 Do not update SSA form here or do intermediate IL verification.
1645 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1646 (graphite_initialize): Remove check on the number of loops in
1647 the function and inline into graphite_transform_loops.
1648 (graphite_finalize): Inline into graphite_transform_loops.
1649 (graphite_transform_loops): Perform SSA update and IL verification
1651 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1653 2017-10-13 Richard Biener <rguenther@suse.de>
1655 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1656 graphite_expression_type_precision): Avoid global constructor
1658 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1659 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1660 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1661 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1662 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1663 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1665 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1668 * config/i386/i386.h (ix86_red_zone_size): New.
1669 * config/i386/i386.md (push peephole2s): Replace
1670 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1672 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1673 Alan Hayward <alan.hayward@arm.com>
1674 David Sherwood <david.sherwood@arm.com>
1676 * combine.c (can_change_dest_mode): Reject changes in
1677 REGMODE_NATURAL_SIZE.
1679 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1680 Alan Hayward <alan.hayward@arm.com>
1681 David Sherwood <david.sherwood@arm.com>
1683 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1684 (expand_debug_source_expr): Likewise.
1685 * combine.c (combine_simplify_rtx): Likewise.
1686 * cse.c (fold_rtx): Likewise.
1687 * fwprop.c (canonicalize_address): Likewise.
1688 * targhooks.c (default_shift_truncation_mask): Likewise.
1690 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1691 Alan Hayward <alan.hayward@arm.com>
1692 David Sherwood <david.sherwood@arm.com>
1694 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1695 (widened_mode): Likewise.
1696 (expand_unop): Likewise.
1697 * ree.c (transform_ifelse): Likewise.
1698 (merge_def_and_ext): Likewise.
1699 (combine_reaching_defs): Likewise.
1700 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1702 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1703 Alan Hayward <alan.hayward@arm.com>
1704 David Sherwood <david.sherwood@arm.com>
1706 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1707 * combine.c (gen_lowpart_for_combine): Likewise.
1708 * dwarf2out.c (rtl_for_decl_location): Likewise.
1709 * final.c (alter_subreg): Likewise.
1710 * rtlhooks.c (gen_lowpart_general): Likewise.
1711 (gen_lowpart_if_possible): Likewise.
1713 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1714 Alan Hayward <alan.hayward@arm.com>
1715 David Sherwood <david.sherwood@arm.com>
1717 * calls.c (expand_call): Use subreg_lowpart_offset.
1718 * cse.c (cse_insn): Likewise.
1719 * regcprop.c (copy_value): Likewise.
1720 (copyprop_hardreg_forward_1): Likewise.
1722 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1725 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1726 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1727 =Q constraints to +Q and into insn condition add check
1728 that operands[0] and operands[1] are equal.
1729 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1730 =Q constraints to +Q and into insn condition add check
1731 that operands[0] is equal to either operands[1] or operands[2].
1734 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1735 instead of handling MINUS_EXPR twice (once for each argument),
1736 canonicalize operand order and handle just once, use rtype where
1737 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1740 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1741 any values of __C while still being pattern recognizable as a simple
1744 2017-10-13 Richard Biener <rguenther@suse.de>
1746 PR tree-optimization/82451
1748 2017-10-02 Richard Biener <rguenther@suse.de>
1750 PR tree-optimization/82355
1751 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1752 a mapping for the enclosing loop but avoid generating one for
1754 (copy_bb_and_scalar_dependences): Remove premature codegen
1755 error on PHIs in blocks duplicated into multiple places.
1756 * graphite-scop-detection.c
1757 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1758 in the region use it as loop and nest to analyze the DR in.
1759 (try_generate_gimple_bb): Likewise.
1760 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1761 (add_loop_constraints): For blocks in a loop not in the region
1762 create a dimension with a single iteration.
1763 * sese.h (gbb_loop_at_index): Remove assert.
1765 * cfgloop.c (loop_preheader_edge): For the loop tree root
1766 return the single successor of the entry block.
1767 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1768 Reset the SCEV hashtable and niters.
1769 * graphite-scop-detection.c
1770 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1771 assert that we only have POLYNOMIAL_CHREC that vary in loops
1772 contained in the region.
1773 (scop_detection::graphite_can_represent_expr): Adjust.
1774 (scop_detection::stmt_has_simple_data_refs_p): For loops
1775 not in the region set loop to NULL. The nest is now the
1776 entry edge to the region.
1777 (try_generate_gimple_bb): Likewise.
1778 * sese.c (scalar_evolution_in_region): Adjust for
1779 instantiate_scev change.
1780 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1781 Make nest parameter the edge into the region.
1782 (create_data_ref): Likewise.
1783 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1784 entry edge into a region and adjust instantiate_scev calls.
1785 (create_data_ref): Likewise.
1786 (graphite_find_data_references_in_stmt): Likewise.
1787 (find_data_references_in_stmt): Pass the loop preheader edge
1788 from the nest argument.
1789 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1790 parameter the edge into the region.
1791 (instantiate_parameters): Use the loop preheader edge as entry.
1792 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1794 (get_instantiated_value_entry): Make instantiate_below parameter
1795 the edge into the region.
1796 (instantiate_scev_name): Likewise. Adjust dominance checks,
1797 when we cannot use loop-based instantiation instantiate by
1798 walking use-def chains.
1799 (instantiate_scev_poly): Adjust.
1800 (instantiate_scev_binary): Likewise.
1801 (instantiate_scev_convert): Likewise.
1802 (instantiate_scev_not): Likewise.
1803 (instantiate_array_ref): Remove.
1804 (instantiate_scev_3): Likewise.
1805 (instantiate_scev_2): Likewise.
1806 (instantiate_scev_1): Likewise.
1807 (instantiate_scev_r): Do not blindly handle N-operand trees.
1808 Do not instantiate array-refs. Handle all constants and invariants.
1809 (instantiate_scev): Make instantiate_below parameter
1810 the edge into the region.
1811 (resolve_mixers): Use the loop preheader edge for the region
1812 parameter to instantiate_scev_r.
1813 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1815 2017-10-13 Richard Biener <rguenther@suse.de>
1817 PR tree-optimization/82525
1818 * graphite-isl-ast-to-gimple.c
1819 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1821 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1822 Fail code generation when we cannot represent the isl integer.
1823 (binary_op_to_tree): Elide modulo operations that are no-ops
1824 in the type we code generate. Remove now superfluous code
1827 2017-10-13 Richard Biener <rguenther@suse.de>
1829 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1830 (scop_detection::harmful_loop_in_region): Remove premature
1831 IV type restriction.
1832 (scop_detection::graphite_can_represent_scev): We can handle
1833 pointer IVs just fine.
1835 2017-10-13 Alan Modra <amodra@gmail.com>
1837 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1838 "Clobbers and Scratch Registers". Add paragraph on
1839 alternative to clobbers for scratch registers and OpenBLAS
1842 2017-10-13 Alan Modra <amodra@gmail.com>
1844 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1845 example of a memory input for a string of known length. Move
1846 commentary out of table. Add a number of new examples
1847 covering array memory inputs.
1849 2017-10-12 Martin Liska <mliska@suse.cz>
1851 PR tree-optimization/82493
1852 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1853 (test_range_functions): New function.
1854 (sbitmap_c_tests): Likewise.
1855 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1856 * selftest.h (sbitmap_c_tests): New function.
1858 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1860 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1862 * config/rs6000/amo.h: Fix spacing issue.
1864 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1867 * config/i386/i386.md (*ashl<mode>3_mask_1,
1868 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1869 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1872 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1874 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1875 (profile_probability): Set max_probability
1876 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1878 * profile-count.c (profile_probability::differs_from_p): Do not
1879 rely on max_probaiblity == 10000
1881 2017-10-12 Jeff Law <law@redhat.com>
1883 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1886 2017-10-12 Martin Sebor <msebor@redhat.com>
1890 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1891 (handle_alias_pairs): Call it.
1892 * common.opt (-Wattribute-alias): New option.
1893 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1894 * doc/invoke.texi (-Wattribute-alias): Document.
1896 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1899 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1901 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1903 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1904 (make_hard_regno_born, make_hard_regno_dead): Update
1905 bb_killed_pseudos and bb_gen_pseudos.
1907 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1909 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1911 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1913 * config/alpha/alpha.c (alpha_split_conditional_move):
1914 Use std::swap instead of manually swapping.
1915 (alpha_stdarg_optimize_hook): Ditto.
1916 (alpha_canonicalize_comparison): Ditto.
1918 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1920 * tree-loop-distribution.c (struct builtin_info): New struct.
1921 (struct partition): Refactor fields into struct builtin_info.
1922 (partition_free): Free struct builtin_info.
1923 (build_size_arg_loc, build_addr_arg_loc): Delete.
1924 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1925 information from struct builtin_info.
1926 (find_single_drs): New function refactored from classify_partition.
1927 Also moved builtin validity checks to this function.
1928 (compute_access_range, alloc_builtin): New functions.
1929 (classify_builtin_st, classify_builtin_ldst): New functions.
1930 (classify_partition): Refactor code into functions find_single_drs,
1931 classify_builtin_st and classify_builtin_ldst.
1932 (distribute_loop): Don't do runtime alias check when distributing
1934 (find_seed_stmts_for_distribution): New function.
1935 (pass_loop_distribution::execute): Refactor code finding seed
1936 stmts into above function. Support distribution for the innermost
1937 two-level loop nest. Adjust dump information.
1939 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1941 * tree-loop-distribution.c: Adjust the general comment.
1942 (NUM_PARTITION_THRESHOLD): New macro.
1943 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1944 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1945 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1946 in call to build_partition_graph.
1947 (finalize_partitions): New parameter. Make loop distribution more
1948 conservative by fusing more partitions.
1949 (distribute_loop): Don't do runtime alias check in case of loop nest
1951 (find_seed_stmts_for_distribution): New function.
1952 (prepare_perfect_loop_nest): New function.
1953 (pass_loop_distribution::execute): Refactor code finding seed stmts
1954 and loop nest into above functions. Support loop nest distribution.
1955 Adjust dump information accordingly.
1957 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1959 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1960 and set PTYPE_SEQUENTIAL for merged partition.
1962 2017-10-12 Richard Biener <rguenther@suse.de>
1964 PR tree-optimization/69728
1966 2017-09-19 Richard Biener <rguenther@suse.de>
1968 PR tree-optimization/69728
1969 * graphite-sese-to-poly.c (schedule_error): New global.
1970 (add_loop_schedule): Handle empty domain by failing the
1972 (build_original_schedule): Handle schedule_error.
1974 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1975 domain by returning an unchanged schedule.
1977 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1979 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1980 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1982 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1984 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1987 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1990 * expr.c (store_field): Don't optimize away bitsize == 0 store
1991 from CALL_EXPR with addressable return type.
1993 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1995 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1996 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1997 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1998 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1999 TARGET_ISEL instead of TARGET_ISEL<sel>.
2001 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2003 * config/rs6000/rs6000.c
2004 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2006 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2008 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2010 (reg_or_cint_operand): Fix comment.
2011 (reg_or_zero_operand): New predicate.
2012 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2013 * config/rs6000/rs6000.c (output_isel): Delete.
2014 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2015 instead of reg_or_cint_operand. Output instruction directly (not via
2017 (isel_unsigned_<mode>): Ditto.
2018 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2019 gpc_reg_operand. Add an instruction alternative for this. Output
2020 instruction directly.
2021 (*isel_reversed_unsigned_<mode>): Ditto.
2023 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2025 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2026 (TARGET_CANONICALIZE_COMPARISON): Define.
2028 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2031 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2032 Check whether the dest is REG before adding REG_EQUIV note.
2034 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2037 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2039 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2040 (make_hard_regno_born, make_hard_regno_dead): Update
2041 bb_killed_pseudos and bb_gen_pseudos.
2043 2017-10-11 Nathan Sidwell <nathan@acm.org>
2045 * incpath.h (enum incpath_kind): Name enum, prefix values.
2046 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2047 * incpath.c (heads, tails): Use INC_MAX.
2048 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2049 (merge_include_chains, split_quote_chain,
2050 register_include_chains): Update incpath_kind names.
2051 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2052 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2054 (add_framework_path, darwin_register_objc_includes): Likewise.
2055 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2057 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2059 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2060 Do not use float_operator operator predicate.
2061 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2062 * config/i386/predicates.md (float_operator): Remove predicate.
2064 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2066 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2067 (*jccxf_i387): Ditto.
2068 (*jcc<mode>_i387): Ditto.
2069 (*jccu<mode>_i387): Ditto.
2070 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2071 (*jcc_*_i387 splitters): Remove.
2072 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2073 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2074 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2077 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2079 * profile-count.h (slow_safe_scale_64bit): New function.
2080 (safe_scale_64bit): New inline.
2081 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2082 * profile-count.c: Include wide-int.h
2083 (slow_safe_scale_64bit): New.
2085 2017-10-11 Nathan Sidwell <nathan@acm.org>
2087 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2088 HAS_DECL_ASSEMBLER_NAME_P.
2089 * gimple-expr.c (gimple_decl_printable_name: Check
2090 HAS_DECL_ASSEMBLER_NAME_P too.
2091 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2092 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2093 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2094 * passes.c (rest_of_decl_compilation): Check
2095 HAS_DECL_ASSEMBLER_NAME_P too.
2096 * recog.c (verify_changes): Likewise.
2097 * tree-pretty-print.c (dump_decl_name): Likewise.
2098 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2100 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2101 (SET_DECL_ASSEMBLER_NAME): Use it.
2102 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2103 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2104 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2106 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2108 * config.gcc (i386, x86_64): Add extra objects.
2109 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2110 (ix86_min_insn_size): Declare.
2111 (ix86_issue_rate): Declare.
2112 (ix86_adjust_cost): Declare.
2113 (ia32_multipass_dfa_lookahead): Declare.
2114 (ix86_macro_fusion_p): Declare.
2115 (ix86_macro_fusion_pair_p): Declare.
2116 (ix86_bd_has_dispatch): Declare.
2117 (ix86_bd_do_dispatch): Declare.
2118 (ix86_core2i7_init_hooks): Declare.
2119 (ix86_atom_sched_reorder): Declare.
2120 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2121 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2122 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2123 (rip_relative_addr_p): Rename to ...
2124 (ix86_rip_relative_addr_p): ... this one; export.
2125 (memory_address_length): Update.
2126 (ix86_issue_rate): Move to x86-tune-sched.c.
2127 (ix86_flags_dependent): Move to x86-tune-sched.c.
2128 (ix86_agi_dependent): Move to x86-tune-sched.c.
2129 (exact_dependency_1): Move to x86-tune-sched.c.
2130 (exact_store_load_dependency): Move to x86-tune-sched.c.
2131 (ix86_adjust_cost): Move to x86-tune-sched.c.
2132 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2133 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2134 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2135 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2136 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2137 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2138 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2139 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2140 (min_insn_size): Rename to ...
2141 (ix86_min_insn_size): ... this one; export.
2142 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2143 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2144 (core2i7_first_cycle_multipass_backtrack): Move to
2145 x86-tune-sched-core.c.
2146 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2147 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2148 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2149 (ix86_avoid_jump_mispredicts): Update.
2150 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2151 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2152 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2153 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2154 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2155 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2156 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2157 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2158 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2159 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2160 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2161 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2162 (BIG): Move to ix86-tune-sched-bd.c.
2163 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2164 (enum insn_path): Move to ix86-tune-sched-bd.c.
2165 (get_mem_group): Move to ix86-tune-sched-bd.c.
2166 (is_cmp): Move to ix86-tune-sched-bd.c.
2167 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2168 (is_branch): Move to ix86-tune-sched-bd.c.
2169 (is_prefetch): Move to ix86-tune-sched-bd.c.
2170 (init_window): Move to ix86-tune-sched-bd.c.
2171 (allocate_window): Move to ix86-tune-sched-bd.c.
2172 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2173 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2174 (process_end_window): Move to ix86-tune-sched-bd.c.
2175 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2176 (find_constant): Move to ix86-tune-sched-bd.c.
2177 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2178 (has_immediate): Move to ix86-tune-sched-bd.c.
2179 (get_insn_path): Move to ix86-tune-sched-bd.c.
2180 (get_insn_group): Move to ix86-tune-sched-bd.c.
2181 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2182 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2183 (add_insn_window): Move to ix86-tune-sched-bd.c.
2184 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2185 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2186 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2187 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2188 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2189 (do_dispatch): Move to ix86-tune-sched-bd.c.
2190 (has_dispatch): Move to ix86-tune-sched-bd.c.
2191 * config/i386/t-i386: Add new object files.
2192 * config/i386/x86-tune-costs.h: New file.
2193 * config/i386/x86-tune-sched-atom.c: New file.
2194 * config/i386/x86-tune-sched-bd.c: New file.
2195 * config/i386/x86-tune-sched-core.c: New file.
2196 * config/i386/x86-tune-sched.c: New file.
2198 2017-10-11 Liu Hao <lh_mouse@126.com>
2200 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2201 the generic version below instead.
2202 (should_colorize): Recognize Windows consoles as terminals
2204 * pretty-print.c [__MINGW32__] (write_all): New function.
2205 [__MINGW32__] (find_esc_head): Likewise.
2206 [__MINGW32__] (find_esc_terminator): Likewise.
2207 [__MINGW32__] (eat_esc_sequence): Likewise.
2208 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2210 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2213 2017-10-11 Richard Biener <rguenther@suse.de>
2215 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2216 Properly call analyze_scalar_evolution with the loop of the stmt.
2218 2017-10-11 Richard Biener <rguenther@suse.de>
2220 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2221 * tree-core.h (tree_base): Add chrec_var union member.
2222 * tree.h (CHREC_VAR): Remove.
2223 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2224 * tree-chrec.h (build_polynomial_chrec): Adjust.
2225 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2226 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2228 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2230 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2231 * match.pd: ... here.
2232 ((T) X == (T) Y): Relax condition.
2234 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2236 PR tree-optimization/82472
2237 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2239 (break_alias_scc_partitions): Update postorder number.
2241 2017-10-11 Martin Liska <mliska@suse.cz>
2244 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2246 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2248 2017-10-11 Martin Liska <mliska@suse.cz>
2250 * print-rtl.c (print_insn): Move declaration of idbuf
2251 to same scope as name.
2253 2017-10-11 Martin Liska <mliska@suse.cz>
2258 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2261 2017-10-11 Martin Liska <mliska@suse.cz>
2264 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2267 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2269 * config.gcc (default_gnu_indirect_function): Default to yes for
2270 arm*-*-linux* with glibc.
2272 2017-10-11 Richard Biener <rguenther@suse.de>
2274 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2275 default-defs and types we do not want to analyze.
2276 (interpret_loop_phi): Replace unreachable code with an assert.
2277 (compute_scalar_evolution_in_loop): Remove and inline ...
2278 (analyze_scalar_evolution_1): ... here, replacing condition with
2279 what makes the intent clearer. Remove handling of cases
2280 get_scalar_evolution now handles.
2282 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2284 PR rtl-optimization/81434
2285 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2286 comment for main loop. In sched_group_found if, also add checks for
2287 pass and min_cost_group.
2289 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2291 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2292 (rs6000_insn_cost): New function.
2293 * config/rs6000/rs6000.md (cost): New attribute.
2295 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2296 H.J. Lu <hongjiu.lu@intel.com>
2300 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2301 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2302 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2303 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2304 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2305 OPTION_MASK_ISA_MMX and some other bit set require both
2306 mmx and the ISAs without the mmx bit.
2307 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2308 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2309 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2310 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2311 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2312 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2313 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2314 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2315 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2316 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2317 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2318 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2319 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2320 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2321 Add OPTION_MASK_ISA_MMX.
2323 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2325 * config.gcc (armv7*-*-freebsd*): New target.
2326 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2328 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2330 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2331 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2334 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2336 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2337 a template parameter.
2338 (WIDE_INT_REF_FOR): Update accordingly.
2339 * tree.h (wi::int_traits <const_tree>): Delete.
2340 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2341 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2342 (wi::tree_to_wide_ref): New typedef.
2343 (wi::to_wide): New function.
2344 * calls.c (get_size_range): Use wi::to_wide when operating on
2346 * cgraph.c (cgraph_node::create_thunk): Likewise.
2347 * config/i386/i386.c (ix86_data_alignment): Likewise.
2348 (ix86_local_alignment): Likewise.
2349 * dbxout.c (stabstr_O): Likewise.
2350 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2351 * expr.c (const_vector_from_tree): Likewise.
2352 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2353 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2354 (fold_negate_expr_1, int_const_binop_1, const_binop)
2355 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2356 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2357 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2358 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2359 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2360 (fold_not_const, round_up_loc): Likewise.
2361 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2362 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2363 (alloca_call_type): Likewise.
2364 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2365 * godump.c (go_output_typedef): Likewise.
2366 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2367 * internal-fn.c (get_min_precision): Likewise.
2368 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2369 * ipa-polymorphic-call.c
2370 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2371 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2372 (ipa_modify_call_arguments): Likewise.
2373 * match.pd: Likewise.
2374 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2375 * print-tree.c (print_node_brief, print_node): Likewise.
2376 * stmt.c (expand_case): Likewise.
2377 * stor-layout.c (layout_type): Likewise.
2378 * tree-affine.c (tree_to_aff_combination): Likewise.
2379 * tree-cfg.c (group_case_labels_stmt): Likewise.
2380 * tree-data-ref.c (dr_analyze_indices): Likewise.
2381 (prune_runtime_alias_test_list): Likewise.
2382 * tree-dump.c (dequeue_and_dump): Likewise.
2383 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2384 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2385 * tree-pretty-print.c (dump_generic_node): Likewise.
2386 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2387 (simple_iv_with_niters): Likewise.
2388 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2389 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2390 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2391 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2392 (refine_value_range_using_guard, number_of_iterations_ne_max)
2393 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2394 (get_cst_init_from_scev, record_nonwrapping_iv)
2395 (scev_var_range_cant_overflow): Likewise.
2396 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2397 * tree-ssa-pre.c (compute_avail): Likewise.
2398 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2399 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2400 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2401 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2402 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2403 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2404 (dump_case_nodes, try_switch_expansion): Likewise.
2405 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2406 (vect_do_peeling): Likewise.
2407 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2408 * tree-vect-stmts.c (vectorizable_load): Likewise.
2409 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2410 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2411 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2412 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2413 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2414 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2415 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2416 (evrp_dom_walker::before_dom_children): Likewise.
2417 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2418 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2419 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2420 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2421 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2422 (get_range_pos_neg): Likewise.
2423 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2424 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2425 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2426 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2427 * config/avr/avr.c (avr_fold_builtin): Likewise.
2428 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2429 * config/msp430/msp430.c (msp430_attr): Likewise.
2430 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2431 * config/powerpcspe/powerpcspe-c.c
2432 (altivec_resolve_overloaded_builtin): Likewise.
2433 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2434 (rs6000_expand_ternop_builtin): Likewise.
2435 * config/rs6000/rs6000-c.c
2436 (altivec_resolve_overloaded_builtin): Likewise.
2437 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2438 (rs6000_expand_ternop_builtin): Likewise.
2439 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2441 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2443 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2444 when copying loop nest with only one inner loop.
2446 2017-10-10 Richard Biener <rguenther@suse.de>
2448 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2449 blocks if SCEV is active.
2450 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2452 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2453 (scev_initialize): Assert we are not yet initialized.
2455 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2457 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2458 inner loop's exit stmt by making it always exit the loop, otherwise
2459 we would generate an infinite empty loop.
2461 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2463 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2464 renaming variables in new preheader if it's deleted.
2466 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2468 * tree-loop-distribution.c (struct partition): Remove unused field
2469 loops of the structure.
2470 (partition_alloc, partition_free): Ditto.
2471 (build_rdg_partition_for_vertex): Ditto.
2473 2017-10-09 Jeff Law <law@redhat.com>
2475 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2476 return type to match prototype and documentation.
2478 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2480 * config/rs6000/rs6000.c (processor_costs): Move to ...
2481 * config/rs6000/rs6000.h: ... here.
2482 (rs6000_cost): Declare.
2484 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2486 * except.c (setjmp_fn): New global variable.
2487 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2488 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2489 if DONT_USE_BUILTIN_SETJMP is defined.
2491 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2493 * target.def (insn_cost): New hook.
2494 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2495 * doc/tm.texi: Regenerate.
2496 * rtlanal.c (insn_cost): Use the new hook.
2498 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2500 * combine.c (combine_validate_cost): Compute the new insn_cost,
2501 not just pattern_cost.
2502 (try_combine): Adjust comment.
2504 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2506 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2508 * combine.c (uid_insn_cost): Adjust comment.
2509 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2511 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2512 * dse.c (find_shift_sequence): Ditto.
2513 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2514 (bb_valid_for_noce_process_p): Use pattern_cost.
2515 * rtl.h (insn_rtx_cost): Delete.
2516 (pattern_cost): New prototype.
2517 (insn_cost): New prototype.
2518 * rtlanal.c (insn_rtx_cost): Rename to...
2519 (pattern_cost): ... this.
2522 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
2524 * config/i386/i386.md (*jcc_2): Remove insn pattern.
2525 (*jcc<mode>_0_r_i387): Ditto.
2526 (*jccxf_r_i387): Ditto.
2527 (*jcc<mode>_r_i387): Ditto.
2528 (*jccu<mode>_r_i387): Ditto.
2529 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
2530 (*jcc): Rename from *jcc_1.
2532 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2534 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
2535 deferred rescans after the lvx/stvx recombination pre-pass.
2537 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
2539 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
2540 memory operation instruction support.
2541 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
2542 (rs6000-ibm-aix[789]*): Likewise.
2543 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
2544 Document new functions.
2546 2017-10-09 Richard Biener <rguenther@suse.de>
2548 PR tree-optimization/82397
2549 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
2550 equality only for semantically equal trees.
2552 2017-10-09 Richard Biener <rguenther@suse.de>
2554 PR tree-optimization/82449
2555 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
2556 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
2557 allow constant addresses.
2558 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
2561 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2563 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
2566 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2569 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
2572 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2575 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
2577 2017-10-09 Jakub Jelinek <jakub@redhat.com>
2580 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
2581 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
2583 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
2585 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
2586 (WI_BINARY_PREDICATE_RESULT): Likewise.
2587 (wi::binary_traits::operator_result): New type.
2588 (wi::binary_traits::predicate_result): Likewise.
2589 (generic_wide_int::operator~, unary generic_wide_int::operator-)
2590 (generic_wide_int::operator==, generic_wide_int::operator!=)
2591 (generic_wide_int::operator&, generic_wide_int::and_not)
2592 (generic_wide_int::operator|, generic_wide_int::or_not)
2593 (generic_wide_int::operator^, generic_wide_int::operator+
2594 (binary generic_wide_int::operator-, generic_wide_int::operator*):
2596 (operator~, unary operator-, operator==, operator!=, operator&)
2597 (operator|, operator^, operator+, binary operator-, operator*): New
2599 * expr.c (get_inner_reference): Use wi::bit_and_not.
2600 * fold-const.c (fold_binary_loc): Likewise.
2601 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
2602 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2603 (bit_value_binop): Likewise.
2604 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2605 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2606 (extract_range_from_binary_expr_1): Likewise.
2607 (masked_increment): Likewise.
2608 (simplify_bit_ops_using_ranges): Likewise.
2610 2017-10-09 Martin Jambor <mjambor@suse.cz>
2613 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2614 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2615 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2617 (hsa_fixup_mov_insn_type): New function.
2618 (hsa_op_with_type::get_in_type): Use it.
2619 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2620 immediates in an assert.
2621 (hsa_op_with_type::extend_int_to_32bit): New method.
2622 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2623 types. Convert to dest type if necessary.
2624 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2625 (reg_for_gimple_ssa): Pass false as min32int to
2626 hsa_type_for_scalar_tree_type.
2627 (gen_hsa_addr): Fixup type when creating addresable temporary.
2628 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2629 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2630 necessary. Call hsa_fixup_mov_insn_type.
2631 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2632 extend operands and convert to dest type if necessary.
2633 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2634 to dest type if necessary.
2635 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2636 if conversion nt necessary and size matches.
2637 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2638 to dest type if necessary.
2639 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2640 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2642 (gen_hsa_clrsb): Likewise.
2643 (gen_hsa_ffs): Likewise.
2644 (gen_hsa_divmod): Extend operands and convert to dest type if
2646 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2648 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2650 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2651 Remove empty default arguments. Use a brace block as output
2653 (conditional return): Ditto.
2655 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2656 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2658 (group_ending_nop): Ditto.
2659 (doloop_end): Ditto.
2660 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2661 (splitters for those): Ditto.
2663 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2665 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2666 a conditional jump (and the compare for it) so that pc_rtx is the
2668 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2669 for the deleted and renamed ctr<mode>_internal[234] patterns.
2670 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2671 Delete second conditional return pattern.
2672 (ctr<mode>_internal2): Delete this second bdnz pattern.
2673 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2674 (ctr<mode>_internal4): Delete this second bdz pattern.
2676 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2678 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2679 (always_initialized_rtx_for_ssa_name_p): New predicate.
2680 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2681 (finish_out_of_ssa): Free new field of SA.
2682 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2683 * tree-ssa-coalesce.c: Include tree-ssa.h.
2684 (get_parm_default_def_partitions): Remove extern keyword.
2685 (get_undefined_value_partitions): New function.
2686 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2687 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2689 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2691 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2693 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2695 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2697 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2698 for targets that preffer 128bit.
2700 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2702 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2704 2017-10-08 Olivier Hainque <hainque@adacore.com>
2706 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2707 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2709 (thumb_set_return_address): Likewise.
2711 2017-10-08 Olivier Hainque <hainque@adacore.com>
2713 * common/config/arm/arm-common.c (arm_except_unwind_info):
2714 Handle DWARF2_UNWIND_INFO.
2716 2017-10-07 Michael Collison <michael.collison@arm.com>
2718 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2721 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2723 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2724 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2725 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2726 defined, force the creation of a new block for a dispatch label.
2728 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2730 * invoke.texi (Wsuggest-attribute=cold): Document.
2731 * common.opt (Wsuggest-attribute=cold): New
2732 * ipa-pure-const.c (warn_function_cold): New function.
2733 * predict.c (compute_function_frequency): Use it.
2734 * predict.h (warn_function_cold): Declare.
2736 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2738 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2741 2017-10-06 Martin Liska <mliska@suse.cz>
2743 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2744 keyword for member functions.
2745 (struct sanopt_tree_couple): New struct.
2746 (struct sanopt_tree_couple_hash): New function.
2747 (struct sanopt_ctx): Add new hash_map.
2748 (has_dominating_ubsan_ptr_check): New function.
2749 (record_ubsan_ptr_check_stmt): Likewise.
2750 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2751 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2752 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2754 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2757 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2758 aarch64_simd_valid_immediate on CONST_VECTORs.
2759 (aarch64_reg_or_bic_imm): Likewise.
2761 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2763 PR rtl-optimization/82396
2764 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2766 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2768 * graphite-dependences.c (scop_get_reads): Move code to...
2769 (scop_get_must_writes): Move code to...
2770 (scop_get_may_writes): Move code to...
2771 (scop_get_reads_and_writes): ... here.
2772 (scop_get_dependences): Call scop_get_reads_and_writes.
2774 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2776 PR tree-optimization/82434
2777 * fold-const.h (can_native_encode_type_p,
2778 can_native_encode_string_p): Remove.
2779 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2780 don't encode anything, just return what would be otherwise returned.
2781 (native_encode_fixed, native_encode_complex, native_encode_vector):
2783 (native_encode_string): Likewise. Inline by hand
2784 can_native_encode_string_p.
2785 (can_native_encode_type_p): Remove.
2786 (can_native_encode_string_p): Remove.
2787 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2788 STRING_CSTs using can_native_encode_string_p, test all
2789 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2790 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2791 argument from native_encode_expr.
2792 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2793 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2796 2017-10-06 Richard Biener <rguenther@suse.de>
2798 PR tree-optimization/82397
2799 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2800 operand_equal_p but rely on data_ref_compare_tree for detecting
2802 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2803 to match up with dr_group_sort_cmp.
2805 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2808 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2810 * config/s390/s390-builtin-types.def: Regenerate.
2812 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2815 * config/s390/s390-builtin-types.def: Regenerate.
2816 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2817 Change flag from B_VXE to B_VX.
2818 (s390_vec_min_dbl): Remove B_VXE flag.
2820 2017-10-06 Richard Biener <rguenther@suse.de>
2822 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2823 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2824 translate_isl_ast_to_gimple::is_valid_rename,
2825 translate_isl_ast_to_gimple::get_rename,
2826 translate_isl_ast_to_gimple::get_def_bb_for_const,
2827 translate_isl_ast_to_gimple::get_new_name,
2828 translate_isl_ast_to_gimple::collect_all_ssa_names,
2829 translate_isl_ast_to_gimple::copy_loop_phi_args,
2830 translate_isl_ast_to_gimple::collect_all_ssa_names,
2831 translate_isl_ast_to_gimple::copy_loop_phi_args,
2832 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2833 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2834 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2835 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2836 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2837 translate_isl_ast_to_gimple::copy_cond_phi_args,
2838 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2839 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2840 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2841 translate_isl_ast_to_gimple::rename_uses,
2842 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2843 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2844 (set_rename_for_each_def): Likewise.
2845 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2846 here. Handle rewriting SCEV analyzable uses here.
2847 (copy_bb_and_scalar_dependences): Generate code for PHI
2849 (graphite_regenerate_ast_isl): Adjust.
2850 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2851 (add_write, add_read): New functions.
2852 (build_cross_bb_scalars_def): Use it and simplify.
2853 (build_cross_bb_scalars_use): Likewise.
2854 (graphite_find_cross_bb_scalar_vars): Inline into...
2855 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2856 simulating out-of-SSA. Compute liveout and add dependencies.
2857 (build_scops): Force an empty entry block.
2858 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2860 (sese_build_liveouts): Declare.
2861 (sese_trivially_empty_bb_p): Likewise.
2862 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2863 compute liveout and debug_liveout.
2864 (sese_bad_liveouts_use): Remove.
2865 (sese_reset_debug_liveouts_bb): Likewise.
2866 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2867 (sese_build_liveouts): Build liveout and debug_liveout and store
2869 (new_sese_info): Adjust.
2870 (free_sese_info): Likewise.
2871 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2872 do not build liveout here.
2873 (move_sese_in_condition): Adjust region entry.
2874 (scev_analyzable_p): Match up with chrec_apply requirements.
2875 (sese_trivially_empty_bb_p): New.
2876 * tree-into-ssa.c (get_reaching_def): Properly support generating
2877 default-defs for incremental rewrite of anonymous names.
2879 2017-10-06 Richard Biener <rguenther@suse.de>
2881 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2882 precision do not perform modulo reduction.
2884 2017-10-06 Richard Biener <rguenther@suse.de>
2886 PR tree-optimization/82436
2887 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2888 conservatively choose the vectorization factor when checking
2889 whether we can perform the required load permutation.
2890 (vect_transform_slp_perm_load): Assert when we may not fail.
2892 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2894 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2895 message for incompatible -msdata=* and -mcall-* options.
2897 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2899 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2900 rate for post-reload scheduling.
2902 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2904 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2906 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2908 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2909 to improve monte carlo in scimark.
2911 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2913 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2914 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2915 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2916 pentium4_cost, nocona_cost): Set reassociation width to 1.
2917 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2918 width to 2 for fp operations and 1 otherwise.
2919 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2921 (atom_cost): Set reassociation width to 2.
2922 (slm_cost, generic_cost): Set fp reassociation width
2923 to 2 and 1 otherwise.
2924 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2925 (core_cost): Set fp reassociation width to 4 and vector to 2.
2926 (ix86_reassociation_width): Rewrite using cost table; special case
2927 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2928 and TARGET_AVX128_OPTIMAL.
2929 * config/i386/i386.h (processor_costs): Add
2930 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2931 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2932 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2933 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2934 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2935 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2937 2017-10-05 Nathan Sidwell <nathan@acm.org>
2939 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2941 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2943 * config/arm/arm.c (arm_test_fpu_data): New.
2944 (arm_run_selftests): Call arm_test_fpu_data.
2946 2017-10-04 Nathan Sidwell <nathan@acm.org>
2948 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2950 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2952 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2954 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2955 check type for aarch64_simd_valid_immediate.
2956 (aarch64_output_simd_mov_immediate): Update prototype.
2957 (aarch64_simd_valid_immediate): Update prototype.
2958 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2959 support for ORR-immediate.
2960 (and<mode>3): modified pattern to add support for BIC-immediate.
2961 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2962 now checks for valid immediate for BIC and ORR based on new enum
2964 (aarch64_output_simd_mov_immediate): Function now used to output
2965 BIC/ORR imm as well based on new enum argument.
2966 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2968 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2969 (aarch64_reg_or_bic_imm): Likewise.
2971 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2973 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2974 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2976 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2980 PR rtl-optimization/82396
2981 * haifa-sched.c (autopref_multipass_init): Simplify
2983 (autopref_rank_data): Simplify sort order.
2984 * sched-int.h (autopref_multipass_data_): Remove
2985 multi_mem_insn_p, min_offset and max_offset.
2987 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2989 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2991 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2993 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2996 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2998 * doc/sourcebuild.texi: Document vect_long_mult.
3000 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3002 PR tree-optimization/82413
3003 * fold-const.c (build_range_check): Use widest_int when comparing
3004 the maximum ETYPE value with HIGH.
3006 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3008 PR rtl-optimization/82396
3009 * haifa-sched.c (autopref_multipass_init): Simplify
3011 (autopref_rank_data): Simplify sort order.
3012 * sched-int.h (autopref_multipass_data_): Remove
3013 multi_mem_insn_p, min_offset and max_offset.
3015 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3017 PR tree-optimization/82381
3018 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3019 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3022 PR tree-optimization/82374
3023 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3024 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3025 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3026 current_function_decl to the new decl.
3028 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3030 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3031 helper macro for IEEE float128 hardware built-in functions.
3032 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3034 (TRUNCF128_ODD): Likewise.
3035 (ADDF128_ODD): Likewise.
3036 (SUBF128_ODD): Likewise.
3037 (MULF128_ODD): Likewise.
3038 (DIVF128_ODD): Likewise.
3039 (FMAF128_ODD): Likewise.
3040 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3041 UNSPEC_TRUNC_ROUND_TO_ODD.
3042 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3043 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3044 floating point round to odd instructions.
3045 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3046 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3047 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3048 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3049 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3050 (trunc<mode>sf2_hw): Change the truncate with round to odd
3051 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3052 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3053 to odd hardware instructions.
3054 (sub<mode>3_odd): Likewise.
3055 (mul<mode>3_odd): Likewise.
3056 (div<mode>3_odd): Likewise.
3057 (sqrt<mode>2_odd): Likewise.
3058 (fma<mode>4_odd): Likewise.
3059 (fms<mode>4_odd): Likewise.
3060 (nfma<mode>4_odd): Likewise.
3061 (nfms<mode>4_odd): Likewise.
3062 (trunc<mode>df2_odd): Change the truncate with round to odd
3063 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3065 * doc/extend.texi (PowerPC built-in functions): Update documentation
3066 for existing IEEE float128-bit built-in functions. Add built-in
3067 functions that generate the IEEE 128-bit floating point round to
3070 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3072 PR rtl-optimization/77729
3073 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3074 to (X&(C1&~C2))|C2 transformations.
3076 2017-10-03 Martin Jambor <mjambor@suse.cz>
3078 PR tree-optimization/82363
3079 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3080 mismatch, mark lacc written regardless of racc.
3082 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3084 PR tree-optimization/82381
3085 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3086 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3087 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3088 fallthrough into reassoc_stmt_dominates_stmt_p.
3091 * combine.c (combine_instructions): Don't combine in unreachable
3094 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3097 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3098 function to not use the have_cpu variable. Do not set cpu_index,
3099 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3101 (rs6000_valid_attribute_p): Remove duplicate initializations of
3102 old_optimize and func_optimize.
3103 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3104 (rs6000_activate_target_options): Make global.
3105 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3108 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3110 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3111 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3112 Set *poffset to 0 and *psize and *pmax_size to -1 if
3113 *poffset + *psize overflows in HOST_WIDE_INT.
3115 PR tree-optimization/82387
3116 PR tree-optimization/82388
3117 PR tree-optimization/82389
3118 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3119 instead of live_bytes non-NULL.
3121 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3124 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3127 2017-10-02 Richard Biener <rguenther@suse.de>
3129 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3130 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3131 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3133 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3135 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3136 requested precision matches the type's.
3137 * calls.c (alloc_max_size): Calculate the new candidate size as
3138 a widest_int and use wi::to_widest when comparing it with the
3139 current candidate size.
3140 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3141 zero rather than integer_zero_node.
3142 * match.pd: Check for a no-op conversion before using wi::add
3143 rather than after. Use tree_to_uhwi when summing small shift
3144 counts into an unsigned int.
3146 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3147 Alan Hayward <alan.hayward@arm.com>
3148 David Sherwood <david.sherwood@arm.com>
3151 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3152 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3153 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3154 POINTER_AND_FP_REGS.
3156 2017-10-02 Richard Biener <rguenther@suse.de>
3158 PR tree-optimization/82355
3159 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3160 a mapping for the enclosing loop but avoid generating one for
3162 (copy_bb_and_scalar_dependences): Remove premature codegen
3163 error on PHIs in blocks duplicated into multiple places.
3164 * graphite-scop-detection.c
3165 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3166 in the region use it as loop and nest to analyze the DR in.
3167 (try_generate_gimple_bb): Likewise.
3168 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3169 (add_loop_constraints): For blocks in a loop not in the region
3170 create a dimension with a single iteration.
3171 * sese.h (gbb_loop_at_index): Remove assert.
3173 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3175 * omp-expand.c (adjust_context_scope): New function.
3176 (expand_parallel_call): Call adjust_context_scope.
3178 2017-10-01 Jeff Law <law@redhat.com>
3180 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3181 dom_opt_dom_walker class with direct access to private members.
3182 Add comments. Call test_for_singularity.
3183 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3184 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3185 m_dummy_cond anymore.
3186 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3188 (pass_dominator:execute): Build the dummy_cond here and pass it
3189 to the dom_opt_dom_walker ctor.
3190 (test_for_singularity): New function.
3192 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3193 Maya Rashish <coypu@sdf.org>
3195 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3196 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3197 (alpha*-*-netbsd*) Use nbsd_tm_file.
3198 (arm*-*-netbsdelf*) Likewise.
3199 (i[34567]86-*-netbsdelf*) Likewise.
3200 (x86_64-*-netbsd*) Likewise.
3201 (mips*-*-netbsd*) Likewise.
3202 (powerpc-*-netbsd*) Likewise.
3203 (sh*-*-netbsd*) Likewise.
3204 (sparc-*-netbsdelf*) Likewise.
3205 (sparc64-*-netbsd*) Likewise.
3206 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3208 (vax-*-netbsdelf*) Likewise.
3209 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3210 (UINT_FAST8_TYPE) Likewise.
3211 (INT_FAST16_TYPE) Check CHAR_FAST16.
3212 (UINT_FAST16_TYPE) Likewise.
3214 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3217 * config/i386/i386.md
3218 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3219 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3220 *divmodsi4_zext_2): New define_insn_and_split.
3221 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3222 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3223 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3224 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3225 New define_insn_and_split.
3226 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3227 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3228 operands[1] having DImode when mode is SImode.
3230 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3231 always SImode for DIV and MOD in REG_EQUAL notes.
3233 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3236 * match.pd: Fix handling of NaNs in pattern.
3238 2017-09-29 Jeff Law <law@redhat.com>
3240 * sbitmap.c (bitmap_bit_in_range_p): New function.
3241 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3242 * tree-ssa-dse.c (live_bytes_read): New function.
3243 (dse_classify_store): Ignore reads of dead bytes.
3245 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3246 typos and whitespace errors.
3247 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3248 * config/s390/s390.c (s390_emit_prologue): Likewise.
3250 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3253 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3254 with a symbol for LRA.
3256 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3258 PR rtl-optimization/82338
3259 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3261 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3263 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3264 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3265 (qsort_chk): Declare.
3266 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3267 (qsort_chk): New function.
3269 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3271 PR tree-optimization/82337
3272 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3273 phi definition if the PHI result appears in an abnormal PHI.
3274 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3275 basis appears in an abnormal PHI.
3277 2017-09-29 Richard Biener <rguenther@suse.de>
3279 * graphite-isl-ast-to-gimple.c
3280 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3281 (binary_op_to_tree): Use it.
3282 (get_rename_from_scev): Likewise.
3283 (copy_loop_phi_nodes): Likewise.
3284 (copy_bb_and_scalar_dependences): Likewise.
3285 (translate_pending_phi_nodes): Likewise.
3287 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3290 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3291 for movabsq $(i32 << shift), r64.
3293 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3295 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3296 index when encoding %esp as %rsp to avoid 0x67 prefix.
3298 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3300 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3301 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3303 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3305 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3306 Extensions with more than 16 double VFP registers.
3307 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3308 to_clear_mask and all code related to it. Replace the remaining
3309 entry by a sbitmap and adapt code accordingly.
3311 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3313 * brig-builtins.def: Change pure attributes to const.
3315 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3317 * config.gcc (default_gnu_indirect_function): Default to yes for
3318 sparc*-*-linux* with glibc.
3320 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3322 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3323 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3324 when creating .init_array and .fini_array sections with priority
3327 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3330 * config/aarch64/aarch64.c
3331 (aarch64_builtin_support_vector_misalignment): Always return false
3332 when misalignment is unknown.
3334 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3336 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3337 this function to return false if the definition used by the swap
3338 instruction is artificial, or if the memory address from which the
3339 constant value is loaded is not represented by a base address held
3340 in a register or if the base address register is a frame or stack
3341 pointer. Additionally, return false if the base address of the
3342 loaded constant is a SYMBOL_REF but is not considered to be a
3344 (replace_swapped_load_constant): New function.
3345 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3346 loaded constant vector with a load of a swapped constant vector.
3348 2017-09-27 Carl Love <cel@us.ibm.com>
3350 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3351 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3352 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3355 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3357 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3358 first, always call autopref_rank_data otherwise.
3360 2017-09-27 Richard Biener <rguenther@suse.de>
3362 * graphite-scop-detection.c (find_scop_parameters): Move
3363 loop bound handling ...
3364 (gather_bbs::before_dom_children): ... here, avoiding the need
3365 to build scop_info->loop_nest.
3366 (record_loop_in_sese): Remove.
3367 * sese.h (sese_info_t::loop_nest): Remove.
3368 * sese.c (new_sese_info): Do not allocate loop_nest.
3369 (free_sese_info): Do not free loop_nest.
3371 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3374 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3375 lhs from calls if the lhs has addressable type.
3377 2017-09-27 Richard Biener <rguenther@suse.de>
3379 * graphite.h (scop::max_alias_set): New member.
3380 * graphite-scop-detection.c: Remove references to non-existing
3381 --param in comments.
3382 (build_alias_sets): Record the maximum alias set used for drs.
3383 (build_scops): Support zero as unlimited for
3384 --param graphite-max-arrays-per-scop.
3385 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3387 (build_poly_sr_1): ... here. Compute alias set based on the
3388 maximum alias set used for drs rather than
3389 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3391 2017-09-27 Richard Biener <rguenther@suse.de>
3393 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3394 --param loop-block-tile-size=0 to disable tiling.
3396 2017-09-27 Richard Biener <rguenther@suse.de>
3398 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3399 (graphite-max-nb-scop-params): Document special value zero.
3400 * domwalk.h (dom_walker::STOP): New symbolical constant.
3401 (dom_walker::dom_walker): Add optional parameter for bb to
3403 (dom_walker::~dom_walker): Declare.
3404 (dom_walker::before_dom_children): Document STOP return value.
3405 (dom_walker::m_user_bb_to_rpo): New member.
3406 (dom_walker::m_bb_to_rpo): Likewise.
3407 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3408 mapping here if not provided by the user.
3409 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3410 provided by the user.
3411 (dom_walker::STOP): Define.
3412 (dom_walker::walk): Do not compute bb to RPO mapping here.
3413 Support STOP return value from before_dom_children to stop
3415 * graphite-optimize-isl.c (optimize_isl): If the schedule
3416 is the same still generate code if -fgraphite-identity
3417 or -floop-parallelize-all are given.
3418 * graphite-scop-detection.c: Include cfganal.h.
3419 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3421 (gather_bbs::before_dom_children): Return STOP for BBs
3423 (build_scops): Compute bb to RPO mapping and pass it to
3424 the domwalk. Treat --param graphite-max-nb-scop-params=0
3425 as not limiting the number of params.
3426 * graphite.c (graphite_initialize): Remove limit on the
3427 number of basic-blocks in a function.
3428 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3429 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3430 default value of 10.
3432 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3434 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3435 Adjust code to eliminate needing to do the shift right 32-bits
3436 operation after XSCVDPSPN.
3438 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3440 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3441 ((X / Y) != 0 -> X >= Y): Likewise.
3443 2017-09-26 Carl Love <cel@us.ibm.com>
3445 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3446 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3447 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3448 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3449 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3450 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3451 definitions and overloading.
3452 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3453 statement for P9V_BUILTIN_XST_LEN_R.
3454 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3455 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3456 define_expand and define_insn for the instructions and builtins.
3457 * doc/extend.texi: Update the built-in documentation file for the new
3459 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3460 define_insn for the instructions
3462 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3465 * gcc/config/netbsd-protos.h: New file.
3466 * gcc/config/netbsd.c: New file.
3467 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3468 * gcc/config/t-netbsd: New file.
3469 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3470 (tmake_file) Add t-netbsd.
3471 (extra_objs) Add netbsd.o.
3473 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3477 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3479 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3481 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3482 sign extension from a vector register to a GPR by doing a 32-bit
3483 direct move and then an EXTSW.
3484 (extendsi<mode>2 splitter): Likewise.
3485 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3486 right or vector extract after doing XSCVDPSPN. Use
3487 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3489 (movdi_from_sf_zero_ext): Likewise.
3490 (reload_gpr_from_vsxsf): Likewise.
3491 (p8_mfvsrd_4_disf): Delete, no longer used.
3492 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3493 then needing to move the SFmode to a GPR to use the XSCVDPSP
3494 instruction instead of FRSP and XSCVDPSPN.
3495 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3496 it is adjacent to the other XSCVSPDP insns.
3497 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3498 SFmode to be in traditional Altivec registers.
3499 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3500 (vsx_xscvspdpn): Likewise.
3501 (vsx_xscvspdpn_scalar): Likewise.
3503 2017-09-26 Martin Jambor <mjambor@suse.cz>
3505 * tree-sra.c (compare_access_positions): Put integral types first,
3506 stabilize sorting of integral types, remove conditions putting
3507 non-full-precision integers last.
3508 (sort_and_splice_var_accesses): Disable scalarization if a
3509 non-integert would be represented by a non-full-precision integer.
3511 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3513 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3514 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3515 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3516 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
3517 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
3518 conditionals inside the function instead of around it. Call
3519 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
3520 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
3522 2017-09-26 Richard Biener <rguenther@suse.de>
3524 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
3526 (scop_detection::build_scop_breadth): ... this. Removed.
3527 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
3528 (scop_detection::harmful_stmt_in_bb): Likewise.
3529 (scop_detection::graphite_can_represent_stmt): Likewise.
3530 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
3531 (scop_detection::can_represent_loop): Remove recursion, fold in ...
3532 (scop_detection::can_represent_loop_1): ... this. Removed.
3533 (scop_detection::harmful_loop_in_region): Simplify after inlining
3534 the above and remove more quadraticness.
3535 (build_scops): Adjust.
3536 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
3539 2017-09-26 Jakub Jelinek <jakub@redhat.com>
3542 * config/i386/i386.c (ix86_print_operand_address_as): Only test
3543 REGNO (base) == SP_REG if base is a REG.
3546 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
3547 if it is different SSA_NAME.
3548 (optimize_range_tests_cmp_bitwise): New function.
3549 (optimize_range_tests): Call it.
3551 2017-09-26 Richard Biener <rguenther@suse.de>
3553 PR tree-optimization/82321
3554 * graphite.c (canonicalize_loop_closed_ssa): Properly check
3555 for the def being inside the loop.
3557 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3559 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
3561 * config/s390/s390-builtins.def: Fix constraint on op4.
3563 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3565 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
3566 independent expanders.
3567 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
3568 ("vec_ordered", "vec_unordered"): New expanders.
3570 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3572 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
3575 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3577 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
3578 vec_unpacks_lo_v16qi.
3579 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
3581 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3583 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
3584 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
3585 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
3587 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3589 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
3591 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
3593 ("*vec_slb<mode>"): New insn pattern.
3594 ("vec_shr_<mode>"): New expander.
3595 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
3596 and force the shift count operand to V16QImode.
3597 ("vec_srb<mode>"): Set shift count mode to V16QI.
3599 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3601 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
3602 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3603 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3605 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3608 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3610 2017-09-26 Richard Biener <rguenther@suse.de>
3612 PR tree-optimization/82320
3613 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3616 2017-09-25 Jeff Law <law@redhat.com>
3618 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3619 prototype for new argument.
3620 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3621 mostly extracted from rs6000_emit_allocate_stack.
3622 (rs6000_emit_probe_stack_range_stack_clash): New function.
3623 (rs6000_emit_allocate_stack): Call
3624 rs6000_emit_probe_stack_range_stack_clash as needed.
3625 (rs6000_emit_probe_stack_range): Add additional argument
3626 to call to gen_probe_stack_range{si,di}.
3627 (output_probe_stack_range): New.
3628 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3629 (output_probe_stack_range_stack_clash): New.
3630 (rs6000_emit_prologue): Emit notes into dump file as requested.
3631 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3632 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3633 Add additional operand and pass it to output_probe_stack_range.
3635 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3637 PR tree-optimization/82163
3638 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3639 (checking_verify_loop_closed_ssa): New parameter.
3640 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3641 (check_loop_closed_ssa_stmt): Delete.
3642 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3643 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3644 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3647 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3649 * brig-builtins.def: Treat HSAIL barrier builtins as
3650 setjmp/longjump style functions.
3652 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3654 * target.def (constant_alignment): New hook.
3655 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3656 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3657 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3658 * doc/tm.texi: Regenerate.
3659 * targhooks.h (default_constant_alignment): Declare.
3660 (constant_alignment_word_strings): Likewise.
3661 * targhooks.c (default_constant_alignment): New function.
3662 (constant_alignment_word_strings): Likewise.
3663 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3664 instead of CONSTANT_ALIGNMENT.
3665 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3666 (force_const_mem): Likewise.
3667 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3668 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3669 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3670 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3671 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3673 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3674 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3675 constant_alignment_word_strings.
3676 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3677 (CONSTANT_ALIGNMENT): Likewise.
3678 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3679 (arm_constant_alignment): New function.
3680 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3681 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3682 constant_alignment_word_strings.
3683 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3684 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3685 constant_alignment_word_strings.
3686 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3687 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3688 (cris_constant_alignment): New function.
3689 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3690 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3691 (epiphany_constant_alignment): New function.
3692 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3693 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3694 constant_alignment_word_strings.
3695 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3696 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3697 constant_alignment_word_strings.
3698 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3699 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3700 constant_alignment_word_strings.
3701 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3702 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3703 * config/i386/i386.c (ix86_constant_alignment): Make static.
3704 Use the same interface as the target hook.
3705 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3706 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3707 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3708 constant_alignment_word_strings.
3709 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3710 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3711 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3712 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3713 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3714 constant_alignment_word_strings.
3715 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3716 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3717 constant_alignment_word_strings.
3718 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3719 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3720 constant_alignment_word_strings.
3721 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3722 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3724 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3725 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3726 * config/mips/mips.c (mips_constant_alignment): New function.
3727 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3728 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3729 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3730 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3731 (mmix_constant_alignment): Make static. Use the same interface
3733 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3734 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3735 constant_alignment_word_strings.
3736 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3737 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3738 constant_alignment_word_strings.
3739 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3740 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3741 constant_alignment_word_strings.
3742 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3743 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3744 (rs6000_constant_alignment): New function.
3745 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3746 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3747 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3748 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3749 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3750 (rs6000_constant_alignment): New function.
3751 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3752 * config/s390/s390.c (s390_constant_alignment): New function.
3753 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3754 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3755 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3756 constant_alignment_word_strings.
3757 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3758 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3759 (sparc_constant_alignment): New function.
3760 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3761 * config/spu/spu.c (spu_constant_alignment): New function.
3762 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3763 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3764 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3765 constant_alignment_word_strings.
3766 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3767 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3768 constant_alignment_word_strings.
3769 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3770 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3771 constant_alignment_word_strings.
3772 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3773 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3774 (visium_constant_alignment): New function.
3775 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3776 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3777 (xtensa_constant_alignment): New function.
3778 * system.h (CONSTANT_ALIGNMENT): Poison.
3780 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3782 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3783 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3784 (rs6000_builtin_valid_without_lhs): New helper function.
3785 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3786 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3788 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3790 * target.h (vec_perm_indices): Use unsigned short rather than
3792 (auto_vec_perm_indices): Likewise.
3793 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3794 Use unsigned int rather than unsigned char.
3795 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3797 2017-09-25 Richard Biener <rguenther@suse.de>
3799 * cfgloop.h (sort_sibling_loops): Declare.
3800 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3801 (sort_sibling_loops): New function sorting the sibling loop list
3803 * graphite.c (graphite_transform_loops): Sort sibling loops.
3805 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3807 * target.def (vec_perm_const_ok): Change sel parameter to
3809 * optabs-query.c (can_vec_perm_p): Update accordingly.
3810 * doc/tm.texi: Regenerate.
3811 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3812 auto_vec_perm_indices and remove separate nelt field.
3813 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3814 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3815 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3816 (aarch64_expand_vec_perm_const): Update accordingly.
3817 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3818 to vec_perm_indices.
3819 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3820 auto_vec_perm_indices and remove separate nelt field.
3821 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3822 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3823 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3825 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3826 to vec_perm_indices.
3827 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3828 sel to vec_perm_indices.
3829 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3830 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3831 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3833 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3836 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3839 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3840 on the FUNCTION_DECL function context if it has a DIE that is a
3843 2017-09-25 Richard Biener <rguenther@suse.de>
3845 PR tree-optimization/82285
3846 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3849 2017-09-25 Tom de Vries <tom@codesourcery.com>
3853 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3856 2017-09-25 Richard Biener <rguenther@suse.de>
3858 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3859 ISL errors other than isl_error_quota happen. Dump if the
3860 schedule is the same.
3861 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3862 errors instead of aborting inside ISL.
3864 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3867 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3868 of libgcc_eh for m64.
3869 * config/i386/darwin64.h: Likewise.
3871 2017-09-25 Richard Biener <rguenther@suse.de>
3874 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3875 attribute for incomplete types nor twice for complete ones.
3877 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3880 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3881 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3884 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3887 * config/i386/i386.opt (mprefer-avx256): Use
3888 ix86_target_flags variable.
3889 * config/i386/i386.c (ix86_target_string): Move
3890 -mprefer-avx256 to flag2_opts.
3892 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3895 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3896 and x != -1 | y != -1 into (x & y) != -1.
3898 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3900 * config.gcc: Add new case statement to set
3901 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3902 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3903 s390x-*-linux* case statements. Added aarch64 to the list of
3904 supported architectures.
3906 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3908 PR tree-optimization/82289
3909 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3910 STMT_VINFO_RELEVANT_P.
3912 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3913 Alan Hayward <alan.hayward@arm.com>
3914 David Sherwood <david.sherwood@arm.com>
3916 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3917 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3918 (extract_range_from_binary_expr_1): Don't call
3919 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3921 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3922 Alan Hayward <alan.hayward@arm.com>
3923 David Sherwood <david.sherwood@arm.com>
3925 * target.def (preferred_vector_alignment): New hook.
3926 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3928 * doc/tm.texi: Regenerate.
3929 * targhooks.h (default_preferred_vector_alignment): Declare.
3930 * targhooks.c (default_preferred_vector_alignment): New function.
3931 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3933 (DR_TARGET_ALIGNMENT): New macro.
3934 (aligned_access_p): Update commentary.
3935 (vect_known_alignment_in_bytes): New function.
3936 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3938 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3939 Calculate the misalignment based on the target alignment rather than
3941 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3942 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3943 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3944 the target alignment, rather than masking the element misalignment
3945 with the number of elements in a vector. Also use the target
3946 alignment when calculating the maximum number of peels.
3947 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3948 instead of TYPE_ALIGN_UNIT.
3949 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3950 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3951 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3952 (vect_create_data_ref_ptr): Likewise.
3953 (vect_setup_realignment): Realign by ANDing with
3954 -DR_TARGET_MISALIGNMENT.
3955 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3956 the number of peels based on DR_TARGET_ALIGNMENT.
3957 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3958 with the guaranteed alignment boundary when deciding whether
3960 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3961 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3962 (ensure_base_align): Remove stmt_info parameter. Get the
3963 target base alignment from DR_TARGET_ALIGNMENT.
3964 (vectorizable_store): Update call accordingly. Interpret
3965 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3967 (vectorizable_load): Likewise.
3969 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3970 Alan Hayward <alan.hayward@arm.com>
3971 David Sherwood <david.sherwood@arm.com>
3973 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3974 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3975 (vect_enhance_data_refs_alignment): Likewise.
3977 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3979 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3980 error. Only quit immediately if parsing is complete.
3981 (BEGIN): Initialize fatal_err and parse_done.
3982 (begin fpu, end fpu): Check number of arguments.
3983 (begin arch, end arch): Likewise.
3984 (begin cpu, end cpu): Likewise.
3985 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3986 (optalias): Likewise.
3988 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3990 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3991 * config/arm/arm-isa.h: Delete. Move definitions to ...
3992 * arm-cpus.in: ... here. Use new feature and fgroup values.
3993 * config/arm/arm.c (arm_option_override): Use lower case for feature
3995 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3996 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3997 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3999 (print_isa_bits_for): New function.
4000 (gen_isa): New function.
4001 (gen_comm_data): Use print_isa_bits_for.
4002 (define feature): New keyword.
4003 (define fgroup): New keyword.
4004 * config/arm/t-arm (TM_H): Remove.
4005 (GTM_H): Add arm-isa.h.
4006 (arm-isa.h): Add rule to generate file.
4007 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4008 case for feature bit names.
4010 2017-09-22 Richard Biener <rguenther@suse.de>
4012 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4014 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4015 print of no dependency loops ...
4016 * graphite.c (graphite_transform_loops): ... here.
4017 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4019 (same_close_phi_node, remove_duplicate_close_phi,
4020 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4021 (canonicalize_loop_closed_ssa): ... here and simplify.
4022 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4023 (optimize_isl): Use dump_printf_loc to tell when we stopped
4024 optimizing because of an ISL timeout.
4026 2017-09-22 Richard Biener <rguenther@suse.de>
4028 PR tree-optimization/82291
4029 * tree-if-conv.c (predicate_mem_writes): Make sure to
4030 remove writes in blocks predicated with false.
4032 2017-09-22 Richard Biener <rguenther@suse.de>
4034 * sese.c: Include cfganal.h.
4035 (if_region_set_false_region): Remove.
4036 (create_if_region_on_edge): Likewise.
4037 (move_sese_in_condition): Re-implement without destroying
4040 2017-09-22 Richard Biener <rguenther@suse.de>
4042 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4043 Verify both BBs contain loop PHI nodes before dispatching to
4045 (graphite_regenerate_ast_isl): Do not recompute dominators,
4046 do not verify three times. Restructure for clarity.
4047 * graphite-scop-detection.c (same_close_phi_node,
4048 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4049 defined_in_loop_p, canonicalize_loop_closed_ssa,
4050 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4051 checking and SSA rewrite, move to ...
4052 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4053 (graphite_initialize): Do not pass in ctx, do not reset the
4054 SCEV cache, compute only dominators.
4055 (graphite_transform_loops): Allocate ISL ctx after
4056 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4057 Maintain post-dominators only around build_scops.
4058 * sese.c (if_region_set_false_region): Make static. Free
4059 and recompute dominators.
4060 (move_sese_in_condition): Assert we don't get called with
4061 post-dominators computed.
4062 * sese.h (if_region_set_false_region): Remove.
4064 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4066 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4067 mode attribute for TARGET_AVX512VL.
4069 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4071 * config/i386/i386.opt (mprefer-avx256): New option.
4072 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4074 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4075 for TARGET_PREFER_AVX256.
4076 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4078 2017-09-21 Jeff Law <law@redhat.com>
4080 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4081 Fix dump output if the only stack space is for pushed registers.
4083 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4085 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4088 2017-09-21 Martin Sebor <msebor@redhat.com>
4091 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4092 code (in C++) or code that triggers warnings.
4094 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4096 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4098 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4100 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4101 * sched-rgn.c: Ditto.
4102 * sel-sched-ir.c: Ditto.
4104 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4106 * toplev.h (set_random_seed): Adjust return type.
4107 * toplev.c (init_local_tick): Move eager initialization of random_seed
4108 to get_random_seed. Adjust comment.
4109 (init_random_seed): Inline to get_random_seed, delete.
4110 (get_random_seed): Initialize random_seed lazily.
4111 (set_random_seed): Do not return previous value.
4112 (print_switch_value): Do not call get_random_seed.
4114 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4116 * cgraph.c (delete_function_version): New, broken out from...
4117 (cgraph_node::delete_function_version): ...here. Rename to
4118 cgraph_node::delete_function_version_by_decl. Update all uses.
4119 (cgraph_node::remove): Call delete_function_version.
4121 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4124 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4125 VAR_DECLs to which addressable non-volatile parameters are mapped
4126 and for id->retvar after the return value assignment. Clear
4127 id->retval and id->retbnd after inlining.
4129 2017-09-21 Richard Biener <rguenther@suse.de>
4131 PR tree-optimization/82276
4132 PR tree-optimization/82244
4133 * tree-vrp.c (build_assert_expr_for): Set
4134 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4136 (remove_range_assertions): Revert earlier change.
4138 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4141 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4143 2017-09-21 Richard Biener <rguenther@suse.de>
4145 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4146 Restore valid IL after code generation errors.
4147 * graphite.c (graphite_transform_loops): Diagnose code
4148 generation issues as MSG_MISSED_OPTIMIZATION and continue
4149 with processing SCOPs.
4151 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4152 Alan Hayward <alan.hayward@arm.com>
4153 David Sherwood <david.sherwood@arm.com>
4155 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4156 rather than choosing between plus_constant and gen_rtx_<CODE>.
4157 * expr.c (emit_push_insn): Likewise.
4158 (expand_expr_real_2): Likewise.
4160 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4161 Alan Hayward <alan.hayward@arm.com>
4162 David Sherwood <david.sherwood@arm.com>
4164 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4166 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4167 Alan Hayward <alan.hayward@arm.com>
4168 David Sherwood <david.sherwood@arm.com>
4170 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4171 calling tree_to_uhwi.
4173 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4174 Alan Hayward <alan.hayward@arm.com>
4175 David Sherwood <david.sherwood@arm.com>
4177 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4178 INTEGER_CST rather than a negative test for ADDR_EXPR.
4180 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4181 Alan Hayward <alan.hayward@arm.com>
4182 David Sherwood <david.sherwood@arm.com>
4184 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4185 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4187 2017-09-21 Richard Biener <rguenther@suse.de>
4189 PR tree-optimization/71351
4190 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4191 graphite_create_new_loop_guard): Remove, fold remaining parts
4193 (translate_isl_ast_node_for): ... here and simplify.
4195 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4198 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4199 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4200 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4201 alternative always use QI mode, for -Os imov (=R,R) alternative
4202 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4205 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4206 Jeff Law <law@redhat.com>
4208 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4209 (allocate_stack_space): New function, partially extracted from
4211 (s390_emit_prologue): Track offset to most recent stack probe.
4212 Code to allocate space moved into allocate_stack_space.
4213 Dump actions when no stack is allocated.
4214 (s390_prologue_plus_offset): New function.
4215 (s390_emit_stack_probe): Likewise.
4217 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4219 * common.opt (Wa, Wl, Wp, g, gz=): Add
4221 (gno-column-info): Remove.
4222 (gcolumn-info): Drop RejectNegative.
4224 (gno-record-gcc-switches): Remove.
4225 (grecord-gcc-switches): Drop RejectNegative.
4226 (gno-split-dwarf): Remove.
4227 (gsplit-dwarf): Drop RejectNegative.
4228 (gno-strict-dwarf): Remove.
4229 (gstrict-dwarf): Drop RejectNegative.
4230 * config/darwin.opt (gfull, gused): Add RejectNegative.
4231 * dwarf2out.c (gen_producer_string): Drop
4232 gno-record-gcc-switches handler.
4233 * optc-gen.awk: Add g to prefixes with negative forms.
4234 * opts-common.c (remapping_prefix_p): New.
4235 (find_opt): Check it.
4236 (generate_canonical_option): Test g prefix.
4237 (option_map): Add -gno- mapping.
4238 (add_misspelling_candidates): Check remapping_prefix_p.
4240 2017-09-20 Jeff Law <law@redhat.com>
4242 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4243 thinko in stack clash protection support.
4245 * explow.c (compute_stack_clash_protection_loop_data): Use
4246 CONST_INT_P instead of explicit test. Verify object is a
4247 CONST_INT_P before looking at INTVAL.
4248 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4249 instead of explicit test.
4251 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4254 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4255 address instead of to r1 and r11.
4257 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4259 * config.gcc: Support "knm".
4260 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4261 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4263 * config/i386/i386.c (m_KNM): Define.
4264 (processor_target_table): Add "knm".
4266 (ix86_option_override_internal): Add "knm".
4267 (ix86_issue_rate): Add PROCESSOR_KNM.
4268 (ix86_adjust_cost): Ditto.
4269 (ia32_multipass_dfa_lookahead): Ditto.
4270 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4271 (fold_builtin_cpu): Add M_INTEL_KNM.
4272 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4273 (processor_type): Add PROCESSOR_KNM.
4274 * config/i386/x86-tune.def: Add m_KNM.
4275 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4277 2017-09-20 Richard Biener <rguenther@suse.de>
4279 PR tree-optimization/80213
4280 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4281 are allowed in empty BBs as well.
4282 (canonicalize_loop_closed_ssa): Also look for other complex
4284 (scop_detection::get_sese): Include the loop-closed PHI block
4286 (scop_detection::merge_sese): Remove code adding extra blocks.
4287 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4288 (build_scops): Assert the final returned scop is invalid.
4290 2017-09-20 Richard Biener <rguenther@suse.de>
4292 PR tree-optimization/82264
4293 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4295 (vn_phi_lookup): Likewise.
4296 (vn_phi_insert): Likewise.
4298 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4300 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4301 that fits into uhwi or shwi, add DW_AT_const_value regardless
4302 of early_dwarf without going through RTL, using add_AT_unsigned
4305 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4306 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4307 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4308 (DEBUG_MACRO_SECTION): Likewise.
4309 (DEBUG_LTO_MACRO_SECTION): Likewise.
4310 (DEBUG_STR_DWO_SECTION): Likewise.
4311 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4312 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4313 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4314 (DEBUG_LTO_LINE_STR_SECTION): Define.
4315 (init_sections_and_labels): Initialize debug_line_str_section
4316 variable. Initialize debug_loc_section for -gdwarf-5 to
4317 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4319 2017-09-20 Richard Biener <rguenther@suse.de>
4321 * graphite-sese-to-poly.c (extract_affine): Properly handle
4322 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4324 2017-09-20 Richard Biener <rguenther@suse.de>
4326 PR tree-optimization/81373
4327 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4328 Force SESE live-out defs to be handled even if they are
4331 2017-09-19 Jeff Law <law@redhat.com>
4333 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4334 nothing for stack adjustments with REG_STACK_CHECK.
4335 * sched-deps.c (parse_add_or_inc): Reject insns with
4336 REG_STACK_CHECK from dependency breaking.
4337 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4338 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4339 * reg-notes.def (STACK_CHECK): New note.
4341 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4342 (ix86_expand_prologue): Dump stack clash info as needed.
4343 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4345 * function.c (dump_stack_clash_frame_info): New function.
4346 * function.h (dump_stack_clash_frame_info): Prototype.
4347 (enum stack_clash_probes): New enum.
4349 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4350 flag_stack_clash_protection.
4351 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4352 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4353 (arm_frame_pointer_required): Likewise.
4354 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4355 (ia64_expand_prologue): Likewise.
4356 * config/mips/mips.c (mips_expand_prologue): Likewise.
4357 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4358 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4359 (sparc_flat_expand_prologue): Likewise.
4360 * config/spu/spu.c (spu_expand_prologue): Likewise.
4362 * explow.c: Include "params.h".
4363 (anti_adjust_stack_and_probe_stack_clash): New function.
4364 (get_stack_check_protect): Likewise.
4365 (compute_stack_clash_protection_loop_data): Likewise.
4366 (emit_stack_clash_protection_loop_start): Likewise.
4367 (emit_stack_clash_protection_loop_end): Likewise.
4368 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4369 Use anti_adjust_stack_and_probe_stack_clash.
4370 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4371 (emit_stack_clash_protection_loop_start): Likewise.
4372 (emit_stack_clash_protection_loop_end): Likewise.
4373 * rtl.h (get_stack_check_protect): Prototype.
4374 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4375 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4376 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4378 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4380 * doc/tm.texi: Rebuilt.
4381 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4382 get_stack_check_protect.
4383 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4384 * config/arm/arm.c (arm_expand_prologue): Likewise.
4385 (arm_frame_pointer_required): Likewise.
4386 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4387 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4388 * config/mips/mips.c (mips_expand_prologue): Likewise.
4389 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4390 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4391 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4392 (sparc_flat_expand_prologue): Likewise.
4394 * common.opt (-fstack-clash-protection): New option.
4395 * flag-types.h (enum stack_check_type): Note difference between
4396 -fstack-check= and -fstack-clash-protection.
4397 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4398 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4399 * toplev.c (process_options): Issue warnings/errors for cases
4400 not handled with -fstack-clash-protection.
4401 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4402 (-fstack-check): Note additional problem with -fstack-check=generic.
4403 Note that -fstack-check is primarily for Ada and refer users
4404 to -fstack-clash-protection for stack-clash-protection.
4405 Document new params for stack clash protection.
4407 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4409 * config/i386/i386.c (ix86_split_long_move): Do not handle
4410 address used for LEA in a special way.
4412 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4414 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4416 2017-09-19 Martin Sebor <msebor@redhat.com>
4419 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4420 of incompatible types.
4422 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4424 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4425 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4426 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4427 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4429 2017-09-19 Richard Biener <rguenther@suse.de>
4431 PR tree-optimization/82244
4432 * tree-vrp.c (remove_range_assertions): Do not propagate
4433 a constant to abnormals but replace the assert with a copy.
4435 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4437 PR rtl-optimization/57878
4438 PR rtl-optimization/68988
4439 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4440 avoidance test involving non_reload_pseudos. Move frequency test
4441 below the general fragmentation avoidance test.
4443 2017-09-19 Richard Biener <rguenther@suse.de>
4445 PR tree-optimization/69728
4446 * graphite-sese-to-poly.c (schedule_error): New global.
4447 (add_loop_schedule): Handle empty domain by failing the
4449 (build_original_schedule): Handle schedule_error.
4451 2017-09-19 Richard Biener <rguenther@suse.de>
4453 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4454 Do not iterate to sibling loops but only to siblings of inner
4457 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4460 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4461 that the registers are different.
4463 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4465 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4466 to processor_model and "amdfam17h" to arch_names_table.
4467 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4469 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4472 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4474 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4475 Alan Hayward <alan.hayward@arm.com>
4476 David Sherwood <david.sherwood@arm.com>
4478 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4480 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4482 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4483 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4484 vect_schedule_slp_instance.
4485 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4486 Update call to vect_slp_analyze_node_operations. Simplify return
4488 (vect_slp_analyze_bb_1): Update call accordingly.
4489 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4490 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4491 (vect_schedule_slp): Update call accordingly.
4493 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4494 Alan Hayward <alan.hayward@arm.com>
4495 David Sherwood <david.sherwood@arm.com>
4497 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4498 with types that aren't in fact scalar.
4500 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4502 * tree-vect-slp.c (vect_record_max_nunits): New function,
4504 (vect_build_slp_tree_1): ...here.
4505 (vect_build_slp_tree_2): Call it for phis too.
4507 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4509 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4510 to vect_get_vec_def_for_operand when getting the mask operand.
4512 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4513 Alan Hayward <alan.hayward@arm.com>
4514 David Sherwood <david.sherwood@arm.com>
4516 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
4519 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4520 Alan Hayward <alan.hayward@arm.com>
4521 David Sherwood <david.sherwood@arm.com>
4523 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
4524 calculation for vector booleans.
4526 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4527 Alan Hayward <alan.hayward@arm.com>
4528 David Sherwood <david.sherwood@arm.com>
4530 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
4532 (vect_transform_stmt): ...here.
4533 (vect_analyze_stmt): Use it instead of calling
4534 vectorizable_live_operation directly.
4536 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
4538 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
4539 non-SIMT targets in acc vector loops.
4541 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
4543 * configure.ac: Add arc and check if assembler supports gdwarf2.
4544 * configure: Regenerate.
4546 2017-09-18 Richard Biener <rguenther@suse.de>
4548 PR tree-optimization/82220
4549 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
4550 epilogue niters from the min_profitable_iters compute.
4552 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4555 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
4556 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
4557 (ix86_init_pic_reg): Revert 2017-09-01 changes.
4559 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4562 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
4563 switching to a new text section.
4565 2017-09-18 Richard Biener <rguenther@suse.de>
4567 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
4569 (build_alias_set): Reject aliases with no access function.
4571 2017-09-18 Richard Biener <rguenther@suse.de>
4573 PR tree-optimization/79622
4574 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
4576 (build_cross_bb_scalars_use): Likewise.
4578 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
4580 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
4582 2017-09-18 Alan Modra <amodra@gmail.com>
4585 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
4586 stack_pointer_rtx for count 0. Update comments. Break up
4587 large rtl expression.
4589 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
4591 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
4592 Increase to 20 bytes.
4593 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
4594 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
4595 or avx version of the stub.
4597 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
4600 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
4601 compute the minimum stack alignment. Also update preferred stack
4602 boundary for leaf functions.
4604 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4606 PR tree-optimization/82228
4607 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4610 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4612 * common/config/nds32/nds32-common.c
4613 (nds32_option_optimization_table): Refine formatting.
4614 (nds32_option_optimization_table): Use -fsched-pressure and
4615 -fomit-frame-pointer for specific optimization level.
4617 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4619 * config/nds32/nds32.c: Refine formatting and comments.
4620 * config/nds32/nds32.h: Likewise.
4621 * config/nds32/nds32.md: Likewise.
4622 * config/nds32/nds32-cost.c: Likewise.
4623 * config/nds32/nds32-isr.c: Likewise.
4624 * config/nds32/nds32-md-auxiliary.c: Likewise.
4625 * config/nds32/nds32-multiple.md: Likewise.
4626 * config/nds32/nds32-predicates.c: Likewise.
4628 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4629 Jakub Jelinek <jakub@redhat.com>
4631 Add support for -std=c++2a.
4632 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4634 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4636 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4639 * doc/extend.texi (Common Function Attributes): Add
4640 references to ARM, AArch64, and S/390 specific attributes.
4641 (Function Specific Option Pragmas): Add AArch64 and S/390
4642 to list of back ends that support the target pragma.
4644 2017-09-15 Nathan Sidwell <nathan@acm.org>
4646 * doc/standards.texi: Fix C++17 description. Update URLs for
4649 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4651 * common.opt (Wcast-align=strict): New warning option.
4652 * doc/invoke.texi: Document -Wcast-align=strict.
4654 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4656 * cgraph.h (cgraph_thunk_info): Add comments.
4657 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4658 assert for VIRTUAL_* arguments stricter.
4660 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4662 PR tree-optimization/71026
4663 * match.pd: Move RDIV patterns from fold-const.c
4664 * fold-const.c (distribute_real_division): Removed.
4665 (fold_binary_loc): Remove calls to distribute_real_divison.
4667 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4669 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4670 c++1z and gnu++1z as deprecated. Change other references to
4671 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4672 Change -Wc++1z-compat to -Wc++17-compat.
4673 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4674 * dwarf2out.c (highest_c_language): Handle C++17.
4675 (gen_compile_unit_die): Likewise.
4677 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4679 PR rtl-optimization/82192
4680 * combine.c (make_extraction): Don't look through non-paradoxical
4681 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4684 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4685 Alan Hayward <alan.hayward@arm.com>
4686 David Sherwood <david.sherwood@arm.com>
4688 * target.def (function_arg_offset): New hook.
4689 * targhooks.h (default_function_arg_offset): Declare.
4690 * targhooks.c (default_function_arg_offset): New function.
4691 * function.c (locate_and_pad_parm): Use
4692 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4693 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4694 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4695 * doc/tm.texi: Regenerate.
4696 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4697 * config/spu/spu.c (spu_function_arg_offset): New function.
4698 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4699 * system.h (FUNCTION_ARG_OFFSET): Poison.
4701 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4702 Alan Hayard <alan.hayward@arm.com>
4703 David Sherwood <david.sherwood@arm.com>
4705 * target.def (truly_noop_truncation): New hook.
4706 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4707 than TRULY_NOOP_TRUNCATION.
4708 * hooks.h (hook_bool_uint_uint_true): Declare.
4709 * hooks.c (hook_bool_uint_uint_true): New function.
4710 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4711 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4712 * doc/tm.texi: Regenerate.
4713 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4714 rather than TRULY_NOOP_TRUNCATION in comments.
4715 (simplify_comparison): Likewise.
4716 (record_truncated_value): Likewise.
4717 * expmed.c (extract_bit_field_1): Likewise.
4718 (extract_split_bit_field): Likewise.
4719 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4720 instead of TRULY_NOOP_TRUNCATION.
4721 * function.c (assign_parm_setup_block): Likewise.
4722 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4723 * rtlhooks.c: Include target.h.
4724 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4725 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4726 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4727 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4728 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4729 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4730 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4731 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4732 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4733 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4734 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4735 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4736 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4737 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4738 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4739 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4740 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4741 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4742 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4743 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4744 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4745 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4746 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4747 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4748 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4749 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4750 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4751 rather than TRULY_NOOP_TRUNCATION in comments.
4752 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4753 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4754 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4755 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4756 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4757 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4758 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4759 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4760 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4761 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4762 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4763 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4764 rather than TRULY_NOOP_TRUNCATION in comments.
4765 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4766 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4767 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4768 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4769 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4770 TRULY_NOOP_TRUNCATION condition.
4771 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4772 (TRULY_NOOP_TRUNCATION): Delete.
4773 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4774 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4775 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4776 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4777 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4778 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4779 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4780 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4781 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4782 rather than TRULY_NOOP_TRUNCATION in comments.
4783 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4784 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4785 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4786 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4787 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4788 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4790 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4793 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4794 (*cmp_ior): Likewise.
4795 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4796 (*ior_scc_scc_cmp): Likewise.
4797 (*and_scc_scc): Likewise.
4798 (*and_scc_scc_cmp): Likewise.
4800 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4801 Alan Hayard <alan.hayward@arm.com>
4802 David Sherwood <david.sherwood@arm.com>
4804 * target.def (can_change_mode_class): New hook.
4805 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4806 (hard_regno_nregs): Likewise.
4807 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4808 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4809 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4810 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4811 (LOAD_EXTEND_OP): Update accordingly.
4812 * doc/tm.texi: Regenerate.
4813 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4814 CANNOT_CHANGE_MODE_CLASS.
4815 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4816 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4817 * combine.c (simplify_set): Update accordingly.
4818 * emit-rtl.c (validate_subreg): Likewise.
4819 * recog.c (general_operand): Likewise.
4820 * regcprop.c (mode_change_ok): Likewise.
4821 * reload1.c (choose_reload_regs): Likewise.
4822 (inherit_piecemeal_p): Likewise.
4823 * rtlanal.c (simplify_subreg_regno): Likewise.
4824 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4825 instead of CANNOT_CHANGE_MODE_CLASS.
4826 (reload_cse_simplify_operands): Likewise.
4827 * reload.c (push_reload): Use targetm.can_change_mode_class
4828 instead of CANNOT_CHANGE_MODE_CLASS.
4829 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4830 REG_CANNOT_CHANGE_MODE_P.
4831 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4832 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4833 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4834 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4835 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4836 (arm_can_change_mode_class): New function.
4837 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4838 than CANNOT_CHANGE_MODE_CLASS in comments.
4839 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4840 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4841 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4842 (ix86_can_change_mode_class): ...this new function, inverting the
4843 sense of the return value.
4844 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4845 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4846 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4847 (ia64_can_change_mode_class): New function.
4848 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4849 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4850 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4851 (m32c_can_change_mode_class): ...this new function, inverting the
4852 sense of the return value.
4853 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4854 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4855 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4856 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4857 (mips_can_change_mode_class): ...this new function, inverting the
4858 sense of the return value.
4859 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4860 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4861 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4862 (msp430_can_change_mode_class): New function.
4863 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4864 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4865 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4866 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4867 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4868 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4869 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4870 (pa_cannot_change_mode_class): Replace with...
4871 (pa_can_change_mode_class): ...this new function, inverting the
4872 sense of the return value.
4873 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4874 than CANNOT_CHANGE_MODE_CLASS in comments.
4875 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4876 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4877 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4878 (pdp11_cannot_change_mode_class): Replace with...
4879 (pdp11_can_change_mode_class): ...this new function, inverting the
4880 sense of the return value.
4881 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4882 * config/powerpcspe/powerpcspe-protos.h
4883 (rs6000_cannot_change_mode_class_ptr): Delete.
4884 * config/powerpcspe/powerpcspe.c
4885 (rs6000_cannot_change_mode_class_ptr): Delete.
4886 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4887 (rs6000_option_override_internal): Assign to
4888 targetm.can_change_mode_class instead of
4889 rs6000_cannot_change_mode_class_ptr.
4890 (rs6000_cannot_change_mode_class): Replace with...
4891 (rs6000_can_change_mode_class): ...this new function, inverting the
4892 sense of the return value.
4893 (rs6000_debug_cannot_change_mode_class): Replace with...
4894 (rs6000_debug_can_change_mode_class): ...this new function.
4895 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4896 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4897 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4898 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4899 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4901 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4902 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4903 (rs6000_option_override_internal): Assign to
4904 targetm.can_change_mode_class instead of
4905 rs6000_cannot_change_mode_class_ptr.
4906 (rs6000_cannot_change_mode_class): Replace with...
4907 (rs6000_can_change_mode_class): ...this new function, inverting the
4908 sense of the return value.
4909 (rs6000_debug_cannot_change_mode_class): Replace with...
4910 (rs6000_debug_can_change_mode_class): ...this new function.
4911 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4912 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4913 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4914 (s390_can_change_mode_class): ...this new function, inverting the
4915 sense of the return value.
4916 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4917 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4918 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4919 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4920 (sh_cannot_change_mode_class): Replace with...
4921 (sh_can_change_mode_class): ...this new function, inverting the
4922 sense of the return value.
4923 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4924 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4925 (sparc_can_change_mode_class): New function.
4926 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4927 * config/spu/spu.c (spu_can_change_mode_class): New function.
4928 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4929 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4930 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4931 (visium_can_change_mode_class): New function.
4932 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4934 2017-09-15 Richard Biener <rguenther@suse.de>
4936 PR tree-optimization/82217
4937 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4938 but not undefined case.
4940 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4943 * postreload.c (reload_cse_simplify_operands): Skip
4944 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4946 2017-09-15 Richard Biener <rguenther@suse.de>
4948 PR tree-optimization/68823
4949 * graphite-scop-detection.c (build_alias_set): If we have a
4950 possible dependence check whether we can handle them by just
4951 looking at the DRs DR_ACCESS_FNs.
4952 (build_scops): If build_alias_set fails, fail the SCOP.
4954 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4956 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4957 to support float128 built-in functions that require the ISA 3.0
4959 (BU_FLOAT128_3_HW): Likewise.
4960 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4962 (FMAF128): Likewise.
4964 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4965 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4966 floating point instructions.
4967 (rs6000_invalid_builtin): Likewise.
4968 (rs6000_builtin_mask_names): Likewise.
4969 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4970 (RS6000_BTM_FLOAT128_HW): Likewise.
4971 (RS6000_BTM_COMMON): Likewise.
4972 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4974 * doc/extend.texi (RS/6000 built-in functions): Document the
4975 IEEE 128-bit floating point square root and fused multiply-add
4978 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4980 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4981 reg (r2) isn't in the set of registers defined in the prologue.
4983 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4984 Alan Hayward <alan.hayward@arm.com>
4985 David Sherwood <david.sherwood@arm.com>
4987 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4988 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4989 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4990 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4991 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4993 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4994 max_vectorization_factor.
4995 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4997 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4998 Alan Hayward <alan.hayward@arm.com>
4999 David Sherwood <david.sherwood@arm.com>
5001 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5002 (vect_worthwhile_without_simd_p): Declare.
5003 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5004 (vectorizable_reduction): Use it.
5005 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5006 (vectorizable_operation): Likewise.
5008 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5009 Alan Hayward <alan.hayward@arm.com>
5010 David Sherwood <david.sherwood@arm.com>
5012 * tree-vectorizer.h (vect_get_num_copies): New function.
5013 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5014 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5015 (vectorizable_induction): Likewise.
5016 (vectorizable_live_operation): Likewise.
5017 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5018 (vectorizable_bswap): Likewise.
5019 (vectorizable_call): Likewise.
5020 (vectorizable_conversion): Likewise.
5021 (vectorizable_assignment): Likewise.
5022 (vectorizable_shift): Likewise.
5023 (vectorizable_operation): Likewise.
5024 (vectorizable_store): Likewise.
5025 (vectorizable_load): Likewise.
5026 (vectorizable_condition): Likewise.
5027 (vectorizable_comparison): Likewise.
5028 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5030 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5031 Alan Hayward <alan.hayward@arm.com>
5032 David Sherwood <david.sherwood@arm.com>
5034 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5035 of vect_init_vector.
5037 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5038 Alan Hayward <alan.hayward@arm.com>
5039 David Sherwood <david.sherwood@arm.com>
5041 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5042 an inline wrapper that provides a location.
5043 (gimple_build_vector): Likewise.
5044 * gimple-fold.c (gimple_build_vector_from_val): New function.
5045 (gimple_build_vector): Likewise.
5046 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5047 functions to build the initial value. Always return a gimple value.
5048 (get_initial_defs_for_reduction): Likewise. Only compute
5050 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5051 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5052 (vectorizable_induction): Use gimple_build_vector rather than
5055 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5056 Alan Hayward <alan.hayward@arm.com>
5057 David Sherwood <david.sherwood@arm.com>
5059 * target.h (vec_perm_indices): New typedef.
5060 (auto_vec_perm_indices): Likewise.
5061 * optabs-query.h: Include target.h
5062 (can_vec_perm_p): Take a vec_perm_indices *.
5063 * optabs-query.c (can_vec_perm_p): Likewise.
5064 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5065 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5066 * tree-vect-generic.c (lower_vec_perm): Likewise.
5067 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5068 (vect_grouped_load_supported): Likewise.
5069 (vect_shift_permute_load_chain): Likewise.
5070 (vect_permute_store_chain): Use auto_vec_perm_indices.
5071 (vect_permute_load_chain): Likewise.
5072 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5073 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5074 Update uses of can_vec_perm_p.
5075 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5076 mode with a number of elements. Take a vec_perm_indices *.
5077 (vect_create_epilog_for_reduction): Update accordingly.
5078 Use auto_vec_perm_indices.
5079 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5080 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5081 (vect_transform_slp_perm_load): Likewise.
5082 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5083 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5084 (vect_gen_perm_mask_checked): Likewise.
5085 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5086 (vect_gen_perm_mask_checked): Likewise.
5087 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5088 (vectorizable_store): Likewise.
5089 (vectorizable_load): Likewise.
5090 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5091 (vectorizable_bswap): Likewise.
5093 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5094 Alan Hayward <alan.hayward@arm.com>
5095 David Sherwood <david.sherwood@arm.com>
5097 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5098 * tree.c (build_vector): Likewise.
5099 (build_vector_from_ctor): Update accordingly.
5100 (build_vector_from_val): Likewise.
5101 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5102 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5103 * tree-vect-generic.c (add_rshift): Likewise.
5104 (expand_vector_divmod): Likewise.
5105 (optimize_vector_constructor): Likewise.
5106 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5107 (vect_transform_slp_perm_load): Likewise.
5108 (vect_schedule_slp_instance): Likewise.
5109 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5110 (vectorizable_call): Likewise.
5111 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5112 * expmed.c (make_tree): Likewise.
5113 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5114 a vector passed to build_vector.
5115 (fold_convert_const): Likewise.
5116 (exact_inverse): Likewise.
5117 (fold_ternary_loc): Likewise.
5118 (fold_relational_const): Likewise.
5119 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5120 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5121 (const_unop): Likewise. Store the reduction accumulator in a
5122 variable rather than an array.
5123 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5124 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5125 the new vector, rather than constructing it after the input arrays.
5126 (native_interpret_vector): Use auto_vec<tree> when building
5127 a vector passed to build_vector. Add elements in order.
5128 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5129 auto_vec<tree> when building a vector passed to build_vector.
5130 (vect_create_epilog_for_reduction): Likewise.
5131 (vectorizable_induction): Likewise.
5132 (get_initial_def_for_reduction): Likewise. Fix indentation of
5134 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5136 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5137 passed to build_vector.
5139 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5140 Alan Hayward <alan.hayward@arm.com>
5141 David Sherwood <david.sherwood@arm.com>
5143 * tree-core.h (tree_base::u): Add an "nelts" field.
5144 (tree_vector): Use VECTOR_CST_NELTS as the length.
5145 * tree.c (tree_size): Likewise.
5146 (make_vector): Initialize VECTOR_CST_NELTS.
5147 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5148 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5149 TYPE_VECTOR_SUBPARTS.
5150 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5151 as the number of units, setting it from VECTOR_CST_NELTS.
5152 (const_vector_from_tree): Likewise.
5153 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5154 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5155 (fold_negate_expr_1): Likewise.
5156 (fold_convert_const): Likewise.
5157 (const_binop): Likewise. Differentiate the number of output and
5159 (const_unop): Likewise.
5160 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5161 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5162 in cases that did the opposite.
5164 2017-09-14 Richard Biener <rguenther@suse.de>
5166 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5169 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5171 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5173 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5176 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5177 if and where to split a bb, except for splitting before debug insn
5178 sequences followed by non-label real insn. Delete debug insns
5179 in between basic blocks.
5181 * combine.c (make_compound_operation_int): Formatting fixes.
5183 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5184 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5185 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5186 * config/sol2.h (LINK_EH_SPEC): Likewise.
5187 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5188 * config/s390/linux.h (LINK_SPEC): Likewise.
5189 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5190 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5191 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5192 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5193 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5194 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5195 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5196 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5198 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5200 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5202 (ENDFILE_LINUX_SPEC): Likewise.
5203 (LINK_EH_SPEC): Likewise.
5204 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5205 (LINK_OS_LINUX_SPEC32): Likewise.
5206 (LINK_OS_LINUX_SPEC64): Likewise.
5207 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5208 (LINK_OS_LINUX_SPEC): Likewise.
5210 2017-09-13 Martin Liska <mliska@suse.cz>
5213 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5214 CASE_HIGH is NULL_TREE.
5216 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5217 Alan Hayward <alan.hayward@arm.com>
5218 David Sherwood <david.sherwood@arm.com>
5220 * target.def (secondary_memory_needed): New hook.
5221 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5222 instead of SECONDARY_MEMORY_NEEDED.
5223 (secondary_memory_needed_mode): Likewise.
5224 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5225 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5226 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5227 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5228 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5229 * doc/tm.texi: Regenerate.
5230 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5231 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5232 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5233 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5234 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5235 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5236 mode argument first and change the reg_class arguments to reg_class_t.
5237 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5238 Make static. Update the call to inline_secondary_memory_needed.
5239 (ix86_register_move_cost): Update the call to
5240 inline_secondary_memory_needed.
5241 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5242 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5244 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5245 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5247 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5248 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5249 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5250 and match hook interface. Add comment from mips.h.
5251 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5252 * config/mmix/mmix.md (truncdfsf2): Refer to
5253 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5255 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5256 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5257 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5258 (pa_secondary_memory_needed): New function.
5259 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5260 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5261 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5262 (pdp11_secondary_memory_needed): Make static and match hook interface.
5263 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5264 * config/powerpcspe/powerpcspe-protos.h
5265 (rs6000_secondary_memory_needed_ptr): Delete.
5266 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5268 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5269 (rs6000_option_override_internal): Assign to
5270 targetm.secondary_memory_needed rather than
5271 rs6000_secondary_memory_needed_ptr.
5272 (rs6000_secondary_memory_needed): Match hook interface.
5273 (rs6000_debug_secondary_memory_needed): Likewise.
5274 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5275 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5276 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5277 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5278 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5279 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5281 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5282 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5283 (rs6000_option_override_internal): Assign to
5284 targetm.secondary_memory_needed rather than
5285 rs6000_secondary_memory_needed_ptr.
5286 (rs6000_secondary_memory_needed): Match hook interface.
5287 (rs6000_debug_secondary_memory_needed): Likewise.
5288 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5289 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5290 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5291 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5292 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5293 (sparc_secondary_memory_needed): New function.
5294 * lra-constraints.c (check_and_process_move): Refer to
5295 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5297 (curr_insn_transform): Likewise.
5298 (process_alt_operands): Use targetm.secondary_memory_needed
5299 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5300 (check_secondary_memory_needed_p): Likewise.
5301 (choose_split_class): Likewise.
5302 * reload.c: Unconditionally include code that was previously
5303 conditional on SECONDARY_MEMORY_NEEDED.
5304 (push_secondary_reload): Use targetm.secondary_memory_needed
5305 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5306 (push_reload): Likewise.
5307 * reload1.c: Unconditionally include code that was previously
5308 conditional on SECONDARY_MEMORY_NEEDED.
5309 (choose_reload_regs): Use targetm.secondary_memory_needed
5310 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5311 (gen_reload): Likewise.
5312 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5314 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5315 Alan Hayward <alan.hayward@arm.com>
5316 David Sherwood <david.sherwood@arm.com>
5318 * target.def (secondary_memory_needed_mode): New hook:
5319 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5320 * targhooks.h (default_secondary_memory_needed_mode): New function.
5321 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5322 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5323 * doc/tm.texi: Regenerate.
5324 * lra-constraints.c (check_and_process_move): Use
5325 targetm.secondary_memory_needed_mode instead of
5326 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5327 (curr_insn_transform): Likewise.
5328 * reload.c (get_secondary_mem): Likewise.
5329 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5330 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5332 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5333 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5334 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5335 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5336 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5338 * config/powerpcspe/powerpcspe-protos.h
5339 (rs6000_secondary_memory_needed_mode): Delete.
5340 * config/powerpcspe/powerpcspe.c
5341 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5342 (rs6000_secondary_memory_needed_mode): Make static.
5343 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5344 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5346 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5348 (rs6000_secondary_memory_needed_mode): Make static.
5349 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5350 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5351 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5352 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5353 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5355 (sparc_secondary_memory_needed_mode): New function.
5356 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5358 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5360 * config/aarch64/constraints.md (Umq): New constraint.
5361 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5363 (mov<mode>): Update condition.
5365 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5367 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5368 when bitposition is the same.
5370 2017-09-13 Richard Biener <rguenther@suse.de>
5372 * dwarf2out.c (output_die_symbol): Remove.
5373 (output_die): Do not output a DIEs symbol.
5375 2017-09-13 Richard Biener <rguenther@suse.de>
5378 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5379 default-def to avoid breaking iterator update with the weird
5380 interaction with cgraph_update_edges_for_call_stmt_node.
5382 2017-09-13 Richard Biener <rguenther@suse.de>
5384 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5385 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5386 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5387 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5389 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5391 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5392 Disable pc relative literal load irrespective of
5393 TARGET_FIX_ERR_A53_84341 for default.
5395 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5397 * config/sparc/sparc.c (output_return): Output the source location of
5398 the insn in the delay slot, if any.
5399 (output_sibcall): Likewise.
5401 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5404 * common.opt (-static-pie): New alias.
5405 (shared): Negate static-pie.
5406 (-no-pie): Update help text.
5408 (static-pie): New option.
5409 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5410 -static-pie support.
5411 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5412 (LINK_EH_SPEC): Likewise.
5413 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5414 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5415 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5416 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5417 (init_gcc_specs): Likewise.
5418 (init_spec): Likewise.
5419 (display_help): Update help message for -pie.
5420 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5423 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5425 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5426 (movdi_aarch64): Likewise.
5427 (movti_aarch64): Likewise.
5429 2017-09-12 Simon Wright <simon@pushface.org>
5432 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5433 calculation of the minor version, always output as 0.
5435 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5438 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5439 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5440 on it early, rather than manual conversion late. For
5441 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5442 instead of performing manual conversion.
5444 2017-09-12 Carl Love <cel@us.ibm.com>
5446 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5447 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5449 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5450 VMULOSW): Add definitions.
5451 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5452 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5453 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5454 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5455 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5457 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5459 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5461 (movti_aarch64): Likewise.
5462 (movdf_aarch64): Likewise.
5463 (movtf_aarch64): Likewise.
5464 (load_pairdi): Likewise.
5465 (store_pairdi): Likewise.
5466 (load_pairdf): Likewise.
5467 (store_pairdf): Likewise.
5468 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5469 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5470 (ldr_got_small_<mode>): Likewise.
5471 (ldr_got_small_28k_<mode>): Likewise.
5472 (ldr_got_tiny): Likewise.
5473 * config/aarch64/iterators.md (ldst_sz): New.
5474 (ldpstp_sz): Likewise.
5475 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5477 (thunderx_load): Split load_8 to load_16.
5478 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5480 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5481 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5482 (xgene1_store_pair): Split store_8 to store_16.
5483 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5484 (falkor_st_0_st_sd): Split store_8 to store_16.
5486 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5488 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5489 and store1/2/3/4 to store_4/8/12/16.
5490 * config/aarch64/aarch64.md: Update for rename.
5491 * config/arm/arm.md: Likewise.: Likewise.
5492 * config/arm/arm.c: Likewise.
5493 * config/arm/thumb1.md: Likewise.
5494 * config/arm/thumb2.md: Likewise.
5495 * config/arm/vfp.md: Likewise.
5496 * config/arm/arm-generic.md: Likewise.
5497 * config/arm/arm1020e.md: Likewise.
5498 * config/arm/arm1026ejs.md: Likewise.
5499 * config/arm/arm1136jfs.md: Likewise.
5500 * config/arm/arm926ejs.md: Likewise.
5501 * config/arm/cortex-a15.md: Likewise.
5502 * config/arm/cortex-a17.md: Likewise.
5503 * config/arm/cortex-a5.md: Likewise.
5504 * config/arm/cortex-a53.md: Likewise.
5505 * config/arm/cortex-a57.md: Likewise.
5506 * config/arm/cortex-a7.md: Likewise.
5507 * config/arm/cortex-a8.md: Likewise.
5508 * config/arm/cortex-a9.md: Likewise.
5509 * config/arm/cortex-m4.md: Likewise.
5510 * config/arm/cortex-m7.md: Likewise.
5511 * config/arm/cortex-r4.md: Likewise.
5512 * config/arm/exynos-m1.md: Likewise.
5513 * config/arm/fa526.md: Likewise.
5514 * config/arm/fa606te.md: Likewise.
5515 * config/arm/fa626te.md: Likewise.
5516 * config/arm/fa726te.md: Likewise.
5517 * config/arm/fmp626.md: Likewise.
5518 * config/arm/iwmmxt.md: Likewise.
5519 * config/arm/ldmstm.md: Likewise.
5520 * config/arm/marvell-pj4.md: Likewise.
5521 * config/arm/xgene1.md: Likewise.
5522 * config/aarch64/thunderx.md: Likewise.
5523 * config/aarch64/thunderx2t99.md: Likewise.
5524 * config/aarch64/falkor.md: Likewise.
5526 2017-09-12 Martin Liska <mliska@suse.cz>
5528 * attribs.c (private_lookup_attribute): New function.
5529 * attribs.h (private_lookup_attribute): Declared here.
5530 (lookup_attribute): Called from this place.
5532 2017-09-12 Richard Biener <rguenther@suse.de>
5534 PR tree-optimization/82157
5535 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
5536 stmts with side-effects.
5538 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5539 Alan Hayward <alan.hayward@arm.com>
5540 David Sherwood <david.sherwood@arm.com>
5542 * target.def (hard_regno_nregs): New hook.
5543 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
5544 * targhooks.h (default_hard_regno_nregs): Declare.
5545 * targhooks.c (default_hard_regno_nregs): New function.
5546 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
5547 (TARGET_HARD_REGNO_NREGS): ...this hook.
5548 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
5549 (CLASS_MAX_NREGS): Likewise.
5550 * doc/tm.texi: Regenerate.
5551 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
5552 instead of HARD_REGNO_NREGS.
5553 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
5554 HARD_REGNO_NREGS in the comment.
5555 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
5556 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
5557 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
5558 Return an unsigned int.
5559 (TARGET_HARD_REGNO_NREGS): Redefine.
5560 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
5561 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
5562 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5563 (arc_hard_regno_nregs): New function.
5564 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
5565 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
5566 (arm_hard_regno_nregs): New function.
5567 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
5568 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
5569 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
5570 (TARGET_HARD_REGNO_NREGS): Redefine.
5571 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
5572 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
5573 (HARD_REGNO_NREGS): Delete.
5574 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
5575 (cr16_hard_regno_nregs): New function.
5576 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
5577 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
5578 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
5579 (cris_hard_regno_nregs): New function.
5580 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
5581 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
5582 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5583 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
5584 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
5585 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
5586 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
5587 (frv_hard_regno_nregs): Make static. Take and return an
5589 (frv_class_max_nregs): Remove outdated copy of documentation.
5590 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
5591 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
5592 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
5593 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
5594 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
5595 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
5596 (TARGET_HARD_REGNO_NREGS): Redefine.
5597 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
5598 (CLASS_MAX_NREGS): Update comment.
5599 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
5600 (ia64_hard_regno_nregs): New function.
5601 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
5602 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5603 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5604 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5605 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5607 (m32c_hard_regno_nregs): Likewise. Make static.
5608 (TARGET_HARD_REGNO_NREGS): Redefine.
5609 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5610 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5611 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5612 (m68k_hard_regno_nregs): New function.
5613 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5614 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5615 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5616 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5617 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5618 Take and return an unsigned int.
5619 (TARGET_HARD_REGNO_NREGS): Redefine.
5620 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5621 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5622 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5623 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5624 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5625 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5626 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5627 (msp430_hard_regno_nregs): Make static. Take and return an
5629 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5630 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5631 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5632 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5633 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5634 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5635 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5636 (TARGET_HARD_REGNO_NREGS): Redefine.
5637 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5638 (PA_HARD_REGNO_NREGS): ...this.
5639 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5640 (PA_HARD_REGNO_NREGS): ...this.
5641 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5642 (pa_hard_regno_nregs): New function.
5643 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5644 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5645 (pdp11_hard_regno_nregs): New function.
5646 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5647 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5648 (rs6000_hard_regno_nregs_hook): New function.
5649 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5650 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5651 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5652 Take and return an unsigned int. Move earlier in file.
5653 (TARGET_HARD_REGNO_NREGS): Redefine.
5654 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5655 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5656 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5657 (rl78_hard_regno_nregs): Make static. Take and return an
5659 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5660 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5661 (rs6000_hard_regno_nregs_hook): New function.
5662 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5663 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5664 (TARGET_HARD_REGNO_NREGS): Redefine.
5665 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5666 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5667 instead of HARD_REGNO_NREGS.
5668 (s390_hard_regno_nregs): New function.
5669 (s390_hard_regno_mode_ok): Add comment from s390.h.
5670 (TARGET_HARD_REGNO_NREGS): Redefine.
5671 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5672 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5673 (sh_hard_regno_nregs): New function.
5674 (sh_pass_in_reg_p): Use it.
5675 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5676 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5677 (sparc_hard_regno_nregs): New function.
5678 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5679 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5680 (spu_function_arg_advance): Use it, supplying a valid register number.
5681 (TARGET_HARD_REGNO_NREGS): Redefine.
5682 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5683 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5684 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5685 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5686 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5687 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5688 (CLASS_MAX_NREGS): Remove copy of old documentation.
5689 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5690 (visium_hard_regno_nregs): New function.
5691 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5692 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5693 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5694 xtensa_hard_regno_nregs): New function.
5695 * system.h (HARD_REGNO_NREGS): Poison.
5697 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5699 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5700 hard_regno_nregs instead of HARD_REGNO_NREGS.
5701 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5702 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5703 (c6x_expand_epilogue): Likewise.
5704 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5705 (frv_read_iacc_argument): Likewise.
5706 * config/sh/sh.c: Include regs.h.
5707 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5708 (regs_used): Likewise.
5709 (output_stack_adjust): Likewise.
5710 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5711 * expmed.c: Include regs.h.
5712 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5713 * ree.c: Include regs.h.
5714 (combine_reaching_defs): Use hard_regno_nregs instead of
5716 (add_removable_extension): Likewise.
5718 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5720 * regs.h (hard_regno_nregs): Turn into a function.
5721 (end_hard_regno): Update accordingly.
5722 * caller-save.c (setup_save_areas): Likewise.
5723 (save_call_clobbered_regs): Likewise.
5724 (replace_reg_with_saved_mem): Likewise.
5725 (insert_restore): Likewise.
5726 (insert_save): Likewise.
5727 * combine.c (can_change_dest_mode): Likewise.
5728 (move_deaths): Likewise.
5729 (distribute_notes): Likewise.
5730 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5731 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5732 (rs6000_split_multireg_move): Likewise.
5733 (rs6000_register_move_cost): Likewise.
5734 (rs6000_memory_move_cost): Likewise.
5735 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5736 (rs6000_split_multireg_move): Likewise.
5737 (rs6000_register_move_cost): Likewise.
5738 (rs6000_memory_move_cost): Likewise.
5739 * cselib.c (cselib_reset_table): Likewise.
5740 (cselib_lookup_1): Likewise.
5741 * emit-rtl.c (set_mode_and_regno): Likewise.
5742 * function.c (aggregate_value_p): Likewise.
5743 * ira-color.c (setup_profitable_hard_regs): Likewise.
5744 (check_hard_reg_p): Likewise.
5745 (calculate_saved_nregs): Likewise.
5746 (assign_hard_reg): Likewise.
5747 (improve_allocation): Likewise.
5748 (calculate_spill_cost): Likewise.
5749 * ira-emit.c (modify_move_list): Likewise.
5750 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5751 (ira_hard_reg_in_set_p): Likewise.
5752 * ira.c (setup_reg_mode_hard_regset): Likewise.
5753 (clarify_prohibited_class_mode_regs): Likewise.
5754 (check_allocation): Likewise.
5755 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5756 (lra_setup_reg_renumber): Likewise.
5757 (setup_try_hard_regno_pseudos): Likewise.
5758 (spill_for): Likewise.
5759 (assign_hard_regno): Likewise.
5760 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5761 * lra-constraints.c (in_class_p): Likewise.
5762 (lra_constraint_offset): Likewise.
5763 (simplify_operand_subreg): Likewise.
5764 (lra_constraints): Likewise.
5765 (split_reg): Likewise.
5766 (split_if_necessary): Likewise.
5767 (invariant_p): Likewise.
5768 (inherit_in_ebb): Likewise.
5769 * lra-lives.c (process_bb_lives): Likewise.
5770 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5771 (get_hard_regs): Likewise.
5772 (do_remat): Likewise.
5773 * lra-spills.c (assign_spill_hard_regs): Likewise.
5774 * mode-switching.c (create_pre_exit): Likewise.
5775 * postreload.c (reload_combine_recognize_pattern): Likewise.
5776 * recog.c (peep2_find_free_register): Likewise.
5777 * regcprop.c (kill_value_regno): Likewise.
5778 (set_value_regno): Likewise.
5779 (copy_value): Likewise.
5780 (maybe_mode_change): Likewise.
5781 (find_oldest_value_reg): Likewise.
5782 (copyprop_hardreg_forward_1): Likewise.
5783 * regrename.c (check_new_reg_p): Likewise.
5784 (regrename_do_replace): Likewise.
5785 * reload.c (push_reload): Likewise.
5786 (combine_reloads): Likewise.
5787 (find_dummy_reload): Likewise.
5788 (operands_match_p): Likewise.
5789 (find_reloads): Likewise.
5790 (find_equiv_reg): Likewise.
5791 (reload_adjust_reg_for_mode): Likewise.
5792 * reload1.c (count_pseudo): Likewise.
5793 (count_spilled_pseudo): Likewise.
5794 (find_reg): Likewise.
5795 (clear_reload_reg_in_use): Likewise.
5796 (free_for_value_p): Likewise.
5797 (allocate_reload_reg): Likewise.
5798 (choose_reload_regs): Likewise.
5799 (reload_adjust_reg_for_temp): Likewise.
5800 (emit_reload_insns): Likewise.
5801 (delete_output_reload): Likewise.
5802 * rtlanal.c (subreg_get_info): Likewise.
5803 * sched-deps.c (sched_analyze_reg): Likewise.
5804 * sel-sched.c (init_regs_for_mode): Likewise.
5805 (mark_unavailable_hard_regs): Likewise.
5806 (choose_best_reg_1): Likewise.
5807 (verify_target_availability): Likewise.
5808 * valtrack.c (dead_debug_insert_temp): Likewise.
5809 * var-tracking.c (track_loc_p): Likewise.
5810 (emit_note_insn_var_location): Likewise.
5811 * varasm.c (make_decl_rtl): Likewise.
5812 * reginfo.c (choose_hard_reg_mode): Likewise.
5813 (init_reg_modes_target): Refer directly to
5814 this_target_regs->x_hard_regno_nregs.
5816 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5818 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5819 instead of hard_regno_nregs.
5821 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5823 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5824 end_hard_regno instead of hard_regno_nregs.
5825 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5826 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5827 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5828 * ira-color.c (improve_allocation): Likewise.
5829 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5830 * lra-lives.c (mark_regno_live): Likewise.
5831 (mark_regno_dead): Likewise.
5832 * lra-remat.c (operand_to_remat): Likewise.
5833 * lra.c (collect_non_operand_hard_regs): Likewise.
5834 * postreload.c (reload_combine_note_store): Likewise.
5835 (move2add_valid_value_p): Likewise.
5836 * reload.c (regno_clobbered_p): Likewise.
5838 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5840 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5842 * config/v850/v850.c (v850_reorg): Likewise.
5843 * reload.c (refers_to_regno_for_reload_p): Likewise.
5844 (find_equiv_reg): Likewise.
5845 * reload1.c (reload_reg_reaches_end_p): Likewise.
5847 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5849 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5851 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5852 * config/arm/arm.c (output_move_neon): Likewise.
5853 (arm_attr_length_move_neon): Likewise.
5854 (neon_split_vcombine): Likewise.
5855 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5856 (c6x_mark_reg_written): Likewise.
5857 (c6x_dwarf_register_span): Likewise.
5858 * config/i386/i386.c (ix86_save_reg): Likewise.
5859 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5860 (rws_access_reg): Likewise.
5861 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5862 * mode-switching.c (create_pre_exit): Likewise.
5863 * ree.c (combine_reaching_defs): Likewise.
5864 (add_removable_extension): Likewise.
5865 * regcprop.c (find_oldest_value_reg): Likewise.
5866 (copyprop_hardreg_forward_1): Likewise.
5867 * reload.c (reload_inner_reg_of_subreg): Likewise.
5868 (push_reload): Likewise.
5869 (combine_reloads): Likewise.
5870 (find_dummy_reload): Likewise.
5871 (reload_adjust_reg_for_mode): Likewise.
5872 * reload1.c (find_reload_regs): Likewise.
5873 (forget_old_reloads_1): Likewise.
5874 (reload_reg_free_for_value_p): Likewise.
5875 (reload_adjust_reg_for_temp): Likewise.
5876 (emit_reload_insns): Likewise.
5877 (delete_output_reload): Likewise.
5878 * sel-sched.c (choose_best_reg_1): Likewise.
5879 (choose_best_pseudo_reg): Likewise.
5881 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5882 Alan Hayward <alan.hayward@arm.com>
5883 David Sherwood <david.sherwood@arm.com>
5885 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5886 * target.def (slow_unaligned_access): New hook.
5887 * targhooks.h (default_slow_unaligned_access): Declare.
5888 * targhooks.c (default_slow_unaligned_access): New function.
5889 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5890 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5891 * doc/tm.texi: Regenerate.
5892 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5893 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5894 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5896 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5897 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5899 (rs6000_slow_unaligned_access): New function.
5900 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5901 (expand_block_compare): Likewise.
5902 (expand_strn_compare): Likewise.
5903 (rs6000_rtx_costs): Likewise.
5904 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5905 (riscv_slow_unaligned_access): Likewise.
5906 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5907 (riscv_slow_unaligned_access_p): ...this and make static.
5908 (riscv_option_override): Update accordingly.
5909 (riscv_slow_unaligned_access): New function.
5910 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5911 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5912 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5913 (rs6000_slow_unaligned_access): New function.
5914 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5915 (rs6000_rtx_costs): Likewise.
5916 * config/rs6000/rs6000-string.c (expand_block_compare)
5917 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5918 of SLOW_UNALIGNED_ACCESS.
5919 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5920 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5921 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5922 of SLOW_UNALIGNED_ACCESS.
5923 * expmed.c (simple_mem_bitfield_p): Likewise.
5924 * expr.c (alignment_for_piecewise_move): Likewise.
5925 (emit_group_load_1): Likewise.
5926 (emit_group_store): Likewise.
5927 (copy_blkmode_from_reg): Likewise.
5928 (emit_push_insn): Likewise.
5929 (expand_assignment): Likewise.
5930 (store_field): Likewise.
5931 (expand_expr_real_1): Likewise.
5932 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5933 * lra-constraints.c (simplify_operand_subreg): Likewise.
5934 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5935 * gimple-ssa-store-merging.c: Likewise in block comment at start
5937 * tree-ssa-strlen.c: Include target.h.
5938 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5939 of SLOW_UNALIGNED_ACCESS.
5940 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5942 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5944 PR rtl-optimization/82185
5945 * expmed.c (emit_store_flag_int): Only test tem if it has been
5948 2017-09-12 Richard Biener <rguenther@suse.de>
5951 * match.pd ((FTYPE) N CMP CST): Fix typo.
5953 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5955 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5957 (mips_near_type_p): Add 'short_call' attribute as a synonym
5959 * doc/extend.texi (short_call): Document new function attribute.
5961 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5964 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5965 assertion check that in the condition.
5966 (get_atomic_generic_size): Likewise. Before testing if parameter
5967 has pointer type, if it has array type, call for C++
5968 default_conversion to perform array-to-pointer conversion.
5970 2017-09-12 Richard Biener <rguenther@suse.de>
5972 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5973 for operations we cannot scalarize.
5975 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5977 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5978 vectors heap vectors. Clean up comments.
5979 Make visited_bbs a reference.
5980 (profitable_jump_thread_path): Make GC
5981 vectors heap vectors. Clean up comments.
5983 (convert_and_register_jump_thread_path): Make GC vectors heap
5985 (check_subpath_and_update_thread_path): Same. Clean up comments.
5986 Make visited_bbs a reference.
5987 (handle_phi): Abstract common code to to
5988 register_jump_thread_path_if_profitable.
5989 Rename VAR_BB to DEF_BB.
5991 Make GC vectors heap vectors.
5992 Make visited_bbs a reference.
5993 (handle_assignment): Same.
5994 (register_jump_thread_path_if_profitable): New.
5995 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5997 Make GC vectors heap vectors. Clean up comments.
5998 Make visited_bbs a reference.
5999 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6000 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6003 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6006 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6007 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6009 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6011 Revert r251800 and r251799.
6013 2017-09-11 Martin Jambor <mjambor@suse.cz>
6016 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6017 arguments in advance.
6018 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6019 use it to find predecessor edges.
6020 (naive_outof_ssa): Collect vector of predecessors.
6022 2017-09-08 Jason Merrill <jason@redhat.com>
6024 PR c++/70029 - ICE with ref-qualifier and -flto
6025 * langhooks.h (struct lang_hooks_for_types): Add
6026 copy_lang_qualifiers.
6027 * attribs.c (build_type_attribute_qual_variant): Use it.
6028 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6030 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6031 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6033 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6036 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6037 (*mulsi3_sp64): New instruction.
6038 (mulsi3): New expander.
6040 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6042 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6044 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6046 * sancov.c: Include memmodel.h.
6048 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6051 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6054 2017-09-07 Carl Love <cel@us.ibm.com>
6056 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6057 the sldi instruction.
6059 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6061 * sancov.c: Include tm_p.h.
6063 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6066 * output.h (switch_to_other_text_partition): New declaration.
6067 * varasm.c (switch_to_other_text_partition): New function.
6068 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6069 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6070 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6071 to the other text partition before emitting LCL label and switch back
6072 after emitting the word after it.
6074 2017-09-07 Richard Biener <rguenther@suse.de>
6076 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6077 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6078 critical edges here, after loop init.
6079 (pass_data_pre): Remove PROP_no_crit_edges flags.
6080 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6081 for valueization of call args to avoid leaking VN_TOP.
6082 (visit_use): Assert we do not visit default defs.
6083 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6084 Use error_mark_node to more easily detect leaking VN_TOP.
6085 All default-defs are varying, not VN_TOP. Mark them visited.
6086 (run_scc_vn): Make code match comment.
6088 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6090 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6091 OPTION_MASK_FLOAT128_KEYWORD.
6092 (POWERPC_MASKS): Likewise.
6093 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6094 support for the -mfloat128-type option, and make -mfloat128
6095 default on PowerPC Linux systems. Define or undefine
6096 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6097 Define __float128 to be __ieee128 if IEEE 128-bit support is
6098 enabled, or undefine it.
6099 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6100 Delete defining __FLOAT128_TYPE__.
6101 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6102 -mfloat128-type option and make -mfloat128 default on PowerPC
6104 (TARGET_FLOAT128_TYPE): Likewise.
6105 (-mfloat128-type): Likewise.
6106 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6107 Delete the -mfloat128-type option and make -mfloat128 default on
6108 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6109 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6110 128-bit floating point is enabled. Change tests from using
6111 -mfloat128-type to -mfloat128.
6112 (rs6000_mangle_type): Use the correct mangling for the __float128
6113 type even if normal long double is restricted to 64-bits.
6114 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6116 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6117 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6118 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6120 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6121 documentation for -mfloat128.
6123 2017-09-06 Olivier Hainque <hainque@adacore.com>
6125 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6127 2017-09-06 Wish Wu <wishwu007@gmail.com>
6128 Jakub Jelinek <jakub@redhat.com>
6130 * asan.c (initialize_sanitizer_builtins): Add
6131 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6132 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6133 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6134 BT_FN_VOID_UINT64_PTR variables.
6135 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6136 (BT_FN_VOID_UINT16_UINT16): Likewise.
6137 (BT_FN_VOID_UINT32_UINT32): Likewise.
6138 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6139 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6140 (BT_FN_VOID_UINT64_PTR): Likewise.
6141 * common.opt (flag_sanitize_coverage): New variable.
6142 (fsanitize-coverage=trace-pc): Remove.
6143 (fsanitize-coverage=): Add.
6144 * flag-types.h (enum sanitize_coverage_code): New enum.
6145 * fold-const.c (fold_range_test): Disable non-short-circuit
6146 optimization if flag_sanitize_coverage.
6147 (fold_truth_andor): Likewise.
6148 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6149 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6150 (coverage_sanitizer_opts): New array.
6151 (get_closest_sanitizer_option): Add OPTS argument, handle also
6152 OPT_fsanitize_coverage_.
6153 (parse_sanitizer_options): Adjusted to also handle
6154 OPT_fsanitize_coverage_.
6155 (common_handle_option): Add OPT_fsanitize_coverage_.
6156 * sancov.c (instrument_comparison, instrument_switch): New function.
6157 (sancov_pass): Add trace-cmp support.
6158 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6159 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6160 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6161 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6162 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6163 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6164 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6165 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6166 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6167 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6169 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6171 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6172 error. Only quit immediately if parsing is complete.
6173 (BEGIN): Initialize fatal_err and parse_done.
6174 (begin fpu, end fpu): Check number of arguments.
6175 (begin arch, end arch): Likewise.
6176 (begin cpu, end cpu): Likewise.
6177 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6178 (optalias): Likewise.
6180 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6182 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6183 * config/arm/arm-isa.h: Delete. Move definitions to ...
6184 * arm-cpus.in: ... here. Use new feature and fgroup values.
6185 * config/arm/arm.c (arm_option_override): Use lower case for feature
6187 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6188 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6189 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6191 (print_isa_bits_for): New function.
6192 (gen_isa): New function.
6193 (gen_comm_data): Use print_isa_bits_for.
6194 (define feature): New keyword.
6195 (define fgroup): New keyword.
6196 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6197 (arm-isa.h): Add rule to generate file.
6198 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6199 case for feature bit names.
6201 2017-09-06 Richard Biener <rguenther@suse.de>
6203 * tree-ssa-pre.c (NECESSARY): Remove.
6204 (create_expression_by_pieces): Do not touch pass-local flags.
6205 (insert_into_preds_of_block): Likewise.
6206 (do_pre_regular_insertion): Likewise.
6207 (eliminate_insert): Likewise.
6208 (eliminate_dom_walker::before_dom_children): Likewise.
6209 (fini_eliminate): Do not look at inserted_exprs.
6210 (mark_operand_necessary): Remove.
6211 (remove_dead_inserted_code): Replace with simple work-list
6212 algorithm based on inserted_exprs and SSA uses.
6213 (pass_pre::execute): Re-order fini_eliminate and
6214 remove_dead_inserted_code.
6216 2017-09-06 Olivier Hainque <hainque@adacore.com>
6218 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6219 for VxWorks 7. Adjust surrounding comments.
6221 2017-09-06 Richard Biener <rguenther@suse.de>
6223 * gimple-ssa-strength-reduction.c
6224 (find_candidates_dom_walker::before_dom_children): Also allow
6227 2017-09-06 Richard Biener <rguenther@suse.de>
6229 PR tree-optimization/82108
6230 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6231 for gap in the non-permutation SLP case.
6233 2017-09-06 Martin Jambor <mjambor@suse.cz>
6235 PR tree-optimization/82078
6236 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6237 add_access_to_work_queue...
6238 (build_accesses_from_assign): ...here.
6239 (propagate_all_subaccesses): Make sure racc is the group
6240 representative, if there is one.
6242 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6245 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6248 2017-09-06 Richard Biener <rguenther@suse.de>
6250 * gimple-ssa-strength-reduction.c
6251 (find_candidates_dom_walker::before_doom_children): Use a
6252 type and not a mode check.
6254 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6257 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6259 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6261 2017-09-05 Jeff Law <law@redhat.com>
6263 PR tree-optimization/64910
6264 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6265 cases where we have 3 or more operands.
6267 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6270 * omp-low.c (lower_omp_for): Recompute tree invariant if
6271 gimple_omp_for_initial/final is ADDR_EXPR.
6274 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6275 into gimple val before gimplification fo the COND_EXPR.
6277 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6279 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6280 REGION_COPY argument.
6281 (thread_through_all_blocks): Remove unused argument to
6282 duplicate_thread_path.
6284 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6285 Alan Hayward <alan.hayward@arm.com>
6286 David Sherwood <david.sherwood@arm.com>
6288 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6289 Take a scalar_mode rather than a machine_mode.
6290 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6291 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6292 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6293 (aarch64_gen_adjusted_ldpstp): Likewise.
6294 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6296 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6297 Alan Hayward <alan.hayward@arm.com>
6298 David Sherwood <david.sherwood@arm.com>
6300 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6301 Take a scalar_int_mode instead of a machine_mode.
6302 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6303 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6304 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6305 (aarch64_simd_attr_length_rglist): Delete.
6306 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6307 a scalar_int_mode instead of a machine_mode.
6308 (aarch64_add_offset): Likewise.
6309 (aarch64_internal_mov_immediate): Likewise
6310 (aarch64_add_constant_internal): Likewise.
6311 (aarch64_add_constant): Likewise.
6312 (aarch64_movw_imm): Likewise.
6313 (aarch64_rtx_arith_op_extract_p): Likewise.
6314 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6315 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6316 Remove assert that the mode isn't a vector.
6317 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6318 (aarch64_expand_mov_immediate): Update calls after above changes.
6319 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6320 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6321 (aarch64_move_imm): Likewise.
6322 (aarch64_can_const_movi_rtx_p): Likewise.
6323 (aarch64_strip_extend): Likewise.
6324 (aarch64_extr_rtx_p): Likewise.
6325 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6326 a CONST_INT when the mode parameter is VOIDmode.
6327 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6329 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6331 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6332 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6333 (bitwise_type_for_mode): Update accordingly.
6335 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6337 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6338 * stor-layout.c (mode_for_size_tree): Likewise.
6339 (mode_for_array): Update accordingly.
6340 (layout_decl): Likewise.
6341 (compute_record_mode): Likewise. Only set the mode once.
6343 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6345 * target.def (get_mask_mode): Change return type to opt_mode.
6347 * doc/tm.texi: Regenerate.
6348 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6349 * targhooks.c (default_get_mask_mode): Likewise.
6350 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6351 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6352 targetm.get_mask_mode.
6353 * tree.c (build_truth_vector_type): Likewise.
6355 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6357 * machmode.h (mode_for_vector): Return an opt_mode.
6358 * stor-layout.c (mode_for_vector): Likewise.
6359 (mode_for_int_vector): Update accordingly.
6360 (layout_type): Likewise.
6361 * config/i386/i386.c (emit_memmov): Likewise.
6362 (ix86_expand_set_or_movmem): Likewise.
6363 (ix86_expand_vector_init): Likewise.
6364 (ix86_get_mask_mode): Likewise.
6365 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6367 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6368 * expmed.c (extract_bit_field_1): Likewise.
6369 * expr.c (expand_expr_real_2): Likewise.
6370 * optabs-query.c (can_vec_perm_p): Likewise.
6371 (can_vec_mask_load_store_p): Likewise.
6372 * optabs.c (expand_vec_perm): Likewise.
6373 * targhooks.c (default_get_mask_mode): Likewise.
6374 * tree-vect-stmts.c (vectorizable_store): Likewise.
6375 (vectorizable_load): Likewise.
6376 (get_vectype_for_scalar_type_and_size): Likewise.
6378 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6380 * machmode.h (mode_for_int_vector): New function.
6381 * stor-layout.c (mode_for_int_vector): Likewise.
6382 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6383 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6384 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6385 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6386 (s390_expand_vcond): Likewise.
6388 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6390 * machmode.h (opt_machine_mode): New type.
6391 (opt_mode<T>): Allow construction from anything that can be
6393 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6394 (mode_for_size): Return an opt_machine_mode.
6395 * stor-layout.c (mode_for_size): Likewise.
6396 (mode_for_size_tree): Update call accordingly.
6397 (bitwise_mode_for_mode): Likewise.
6398 (make_fract_type): Likewise.
6399 (make_accum_type): Likewise.
6400 * caller-save.c (replace_reg_with_saved_mem): Update call
6402 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6403 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6404 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6405 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6406 * expmed.c (extract_bit_field_1): Likewise.
6407 * reload.c (get_secondary_mem): Likewise.
6408 * varasm.c (assemble_integer): Likewise.
6409 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6412 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6414 * machmode.h (decimal_float_mode_for_size): New function.
6415 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6416 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6417 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6418 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6419 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6420 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6422 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6424 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6425 (get_builtin_sync_mode): Likewise.
6426 (expand_ifn_atomic_compare_exchange): Likewise.
6427 (expand_builtin_atomic_clear): Likewise.
6428 (expand_builtin_atomic_test_and_set): Likewise.
6429 (fold_builtin_atomic_always_lock_free): Likewise.
6430 * calls.c (compute_argument_addresses): Likewise.
6431 (emit_library_call_value_1): Likewise.
6432 (store_one_arg): Likewise.
6433 * combine.c (combine_instructions): Likewise.
6434 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6435 * config/arm/arm.c (arm_function_value): Likewise.
6436 (aapcs_allocate_return_reg): Likewise.
6437 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6438 * config/i386/i386.c (construct_container): Likewise.
6439 (ix86_gimplify_va_arg): Likewise.
6440 (ix86_expand_sse_cmp): Likewise.
6441 (emit_memmov): Likewise.
6442 (emit_memset): Likewise.
6443 (expand_small_movmem_or_setmem): Likewise.
6444 (ix86_expand_pextr): Likewise.
6445 (ix86_expand_pinsr): Likewise.
6446 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6447 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6449 * config/mips/mips.c (mips_function_value_1) Likewise.
6450 (mips_block_move_straight): Likewise.
6451 (mips_expand_ins_as_unaligned_store): Likewise.
6452 * config/powerpcspe/powerpcspe.c
6453 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6454 (rs6000_darwin64_record_arg_flush): Likewise.
6455 * config/rs6000/rs6000.c
6456 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6457 (rs6000_darwin64_record_arg_flush): Likewise.
6458 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6459 (sparc_function_value_1): Likewise.
6460 * config/spu/spu.c (adjust_operand): Likewise.
6461 (spu_emit_branch_or_set): Likewise.
6462 (arith_immediate_p): Likewise.
6463 * emit-rtl.c (gen_lowpart_common): Likewise.
6464 * expr.c (expand_expr_real_1): Likewise.
6465 * function.c (assign_parm_setup_block): Likewise.
6466 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6467 * reload1.c (alter_reg): Likewise.
6468 * stor-layout.c (mode_for_vector): Likewise.
6469 (layout_type): Likewise.
6471 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6473 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6474 (spu_convert_move): Likewise.
6475 * lower-subreg.c (resolve_simple_move): Likewise.
6477 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6480 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6481 define_insn to a define_expand.
6482 (altivec_vsum2sws_direct): New define_insn.
6483 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6485 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6487 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6490 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6495 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6496 realign stack if stack alignment needed is less than incoming
6499 2017-09-05 Marek Polacek <polacek@redhat.com>
6502 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6505 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6507 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6509 2017-09-05 Richard Biener <rguenther@suse.de>
6511 PR tree-optimization/82084
6512 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6514 2017-09-05 Richard Biener <rguenther@suse.de>
6516 PR tree-optimization/82102
6517 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
6519 2017-09-05 Martin Liska <mliska@suse.cz>
6521 PR tree-optimization/82032
6522 * tree-cfg.c (generate_range_test): New function.
6523 * tree-cfg.h (generate_range_test): Declared here.
6524 * tree-cfgcleanup.c (convert_single_case_switch): New function.
6525 (cleanup_control_expr_graph): Use it.
6526 * tree-switch-conversion.c (try_switch_expansion): Remove
6528 (emit_case_nodes): Use generate_range_test.
6530 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6533 * config/i386/i386.md (*<btsc><mode>_mask): Add
6534 TARGET_USE_BT to insn constraint.
6535 (*btr<mode>_mask): Ditto.
6537 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
6539 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
6540 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
6542 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6545 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
6546 TARGET_NEON and TARGET_IWMMXT.
6547 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
6548 TARGET_NEON and TARGET_IWMMXT.
6549 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
6551 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6553 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
6554 (ix86_rewrite_tls_address): Ditto.
6555 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
6556 (ix86_rewrite_tls_address_1): Ditto.
6557 (ix86_rewrite_tls_address): Ditto.
6558 * config/i386/predicates.md (tls_address_pattern): New predicate.
6559 * config/i386/i386.md (TLS address splitter): New splitter.
6561 2017-09-04 Richard Biener <rguenther@suse.de>
6563 PR tree-optimization/82084
6564 * fold-const.h (can_native_encode_string_p): Declare.
6565 * fold-const.c (can_native_encode_string_p): Factor out from ...
6566 (native_encode_string): ... here.
6567 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
6568 vectorizing stores from constants we later cannot handle.
6570 2017-09-04 Marek Polacek <polacek@redhat.com>
6573 * doc/invoke.texi: Update -Wtautological-compare documentation.
6575 2017-09-04 Jeff Law <law@redhat.com>
6577 PR tree-optimization/64910
6578 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
6579 swap the first and last operand if the last is a constant.
6581 2017-09-04 Marek Polacek <polacek@redhat.com>
6584 * convert.c (do_narrow): When sanitizing signed integer overflows,
6585 bail out for signed types.
6586 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
6588 2017-09-04 Richard Biener <rguenther@suse.de>
6590 PR tree-optimization/82060
6591 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
6592 Move devirtualization after stmt folding and before EH/AB/noreturn
6593 cleanup to get the stmt refs canonicalized. Use a bool instead
6594 of gimple_modified_p since that doesn't work for NOPs. Schedule
6595 NOPs generated by folding for removal.
6597 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6598 Alan Hayward <alan.hayward@arm.com>
6599 David Sherwood <david.sherwood@arm.com>
6601 * coretypes.h (pad_direction): New enum.
6602 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6603 (FUNCTION_ARG_PADDING): Likewise.
6604 * target.def (function_arg_padding): New hook.
6605 * targhooks.h (default_function_arg_padding): Declare.
6606 * targhooks.c (default_function_arg_padding): New function.
6607 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6608 (TARGET_FUNCTION_ARG_PADDING): ...this.
6609 * doc/tm.texi: Regenerate.
6610 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6611 instead of direction.
6612 (compute_argument_addresses): Likewise.
6613 (load_register_parameters): Likewise.
6614 (emit_library_call_value_1): Likewise.
6615 (store_one_arg): Use targetm.calls.function_arg_padding instead
6616 of FUNCTION_ARG_PADDING.
6617 (must_pass_in_stack_var_size_or_pad): Likewise.
6618 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6619 (emit_group_store): Likewise.
6620 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6621 instead of FUNCTION_ARG_PADDING.
6622 (emit_push_insn): Likewise, and propagate enum change throughout
6624 * function.h (direction): Delete.
6625 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6627 * function.c (assign_parm_find_stack_rtl): Likewise.
6628 (assign_parm_setup_block_p): Likewise.
6629 (assign_parm_setup_block): Likewise.
6630 (gimplify_parameters): Likewise.
6631 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6632 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6634 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6635 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6636 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6637 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6638 (aarch64_function_arg_padding): ...this new function.
6639 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6640 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6641 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6642 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6643 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6644 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6645 (arm_pad_arg_upward): Replace with...
6646 (arm_function_arg_padding): ...this new function.
6647 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6649 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6650 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6651 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6652 (ia64_hpux_function_arg_padding): Replace with...
6653 (ia64_function_arg_padding): ...this new function. Use pad_direction
6654 instead of direction. Check for TARGET_HPUX.
6655 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6656 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6657 (iq2000_function_arg_padding): New function.
6658 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6659 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6660 (mips_function_arg_padding): ...this new function.
6661 (mips_pad_reg_upward): Update accordingly.
6662 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6663 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6664 targetm.calls.function_arg_padding.
6665 (FUNCTION_ARG_PADDING): Delete.
6666 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6667 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6668 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6669 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6670 (nios2_block_reg_padding): Return pad_direction instead of direction.
6671 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6672 instead of direction.
6673 (nios2_function_arg_padding): Likewise. Make static.
6674 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6675 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6676 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6677 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6678 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6679 (pa_function_arg_padding): Make static. Return pad_direction instead
6681 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6682 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6683 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6684 instead of direction. Use targetm.calls.function_arg_padding.
6685 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6686 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6687 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6688 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6689 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6691 (function_arg_padding): Rename to...
6692 (rs6000_function_arg_padding): ...this. Make static. Return
6693 pad_direction instead of direction.
6694 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6695 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6696 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6697 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6698 instead of direction. Use targetm.calls.function_arg_padding.
6699 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6700 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6701 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6702 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6703 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6704 (function_arg_padding): Rename to...
6705 (rs6000_function_arg_padding): ...this. Make static. Return
6706 pad_direction instead of direction.
6707 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6708 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6709 * config/s390/s390.c (s390_function_arg_padding): New function.
6710 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6711 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6712 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6713 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6714 (function_arg_padding): Rename to...
6715 (sparc_function_arg_padding): ...this. Make static. Return
6716 pad_direction instead of direction.
6717 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6718 * config/spu/spu.c (spu_function_arg_padding): New function.
6719 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6720 * system.h (FUNCTION_ARG_PADDING): Poison.
6722 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6723 Alan Hayward <alan.hayward@arm.com>
6724 David Sherwood <david.sherwood@arm.com>
6726 * target.def (modes_tieable_p): New hook.
6727 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6728 (TARGET_MODES_TIEABLE_P): ...this.
6729 * doc/tm.texi.in: Regenerate.
6730 * hooks.h (hook_bool_mode_mode_true): Declare.
6731 * hooks.c (hook_bool_mode_mode_true): New function.
6732 * combine.c (subst): Use targetm.modes_tieable_p instead of
6734 * dse.c (find_shift_sequence): Likewise.
6735 * expmed.c (extract_low_bits): Likewise.
6736 * lower-subreg.c: Include target.h.
6737 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6739 * rtlanal.c (rtx_cost): Likewise.
6740 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6741 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6742 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6743 (TARGET_MODES_TIEABLE_P): Redefine.
6744 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6745 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6746 (TARGET_MODES_TIEABLE_P): Redefine.
6747 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6748 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6749 (arc_modes_tieable_p): New function.
6750 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6751 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6752 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6753 (arm_modes_tieable_p): Make static.
6754 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6755 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6756 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6757 (TARGET_MODES_TIEABLE_P): Redefine.
6758 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6759 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6760 (TARGET_MODES_TIEABLE_P): Redefine.
6761 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6762 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6763 (cr16_modes_tieable_p): New function.
6764 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6765 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6766 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6767 (TRULY_NOOP_TRUNCATION): Update comment.
6768 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6769 (TRULY_NOOP_TRUNCATION): Update comment.
6770 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6771 (frv_modes_tieable_p): New function.
6772 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6773 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6774 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6775 (TARGET_MODES_TIEABLE_P): Redefine.
6776 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6777 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6778 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6779 (TARGET_MODES_TIEABLE_P): Redefine.
6780 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6781 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6782 (ia64_modes_tieable_p): New function.
6783 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6784 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6785 (iq2000_modes_tieable_p): New function.
6786 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6787 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6788 (lm32_modes_tieable_p): New function.
6789 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6790 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6791 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6792 (TARGET_MODES_TIEABLE_P): Redefine.
6793 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6794 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6795 (m32r_modes_tieable_p): New function.
6796 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6797 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6798 (m68k_modes_tieable_p): New function.
6799 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6800 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6801 (mcore_modes_tieable_p): New function.
6802 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6803 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6805 (TARGET_MODES_TIEABLE_P): Redefine.
6806 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6807 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6808 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6809 (TARGET_MODES_TIEABLE_P): Redefine.
6810 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6811 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6812 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6813 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6814 (mn10300_modes_tieable_p): ...this and make static.
6815 (TARGET_MODES_TIEABLE_P): Redefine.
6816 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6817 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6818 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6819 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6820 (msp430_modes_tieable_p): Make static.
6821 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6822 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6823 (TARGET_MODES_TIEABLE_P): Redefine.
6824 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6825 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6826 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6827 (TARGET_MODES_TIEABLE_P): Redefine.
6828 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6829 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6830 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6831 (TARGET_MODES_TIEABLE_P): Redefine.
6832 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6833 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6834 (pdp11_modes_tieable_p): New function.
6835 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6836 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6837 (rs6000_modes_tieable_p): New function.
6838 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6839 * config/powerpcspe/powerpcspe.md: Update comment.
6840 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6841 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6842 (TARGET_MODES_TIEABLE_P): Redefine.
6843 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6844 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6845 (rl78_modes_tieable_p): New function.
6846 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6847 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6848 (rs6000_modes_tieable_p): New function.
6849 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6850 * config/rs6000/rs6000.md: Update comment.
6851 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6852 * config/rx/rx.c (rx_modes_tieable_p): New function.
6853 (TARGET_MODES_TIEABLE_P): Redefine.
6854 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6855 * config/s390/s390.c (s390_modes_tieable_p): New function.
6856 (TARGET_MODES_TIEABLE_P): Redefine.
6857 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6858 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6859 (sh_modes_tieable_p): New function.
6860 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6861 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6862 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6863 (sparc_modes_tieable_p): Make static.
6864 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6865 * config/spu/spu.c (spu_modes_tieable_p): New function.
6866 (TARGET_MODES_TIEABLE_P): Redefine.
6867 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6868 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6869 (TARGET_MODES_TIEABLE_P): Redefine.
6870 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6871 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6872 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6873 * config/v850/v850.c (v850_modes_tieable_p): New function.
6874 (TARGET_MODES_TIEABLE_P): Redefine.
6875 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6876 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6877 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6878 (visium_modes_tieable_p): New function.
6879 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6880 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6881 (xtensa_modes_tieable_p): New function.
6882 * system.h (MODES_TIEABLE_P): Poison.
6884 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6885 Alan Hayward <alan.hayward@arm.com>
6886 David Sherwood <david.sherwood@arm.com>
6888 * target.def (hard_regno_mode_ok): New hook.
6889 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6890 (TARGET_HARD_REGNO_MODE_OK): ...this.
6891 * doc/tm.texi.in: Regenerate.
6892 * hooks.h (hook_bool_uint_mode_true): Declare.
6893 * hooks.c (hook_bool_uint_mode_true): New function.
6894 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6896 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6897 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6898 instead of HARD_REGNO_MODE_OK.
6899 * caller-save.c: Include target.h.
6900 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6902 * combine.c (can_combine_p): Likewise.
6903 (combinable_i3pat): Likewise.
6904 (can_change_dest_mode): Likewise.
6905 * expr.c (init_expr_target): Likewise.
6906 (convert_move): Likewise.
6907 (convert_modes): Likewise.
6908 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6909 (setup_prohibited_mode_move_regs): Likewise.
6910 * ira.h (target_ira): Likewise.
6911 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6912 * lra-constraints.c (process_alt_operands): Likewise.
6913 (split_reg): Likewise.
6914 * recog.c (peep2_find_free_register): Likewise.
6915 * ree.c (combine_reaching_defs): Likewise.
6916 * regcprop.c (maybe_mode_change): Likewise.
6917 * reginfo.c (init_reg_sets_1): Likewise.
6918 (choose_hard_reg_mode): Likewise.
6919 (simplifiable_subregs): Likewise.
6920 * regrename.c (check_new_reg_p): Likewise.
6921 * reload.c (find_valid_class): Likewise.
6922 (find_valid_class_1): Likewise.
6923 (reload_inner_reg_of_subreg): Likewise.
6924 (push_reload): Likewise.
6925 (combine_reloads): Likewise.
6926 (find_dummy_reload): Likewise.
6927 (find_reloads): Likewise.
6928 * reload1.c (find_reg): Likewise.
6929 (set_reload_reg): Likewise.
6930 (allocate_reload_reg): Likewise.
6931 (choose_reload_regs): Likewise.
6932 (reload_adjust_reg_for_temp): Likewise.
6933 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6934 (simplify_subreg_regno): Likewise.
6935 * sel-sched.c (init_regs_for_mode): Likewise.
6936 * varasm.c (make_decl_rtl): Likewise.
6937 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6938 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6940 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6941 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6942 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6943 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6944 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6945 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6946 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6947 (arc_mode_class): Delete.
6948 (HARD_REGNO_MODE_OK): Delete.
6949 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6950 (arc_hard_regno_mode_ok): Rename old array to...
6951 (arc_hard_regno_mode_ok_modes): ...this.
6952 (arc_conditional_register_usage): Update accordingly.
6953 (arc_mode_class): Make static.
6954 (arc_hard_regno_mode_ok): New function.
6955 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6956 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6957 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6958 (arm_hard_regno_mode_ok): Make static.
6959 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6961 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6962 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6963 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6965 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6966 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6967 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6968 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6969 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6970 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6971 * config/bfin/predicates.md (valid_reg_operand): Use
6972 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6973 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6974 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6975 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6976 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6977 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6978 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6979 (cr16_hard_regno_mode_ok): Make static and return a bool.
6980 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6981 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6982 (cris_hard_regno_mode_ok): New function.
6983 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6984 (epiphany_mode_class): Delete.
6985 (HARD_REGNO_MODE_OK): Delete.
6986 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6987 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6988 (hard_regno_mode_ok): Rename to...
6989 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6990 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6991 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6993 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6994 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6995 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6996 (frv_hard_regno_mode_ok): Make static and return a bool.
6997 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6999 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7000 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7001 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7002 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7004 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7005 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7006 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7007 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7009 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7010 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7011 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7012 (ia64_hard_regno_mode_ok): New function.
7013 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7014 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7015 (iq2000_hard_regno_mode_ok): New function.
7016 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7017 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7018 (lm32_hard_regno_mode_ok): New function.
7019 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7020 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7021 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7022 instead of HARD_REGNO_MODE_OK.
7023 (m32c_hard_regno_ok): Rename to...
7024 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7025 (m32c_cannot_change_mode_class): Update accordingly.
7026 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7027 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7028 (m32r_mode_class): Delete.
7029 (HARD_REGNO_MODE_OK): Delete.
7030 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7031 (m32r_hard_regno_mode_ok): Rename to...
7032 (m32r_hard_regno_modes): ...this.
7033 (m32r_mode_class): Make static.
7034 (m32r_hard_regno_mode_ok): New function.
7035 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7036 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7037 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7038 (m68k_hard_regno_mode_ok): Make static.
7039 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7040 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7041 (mcore_hard_regno_mode_ok): New function.
7042 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7043 (HARD_REGNO_MODE_OK): Delete.
7044 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7046 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7047 (microblaze_hard_regno_mode_ok): New function.
7048 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7049 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7050 (mips_hard_regno_mode_ok): Delete.
7051 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7052 (mips_hard_regno_mode_ok_p): ...this and make static.
7053 (mips_hard_regno_mode_ok_p): Rename to...
7054 (mips_hard_regno_mode_ok_uncached): ...this.
7055 (mips_hard_regno_mode_ok): New function.
7056 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7057 of HARD_REGNO_MODE_OK.
7058 (mips_option_override): Update after above name changes.
7059 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7060 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7061 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7062 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7063 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7064 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7065 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7066 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7067 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7068 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7069 (msp430_hard_regno_mode_ok): Make static and return a bool.
7070 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7071 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7072 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7074 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7075 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7076 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7077 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7078 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7079 (PA_HARD_REGNO_MODE_OK): ...this
7080 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7081 (PA_HARD_REGNO_MODE_OK): ...this.
7082 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7083 (pa_hard_regno_mode_ok): New function.
7084 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7085 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7086 (pdp11_hard_regno_mode_ok): New function.
7087 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7088 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7090 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7092 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7093 (rs6000_hard_regno_mode_ok): Rename to...
7094 (rs6000_hard_regno_mode_ok_uncached): ...this.
7095 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7096 (rs6000_hard_regno_mode_ok): New function.
7097 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7098 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7099 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7100 (riscv_hard_regno_mode_ok): ...this and make static.
7101 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7102 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7103 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7104 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7105 (rl78_hard_regno_mode_ok): Make static and return bool.
7106 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7107 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7109 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7110 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7111 (rs6000_hard_regno_mode_ok): Rename to...
7112 (rs6000_hard_regno_mode_ok_uncached): ...this.
7113 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7114 (rs6000_hard_regno_mode_ok): New function.
7115 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7116 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7117 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7118 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7119 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7120 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7121 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7122 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7123 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7124 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7125 (sh_hard_regno_mode_ok): Make static.
7126 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7127 instead of HARD_REGNO_MODE_OK.
7128 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7129 (sparc_mode_class): Delete.
7130 (HARD_REGNO_MODE_OK): Delete.
7131 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7132 (hard_regno_mode_classes): Make static.
7133 (sparc_mode_class): Likewise.
7134 (sparc_hard_regno_mode_ok): New function.
7135 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7136 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7137 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7139 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7140 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7141 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7142 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7143 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7144 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7145 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7146 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7147 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7148 (visium_hard_regno_mode_ok): New function.
7149 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7150 instead of HARD_REGNO_MODE_OK.
7151 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7152 (HARD_REGNO_MODE_OK): Delete.
7153 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7154 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7155 (xtensa_option_override): Update accordingly.
7156 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7157 (xtensa_hard_regno_mode_ok): New function.
7158 * system.h (HARD_REGNO_MODE_OK): Poison.
7160 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7161 Alan Hayward <alan.hayward@arm.com>
7162 David Sherwood <david.sherwood@arm.com>
7164 * target.def (hard_regno_call_part_clobbered): New hook.
7165 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7166 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7167 * doc/tm.texi: Regenerate.
7168 * hooks.h (hook_bool_uint_mode_false): Declare.
7169 * hooks.c (hook_bool_uint_mode_false): New function.
7170 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7171 * cselib.c (cselib_process_insn): Use
7172 targetm.hard_regno_call_part_clobbered instead of
7173 HARD_REGNO_CALL_PART_CLOBBERED.
7174 * ira-conflicts.c (ira_build_conflicts): Likewise.
7175 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7176 * lra-constraints.c (need_for_call_save_p): Likewise.
7177 * lra-lives.c: Include target.h.
7178 (check_pseudos_live_through_calls): Use
7179 targetm.hard_regno_call_part_clobbered instead of
7180 HARD_REGNO_CALL_PART_CLOBBERED.
7181 * regcprop.c: Include target.h.
7182 (copyprop_hardreg_forward_1): Use
7183 targetm.hard_regno_call_part_clobbered instead of
7184 HARD_REGNO_CALL_PART_CLOBBERED.
7185 * reginfo.c (choose_hard_reg_mode): Likewise.
7186 * regrename.c (check_new_reg_p): Likewise.
7187 * reload.c (find_equiv_reg): Likewise.
7188 * reload1.c (emit_reload_insns): Likewise.
7189 * sched-deps.c (deps_analyze_insn): Likewise.
7190 * sel-sched.c (init_regs_for_mode): Likewise.
7191 (mark_unavailable_hard_regs): Likewise.
7192 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7193 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7194 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7196 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7197 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7198 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7200 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7202 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7203 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7204 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7206 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7207 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7208 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7210 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7211 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7213 * config/powerpcspe/powerpcspe.c
7214 (rs6000_hard_regno_call_part_clobbered): New function.
7215 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7216 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7217 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7219 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7220 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7221 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7223 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7224 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7225 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7227 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7228 Alan Hayward <alan.hayward@arm.com>
7229 David Sherwood <david.sherwood@arm.com>
7231 * rtl.h (subreg_memory_offset): Declare.
7232 * emit-rtl.c (subreg_memory_offset): New function.
7233 * expmed.c (store_bit_field_1): Use it.
7234 * expr.c (undefined_operand_subword_p): Likewise.
7235 * simplify-rtx.c (simplify_subreg): Likewise.
7237 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7239 PR rtl-optimization/57448
7242 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7243 using atomic_load pattern.
7244 (expand_atomic_store): Likewise.
7246 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7249 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7250 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7251 handling. Use replace_call_with_value with NULL instead of
7252 gsi_replace, unlink_stmt_vdef and release_defs.
7254 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7257 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7259 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7262 * rtl.h (emit_library_call_value_1): Declare.
7263 (emit_library_call): Replace declaration with a series of overloads.
7264 Remove the parameter count argument.
7265 (emit_library_call_value): Likewise.
7266 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7267 with an "rtx_mode_t *".
7268 (emit_library_call_value): Delete.
7269 (emit_library_call): Likewise.
7270 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7271 (asan_emit_allocas_unpoison): Likewise.
7272 * builtins.c (expand_builtin_powi): Likewise.
7273 (expand_asan_emit_allocas_unpoison): Likewise.
7274 * cfgexpand.c (expand_main_function): Likewise.
7275 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7276 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7277 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7278 * config/arm/arm.c (arm_trampoline_init): Likewise.
7279 (arm_call_tls_get_addr): Likewise.
7280 (arm_expand_divmod_libfunc): Likewise.
7281 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7282 (smulsi3_highpart): Likewise.
7283 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7284 (c6x_expand_compare): Likewise.
7285 (c6x_expand_movmem): Likewise.
7286 * config/frv/frv.c (frv_trampoline_init): Likewise.
7287 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7288 (ix86_expand_divmod_libfunc): Likewise.
7289 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7290 (ia64_expand_compare): Likewise.
7291 (ia64_profile_hook): Likewise.
7292 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7293 (nonlocal_goto): Likewise.
7294 (restore_stack_nonlocal): Likewise.
7295 * config/m32r/m32r.c (block_move_call): Likewise.
7296 (m32r_trampoline_init): Likewise.
7297 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7298 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7299 (m68k_call_m68k_read_tp): Likewise.
7300 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7301 (microblaze_expand_divide): Likewise.
7302 * config/mips/mips.h (mips_args): Likewise.
7303 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7304 (MIPS_ICACHE_SYNC): Likewise.
7305 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7306 (nios2_trampoline_init): Likewise.
7307 * config/pa/pa.c (hppa_tls_call): Likewise.
7308 (pa_trampoline_init): Likewise.
7309 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7310 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7311 (expand_strn_compare): Likewise.
7312 (rs6000_generate_compare): Likewise.
7313 (rs6000_expand_float128_convert): Likewise.
7314 (output_profile_hook): Likewise.
7315 (rs6000_trampoline_init): Likewise.
7316 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7317 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7318 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7319 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7320 (rs6000_generate_compare): Likewise.
7321 (rs6000_expand_float128_convert): Likewise.
7322 (output_profile_hook): Likewise.
7323 (rs6000_trampoline_init): Likewise.
7324 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7325 * config/sh/sh.c (sh_trampoline_init): Likewise.
7326 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7327 (sparc_emit_float_lib_cmp): Likewise.
7328 (sparc32_initialize_trampoline): Likewise.
7329 (sparc64_initialize_trampoline): Likewise.
7330 (sparc_profile_hook): Likewise.
7331 * config/spu/spu.c (ea_load_store): Likewise.
7332 * config/spu/spu.md (floatunssidf2): Likewise.
7333 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7334 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7335 * config/visium/visium.c (expand_block_move_4): Likewise.
7336 (expand_block_move_2): Likewise.
7337 (expand_block_move_1): Likewise.
7338 (expand_block_set_4): Likewise.
7339 (expand_block_set_2): Likewise.
7340 (expand_block_set_1): Likewise.
7341 (visium_trampoline_init): Likewise.
7342 (visium_profile_hook): Likewise.
7343 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7344 (xtensa_setup_frame_addresses): Likewise.
7345 (xtensa_trampoline_init): Likewise.
7346 * except.c (sjlj_emit_function_enter): Likewise.
7347 (sjlj_emit_function_exit): Likewise.
7348 * explow.c (allocate_dynamic_stack_space): Likewise.
7349 (probe_stack_range): Likewise.
7350 * expr.c (convert_mode_scalar): Likewise.
7351 * optabs.c (expand_binop): Likewise.
7352 (expand_twoval_binop_libfunc): Likewise.
7353 (expand_unop): Likewise.
7354 (prepare_cmp_insn): Likewise.
7355 (prepare_float_lib_cmp): Likewise.
7356 (expand_float): Likewise.
7357 (expand_fix): Likewise.
7358 (expand_fixed_convert): Likewise.
7359 (maybe_emit_sync_lock_test_and_set): Likewise.
7360 (expand_atomic_compare_and_swap): Likewise.
7361 (expand_mem_thread_fence): Likewise.
7362 (expand_atomic_fetch_op): Likewise.
7364 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7366 * doc/generic.texi (OpenACC): Adjust URL.
7367 * doc/invoke.texi (C Dialect Options): Ditto.
7369 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7371 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7372 predicate for operand 1. Add (m,<S>) constraint.
7373 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7374 Prevent memory operand 1 with register operand 2.
7376 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7378 PR rtl-optimization/82024
7379 * combine.c (try_combine): If the combination result is a PARALLEL,
7380 and we only need to retain the SET in there that would be placed
7381 at I2, check that we can place that at I3 instead, before doing so.
7383 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7386 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7388 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7389 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7392 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7393 Jeff Law <law@redhat.com>
7395 * varasm.c (bss_initializer_p): Do not put constants into .bss
7396 (categorize_decl_for_section): Handle bss_initializer_p returning
7397 false when DECL_INITIAL is NULL.
7399 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7402 * config/s390/s390.c (s390_can_inline_p): New function.
7404 2017-09-01 Jeff Law <law@redhat.com>
7406 PR tree-optimization/82052
7407 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7408 Always initialize the returned slot after a hash table miss
7409 when INSERT is true.
7411 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7413 * config/s390/s390.md (mem_signal_fence): Remove.
7414 * doc/md.texi (mem_signal_fence): Remove.
7415 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7417 * target-insns.def (mem_signal_fence): Remove.
7419 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7422 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7425 * asan.c (create_odr_indicator): Strip name encoding from assembler
7426 name before appending it after __odr_asan_.
7428 2017-09-01 Martin Liska <mliska@suse.cz>
7430 PR tree-optimization/82059
7431 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7432 frequency only when an edge is redirected.
7434 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7436 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7437 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7438 (arc_conditional_register_usage): Remove ARC600 lp_count
7440 (arc_file_start): Emit Tag_ARC_CPU_variation.
7441 (arc_can_use_doloop_p): New conditions to use ZOLs.
7442 (hwloop_fail): New function.
7443 (hwloop_optimize): Likewise.
7444 (hwloop_pattern_reg): Likewise.
7445 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7446 (arc_reorg_loops): New function, calls reorg_loops.
7447 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7448 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7450 (arc_loop_hazard): Remove function, functionality moved into
7452 (arc_hazard): Remove arc_loop_hazard call.
7453 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7454 into hwloop_optimize.
7455 (arc_label_align): Remove ZOL handling.
7456 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7457 * config/arc/arc.md (doloop_begin): Remove pattern.
7458 (doloop_begin_i): Likewise.
7459 (doloop_end_i): Likewise.
7460 (doloop_fallback): Likewise.
7461 (doloop_fallback_m): Likewise.
7462 (doloop_end): Reimplement expand.
7463 (arc_lp): New pattern for LP instruction.
7464 (loop_end): New pattern.
7465 (loop_fail): Likewise.
7466 (decrement_and_branch_until_zero): Likewise.
7467 * config/arc/arc.opt (mlpc-width): New option.
7468 * doc/invoke.texi (mlpc-width): Document option.
7470 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7472 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7473 (arc_ccfsm_advance): Fix checking for delay slots.
7474 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7476 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7478 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7480 (movhi_insn): Update store instruction constraint which are saving
7481 6-bit short immediates.
7482 (movsi_insn): Consider also short scaled load operations.
7483 (zero_extendhisi2_i): Use Usd constraint instead of T.
7484 (extendhisi2_i): Add q constraint.
7485 (arc_clzsi2): Add type and length attributes.
7486 (arc_ctzsi2): Likewise.
7487 * config/arc/constraints.md (Usc): Update constraint, the
7488 assembler can parse two relocations for a single instruction.
7490 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7492 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7493 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7495 2017-08-31 Olivier Hainque <hainque@adacore.com>
7497 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7498 match as powerpc-wrs-vxworks*.
7500 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7502 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7503 register constraint for by-element operand.
7504 (aarch64_mls_elt_merge<mode>): Likewise.
7506 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7508 * config/arc/arc.c (arc_can_follow_jump): Check for short
7511 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7513 * config.gcc: Use g.opt for arc.
7514 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7515 functionality moved to ...
7516 (legitimate_scaled_address_p): New function, ...here.
7517 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
7518 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
7519 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
7521 (arc_override_options): Handle G option.
7522 (arc_output_pic_addr_const): Correct function definition.
7523 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
7524 (arc_decl_anon_ns_mem_p): Delete.
7525 (arc_in_small_data_p): Overhaul this function to take into
7526 consideration the value given via G option.
7527 (arc_rewrite_small_data_1): Renamed and corrected old
7528 arc_rewrite_small_data function.
7529 (arc_rewrite_small_data): New function.
7530 (small_data_pattern): Don't use pic_offset_table_rtx.
7531 * config/arc/arc.h (CC1_SPEC): Recognize G option.
7532 * config/arc/simdext.md (movmisalignv2hi): Use
7533 prepare_move_operands function.
7535 (movmisalign*): Likewise.
7536 * doc/invoke.texi (ARC options): Document -G option.
7538 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7540 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
7542 * config/arc/arc.c (arc_print_operand): Output scalled address for
7543 sdata whenever is possible.
7544 (arc_in_small_data_p): Allow sdata for 64bit datum when double
7545 load/stores are available.
7546 (compact_sda_memory_operand): Check for the alignment required by
7547 code density instructions.
7548 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
7550 * config/arc/constraints.md (Usd): Update constraint.
7551 (Us0): New constraint.
7552 (Usc): Update constraint.
7554 2017-08-31 Richard Biener <rguenther@suse.de>
7557 * dwarf2out.c (dwarf2out_early_global_decl): Process each
7560 2017-08-31 Tamar Christina <tamar.christina@arm.com>
7562 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
7563 Resize type_signature.
7565 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7566 Alan Hayward <alan.hayward@arm.com>
7567 David Sherwood <david.sherwood@arm.com>
7569 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
7570 subregs whose inner modes can be stored in GPRs.
7571 (aarch64_classify_index): Likewise.
7573 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7574 Alan Hayward <alan.hayward@arm.com>
7575 David Sherwood <david.sherwood@arm.com>
7577 * config/aarch64/iterators.md (V_cmp_result): Rename to...
7578 (V_INT_EQUIV): ...this.
7579 (v_cmp_result): Rename to...
7580 (v_int_equiv): ...this.
7581 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
7582 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
7583 (copysign<mode>3): Likewise.
7584 (aarch64_simd_bsl<mode>_internal): Likewise.
7585 (aarch64_simd_bsl<mode>): Likewise.
7586 (vec_cmp<mode><mode>): Likewise.
7587 (vcond<mode><mode>): Likewise.
7588 (vcond<v_cmp_mixed><mode>): Likewise.
7589 (vcondu<mode><v_cmp_mixed>): Likewise.
7590 (aarch64_cm<optab><mode>): Likewise.
7591 (aarch64_cmtst<mode>): Likewise.
7592 (aarch64_fac<optab><mode>): Likewise.
7593 (vec_perm_const<mode>): Likewise.
7594 (vcond_mask_<mode><v_cmp_result>): Rename to...
7595 (vcond_mask_<mode><v_int_equiv>): ...this.
7596 (vec_cmp<mode><v_cmp_result>): Rename to...
7597 (vec_cmp<mode><v_int_equiv>): ...this.
7599 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7600 Alan Hayward <alan.hayward@arm.com>
7601 David Sherwood <david.sherwood@arm.com>
7603 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7605 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7606 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7607 (UNSPEC_LD4_DREG): New unspecs.
7608 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7609 (aarch64_ld2<mode>_dreg_be): Replace with...
7610 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7612 (aarch64_ld3<mode>_dreg_le)
7613 (aarch64_ld3<mode>_dreg_be): Replace with...
7614 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7616 (aarch64_ld4<mode>_dreg_le)
7617 (aarch64_ld4<mode>_dreg_be): Replace with...
7618 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7621 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7623 PR tree-optimization/81987
7624 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7625 insert an initializer in a location not dominated by the stride
7628 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7630 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7631 on the entire header of the finally block in the fallthru case.
7633 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7635 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7637 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7639 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7640 rs6000_emit_move_from_cr and call renamed function.
7641 (rs6000_emit_prologue): Call renamed functions.
7642 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7643 movesi_from_cr, remove volatile CRs.
7645 2017-08-30 Jon Beniston <jon@beniston.com>
7646 Richard Biener <rguenther@suse.de>
7648 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7649 of VECTOR_MODE_P check.
7650 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7651 element vector types.
7653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7655 * df.h (df_read_modify_subreg_p): Remove in favor of...
7656 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7657 const_rtx instead of an rtx.
7658 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7659 * df-problems.c (df_word_lr_mark_ref): Likewise.
7660 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7661 (mark_pseudo_reg_dead): Likewise.
7662 (mark_ref_dead): Likewise.
7663 * reginfo.c (init_subregs_of_mode): Likewise.
7664 * sched-deps.c (sched_analyze_1): Likewise.
7665 * df-scan.c (df_def_record_1): Likewise.
7666 (df_uses_record): Likewise.
7667 (df_read_modify_subreg_p): Remove in favor of...
7668 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7669 const_rtx instead of an rtx.
7671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7672 Alan Hayward <alan.hayward@arm.com>
7673 David Sherwood <david.sherwood@arm.com>
7675 * rtl.h (partial_subreg_p): New function.
7676 * caller-save.c (save_call_clobbered_regs): Use it.
7677 * calls.c (expand_call): Likewise.
7678 * combine.c (combinable_i3pat): Likewise.
7679 (simplify_set): Likewise.
7680 (make_extraction): Likewise.
7681 (make_compound_operation_int): Likewise.
7682 (gen_lowpart_or_truncate): Likewise.
7683 (force_to_mode): Likewise.
7684 (make_field_assignment): Likewise.
7685 (reg_truncated_to_mode): Likewise.
7686 (record_truncated_value): Likewise.
7687 (move_deaths): Likewise.
7688 * cse.c (record_jump_cond): Likewise.
7689 (cse_insn): Likewise.
7690 * cselib.c (cselib_lookup_1): Likewise.
7691 * expmed.c (extract_bit_field_using_extv): Likewise.
7692 * function.c (assign_parm_setup_reg): Likewise.
7693 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7694 * ira-build.c (create_insn_allocnos): Likewise.
7695 * lra-coalesce.c (merge_pseudos): Likewise.
7696 * lra-constraints.c (match_reload): Likewise.
7697 (simplify_operand_subreg): Likewise.
7698 (curr_insn_transform): Likewise.
7699 * lra-lives.c (process_bb_lives): Likewise.
7700 * lra.c (new_insn_reg): Likewise.
7701 (lra_substitute_pseudo): Likewise.
7702 * regcprop.c (mode_change_ok): Likewise.
7703 (maybe_mode_change): Likewise.
7704 (copyprop_hardreg_forward_1): Likewise.
7705 * reload.c (push_reload): Likewise.
7706 (find_reloads): Likewise.
7707 (find_reloads_subreg_address): Likewise.
7708 * reload1.c (alter_reg): Likewise.
7709 (eliminate_regs_1): Likewise.
7710 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7712 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7714 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7715 back to if statements, including unpack.
7717 2017-08-30 Martin Liska <mliska@suse.cz>
7720 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7722 (func_checker::compare_asm_inputs_outputs): ... this function.
7723 (func_checker::compare_gimple_asm): Use the function to compare
7724 also ASM constrains.
7725 * ipa-icf-gimple.h: Rename the function.
7727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7728 Alan Hayward <alan.hayward@arm.com>
7729 David Sherwood <david.sherwood@arm.com>
7731 * coretypes.h (complex_mode): New type.
7732 * gdbhooks.py (build_pretty_printer): Handle it.
7733 * machmode.h (complex_mode): New class.
7734 (complex_mode::includes_p): New function.
7735 (is_complex_int_mode): Likewise.
7736 (is_complex_float_mode): Likewise.
7737 * genmodes.c (get_mode_class): Handle complex mode classes.
7738 * function.c (expand_function_end): Use is_complex_int_mode.
7740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7741 Alan Hayward <alan.hayward@arm.com>
7742 David Sherwood <david.sherwood@arm.com>
7744 * coretypes.h (scalar_mode_pod): New typedef.
7745 * gdbhooks.py (build_pretty_printer): Handle it.
7746 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7747 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7748 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7749 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7753 Alan Hayward <alan.hayward@arm.com>
7754 David Sherwood <david.sherwood@arm.com>
7756 * machmode.h (mode_for_vector): Take a scalar_mode instead
7758 * stor-layout.c (mode_for_vector): Likewise.
7759 * explow.c (promote_mode): Use as_a <scalar_mode>.
7760 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7763 Alan Hayward <alan.hayward@arm.com>
7764 David Sherwood <david.sherwood@arm.com>
7766 * target.def (preferred_simd_mode): Take a scalar_mode
7767 instead of a machine_mode.
7768 * targhooks.h (default_preferred_simd_mode): Likewise.
7769 * targhooks.c (default_preferred_simd_mode): Likewise.
7770 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7771 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7772 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7773 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7774 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7775 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7776 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7777 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7779 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7780 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7781 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7782 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7783 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7784 * doc/tm.texi: Regenerate.
7785 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7789 Alan Hayward <alan.hayward@arm.com>
7790 David Sherwood <david.sherwood@arm.com>
7792 * target.def (scalar_mode_supported_p): Take a scalar_mode
7793 instead of a machine_mode.
7794 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7795 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7796 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7797 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7798 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7799 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7800 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7801 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7802 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7803 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7804 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7805 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7806 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7807 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7809 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7810 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7811 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7812 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7813 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7815 * doc/tm.texi: Regenerate.
7817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7818 Alan Hayward <alan.hayward@arm.com>
7819 David Sherwood <david.sherwood@arm.com>
7821 * coretypes.h (opt_scalar_mode): New typedef.
7822 * gdbhooks.py (build_pretty_printers): Handle it.
7823 * machmode.h (mode_iterator::get_2xwider): Add overload for
7825 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7827 * expr.c (convert_mode_scalar): Likewise.
7828 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7829 * optabs.c (expand_float): Likewise.
7830 (expand_fix): Likewise.
7831 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7834 Alan Hayward <alan.hayward@arm.com>
7835 David Sherwood <david.sherwood@arm.com>
7837 * optabs.c (expand_float): Explicitly check for scalars before
7838 using a branching expansion.
7839 (expand_fix): Likewise.
7841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7842 Alan Hayward <alan.hayward@arm.com>
7843 David Sherwood <david.sherwood@arm.com>
7845 * expr.c (convert_mode): Split scalar handling out into...
7846 (convert_mode_scalar): ...this new function. Treat the modes
7849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7850 Alan Hayward <alan.hayward@arm.com>
7851 David Sherwood <david.sherwood@arm.com>
7853 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7855 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7858 Alan Hayward <alan.hayward@arm.com>
7859 David Sherwood <david.sherwood@arm.com>
7861 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7862 rather than a machine_mode.
7863 (fixed_from_string): Likewise.
7864 (fixed_convert): Likewise.
7865 (fixed_convert_from_int): Likewise.
7866 (fixed_convert_from_real): Likewise.
7867 (real_convert_from_fixed): Likewise.
7868 * fixed-value.c (fixed_from_double_int): Likewise.
7869 (fixed_from_string): Likewise.
7870 (fixed_convert): Likewise.
7871 (fixed_convert_from_int): Likewise.
7872 (fixed_convert_from_real): Likewise.
7873 (real_convert_from_fixed): Likewise.
7874 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7877 Alan Hayward <alan.hayward@arm.com>
7878 David Sherwood <david.sherwood@arm.com>
7880 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7881 of separate mode class checks. Do not allow vector modes here.
7882 (immed_wide_int_const): Use as_a <scalar_mode>.
7883 * explow.c (trunc_int_for_mode): Likewise.
7884 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7885 (wi::shwi): Likewise.
7886 (wi::min_value): Likewise.
7887 (wi::max_value): Likewise.
7888 * dwarf2out.c (loc_descriptor): Likewise.
7889 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7893 Alan Hayward <alan.hayward@arm.com>
7894 David Sherwood <david.sherwood@arm.com>
7896 * tree.h (SCALAR_TYPE_MODE): New macro.
7897 * expr.c (expand_expr_addr_expr_1): Use it.
7898 (expand_expr_real_2): Likewise.
7899 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7900 (fold_convert_const_fixed_from_int): Likewise.
7901 (fold_convert_const_fixed_from_real): Likewise.
7902 (native_encode_fixed): Likewise
7903 (native_encode_complex): Likewise
7904 (native_encode_vector): Likewise.
7905 (native_interpret_fixed): Likewise.
7906 (native_interpret_real): Likewise.
7907 (native_interpret_complex): Likewise.
7908 (native_interpret_vector): Likewise.
7909 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7910 (simd_clone_adjust_argument_types): Likewise.
7911 (simd_clone_init_simd_arrays): Likewise.
7912 (simd_clone_adjust): Likewise.
7913 * stor-layout.c (layout_type): Likewise.
7914 * tree.c (build_minus_one_cst): Likewise.
7915 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7916 * tree-inline.c (estimate_move_cost): Likewise.
7917 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7918 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7919 (vectorizable_reduction): Likewise.
7920 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7921 (vect_recog_mixed_size_cond_pattern): Likewise.
7922 (check_bool_pattern): Likewise.
7923 (adjust_bool_pattern): Likewise.
7924 (search_type_for_mask_1): Likewise.
7925 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7926 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7927 (vectorizable_load): Likewise.
7928 (vectorizable_store): Likewise.
7929 * ubsan.c (ubsan_encode_value): Likewise.
7930 * varasm.c (output_constant): Likewise.
7932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7933 Alan Hayward <alan.hayward@arm.com>
7934 David Sherwood <david.sherwood@arm.com>
7936 * coretypes.h (scalar_mode): New class.
7937 * machmode.h (scalar_mode): Likewise.
7938 (scalar_mode::includes_p): New function.
7939 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7940 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7941 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7942 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7943 * expmed.c (store_bit_field_1): Likewise.
7944 (extract_bit_field_1): Likewise.
7945 * expr.c (write_complex_part): Likewise.
7946 (read_complex_part): Likewise.
7947 (emit_move_complex_push): Likewise.
7948 (expand_expr_real_2): Likewise.
7949 * function.c (assign_parm_setup_reg): Likewise.
7950 (assign_parms_unsplit_complex): Likewise.
7951 * optabs.c (expand_binop): Likewise.
7952 * rtlanal.c (subreg_get_info): Likewise.
7953 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7954 * varasm.c (output_constant_pool_2): Likewise.
7956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7957 Alan Hayward <alan.hayward@arm.com>
7958 David Sherwood <david.sherwood@arm.com>
7960 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7962 (expmed_mult_highpart_optab): Likewise.
7963 (expmed_mult_highpart): Likewise.
7965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7966 Alan Hayward <alan.hayward@arm.com>
7967 David Sherwood <david.sherwood@arm.com>
7969 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7970 instead of a machine_mode.
7971 (builtin_memset_read_str): Likewise.
7972 * builtins.c (c_readstr): Likewise.
7973 (builtin_memcpy_read_str): Likewise.
7974 (builtin_strncpy_read_str): Likewise.
7975 (builtin_memset_read_str): Likewise.
7976 (builtin_memset_gen_str): Likewise.
7977 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7978 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7979 instead of a machine_mode.
7980 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7982 (make_extraction): Likewise.
7983 (try_widen_shift_mode): Take and return scalar_int_modes instead
7985 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7986 a scalar_int_mode instead of a machine_mode.
7987 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7988 (avr_addr_space_pointer_mode): Likewise.
7989 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7990 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7991 (msp430_unwind_word_mode): Likewise.
7992 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7993 (spu_addr_space_pointer_mode): Likewise.
7994 (spu_addr_space_address_mode): Likewise.
7995 (spu_libgcc_cmp_return_mode): Likewise.
7996 (spu_libgcc_shift_count_mode): Likewise.
7997 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7998 (rl78_addr_space_pointer_mode): Likewise.
7999 (fl78_unwind_word_mode): Likewise.
8000 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8002 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8003 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8004 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8005 (mips_valid_pointer_mode): Likewise.
8006 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8007 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8008 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8010 (ft32_addr_space_address_mode): Likewise.
8011 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8012 scalar_int_mode instead of a machine_mode.
8013 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8015 (m32c_addr_space_address_mode): Likewise.
8016 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8017 (rs6000_eh_return_filter_mode): Likewise.
8018 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8019 (rs6000_eh_return_filter_mode): Likewise.
8020 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8021 (s390_libgcc_shift_count_mode): Likewise.
8022 (s390_unwind_word_mode): Likewise.
8023 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8025 * target.def (mode_rep_extended): Likewise.
8026 (valid_pointer_mode): Likewise.
8027 (addr_space.valid_pointer_mode): Likewise.
8028 (eh_return_filter_mode): Return a scalar_int_mode rather than
8030 (libgcc_cmp_return_mode): Likewise.
8031 (libgcc_shift_count_mode): Likewise.
8032 (unwind_word_mode): Likewise.
8033 (addr_space.pointer_mode): Likewise.
8034 (addr_space.address_mode): Likewise.
8035 * doc/tm.texi: Regenerate.
8036 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8038 (do_jump): Use scalar_int_mode for local variables.
8039 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8040 rather than a machine_mode.
8041 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8042 (scompare_loc_descriptor_wide): Likewise.
8043 (scompare_loc_descriptor_narrow): Likewise.
8044 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8046 * except.c (sjlj_emit_dispatch_table): Likewise.
8047 (expand_builtin_eh_copy_values): Likewise.
8048 * explow.c (convert_memory_address_addr_space_1): Likewise.
8049 Take a scalar_int_mode rather than a machine_mode.
8050 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8051 than a machine_mode.
8052 (memory_address_addr_space): Use scalar_int_mode for local variables.
8053 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8054 rather than a machine_mode.
8055 * expmed.c (mask_rtx): Likewise.
8056 (init_expmed_one_conv): Likewise.
8057 (expand_mult_highpart_adjust): Likewise.
8058 (extract_high_half): Likewise.
8059 (expmed_mult_highpart_optab): Likewise.
8060 (expmed_mult_highpart): Likewise.
8061 (expand_smod_pow2): Likewise.
8062 (expand_sdiv_pow2): Likewise.
8063 (emit_store_flag_int): Likewise.
8064 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8066 (extract_low_bits): Likewise.
8067 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8069 * expr.c (pieces_addr::adjust): Likewise.
8070 (can_store_by_pieces): Likewise.
8071 (store_by_pieces): Likewise.
8072 (clear_by_pieces_1): Likewise.
8073 (expand_expr_addr_expr_1): Likewise.
8074 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8075 (expand_expr_real_1): Likewise.
8076 (try_casesi): Likewise.
8077 * final.c (shorten_branches): Likewise.
8078 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8079 type of "mode" to machine_mode.
8080 * internal-fn.c (expand_arith_overflow_result_store): Take a
8081 scalar_int_mode rather than a machine_mode.
8082 (expand_mul_overflow): Use scalar_int_mode for local variables.
8083 * loop-doloop.c (doloop_modify): Likewise.
8084 (doloop_optimize): Likewise.
8085 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8086 than a machine_mode.
8087 (expand_doubleword_shift_condmove): Likewise.
8088 (expand_doubleword_shift): Likewise.
8089 (expand_doubleword_clz): Likewise.
8090 (expand_doubleword_popcount): Likewise.
8091 (expand_doubleword_parity): Likewise.
8092 (expand_absneg_bit): Use scalar_int_mode for local variables.
8093 (prepare_float_lib_cmp): Likewise.
8094 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8095 rather than a machine_mode.
8096 (convert_memory_address_addr_space): Likewise.
8097 (get_mode_bounds): Likewise.
8098 (get_address_mode): Return a scalar_int_mode rather than a
8100 * rtlanal.c (get_address_mode): Likewise.
8101 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8102 than a machine_mode.
8103 * targhooks.c (default_mode_rep_extended): Likewise.
8104 (default_valid_pointer_mode): Likewise.
8105 (default_addr_space_valid_pointer_mode): Likewise.
8106 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8107 than a machine_mode.
8108 (default_libgcc_cmp_return_mode): Likewise.
8109 (default_libgcc_shift_count_mode): Likewise.
8110 (default_unwind_word_mode): Likewise.
8111 (default_addr_space_pointer_mode): Likewise.
8112 (default_addr_space_address_mode): Likewise.
8113 * targhooks.h (default_eh_return_filter_mode): Likewise.
8114 (default_libgcc_cmp_return_mode): Likewise.
8115 (default_libgcc_shift_count_mode): Likewise.
8116 (default_unwind_word_mode): Likewise.
8117 (default_addr_space_pointer_mode): Likewise.
8118 (default_addr_space_address_mode): Likewise.
8119 (default_mode_rep_extended): Take a scalar_int_mode rather than
8121 (default_valid_pointer_mode): Likewise.
8122 (default_addr_space_valid_pointer_mode): Likewise.
8123 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8125 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8126 rather than a machine_mode.
8127 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8128 for local variables.
8129 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8130 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8131 than a machine_mode.
8133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8134 Alan Hayward <alan.hayward@arm.com>
8135 David Sherwood <david.sherwood@arm.com>
8137 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8138 the mode argument to scalar_int_mode.
8139 (do_jump_by_parts_zero_rtx): Likewise.
8140 (do_jump_by_parts_equality_rtx): Likewise.
8141 (do_jump_by_parts_greater): Take a mode argument.
8142 (do_jump_by_parts_equality): Likewise.
8143 (do_jump_1): Update calls accordingly.
8145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8146 Alan Hayward <alan.hayward@arm.com>
8147 David Sherwood <david.sherwood@arm.com>
8149 * is-a.h (safe_dyn_cast): New function.
8150 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8151 (jump_table_for_label): Likewise.
8152 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8153 instead of an rtx_insn *.
8154 (shorten_branches): Use dyn_cast instead of LABEL_P and
8155 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8156 rtx_jump_table_data::get_data_mode.
8157 (final_scan_insn): Likewise.
8159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8160 Alan Hayward <alan.hayward@arm.com>
8161 David Sherwood <david.sherwood@arm.com>
8163 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8164 trying to combine a full-register integer set with a subreg
8167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8168 Alan Hayward <alan.hayward@arm.com>
8169 David Sherwood <david.sherwood@arm.com>
8171 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8172 that is always either address_mode or pointer_mode.
8174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8175 Alan Hayward <alan.hayward@arm.com>
8176 David Sherwood <david.sherwood@arm.com>
8178 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8179 when the two are known to be equal.
8181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8182 Alan Hayward <alan.hayward@arm.com>
8183 David Sherwood <david.sherwood@arm.com>
8185 * simplify-rtx.c (simplify_const_unary_operation): Use
8186 is_a <scalar_int_mode> instead of checking for a nonzero
8187 precision. Forcibly convert op_mode to a scalar_int_mode
8188 in that case. More clearly differentiate the operand and
8189 result modes and use the former when deciding what the value
8190 of a count-bits operation should be. Use is_int_mode instead
8191 of checking for a MODE_INT. Remove redundant check for whether
8192 this mode has a zero precision.
8194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8195 Alan Hayward <alan.hayward@arm.com>
8196 David Sherwood <david.sherwood@arm.com>
8198 * optabs.c (widen_leading): Change the type of the mode argument
8199 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8200 (widen_bswap): Likewise.
8201 (expand_parity): Likewise.
8202 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8203 (expand_ffs): Likewise.
8204 (epand_unop): Check for scalar integer modes before calling the
8207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8208 Alan Hayward <alan.hayward@arm.com>
8209 David Sherwood <david.sherwood@arm.com>
8211 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8213 (expand_expr_real_1): Update call accordingly.
8215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8216 Alan Hayward <alan.hayward@arm.com>
8217 David Sherwood <david.sherwood@arm.com>
8219 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8220 value_mode arguments. Use scalar_int_mode internally.
8221 (store_bit_field_1): Rename the new integer mode from imode
8222 to op0_mode and use it instead of GET_MODE (op0). Update calls
8223 to store_split_bit_field, store_bit_field_using_insv and
8224 store_fixed_bit_field.
8225 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8226 Use scalar_int_mode internally. Use a bit count rather than a mode
8227 when calculating the largest bit size for get_best_mode.
8228 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8229 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8230 Remove assertion that OP0 has a scalar integer mode.
8231 (store_split_bit_field): Add op0_mode and value_mode arguments.
8232 Update calls to extract_fixed_bit_field.
8233 (extract_bit_field_using_extv): Add an op0_mode argument.
8234 Use scalar_int_mode internally.
8235 (extract_bit_field_1): Rename the new integer mode from imode to
8236 op0_mode and use it instead of GET_MODE (op0). Update calls to
8237 extract_split_bit_field, extract_bit_field_using_extv and
8238 extract_fixed_bit_field.
8239 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8240 to extract_split_bit_field and extract_fixed_bit_field_1.
8241 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8242 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8244 (extract_split_bit_field): Add an op0_mode argument. Update call
8245 to extract_fixed_bit_field.
8247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8248 Alan Hayward <alan.hayward@arm.com>
8249 David Sherwood <david.sherwood@arm.com>
8251 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8252 * explow.c (hard_function_value): Likewise.
8253 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8254 convert_to_mode call outside the loop.
8255 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8256 for the mode iterator. Require the mode specified by max_pieces
8258 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8260 (copy_blkmode_to_reg): Likewise.
8261 (set_storage_via_setmem): Likewise.
8262 * optabs.c (prepare_cmp_insn): Likewise.
8263 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8264 * stor-layout.c (finish_bitfield_representative): Likewise.
8266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8267 Alan Hayward <alan.hayward@arm.com>
8268 David Sherwood <david.sherwood@arm.com>
8270 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8271 * expr.c (convert_move): Use them.
8272 (convert_modes): Likewise.
8273 (store_expr_with_bounds): Likewise.
8275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8276 Alan Hayward <alan.hayward@arm.com>
8277 David Sherwood <david.sherwood@arm.com>
8279 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8280 parameter for the mode of "x". Remove the "known_x", "known_mode"
8281 and "known_ret" arguments. Change the type of the mode argument
8283 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8284 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8285 (reg_num_sign_bit_copies_for_combine): Likewise.
8286 * rtlanal.c (nonzero_bits1): Likewise.
8287 (num_sign_bit_copies1): Likewise.
8288 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8289 (reg_num_sign_bit_copies_general): Likewise.
8290 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8291 (reg_nonzero_bits_general): Likewise.
8293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8294 Alan Hayward <alan.hayward@arm.com>
8295 David Sherwood <david.sherwood@arm.com>
8297 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8298 than in subroutines. Return 1 for non-integer modes.
8299 (cached_num_sign_bit_copies): Change the type of the mode parameter
8301 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8302 classes. Handle CONST_INT_P first and then check whether X also
8303 has a scalar integer mode. Check the same thing for inner registers
8304 of a SUBREG and for values that are being extended or truncated.
8306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8307 Alan Hayward <alan.hayward@arm.com>
8308 David Sherwood <david.sherwood@arm.com>
8310 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8311 in subroutines. Return the mode mask for non-integer modes.
8312 (cached_nonzero_bits): Change the type of the mode parameter
8314 (nonzero_bits1): Likewise. Remove early exit for other mode
8315 classes. Handle CONST_INT_P first and then check whether X
8316 also has a scalar integer mode.
8318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8319 Alan Hayward <alan.hayward@arm.com>
8320 David Sherwood <david.sherwood@arm.com>
8322 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8323 Return a scalar_int_mode and assert that the size is greater than
8325 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8326 (op_by_pieces_d::op_by_pieces_d): Likewise.
8327 (op_by_pieces_d::run): Likewise.
8328 (can_store_by_pieces): Likewise.
8330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8331 Alan Hayward <alan.hayward@arm.com>
8332 David Sherwood <david.sherwood@arm.com>
8334 * combine.c (extract_left_shift): Add a mode argument and update
8336 (make_compound_operation_int): Change the type of the mode parameter
8337 to scalar_int_mode and update the call to extract_left_shift.
8339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8340 Alan Hayward <alan.hayward@arm.com>
8341 David Sherwood <david.sherwood@arm.com>
8343 * combine.c (simplify_and_const_int): Change the type of the mode
8344 parameter to scalar_int_mode.
8345 (simplify_and_const_int_1): Likewise. Update recursive call.
8347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8348 Alan Hayward <alan.hayward@arm.com>
8349 David Sherwood <david.sherwood@arm.com>
8351 * combine.c (simplify_compare_const): Check that the mode is a
8352 scalar_int_mode (rather than VOIDmode) before testing its
8354 (simplify_comparison): Move COMPARISON_P handling out of the
8355 loop and restrict the latter part of the loop to scalar_int_modes.
8356 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8357 and when considering SUBREG_REGs. Use is_int_mode instead of
8358 checking GET_MODE_CLASS against MODE_INT.
8360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8361 Alan Hayward <alan.hayward@arm.com>
8362 David Sherwood <david.sherwood@arm.com>
8364 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8365 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8366 shift_unit_mode and for modes involved in scalar shifts.
8368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8369 Alan Hayward <alan.hayward@arm.com>
8370 David Sherwood <david.sherwood@arm.com>
8372 * combine.c (force_int_to_mode): New function, split out from...
8373 (force_to_mode): ...here. Keep xmode up-to-date and use it
8374 instead of GET_MODE (x).
8376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8377 Alan Hayward <alan.hayward@arm.com>
8378 David Sherwood <david.sherwood@arm.com>
8380 * optabs-query.h (extraction_insn::struct_mode): Change type to
8381 opt_scalar_int_mode and update comment.
8382 (extraction_insn::field_mode): Change type to scalar_int_mode.
8383 (extraction_insn::pos_mode): Likewise.
8384 * combine.c (make_extraction): Update accordingly.
8385 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8386 (get_optab_extraction_insn): Likewise.
8387 * recog.c (simplify_while_replacing): Likewise.
8388 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8389 parameter to opt_scalar_int_mode.
8391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8392 Alan Hayward <alan.hayward@arm.com>
8393 David Sherwood <david.sherwood@arm.com>
8395 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8396 to a scalar_int_mode instead of a machine_mode.
8397 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8398 (get_best_mode): Return a boolean and use a pointer argument to store
8399 the selected mode. Replace the limit mode parameter with a bit limit.
8400 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8401 for the values returned by bit_field_mode_iterator::next_mode.
8402 (store_bit_field): Update call to get_best_mode.
8403 (store_fixed_bit_field): Likewise.
8404 (extract_fixed_bit_field): Likewise.
8405 * expr.c (optimize_bitfield_assignment_op): Likewise.
8406 * fold-const.c (optimize_bit_field_compare): Likewise.
8407 (fold_truth_andor_1): Likewise.
8408 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8409 Update for new type of m_mode.
8410 (get_best_mode): As above.
8412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8413 Alan Hayward <alan.hayward@arm.com>
8414 David Sherwood <david.sherwood@arm.com>
8416 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8417 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8418 (store_bit_field): Check is_a <scalar_int_mode> before calling
8419 strict_volatile_bitfield_p.
8420 (extract_bit_field): Likewise.
8422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8423 Alan Hayward <alan.hayward@arm.com>
8424 David Sherwood <david.sherwood@arm.com>
8426 * target.def (cstore_mode): Return a scalar_int_mode.
8427 * doc/tm.texi: Regenerate.
8428 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8429 * targhooks.h (default_cstore_mode): Likewise.
8430 * targhooks.c (default_cstore_mode): Likewise, using a forced
8432 * expmed.c (emit_cstore): Expect the target of the cstore to be
8435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8436 Alan Hayward <alan.hayward@arm.com>
8437 David Sherwood <david.sherwood@arm.com>
8439 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8441 (niter_desc): Likewise mode.
8442 (iv_analyze): Add a mode parameter.
8444 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8445 and change its type to scalar_int_mode.
8446 * loop-iv.c: Update commentary at head of file.
8447 (iv_constant): Pass the mode paraeter before the rtx it describes
8448 and change its type to scalar_int_mode. Remove VOIDmode handling.
8449 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8450 (iv_extend): Likewise.
8451 (shorten_into_mode): Likewise.
8452 (iv_add): Use scalar_int_mode.
8453 (iv_mult): Likewise.
8454 (iv_shift): Likewise.
8455 (canonicalize_iv_subregs): Likewise.
8456 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8457 it describes and change its mode to scalar_int_mode. Also change
8458 the type of the returned inner_mode to scalar_int_mode.
8459 (get_biv_step): Likewise, turning outer_mode from a pointer
8460 into a direct parameter. Update call to get_biv_step_1.
8461 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8462 iv_constant and get_biv_step.
8463 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8464 and change its type to scalar_int_mode. Don't initialise iv->mode
8465 to VOIDmode and remove later checks for its still being VOIDmode.
8466 Update calls to iv_analyze_op and iv_analyze_expr. Check
8467 is_a <scalar_int_mode> when changing the mode under consideration.
8468 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8469 Update call to iv_analyze_expr.
8470 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8471 inner register is not also a scalar_int_mode. Update call to
8473 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8474 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8475 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8476 separate mode class checks. Update calls to iv_analyze. Remove
8477 fix-up of VOIDmodes after iv_analyze_biv.
8478 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8479 don't have a scalar_int_mode. Update call to biv_p.
8481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8482 Alan Hayward <alan.hayward@arm.com>
8483 David Sherwood <david.sherwood@arm.com>
8485 * cfgexpand.c (convert_debug_memory_address): Use
8486 as_a <scalar_int_mode>.
8487 * combine.c (expand_compound_operation): Likewise.
8488 (make_extraction): Likewise.
8489 (change_zero_ext): Likewise.
8490 (simplify_comparison): Likewise.
8491 * cse.c (cse_insn): Likewise.
8492 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8493 (mem_loc_descriptor): Likewise.
8494 (loc_descriptor): Likewise.
8495 * expmed.c (init_expmed_one_mode): Likewise.
8496 (synth_mult): Likewise.
8497 (emit_store_flag_1): Likewise.
8498 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8499 of a comparison with size.
8500 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8501 (reduce_to_bit_field_precision): Likewise.
8502 * function.c (expand_function_end): Likewise.
8503 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8504 * loop-doloop.c (doloop_modify): Likewise.
8505 * optabs.c (expand_binop): Likewise.
8506 (expand_unop): Likewise.
8507 (expand_copysign_absneg): Likewise.
8508 (prepare_cmp_insn): Likewise.
8509 (maybe_legitimize_operand): Likewise.
8510 * recog.c (const_scalar_int_operand): Likewise.
8511 * rtlanal.c (get_address_mode): Likewise.
8512 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8513 (simplify_cond_clz_ctz): Likewise.
8514 * tree-nested.c (get_nl_goto_field): Likewise.
8515 * tree.c (build_vector_type_for_mode): Likewise.
8516 * var-tracking.c (use_narrower_mode): Likewise.
8518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8519 Alan Hayward <alan.hayward@arm.com>
8520 David Sherwood <david.sherwood@arm.com>
8522 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
8523 * builtins.c (expand_builtin_signbit): Use it.
8524 * cfgexpand.c (expand_debug_expr): Likewise.
8525 * dojump.c (do_jump): Likewise.
8526 (do_compare_and_jump): Likewise.
8527 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
8528 * expmed.c (make_tree): Likewise.
8529 * expr.c (expand_expr_real_2): Likewise.
8530 (expand_expr_real_1): Likewise.
8531 (try_casesi): Likewise.
8532 * fold-const-call.c (fold_const_call_ss): Likewise.
8533 * fold-const.c (unextend): Likewise.
8534 (extract_muldiv_1): Likewise.
8535 (fold_single_bit_test): Likewise.
8536 (native_encode_int): Likewise.
8537 (native_encode_string): Likewise.
8538 (native_interpret_int): Likewise.
8539 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
8540 * internal-fn.c (expand_addsub_overflow): Likewise.
8541 (expand_neg_overflow): Likewise.
8542 (expand_mul_overflow): Likewise.
8543 (expand_arith_overflow): Likewise.
8544 * match.pd: Likewise.
8545 * stor-layout.c (layout_type): Likewise.
8546 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8547 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
8548 * tree-ssanames.c (get_range_info): Likewise.
8549 * tree-switch-conversion.c (array_value_type) Likewise.
8550 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
8551 (vect_recog_divmod_pattern): Likewise.
8552 (vect_recog_mixed_size_cond_pattern): Likewise.
8553 * tree-vrp.c (extract_range_basic): Likewise.
8554 (simplify_float_conversion_using_ranges): Likewise.
8555 * tree.c (int_fits_type_p): Likewise.
8556 * ubsan.c (instrument_bool_enum_load): Likewise.
8557 * varasm.c (mergeable_string_section): Likewise.
8558 (narrowing_initializer_constant_valid_p): Likewise.
8559 (output_constant): Likewise.
8561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8562 Alan Hayward <alan.hayward@arm.com>
8563 David Sherwood <david.sherwood@arm.com>
8565 * machmode.h (NARROWEST_INT_MODE): New macro.
8566 * expr.c (alignment_for_piecewise_move): Use it instead of
8567 GET_CLASS_NARROWEST_MODE (MODE_INT).
8568 (push_block): Likewise.
8569 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
8571 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8574 Alan Hayward <alan.hayward@arm.com>
8575 David Sherwood <david.sherwood@arm.com>
8577 * postreload.c (move2add_valid_value_p): Change the type of the
8578 mode parameter to scalar_int_mode.
8579 (move2add_use_add2_insn): Add a mode parameter and use it instead
8581 (move2add_use_add3_insn): Likewise.
8582 (reload_cse_move2add): Update accordingly.
8584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8585 Alan Hayward <alan.hayward@arm.com>
8586 David Sherwood <david.sherwood@arm.com>
8588 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
8590 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
8591 * optabs.c (expand_unop): Likewise.
8593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8594 Alan Hayward <alan.hayward@arm.com>
8595 David Sherwood <david.sherwood@arm.com>
8597 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
8598 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
8599 (popcount_loc_descriptor): Likewise.
8600 (bswap_loc_descriptor): Likewise.
8601 (rotate_loc_descriptor): Likewise.
8602 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8603 calling the functions above.
8605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8606 Alan Hayward <alan.hayward@arm.com>
8607 David Sherwood <david.sherwood@arm.com>
8609 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8611 (try_combine): Likewise.
8612 (simplify_if_then_else): Likewise.
8613 * cse.c (cse_insn): Likewise.
8614 * dwarf2out.c (mem_loc_descriptor): Likewise.
8615 * emit-rtl.c (gen_lowpart_common): Likewise.
8616 * simplify-rtx.c (simplify_truncation): Likewise.
8617 (simplify_binary_operation_1): Likewise.
8618 (simplify_const_relational_operation): Likewise.
8619 (simplify_ternary_operation): Likewise.
8620 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
8622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8623 Alan Hayward <alan.hayward@arm.com>
8624 David Sherwood <david.sherwood@arm.com>
8626 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8627 * reload.c (push_reload): Likewise.
8628 (find_reloads): Likewise.
8630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8631 Alan Hayward <alan.hayward@arm.com>
8632 David Sherwood <david.sherwood@arm.com>
8634 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8635 (make_compound_operation_int): Likewise.
8636 (change_zero_ext): Likewise.
8637 * expr.c (convert_move): Likewise.
8638 (convert_modes): Likewise.
8639 * fwprop.c (forward_propagate_subreg): Likewise.
8640 * loop-iv.c (get_biv_step_1): Likewise.
8641 * optabs.c (widen_operand): Likewise.
8642 * postreload.c (move2add_valid_value_p): Likewise.
8643 * recog.c (simplify_while_replacing): Likewise.
8644 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8645 (simplify_binary_operation_1): Likewise. Remove redundant
8646 mode equality check.
8648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8649 Alan Hayward <alan.hayward@arm.com>
8650 David Sherwood <david.sherwood@arm.com>
8652 * combine.c (combine_simplify_rtx): Add checks for
8653 is_a <scalar_int_mode>.
8654 (simplify_if_then_else): Likewise.
8655 (make_field_assignment): Likewise.
8656 (simplify_comparison): Likewise.
8657 * ifcvt.c (noce_try_bitop): Likewise.
8658 * loop-invariant.c (canonicalize_address_mult): Likewise.
8659 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8662 Alan Hayward <alan.hayward@arm.com>
8663 David Sherwood <david.sherwood@arm.com>
8665 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8666 is_a <scalar_int_mode> instead of != BLKmode.
8668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8669 Alan Hayward <alan.hayward@arm.com>
8670 David Sherwood <david.sherwood@arm.com>
8672 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8673 instead of != VOIDmode.
8674 * combine.c (if_then_else_cond): Likewise.
8675 (change_zero_ext): Likewise.
8676 * dwarf2out.c (mem_loc_descriptor): Likewise.
8677 (loc_descriptor): Likewise.
8678 * rtlanal.c (canonicalize_condition): Likewise.
8679 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
8681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8682 Alan Hayward <alan.hayward@arm.com>
8683 David Sherwood <david.sherwood@arm.com>
8685 * simplify-rtx.c (simplify_binary_operation_1): Use
8686 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8689 Alan Hayward <alan.hayward@arm.com>
8690 David Sherwood <david.sherwood@arm.com>
8692 * wide-int.h (int_traits<unsigned char>) New class.
8693 (int_traits<unsigned short>) Likewise.
8694 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8695 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8697 * combine.c (set_nonzero_bits_and_sign_copies): Use
8698 is_a <scalar_int_mode>.
8699 (find_split_point): Likewise.
8700 (combine_simplify_rtx): Likewise.
8701 (simplify_logical): Likewise.
8702 (expand_compound_operation): Likewise.
8703 (expand_field_assignment): Likewise.
8704 (make_compound_operation): Likewise.
8705 (extended_count): Likewise.
8706 (change_zero_ext): Likewise.
8707 (simplify_comparison): Likewise.
8708 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8709 (ucompare_loc_descriptor): Likewise.
8710 (minmax_loc_descriptor): Likewise.
8711 (mem_loc_descriptor): Likewise.
8712 (loc_descriptor): Likewise.
8713 * expmed.c (init_expmed_one_mode): Likewise.
8714 * lra-constraints.c (lra_constraint_offset): Likewise.
8715 * optabs.c (prepare_libcall_arg): Likewise.
8716 * postreload.c (move2add_note_store): Likewise.
8717 * reload.c (operands_match_p): Likewise.
8718 * rtl.h (load_extend_op): Likewise.
8719 * rtlhooks.c (gen_lowpart_general): Likewise.
8720 * simplify-rtx.c (simplify_truncation): Likewise.
8721 (simplify_unary_operation_1): Likewise.
8722 (simplify_binary_operation_1): Likewise.
8723 (simplify_const_binary_operation): Likewise.
8724 (simplify_const_relational_operation): Likewise.
8725 (simplify_subreg): Likewise.
8726 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8727 * var-tracking.c (adjust_mems): Likewise.
8728 (prepare_call_arguments): Likewise.
8730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8731 Alan Hayward <alan.hayward@arm.com>
8732 David Sherwood <david.sherwood@arm.com>
8734 * machmode.h (is_int_mode): New fuction.
8735 * combine.c (find_split_point): Use it.
8736 (combine_simplify_rtx): Likewise.
8737 (simplify_if_then_else): Likewise.
8738 (simplify_set): Likewise.
8739 (simplify_shift_const_1): Likewise.
8740 (simplify_comparison): Likewise.
8741 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8742 * cse.c (notreg_cost): Likewise.
8743 (cse_insn): Likewise.
8744 * cselib.c (cselib_lookup_1): Likewise.
8745 * dojump.c (do_jump_1): Likewise.
8746 (do_compare_rtx_and_jump): Likewise.
8747 * dse.c (get_call_args): Likewise.
8748 * dwarf2out.c (rtl_for_decl_init): Likewise.
8749 (native_encode_initializer): Likewise.
8750 * expmed.c (emit_store_flag_1): Likewise.
8751 (emit_store_flag): Likewise.
8752 * expr.c (convert_modes): Likewise.
8753 (store_field): Likewise.
8754 (expand_expr_real_1): Likewise.
8755 * fold-const.c (fold_read_from_constant_string): Likewise.
8756 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8757 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8758 * optabs.c (expand_binop): Likewise.
8759 (expand_unop): Likewise.
8760 (expand_abs_nojump): Likewise.
8761 (expand_one_cmpl_abs_nojump): Likewise.
8762 * simplify-rtx.c (mode_signbit_p): Likewise.
8763 (val_signbit_p): Likewise.
8764 (val_signbit_known_set_p): Likewise.
8765 (val_signbit_known_clear_p): Likewise.
8766 (simplify_relational_operation_1): Likewise.
8767 * tree.c (vector_type_mode): Likewise.
8769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8770 Alan Hayward <alan.hayward@arm.com>
8771 David Sherwood <david.sherwood@arm.com>
8773 * machmode.h (smallest_mode_for_size): Fix formatting.
8774 (smallest_int_mode_for_size): New function.
8775 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8776 instead of smallest_mode_for_size.
8777 * combine.c (make_extraction): Likewise.
8778 * config/arc/arc.c (arc_expand_movmem): Likewise.
8779 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8780 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8781 * config/s390/s390.c (s390_expand_insv): Likewise.
8782 * config/sparc/sparc.c (assign_int_registers): Likewise.
8783 * config/spu/spu.c (spu_function_value): Likewise.
8784 (spu_function_arg): Likewise.
8785 * coverage.c (get_gcov_type): Likewise.
8786 (get_gcov_unsigned_t): Likewise.
8787 * dse.c (find_shift_sequence): Likewise.
8788 * expmed.c (store_bit_field_1): Likewise.
8789 * expr.c (convert_move): Likewise.
8790 (store_field): Likewise.
8791 * internal-fn.c (expand_arith_overflow): Likewise.
8792 * optabs-query.c (get_best_extraction_insn): Likewise.
8793 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8794 * stor-layout.c (layout_type): Likewise.
8795 (initialize_sizetypes): Likewise.
8796 * targhooks.c (default_get_mask_mode): Likewise.
8797 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8800 Alan Hayward <alan.hayward@arm.com>
8801 David Sherwood <david.sherwood@arm.com>
8803 * machmode.h (opt_mode::else_blk): New function.
8804 (int_mode_for_mode): Declare.
8805 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8806 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8808 * cfgexpand.c (expand_debug_expr): Likewise.
8809 * combine.c (gen_lowpart_or_truncate): Likewise.
8810 (gen_lowpart_for_combine): Likewise.
8811 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8812 * config/avr/avr.c (avr_to_int_mode): Likewise.
8813 (avr_out_plus_1): Likewise.
8814 (avr_out_plus): Likewise.
8815 (avr_out_round): Likewise.
8816 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8817 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8818 (s390_expand_vcond): Likewise.
8819 * config/spu/spu.c (spu_split_immediate): Likewise.
8820 (spu_expand_mov): Likewise.
8821 * dse.c (get_stored_val): Likewise.
8822 * expmed.c (store_bit_field_1): Likewise.
8823 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8825 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8826 (extract_low_bits): Likewise.
8827 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8828 handling rather than repeating the check.
8829 (emit_group_store): Likewise.
8830 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8831 * optabs.c (expand_absneg_bit): Likewise.
8832 (expand_copysign_absneg): Likewise.
8833 (expand_copysign_bit): Likewise.
8834 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8835 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8836 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8837 * var-tracking.c (prepare_call_arguments): Likewise.
8838 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8839 int_mode_for_mode instead of mode_for_size.
8840 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8843 Alan Hayward <alan.hayward@arm.com>
8844 David Sherwood <david.sherwood@arm.com>
8846 * machmode.h (int_mode_for_size): New function.
8847 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8848 instead of mode_for_size.
8849 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8851 * combine.c (expand_field_assignment): Use int_mode_for_size
8852 instead of mode_for_size.
8853 (make_extraction): Likewise.
8854 (simplify_shift_const_1): Likewise.
8855 (simplify_comparison): Likewise.
8856 * dojump.c (do_jump): Likewise.
8857 * dwarf2out.c (mem_loc_descriptor): Likewise.
8858 * emit-rtl.c (init_derived_machine_modes): Likewise.
8859 * expmed.c (flip_storage_order): Likewise.
8860 (convert_extracted_bit_field): Likewise.
8861 * expr.c (copy_blkmode_from_reg): Likewise.
8862 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8863 * internal-fn.c (expand_mul_overflow): Likewise.
8864 * lower-subreg.c (simple_move): Likewise.
8865 * optabs-libfuncs.c (init_optabs): Likewise.
8866 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8867 * tree.c (vector_type_mode): Likewise.
8868 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8869 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8870 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8871 * tree-vect-stmts.c (vectorizable_load): Likewise.
8872 (vectorizable_store): Likewise.
8874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8875 Alan Hayward <alan.hayward@arm.com>
8876 David Sherwood <david.sherwood@arm.com>
8878 * coretypes.h (pod_mode): New type.
8879 (scalar_int_mode_pod): New typedef.
8880 * machmode.h (pod_mode): New class.
8881 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8882 * genmodes.c (emit_mode_int_n): Update accordingly.
8883 * lower-subreg.h (target_lower_subreg): Change type to
8884 scalar_int_mode_pod.
8885 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8886 scalar_int_mode_pod.
8888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8889 Alan Hayward <alan.hayward@arm.com>
8890 David Sherwood <david.sherwood@arm.com>
8892 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8893 machine_mode to scalar_int_mode.
8894 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8895 (rs6000_option_override_internal): Remove cast to int.
8896 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8897 machine_mode to scalar_int_mode.
8898 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8899 (rs6000_option_override_internal): Remove cast to int.
8900 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8901 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8903 * config/s390/s390.c (s390_expand_builtin): Likewise.
8904 * coretypes.h (scalar_int_mode): New type.
8905 (opt_scalar_int_mode): New typedef.
8906 * machmode.h (scalar_int_mode): New class.
8907 (scalar_int_mode::includes_p): New function.
8908 (byte_mode): Change type to scalar_int_mode.
8909 (word_mode): Likewise.
8910 (ptr_mode): Likewise.
8911 * emit-rtl.c (byte_mode): Likewise.
8912 (word_mode): Likewise.
8913 (ptr_mode): Likewise.
8914 (init_derived_machine_modes): Update accordingly.
8915 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8916 and MODE_PARTIAL_INT.
8917 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8918 opt_scalar_int_mode.
8920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8921 Alan Hayward <alan.hayward@arm.com>
8922 David Sherwood <david.sherwood@arm.com>
8924 * target.def (libgcc_floating_mode_supported_p): Take a
8926 * doc/tm.texi: Regenerate.
8927 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8929 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8930 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8934 Alan Hayward <alan.hayward@arm.com>
8935 David Sherwood <david.sherwood@arm.com>
8937 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8938 * doc/tm.texi: Regenerate.
8939 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8940 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8941 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8942 * targhooks.h (default_floatn_mode): Likewise.
8943 * targhooks.c (default_floatn_mode): Likewise.
8944 * tree.c (build_common_tree_nodes): Update accordingly.
8946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8947 Alan Hayward <alan.hayward@arm.com>
8948 David Sherwood <david.sherwood@arm.com>
8950 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8951 (mode_iterator::iterate_p): Likewise.
8952 (mode_iterator::get_wider): Likewise.
8953 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8956 Alan Hayward <alan.hayward@arm.com>
8957 David Sherwood <david.sherwood@arm.com>
8959 * coretypes.h (opt_scalar_float_mode): New typedef.
8960 * machmode.h (float_mode_for_size): New function.
8961 * emit-rtl.c (double_mode): Delete.
8962 (init_emit_once): Use float_mode_for_size.
8963 * stor-layout.c (layout_type): Likewise.
8964 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8967 Alan Hayward <alan.hayward@arm.com>
8968 David Sherwood <david.sherwood@arm.com>
8970 * output.h (assemble_real): Take a scalar_float_mode.
8971 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8972 * config/arm/arm.md (consttable_4): Likewise.
8973 (consttable_8): Likewise.
8974 (consttable_16): Likewise.
8975 * config/mips/mips.md (consttable_float): Likewise.
8976 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8977 * varasm.c (assemble_real): Take a scalar_float_mode.
8978 (output_constant_pool_2): Update accordingly.
8979 (output_constant): Likewise.
8981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8982 Alan Hayward <alan.hayward@arm.com>
8983 David Sherwood <david.sherwood@arm.com>
8985 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8986 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8987 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8988 (native_encode_real): Likewise.
8989 (native_interpret_real): Likewise.
8990 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8991 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8994 Alan Hayward <alan.hayward@arm.com>
8995 David Sherwood <david.sherwood@arm.com>
8997 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8998 <scalar_float_mode>. Simplify.
8999 (gen_extend_conv_libfunc): Likewise.
9001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9002 Alan Hayward <alan.hayward@arm.com>
9003 David Sherwood <david.sherwood@arm.com>
9005 * coretypes.h (scalar_float_mode): New type.
9006 * machmode.h (mode_traits::from_int): Use machine_mode if
9007 USE_ENUM_MODES is defined.
9008 (is_a): New function.
9010 (dyn_cast): Likewise.
9011 (scalar_float_mode): New class.
9012 (scalar_float_mode::includes_p): New function.
9013 (is_float_mode): Likewise.
9014 * gdbhooks.py (MachineModePrinter): New class.
9015 (build_pretty_printer): Use it for scalar_float_mode.
9016 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9017 (format_helper::format_helper): Turn into a template.
9018 * genmodes.c (get_mode_class): New function.
9019 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9020 or machine_mode if none.
9021 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9022 as_a <scalar_float_mode>.
9023 * dwarf2out.c (mem_loc_descriptor): Likewise.
9024 (insert_float): Likewise.
9025 (add_const_value_attribute): Likewise.
9026 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9027 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9028 (expand_unop): Update accordingly.
9029 (expand_abs_nojump): Likewise.
9030 (expand_copysign_absneg): Take a scalar_float_mode.
9031 (expand_copysign_bit): Likewise.
9032 (expand_copysign): Update accordingly.
9034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9035 Alan Hayward <alan.hayward@arm.com>
9036 David Sherwood <david.sherwood@arm.com>
9038 * coretypes.h (opt_mode): New class.
9039 * machmode.h (opt_mode): Likewise.
9040 (opt_mode::else_void): New function.
9041 (opt_mode::require): Likewise.
9042 (opt_mode::exists): Likewise.
9043 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9044 (GET_MODE_2XWIDER_MODE): Likewise.
9045 (mode_iterator::get_wider): Update accordingly.
9046 (mode_iterator::get_2xwider): Likewise.
9047 (mode_iterator::get_known_wider): Likewise, turning into a template.
9048 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9049 forcing a wider mode to exist.
9050 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9051 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9052 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9053 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9054 * lower-subreg.c (init_lower_subreg): Likewise.
9055 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9056 on the final iteration.
9057 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9058 a wider mode exists before asking for a move pattern.
9059 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9060 forcing a wider mode to exist.
9061 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9062 returning false if no such mode exists.
9063 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9064 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9065 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9066 Avoid checking for a MODE_INT if we already know the mode is not a
9068 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9069 forcing a wider mode to exist.
9070 (expmed_mult_highpart_optab): Likewise.
9071 (expmed_mult_highpart): Likewise.
9072 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9074 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9075 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9076 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9077 * internal-fn.c (expand_mul_overflow): Update use of
9078 GET_MODE_2XWIDER_MODE.
9079 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9080 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9081 GET_MODE_WIDER_MODE.
9082 (convert_plusminus_to_widen): Likewise.
9083 * tree-switch-conversion.c (array_value_type): Likewise.
9084 * var-tracking.c (emit_note_insn_var_location): Likewise.
9085 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9086 Return false inside rather than outside the loop if no wider mode
9088 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9089 and GET_MODE_2XWIDER_MODE
9090 (can_compare_p): Use else_void.
9091 * gdbhooks.py (OptMachineModePrinter): New class.
9092 (build_pretty_printer): Use it for opt_mode.
9094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9095 Alan Hayward <alan.hayward@arm.com>
9096 David Sherwood <david.sherwood@arm.com>
9098 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9099 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9102 Alan Hayward <alan.hayward@arm.com>
9103 David Sherwood <david.sherwood@arm.com>
9105 * machmode.h (mode_traits): New structure.
9106 (get_narrowest_mode): New function.
9107 (mode_iterator::start): Likewise.
9108 (mode_iterator::iterate_p): Likewise.
9109 (mode_iterator::get_wider): Likewise.
9110 (mode_iterator::get_known_wider): Likewise.
9111 (mode_iterator::get_2xwider): Likewise.
9112 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9113 (FOR_EACH_MODE): Likewise.
9114 (FOR_EACH_MODE_FROM): Likewise.
9115 (FOR_EACH_MODE_UNTIL): Likewise.
9116 (FOR_EACH_WIDER_MODE): Likewise.
9117 (FOR_EACH_2XWIDER_MODE): Likewise.
9118 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9119 * combine.c (simplify_comparison): Likewise
9120 * config/i386/i386.c (type_natural_mode): Likewise.
9121 * cse.c (cse_insn): Likewise.
9122 * dse.c (find_shift_sequence): Likewise.
9123 * emit-rtl.c (init_derived_machine_modes): Likewise.
9124 (init_emit_once): Likewise.
9125 * explow.c (hard_function_value): Likewise.
9126 * expmed.c (extract_fixed_bit_field_1): Likewise.
9127 (extract_bit_field_1): Likewise.
9128 (expand_divmod): Likewise.
9129 (emit_store_flag_1): Likewise.
9130 * expr.c (init_expr_target): Likewise.
9131 (convert_move): Likewise.
9132 (alignment_for_piecewise_move): Likewise.
9133 (widest_int_mode_for_size): Likewise.
9134 (emit_block_move_via_movmem): Likewise.
9135 (copy_blkmode_to_reg): Likewise.
9136 (set_storage_via_setmem): Likewise.
9137 (compress_float_constant): Likewise.
9138 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9139 * optabs-query.c (get_best_extraction_insn): Likewise.
9140 * optabs.c (expand_binop): Likewise.
9141 (expand_twoval_unop): Likewise.
9142 (expand_twoval_binop): Likewise.
9143 (widen_leading): Likewise.
9144 (widen_bswap): Likewise.
9145 (expand_parity): Likewise.
9146 (expand_unop): Likewise.
9147 (prepare_cmp_insn): Likewise.
9148 (prepare_float_lib_cmp): Likewise.
9149 (expand_float): Likewise.
9150 (expand_fix): Likewise.
9151 (expand_sfix_optab): Likewise.
9152 * postreload.c (move2add_use_add2_insn): Likewise.
9153 * reg-stack.c (reg_to_stack): Likewise.
9154 * reginfo.c (choose_hard_reg_mode): Likewise.
9155 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9156 * stor-layout.c (mode_for_size): Likewise.
9157 (smallest_mode_for_size): Likewise.
9158 (mode_for_vector): Likewise.
9159 (finish_bitfield_representative): Likewise.
9160 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9161 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9162 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9163 * var-tracking.c (prepare_call_arguments): Likewise.
9165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9166 Alan Hayward <alan.hayward@arm.com>
9167 David Sherwood <david.sherwood@arm.com>
9169 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9170 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9171 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9172 * machmode.h (mode_size): Move earlier in file.
9173 (mode_precision): Likewise.
9174 (mode_inner): Likewise.
9175 (mode_nunits): Likewise.
9176 (mode_unit_size): Likewise.
9177 (unit_unit_precision): Likewise.
9178 (mode_wider): Likewise.
9179 (mode_2xwider): Likewise.
9180 (machine_mode): New class.
9181 (mode_to_bytes): New function.
9182 (mode_to_bits): Likewise.
9183 (mode_to_precision): Likewise.
9184 (mode_to_inner): Likewise.
9185 (mode_to_unit_size): Likewise.
9186 (mode_to_unit_precision): Likewise.
9187 (mode_to_nunits): Likewise.
9188 (GET_MODE_SIZE): Use mode_to_bytes.
9189 (GET_MODE_BITSIZE): Use mode_to_bits.
9190 (GET_MODE_PRECISION): Use mode_to_precision.
9191 (GET_MODE_INNER): Use mode_to_inner.
9192 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9193 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9194 (GET_MODE_NUNITS): Use mode_to_nunits.
9195 * system.h (ALWAYS_INLINE): New macro.
9196 * config/powerpcspe/powerpcspe-c.c
9197 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9198 int for arg1_mode and arg2_mode.
9200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9201 Alan Hayward <alan.hayward@arm.com>
9202 David Sherwood <david.sherwood@arm.com>
9204 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9205 Prefix mode names with E_ in case statements.
9206 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9207 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9208 (aarch64_split_simd_move): Likewise.
9209 (aarch64_gen_storewb_pair): Likewise.
9210 (aarch64_gen_loadwb_pair): Likewise.
9211 (aarch64_gen_store_pair): Likewise.
9212 (aarch64_gen_load_pair): Likewise.
9213 (aarch64_get_condition_code_1): Likewise.
9214 (aarch64_constant_pool_reload_icode): Likewise.
9215 (get_rsqrte_type): Likewise.
9216 (get_rsqrts_type): Likewise.
9217 (get_recpe_type): Likewise.
9218 (get_recps_type): Likewise.
9219 (aarch64_gimplify_va_arg_expr): Likewise.
9220 (aarch64_simd_container_mode): Likewise.
9221 (aarch64_emit_load_exclusive): Likewise.
9222 (aarch64_emit_store_exclusive): Likewise.
9223 (aarch64_expand_compare_and_swap): Likewise.
9224 (aarch64_gen_atomic_cas): Likewise.
9225 (aarch64_emit_bic): Likewise.
9226 (aarch64_emit_atomic_swap): Likewise.
9227 (aarch64_emit_atomic_load_op): Likewise.
9228 (aarch64_evpc_trn): Likewise.
9229 (aarch64_evpc_uzp): Likewise.
9230 (aarch64_evpc_zip): Likewise.
9231 (aarch64_evpc_ext): Likewise.
9232 (aarch64_evpc_rev): Likewise.
9233 (aarch64_evpc_dup): Likewise.
9234 (aarch64_gen_ccmp_first): Likewise.
9235 (aarch64_gen_ccmp_next): Likewise.
9236 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9237 (alpha_emit_xfloating_libcall): Likewise.
9238 (emit_insxl): Likewise.
9239 (alpha_arg_type): Likewise.
9240 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9241 (arc_preferred_simd_mode): Likewise.
9242 (arc_secondary_reload): Likewise.
9243 (get_arc_condition_code): Likewise.
9244 (arc_print_operand): Likewise.
9245 (arc_legitimate_constant_p): Likewise.
9246 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9247 * config/arc/arc.md (casesi_load): Likewise.
9248 (casesi_compact_jump): Likewise.
9249 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9250 (cc_use_register): Likewise.
9251 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9252 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9253 (arm_init_iwmmxt_builtins): Likewise.
9254 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9255 (neon_expand_vector_init): Likewise.
9256 (arm_attr_length_move_neon): Likewise.
9257 (maybe_get_arm_condition_code): Likewise.
9258 (arm_emit_vector_const): Likewise.
9259 (arm_preferred_simd_mode): Likewise.
9260 (arm_output_iwmmxt_tinsr): Likewise.
9261 (thumb1_output_casesi): Likewise.
9262 (thumb2_output_casesi): Likewise.
9263 (arm_emit_load_exclusive): Likewise.
9264 (arm_emit_store_exclusive): Likewise.
9265 (arm_expand_compare_and_swap): Likewise.
9266 (arm_evpc_neon_vuzp): Likewise.
9267 (arm_evpc_neon_vzip): Likewise.
9268 (arm_evpc_neon_vrev): Likewise.
9269 (arm_evpc_neon_vtrn): Likewise.
9270 (arm_evpc_neon_vext): Likewise.
9271 (arm_validize_comparison): Likewise.
9272 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9273 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9274 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9275 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9276 (c6x_preferred_simd_mode): Likewise.
9277 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9278 (epiphany_rtx_costs): Likewise.
9279 * config/epiphany/predicates.md (proper_comparison_operator):
9281 * config/frv/frv.c (condexec_memory_operand): Likewise.
9282 (frv_emit_move): Likewise.
9283 (output_move_single): Likewise.
9284 (output_condmove_single): Likewise.
9285 (frv_hard_regno_mode_ok): Likewise.
9286 (frv_matching_accg_mode): Likewise.
9287 * config/h8300/h8300.c (split_adds_subs): Likewise.
9288 (h8300_rtx_costs): Likewise.
9289 (h8300_print_operand): Likewise.
9290 (compute_mov_length): Likewise.
9291 (output_logical_op): Likewise.
9292 (compute_logical_op_length): Likewise.
9293 (compute_logical_op_cc): Likewise.
9294 (h8300_shift_needs_scratch_p): Likewise.
9295 (output_a_shift): Likewise.
9296 (compute_a_shift_length): Likewise.
9297 (compute_a_shift_cc): Likewise.
9298 (expand_a_rotate): Likewise.
9299 (output_a_rotate): Likewise.
9300 * config/i386/i386.c (classify_argument): Likewise.
9301 (function_arg_advance_32): Likewise.
9302 (function_arg_32): Likewise.
9303 (function_arg_64): Likewise.
9304 (function_value_64): Likewise.
9305 (ix86_gimplify_va_arg): Likewise.
9306 (ix86_legitimate_constant_p): Likewise.
9307 (put_condition_code): Likewise.
9308 (split_double_mode): Likewise.
9309 (ix86_avx256_split_vector_move_misalign): Likewise.
9310 (ix86_expand_vector_logical_operator): Likewise.
9311 (ix86_split_idivmod): Likewise.
9312 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9313 (ix86_build_const_vector): Likewise.
9314 (ix86_build_signbit_mask): Likewise.
9315 (ix86_match_ccmode): Likewise.
9316 (ix86_cc_modes_compatible): Likewise.
9317 (ix86_expand_branch): Likewise.
9318 (ix86_expand_sse_cmp): Likewise.
9319 (ix86_expand_sse_movcc): Likewise.
9320 (ix86_expand_int_sse_cmp): Likewise.
9321 (ix86_expand_vec_perm_vpermi2): Likewise.
9322 (ix86_expand_vec_perm): Likewise.
9323 (ix86_expand_sse_unpack): Likewise.
9324 (ix86_expand_int_addcc): Likewise.
9325 (ix86_split_to_parts): Likewise.
9326 (ix86_vectorize_builtin_gather): Likewise.
9327 (ix86_vectorize_builtin_scatter): Likewise.
9328 (avx_vpermilp_parallel): Likewise.
9329 (inline_memory_move_cost): Likewise.
9330 (ix86_tieable_integer_mode_p): Likewise.
9331 (x86_maybe_negate_const_int): Likewise.
9332 (ix86_expand_vector_init_duplicate): Likewise.
9333 (ix86_expand_vector_init_one_nonzero): Likewise.
9334 (ix86_expand_vector_init_one_var): Likewise.
9335 (ix86_expand_vector_init_concat): Likewise.
9336 (ix86_expand_vector_init_interleave): Likewise.
9337 (ix86_expand_vector_init_general): Likewise.
9338 (ix86_expand_vector_set): Likewise.
9339 (ix86_expand_vector_extract): Likewise.
9340 (emit_reduc_half): Likewise.
9341 (ix86_emit_i387_round): Likewise.
9342 (ix86_mangle_type): Likewise.
9343 (ix86_expand_round_sse4): Likewise.
9344 (expand_vec_perm_blend): Likewise.
9345 (canonicalize_vector_int_perm): Likewise.
9346 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9347 (expand_vec_perm_1): Likewise.
9348 (expand_vec_perm_interleave3): Likewise.
9349 (expand_vec_perm_even_odd_pack): Likewise.
9350 (expand_vec_perm_even_odd_1): Likewise.
9351 (expand_vec_perm_broadcast_1): Likewise.
9352 (ix86_vectorize_vec_perm_const_ok): Likewise.
9353 (ix86_expand_vecop_qihi): Likewise.
9354 (ix86_expand_mul_widen_hilo): Likewise.
9355 (ix86_expand_sse2_abs): Likewise.
9356 (ix86_expand_pextr): Likewise.
9357 (ix86_expand_pinsr): Likewise.
9358 (ix86_preferred_simd_mode): Likewise.
9359 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9360 * config/i386/sse.md (*andnot<mode>3): Likewise.
9361 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9362 (*<code><mode>3): Likewise.
9363 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9364 (ia64_expand_atomic_op): Likewise.
9365 (ia64_arg_type): Likewise.
9366 (ia64_mode_to_int): Likewise.
9367 (ia64_scalar_mode_supported_p): Likewise.
9368 (ia64_vector_mode_supported_p): Likewise.
9369 (expand_vec_perm_broadcast): Likewise.
9370 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9371 (iq2000_function_arg_advance): Likewise.
9372 (iq2000_function_arg): Likewise.
9373 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9374 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9375 (m68k_libcall_value): Likewise.
9376 (m68k_function_value): Likewise.
9377 (sched_attr_op_type): Likewise.
9378 * config/mcore/mcore.c (mcore_output_move): Likewise.
9379 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9381 (microblaze_function_arg): Likewise.
9382 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9383 (mips_print_operand): Likewise.
9384 (mips_mode_ok_for_mov_fmt_p): Likewise.
9385 (mips_vector_mode_supported_p): Likewise.
9386 (mips_preferred_simd_mode): Likewise.
9387 (mips_expand_vpc_loongson_even_odd): Likewise.
9388 (mips_expand_vec_unpack): Likewise.
9389 (mips_expand_vi_broadcast): Likewise.
9390 (mips_expand_vector_init): Likewise.
9391 (mips_expand_vec_reduc): Likewise.
9392 (mips_expand_msa_cmp): Likewise.
9393 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9394 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9395 (cc_flags_for_mode): Likewise.
9396 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9397 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9398 (nds32_output_casesi_pc_relative): Likewise.
9399 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9400 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9401 (nvptx_gen_unpack): Likewise.
9402 (nvptx_gen_pack): Likewise.
9403 (nvptx_gen_shuffle): Likewise.
9404 (nvptx_gen_wcast): Likewise.
9405 (nvptx_preferred_simd_mode): Likewise.
9406 * config/pa/pa.c (pa_secondary_reload): Likewise.
9407 * config/pa/predicates.md (base14_operand): Likewise.
9408 * config/powerpcspe/powerpcspe-c.c
9409 (altivec_resolve_overloaded_builtin): Likewise.
9410 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9412 (rs6000_preferred_simd_mode): Likewise.
9413 (output_vec_const_move): Likewise.
9414 (rs6000_expand_vector_extract): Likewise.
9415 (rs6000_split_vec_extract_var): Likewise.
9416 (reg_offset_addressing_ok_p): Likewise.
9417 (rs6000_legitimate_offset_address_p): Likewise.
9418 (rs6000_legitimize_address): Likewise.
9419 (rs6000_emit_set_const): Likewise.
9420 (rs6000_const_vec): Likewise.
9421 (rs6000_emit_move): Likewise.
9422 (spe_build_register_parallel): Likewise.
9423 (rs6000_darwin64_record_arg_recurse): Likewise.
9424 (swap_selector_for_mode): Likewise.
9425 (spe_init_builtins): Likewise.
9426 (paired_init_builtins): Likewise.
9427 (altivec_init_builtins): Likewise.
9428 (do_load_for_compare): Likewise.
9429 (rs6000_generate_compare): Likewise.
9430 (rs6000_expand_float128_convert): Likewise.
9431 (emit_load_locked): Likewise.
9432 (emit_store_conditional): Likewise.
9433 (rs6000_output_function_epilogue): Likewise.
9434 (rs6000_handle_altivec_attribute): Likewise.
9435 (rs6000_function_value): Likewise.
9436 (emit_fusion_gpr_load): Likewise.
9437 (emit_fusion_p9_load): Likewise.
9438 (emit_fusion_p9_store): Likewise.
9439 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9440 (fusion_gpr_mem_load): Likewise.
9441 (fusion_addis_mem_combo_load): Likewise.
9442 (fusion_addis_mem_combo_store): Likewise.
9443 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9444 (fusion_gpr_mem_load): Likewise.
9445 (fusion_addis_mem_combo_load): Likewise.
9446 (fusion_addis_mem_combo_store): Likewise.
9447 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9449 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9450 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9451 (rs6000_preferred_simd_mode): Likewise.
9452 (output_vec_const_move): Likewise.
9453 (rs6000_expand_vector_extract): Likewise.
9454 (rs6000_split_vec_extract_var): Likewise.
9455 (reg_offset_addressing_ok_p): Likewise.
9456 (rs6000_legitimate_offset_address_p): Likewise.
9457 (rs6000_legitimize_address): Likewise.
9458 (rs6000_emit_set_const): Likewise.
9459 (rs6000_const_vec): Likewise.
9460 (rs6000_emit_move): Likewise.
9461 (rs6000_darwin64_record_arg_recurse): Likewise.
9462 (swap_selector_for_mode): Likewise.
9463 (paired_init_builtins): Likewise.
9464 (altivec_init_builtins): Likewise.
9465 (rs6000_expand_float128_convert): Likewise.
9466 (emit_load_locked): Likewise.
9467 (emit_store_conditional): Likewise.
9468 (rs6000_output_function_epilogue): Likewise.
9469 (rs6000_handle_altivec_attribute): Likewise.
9470 (rs6000_function_value): Likewise.
9471 (emit_fusion_gpr_load): Likewise.
9472 (emit_fusion_p9_load): Likewise.
9473 (emit_fusion_p9_store): Likewise.
9474 * config/rx/rx.c (rx_gen_move_template): Likewise.
9475 (flags_from_mode): Likewise.
9476 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9477 (s390_slb_comparison): Likewise.
9478 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9479 (s390_vector_mode_supported_p): Likewise.
9480 (s390_cc_modes_compatible): Likewise.
9481 (s390_match_ccmode_set): Likewise.
9482 (s390_canonicalize_comparison): Likewise.
9483 (s390_emit_compare_and_swap): Likewise.
9484 (s390_branch_condition_mask): Likewise.
9485 (s390_rtx_costs): Likewise.
9486 (s390_secondary_reload): Likewise.
9487 (__SECONDARY_RELOAD_CASE): Likewise.
9488 (s390_expand_cs): Likewise.
9489 (s390_preferred_simd_mode): Likewise.
9490 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9491 * config/sh/sh.c (sh_print_operand): Likewise.
9492 (dump_table): Likewise.
9493 (sh_secondary_reload): Likewise.
9494 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9495 * config/sh/sh.md (casesi_worker_1): Likewise.
9496 (casesi_worker_2): Likewise.
9497 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9498 (fcc_comparison_operator): Likewise.
9499 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9500 (emit_soft_tfmode_cvt): Likewise.
9501 (sparc_preferred_simd_mode): Likewise.
9502 (output_cbranch): Likewise.
9503 (sparc_print_operand): Likewise.
9504 (sparc_expand_vec_perm_bmask): Likewise.
9505 (vector_init_bshuffle): Likewise.
9506 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9507 (spu_vector_mode_supported_p): Likewise.
9508 (spu_expand_insv): Likewise.
9509 (spu_emit_branch_or_set): Likewise.
9510 (spu_handle_vector_attribute): Likewise.
9511 (spu_builtin_splats): Likewise.
9512 (spu_builtin_extract): Likewise.
9513 (spu_builtin_promote): Likewise.
9514 (spu_expand_sign_extend): Likewise.
9515 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9516 (tilegx_simd_int): Likewise.
9517 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
9518 (tilepro_simd_int): Likewise.
9519 * config/v850/v850.c (const_double_split): Likewise.
9520 (v850_print_operand): Likewise.
9521 (ep_memory_offset): Likewise.
9522 * config/vax/vax.c (vax_rtx_costs): Likewise.
9523 (vax_output_int_move): Likewise.
9524 (vax_output_int_add): Likewise.
9525 (vax_output_int_subtract): Likewise.
9526 * config/visium/predicates.md (visium_branch_operator): Likewise.
9527 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
9528 (visium_print_operand_address): Likewise.
9529 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9530 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
9531 (xtensa_expand_conditional_branch): Likewise.
9532 (xtensa_copy_incoming_a7): Likewise.
9533 (xtensa_output_literal): Likewise.
9534 * dfp.c (decimal_real_maxval): Likewise.
9535 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9538 Alan Hayward <alan.hayward@arm.com>
9539 David Sherwood <david.sherwood@arm.com>
9541 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
9542 (mode_nunits_inline): Likewise.
9543 (mode_inner_inline): Likewise.
9544 (mode_unit_size_inline): Likewise.
9545 (mode_unit_precision_inline): Likewise.
9546 (emit_insn_modes_h): Likewise. Also emit a #define of the
9548 (emit_mode_wider): Add an E_ prefix to mode names.
9549 (emit_mode_complex): Likewise.
9550 (emit_mode_inner): Likewise.
9551 (emit_mode_adjustments): Likewise.
9552 (emit_mode_int_n): Likewise.
9553 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
9554 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
9555 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
9556 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
9557 (CRC32_BUILTIN, ENTRY): Likewise.
9558 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
9559 (aarch64_pop_regs): Likewise.
9560 (aarch64_process_components): Likewise.
9561 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
9562 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
9563 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
9564 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
9565 * config/arm/arm.c (arm_init_libfuncs): Likewise.
9566 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
9568 * config/i386/i386-builtin.def (pcmpestr): Likewise.
9569 (pcmpistr): Likewise.
9570 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
9571 * config/mmix/mmix.c (mmix_output_condition): Likewise.
9572 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9574 * config/rl78/rl78.c (mduc_regs): Likewise.
9575 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9576 (htm_expand_builtin): Likewise.
9577 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
9578 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
9579 * config/xtensa/xtensa.c (print_operand): Likewise.
9580 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
9581 (NUM_MODE_VECTOR_INT): Likewise.
9582 * genoutput.c (null_operand): Likewise.
9583 (output_operand_data): Likewise.
9584 * genrecog.c (print_parameter_value): Likewise.
9585 * lra.c (debug_operand_data): Likewise.
9587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9589 * dwarf2out.c (scompare_loc_descriptor_wide)
9590 (scompare_loc_descriptor_narrow): New functions, split out from...
9591 (scompare_loc_descriptor): ...here.
9592 * expmed.c (emit_store_flag_int): New function, split out from...
9593 (emit_store_flag): ...here.
9595 2017-08-30 Richard Biener <rguenther@suse.de>
9597 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
9598 (dwarf2out_early_finish): Move setting of AT_pubnames from
9599 early debug output to early finish.
9601 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9603 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9604 and -mdata-region to the assembler.
9606 2017-08-30 Richard Biener <rguenther@suse.de>
9608 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9610 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9611 (dwarf2out_early_global_decl): Only generate a DIE for the
9612 abstract origin if it doesn't already exist or is a declaration DIE.
9613 (resolve_addr): Do not add the linkage name twice when
9614 generating a stub DIE for the DW_TAG_GNU_call_site target.
9616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9618 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9619 Use machine_mode rather than int for arg1_mode.
9621 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9624 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9625 that the second argument of the built-in functions to unpack
9626 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9627 switch statement instead a lot of if statements.
9628 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9629 Allow 64-bit values to be in Altivec registers as well as
9630 traditional floating point registers.
9631 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9633 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9635 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9636 MAX_REGS_PER_ADDRESS == 1.
9638 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9640 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9641 * config/i386/i386.c (ix86_option_override_internal): Simplify
9642 setting of opts->x_flag_entry.
9644 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9645 Jakub Jelinek <jakub@redhat.com>
9646 Richard Biener <rguenther@suse.de>
9648 PR tree-optimization/81503
9649 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9650 folded constant fits in the target type; reorder tests for clarity.
9652 2017-08-29 Martin Liska <mliska@suse.cz>
9654 * passes.def: Include pass_lower_switch.
9655 * stmt.c (dump_case_nodes): Remove and move to
9656 tree-switch-conversion.
9657 (case_values_threshold): Likewise.
9658 (expand_switch_as_decision_tree_p): Likewise.
9659 (emit_case_decision_tree): Likewise.
9660 (expand_case): Likewise.
9661 (balance_case_nodes): Likewise.
9662 (node_has_low_bound): Likewise.
9663 (node_has_high_bound): Likewise.
9664 (node_is_bounded): Likewise.
9665 (emit_case_nodes): Likewise.
9666 (struct simple_case_node): New struct.
9667 (add_case_node): Remove.
9668 (emit_case_dispatch_table): Use vector instead of case_list.
9669 (reset_out_edges_aux): Remove.
9670 (compute_cases_per_edge): Likewise.
9671 (expand_case): Build list of simple_case_node.
9672 (expand_sjlj_dispatch_table): Use it.
9673 * tree-switch-conversion.c (struct case_node): Moved from
9674 stmt.c and adjusted.
9675 (emit_case_nodes): Likewise.
9676 (node_has_low_bound): Likewise.
9677 (node_has_high_bound): Likewise.
9678 (node_is_bounded): Likewise.
9679 (case_values_threshold): Likewise.
9680 (reset_out_edges_aux): Likewise.
9681 (compute_cases_per_edge): Likewise.
9682 (add_case_node): Likewise.
9683 (dump_case_nodes): Likewise.
9684 (balance_case_nodes): Likewise.
9685 (expand_switch_as_decision_tree_p): Likewise.
9686 (emit_jump): Likewise.
9687 (emit_case_decision_tree): Likewise.
9688 (try_switch_expansion): Likewise.
9689 (do_jump_if_equal): Likewise.
9690 (emit_cmp_and_jump_insns): Likewise.
9691 (fix_phi_operands_for_edge): New function.
9692 (record_phi_operand_mapping): Likewise.
9693 (class pass_lower_switch): New pass.
9694 (pass_lower_switch::execute): New function.
9695 (make_pass_lower_switch): Likewise.
9696 (conditional_probability):
9697 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9698 * tree-pass.h: Add make_pass_lower_switch.
9700 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9703 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9706 2017-08-29 Richard Biener <rguenther@suse.de>
9708 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9709 we do not add a DW_AT_inline attribute twice.
9710 (gen_subprogram_die): Remove code setting DW_AT_inline on
9711 DECL_ABSTRACT_P nodes.
9713 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9715 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9716 calls to internal functions.
9717 (gimplify_modify_expr): Likewise.
9718 * tree-call-cdce.c (use_internal_fn): Likewise.
9719 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9720 (convert_to_divmod): Set the nothrow flag.
9721 * tree-if-conv.c (predicate_mem_writes): Likewise.
9722 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9723 (vectorizable_call): Likewise.
9724 (vectorizable_store): Likewise.
9725 (vectorizable_load): Likewise.
9726 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9727 (vect_recog_mask_conversion_pattern): Likewise.
9729 2017-08-29 Martin Liska <mliska@suse.cz>
9732 * gcc.c (driver_handle_option): Add new argument.
9733 * opts-common.c (handle_option): Pass
9734 target_option_override_hook.
9735 * opts-global.c (lang_handle_option): Add new option.
9736 (set_default_handlers): Add new argument.
9737 (decode_options): Likewise.
9738 * opts.c (target_handle_option): Likewise.
9739 (common_handle_option): Call target_option_override_hook.
9740 * opts.h (struct cl_option_handler_func): Add hook for
9741 target option override.
9742 (struct cl_option_handlers): Likewise.
9743 (set_default_handlers): Add new argument.
9744 (decode_options): Likewise.
9745 (common_handle_option): Likewise.
9746 (target_handle_option): Likewise.
9747 * toplev.c (toplev::main): Pass targetm.target_option.override
9750 2017-08-29 Richard Biener <rguenther@suse.de>
9751 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9753 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9754 life to the active subtree.
9756 2017-08-28 Jeff Law <law@redhat.com>
9758 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9759 derive_equivalences.
9760 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9762 (edge_info::derive_equivalences): New private member function
9764 * tree-ssa-dom.c (class edge_info): Changed from a struct
9765 to a class. Add ctor/dtor, methods and data members.
9766 (edge_info::edge_info): Renamed from allocate_edge_info.
9767 Initialize additional members.
9768 (edge_info::~edge_info): New.
9769 (free_dom_edge_info): Delete the edge info.
9770 (record_edge_info): Use new class & associated member functions.
9771 Tighten forms for testing for edge equivalences.
9772 (record_temporary_equivalences): Iterate over the simple
9773 equivalences rather than assuming there's only one per edge.
9774 (cprop_into_successor_phis): Iterate over the simple
9775 equivalences rather than assuming there's only one per edge.
9776 (optimize_stmt): Use operand_equal_p rather than pointer
9777 equality for mini-DSE code.
9779 2017-08-28 Nathan Sidwell <nathan@acm.org>
9781 * gcc.c (execute): Fold SIGPIPE handling into switch
9782 statement. Adjust internal error message.
9784 2017-08-28 Richard Biener <rguenther@suse.de>
9787 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9788 Do nothing for removed DIEs.
9790 2017-08-28 Richard Biener <rguenther@suse.de>
9792 PR tree-optimization/81977
9793 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9796 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9799 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9800 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9801 using targetm.gen_mem_thread_fence.
9803 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9806 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9807 predicate to register_operand. Reorder operands.
9808 (*btr<mode>): Ditto.
9809 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9810 (*btr<mode>_mask): Ditto.
9812 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9814 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9815 * config/rs6000/xmmintrin.h: New file.
9816 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9818 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9821 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9822 parameter and_insn and return it.
9823 (recombine_lvx_pattern): Insert a copy to ensure availability of
9824 the base register of the copied masking operation at the point of
9825 the instruction replacement.
9826 (recombine_stvx_pattern): Likewise.
9828 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9830 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9831 undocumented switches.
9832 (-mpower9-dform-vector): Likewise.
9833 (-mpower9-dform): Likewise.
9834 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9835 comments to delete references to -mpower9-dform* switches.
9836 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9837 Delete reference to -mpower9-dform* switches, test for
9838 -mpower9-vector instead.
9839 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9840 (OTHER_P9_VECTOR_MASKS): Likewise.
9841 (POWERPC_MASKS): Likewise.
9842 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9843 tests against -mpower9-dform* to -mpower9-vector. Delete code
9844 that checked for -mpower9-dform* consistancy with other options.
9845 Add test for -mpower9-misc to enable other power9 switches.
9846 (rs6000_init_hard_regno_mode_ok): Likewise.
9847 (rs6000_option_override_internal): Likewise.
9848 (rs6000_emit_prologue): Likewise.
9849 (rs6000_emit_epilogue): Likewise.
9850 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9851 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9852 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9854 (emit_fusion_p9_store): Likewise.
9855 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9856 resetting these macros if the assembler does not support ISA 3.0
9858 (TARGET_P9_DFORM_VECTOR): Likewise.
9859 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9860 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9862 2017-08-25 Alan Modra <amodra@gmail.com>
9864 PR rtl-optimization/81747
9865 * cse.c (cse_extended_basic_block): Don't attempt to record
9866 equivalences for degenerate conditional branches that branch
9867 to their fall-through.
9869 2017-08-24 Martin Sebor <msebor@redhat.com>
9872 * gimple-fold.c (size_must_be_zero_p): New function.
9873 (gimple_fold_builtin_memory_op): Call it.
9875 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9877 * config/rs6000/mm_malloc.h: New file.
9879 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9881 PR tree-optimization/81913
9882 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9883 analysis when either IVs in condition can wrap.
9885 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9887 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9888 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9890 2017-08-24 Richard Biener <rguenther@suse.de>
9893 * targhooks.c (default_target_can_inline_p): Properly
9894 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9895 is present and always compare.
9896 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9897 infer -mfpmath=sse from TARGET_SSE_P.
9898 (ix86_can_inline_p): Properly use target_option_default_node when
9899 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9901 2017-08-24 Richard Biener <rguenther@suse.de>
9904 * dwarf2out.c (output_die): Handle flag_generate_offload like
9906 (output_comp_unit): Likewise.
9907 (gen_array_type_die): Likewise.
9908 (dwarf2out_early_finish): Likewise.
9909 (note_variable_value_in_expr): Likewise.
9910 (dwarf2out_finish): Likewise. Adjust assert.
9911 * cgraphunit.c (symbol_table::compile): Move setting of
9912 flag_generate_offload earlier ...
9913 (symbol_table::finalize_compilation_unit): ... here, before
9914 early debug finalization.
9916 2017-08-24 Richard Biener <rguenther@suse.de>
9918 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9919 and ipa-fnsummary.h.
9920 (ix86_can_inline_p): When ix86_fpmath flags do not match
9921 check whether the callee uses FP math at all.
9923 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9926 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9927 instead of TYPE_PRECISION.
9929 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9930 Alan Hayward <alan.hayward@arm.com>
9931 David Sherwood <david.sherwood@arm.com>
9933 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9935 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9937 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9939 (ix86_frame::stack_realign_allocate): New field.
9940 (struct machine_frame_state): Modify comments.
9941 (machine_frame_state::sp_realigned_fp_end): New field.
9942 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9944 (sp_valid_at): Add assertion to assure no attempt to access invalid
9945 offset of a realigned stack.
9946 (fp_valid_at): Likewise.
9947 (choose_baseaddr): Modify comments.
9948 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9949 ix86_expand_prologue.
9950 (ix86_expand_prologue): Modify stack realignment and allocation.
9951 (ix86_expand_epilogue): Modify comments.
9952 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9953 avx2_runtime, avx512f, and avx512f_runtime.
9955 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9957 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9958 (mstackrealign): Do not init to -1.
9959 * config/i386/i386.c (ix86_option_override_internal):
9960 Check opts_set, not opts when setting default value of
9961 opts->x_ix86_force_align_arg_pointer.
9963 2017-08-23 Richard Biener <rguenther@suse.de>
9965 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9966 lang_hooks.decl_printable_name.
9967 * print-rtl-function.c (print_rtx_function): Likewise.
9968 * tree-pretty-print.c (dump_function_header): Likewise.
9970 2017-08-23 Richard Biener <rguenther@suse.de>
9973 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9974 -g0 at compile-time.
9976 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9979 * doc/sourcebuild.texi (Other hardware attributes):
9982 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9985 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9986 Fix single-use check.
9988 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9990 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9992 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9994 * doc/install.texi: Modify to add more details on running selected
9997 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9999 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10000 is combined with -mabi=ms.
10001 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10004 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10006 PR tree-optimization/81488
10007 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10008 and cached_basis fields.
10009 (MAX_SPREAD): New constant.
10010 (alloc_cand_and_find_basis): Initialize new fields.
10011 (clear_visited): New function.
10012 (create_phi_basis_1): Rename from create_phi_basis, set visited
10013 and cached_basis fields.
10014 (create_phi_basis): New wrapper function.
10015 (phi_add_costs_1): Rename from phi_add_costs, add spread
10016 parameter, set visited field, short-circuit when limits reached.
10017 (phi_add_costs): New wrapper function.
10018 (record_phi_increments_1): Rename from record_phi_increments, set
10020 (record_phi_increments): New wrapper function.
10021 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10022 (phi_incr_cost): New wrapper function.
10023 (all_phi_incrs_profitable_1): Rename from
10024 all_phi_incrs_profitable, set visited field.
10025 (all_phi_incrs_profitable): New wrapper function.
10027 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10028 Alan Hayward <alan.hayward@arm.com>
10029 David Sherwood <david.sherwood@arm.com>
10031 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10032 that takes the outer and inner modes.
10033 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10034 comparison as the canonical test for a paradoxical subreg.
10035 * combine.c (simplify_set): Use paradoxical_subreg_p.
10036 (make_extraction): Likewise.
10037 (force_to_mode): Likewise.
10038 (rtx_equal_for_field_assignment_p): Likewise.
10039 (gen_lowpart_for_combine): Likewise.
10040 (simplify_comparison): Likewise.
10041 * cse.c (equiv_constant): Likewise.
10042 * expmed.c (store_bit_field_1): Likewise.
10043 * final.c (alter_subreg): Likewise.
10044 * fwprop.c (propagate_rtx): Likewise.
10045 (forward_propagate_subreg): Likewise.
10046 * ira-conflicts.c (ira_build_conflicts): Likewise.
10047 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10048 * lra-constraints.c (curr_insn_transform): Likewise.
10049 (split_reg): Likewise.
10050 * lra-eliminations.c (move_plus_up): Likewise.
10051 (lra_eliminate_regs_1): Likewise.
10052 * recog.c (general_operand): Likewise.
10053 * ree.c (combine_reaching_defs): Likewise.
10054 * reload.c (push_reload): Likewise.
10055 (find_reloads): Likewise.
10056 * reload1.c (elimination_effects): Likewise.
10057 (compute_reload_subreg_offset): Likewise.
10058 (choose_reload_regs): Likewise.
10059 * rtlanal.c (subreg_lsb_1): Likewise.
10060 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10061 (simplify_subreg): Likewise.
10062 * var-tracking.c (track_loc_p): Likewise.
10063 * emit-rtl.c (byte_lowpart_offset): Likewise.
10064 (paradoxical_subreg_p): Delete out-of-line definition.
10066 2017-08-22 Jeff Law <law@redhat.com>
10068 PR tree-optimization/81741
10069 PR tree-optimization/71947
10070 * tree-ssa-dom.c: Include tree-inline.h.
10071 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10072 equivalences if one is more expensive to compute than the other.
10073 * tree-ssa-scopedtables.h (class const_or_copies): Make
10074 record_const_or_copy_raw method private.
10075 (class avail_exprs_stack): New method simplify_binary_operation.
10076 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10077 avail_exprs_stack::simplify_binary_operation as needed.
10078 (avail_exprs_stack::simplify_binary_operation): New function.
10080 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10082 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10083 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10084 (DOT_SYMBOLS): Likewise.
10085 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10086 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10087 (RS6000_ABI_NAME): Likewise.
10088 (TARGET_CMODEL): Likewise.
10089 (TOC_SECTION_ASM_OP): Likewise.
10090 (SET_CMODEL): New macro.
10091 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10093 2017-08-22 Richard Biener <rguenther@suse.de>
10095 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10096 to allow for free-lang-data replacements similar to verify_type_variant.
10098 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10099 Alan Hayward <alan.hayward@arm.com>
10100 David Sherwood <david.sherwood@arm.com>
10102 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10103 VOIDmode for the LABEL_REF.
10105 2017-08-22 Richard Biener <rguenther@suse.de>
10107 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10109 2017-08-22 Richard Biener <rguenther@suse.de>
10111 * common.opt (feliminate-dwarf2-dups): Ignore.
10112 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10113 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10114 same_die_p_wrap, compute_section_prefix,
10115 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10116 (comdat_symbol_id, comdat_symbol_number): Likewise.
10117 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10119 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10120 (output_die): Mark unreachable path unreachable.
10121 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10122 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10123 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10124 (dwarf2out_early_finish): Likewise.
10126 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10128 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10130 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10133 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10134 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10135 (avr_attribute_table) <io, io_low, address>: Initialize
10136 .decl_required with true.
10138 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10140 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10141 undocumented debugging options.
10142 (-mvsx-scalar-double): Likewise.
10143 (-mallow-df-permute): Likewise.
10144 (-mvectorize-builtins): Likewise.
10145 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10146 (rs6000_builtin_vectorized_function): Likewise.
10147 (rs6000_builtin_md_vectorized_function): Likewise.
10148 (rs6000_opt_vars): Likewise.
10150 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10153 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10154 (*btrq_imm): Rename from *btrq.
10155 (*btcq_imm): Rename from *btcq.
10156 (btsc): New code attribute.
10157 (*<btsc><mode>): New insn pattern.
10158 (*btr<mode>): Ditto.
10159 (*<btsc><mode>_mask): New insn_and_split pattern.
10160 (*btr<mode>_mask): Ditto.
10162 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10163 Alan Hayward <alan.hayward@arm.com>
10164 David Sherwood <david.sherwood@arm.com>
10166 * function.c (pad_below): Simplify padding calculation.
10168 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10169 Alan Hayward <alan.hayward@arm.com>
10170 David Sherwood <david.sherwood@arm.com>
10172 * target.def (function_prologue): Remove frame size argument.
10173 (function_epilogue): Likewise.
10174 * doc/tm.texi: Regenerate.
10175 * final.c (final_start_function): Update call to function_prologue.
10176 (final_end_function): Update call to function_epilogue.
10177 (default_function_pro_epilogue): Remove frame size argument.
10178 * output.h (default_function_pro_epilogue): Likewise.
10179 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10180 (arm_output_function_prologue): Likewise.
10181 * config/frv/frv.c (frv_function_prologue): Likewise.
10182 (frv_function_epilogue): Likewise.
10183 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10184 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10185 (ia64_output_function_epilogue): Likewise.
10186 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10187 (m32r_output_function_epilogue): Likewise.
10188 * config/microblaze/microblaze.c (microblaze_function_prologue)
10189 (microblaze_function_epilogue): Likewise.
10190 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10191 (mips_output_function_epilogue): Likewise.
10192 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10193 (mmix_target_asm_function_epilogue): Likewise.
10194 * config/msp430/msp430.c (msp430_start_function): Likewise.
10195 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10196 (nds32_asm_function_epilogue): Likewise.
10197 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10198 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10199 (pa_output_function_epilogue): Likewise.
10200 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10201 (rs6000_output_function_epilogue): Likewise.
10202 * config/rl78/rl78.c (rl78_start_function): Likewise.
10203 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10204 (rs6000_output_function_epilogue): Likewise.
10205 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10206 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10207 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10208 (sparc_asm_function_epilogue): Likewise.
10210 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10212 * tree.h (type_has_mode_precision_p): New function.
10213 * convert.c (convert_to_integer_1): Use it.
10214 * expr.c (expand_expr_real_2): Likewise.
10215 (expand_expr_real_1): Likewise.
10216 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10217 * match.pd: Likewise.
10218 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10219 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10220 * tree-tailcall.c (process_assignment): Likewise.
10221 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10222 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10223 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10224 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10225 (vectorizable_assignment): Likewise.
10226 (vectorizable_shift): Likewise.
10227 (vectorizable_operation): Likewise.
10228 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10230 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10232 * match.pd: Add pow (C, x) simplification.
10234 2017-08-21 Richard Biener <rguenther@suse.de>
10236 PR tree-optimization/81900
10237 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10238 for blocks with abnormal predecessors.
10239 (compute_antic): Do not set visited flag prematurely.
10241 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10244 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10246 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10248 * stor-layout.h (vector_type_mode): Move to...
10249 * tree.h (vector_type_mode): ...here.
10250 * stor-layout.c (vector_type_mode): Move to...
10251 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10253 2017-08-21 Richard Biener <rguenther@suse.de>
10255 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10256 register_external_die hooks.
10257 (debug_false_tree_charstarstar_uhwistar): Declare.
10258 (debug_nothing_tree_charstar_uhwi): Likewise.
10259 * debug.c (do_nothing_debug_hooks): Adjust.
10260 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10261 (debug_nothing_tree_charstar_uhwi): Likewise.
10262 * dbxout.c (dbx_debug_hooks): Adjust.
10263 (xcoff_debug_hooks): Likewise.
10264 * sdbout.c (sdb_debug_hooks): Likewise.
10265 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10266 * dwarf2out.c (macinfo_label_base): New global.
10267 (dwarf2out_register_external_die): New function for the
10268 register_external_die hook.
10269 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10270 (dwarf2_debug_hooks): Use them.
10271 (dwarf2_lineno_debug_hooks): Adjust.
10272 (struct die_struct): Add with_offset flag.
10273 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10274 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10275 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10276 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10277 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10278 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10279 defining section names for the early LTO debug variants.
10280 (reset_indirect_string): New helper.
10281 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10282 (print_dw_val): Add support for offsetted symbol references.
10283 (get_ultimate_context): Split out from is_cxx.
10284 (is_cxx): Use get_ultimate_context.
10285 (is_fortran): Add decl overload.
10286 (compute_comp_unit_symbol): Split out worker from
10287 compute_section_prefix.
10288 (compute_section_prefix): Call compute_comp_unit_symbol and
10289 set comdat_type_p here.
10290 (output_die): Skip DIE symbol output for the LTO added one.
10291 Handle DIE symbol references with offset.
10292 (output_comp_unit): Guard section name mangling properly.
10293 For LTO debug sections emit a symbol at the section beginning
10294 which we use to refer to its DIEs.
10295 (add_abstract_origin_attribute): For DIEs registered via
10296 dwarf2out_register_external_die directly refer to the early
10297 DIE rather than indirectly through the shadow one we created.
10298 Remove obsolete call to dwarf2out_abstract_function for
10299 non-function/block origins.
10300 (gen_array_type_die): When generating early LTO debug do
10301 not emit DW_AT_string_length.
10302 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10303 late when in LTO. As suggested place a gcc_unreachable for
10304 the DECL_ABSTRACT_P case.
10305 (gen_subprogram_die): Avoid another specification DIE
10306 for early built declarations/definitions for the late LTO case.
10307 (gen_variable_die): Add type references for late duplicated VLA dies
10309 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10310 we have the abstract instance already.
10311 (process_scope_var): Adjust decl DIE contexts in LTO which
10312 first puts them in limbo.
10313 (gen_decl_die): Do not generate type DIEs late apart from
10314 types for VLAs or for decls we do not yet have a DIE. Do not
10315 call dwarf2out_abstract_function late.
10316 (dwarf2out_early_global_decl): Make sure to create DIEs
10317 for abstract instances of a decl first.
10318 (dwarf2out_late_global_decl): Adjust comment.
10319 (output_macinfo_op): With multiple macro sections use
10320 macinfo_label_base to distinguish labels.
10321 (output_macinfo): Likewise. Update macinfo_label_base.
10322 Pass in the line info label.
10323 (note_variable_value_in_expr): When generating LTO resolve
10324 all variable values here by generating DIEs as needed.
10325 (init_sections_and_labels): Add early LTO debug flag parameter
10326 and generate different sections and names if set. Add generation
10327 counter for the labels so we can have multiple of them.
10328 (reset_dies): Helper to allow DIEs to be output multiple times.
10329 (dwarf2out_finish): When outputting DIEs to the fat part of an
10330 LTO object first reset DIEs.
10331 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10332 (modified_type_die): Check for decl_ultimate_origin being self
10334 (gen_type_die_with_usage): Likewise.
10335 (gen_typedef_die): Allow decl_ultimate_origin being self.
10336 (set_decl_abstract_flags): Remove.
10337 (set_block_abstract_flags): Likewise.
10338 (dwarf2out_abstract_function): Treat the early generated DIEs
10339 as the abstract copy and only add DW_AT_inline and
10340 DW_AT_artificial here and call set_decl_origin_self.
10341 If the DIE has an abstract origin don't do anything.
10342 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10343 if we have none yet (Go fails to build one, PR78628).
10344 (variably_modified_type_p): Prevent endless recursion for Ada
10345 cyclic pointer types.
10346 * lto-streamer-in.c: Include debug.h.
10347 (dref_queue): New global.
10348 (lto_read_tree_1): Stream in DIE references.
10349 (lto_input_tree): Register DIE references.
10350 (input_function): Stream DECL_DEBUG_ARGS.
10351 * lto-streamer-out.c: Include debug.h.
10352 (lto_write_tree_1): Output DIE references.
10353 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10354 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10355 (output_function): Stream DECL_DEBUG_ARGS.
10356 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10357 Stream DECL_ABSTRACT_ORIGIN.
10358 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10359 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10360 DECL_CONTEXT for file-scope decls.
10361 * lto-streamer.h (struct dref_entry): Declare.
10362 (dref_queue): Likewise.
10363 * cfgexpand.c (pass_expand::execute): Do not call the
10364 outlining_inline_function hook here.
10365 * lto-wrapper.c (debug_obj): New global.
10366 (tool_cleanup): Unlink it if required.
10367 (debug_objcopy): New function.
10368 (run_gcc): Handle early debug sections in the IL files by
10369 extracting them to separate files, partially linkin them and
10370 feeding the result back as result to the linker.
10371 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10372 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10373 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10374 sections into a separate segment.
10375 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10377 (darwin_asm_dwarf_section): Likewise.
10378 (darwin_asm_output_dwarf_offset): Likewise.
10379 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10381 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10382 Alan Hayward <alan.hayward@arm.com>
10383 David Sherwood <david.sherwood@arm.com>
10385 * read-md.h (md_reader::record_potential_iterator_use): Replace
10386 pointer argument with an rtx and an index.
10387 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10388 (apply_mode_iterator): Likewise.
10389 (apply_code_iterator): Likewise.
10390 (apply_int_iterator): Likewise.
10391 (apply_subst_iterator): Likewise.
10392 (record_iterator_use): Likewise.
10393 (record_attribute_use): Likewise.
10394 (md_reader::record_potential_iterator_use): Likewise. Update calls
10395 to record_iterator_use and apply_iterator.
10396 (iterator_use): Replace ptr with x and index.
10397 (attribute_use): Likewise.
10398 (apply_attribute_uses): Update calls to apply_iterator.
10399 (apply_iterators): Likewise. Update initialization of iterator_use.
10400 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10401 and record_potential_iterator_use.
10402 (rtx_reader::read_rtx_operand): Likewise.
10404 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10405 Alan Hayward <alan.hayward@arm.com>
10406 David Sherwood <david.sherwood@arm.com>
10408 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10411 2017-08-21 Richard Biener <rguenther@suse.de>
10413 PR middle-end/81884
10414 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10415 at struct end conservatively when comparing common bases.
10417 2017-08-21 Richard Biener <rguenther@suse.de>
10419 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10420 (mem_ref_in_stmt): Remove.
10421 (determine_max_movement): Use ref index to get at the reference.
10422 (invariantness_dom_walker::before_dom_children): Deal with
10423 lim data already initialized.
10424 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10426 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10428 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10429 (TARGET_ISA_ROUND): Ditto.
10430 (TARGET_ROUND): Ditto.
10431 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10432 * config/i386/i386.md: Ditto.
10433 * config/i386/sse.md: Ditto.
10434 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10435 with OPTION_MASK_ISA_SSE4_1.
10437 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10440 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10441 __builtin_ia32_lzcnt_u16.
10443 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10446 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10447 (rs6000_set_current_function): Rewrite function to use it.
10449 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10452 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10453 and TYPE_WARN_IF_NOT_ALIGN.
10454 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10455 (handle_warn_if_not_align): New.
10456 (place_union_field): Call handle_warn_if_not_align.
10457 (place_field): Call handle_warn_if_not_align.
10458 Copy TYPE_WARN_IF_NOT_ALIGN.
10459 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10460 (layout_type): Likewise.
10461 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10463 (tree_decl_common): Add warn_if_not_align.
10464 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10465 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10466 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10467 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10468 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10469 * doc/extend.texi: Document warn_if_not_aligned attribute.
10470 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10472 2017-08-17 Martin Liska <mliska@suse.cz>
10475 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10476 (get_data_dependence): Use it as pointer type.
10477 (distribute_loop): Likewise.
10479 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10481 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10482 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10483 (p8_vmrgow_<mode>_direct): New define_insn.
10484 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10485 handle endianness for vmrgew and vmrgow permute patterns.
10487 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10489 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10490 * config/rs6000/rs6000-cpus.def: Remove comment.
10491 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10492 (POWERPC_MASKS): Likewise.
10493 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10494 use of TARGET_VSX_TIMODE.
10495 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10496 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10497 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10498 (rs6000_option_override_internal): Remove dead code.
10499 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10500 (rs6000_legitimize_reload_address): Likewise.
10501 (rs6000_legitimate_address_p): Likewise.
10502 (rs6000_opt_masks): Delete "vsx-timode".
10503 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10504 from function comment.
10505 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10506 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10507 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10509 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10510 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10511 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10512 (VSX_TI): Likewise.
10514 (define_peephole2): Likewise.
10516 2017-08-17 Martin Sebor <msebor@redhat.com>
10519 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
10520 past the end of an array.
10521 (test_pp_format): Add test cases.
10523 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
10525 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
10526 missing ECF_NOTHROW flags.
10528 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10531 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
10532 operands residing in integer registers.
10533 (*vsx_le_perm_load_<mode>): Likewise.
10534 (*vsx_le_perm_store_<mode>): Likewise.
10535 (define_peephole2): Add peepholes to optimize the above.
10537 2017-08-17 Marek Polacek <polacek@redhat.com>
10539 PR middle-end/81814
10540 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
10541 to mimic what shorten_compare did. Change the return type to bool.
10542 (fold_cond_expr_with_comparison): Update call to
10543 operand_equal_for_comparison_p.
10544 (fold_ternary_loc): Likewise.
10546 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
10548 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
10550 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
10552 2017-08-17 Richard Biener <rguenther@suse.de>
10554 * tree-ssa-structalias.c (solve_graph): When propagating
10555 to successors update the graphs succ edges and avoid duplicate work.
10557 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
10560 * config/i386/i386.c (ix86_option_override_internal): Save target
10561 specific options after ix86_stack_protector_guard_reg was changed.
10563 2017-08-17 Richard Biener <rguenther@suse.de>
10565 PR tree-optimization/81827
10566 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
10567 (new_var_info): Initialize it conservatively.
10568 (get_call_vi): Mark register vars.
10569 (new_scalar_tmp_constraint_exp): Likewise.
10570 (handle_rhs_call): Likewise.
10571 (handle_const_call): Likewise.
10572 (create_function_info_for): Likewise.
10573 (solve_constraints): Sort varinfos to separate register from
10574 non-register vars to pack points-to solution bitmaps during
10577 2017-08-17 Marek Polacek <polacek@redhat.com>
10579 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
10581 2017-08-17 Richard Biener <rguenther@suse.de>
10583 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
10584 to true when overflow is undefined and we saturated the result.
10586 2017-08-17 Alan Modra <amodra@gmail.com>
10589 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
10590 Don't use store multiple if only one reg needs saving.
10591 (interesting_frame_related_regno): New function.
10592 (rs6000_frame_related): Don't emit frame info for regs that
10594 (rs6000_emit_epilogue): Likewise.
10596 2017-08-16 Nathan Sidwell <nathan@acm.org>
10598 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
10599 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
10600 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
10601 * tree.c (free_lang_data_in_type): Use else-if chain. Always
10602 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10603 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10604 (verify_type): Adjust for TYPE_BINFO move.
10605 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10606 process TYPE_BINFO directly.
10607 (hash_tree): Likewise.
10608 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10610 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10613 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10615 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10617 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10620 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10621 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10622 Add ix86_binary_operator_ok to insn constraint.
10623 (*iordi_1_bts): Ditto.
10624 (*xordi_1_btc): Ditto.
10625 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10626 Update corresponding peephole2 pattern.
10630 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10632 PR tree-optimization/81832
10633 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10634 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10636 2017-08-16 Marek Polacek <polacek@redhat.com>
10639 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10640 regarding size_zero_node.
10642 2017-08-16 Martin Liska <mliska@suse.cz>
10645 * config.gcc: Respect previously set extra_objs in case
10648 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10650 PR tree-optimization/81835
10651 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10652 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10653 not depend on the phi.
10655 2017-08-16 Alan Modra <amodra@gmail.com>
10657 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10660 2017-08-16 Alan Modra <amodra@gmail.com>
10662 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10663 (save_reg_p): ..into this. Update all callers.
10664 (first_reg_to_save): Simplify.
10666 2017-08-16 Alan Modra <amodra@gmail.com>
10668 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10671 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10675 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10676 constant count if that count is less than 32.
10678 2017-08-15 Nathan Sidwell <nathan@acm.org>
10680 * gcc.c (execute): Emit friendlier message if inferior is killed
10681 by an external cause.
10683 2017-08-15 Richard Biener <rguenther@suse.de>
10685 PR tree-optimization/81790
10686 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10687 CONSTRUCTORs from simplifying and VN.
10689 2017-08-14 Martin Sebor <msebor@redhat.com>
10691 * builtin-attrs.def: Add comments.
10693 2017-08-14 Martin Sebor <msebor@redhat.com>
10696 * doc/extend.texi (attribute nonstring): Document new attribute.
10698 2017-08-14 Martin Sebor <msebor@redhat.com>
10701 * tree-diagnostic.c (default_tree_printer): Handle %G.
10702 * gimple-pretty-print.h (percent_G_format): Declare new function.
10703 * gimple-pretty-print.c (percent_G_format): Define.
10704 * tree-pretty-print.c (percent_K_format): Add argument.
10706 2017-08-14 Martin Sebor <msebor@redhat.com>
10708 PR translation/79998
10709 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10710 Remove a stray space.
10712 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10715 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10716 (*iordi_1_bts): Ditto.
10717 (*xordi_1_btc): Ditto.
10719 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10722 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10723 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10725 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10726 quoted strings, and make more translator-friendly.
10727 (darwin_rs6000_override_options): Likewise.
10728 (rs6000_option_override_internal): Likewise.
10729 (rs6000_return_in_memory): Fix overlong line.
10730 (init_cmulative_args): Use quoted strings, and make more
10731 translator-friendly.
10732 (rs6000_pass_by_reference): Fix overlong line.
10733 (def_builtin): Use quoted strings.
10734 (altivec_expand_predicate_builtin): Use quoted strings, and make
10735 more translator-friendly.
10736 (htm_expand_builtin): Use quoted strings.
10737 (cpu_expand_builtin): Use quoted strings, and make more
10738 translator-friendly.
10739 (altivec_expand_builtin): Likewise.
10740 (paired_expand_predicate_builtin): Likewise.
10741 (rs6000_invalid_builtin): Likewise.
10742 (builtin_function_type): Use quoted strings.
10743 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10744 more translator-friendly.
10745 (rs6000_trampoline_init): Likewise.
10746 (rs6000_handle_altivec_attribute): Likewise.
10747 (rs6000_inner_target_options): Use quoted strings.
10748 (rs6000_disable_incompatible_switches): Likewise.
10749 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10750 strings, and make more translator-friendly.
10751 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10753 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10755 PR tree-optimization/81799
10756 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10757 cond_expr to simple gimple operand.
10759 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10761 PR middle-end/46932
10762 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10764 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10768 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10769 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10770 TARGET_GASISR_PROLOGUES.
10771 * config/avr/avr.c (avr_option_override): Same.
10772 (avr_pass_pre_proep::execute): Same.
10774 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10777 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10778 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10780 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10782 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10783 define_insn_and_split. Split to a memory load from 0 in
10784 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10785 using PTR mode iterator.
10786 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10787 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10788 (*add_tp_<mode>): Redefine as define_insn_and_split.
10789 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10790 address space. Merge with *add_tp_x32 using PTR mode iterator.
10791 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10792 Split to an add with a memory load from 0 in
10793 DEFAULT_TLS_SEG_REG address space.
10795 2017-08-12 Andrew Pinski <apinski@cavium.com>
10797 * config/aarch64/aarch64-option-extensions.def (rdma):
10798 Fix feature string to what Linux prints out in /proc/cpuinfo.
10800 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10803 * dwarf2out.c (modified_type_die): For C typedef types that have
10804 an ultimate origin, process the ultimate origin instead of the
10806 (gen_typedef_die): Assert that input DECLs have no ultimate
10808 (gen_type_die_with_usage): For typedef variants that have an
10809 ultimate origin, just call gen_decl_die on the original DECL.
10810 (process_scope_var): Avoid creating DIEs for local typedefs and
10811 concrete static variables.
10813 2017-08-12 Alan Modra <amodra@gmail.com>
10817 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10818 match gnu-user.h startfile.
10819 (ENDFILE_LINUX_SPEC): Similarly.
10821 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10824 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10826 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10828 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10829 * config/aarch64/aarch64.md (mov<mode>): Change.
10830 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10831 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10832 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10834 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10836 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10837 for storage order barriers.
10839 2017-08-11 Martin Liska <mliska@suse.cz>
10842 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10843 variables of void type.
10845 2017-08-11 Martin Liska <mliska@suse.cz>
10847 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10848 TARGET_SUPPORTS_ALIASES.
10849 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10850 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10851 (optimize_weakref): Likewise.
10852 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10853 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10854 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10856 2017-08-11 Martin Liska <mliska@suse.cz>
10859 * config/i386/i386.c (make_resolver_func): Do complete
10860 refactoring of the function.
10862 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10865 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10866 * config/i386/i386.c (ix86_stack_protect_guard): Use
10867 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10868 * doc/invoke.texi (x86 Options): Document
10869 -mstack-protector-guard-symbol= option.
10871 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10873 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10874 * config/i386/i386.c (ix86_split_stack_guard): New function.
10875 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10876 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10877 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10878 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10879 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10880 (split_stack_space_check): Call ix86_split_stack_guard.
10882 2017-08-10 Martin Sebor <msebor@redhat.com>
10884 * print-tree.c (print_node): Print location using the established
10886 Replace spaces with colons.
10887 (debug_raw, debug): Ditto.
10889 2017-08-10 Martin Sebor <msebor@redhat.com>
10892 * pretty-print.c (pp_format): Correct the handling of %s precision.
10894 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10897 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10899 (ix86_finalize_stack_frame_flags): This. Also clear
10900 frame_pointer_needed if -fno-omit-frame-pointer is used without
10902 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10903 with ix86_finalize_stack_frame_flags.
10904 (ix86_expand_epilogue): Likewise.
10905 (ix86_expand_split_stack_prologue): Likewise.
10906 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10908 2017-08-10 Martin Liska <mliska@suse.cz>
10911 * c-attribs.c (handle_target_attribute):
10912 Report warning for an empty string argument of target attribute.
10914 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10917 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10919 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10920 or DECL_NONLOCAL labels.
10921 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10922 or DECL_NONLOCAL labels here.
10924 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10926 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10927 to indicate when early gimple folding has been disabled.
10928 (rs6000_gimple_fold_builtin): Add debug content.
10929 (rs6000_invalid_builtin): Fix whitespace.
10930 (rs6000_expand_builtin): Fix whitespace.
10931 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10933 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10936 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10937 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10939 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10941 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10942 (qdf24xx): Likewise.
10943 * config/aarch64/aarch64.md: Include falkor.md.
10944 * config/aarch64/falkor.md: New.
10946 2017-08-09 Marek Polacek <polacek@redhat.com>
10949 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10950 * diagnostic.c (emit_diagnostic): Add a comment.
10951 (emit_diagnostic_valist): New function.
10953 2017-08-09 Marek Polacek <polacek@redhat.com>
10956 * input.c (make_location): New overload.
10957 * input.h (make_location): Declare.
10959 2017-08-08 Alan Modra <amodra@gmail.com>
10960 H.J. Lu <hongjiu.lu@intel.com>
10963 * gcc.c (NO_PIE_SPEC): Delete.
10964 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10966 (LINK_PIE_SPEC): ..to here.
10967 (LINK_COMMAND_SPEC): Support -no-pie.
10968 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10969 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10971 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10972 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10973 (ENDFILE_CRTEND_SPEC): Similarly.
10975 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10978 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10979 (mstack-protector-guard-offset=): Ditto.
10980 * config/i386/i386.c (ix86_option_override): Handle
10981 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10983 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10984 ix86_stack_protect_guard_offset variables.
10985 (TARGET_STACK_PROTECT_GUARD): Always define.
10986 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10987 and -mstack-protector-guard-offset= options.
10989 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10991 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10992 boundary case for the last candidate.
10994 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10996 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10997 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10999 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11001 PR middle-end/19706
11002 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11003 * config/aarch64/aarch64-builtins.c
11004 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11005 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11006 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11008 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11009 Andrew Pinski <pinskia@gmail.com>
11011 PR middle-end/19706
11012 * internal-fn.def (XORSIGN): New.
11013 * optabs.def (xorsign_optab): New.
11014 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11015 (convert_expand_mult_copysign): New.
11016 (pass_optimize_widening_mul::execute): Call
11017 convert_expand_mult_copysign.
11019 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11021 PR tree-optimization/81354
11022 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11023 Insert on edges rather than explicitly creating landing pads.
11024 (analyze_candidates_and_replace): Commit edge inserts.
11026 2017-08-08 Richard Biener <rguenther@suse.de>
11028 PR middle-end/81719
11029 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11030 (expand_simple_operations): Also look through ADDR_EXPRs with
11031 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11033 2017-08-08 Richard Biener <rguenther@suse.de>
11035 PR tree-optimization/81723
11036 * tree-vect-slp.c (struct bst_traits): New hash traits.
11037 (bst_fail): New global.
11038 (vect_build_slp_tree_2): New worker, split out from ...
11039 (vect_build_slp_tree): ... this now wrapping it with using
11040 bst_fail set to cache SLP tree build fails. Properly handle
11042 (vect_analyze_slp_instance): Allocate and free bst_fail.
11044 2017-08-08 Martin Liska <mliska@suse.cz>
11047 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11048 LABEL_DECLs that can be from a different function.
11050 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11052 PR tree-optimization/81744
11053 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11054 loop's number of iterations.
11056 2017-08-08 Martin Liska <mliska@suse.cz>
11058 * asan.c: Include header files.
11059 * attribs.c (build_decl_attribute_variant): New function moved
11061 (build_type_attribute_qual_variant): Likewise.
11062 (cmp_attrib_identifiers): Likewise.
11063 (simple_cst_list_equal): Likewise.
11064 (omp_declare_simd_clauses_equal): Likewise.
11065 (attribute_value_equal): Likewise.
11066 (comp_type_attributes): Likewise.
11067 (build_type_attribute_variant): Likewise.
11068 (lookup_ident_attribute): Likewise.
11069 (remove_attribute): Likewise.
11070 (merge_attributes): Likewise.
11071 (merge_type_attributes): Likewise.
11072 (merge_decl_attributes): Likewise.
11073 (merge_dllimport_decl_attributes): Likewise.
11074 (handle_dll_attribute): Likewise.
11075 (attribute_list_equal): Likewise.
11076 (attribute_list_contained): Likewise.
11077 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11078 (lookup_attribute_by_prefix): Likewise.
11079 * bb-reorder.c: Include header files.
11080 * builtins.c: Likewise.
11081 * calls.c: Likewise.
11082 * cfgexpand.c: Likewise.
11083 * cgraph.c: Likewise.
11084 * cgraphunit.c: Likewise.
11085 * convert.c: Likewise.
11086 * dwarf2out.c: Likewise.
11087 * final.c: Likewise.
11088 * fold-const.c: Likewise.
11089 * function.c: Likewise.
11090 * gimple-expr.c: Likewise.
11091 * gimple-fold.c: Likewise.
11092 * gimple-pretty-print.c: Likewise.
11093 * gimple.c: Likewise.
11094 * gimplify.c: Likewise.
11095 * hsa-common.c: Likewise.
11096 * hsa-gen.c: Likewise.
11097 * internal-fn.c: Likewise.
11098 * ipa-chkp.c: Likewise.
11099 * ipa-cp.c: Likewise.
11100 * ipa-devirt.c: Likewise.
11101 * ipa-fnsummary.c: Likewise.
11102 * ipa-inline.c: Likewise.
11103 * ipa-visibility.c: Likewise.
11105 * lto-cgraph.c: Likewise.
11106 * omp-expand.c: Likewise.
11107 * omp-general.c: Likewise.
11108 * omp-low.c: Likewise.
11109 * omp-offload.c: Likewise.
11110 * omp-simd-clone.c: Likewise.
11111 * opts-global.c: Likewise.
11112 * passes.c: Likewise.
11113 * predict.c: Likewise.
11114 * sancov.c: Likewise.
11115 * sanopt.c: Likewise.
11116 * symtab.c: Likewise.
11117 * toplev.c: Likewise.
11118 * trans-mem.c: Likewise.
11119 * tree-chkp.c: Likewise.
11120 * tree-eh.c: Likewise.
11121 * tree-into-ssa.c: Likewise.
11122 * tree-object-size.c: Likewise.
11123 * tree-parloops.c: Likewise.
11124 * tree-profile.c: Likewise.
11125 * tree-ssa-ccp.c: Likewise.
11126 * tree-ssa-live.c: Likewise.
11127 * tree-ssa-loop.c: Likewise.
11128 * tree-ssa-sccvn.c: Likewise.
11129 * tree-ssa-structalias.c: Likewise.
11130 * tree-ssa.c: Likewise.
11131 * tree-streamer-in.c: Likewise.
11132 * tree-vectorizer.c: Likewise.
11133 * tree-vrp.c: Likewise.
11134 * tsan.c: Likewise.
11135 * ubsan.c: Likewise.
11136 * varasm.c: Likewise.
11137 * varpool.c: Likewise.
11138 * tree.c: Remove functions moved to attribs.[ch].
11139 * tree.h: Likewise.
11140 * config/aarch64/aarch64.c: Add attrs.h header file.
11141 * config/alpha/alpha.c: Likewise.
11142 * config/arc/arc.c: Likewise.
11143 * config/arm/arm.c: Likewise.
11144 * config/avr/avr.c: Likewise.
11145 * config/bfin/bfin.c: Likewise.
11146 * config/c6x/c6x.c: Likewise.
11147 * config/cr16/cr16.c: Likewise.
11148 * config/cris/cris.c: Likewise.
11149 * config/darwin.c: Likewise.
11150 * config/epiphany/epiphany.c: Likewise.
11151 * config/fr30/fr30.c: Likewise.
11152 * config/frv/frv.c: Likewise.
11153 * config/ft32/ft32.c: Likewise.
11154 * config/h8300/h8300.c: Likewise.
11155 * config/i386/winnt.c: Likewise.
11156 * config/ia64/ia64.c: Likewise.
11157 * config/iq2000/iq2000.c: Likewise.
11158 * config/lm32/lm32.c: Likewise.
11159 * config/m32c/m32c.c: Likewise.
11160 * config/m32r/m32r.c: Likewise.
11161 * config/m68k/m68k.c: Likewise.
11162 * config/mcore/mcore.c: Likewise.
11163 * config/microblaze/microblaze.c: Likewise.
11164 * config/mips/mips.c: Likewise.
11165 * config/mmix/mmix.c: Likewise.
11166 * config/mn10300/mn10300.c: Likewise.
11167 * config/moxie/moxie.c: Likewise.
11168 * config/msp430/msp430.c: Likewise.
11169 * config/nds32/nds32-isr.c: Likewise.
11170 * config/nds32/nds32.c: Likewise.
11171 * config/nios2/nios2.c: Likewise.
11172 * config/nvptx/nvptx.c: Likewise.
11173 * config/pa/pa.c: Likewise.
11174 * config/pdp11/pdp11.c: Likewise.
11175 * config/powerpcspe/powerpcspe.c: Likewise.
11176 * config/riscv/riscv.c: Likewise.
11177 * config/rl78/rl78.c: Likewise.
11178 * config/rx/rx.c: Likewise.
11179 * config/s390/s390.c: Likewise.
11180 * config/sh/sh.c: Likewise.
11181 * config/sol2.c: Likewise.
11182 * config/sparc/sparc.c: Likewise.
11183 * config/spu/spu.c: Likewise.
11184 * config/stormy16/stormy16.c: Likewise.
11185 * config/tilegx/tilegx.c: Likewise.
11186 * config/tilepro/tilepro.c: Likewise.
11187 * config/v850/v850.c: Likewise.
11188 * config/vax/vax.c: Likewise.
11189 * config/visium/visium.c: Likewise.
11190 * config/xtensa/xtensa.c: Likewise.
11192 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11195 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11196 constraints since the -mupper-regs-* switches have been
11198 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11199 into a vector from a double word element that was extracted from
11200 another vector, and eliminate extra XXPERMDI instructions.
11201 (vsx_concat_<mode>_2): Likewise.
11202 (vsx_concat_<mode>_3): Likewise.
11203 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11204 concat to allow optimizing inserts from previous extracts.
11206 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11208 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11209 memory reference to a SSP offset in TLS address space.
11210 (ix86_print_operand) <case '@'>: Remove.
11211 (ix86_print_operand_punct_valid_p): Remove '@' code.
11212 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11213 UNSPEC_SP_TLS_TEST.
11214 (stack_tls_protect_set_<mode>): Remove.
11215 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11216 (stack_tls_protect_test_<mode>): Remove.
11217 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11219 2017-08-07 Olivier Hainque <hainque@adacore.com>
11222 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11224 2017-08-07 Douglas Rupp <rupp@adacore.com>
11226 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11227 variable was referenced as multidir in command.
11229 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11232 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11235 2017-08-07 Martin Liska <mliska@suse.cz>
11237 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11238 * config/rl78/rl78.c: Add include of attribs.h.
11239 * config/sh/sh.c: Likewise.
11240 * config/v850/v850.c: Likewise.
11242 2017-08-07 Tom de Vries <tom@codesourcery.com>
11244 PR middle-end/78266
11245 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11247 2017-08-07 Martin Liska <mliska@suse.cz>
11249 * config/mips/mips.c: Include attribs.h.
11251 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11254 * doc/invoke.texi: Document change in behvaior for -Ofast for
11257 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11259 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11261 (aarch64_pop_regs): Likewise.
11262 (aarch64_gen_load_pair): Likewise.
11263 (aarch64_save_callee_saves): Likewise.
11264 (aarch64_restore_callee_saves): Likewise.
11266 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11268 * config/i386/i386.c: Revert the last change.
11270 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11273 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11275 (ix86_finalize_stack_frame_flags): This. Also clear
11276 frame_pointer_needed if -fno-omit-frame-pointer is used without
11278 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11279 with ix86_finalize_stack_frame_flags.
11280 (ix86_expand_epilogue): Likewise.
11281 (ix86_expand_split_stack_prologue): Likewise.
11283 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11286 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11287 to P_AES for Westmere.
11289 2017-08-07 Jonathan Yong <10walls@gmail.com>
11291 * config/i386/mingw.opt (fset-stack-executable): Removed.
11292 * config/i386/cygming.opt (fset-stack-executable): Moved
11294 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11296 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11298 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11300 2017-08-07 Marek Polacek <polacek@redhat.com>
11302 PR middle-end/81737
11303 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11305 2017-08-07 Martin Liska <mliska@suse.cz>
11307 * attribs.h (canonicalize_attr_name): New function.
11308 (cmp_attribs): Move from c-format.c and adjusted.
11309 (is_attribute_p): Moved from tree.h.
11310 * tree-inline.c: Add new includes.
11311 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11312 (private_is_attribute_p): Remove.
11313 (private_lookup_attribute): Likewise.
11314 (private_lookup_attribute_by_prefix): Simplify.
11315 (remove_attribute): Use is_attribute_p.
11316 * tree.h: Remove removed declarations.
11318 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11320 PR middle-end/81698
11321 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11322 instead of computing it in the function. Formatting fix.
11323 (expand_case): Don't rely on default_edge being the first edge,
11324 clear it if removing it, pass default_edge to
11325 emit_case_dispatch_table.
11326 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11329 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11331 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11332 insn in the function, emit NOP after the insn.
11334 2017-08-06 Tom de Vries <tom@codesourcery.com>
11336 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11339 2017-08-06 Tom de Vries <tom@codesourcery.com>
11341 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11344 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11346 PR tree-optimization/57371
11347 * match.pd: New pattern.
11349 2017-08-04 Marek Polacek <polacek@redhat.com>
11351 PR middle-end/81695
11352 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11353 perform the computation in offset_int.
11355 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11357 PR tree-optimization/81136
11358 * tree-vectorizer.h: Include tree-hash-traits.h.
11359 (vec_base_alignments): New typedef.
11360 (vec_info): Add a base_alignments field.
11361 (vect_record_base_alignments): Declare.
11362 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11364 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11365 (create_data_ref): Add an is_conditional_in_stmt argument.
11366 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11367 the is_conditional_in_stmt field.
11368 (data_ref_loc): Add an is_conditional_in_stmt field.
11369 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11370 (find_data_references_in_stmt): Update call to create_data_ref.
11371 (graphite_find_data_references_in_stmt): Likewise.
11372 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11373 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11374 (vect_record_base_alignment): New function.
11375 (vect_record_base_alignments): Likewise.
11376 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11377 for nested statements even if we fail to compute a misalignment.
11378 Use pooled base alignments for unconditional references.
11379 (vect_find_same_alignment_drs): Compare base addresses instead
11381 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11382 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11384 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11386 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11387 Add an explicit name for the enum. Use auto_vec for slp_instances
11388 and grouped_stores.
11389 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11391 (_bb_vec_info): Add a constructor and destructor.
11392 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11393 (destroy_loop_vec_info): Delete.
11394 (vect_destroy_datarefs): Likewise.
11395 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11396 (vec_info::vec_info): New function.
11397 (vec_info::~vec_info): Likewise.
11398 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11399 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11400 destroy_loop_vec_info.
11401 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11402 (_loop_vec_info::_loop_vec_info): ...this.
11403 (destroy_loop_vec_info): Replace with...
11404 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11405 the stmt_vec_infos. Leave handling of vec_info information to its
11406 destructor. Remove explicit vector releases.
11407 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11408 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11409 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11410 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11411 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11412 (destroy_bb_vec_info): Replace with...
11413 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11414 information to its destructor.
11415 (vect_slp_analyze_bb_1): Use new and delete instead of
11416 new_bb_vec_info and destroy_bb_vec_info.
11417 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11420 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11422 * tree-data-ref.h (subscript): Add access_fn field.
11423 (data_dependence_relation): Add could_be_independent_p.
11424 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11425 (same_access_functions): Move to tree-data-ref.c.
11426 * tree-data-ref.c (ref_contains_union_access_p): New function.
11427 (access_fn_component_p): Likewise.
11428 (access_fn_components_comparable_p): Likewise.
11429 (dr_analyze_indices): Add a reference to access_fn_component_p.
11430 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11432 (constant_access_functions): Likewise.
11433 (add_other_self_distances): Likewise.
11434 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11435 (initialize_data_dependence_relation): Use XCNEW and remove
11436 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11437 of access functions that have the same type. Allow the
11438 subsequence to end with different bases in some circumstances.
11439 Record the chosen access functions in SUB_ACCESS_FN.
11440 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11441 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11442 (subscript_dependence_tester_1): Likewise dra and drb.
11443 (build_classic_dist_vector): Update calls accordingly.
11444 (subscript_dependence_tester): Likewise.
11445 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11446 DDR_COULD_BE_INDEPENDENT_P.
11447 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11448 comp_alias_ddrs instead of may_alias_ddrs.
11449 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11451 (vect_analyze_data_ref_dependence): Use it if
11452 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11453 distance vectors if that fails.
11454 (dependence_distance_ge_vf): New function.
11455 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11456 LOOP_VINFO_MAY_ALIAS_DDRS.
11458 2017-08-04 Richard Biener <rguenther@suse.de>
11460 PR middle-end/81705
11461 * fold-const.c (fold_binary_loc): Properly restrict
11462 minus_var0 && minus_var1 case when associating undefined overflow
11465 2017-08-04 Tom de Vries <tom@codesourcery.com>
11467 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11469 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11471 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11472 Don't start diagnostic messages with a capital letter.
11473 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11475 (rs6000_invalid_builtin): Likewise.
11476 (rs6000_trampoline_init): Likewise.
11478 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11481 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11482 after setting changeable df flags.
11484 2017-08-03 Richard Biener <rguenther@suse.de>
11486 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11487 up if the use is in USE - X.
11489 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11491 * toplev.c (dumpfile.h): New include.
11492 (internal_error_reentered): New static function. Use it...
11493 (internal_error_function): ...here to handle reentered internal_error.
11495 2017-08-03 Richard Biener <rguenther@suse.de>
11497 PR middle-end/81148
11498 * fold-const.c (split_tree): Add minus_var and minus_con
11499 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11500 here but always use minus_*.
11501 (associate_trees): Assert we never associate with MINUS_EXPR
11502 and NULL first operand. Do not recurse for PLUS_EXPR operands
11503 when associating as MINUS_EXPR either.
11504 (fold_binary_loc): Track minus_var and minus_con.
11506 2017-08-03 Tom de Vries <tom@codesourcery.com>
11509 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11510 ACCEL_COMPILER, apply finish_options on
11511 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11513 2017-08-03 Tom de Vries <tom@codesourcery.com>
11516 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
11517 function_entry_patch_area_size > 0.
11519 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11522 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
11523 instead of 10??LU, perform unit multiplication in wide_int,
11524 don't change alloc_object_size_limit if the limit is larger
11527 PR tree-optimization/81655
11528 PR tree-optimization/81588
11529 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
11530 the case when ranges[i].low and high are 1 for unsigned type with
11533 PR middle-end/81052
11534 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
11535 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
11537 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11539 * tree-vrp.h: Add include guard.
11541 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
11544 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
11545 (ud2): New insn pattern.
11546 * config/i386/i386.c (ix86_expand_epilogue):
11547 For naked functions, generate ud2 instead of trap insn.
11549 2017-08-02 Marek Polacek <polacek@redhat.com>
11552 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
11554 2017-08-02 Tom de Vries <tom@codesourcery.com>
11555 Cesar Philippidis <cesar@codesourcery.com>
11557 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
11558 Add missing edge probabilities.
11560 2017-08-02 Tamar Christina <tamar.christina@arm.com>
11562 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
11563 Correct endianness.
11565 2017-08-02 Jakub Jelinek <jakub@redhat.com>
11567 PR middle-end/79499
11568 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
11569 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
11570 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
11571 prologue_seq sequences - if any.
11573 2017-08-02 Richard Biener <rguenther@suse.de>
11575 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
11576 via vectors if supported, integer extracts via punning if supported
11577 or otherwise vector extracts.
11579 2017-08-02 Richard Biener <rguenther@suse.de>
11581 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
11583 (bitmap_insert_into_set): ... this.
11585 2017-08-02 Richard Biener <rguenther@suse.de>
11587 PR tree-optimization/81633
11589 2015-08-17 Alan Hayward <alan.hayward@arm.com>
11591 PR tree-optimization/71752
11592 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
11594 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
11596 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
11597 (machine_function::call_ms2sysv_pad_out): Remove field.
11598 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
11599 (ix86_compute_frame_layout): Likewise.
11601 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11604 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11605 attribute with interrupt attribute.
11607 2017-08-01 Andrew Pinski <apinski@cavium.com>
11609 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11610 BIT_INSERT_EXPR's operand 1
11611 to see if the types precision matches.
11613 2017-08-01 Martin Liska <mliska@suse.cz>
11615 PR middle-end/70140
11616 * builtins.c (expand_builtin_memcpy_args): Remove.
11617 (expand_builtin_memcpy): Call newly added function
11618 expand_builtin_memory_copy_args.
11619 (expand_builtin_memcpy_with_bounds): Likewise.
11620 (expand_builtin_mempcpy): Remove last argument.
11621 (expand_builtin_mempcpy_with_bounds): Likewise.
11622 (expand_builtin_memory_copy_args): New function created from
11623 expand_builtin_mempcpy_args with small modifications.
11624 (expand_builtin_mempcpy_args): Remove.
11625 (expand_builtin_stpcpy): Remove unused argument.
11626 (expand_builtin): Likewise.
11627 (expand_builtin_with_bounds): Likewise.
11629 2017-08-01 Martin Liska <mliska@suse.cz>
11632 Make mempcpy more optimal (PR middle-end/70140).
11634 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11637 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11638 __builtin_vec_cmpne verify both arguments are compatible vectors
11639 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11640 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11641 move computation of aligned to after checking the argument types.
11645 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11646 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11649 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11650 Doug Rupp <rupp@adacore.com>
11651 Olivier Hainque <hainque@adacore.com>
11653 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11654 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11656 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11657 for TARGET_OS_CPP_BUILTIN.
11658 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11659 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11661 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11662 passing required abi options to the assembler for EABI configurations.
11663 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11664 of .text.hot and .text.unlikely sections for kernel modules when
11665 using ARM style exceptions.
11666 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11667 options. Add EXTRA_CC1_SPEC.
11668 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11670 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11672 (ARM_TARGET2_DWARF_FORMAT): Define.
11673 * config/arm/t-vxworks: Adjust multilib control to removal of the
11674 Diab command line options.
11676 2017-08-01 Martin Liska <mliska@suse.cz>
11678 PR gcov-profile/81561
11679 * gcov.c (unblock): Make unblocking safe as we need to preserve
11680 index correspondence of blocks and block_lists.
11682 2017-08-01 Richard Biener <rguenther@suse.de>
11684 PR tree-optimization/81181
11685 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11686 (compute_antic): ... end of iteration here.
11688 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11690 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11691 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11692 (ftree-slp-vectorize): Likewise.
11693 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11694 can no longer be set independent of flag_tree_loop_vectorize.
11695 * omp-general.c (emp_max_vf): Likewise.
11696 * opts.c (enable_fdo_optimizations): Remove references to
11697 flag_tree_vectorize, these are now implicit.
11698 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11699 and leave it for the options machinery.
11701 2017-08-01 Martin Liska <mliska@suse.cz>
11703 PR middle-end/70140
11704 * builtins.c (expand_builtin_memcpy_args): Remove.
11705 (expand_builtin_memcpy): Call newly added function
11706 expand_builtin_memory_copy_args.
11707 (expand_builtin_memcpy_with_bounds): Likewise.
11708 (expand_builtin_mempcpy): Remove last argument.
11709 (expand_builtin_mempcpy_with_bounds): Likewise.
11710 (expand_builtin_memory_copy_args): New function created from
11711 expand_builtin_mempcpy_args with small modifications.
11712 (expand_builtin_mempcpy_args): Remove.
11713 (expand_builtin_stpcpy): Remove unused argument.
11714 (expand_builtin): Likewise.
11715 (expand_builtin_with_bounds): Likewise.
11717 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11720 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11721 print "ds:" only for immediates in generic address space.
11723 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11726 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11727 (ix86_function_ok_for_sibcall): Return false for naked functions.
11729 2017-08-01 Richard Biener <rguenther@suse.de>
11731 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11732 (compute_antic): Seed worklist with exit block predecessors.
11733 * cfganal.c (dfs_find_deadend): For a cycle return the source
11734 of the edge closing it.
11736 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11738 * config/aarch64/aarch64.c
11739 (aarch64_can_const_movi_rtx_p): Move 0 check.
11741 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11743 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11744 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11746 * match.pd: Ditto in address comparison pattern.
11748 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11750 PR tree-optimization/81627
11751 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11752 closed ssa form for store-store chain.
11754 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11756 PR tree-optimization/81620
11757 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11758 for store-store chain.
11760 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11762 PR tree-optimization/81588
11763 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11764 ranges[i].in_p, invert comparison code ccode. For >/>=,
11765 swap rhs1 and rhs2 and comparison code unconditionally,
11766 for </<= don't do that. Don't swap rhs1/rhs2 again if
11767 ranges[i].in_p, instead invert comparison code ccode if
11768 opcode or oe->rank is BIT_IOR_EXPR.
11771 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11772 a direct optab to conversion optab.
11773 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11774 with GET_MODE_INNER as last argument instead of optab_handler.
11775 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11776 vector extraction if possible and optab is available.
11777 * expr.c (store_constructor): Use convert_optab_handler instead
11778 of optab_handler. Use vector initialization from smaller
11779 vectors if possible and optab is available.
11780 * tree-vect-stmts.c (vectorizable_load): Likewise.
11781 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11782 now have two modes.
11783 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11784 of vec_init from half-sized vectors with the same element mode.
11785 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11786 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11787 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11788 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11789 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11790 after mode in gen_vec_extract* calls.
11791 (vec_extract<mode>): Renamed to ...
11792 (vec_extract<mode><ssescalarmodelower>): ... this.
11793 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11794 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11795 element mode after mode in gen_vec_init* calls.
11796 (VEC_INIT_HALF_MODE): New mode iterator.
11797 (vec_init<mode>): Renamed to ...
11798 (vec_init<mode><ssescalarmodelower>): ... this.
11799 (vec_init<mode><ssehalfvecmodelower>): New expander.
11800 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11801 (vec_extractv2sfsf): ... this.
11802 (vec_initv2sf): Renamed to ...
11803 (vec_initv2sfsf): ... this.
11804 (vec_extractv2si): Renamed to ...
11805 (vec_extractv2sisi): ... this.
11806 (vec_initv2si): Renamed to ...
11807 (vec_initv2sisi): ... this.
11808 (vec_extractv4hi): Renamed to ...
11809 (vec_extractv4hihi): ... this.
11810 (vec_initv4hi): Renamed to ...
11811 (vec_initv4hihi): ... this.
11812 (vec_extractv8qi): Renamed to ...
11813 (vec_extractv8qiqi): ... this.
11814 (vec_initv8qi): Renamed to ...
11815 (vec_initv8qiqi): ... this.
11816 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11817 (vec_init<mode>): Renamed to ...
11818 (vec_init<mode><VEC_base_l>): ... this.
11819 (vec_extract<mode>): Renamed to ...
11820 (vec_extract<mode><VEC_base_l>): ... this.
11821 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11822 (vec_initv2sfsf): ... this.
11823 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11824 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11825 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11826 element mode after mode in gen_vec_init* calls.
11827 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11828 (vec_init<mode><Vel>): ... this.
11829 (vec_extract<mode>): Renamed to ...
11830 (vec_extract<mode><Vel>): ... this.
11831 * config/aarch64/iterators.md (Vel): New mode attribute.
11832 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11833 Add element mode after mode in gen_vec_extract* calls.
11834 * config/s390/vector.md (non_vec_l): New mode attribute.
11835 (vec_extract<mode>): Renamed to ...
11836 (vec_extract<mode><non_vec_l>): ... this.
11837 (vec_init<mode>): Renamed to ...
11838 (vec_init<mode><non_vec_l>): ... this.
11839 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11840 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11842 * config/arm/iterators.md (V_elem_l): New mode attribute.
11843 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11844 (vec_extract<mode><V_elem_l>): ... this.
11845 (vec_extractv2di): Renamed to ...
11846 (vec_extractv2didi): ... this.
11847 (vec_init<mode>): Renamed to ...
11848 (vec_init<mode><V_elem_l>): ... this.
11849 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11850 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11851 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11852 Add element mode after gen_vec_extract* calls.
11853 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11854 (vec_init<mode><unitmode>): ... this.
11855 (vec_extract<mode>): Renamed to ...
11856 (vec_extract<mode><unitmode>): ... this.
11857 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11858 (vec_init<mode><unitmode>): ... this.
11859 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11860 (vec_initv2sfsf): ... this.
11861 (vec_extractv2sf): Renamed to ...
11862 (vec_extractv2sfsf): ... this.
11863 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11864 Add element mode after gen_vec_extract* calls.
11865 * config/mips/mips.md (unitmode): New mode iterator.
11866 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11867 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11868 * config/spu/spu.md (inner_l): New mode attribute.
11869 (vec_init<mode>): Renamed to ...
11870 (vec_init<mode><inner_l>): ... this.
11871 (vec_extract<mode>): Renamed to ...
11872 (vec_extract<mode><inner_l>): ... this.
11873 * config/sparc/sparc.md (veltmode): New mode iterator.
11874 (vec_init<VMALL:mode>): Renamed to ...
11875 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11876 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11877 (vec_initv2sisi): ... this.
11878 (vec_initv2sf): Renamed to ...
11879 (vec_initv2sfsf): ... this.
11880 (vec_extractv2sf): Renamed to ...
11881 (vec_extractv2sfsf): ... this.
11882 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11883 (vec_init<mode>): Renamed to ...
11884 (vec_init<mode><VEC_base_l>): ... this.
11885 (vec_extract<mode>): Renamed to ...
11886 (vec_extract<mode><VEC_base_l>): ... this.
11887 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11888 (vec_initv2sfsf): ... this.
11889 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11890 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11891 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11892 gen_vec_init* calls.
11894 2017-08-01 Richard Biener <rguenther@suse.de>
11896 PR tree-optimization/81297
11897 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11898 TREE_OVERFLOW from INTEGER_CSTs.
11900 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11902 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11904 2017-07-31 Carl Love <cel@us.ibm.com>
11906 * config/rs6000/rs6000-c: Add support for built-in functions
11907 vector signed char vec_xl_be (signed long long, signed char *);
11908 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11909 vector signed int vec_xl_be (signed long long, signed int *);
11910 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11911 vector signed long long vec_xl_be (signed long long, signed long long *);
11912 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11913 vector signed short vec_xl_be (signed long long, signed short *);
11914 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11915 vector double vec_xl_be (signed long long, double *);
11916 vector float vec_xl_be (signed long long, float *);
11917 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11918 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11919 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11921 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11922 (altivec_expand_builtin): Add switch statement to call
11923 altivec_expand_xl_be for each builtin.
11924 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11925 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11926 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11927 __builtin_vsx_le_be_v16qi.
11928 * doc/extend.texi: Update the built-in documentation file for the
11929 new built-in functions.
11931 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11934 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11936 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11938 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11940 * config.gcc: Add z14.
11941 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11942 CPU model numbers for z13s and z14.
11943 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11945 * config/s390/s390-opts.h (enum processor_type): Rename
11946 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11947 * config/s390/s390.c (processor_table): Add field for CPU name to
11948 be passed to Binutils.
11949 (s390_asm_output_machine_for_arch): Use the new field in
11950 processor_table for Binutils.
11951 (s390_expand_builtin): Replace arch12 with z14.
11952 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11953 (s390_get_sched_attrmask): Likewise.
11954 (s390_get_unit_mask): Likewise.
11955 * config/s390/s390.opt: Add z14 to processor_type enum.
11957 2017-07-31 Martin Jambor <mjambor@suse.cz>
11960 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11961 regardless of optimization level.
11963 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11964 Martin Liska <mliska@suse.cz>
11966 * predict.def: Remove old comment and adjust probability.
11967 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11968 PREDICT statements.
11970 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11973 * config/i386/i386.c (ix86_function_naked): New function.
11974 (ix86_can_use_return_insn_p): Return false for naked functions.
11975 (ix86_expand_prologue): Skip prologue for naked functions.
11976 (ix86_expand_epilogue): Skip epilogue for naked functions
11977 and emit trap instruction.
11978 (ix86_warn_func_return): New function.
11979 (ix86_attribute_table): Add "naked" attribute specification.
11980 (TARGET_WARN_FUNC_RETURN): Define.
11981 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11983 2017-07-31 Martin Liska <mliska@suse.cz>
11985 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11986 (dump_gimple_bb_header): Always dump BB info.
11987 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11989 2017-07-31 Martin Liska <mliska@suse.cz>
11992 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11993 also with current_function_decl non-null equality.
11995 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11998 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11999 change type to the element type, instead add eltype variable and
12000 use it where we are interested in the element type.
12002 PR tree-optimization/81603
12003 * ipa-polymorphic-call.c
12004 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12005 offset arithmetic in offset_int, bail out if the resulting bit offset
12006 doesn't fit into shwi.
12008 2017-07-31 Martin Liska <mliska@suse.cz>
12010 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12011 (gimplify_save_expr): Fix comment.
12013 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12016 * config/i386/i386.c (ix86_function_arg): Update arguments for
12018 (ix86_compute_frame_layout): Set the initial stack offset to
12019 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12020 INCOMING_FRAME_SP_OFFSET.
12021 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12022 stack before exception handler returns.
12023 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12024 the 'ERROR_CODE' for exception handler.
12026 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12028 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12029 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12030 (ASM_OUTPUT_REG_POP): Ditto.
12031 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12032 instead of asm_fprintf to output pure string.
12034 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12036 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12037 to imported_module_or_decl hook.
12038 (debug_nothing_tree_tree_tree_bool): Remove.
12039 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12040 * debug.c (do_nothing_debug_hooks): Use
12041 debug_nothing_tree_tree_tree_bool_bool instead of
12042 debug_nothing_tree_tree_tree_bool.
12043 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12044 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12045 * sdbout.c (sdb_debug_hooks): Likewise.
12046 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12047 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12049 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12050 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12051 attribute, don't add anything.
12053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12055 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12056 (fold_build2_stat_loc): Likewise.
12057 (fold_build3_stat_loc): Likewise.
12058 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12059 (fold_build1_loc): Remove macro.
12060 (fold_build2_loc): Likewise.
12061 (fold_build3_loc): Likewise.
12063 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12065 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12066 (gimple_build_debug_bind_source_stat): Likewise.
12067 * gimple.h (gimple_build_debug_bind): Remove macro.
12068 (gimple_build_debug_bind_source): Likewise.
12070 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12072 * bitmap.c (bitmap_alloc): Adjust.
12073 (bitmap_gc_alloc): Likewise.
12074 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12076 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12078 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12079 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12080 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12081 (bitmap_gc_alloc_stat): Likewise.
12082 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12084 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12086 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12087 * rtl.h (shallow_copy_rtx): Remove macro.
12089 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12091 * emit-rtl.c (gen_raw_REG): Adjust.
12092 * gengenrtl.c (gendef): Likewise.
12093 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12094 * rtl.h (rtx_alloc): Remove macro.
12096 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12098 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12099 (build_tree_list_stat): Likewise.
12100 * tree.h (build_tree_list): Remove macro.
12101 (build_tree_list_vec): Likewise.
12103 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12105 * tree.c (make_vector_stat): Remove _stat from name.
12106 (build_vector_stat): Likewise.
12107 * tree.h (make_vector_stat): Remove macro.
12108 (build_vector_stat): Likewise.
12110 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12112 * tree.h (build_var_debug_value): Remove prototype.
12114 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12116 * tree.c (tree_cons_stat): Remove _stat from name.
12117 * tree.h (tree_cons): Remove macro.
12119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12121 * tree.c (build_vl_exp_stat): Remove _stat from name.
12122 * tree.h (build_vl_exp): Remove macro.
12124 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12126 * tree.c (build_decl_stat): Remove _stat from name.
12127 * tree.h (build_decl): Remove macro.
12129 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12131 * gimple.c (gimple_build_with_ops_stat): Adjust.
12132 (gimple_alloc_stat): Remove _stat from name.
12133 * gimple.h (gimple_alloc): Remove macro.
12135 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12137 * tree.c (make_tree_vec_stat): Remove _stat from name.
12138 (grow_tree_vec_stat): Likewise.
12139 * tree.h (make_tree_vec_stat): Adjust prototype.
12140 (grow_tree_vec_stat): Likewise.
12141 (make_tree_vec): Remove macro.
12142 (grow_tree_vec): Likewise.
12144 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12146 * fold-const.c (fold_build1_stat_loc): Adjust.
12147 (fold_build2_stat_loc): Likewise.
12148 (fold_build3_stat_loc): Likewise.
12149 * tree.c (build0_stat): Remove _stat from name.
12150 (build1_stat): Likewise.
12151 (build2_stat): Likewise.
12152 (build3_stat): Likewise.
12153 (build4_stat): Likewise.
12154 (build5_stat): Likewise.
12155 * tree.h (build1_loc): Remove macro, and rename _stat function
12157 (build2_loc): Likewise.
12158 (build3_loc): Likewise.
12159 (build4_loc): Likewise.
12160 (build5_loc): Likewise.
12162 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12164 * tree.c (make_int_cst_stat): Remove _stat from name.
12165 * tree.h (make_int_cst_stat): Adjust prototype.
12166 (make_int_cst): Remove macro.
12168 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12170 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12171 * tree.h (make_tree_binfo_stat): Adjust prototype.
12172 (make_tree_binfo): Remove.
12174 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12176 * tree.c (copy_node_stat): Rename to copy_node.
12177 (build_distinct_type_copy): Adjust.
12178 * tree.h (copy_node_stat): Adjust prototype.
12179 (copy_node): Remove macro.
12181 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12183 * tree.c (make_node_stat): rename to make_node.
12184 (build_tree_list_stat): Adjust.
12185 (build0_stat): Likewise.
12186 (build2_stat): Likewise.
12187 (build3_stat): Likewise.
12188 (build4_stat): Likewise.
12189 (build5_stat): Likewise.
12190 (build_decl_stat): Likewise.
12191 * tree.h (make_node_stat): Adjust prototype.
12192 (make_node): remove macro.
12194 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12196 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12197 (PPC_FEATURE2_SCV): Likewise.
12198 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12200 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12202 * config/aarch64/aarch64.c
12203 (aarch64_internal_mov_immediate): Add new special pattern.
12204 * config/aarch64/aarch64.md (*movdi_aarch64):
12205 Add reg/32bit const mov case.
12207 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12208 Richard Sandiford <richard.sandiford@linaro.org>
12210 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12211 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12212 Add integer and movi cases.
12213 (movi-split-hf-df-sf split, fp16): New.
12214 (enabled): Added TARGET_FP_F16INST.
12215 * config/aarch64/iterators.md (GPF_HF): New.
12216 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12218 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12220 * config/aarch64/aarch64.c
12221 (aarch64_simd_container_mode): Add prototype.
12222 (aarch64_expand_mov_immediate): Add HI support.
12223 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12224 (aarch64_can_const_movi_rtx_p): New.
12225 (aarch64_preferred_reload_class):
12226 Remove restrictions of using FP registers for certain SIMD operations.
12227 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12228 (aarch64_valid_floating_const): Add integer move validation.
12229 (aarch64_simd_imm_scalar_p): Remove.
12230 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12231 (aarch64_legitimate_constant_p): Expand list of supported cases.
12232 * config/aarch64/aarch64-protos.h
12233 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12234 (aarch64_reinterpret_float_as_int): New.
12235 (aarch64_simd_imm_scalar_p): Remove.
12236 * config/aarch64/constraints.md (Uvi): New.
12237 (Dd): Split into Ds and new Dd.
12238 * config/aarch64/aarch64.md (*movsi_aarch64):
12240 (*movdi_aarch64): Add SIMD mov case.
12242 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12244 * tree-predcom.c: (struct chain): Handle store-store chain in which
12245 stores for elimination only store loop invariant values.
12246 (execute_pred_commoning_chain): Ditto.
12247 (prepare_initializers_chain_store_elim): Ditto.
12248 (prepare_finalizers): Ditto.
12249 (is_inv_store_elimination_chain): New function.
12250 (initialize_root_vars_store_elim_1): New function.
12252 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12254 * tree-predcom.c: Revise general description of the pass.
12255 (enum chain_type): New enum type for store elimination.
12256 (struct chain): New field supporting store elimination.
12257 (struct component): Ditto.
12258 (dump_chain): Dump store-stores chain.
12259 (release_chain): Release resources.
12260 (split_data_refs_to_components): Compute and create component
12261 contains only stores for elimination.
12262 (get_chain_last_ref_at): New function.
12263 (make_invariant_chain): Initialization.
12264 (make_rooted_chain): Specify chain type in parameter and record it.
12265 (add_looparound_copies): Skip for store-stores chain.
12266 (determine_roots_comp): Compute type of chain and pass it to
12268 (initialize_root_vars_store_elim_2): New function.
12269 (finalize_eliminated_stores): New function.
12270 (remove_stmt): Handle store for elimination.
12271 (execute_pred_commoning_chain): Execute predictive commoning on
12272 store-store chains.
12273 (determine_unroll_factor): Skip unroll for store-stores chain.
12274 (prepare_initializers_chain_store_elim): New function.
12275 (prepare_initializers_chain): Hanlde store-store chain.
12276 (prepare_finalizers_chain, prepare_finalizers): New function.
12277 (tree_predictive_commoning_loop): Return integer value indicating
12278 if loop is unrolled or lcssa form is corrupted.
12279 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12281 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12283 * tree-predcom.c (initialize_root): Delete.
12284 (execute_pred_commoning_chain): Initialize root vars and replace
12285 reference of non-combined chain directly, rather than call above
12288 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12290 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12291 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12293 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12295 * tree-predcom.c (struct chain): New field init_seq.
12296 (release_chain): Release init_seq.
12297 (prepare_initializers_chain): Record intialization stmts in above
12299 (insert_init_seqs): New function.
12300 (tree_predictive_commoning_loop): Call insert_init_seqs.
12302 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12304 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12306 2017-07-28 Richard Biener <rguenther@suse.de>
12308 * match.pd: Remove superfluous :c.
12309 * genmatch.c (simplify::id): Add member.
12310 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12312 (current_id): New global.
12313 (dt_node::parent): Move from ...
12314 (dt_operand::parent): ... here. Add for_id member.
12315 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12316 (decision_tree::find_node): Relax order requirement when
12317 merging DT_TRUE nodes to ones inbetween the current simplify
12318 and the one we try to merge with. Add diagnostic whenever
12319 we need to enforce pattern order by not merging.
12320 (decision_tree::insert): Set current_id.
12321 (decision_tree::print_node): Dump parent node and for_id.
12322 (parser::last_id): Add member.
12323 (parser::push_simplify): Assign unique id.
12324 (parser::parser): Initialize last_id.
12326 2017-07-28 Martin Liska <mliska@suse.cz>
12329 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12330 gimple_build_debug_bind.
12332 2017-07-28 Richard Biener <rguenther@suse.de>
12334 PR tree-optimization/81502
12335 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12337 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12339 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12340 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12341 for BIT_FIELD_REF args.
12342 * fold-const.c (make_bit_field_ref): Likewise.
12343 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12345 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12348 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12349 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12350 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12351 Or it into SANITIZER_UNDEFINED.
12352 * ubsan.c: Include gimple-fold.h and varasm.h.
12353 (ubsan_expand_ptr_ifn): New function.
12354 (instrument_pointer_overflow): New function.
12355 (maybe_instrument_pointer_overflow): New function.
12356 (instrument_object_size): Formatting fix.
12357 (pass_ubsan::execute): Call instrument_pointer_overflow
12358 and maybe_instrument_pointer_overflow.
12359 * internal-fn.c (expand_UBSAN_PTR): New function.
12360 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12361 * sanitizer.def (__ubsan_handle_pointer_overflow,
12362 __ubsan_handle_pointer_overflow_abort): New builtins.
12363 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12364 * internal-fn.def (UBSAN_PTR): New internal function.
12365 * opts.c (sanitizer_opts): Add pointer-overflow.
12366 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12367 * fold-const.c (build_range_check): Compute pointer range check in
12368 integral type if pointer arithmetics would be needed. Formatting
12371 2017-07-28 Martin Liska <mliska@suse.cz>
12374 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12375 parameters that are of a variable-length.
12377 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12379 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12381 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12382 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12383 (CRT_CALL_STATIC_FUNCTION): Likewise.
12384 (ASM_DEFAULT_SPEC): New define.
12385 (ASM_SPEC32): Likewise.
12386 (ASM_SPEC64): Likewise.
12387 (ASM_SPEC_COMMON): Likewise.
12388 (ASM_SPEC): Likewise.
12389 (INVALID_64BIT): Likewise.
12390 (LINK_OS_DEFAULT_SPEC): Likewise.
12391 (LINK_OS_SPEC32): Likewise.
12392 (LINK_OS_SPEC64): Likewise.
12393 (POWERPC_LINUX): Likewise.
12394 (PTRDIFF_TYPE): Likewise.
12395 (RESTORE_FP_PREFIX): Likewise.
12396 (RESTORE_FP_SUFFIX): Likewise.
12397 (SAVE_FP_PREFIX): Likewise.
12398 (SAVE_FP_SUFFIX): Likewise.
12399 (SIZE_TYPE): Likewise.
12400 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12401 (TARGET_64BIT): Likewise.
12402 (TARGET_64BIT): Likewise.
12403 (TARGET_AIX): Likewise.
12404 (WCHAR_TYPE_SIZE): Likewise.
12405 (WCHAR_TYPE): Undefine.
12406 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12407 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12408 (CPP_OS_RTEMS_SPEC): Delete.
12409 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12410 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12412 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12414 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12416 PR tree-optimization/81578
12417 * tree-parloops.c (build_new_reduction): Bail out if
12418 reduction_code isn't one of the standard OpenMP reductions.
12419 Move the details printing after that decision.
12421 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12423 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12424 related to reload_in_progress.
12425 (splat_input_operand): Likewise.
12426 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12428 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12430 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12431 (TARGET_INSTANTIATE_DECLS): Likewise.
12432 (legitimate_indexed_address_p): Delete reload_in_progress code.
12433 (rs6000_debug_legitimate_address_p): Likewise.
12434 (rs6000_eliminate_indexed_memrefs): Likewise.
12435 (rs6000_emit_le_vsx_store): Likewise.
12436 (rs6000_emit_move_si_sf_subreg): Likewise.
12437 (rs6000_emit_move): Likewise.
12438 (register_to_reg_type): Likewise.
12439 (rs6000_pre_atomic_barrier): Likewise.
12440 (rs6000_machopic_legitimize_pic_address): Likewise.
12441 (rs6000_allocate_stack_temp): Likewise.
12442 (rs6000_address_for_fpconvert): Likewise.
12443 (rs6000_address_for_altivec): Likewise.
12444 (rs6000_secondary_memory_needed_rtx): Delete function.
12445 (rs6000_check_sdmode): Likewise.
12446 (rs6000_alloc_sdmode_stack_slot): Likewise.
12447 (rs6000_instantiate_decls): Likewise.
12448 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12449 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12450 Delete reload_in_progress.
12451 (*vec_reload_and_plus_<mptrsize>): Likewise.
12452 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12453 (vsx_div_v2di): Likewise.
12454 (vsx_udiv_v2di): Likewise.
12456 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12458 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12459 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12460 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12461 (rs6000_debug_reg_global): Delete print of LRA status.
12462 (rs6000_option_override_internal): Delete dead LRA related code.
12463 (rs6000_lra_p): Delete function.
12464 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12466 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12468 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12469 * config/riscv/rtems.h: New file.
12471 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12472 Sudakshina Das <sudi.das@arm.com>
12474 * config/aarch64/aarch64.md
12475 (define_split for and<mode>3nr_compare): Move
12476 non aarch64_logical_operand to a register.
12477 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12478 register immediate operand to a register.
12479 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12481 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12483 PR middle-end/81564
12484 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12486 2017-07-27 Richard Biener <rguenther@suse.de>
12488 PR tree-optimization/81573
12489 PR tree-optimization/81494
12490 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12491 multi defuse cycle case.
12493 2017-07-27 Richard Biener <rguenther@suse.de>
12495 PR tree-optimization/81571
12496 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12499 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12501 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12502 earlier and only if MASK_FPU is set. Adjust formatting.
12504 2017-07-27 Martin Liska <mliska@suse.cz>
12506 * opt-functions.awk: Add validation of value of Init.
12507 * optc-gen.awk: Pass new argument.
12509 2017-07-27 Martin Liska <mliska@suse.cz>
12511 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12512 Fix wrong condition.
12514 2017-07-27 Martin Liska <mliska@suse.cz>
12516 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
12517 BBs and edges seen by autoFDO.
12519 2017-07-27 Richard Biener <rguenther@suse.de>
12521 PR tree-optimization/81502
12522 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
12523 with incompatible but same sized type.
12524 (execute_update_addresses_taken): Likewise.
12526 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
12528 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
12529 flag_tree_loop_vectorize rather than flag_tree_vectorize.
12531 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12534 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
12535 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
12536 Change s_operand to memory_operand.
12538 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
12540 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
12541 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12542 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
12543 Emit instructions rather than returning an expression. Handle TFmode
12544 and KFmode by casting to TImode.
12545 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
12546 (rs6000_emit_le_vsx_store): Likewise.
12547 * config/rs6000/vsx.md (VSX_TI): New iterator.
12548 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
12549 (*vsx_le_undo_permute_<mode>): Likewise.
12550 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
12551 emit the split sequence.
12552 (*vsx_le_perm_store_<mode>): Likewise.
12554 2017-07-27 Jakub Jelinek <jakub@redhat.com>
12556 PR tree-optimization/81555
12557 PR tree-optimization/81556
12558 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
12559 if true, force CHANGED for the recursive invocation.
12560 (reassociate_bb): Remember original length of ops array, pass
12561 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
12563 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
12564 attributes for noipa attribute. For naked attribute use
12565 lookup_attribute first before lookup_attribute_spec.
12566 * final.c (rest_of_handle_final): Disable IPA RA for functions with
12568 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
12569 for functions with noipa attribute.
12570 (cgraph_externally_visible_p): Return true for functions with noipa
12572 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
12573 for functions with noipa attribute.
12574 * doc/extend.texi: Document noipa function attribute.
12575 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
12576 also for functions with noipa attribute.
12577 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
12579 2017-07-26 Andrew Pinski <apinski@cavium.com>
12581 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
12582 vec_unalign_load_cost and vec_unalign_store_cost.
12584 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12586 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
12587 -mvsx-small-integer option.
12588 (ISA_3_0_MASKS_IEEE): Likewise.
12589 (OTHER_VSX_VECTOR_MASKS): Likewise.
12590 (POWERPC_MASKS): Likewise.
12591 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
12592 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
12593 code, only testing for DImode being allowed in non-VSX floating
12595 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
12596 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
12598 (rs6000_option_override_internal): Delete -mvsx-small-integer.
12599 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
12600 TARGET_P8_VECTOR test.
12601 (rs6000_secondary_reload_simple_move): Likewise.
12602 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12603 since TARGET_P9_VECTOR was already tested.
12604 (rs6000_opt_masks): Remove -mvsx-small-integer.
12605 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12606 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12608 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12609 test for TARGET_VEXTRACTUB was used, and that uses
12611 (p9 extract splitter): Likewise.
12612 (vsx_extract_<mode>_di_p9): Likewise.
12613 (vsx_extract_<mode>_store_p9): Likewise.
12614 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12615 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12616 the elimination of TARGET_VSX_SMALL_INTEGER.
12617 (vsx_extract_<mode>_p8): Likewise.
12618 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12619 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12620 (vsx_set_<mode>_p9): Likewise.
12621 (vsx_set_v4sf_p9): Likewise.
12622 (vsx_set_v4sf_p9_zero): Likewise.
12623 (vsx_insert_extract_v4sf_p9): Likewise.
12624 (vsx_insert_extract_v4sf_p9_2): Likewise.
12625 * config/rs6000/rs6000.md (sign extend splitter): Change
12626 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12627 (floatsi<mode>2_lfiwax_mem): Likewise.
12628 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12629 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12630 since a test for TARGET_P9_VECTOR was used.
12631 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12632 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12633 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12634 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12635 TARGET_P8_VECTOR test.
12636 (fix_trunc<mode>si2_stfiwx): Likewise.
12637 (fix_trunc<mode>si2_internal): Likewise.
12638 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12639 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12641 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12642 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12643 TARGET_P8_VECTOR test.
12644 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12645 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12646 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12648 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12649 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12650 since a test for TARGET_P9_VECTOR was used.
12651 (splitter for loading small constants): Likewise.
12653 2017-07-26 Andrew Pinski <apinski@cavium.com>
12655 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12658 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12661 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12662 (fp_valid_at): Likewise.
12664 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12666 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12667 (qdf24xx_addrcost_table): Likewise.
12668 (cortexa57_tunings): Update to use generic_branch_cost.
12669 (cortexa72_tunings): Likewise.
12670 (cortexa73_tunings): Likewise.
12671 (qdf24xx_tunings): Likewise.
12673 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12675 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12676 (thunderx2t99_branch_cost): Likewise.
12677 (cortexa35_tunings): Update to use generic_branch_cost.
12678 (cortexa53_tunings): Likewise.
12679 (cortexa57_tunings): Likewise.
12680 (cortexa72_tunings): Likewise.
12681 (cortexa73_tunings): Likewise.
12682 (thunderx2t99_tunings): Likewise.
12684 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12686 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12687 (sparc_option_override): Honour MASK_FSMULD.
12688 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12689 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12690 * config/sparc/sparc.opt (mfsmuld): New option.
12691 * doc/invoke.texi (mfsmuld): Document option.
12693 2017-07-26 Marek Polacek <polacek@redhat.com>
12695 PR middle-end/70992
12696 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12698 2017-07-26 Richard Biener <rguenther@suse.de>
12700 * gimple-match-head.c (do_valueize): Return OP if valueize
12702 (get_def): New helper to get at the def stmt of a SSA name
12703 if valueize allows.
12704 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12705 do_valueize to get at the def stmt.
12706 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12708 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12710 PR middle-end/46932
12711 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12713 2017-07-26 Martin Liska <mliska@suse.cz>
12716 * function.c (expand_function_start): Make expansion of
12717 nonlocal_goto_save_area after parm_birth_insn.
12719 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12721 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12722 from all CPU target flags enable members.
12724 2017-07-26 Richard Biener <rguenther@suse.de>
12726 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12727 (decision_tree::gen): Make 'type' const.
12728 (write_predicate): Likewise.
12730 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12732 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12733 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12734 (rs6000_option_override_internal): Likewise.
12735 (rs6000_expand_vector_set): Likewise.
12736 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12737 (TARGET_UPPER_REGS_SF): Likewise.
12738 (TARGET_UPPER_REGS_DI): Likewise.
12739 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12740 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12741 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12742 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12743 (Splitters for DI constants in Altivec registers): Likewise.
12744 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12745 (vsx_set_v4sf_p9): Likewise.
12746 (vsx_set_v4sf_p9_zero): Likewise.
12747 (vsx_insert_extract_v4sf_p9): Likewise.
12748 (vsx_insert_extract_v4sf_p9_2): Likewise.
12750 2017-07-25 Carl Love <cel@us.ibm.com>
12752 * doc/extend.texi: Update the built-in documentation file for the
12753 existing built-in functions
12754 vector signed char vec_cnttz (vector signed char);
12755 vector unsigned char vec_cnttz (vector unsigned char);
12756 vector signed short vec_cnttz (vector signed short);
12757 vector unsigned short vec_cnttz (vector unsigned short);
12758 vector signed int vec_cnttz (vector signed int);
12759 vector unsigned int vec_cnttz (vector unsigned int);
12760 vector signed long long vec_cnttz (vector signed long long);
12761 vector unsigned long long vec_cnttz (vector unsigned long long);
12763 2017-07-25 Andrew Pinski <apinski@cavium.com>
12765 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12766 accesses where the use is for the first operand of a BIT_INSERT.
12768 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12771 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12772 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12774 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12776 * config/i386/gstabs.h: Delete.
12777 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12779 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12781 * config/i386/i386.c (ix86_decompose_address): Do not check for
12782 register RTX when looking at index_reg or base_reg.
12783 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12785 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12787 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12788 to update EH info here.
12790 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12792 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12794 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12796 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12798 2017-07-25 Torsten Duwe <duwe@suse.de>
12800 * common.opt: Introduce -fpatchable-function-entry
12801 command line option, and its variables function_entry_patch_area_size
12802 and function_entry_patch_area_start.
12803 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12804 including a two-value parser.
12805 * target.def (print_patchable_function_entry): New target hook.
12806 * targhooks.h (default_print_patchable_function_entry): New function.
12807 * targhooks.c (default_print_patchable_function_entry): Likewise.
12808 * toplev.c (process_options): Switch off IPA-RA if
12809 patchable function entries are being generated.
12810 * varasm.c (assemble_start_function): Look at the
12811 patchable-function-entry command line switch and current
12812 function attributes and maybe generate NOP instructions by
12813 calling the print_patchable_function_entry hook.
12814 * doc/extend.texi: Document patchable_function_entry attribute.
12815 * doc/invoke.texi: Document -fpatchable_function_entry
12816 command line option.
12817 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12819 * doc/tm.texi: Re-generate.
12821 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12824 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12825 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12827 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12829 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12831 2017-07-25 Richard Biener <rguenther@suse.de>
12833 PR tree-optimization/81455
12834 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12835 not walk in cycles when looking for guards.
12837 2017-07-25 Richard Biener <rguenther@suse.de>
12839 PR tree-optimization/81529
12840 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12841 when optimizing backedge uses.
12843 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12845 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12847 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12848 to dl_section_ref. On AIX, append an expression to subtract
12849 the size of the section length to dl_section_ref.
12851 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12853 * configure.ac: If any of the config.* scripts fail, exit 1.
12854 * configure: Regenerate.
12856 2017-07-25 Richard Biener <rguenther@suse.de>
12858 PR middle-end/81546
12859 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12860 of immediate uses, be more verbose on errors.
12862 2017-07-25 Richard Biener <rguenther@suse.de>
12864 PR tree-optimization/81510
12865 * tree-vect-loop.c (vect_is_simple_reduction): When the
12866 reduction stmt is not inside the loop bail out.
12868 2017-07-25 Richard Biener <rguenther@suse.de>
12870 PR tree-optimization/81303
12871 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12872 profitability check against LOOP_VINFO_NITERSM1.
12874 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12876 * domwalk.c (cmp_bb_postorder): Simplify.
12877 (sort_bbs_postorder): New function. Use it...
12878 (dom_walker::walk): ...here to optimize common cases.
12880 2017-07-25 Martin Liska <mliska@suse.cz>
12883 * ipa-visibility.c (function_and_variable_visibility): Make the
12884 redirection just on target that supports aliasing.
12885 Fix GNU coding style.
12887 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12890 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12892 * config/arm/rtems.h: Likewise.
12893 * config/bfin/rtems.h: Likewise.
12894 * config/i386/rtemself.h: Likewise.
12895 * config/lm32/rtems.h: Likewise.
12896 * config/m32c/rtems.h: Likewise.
12897 * config/m68k/rtemself.h: Likewise.
12898 * config/microblaze/rtems.h: Likewise.
12899 * config/mips/rtems.h: Likewise.
12900 * config/moxie/rtems.h: Likewise.
12901 * config/nios2/rtems.h: Likewise.
12902 * config/powerpcspe/rtems.h: Likewise.
12903 * config/rs6000/rtems.h: Likewise.
12904 * config/rtems.h: Likewise.
12905 * config/sh/rtems.h: Likewise.
12906 * config/sh/rtemself.h: Likewise.
12907 * config/sparc/rtemself.h: Likewise.
12909 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12912 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12913 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12914 * tree-ssa-structalias.c (alias_get_name): Same.
12916 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12919 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12920 instructions if no du chain is found.
12922 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12924 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12926 2017-07-25 Richard Biener <rguenther@suse.de>
12928 PR middle-end/81505
12929 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12932 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12934 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12935 upper-regs options.
12936 (ISA_2_7_MASKS_SERVER): Likewise.
12937 (ISA_3_0_MASKS_IEEE): Likewise.
12938 (OTHER_P8_VECTOR_MASKS): Likewise.
12939 (OTHER_VSX_VECTOR_MASKS): Likewise.
12940 (POWERPC_MASKS): Likewise.
12941 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12942 duplicate list of options.
12943 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12944 explicit -mupper-regs options.
12945 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12946 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12947 alias for -mupper-regs-df.
12948 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12949 (rs6000_init_hard_regno_mode_ok): Likewise.
12950 (rs6000_option_override_internal): Likewise.
12951 (rs6000_opt_masks): Likewise.
12952 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12953 options in terms of whether -mvsx or -mpower8-vector was used.
12954 (TARGET_UPPER_REGS_DI): Likewise.
12955 (TARGET_UPPER_REGS_SF): Likewise.
12956 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12957 -mupper-regs-* options.
12959 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12961 * passes.c (emergency_dump_function): Print some empty lines and a
12962 header before the RTL dump.
12964 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12966 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12968 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12971 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12972 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12974 2017-07-24 Carl Love <cel@us.ibm.com>
12976 * config/rs6000/rs6000-c.c: Add support for built-in functions
12977 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12978 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12979 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12980 vec_extract_fp_from_shortl): Add defines for the two builtins.
12981 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12982 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12984 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12985 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12986 * doc/extend.texi: Update the built-in documentation file for the
12987 new built-in function.
12989 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12992 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12994 * doc/generic.texi: Likewise.
12995 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12996 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12998 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13000 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13001 (aarch64_mls_elt_merge<mode>): Likewise.
13003 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13005 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13006 having __cxa_atexit.
13008 2017-07-23 Michael Collison <michael.collison@arm.com>
13010 * config/arm/arm.c (arm_option_override): Deprecate
13011 use of -mstructure-size-boundary.
13012 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13013 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13015 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13018 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13019 Reduce cost estimate for direct moves.
13021 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13024 * config/i386/i386.c (ix86_option_override_internal): Disable
13025 BMI, BMI2 and TBM instructions for -m16.
13027 2017-07-21 Carl Love <cel@us.ibm.com>
13029 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13030 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13031 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13033 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13034 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13035 VMULOSW): New enum "unspec" values.
13036 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13037 altivec_vmulosw): New patterns.
13038 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13039 VMULOSW): Add definitions.
13041 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13043 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13044 (qdf24xx): Likewise.
13045 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13046 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13047 (AARCH64_FL_V8_1): Renumber.
13048 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13049 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13050 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13051 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13052 rdma to feature modifiers list.
13054 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13056 PR middle-end/56727
13057 * ipa-visibility (function_and_variable_visibility): Convert
13058 recursive PLT call to direct call if appropriate.
13060 2017-07-21 Andrew Pinski <apinski@cavium.com>
13062 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13063 operand 1 to see if the types precision matches.
13064 * fold-const.c (operand_equal_p): Likewise.
13066 2017-07-21 Richard Biener <rguenther@suse.de>
13068 PR tree-optimization/81303
13069 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13070 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13071 (vect_peeling_hash_get_lowest_cost): Adjust.
13072 (vect_enhance_data_refs_alignment): Likewise. Use
13073 vect_get_peeling_costs_all_drs to compute the penalty for no
13074 peeling to match up costs.
13076 2017-07-21 Richard Biener <rguenther@suse.de>
13078 PR tree-optimization/81500
13079 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13080 we didn't identify a reduction path.
13082 2017-07-21 Tom de Vries <tom@codesourcery.com>
13083 Cesar Philippidis <cesar@codesourcery.com>
13085 PR gcov-profile/81442
13086 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13089 2017-07-21 Tom de Vries <tom@codesourcery.com>
13092 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13094 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13095 nvptx_override_options_after_change.
13097 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13099 * dwarf2out.c (output_file_names): Avoid double testing for
13100 dwarf_version >= 5.
13102 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13104 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13106 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13108 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13110 (try_crossjump_to_edge): Do not punt on partitioned functions.
13112 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13114 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13115 Put all BBs reachable only via paths crossing cold region to cold
13117 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13119 2016-07-21 Richard Biener <rguenther@suse.de>
13121 PR tree-optimization/81303
13122 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13123 into account prologue and epilogue iterations when raising
13124 min_profitable_iters to sth at least covering one vector iteration.
13126 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13128 * config/arm/arm.c (arm_test_cpu_arch_dat):
13131 2017-07-20 Nathan Sidwell <nathan@acm.org>
13133 Remove TYPE_METHODS.
13134 * tree.h (TYPE_METHODS): Delete.
13135 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13136 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13137 (dbxout_type_methods): Scan TYPE_FIELDS.
13138 (dbxout_type): Don't check TYPE_METHODS here.
13139 * function.c (use_register_for_decl): Always ignore register for
13140 class types when not optimizing.
13141 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13142 * tree.c (free_lang_data_in_type): Stitch out member functions and
13143 templates from TYPE_FIELDS.
13144 (build_distinct_type_copy, verify_type_variant,
13145 verify_type): Member fns are on TYPE_FIELDS.
13146 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13147 * tree-pretty-print.c (dump_generic_node): Likewise.
13149 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13152 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13153 V2TImode and V4TImode.
13154 (ix86_expand_vector_extract): Likewise.
13155 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13156 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13157 (ssescalarmode): Handle V4TImode and V2TImode.
13158 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13159 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13160 (VEXTRACTI128_MODE): New mode iterator.
13161 (splitter for *vec_extractv?ti first element): New.
13162 (VEC_INIT_MODE): New mode iterator.
13163 (vec_init<mode>): Consolidate 3 expanders into one using
13164 VEC_INIT_MODE mode iterator.
13166 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13168 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13169 non_spilled_static_chain_regno_p.
13171 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13173 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13175 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13177 * bb-reorder.c (connect_traces): Allow copying of blocks within
13180 2017-07-20 Richard Biener <rguenther@suse.de>
13182 * gimple.h (gimple_phi_result): Add gphi * overload.
13183 (gimple_phi_result_ptr): Likewise.
13184 (gimple_phi_arg): Likewise. Adjust index assert to only
13185 allow actual argument accesses rather than all slots available
13187 (gimple_phi_arg_def): Add gphi * overload.
13188 * tree-phinodes.c (make_phi_node): Initialize only actual
13190 (resize_phi_node): Clear memory not covered by old node,
13191 do not initialize excess argument slots.
13192 (reserve_phi_args_for_new_edge): Initialize new argument slot
13195 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13197 PR tree-optimization/81388
13199 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13201 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13202 by removing computation of may_be_zero.
13204 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13205 Tom de Vries <tom@codesourcery.com>
13207 PR middle-end/81030
13208 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13209 when gimple level profile disagrees with what RTL expander did.
13211 2017-07-20 Richard Biener <rguenther@suse.de>
13213 PR tree-optimization/61171
13214 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13215 (vect_analyze_stmt): Add slp instance parameter.
13216 (vectorizable_reduction): Likewise.
13217 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13218 (vect_is_simple_reduction): Deal with chains not detected
13219 as SLP reduction chain, specifically not properly associated
13220 chains containing a mix of plus/minus.
13221 (get_reduction_op): Remove.
13222 (get_initial_defs_for_reduction): Simplify, pass in whether
13223 this is a reduction chain, pass in the SLP node for the PHIs.
13224 (vect_create_epilog_for_reduction): Get the SLP instance as
13226 (vectorizable_reduction): Get the SLP instance as arg.
13227 During analysis remember the SLP node with the PHIs in the
13228 instance. Simplify getting at the vectorized reduction PHIs.
13229 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13230 through SLP instance.
13231 (vect_slp_analyze_operations): Likewise.
13232 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13233 (vect_transform_stmt): Likewise.
13235 2017-07-20 Tom de Vries <tom@codesourcery.com>
13237 PR tree-optimization/81489
13238 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13239 read of phi arg location to before loop that modifies phi.
13241 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13243 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13246 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13248 PR middle-end/81331
13249 * except.c (execute): Fix ordering issue.
13251 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13253 PR rtl-optimization/81423
13254 * combine.c (make_compound_operation_int): Don't try to optimize
13255 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13257 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13259 PR rtl-optimization/81423
13260 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13261 with a constant that is -1 in the truncated to mode.
13263 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13265 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13266 (determine_unlikely_bbs): ... here.
13267 * predict.h (propagate_unlikely_bbs_forward): Declare.
13268 * cfgexpand.c (pass_expand::execute): Use it.
13269 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13271 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13272 propagate_unlikely_bbs_forward.
13274 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13276 PR middle-end/81331
13277 * except.c (maybe_add_nop_after_section_switch): New function.
13280 2017-07-19 Tom de Vries <tom@codesourcery.com>
13282 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13284 2017-07-19 Tom de Vries <tom@codesourcery.com>
13286 * gimple.h (gimple_phi_arg): Make assert more strict.
13288 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13290 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13291 * config/rs6000/mmintrin.h: New file.
13292 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13294 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13296 PR tree-optimization/81346
13297 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13299 2017-07-19 Tom de Vries <tom@codesourcery.com>
13301 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13303 2017-07-19 Tom de Vries <tom@codesourcery.com>
13305 * config/nvptx/nvptx-modes.def: Add V2DImode.
13306 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13307 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13308 (nvptx_output_mov_insn): Handle lack of mov.b128.
13309 (nvptx_print_operand): Handle 'H' and 'L' codes.
13310 (nvptx_vector_mode_supported): Allow V2DImode.
13311 (nvptx_preferred_simd_mode): New function.
13312 (nvptx_data_alignment): New function.
13313 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13314 nvptx_preferred_simd_mode.
13315 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13317 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13319 2017-07-19 Tom de Vries <tom@codesourcery.com>
13321 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13322 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13323 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13324 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13325 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13326 (mov<VECIM>_insn): New define_insn.
13327 (define_expand "mov<VECIM>): New define_expand.
13329 2017-07-19 Tom de Vries <tom@codesourcery.com>
13331 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13333 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13335 PR tree-optimization/81346
13336 * fold-const.h (fold_div_compare, range_check_type): Declare.
13337 * fold-const.c (range_check_type): New function.
13338 (build_range_check): Use range_check_type.
13339 (fold_div_compare): No longer static, rewritten into
13340 a match.pd helper function.
13341 (fold_comparison): Don't call fold_div_compare here.
13342 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13343 as helper function.
13345 2017-07-19 Nathan Sidwell <nathan@acm.org>
13347 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13348 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13349 * tree.c (find_decls_types_r, verify_type): Use
13350 TYPE_{MIN,MAX}_VALUE_RAW.
13351 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13352 (hash_tree): Likewise.
13353 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13355 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13358 2017-07-18 Tom de Vries <tom@codesourcery.com>
13360 PR middle-end/81464
13361 * omp-expand.c (expand_omp_for_static_chunk): Handle
13362 equal-argument loop exit phi.
13364 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13367 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13368 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13369 operand 2 predicate.
13370 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13371 operand 2 predicate.
13372 (ror,rol -> rorx splitters): Use const_int_operand as
13373 operand 2 predicate.
13375 2017-06-18 Richard Biener <rguenther@suse.de>
13377 PR tree-optimization/81410
13378 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13379 the gap in the ! slp_perm SLP case after each group.
13381 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13383 PR middle-end/81463
13384 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13387 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13389 PR middle-end/81462
13390 * predict.c (set_even_probabilities): Cleanup; do not affect
13391 probabilities that are already known.
13392 (combine_predictions_for_bb): Call even when count is set.
13394 2017-07-18 Nathan Sidwell <nathan@acm.org>
13396 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13399 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13402 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13403 optimization for loop niter analysis.
13405 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13408 * config/avr/avr.c (avr_optimize_casesi): Don't use
13409 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13411 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13413 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13414 body_cost_vec from _vect_peel_extended_info.
13415 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13416 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13419 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13421 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13423 2017-07-18 Richard Biener <rguenther@suse.de>
13425 PR tree-optimization/80620
13426 PR tree-optimization/81403
13427 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13428 info when re-using a VN table entry.
13430 2017-07-18 Richard Biener <rguenther@suse.de>
13432 PR tree-optimization/81418
13433 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13434 vectype_in. Verify that with lane-reducing reduction operations
13435 we have a single def-use cycle.
13437 2017-07-17 Carl Love <cel@us.ibm.com>
13439 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13441 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13442 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13443 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13444 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13445 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13446 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13447 VMULOSW): New enum "unspec" values.
13448 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13449 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13450 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13451 altivec_vmulosw): New patterns.
13452 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13453 VMULOSW): Add definitions.
13454 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13456 * config/alpha/alpha.c: Include predict.h.
13458 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13460 * tree-vrp.c (compare_assert_loc): Fix comparison function
13461 to return predictable results.
13463 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13465 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13467 (subdi3): Likewise.
13468 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13469 * doc/invoke.texi (mexpand-adddi): Update text.
13471 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13473 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13474 that also clobbers the CC register. The old expand code is moved
13476 (*arc_clzsi2): ... here.
13477 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13478 the CC register. The old expand code is moved to ...
13479 (arc_ctzsi2): ... here.
13481 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13483 * config/arc/arc.opt (mindexed-loads): Use initial value
13484 TARGET_INDEXED_LOADS_DEFAULT.
13485 (mauto-modify-reg): Use initial value
13486 TARGET_AUTO_MODIFY_REG_DEFAULT.
13487 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13488 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13489 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13490 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13492 2017-07-17 Martin Liska <mliska@suse.cz>
13495 * opts.c (finish_options): Do not allow -fgnu-tm
13496 w/ -fsanitize={kernel-,}address. Say sorry.
13498 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13501 * tree-loop-distribution.c (classify_partition): Only assert on
13502 numer of iterations.
13503 (merge_dep_scc_partitions): Delete prameter. Update function call.
13504 (distribute_loop): Remove code handling loop with unknown niters.
13505 (pass_loop_distribution::execute): Skip loop with unknown niters.
13507 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13510 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13511 function sort_partitions_by_post_order.
13513 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13515 PR tree-optimization/81374
13516 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
13517 the max index of basic blocks, rather than number of basic blocks.
13519 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13521 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
13523 (arc_legitimate_pic_operand_p): Likewise.
13524 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
13526 (arc_needs_pcl_p): Likewise.
13527 (arc_legitimate_pc_offset_p): Likewise.
13528 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
13529 function is also used in constrains.md.
13530 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
13531 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
13532 PLUS. Only return true/false in known cases, otherwise assert.
13533 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
13534 is already called in arc_legitimate_constant_p.
13535 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
13537 (LEGITIMATE_PIC_OPERAND_P): Use
13538 arc_raw_symbolic_reference_mentioned_p function.
13539 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
13544 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13545 Andrew Burgess <andrew.burgess@embecosm.com>
13547 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
13548 (arc_return_address_register): New function.
13549 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
13550 (arc_handle_fndecl_attribute): Add naked attribute.
13551 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13552 (TARGET_WARN_FUNC_RETURN): Likewise.
13553 (arc_allocate_stack_slots_for_args): New function.
13554 (arc_warn_func_return): Likewise.
13555 (machine_function): Change type fn_type.
13556 (arc_compute_function_type): Consider new naked function type,
13557 change function return type.
13558 (arc_must_save_register): Adapt to handle new
13559 arc_compute_function_type's return type.
13560 (arc_expand_prologue): Likewise.
13561 (arc_expand_epilogue): Likewise.
13562 (arc_return_address_regs): Delete.
13563 (arc_return_address_register): New function.
13564 (arc_epilogue_uses): Use above function.
13565 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
13566 (arc_function_type): Change encoding, add naked type.
13567 (ARC_INTERRUPT_P): Change to handle the new encoding.
13568 (ARC_FAST_INTERRUPT_P): Likewise.
13569 (ARC_NORMAL_P): Define.
13570 (ARC_NAKED_P): Likewise.
13571 (arc_compute_function_type): Delete prototype.
13572 * config/arc/arc.md (in_ret_delay_slot): Use
13573 arc_return_address_register function.
13574 (simple_return): Likewise.
13575 (p_return_i): Likewise.
13577 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13579 PR tree-optimization/81428
13580 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
13581 can't be built for those types.
13583 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13585 Remove stuff dead since r239246.
13587 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
13588 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
13589 (avr_inform_devices): Remove dead stuff.
13591 2017-07-17 Tamar Christina <tamar.christina@arm.com>
13593 * config/arm/arm_neon.h: Fix softp typo.
13595 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13597 PR tree-optimization/81365
13598 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
13599 aggregate moves onto bb predecessor edges, make sure there are no
13600 loads that could alias the lhs in between the start of bb and the
13603 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13606 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13607 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13608 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13610 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13612 PR tree-optimization/81396
13613 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13614 (init_symbolic_number): Initialize it to 1.
13615 (perform_symbolic_merge): Add n_ops from both operands into the new
13617 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13618 without base_addr as useless if they need more than one operation.
13619 (bswap_replace): Handle !bswap case for NULL base_addr.
13621 2017-07-17 Tom de Vries <tom@codesourcery.com>
13624 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13627 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13629 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13630 conditional builtin define __FIX_LEON3FT_B2BST.
13632 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13634 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13635 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13638 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13640 PR rtl-optimization/81424
13641 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13642 to remove potential trapping from operands if -fnon-call-exceptions.
13644 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13646 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13647 profile_proability for scalling.
13648 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13650 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13652 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13654 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13656 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13657 fixpoint arithmetics.
13659 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13661 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13662 fixpoint arithmetics.
13664 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13666 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13667 fixpoint arithmetics.
13669 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13671 * profile-count.h (profile_probability::from_reg_br_prob_note,
13672 profile_probability::to_reg_br_prob_note): New functions.
13673 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13674 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13675 * predict.c (probability_reliable_p): Update.
13676 (edge_probability_reliable_p): Update.
13677 (br_prob_note_reliable_p): Update.
13678 (invert_br_probabilities): Update.
13679 (add_reg_br_prob_note): New function.
13680 (combine_predictions_for_insn): Update.
13681 * asan.c (asan_clear_shadow): Update.
13682 * cfgbuild.c (compute_outgoing_frequencies): Update.
13683 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13684 (update_br_prob_note): Update.
13685 (rtl_verify_edges): Update.
13686 (purge_dead_edges): Update.
13687 (fixup_reorder_chain): Update.
13688 * emit-rtl.c (try_split): Update.
13689 * ifcvt.c (cond_exec_process_insns): Update.
13690 (cond_exec_process_if_block): Update.
13691 (dead_or_predicable): Update.
13692 * internal-fn.c (expand_addsub_overflow): Update.
13693 (expand_neg_overflow): Update.
13694 (expand_mul_overflow): Update.
13695 * loop-doloop.c (doloop_modify): Update.
13696 * loop-unroll.c (compare_and_jump_seq): Update.
13697 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13698 * predict.h: Update.
13699 * reorg.c (mostly_true_jump): Update.
13701 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13702 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13703 * config/arc/arc.c: (emit_unlikely_jump): Update.
13704 * config/arm/arm.c: (emit_unlikely_jump): Update.
13705 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13706 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13707 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13708 (ix86_print_operand): Update.
13709 (ix86_split_fp_branch): Update.
13710 (predict_jump): Update.
13711 * config/ia64/ia64.c (ia64_print_operand): Update.
13712 * config/mmix/mmix.c (mmix_print_operand): Update.
13713 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13714 (rs6000_expand_split_stack_prologue): Update.
13715 * config/rs6000/rs6000.c: Update.
13716 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13717 (s390_expand_vec_movstr): Update.
13718 (s390_expand_cs_tdsi): Update.
13719 (s390_expand_split_stack_prologue): Update.
13720 * config/sh/sh.c (sh_print_operand): Update.
13721 (expand_cbranchsi4): Update.
13722 (expand_cbranchdi4): Update.
13723 * config/sparc/sparc.c (output_v9branch): Update.
13724 * config/spu/spu.c (get_branch_target): Update.
13725 (ea_load_store_inline): Update.
13726 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13727 * config/tilepro/tilepro.c: Update.
13729 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13731 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13732 (gimplify_save_expr): Likewise.
13734 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13736 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13738 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13740 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13742 * timevar.def (TV_IPA_FNSUMMARY): Define.
13744 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13746 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13747 to back store errata sensitive sequence from being generated.
13748 (sqrtdf2_fix): Likewise.
13750 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13752 * tree-ssa-threadupdate.c (compute_path_counts,
13753 update_joiner_offpath_counts): Use profile_probability.
13755 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13758 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13760 * config/arm/arm-c.c (arm_cpu_builtins): Define
13761 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13763 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13765 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13766 array entries to represent __ieee128 versions of the
13767 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13768 scalar_extract_sig, and scalar_insert_exp built-in functions.
13769 (altivec_resolve_overloaded_builtin): Add special case handling
13770 for the __builtin_scalar_insert_exp function, as represented by
13771 the P9V_BUILTIN_VEC_VSIEDP constant.
13772 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13773 exponent support for __ieee128 argument.
13774 (VSESQP): Add scalar extract signature support for __ieee128
13776 (VSTDCNQP): Add scalar test negative support for __ieee128
13778 (VSIEQP): Add scalar insert exponent support for __int128 argument
13779 with __ieee128 result.
13780 (VSIEQPF): Add scalar insert exponent support for __ieee128
13781 argument with __ieee128 result.
13782 (VSTDCQP): Add scalar test data class support for __ieee128
13784 (VSTDCNQP): Add overload support for scalar test negative with
13785 __ieee128 argument.
13786 (VSTDCQP): Add overload support for scalar test data class
13787 __ieee128 argument.
13788 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13789 UNSPEC_VSX_SXSIGDP.
13790 (UNSPEC_VSX_SIEXPQP): New constant.
13791 (xsxexpqp): New insn for VSX scalar extract exponent quad
13793 (xsxsigqp): New insn for VSX scalar extract significand quad
13795 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13796 precision with floating point argument.
13797 (xststdcqp): New expand for VSX scalar test data class quad
13799 (xststdcnegqp): New expand for VSX scalar test negative quad
13801 (xststdcqp): New insn to match expansions for VSX scalar test data
13802 class quad precision and VSX scalar test negative quad precision.
13803 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13804 special case operand checking to enforce that second operand of
13805 VSX scalar test data class with quad precision argument is a 7-bit
13807 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13808 prototypes and descriptions of __ieee128 versions of
13809 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13810 scalar_test_data_class, and scalar_test_neg built-in functions.
13812 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13814 PR tree-optimization/81162
13815 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13816 replace a negate with an add.
13818 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13820 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13822 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13824 * config/arm/arm-c.c (arm_cpu_builtins): Define
13825 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13827 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13829 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13830 (armv8-r): Set ARM Cortex-R52 as default CPU.
13831 * config/arm/arm-tables.opt: Regenerate.
13832 * config/arm/arm-tune.md: Regenerate.
13833 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13835 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13836 extension for -mcpu=cortex-r52.
13838 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13840 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13841 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13842 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13843 (fp-armv8): Define it as FP_ARMv8 only.
13844 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13845 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13847 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13848 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13849 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13850 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13851 than TARGET_FPU_ARMV8.
13852 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13853 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13854 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13856 * config/arm/neon.md (neon_vrint): Likewise.
13857 (neon_vcvt): Likewise.
13858 (neon_<fmaxmin_op><mode>): Likewise.
13859 (<fmaxmin><mode>3): Likewise.
13860 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13861 * config/arm/predicates.md (arm_cond_move_operator): Check against
13862 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13864 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13866 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13867 to top of function.
13869 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13871 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13872 loop in comment with memset.
13874 2017-07-14 Martin Liska <mliska@suse.cz>
13876 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13877 * dwarf2out.c (is_java): Remove the function.
13878 (output_pubname): Remove usage of the function.
13879 (lower_bound_default): Remove usage of DW_LANG_Java.
13880 (gen_compile_unit_die): Likewise.
13881 * gcc.c: Remove compiler defaults for .java and .zip files.
13882 * gimple-expr.c (remove_suffix): Change as there's no longer
13883 extension than 4-letter one.
13884 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13885 (gimplify_save_expr): Likewise.
13886 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13887 as it's possible even for other languages than Java.
13888 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13889 * lto-opts.c (lto_write_options): Remove reference to Java.
13890 * opts.c (strip_off_ending): Update file extension handling.
13891 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13892 * tree-eh.c (lower_resx): Likewise.
13893 * tree.c (free_lang_data_in_type): Remove dead code.
13894 (find_decls_types_r): Likewise.
13895 (build_common_builtin_nodes): Remove Java from a comment.
13896 (verify_type): Remove dead code.
13897 * varasm.c (assemble_external): Remove Java from a comment.
13899 2017-07-14 Martin Liska <mliska@suse.cz>
13901 * opts.c (finish_options): Add quotes.
13902 (common_handle_option): Likewise.
13904 2017-07-14 Martin Liska <mliska@suse.cz>
13906 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13907 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13908 Remove N_SO_PASCAL.
13909 * dwarf2out.c (lower_bound_default): Do not handle
13911 (gen_compile_unit_die): Likewise.
13912 * gcc.c: Remove default extension binding for GNU Pascal.
13913 * stmt.c: Remove Pascal language from a comment.
13914 * xcoffout.c: Likewise.
13916 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13919 * diagnostic-show-locus.c (fixit_cmp): New function.
13920 (layout::layout): Sort m_fixit_hints.
13921 (column_range::column_range): Assert that the values are valid.
13922 (struct char_span): New struct.
13923 (correction::overwrite): New method.
13924 (struct source_line): New struct.
13925 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13926 calls in terms of classes source_line and char_span, and
13927 correction::overwrite.
13928 (selftest::test_overlapped_fixit_printing_2): New function.
13929 (selftest::diagnostic_show_locus_c_tests): Call it.
13931 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13933 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13934 early if there is no lhs.
13936 2017-07-13 Martin Liska <mliska@suse.cz>
13938 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13939 (gen_reference_type_die): Likewise.
13940 * stor-layout.c: Remove Pascal-related comment.
13942 2017-07-13 Martin Liska <mliska@suse.cz>
13944 * opts.c (finish_options): Add quotes to error messages.
13945 (parse_sanitizer_options): Likewise.
13947 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13949 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13951 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13953 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13955 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13957 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13959 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13961 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13964 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13965 provides the hardware capability bits, define the macro
13966 __BUILTIN_CPU_SUPPORTS__.
13967 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13968 if GLIBC does not provide the hardware capability bits. Add a
13969 gcc_unreachable call if the built-in cpu function is neither
13970 __builtin_cpu_is nor __builtin_cpu_supports.
13971 (rs6000_get_function_versions_dispatcher): Change the warning
13972 that an old GLIBC is used which does not export the capability
13973 bits to be an error.
13974 * doc/extend.texi (target_clones attribute): Document the
13975 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13976 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13977 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13978 the macros defined by GCC if the newer GLIBC is available.
13980 2017-07-12 Jeff Law <law@redhat.com>
13982 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13983 remaining includes slightly.
13984 * config/riscv/riscv-builtins.c: Include profile-count.h.
13986 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13989 * config/avr/avr.c (avr_set_current_function): In diagnostic
13990 messages: Quote keywords and (parts of) identifiers.
13991 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13994 2017-07-12 Carl Love <cel@us.ibm.com>
13996 * config/rs6000/rs6000-c.c: Add support for built-in functions
13997 vector bool char vec_revb (vector bool char);
13998 vector bool short vec_revb (vector short char);
13999 vector bool int vec_revb (vector bool int);
14000 vector bool long long vec_revb (vector bool long long);
14001 * doc/extend.texi: Update the built-in documentation file for the
14002 new built-in functions.
14004 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14006 * config/s390/s390.md: Remove movcc splitter.
14008 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14010 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14011 load/store on condition.
14013 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14016 * config/avr/avr.c (avr_encode_section_info)
14017 [progmem && !TREE_READONLY]: Error if progmem object needs
14020 2017-07-11 Michael Collison <michael.collison@arm.com>
14022 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14025 2017-07-11 Carl Love <cel@us.ibm.com>
14027 * config/rs6000/rs6000-c.c: Add support for builtins
14028 vector unsigned int vec_parity_lsbb (vector signed int);
14029 vector unsigned int vec_parity_lsbb (vector unsigned int);
14030 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14031 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14032 vector unsigned long long vec_parity_lsbb (vector signed long long);
14033 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14034 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14035 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14036 * doc/extend.texi: Update the built-in documentation file for the
14037 new built-in functions.
14039 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14041 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14042 (layout::m_primary_loc): New field.
14043 (layout::layout): Initialize new field. Move location filtering
14044 logic from here to...
14045 (layout::maybe_add_location_range): ...this new method. Add
14046 support for filtering to just the lines already specified by other
14048 (layout::will_show_line_p): New method.
14049 (gcc_rich_location::add_location_if_nearby): New method.
14050 (selftest::test_add_location_if_nearby): New test function.
14051 (selftest::diagnostic_show_locus_c_tests): Call it.
14052 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14055 2017-07-11 Tom de Vries <tom@codesourcery.com>
14057 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14058 (bb_first_real_insn): New function.
14059 (nvptx_single): Add extra initialization of broadcasted condition
14062 2017-07-11 Nathan Sidwell <nathan@acm.org>
14064 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14066 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14068 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14069 Binutils doc as TEXI will mess them up.
14070 * doc/invoke.texi (AVR Options): Same here.
14072 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14074 * config/sparc/sparc.opt (mfix-ut700): New option.
14075 (mfix-gr712rc): Likewise.
14076 (sparc_fix_b2bst): New variable.
14077 * doc/invoke.texi (SPARC options): Document them.
14078 (ARM options): Fix warnings.
14079 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14080 instructions to prevent sequences that can trigger the store-store
14081 errata for certain LEON3FT processors.
14082 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14083 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14084 * config/sparc/sparc.md (fix_b2bst): New attribute.
14085 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14087 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14090 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14092 (*rsqrtsf2_sse): Ditto.
14094 (div<mode>3): Macroize insn from divdf3 and divsf3
14095 using MODEF mode iterator.
14097 2017-07-10 Martin Sebor <msebor@redhat.com>
14099 PR tree-optimization/80397
14100 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14101 instead of testing for equality to INTEGER_TYPE.
14103 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14105 * config.gcc: Remove uclibc from arc target spec.
14107 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14109 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14111 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14114 * lto-wrapper.c (remove_option): New function.
14115 (merge_and_complain): Merge PIC/PIE options more realistically.
14117 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14119 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14123 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14124 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14125 * config.in: Regenerate.
14126 * configure: Regenerate.
14127 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14128 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14129 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14130 (TARGET_GASISR_PROLOGUES): ...target mask.
14131 * common/config/avr/avr-common.c
14132 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14133 Set -mgas-isr-prologues.
14134 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14135 INSERT_PASS_BEFORE for it.
14136 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14137 * config/avr/avr.c (avr_option_override)
14138 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14139 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14140 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14141 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14142 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14144 (make_avr_pass_pre_proep): New function.
14145 (emit_push_sfr) <treg>: Add argument to function and use it
14146 instead of TMP_REG.
14147 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14148 and set machine->gasisr.yes.
14149 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14150 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14151 __gcc_isr.n_pushed to .L__stack_usage.
14152 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14153 (avr_asm_final_postscan_insn): ...this new static function.
14154 * config/avr/avr.h (machine_function)
14155 <is_no_gccisr, use_L__stack_usage>: New fields.
14156 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14157 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14158 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14159 (gasisr, *gasisr): New expander and insn.
14160 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14161 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14162 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14164 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14166 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14169 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14171 Move jump-tables out of .text again.
14174 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14175 (ASM_OUTPUT_ADDR_VEC): New function.
14176 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14177 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14178 INSN_ADDRESSes as asm comment.
14179 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14180 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14181 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14182 * config/avr/avr.md (*tablejump): Adjust comment.
14183 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14184 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14186 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14187 (avr_output_addr_vec): New proto.
14188 (avr_log_t) <insn_addresses>: New field.
14190 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14193 * config/i386/i386.c (ix86_function_arg_advance): Set
14194 outgoing_args_on_stack to true if there are outgoing arguments
14196 (ix86_function_arg): Likewise.
14197 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14198 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14199 * config/i386/i386.h (machine_function): Add
14200 outgoing_args_on_stack.
14202 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14204 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14205 supporting pthreds.
14206 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14208 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14210 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14211 (REAL_H): Remove $(MACHMODE_H).
14212 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14214 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14215 $(MACHMODE_H) and double-int.h.
14216 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14218 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14221 2017-07-07 Andrew Pinski <apinski@cavium.com>
14223 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14224 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14226 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14228 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14229 Add warning if GCC was not configured to link against a GLIBC that
14230 exports the hardware capability bits.
14231 (make_resolver_func): Make resolver function private and not a
14232 COMDAT function. Create the name with clone_function_name instead
14233 of make_unique_name.
14236 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14237 correct operand in doing the split.
14239 2017-07-07 Carl Love <cel@us.ibm.com>
14241 * config/rs6000/rs6000-c: Add support for built-in function
14242 vector unsigned short vec_pack_to_short_fp32 (vector float,
14244 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14245 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14246 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14247 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14248 (convert_4f32_8i16): Add define_expand.
14249 * doc/extend.texi: Update the built-in documentation file for the
14250 new built-in function.
14252 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14254 * config/sparc/m8.md: New file.
14255 * config/sparc/sparc.md: Include m8.md.
14257 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14259 * config/sparc/sparc.opt: New option -mvis4b.
14260 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14261 (sparc_option_override): Handle VIS4B.
14262 (enum sparc_builtins): Define
14263 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14264 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14265 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14266 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14267 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14268 (check_constant_argument): New function.
14269 (sparc_vis_init_builtins): Define builtins
14270 __builtin_vis_dictunpack{8,16,32},
14271 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14272 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14273 __builtin_vis_fpcmpde{8,16,32}shl and
14274 __builtin_vis_fpcmpur{8,16,32}shl.
14275 (sparc_expand_builtin): Check that the constant operands to
14276 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14277 constant and in range.
14278 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14280 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14281 (SPARC_IMM5_P): Likewise.
14282 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14283 (enabled): Handle vis4b.
14284 (UNSPEC_DICTUNPACK): New unspec.
14285 (UNSPEC_FPCMPSHL): Likewise.
14286 (UNSPEC_FPUCMPSHL): Likewise.
14287 (UNSPEC_FPCMPDESHL): Likewise.
14288 (UNSPEC_FPCMPURSHL): Likewise.
14289 (cpu_feature): New CPU feature `vis4b'.
14290 (dictunpack{8,16,32}): New insns.
14291 (FPCSMODE): New mode iterator.
14292 (fpcscond): New code iterator.
14293 (fpcsucond): Likewise.
14294 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14295 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14296 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14297 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14298 * config/sparc/constraints.md: Define constraints `q' for unsigned
14299 2-bit integer constants and `t' for unsigned 5-bit integer
14301 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14303 (imm5_operand_dictunpack16): Likewise.
14304 (imm5_operand_dictunpack32): Likewise.
14305 (imm2_operand): Likewise.
14306 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14307 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14308 ditunpack* and fpcmp*shl builtins.
14310 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14312 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14313 * config.in: Add HAVE_AS_SPARC6 define.
14314 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14316 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14318 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14319 (CPP_CPU_SPEC): Handle m8.
14320 (ASM_CPU_SPEC): Likewise.
14321 * config/sparc/sparc-opts.h (enum processor_type): Add
14323 * config/sparc/sparc.c (m8_costs): New struct.
14324 (sparc_option_override): Handle TARGET_CPU_m8.
14325 (sparc32_initialize_trampoline): Likewise.
14326 (sparc64_initialize_trampoline): Likewise.
14327 (sparc_issue_rate): Likewise.
14328 (sparc_register_move_cost): Likewise.
14329 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14330 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14331 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14332 (CPP_CPU_SPEC): Handle M8.
14333 (ASM_CPU_SPEC): Likewise.
14334 (AS_M8_FLAG): Define.
14335 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14336 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14337 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14339 * configure: Regenerate.
14340 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14343 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14345 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14347 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14348 ("*movdi_insn_sp32"): Do not set v3pipe.
14349 ("*movsi_insn"): Likewise.
14350 ("*movdi_insn_sp64"): Likewise.
14351 ("*movsf_insn"): Likewise.
14352 ("*movdf_insn_sp32"): Likewise.
14353 ("*movdf_insn_sp64"): Likewise.
14354 ("*zero_extendsidi2_insn_sp64"): Likewise.
14355 ("*sign_extendsidi2_insn"): Likewise.
14356 ("*mov<VM32:mode>_insn"): Likewise.
14357 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14358 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14359 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14360 ("<vlop:code><VL:mode>3"): Likewise.
14361 ("*not_<vlop:code><VL:mode>3"): Likewise.
14362 ("*nand<VL:mode>_vis"): Likewise.
14363 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14364 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14365 ("one_cmpl<VL:mode>2"): Likewise.
14366 ("faligndata<VM64:mode>_vis"): Likewise.
14367 ("alignaddrsi_vis"): Likewise.
14368 ("alignaddrdi_vis"): Likweise.
14369 ("alignaddrlsi_vis"): Likewise.
14370 ("alignaddrldi_vis"): Likewise.
14371 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14372 ("bmaskdi_vis"): Likewise.
14373 ("bmasksi_vis"): Likewise.
14374 ("bshuffle<VM64:mode>_vis"): Likewise.
14375 ("cmask8<P:mode>_vis"): Likewise.
14376 ("cmask16<P:mode>_vis"): Likewise.
14377 ("cmask32<P:mode>_vis"): Likewise.
14378 ("pdistn<P:mode>_vis"): Likewise.
14379 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14381 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14383 * config/sparc/sparc.md ("subtype"): New insn attribute.
14384 ("*wrgsr_sp64"): Set insn subtype.
14385 ("*rdgsr_sp64"): Likewise.
14386 ("alignaddrsi_vis"): Likewise.
14387 ("alignaddrdi_vis"): Likewise.
14388 ("alignaddrlsi_vis"): Likewise.
14389 ("alignaddrldi_vis"): Likewise.
14390 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14391 ("fexpand_vis"): Likewise.
14392 ("fpmerge_vis"): Likewise.
14393 ("faligndata<VM64:mode>_vis"): Likewise.
14394 ("bshuffle<VM64:mode>_vis"): Likewise.
14395 ("cmask8<P:mode>_vis"): Likewise.
14396 ("cmask16<P:mode>_vis"): Likewise.
14397 ("cmask32<P:mode>_vis"): Likewise.
14398 ("fchksm16_vis"): Likewise.
14399 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14400 ("fmean16_vis"): Likewise.
14401 ("fp<plusminus_insn>64_vis"): Likewise.
14402 ("<plusminus_insn>v8qi3"): Likewise.
14403 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14404 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14405 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14406 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14407 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14408 ("*movqi_insn"): Likewise.
14409 ("*movhi_insn"): Likewise.
14410 ("*movsi_insn"): Likewise.
14411 ("movsi_pic_gotdata_op"): Likewise.
14412 ("*movdi_insn_sp32"): Likewise.
14413 ("*movdi_insn_sp64"): Likewise.
14414 ("movdi_pic_gotdata_op"): Likewise.
14415 ("*movsf_insn"): Likewise.
14416 ("*movdf_insn_sp32"): Likewise.
14417 ("*movdf_insn_sp64"): Likewise.
14418 ("*zero_extendhisi2_insn"): Likewise.
14419 ("*zero_extendqihi2_insn"): Likewise.
14420 ("*zero_extendqisi2_insn"): Likewise.
14421 ("*zero_extendqidi2_insn"): Likewise.
14422 ("*zero_extendhidi2_insn"): Likewise.
14423 ("*zero_extendsidi2_insn_sp64"): Likewise.
14424 ("ldfsr"): Likewise.
14425 ("prefetch_64"): Likewise.
14426 ("prefetch_32"): Likewise.
14427 ("tie_ld32"): Likewise.
14428 ("tie_ld64"): Likewise.
14429 ("*tldo_ldub_sp32"): Likewise.
14430 ("*tldo_ldub1_sp32"): Likewise.
14431 ("*tldo_ldub2_sp32"): Likewise.
14432 ("*tldo_ldub_sp64"): Likewise.
14433 ("*tldo_ldub1_sp64"): Likewise.
14434 ("*tldo_ldub2_sp64"): Likewise.
14435 ("*tldo_ldub3_sp64"): Likewise.
14436 ("*tldo_lduh_sp32"): Likewise.
14437 ("*tldo_lduh1_sp32"): Likewise.
14438 ("*tldo_lduh_sp64"): Likewise.
14439 ("*tldo_lduh1_sp64"): Likewise.
14440 ("*tldo_lduh2_sp64"): Likewise.
14441 ("*tldo_lduw_sp32"): Likewise.
14442 ("*tldo_lduw_sp64"): Likewise.
14443 ("*tldo_lduw1_sp64"): Likewise.
14444 ("*tldo_ldx_sp64"): Likewise.
14445 ("*mov<VM32:mode>_insn"): Likewise.
14446 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14447 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14449 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14451 * config/sparc/sparc.md ("type"): New insn type viscmp.
14452 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14454 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14455 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14456 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14457 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14459 ("n7_vis_logical_11cycle"): Likewise.
14460 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14461 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14462 * config/sparc/niagara.md ("niag_vis"): Likewise.
14463 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14464 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14466 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14468 * config/sparc/sparc.md: New instruction type `bmask'.
14469 (bmaskdi_vis): Use the `bmask' type.
14470 (bmasksi_vis): Likewise.
14471 * config/sparc/ultra3.md (us3_array): Likewise.
14472 * config/sparc/niagara7.md (n7_array): Likewise.
14473 * config/sparc/niagara4.md (n4_array): Likewise.
14474 * config/sparc/niagara2.md (niag2_vis): Likewise.
14475 (niag3_vis): Likewise.
14476 * config/sparc/niagara.md (niag_vis): Likewise.
14478 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14480 * ipa-comdats.c: Remove optimize check from gate.
14481 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14482 for functions not optimized.
14483 (ipa_fn_summary_read): Skip optimize check.
14484 (ipa_fn_summary_write): Likewise.
14485 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14487 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14489 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14490 (check_callers): Give up on caller which is not optimized.
14491 (inline_small_functions): Likewise.
14492 (ipa_inline): Do not give up when not optimizing.
14493 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14494 away unoptimizes cdtors.
14495 (whole_program_function_and_variable_visibility): Do
14496 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14497 * ipa.c (process_references): Do not check optimize.
14498 (symbol_table::remove_unreachable_nodes): Update optimize check.
14499 (set_writeonly_bit): Update optimize check.
14500 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14501 (pass_ipa_single_use::gate): Remove.
14503 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14505 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14506 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14507 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14508 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14509 permute_load, permute_store, adjust_extract, adjust_splat,
14510 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14511 replace_swap_with_copy, dump_swap_insn_table,
14512 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14513 recombine_lvx_pattern, recombine_stvx_pattern,
14514 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14515 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14516 to file rs6000-p8swap.c.
14517 * config/rs6000/rs6000-p8swap.c: New file.
14518 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
14519 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
14520 and rs6000*-*-* targets.
14522 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14524 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
14526 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14528 * lto-wrapper.c (merge_and_complain): Do not merge
14529 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
14530 fsigned_zeros, ftrapping_math, fwrapv.
14531 (append_compiler_options): Do not track these options.
14532 (append_linker_options): Likewie
14534 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14536 * cgraphunit.c (cgraph_node::finalize_function): When
14537 !flag_toplevel_reorde set no_reorder flag.
14538 (varpool_node::finalize_decl): Likewise.
14539 (symbol_table::compile): Drop no toplevel reorder path.
14541 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14543 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
14544 edges; zero probability is not better than uninitialized.
14546 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14548 * asan.h (asan_sanitize_allocas_p): Declare.
14549 * asan.c (asan_sanitize_allocas_p): New function.
14550 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
14551 (handle_builtin_alloca): Likewise.
14552 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
14553 if !asan_sanitize_allocas_p.
14554 * params.def (asan-instrument-allocas): Add new option.
14555 * params.h (ASAN_PROTECT_ALLOCAS): Define.
14556 * opts.c (common_handle_option): Disable allocas sanitization for
14559 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14561 * asan.c: Include gimple-fold.h.
14562 (get_last_alloca_addr): New function.
14563 (handle_builtin_stackrestore): Likewise.
14564 (handle_builtin_alloca): Likewise.
14565 (asan_emit_allocas_unpoison): Likewise.
14566 (get_mem_refs_of_builtin_call): Add new parameter, remove const
14567 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
14568 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
14569 (instrument_builtin_call): Pass gimple iterator to
14570 get_mem_refs_of_builtin_call.
14571 (last_alloca_addr): New global.
14572 * asan.h (asan_emit_allocas_unpoison): Declare.
14573 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
14574 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
14575 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
14576 if function calls alloca.
14577 * gimple-fold.c (replace_call_with_value): Remove static keyword.
14578 * gimple-fold.h (replace_call_with_value): Declare.
14579 * internal-fn.c: Include asan.h.
14580 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
14581 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
14583 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14585 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
14586 (C_SELFTEST_FLAGS): New.
14587 (CPP_SELFTEST_FLAGS): New.
14588 (SELFTEST_DEPS): New, from deps of s-selftest.
14589 (C_SELFTEST_DEPS): New, from deps of s-selftest.
14590 (CPP_SELFTEST_DEPS): New.
14591 (selftest): Add dependency on s-selftest-c++.
14592 (s-selftest): Rename to...
14593 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
14594 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
14595 than SELFTEST_FLAGS.
14596 (selftest-gdb): Rename to...
14597 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
14599 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
14600 (selftest-valgrind): Rename to...
14601 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14603 (selftest-valgrind): Reintroduce as an alias for
14604 selftest-c-valgrind.
14605 (s-selftest-c++): New.
14606 (selftest-c++-gdb): New.
14607 (selftest-c++-valgrind): New.
14609 2017-07-06 Olivier Hainque <hainque@adacore.com>
14611 * gcc.c (process_command): When deciding if undefined variables
14612 should be ignored when processing specs, accept "gcc -v" as well.
14614 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14616 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14617 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14619 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14621 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14622 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14623 * config/arm/arm-tables.opt: Regenerate.
14624 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14626 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14628 2017-07-06 Carl Love <cel@us.ibm.com>
14630 * ChangeLog: Clean up from mid air collision
14632 2017-07-06 Carl Love <cel@us.ibm.com>
14634 * config/rs6000/rs6000-c.c: Add support for built-in functions
14635 vector signed int vec_subc (vector signed int, vector signed int);
14636 vector signed __int128 vec_subc (vector signed __int128,
14637 vector signed __int128);
14638 vector unsigned __int128 vec_subc (vector unsigned __int128,
14639 vector unsigned __int128);
14640 vector signed int vec_sube (vector signed int, vector signed int,
14641 vector signed int);
14642 vector unsigned int vec_sube (vector unsigned int,
14643 vector unsigned int,
14644 vector unsigned int);
14645 vector signed __int128 vec_sube (vector signed __int128,
14646 vector signed __int128,
14647 vector signed__int128);
14648 vector unsigned __int128 vec_sube (vector unsigned __int128,
14649 vector unsigned __int128,
14650 vector unsigned __int128);
14651 vector signed int vec_subec (vector signed int, vector signed int,
14652 vector signed int);
14653 vector unsigned int vec_subec (vector unsigned int,
14654 vector unsigned int,
14655 vector unsigned int);
14656 vector signed __int128 vec_subec (vector signed __int128,
14657 vector signed __int128,
14658 vector signed__int128);
14659 vector unsigned __int128 vec_subec (vector unsigned __int128,
14660 vector unsigned __int128,
14661 vector unsigned __int128);
14662 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14663 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14664 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14665 BU_ALTIVEC_OVERLOAD_X definitions.
14666 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14667 * doc/extend.texi: Update the built-in documentation file for the new
14668 built-in functions.
14670 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14673 * diagnostic-show-locus.c (layout::layout): Use start and finish
14674 spelling location for the start and finish of each range.
14675 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14676 Add unused aspect param.
14677 * input.c (expand_location_1): Add "aspect" param, and use it
14678 to access the correct part of the location.
14679 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14681 (expand_location_to_spelling_point): Likewise.
14682 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14683 param, and pass it to expand_location_1.
14685 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14687 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14688 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14689 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14690 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14691 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14692 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14693 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14694 _mm_maskz_getmant_ss): New intrinsics.
14695 (__builtin_ia32_getexpss128_mask): Changed to ...
14696 __builtin_ia32_getexpss128_round ... this.
14697 (__builtin_ia32_getexpsd128_mask): Changed to ...
14698 __builtin_ia32_getexpsd128_round ... this.
14699 * config/i386/i386-builtin-types.def
14700 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14701 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14702 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14703 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14704 __builtin_ia32_getmantss_mask_round): New builtins.
14705 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14706 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14707 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14708 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14709 * config/i386/sse.md
14710 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14711 avx512f_sgetexp<mode><mask_scalar_name>
14712 <round_saeonly_scalar_name> ... this.
14713 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14714 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14715 vgetexp<ssescalarmodesuffix>
14716 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14717 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14718 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14719 avx512f_vgetmant<mode><mask_scalar_name>
14720 <round_saeonly_scalar_name> ... this.
14721 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14722 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14723 vgetmant<ssescalarmodesuffix>
14724 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14725 %0<mask_scalar_operand4>, %1, %2
14726 <round_saeonly_scalar_mask_op4>, %3} ... this.
14727 * config/i386/subst.md (mask_scalar_operand4,
14728 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14729 round_saeonly_scalar_nimm_predicate): New subst attributes.
14731 2017-07-06 Julia Koval <julia.koval@intel.com>
14733 * config/i386/i386.c (ix86_erase_embedded_rounding):
14734 Remove code for old rounding pattern.
14736 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14738 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14740 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14742 * doc/sourcebuild.texi (Test Directives, Variants of
14743 dg-require-support): Add documentation for dg-require-stack-check.
14745 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14747 * config/i386/subst.md (mask_scalar, round_scalar,
14748 round_saeonly_scalar): New meta-templates.
14749 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14750 round_scalar_mask_operand3, round_scalar_mask_op3,
14751 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14752 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14753 round_saeonly_scalar_constraint,
14754 round_saeonly_scalar_prefix): New subst attribute.
14755 * config/i386/sse.md
14756 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14757 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14758 <round_scalar_name> ... this.
14759 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14760 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14761 <round_scalar_name> ... this.
14762 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14763 <sse>_vm<code><mode>3<mask_scalar_name>
14764 <round_saeonly_scalar_name> ... this.
14765 (v<plusminus_mnemonic><ssescalarmodesuffix>
14766 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14767 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14768 v<plusminus_mnemonic><ssescalarmodesuffix>
14769 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14770 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14771 (v<multdiv_mnemonic><ssescalarmodesuffix>
14772 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14773 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14774 v<multdiv_mnemonic><ssescalarmodesuffix>
14775 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14776 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14777 (v<maxmin_float><ssescalarmodesuffix>
14778 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14779 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14780 v<maxmin_float><ssescalarmodesuffix>
14781 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14782 %0<mask_scalar_operand3>, %1, %<iptr>2
14783 <round_saeonly_scalar_mask_op3>} ... this.
14785 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14787 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14788 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14790 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14791 Alan Hayward <alan.hayward@arm.com>
14792 David Sherwood <david.sherwood@arm.com>
14794 * combine.c (simplify_if_then_else): Remove "enum" before
14796 * compare-elim.c (can_eliminate_compare): Likewise.
14797 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14799 (aarch64_lookup_simd_builtin_type): Likewise.
14800 (aarch64_simd_builtin_type): Likewise.
14801 (aarch64_init_simd_builtin_types): Likewise.
14802 (aarch64_simd_expand_args): Likewise.
14803 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14805 (aarch64_reverse_mask): Likewise.
14806 (aarch64_simd_emit_reg_reg_move): Likewise.
14807 (aarch64_gen_adjusted_ldpstp): Likewise.
14808 (aarch64_ccmp_mode_to_code): Likewise.
14809 (aarch64_operands_ok_for_ldpstp): Likewise.
14810 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14811 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14813 (aarch64_min_divisions_for_recip_mul): Likewise.
14814 (aarch64_reassociation_width): Likewise.
14815 (aarch64_get_condition_code_1): Likewise.
14816 (aarch64_simd_emit_reg_reg_move): Likewise.
14817 (aarch64_simd_attr_length_rglist): Likewise.
14818 (aarch64_reverse_mask): Likewise.
14819 (aarch64_operands_ok_for_ldpstp): Likewise.
14820 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14821 (aarch64_gen_adjusted_ldpstp): Likewise.
14822 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14824 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14825 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14826 (arm_lookup_simd_builtin_type): Likewise.
14827 (arm_simd_builtin_type): Likewise.
14828 (arm_init_simd_builtin_types): Likewise.
14829 (arm_expand_builtin_args): Likewise.
14830 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14831 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14832 (ft32_setup_incoming_varargs): Likewise.
14833 (ft32_function_arg): Likewise.
14834 (ft32_function_arg_advance): Likewise.
14835 (ft32_pass_by_reference): Likewise.
14836 (ft32_arg_partial_bytes): Likewise.
14837 (ft32_valid_pointer_mode): Likewise.
14838 (ft32_addr_space_pointer_mode): Likewise.
14839 (ft32_addr_space_legitimate_address_p): Likewise.
14840 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14842 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14843 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14844 (iamcu_alignment): Likewise.
14845 (canonicalize_vector_int_perm): Likewise.
14846 (ix86_noce_conversion_profitable_p): Likewise.
14847 (ix86_mpx_bound_mode): Likewise.
14848 (ix86_operands_ok_for_move_multiple): Likewise.
14849 * config/microblaze/microblaze-protos.h
14850 (microblaze_expand_conditional_branch_reg): Likewise.
14851 * config/microblaze/microblaze.c
14852 (microblaze_expand_conditional_branch_reg): Likewise.
14853 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14855 (rs6000_reassociation_width): Likewise.
14856 (rs6000_invalid_binary_op): Likewise.
14857 (fusion_p9_p): Likewise.
14858 (emit_fusion_p9_load): Likewise.
14859 (emit_fusion_p9_store): Likewise.
14860 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14862 (riscv_hard_regno_mode_ok_p): Likewise.
14863 (riscv_address_insns): Likewise.
14864 (riscv_split_symbol): Likewise.
14865 (riscv_legitimize_move): Likewise.
14866 (riscv_function_value): Likewise.
14867 (riscv_hard_regno_nregs): Likewise.
14868 (riscv_expand_builtin): Likewise.
14869 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14870 (riscv_build_integer): Likewise.
14871 (riscv_split_integer): Likewise.
14872 (riscv_legitimate_constant_p): Likewise.
14873 (riscv_cannot_force_const_mem): Likewise.
14874 (riscv_regno_mode_ok_for_base_p): Likewise.
14875 (riscv_valid_base_register_p): Likewise.
14876 (riscv_valid_offset_p): Likewise.
14877 (riscv_valid_lo_sum_p): Likewise.
14878 (riscv_classify_address): Likewise.
14879 (riscv_legitimate_address_p): Likewise.
14880 (riscv_address_insns): Likewise.
14881 (riscv_load_store_insns): Likewise.
14882 (riscv_force_binary): Likewise.
14883 (riscv_split_symbol): Likewise.
14884 (riscv_force_address): Likewise.
14885 (riscv_legitimize_address): Likewise.
14886 (riscv_move_integer): Likewise.
14887 (riscv_legitimize_const_move): Likewise.
14888 (riscv_legitimize_move): Likewise.
14889 (riscv_address_cost): Likewise.
14890 (riscv_subword): Likewise.
14891 (riscv_output_move): Likewise.
14892 (riscv_canonicalize_int_order_test): Likewise.
14893 (riscv_emit_int_order_test): Likewise.
14894 (riscv_function_arg_boundary): Likewise.
14895 (riscv_pass_mode_in_fpr_p): Likewise.
14896 (riscv_pass_fpr_single): Likewise.
14897 (riscv_pass_fpr_pair): Likewise.
14898 (riscv_get_arg_info): Likewise.
14899 (riscv_function_arg): Likewise.
14900 (riscv_function_arg_advance): Likewise.
14901 (riscv_arg_partial_bytes): Likewise.
14902 (riscv_function_value): Likewise.
14903 (riscv_pass_by_reference): Likewise.
14904 (riscv_setup_incoming_varargs): Likewise.
14905 (riscv_print_operand): Likewise.
14906 (riscv_elf_select_rtx_section): Likewise.
14907 (riscv_save_restore_reg): Likewise.
14908 (riscv_for_each_saved_reg): Likewise.
14909 (riscv_register_move_cost): Likewise.
14910 (riscv_hard_regno_mode_ok_p): Likewise.
14911 (riscv_hard_regno_nregs): Likewise.
14912 (riscv_class_max_nregs): Likewise.
14913 (riscv_memory_move_cost): Likewise.
14914 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14915 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14916 (rl78_addr_space_address_mode): Likewise.
14917 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14919 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14920 (rs6000_reassociation_width): Likewise.
14921 (rs6000_invalid_binary_op): Likewise.
14922 (fusion_p9_p): Likewise.
14923 (emit_fusion_p9_load): Likewise.
14924 (emit_fusion_p9_store): Likewise.
14925 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14926 (ok_for_simple_move_operands): Likewise.
14927 (ok_for_simple_move_strict_operands): Likewise.
14928 (ok_for_simple_arith_logic_operands): Likewise.
14929 (visium_legitimize_reload_address): Likewise.
14930 (visium_select_cc_mode): Likewise.
14931 (output_cbranch): Likewise.
14932 (visium_split_double_move): Likewise.
14933 (visium_expand_copysign): Likewise.
14934 (visium_expand_int_cstore): Likewise.
14935 (visium_expand_fp_cstore): Likewise.
14936 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14937 (visium_function_arg): Likewise.
14938 (visium_function_arg_advance): Likewise.
14939 (visium_libcall_value): Likewise.
14940 (visium_setup_incoming_varargs): Likewise.
14941 (visium_legitimate_constant_p): Likewise.
14942 (visium_legitimate_address_p): Likewise.
14943 (visium_legitimize_address): Likewise.
14944 (visium_secondary_reload): Likewise.
14945 (visium_register_move_cost): Likewise.
14946 (visium_memory_move_cost): Likewise.
14947 (prepare_move_operands): Likewise.
14948 (ok_for_simple_move_operands): Likewise.
14949 (ok_for_simple_move_strict_operands): Likewise.
14950 (ok_for_simple_arith_logic_operands): Likewise.
14951 (visium_function_value_1): Likewise.
14952 (rtx_ok_for_offset_p): Likewise.
14953 (visium_legitimize_reload_address): Likewise.
14954 (visium_split_double_move): Likewise.
14955 (visium_expand_copysign): Likewise.
14956 (visium_expand_int_cstore): Likewise.
14957 (visium_expand_fp_cstore): Likewise.
14958 (visium_split_cstore): Likewise.
14959 (visium_select_cc_mode): Likewise.
14960 (visium_split_cbranch): Likewise.
14961 (output_cbranch): Likewise.
14962 (visium_print_operand_address): Likewise.
14963 * expmed.c (flip_storage_order): Likewise.
14964 * expmed.h (emit_cstore): Likewise.
14965 (flip_storage_order): Likewise.
14966 * genrecog.c (validate_pattern): Likewise.
14967 * hsa-gen.c (gen_hsa_addr): Likewise.
14968 * internal-fn.c (expand_arith_overflow): Likewise.
14969 * ira-color.c (allocno_copy_cost_saving): Likewise.
14970 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14971 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14972 (process_invariant_for_inheritance): Likewise.
14973 * lra-eliminations.c (move_plus_up): Likewise.
14974 * omp-low.c (lower_oacc_reductions): Likewise.
14975 * simplify-rtx.c (simplify_subreg): Likewise.
14976 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14977 (TARGET_CHKP_BOUND_MODE): Likewise..
14978 * targhooks.c (default_chkp_bound_mode): Likewise.
14979 (default_setup_incoming_vararg_bounds): Likewise.
14980 * targhooks.h (default_chkp_bound_mode): Likewise.
14981 (default_setup_incoming_vararg_bounds): Likewise.
14982 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14983 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14984 (have_whole_vector_shift): Likewise.
14985 * tree-vect-stmts.c (vectorizable_load): Likewise.
14986 * doc/tm.texi: Regenerate.
14988 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14990 Graceful degrade if Binutils PR21472 is not available.
14993 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14994 .rodata in flash test fails.
14995 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14996 * confgure: Regenerate.
14997 * config.in: Regenerate.
14998 * config/avr/avr.c (avr_asm_named_section)
14999 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15000 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15001 (avr_asm_init_sections): Same.
15003 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15005 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15006 (fma<VH:mode>4_intrinsic): Likewise.
15007 (*fmsub<VCVTF:mode>4): Likewise.
15008 (*fmsub<VH:mode>4_intrinsic): Likewise.
15010 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15013 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15014 Don't depend on "optimize > 0".
15015 (out_movhi_r_mr, out_movqi_mr_r): Same.
15016 (out_movhi_mr_r, out_movqi_r_mr): Same.
15017 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15018 io_address_operand on "optimize > 0".
15020 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15022 * tree-loop-distribution.c: Add general explanantion on the pass.
15023 (generate_loops_for_partition): Mark distributed loop.
15024 (pg_add_dependence_edges): New parameter. Handle alias data
15025 dependence specially and record it in the parameter if asked.
15026 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15027 (init_partition_graph_vertices, add_partition_graph_edge): New.
15028 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15029 (free_partition_graph_vdata, build_partition_graph): New.
15030 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15031 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15032 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15033 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15034 (version_for_distribution_p, finalize_partitions): New.
15035 (distribute_loop): Handle alias data dependence specially. Factor
15036 out loop fusion code as functions and call these functions.
15038 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15040 * tree-loop-distribution.c (classify_partition): New parameter and
15041 better handle reduction statement.
15042 (rdg_build_partitions): Revise comment.
15043 (distribute_loop): Compute statements in all partitions and pass it
15044 to classify_partition.
15046 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15048 * tree-loop-distribution.c (enum partition_type): New.
15049 (struct partition): New field type.
15050 (partition_merge_into): Add parameter. Update partition type.
15051 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15052 (build_rdg_partition_for_vertex): Compute partition type.
15053 (rdg_build_partitions): Dump partition type.
15054 (distribute_loop): Update calls to partition_merge_into.
15056 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15058 * tree-loop-distribution.c (struct ddr_hasher): New.
15059 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15061 (classify_partition): Call get_data_dependence.
15062 (pg_add_dependence_edges): Ditto.
15063 (distribute_loop): Release data dependence hash table.
15065 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15067 * tree-loop-distribution.c (ref_base_address): Delete.
15068 (similar_memory_accesses): Rename ...
15069 (share_memory_accesses): ... to this. Check if partitions access
15070 the same memory reference.
15071 (distribute_loop): Call share_memory_accesses.
15073 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15075 * tree-loop-distribution.c (struct partition): New field recording
15076 its data reference.
15077 (partition_alloc, partition_free): Init and release data refs.
15078 (partition_merge_into): Merge data refs.
15079 (build_rdg_partition_for_vertex): Collect data refs for partition.
15080 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15082 (distribute_loop): Remve data refs from vertice data of partition
15085 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15087 * tree-loop-distribution.c (params.h): Include header file.
15088 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15089 (datarefs_vec): New global var.
15090 (create_rdg_vertices): Use datarefs_vec directly.
15091 (free_rdg): Don't free data references.
15092 (build_rdg): Update use. Don't free data references.
15093 (distribute_loop): Compute global variable for data references.
15094 Bail out if there are too many data references.
15096 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15098 * tree-loop-distribution.c (loop_nest): New global var.
15099 (build_rdg): Use loop directly, rather than loop nest.
15100 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15102 (distribute_loop): Compute global variable loop nest. Update use.
15104 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15106 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15107 (partition_merge_into): New parameter. Dump reason for fusion.
15108 (distribute_loop): Update use of partition_merge_into.
15110 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15112 * tree-loop-distribution.c (bb_top_order_index): New.
15113 (bb_top_order_index_size, bb_top_order_cmp): New.
15114 (stmts_from_loop): Use topological order.
15115 (pass_loop_distribution::execute): Compute and release topological
15116 order for basic blocks.
15118 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15120 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15123 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15125 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15126 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15127 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15128 * internal-fn.def (LOOP_DIST_ALIAS): New.
15129 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15130 (fold_loop_internal_call): ... this.
15131 (vect_loop_dist_alias_call): New function.
15132 (set_uid_loop_bbs): Call fold_loop_internal_call.
15133 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15136 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15139 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15140 Require dead FLAGS_REG at the beginning of a peephole.
15142 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15145 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15146 arguments in the call to __builtin_ia32_sbb_u32.
15147 (_subborrow_u64): Swap _X and _Y arguments in the call to
15148 __builtin_ia32_sbb_u64.
15150 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15153 * tree-vrp.c (compare_assert_loc): Turn into a function template
15154 with stable template parameter. Only test if a->e is NULL,
15155 !a->e == !b->e has been verified already. Use e == NULL or
15156 e != NULL instead of e or ! e tests. If stable is true, don't use
15157 iterative_hash_expr, on the other side allow a or b or both NULL
15158 and sort the NULLs last.
15159 (process_assert_insertions): Sort using compare_assert_loc<false>
15160 instead of compare_assert_loc, later sort using
15161 compare_assert_loc<true> before calling process_assert_insertions_for
15162 in a loop. Use break instead of continue once seen NULL pointer.
15164 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15166 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15167 Cortex-R7 and Cortex-R8 processors.
15169 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15171 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15172 uninitialized while src is not.
15174 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15176 * common/config/arm/arm-common.c: Adjust include path for
15178 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15179 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15180 (arm-cpu-data.h): Likewise.
15181 (arm-cpu-cdata.h): Likewise.
15182 * config/arm/arm-cpu.h: Delete.
15183 * config/arm/arm-cpu-cdata.h: Delete.
15184 * config/arm/arm-cpu-data.h: Delete.
15186 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15188 * config/arm/arm-cpus.in (cortex-a55): New.
15189 (cortex-a75): Likewise.
15190 (cortex-a75.cortex-a55): Likewise.
15191 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15193 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15194 * config/arm/arm-cpu-cdata.h: Regenerate.
15195 * config/arm/arm-cpu-data.h: Regenerate.
15196 * config/arm/arm-cpu.h: Regenerate.
15197 * config/arm/arm-tables.opt: Regenerate.
15198 * config/arm/arm-tune.md: Regenerate.
15200 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15202 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15204 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15206 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15209 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15211 PR tree-optimization/81292
15212 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15213 full_string_p, also call adjust_related_strinfos if the adjustment
15214 is simple, otherwise invalidate related strinfos.
15216 2017-07-04 Martin Liska <mliska@suse.cz>
15219 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15220 newly created variable as DECL_IGNORED_P.
15222 2017-07-04 Martin Liska <mliska@suse.cz>
15225 * ipa-inline.c (inline_small_functions):
15226 Use xstrdup_for_dump.
15228 2017-07-04 Tom de Vries <tom@codesourcery.com>
15230 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15232 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15235 * config/darwin.c (darwin_function_switched_text_sections):
15238 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15240 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15242 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15244 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15246 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15248 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15249 min_profitable_iters, and th as inclusive lower bounds.
15250 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15251 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15252 for min_profitable_iters and min_profitable_estimate.
15253 (vect_transform_loop): Treat th as an inclusive lower bound.
15254 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15256 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15259 * config/darwin.c (darwin_function_switched_text_sections):
15260 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15261 in two pieces, and suppress the use of buf.
15263 2017-07-03 Nathan Sidwell <nathan@acm.org>
15265 * hash-table.h (hash_table_mod1): Fix indentation.
15267 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15269 PR middle-end/81290
15270 * predict.c (force_edge_cold): Be more careful about propagation
15272 * profile-count.h (profile_probability::guessed,
15273 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15275 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15277 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15279 * doc/invoke.texi (rcpc architecture extension): Document it.
15281 2017-07-03 Richard Biener <rguenther@suse.de>
15283 PR tree-optimization/60510
15284 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15285 the scalar reduction PHI and use it.
15286 (vectorizable_reduction): Properly guard the single_defuse_cycle
15287 path for non-SLP reduction chains where we cannot use it.
15288 Rework reduc_def/index and vector type deduction. Rework
15289 vector operand gathering during reduction op code-gen.
15290 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15291 chains dissolve the chain and leave it to non-SLP reduction
15294 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15296 * tree-data-ref.h (dr_alignment): Declare.
15297 * tree-data-ref.c (dr_alignment): New function.
15298 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15299 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15301 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15303 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15305 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15306 and base_misalignment fields.
15307 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15308 * tree-data-ref.c: Include builtins.h.
15309 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15310 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15311 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15312 * tree-vect-data-refs.c: Include tree-cfg.h.
15313 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15314 fields instead of calculating an alignment here.
15315 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15316 innermost_loop_behavior fields.
15318 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15320 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15322 (DR_STEP_ALIGNMENT): New macro.
15323 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15324 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15325 (create_data_ref): Print it.
15326 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15327 to tell whether the step preserves vector (mis)alignment.
15328 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15329 Move the check for an integer step and generalise to all INTEGER_CST.
15330 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15331 Print the outer step alignment.
15333 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15335 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15336 with offset_alignment.
15337 (DR_ALIGNED_TO): Delete.
15338 (DR_OFFSET_ALIGNMENT): New macro.
15339 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15340 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15341 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15342 (create_data_ref): Likewise.
15343 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15344 (vect_analyze_data_refs): Likewise.
15345 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15346 creating dummy innermost behavior.
15348 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15350 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15351 with a "innermost_loop_behavior *" and refeence tree.
15352 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15353 (create_data_ref): Update call accordingly.
15354 * tree-predcom.c (find_looparound_phi): Likewise.
15356 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15358 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15359 fields with dr_wrt_vec_loop.
15360 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15361 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15362 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15363 (vect_dr_behavior): New function.
15364 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15365 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15366 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15367 track whether the step preserves the misalignment.
15368 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15369 Use vect_dr_behavior.
15370 (vect_setup_realignment): Update call accordingly.
15371 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15372 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15373 call to vect_create_addr_base_for_vector_ref.
15374 (vect_create_cond_for_align_checks): Likewise.
15375 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15376 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15377 (vect_recog_mask_conversion_pattern): Likewise.
15378 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15379 (new_stmt_vec_info): Remove redundant zeroing.
15381 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15383 * common/config/arm/arm-common.c (arm_be8_option): New function.
15384 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15385 (ISA_ARMv6): Add isa_bit_be8.
15386 * config/arm/arm.h (arm_be8_option): Add prototype.
15387 (BE8_SPEC_FUNCTION): New define.
15388 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15389 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15390 (mlittle-endian): Similarly.
15391 (mbe8, mbe32): New options.
15392 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15393 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15395 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15397 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15399 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15401 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15402 (cleanup_tree_cfg_bb): Use it.
15403 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15405 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15407 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15410 * loop-doloop.c (add_test): Update profile.
15412 2017-07-03 Martin Liska <mliska@suse.cz>
15415 * sanopt.c (rewrite_usage_of_param): New function.
15416 (sanitize_rewrite_addressable_params): Likewise.
15417 (pass_sanopt::execute): Call rewrite_usage_of_param.
15419 2017-07-03 Richard Biener <rguenther@suse.de>
15421 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15422 back to using VIEW_CONVERT_EXPR.
15424 2017-07-03 Martin Liska <mliska@suse.cz>
15427 * doc/extend.texi: Document when a resolver function is
15428 generated for target_clones.
15430 2017-07-03 Martin Liska <mliska@suse.cz>
15432 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15433 and shadow memory of auto variables which are subject of
15434 use-after-scope sanitization.
15435 (asan_expand_mark_ifn): Add do set only when is_poison.
15437 2016-07-03 Richard Biener <rguenther@suse.de>
15439 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15441 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15442 (vectorizable_reduction): Perform reduction PHI creation when
15443 visiting a reduction PHI and adjust and simplify code generation
15444 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15445 (vect_transform_loop): Visit reduction PHIs.
15446 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15447 defs into the SLP tree.
15448 (vect_build_slp_tree): Reduction defs terminate the recursion.
15449 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15451 (vect_get_vec_defs_for_stmt_copy): Export.
15452 (vect_get_vec_defs): Likewise.
15453 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15455 (vect_get_vec_defs_for_stmt_copy): Declare.
15456 (vect_get_vec_defs): Likewise.
15458 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15460 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15461 parameter with a "loop" parameter and use it instead of the
15462 loop containing DR_STMT. Don't check simple_iv when doing
15463 BB analysis. Describe the two analysis modes in the comment.
15465 2017-07-03 Tom de Vries <tom@codesourcery.com>
15467 PR tree-optimization/69468
15468 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15469 (find_same_succ_bb): Handle ignore_edge_flags.
15471 2017-07-03 Tom de Vries <tom@codesourcery.com>
15473 PR tree-optimization/81192
15474 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15476 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15478 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15480 2017-07-03 Tom de Vries <tom@codesourcery.com>
15482 PR tree-optimization/81192
15483 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15484 BB_SAME_SUCC (bb) == NULL.
15486 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15488 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15491 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15493 * dumpfile.c: Include profile-count.h
15494 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15496 (insert_cond_bb): Update profile.
15497 * tree-cfg.h (insert_cond_bb): Update prototype.
15498 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15499 * tree-dump.c: Do not include tree-cfg.
15501 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15503 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15505 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15507 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15510 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15512 * tree-complex.c (expand_complex_div_wide): update profile.
15514 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15515 Alan Hayward <alan.hayward@arm.com>
15516 David Sherwood <david.sherwood@arm.com>
15518 * Makefile.in (MACHMODE_H): Remove insn-modes.h
15519 (CORETYPES_H): New define.
15520 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
15521 (insn-modes-inline.h, s-modes-inline-h): New rules.
15522 (generated_files): Add insn-modes-inline.h.
15523 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
15524 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
15525 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
15526 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
15527 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
15528 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
15529 (build/gencodes.o, build/genconditions.o): Likewise.
15530 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
15531 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
15532 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
15533 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
15534 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
15535 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
15536 * coretypes.h: Include everything up to real.h for generators.
15537 Include insn-modes.h first. Include wide-int-print.h after
15538 wide-int.h. Include insn-modes-inline.h and then machmode.h.
15539 * machmode.h: Don't include insn-modes.h here.
15540 * function-tests.c: Remove includes of signop.h, machmode.h,
15541 double-int.h and wide-int.h.
15543 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
15545 * optc-save-gen.awk: Likewise.
15546 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
15547 * godump.c: Remove include of wide-int-print.h.
15548 * pretty-print.h: Likewise.
15549 * wide-int-print.cc: Likewise.
15550 * wide-int.cc: Likewise.
15551 * hash-map-tests.c: Remove include of signop.h.
15552 * hash-set-tests.c: Likewise.
15553 * rtl-tests.c: Likewise.
15554 * mkconfig.sh: Remove include of machmode.h.
15555 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
15557 (emit_insn_modes_inline_h): ...this new function. Emit the code
15558 into an insn-modes-inline.h header file, adding appropriate
15559 include guards and end comments.
15560 (emit_insn_modes_c_header): Remove include of machmode.h.
15561 (emit_min_insn_modes_c_header): Include coretypes.h rather than
15563 (main): Handle -i flag and call emit_insn_modes_inline_h when
15566 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15568 * tree-ssa-strlen.c (strinfo): Rename the length field to
15569 nonzero_chars. Add a full_string_p field.
15570 (compare_nonzero_chars, zero_length_string_p): New functions.
15571 (get_addr_stridx): Add an offset_out parameter.
15572 Use compare_nonzero_chars.
15573 (get_stridx): Update accordingly. Use compare_nonzero_chars.
15574 (new_strinfo): Update after above changes to strinfo.
15575 (set_endptr_and_length): Set full_string_p.
15576 (get_string_length): Update after above changes to strinfo.
15577 (unshare_strinfo): Update call to new_strinfo.
15578 (maybe_invalidate): Likewise.
15579 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
15580 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
15581 as a uhwi instead of an shwi. Update after above changes to
15582 strinfo and new_strinfo.
15583 (zero_length_string): Assert that chainsi contains full strings.
15584 Use zero_length_string_p. Update call to new_strinfo.
15585 (adjust_related_strinfos): Update after above changes to strinfo.
15586 Copy full_string_p from origsi.
15587 (adjust_last_stmt): Use zero_length_string_p.
15588 (handle_builtin_strlen): Update after above changes to strinfo and
15589 new_strinfo. Install the lhs as the string length if the previous
15590 entry didn't describe a full string.
15591 (handle_builtin_strchr): Update after above changes to strinfo
15593 (handle_builtin_strcpy): Likewise.
15594 (handle_builtin_strcat): Likewise.
15595 (handle_builtin_malloc): Likewise.
15596 (handle_pointer_plus): Likewise.
15597 (handle_builtin_memcpy): Likewise. Track nonzero characters
15598 that aren't necessarily followed by a nul terminator.
15599 (handle_char_store): Likewise.
15601 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15603 PR tree-optimization/80769
15604 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15605 for malloc and calloc. Document the new invariant that all related
15606 strinfos have delayed lengths or none do.
15607 (verify_related_strinfos): Move earlier in file.
15608 (set_endptr_and_length): New function, split out from...
15609 (get_string_length): ...here. Also set the lengths of related
15611 (zero_length_string): Assert that chainsi has known (rather than
15613 (adjust_related_strinfos): Likewise.
15615 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15617 PR tree-optimization/81136
15618 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15619 assert that two references with the same misalignment have the same
15620 compile-time misalignment if those compile-time misalignments
15623 2017-07-01 Andi Kleen <ak@linux.intel.com>
15625 * print-tree.c (print_node): Print all attributes.
15627 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15629 * cfg.c (scale_bbs_frequencies): New function.
15630 * cfg.h (scale_bbs_frequencies): Declare it.
15631 * cfgloopanal.c (single_likely_exit): Cleanup.
15632 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15634 (scale_loop_profile): Likewise.
15635 (loop_version): Likewise.
15636 (create_empty_loop_on_edge): Update.
15637 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15638 scale_loop_frequencies, scale_loop_profile, loopify,
15639 loop_version): Update prototypes.
15640 * modulo-sched.c (sms_schedule): Update.
15641 * predict.c (unlikely_executed_edge_p): Also check probability.
15642 (probably_never_executed_edge_p): Fix typo.
15643 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15644 * tree-parloops.c (gen_parallel_loop): Update.
15645 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15646 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15647 * tree-ssa-loop-split.c (split_loop): Update.
15648 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15649 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15650 (vect_loop_versioning): Update.
15651 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15653 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15655 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15657 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15659 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15660 to keep profile consistent.
15662 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15664 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15665 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15666 * profile-count.h (max_safe_multiplier): Make unsigned.
15667 (profile_count::guessed_zero): New.
15669 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15671 * bb-reorder.c (fix_up_crossing_landing_pad,
15672 fix_crossing_conditional_branches): Use make_single_succ_edge
15673 to keep profile consistent.
15675 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15677 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15680 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15683 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15684 the right scopes, make sure cond_jump isn't preserved between multiple
15685 iterations. Search for fallthru edge whenever there are 3+ edges and
15686 use find_fallthru_edge for it.
15688 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15690 Patch by Alexander Monakov <amonakov@ispras.ru>
15691 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15692 probabilities consistently.
15694 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15696 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15697 emit_cmp_and_jump_insns.
15699 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15702 * tree-inline.c (expand_call_inline): Combine profile statuses.
15704 2017-06-30 Andrew Pinski <apinski@cavium.com>
15706 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15707 fold_stmt returned true.
15709 2017-06-30 Nathan Sidwell <nathan@acm.org>
15711 * ggc.h (empty_string): Delete.
15712 * cfgexpand.c (expand_asm_stmt): Use plain "".
15713 * optabs.c (expand_asm_memory_barrier): Likewise.
15714 * stringpool.c (empty_string): Delete.
15715 (digit_vector, digit_string): Delete.
15716 (ggc_alloc_string): Use plain "", don't optimize single digit
15717 strings. Use ggc_alloc_atomic.
15719 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15721 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15722 comparison set and one other set, use the cost of the non-comparison
15725 2017-06-30 Nathan Sidwell <nathan@acm.org>
15727 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15730 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15732 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15733 loops. Remove now unneeded calls to gimple_switch_set_label() that
15734 just set removed labels to NULL_TREE.
15736 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15738 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15739 (set_range_info): ...here. Only call set_range_info_raw if domain
15741 (set_nonzero_bits): Call set_range_info_raw.
15742 * tree-ssanames.h (set_range_info_raw): New.
15744 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15747 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15748 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15749 of nonimmediate_operand and <store_mask_constraint> instead of m
15750 for the input operand. For V8FI iterator, always split if input
15751 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15752 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15753 <store_mask_predicate> instead of register_operand and
15754 <store_mask_constraint> instead of v for the input operand. Make
15755 sure both operands aren't MEMs for if not <mask_applied>.
15757 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15759 * lto-wrapper.c (copy_file) Close both file descriptors before
15762 2017-06-30 Martin Liska <mliska@suse.cz>
15765 * multiple_target.c (create_dispatcher_calls): Make ifunc
15766 also for function that don't have calls or are not referenced.
15768 2017-06-30 Richard Biener <rguenther@suse.de>
15770 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15771 analyze the first scalar stmt. Move vector type computation
15772 for the BB case here from ...
15773 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15774 live operation processing in the SLP case properly.
15776 2017-06-30 Richard Biener <rguenther@suse.de>
15778 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15780 2017-06-30 Martin Liska <mliska@suse.cz>
15783 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15784 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15786 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15788 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15791 2017-06-30 Richard Biener <rguenther@suse.de>
15793 PR tree-optimization/81249
15794 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15795 condition reduction result to original scalar type.
15797 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15799 * profile-count.h (enum profile_quality): Fix typos and whitespace
15802 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15804 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15805 type for branch probabilities.
15807 2017-06-29 Julian Brown <julian@codesourcery.com>
15808 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15810 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15811 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15812 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15813 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15815 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15817 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15818 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15819 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15820 CC usage from generic code to here.
15821 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15822 CC usage into the target macros.
15824 2017-06-29 Maya Rashish <coypu@sdf.org>
15826 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15829 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15831 * arm/arm-builtins.c: Include profile-count.h
15832 * except.c (sjlj_emit_function_enter): Use
15833 profile_probability::unlikely.
15835 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15837 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15838 and tocrel_offset be pointer args rather than implicitly using
15840 (legitimate_constant_pool_address_p, rs6000_emit_move,
15841 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15842 tocrel_offset and use in toc_relative_expr_p call.
15843 (print_operand, print_operand_address): Use static tocrel_base_oac
15844 and tocrel_offset_oac.
15845 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15848 2017-06-29 Maya Rashish <coypu@sdf.org>
15850 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15852 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15854 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15855 objects, take into account only the alignment of 'op0' and 'mode1' if
15858 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15860 * ccmp.c (ccmp_tree_comparison_p): New function.
15861 (ccmp_candidate_p): Update to use above function.
15862 (get_compare_parts): New function.
15863 (expand_ccmp_next): Update to use new functions.
15864 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15866 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15867 take mode as argument.
15868 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15869 * expr.c (expand_expr_real_1): Pass mode as argument.
15871 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15873 * combine.c (combine_instructions): Print insns to dump_file, together
15876 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15878 * asan.c (asan_emit_stack_protection): Update.
15879 (create_cond_insert_point): Update.
15880 * auto-profile.c (afdo_propagate_circuit): Update.
15881 * basic-block.h (struct edge_def): Turn probability to
15882 profile_probability.
15883 (EDGE_FREQUENCY): Update.
15884 * bb-reorder.c (find_traces_1_round): Update.
15885 (better_edge_p): Update.
15886 (sanitize_hot_paths): Update.
15887 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15888 (make_single_succ_edge): Update.
15889 (check_bb_profile): Update.
15890 (dump_edge_info): Update.
15891 (update_bb_profile_for_threading): Update.
15892 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15893 probabilitycount to 0.
15894 * cfgbuild.c (compute_outgoing_frequencies): Update.
15895 * cfgcleanup.c (try_forward_edges): Update.
15896 (outgoing_edges_match): Update.
15897 (try_crossjump_to_edge): Update.
15898 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15899 (expand_gimple_tailcall): Update.
15900 (construct_init_block): Use make_single_succ_edge.
15901 (construct_exit_block): Use make_single_succ_edge.
15902 * cfghooks.c (verify_flow_info): Update.
15903 (redirect_edge_succ_nodup): Update.
15904 (split_edge): Update.
15905 (account_profile_record): Update.
15906 * cfgloopanal.c (single_likely_exit): Update.
15907 * cfgloopmanip.c (scale_loop_profile): Update.
15908 (set_zero_probability): Remove.
15909 (duplicate_loop_to_header_edge): Update.
15910 * cfgloopmanip.h (loop_version): Update prototype.
15911 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15912 (force_nonfallthru_and_redirect): Update.
15913 (update_br_prob_note): Update.
15914 (rtl_verify_edges): Update.
15915 (purge_dead_edges): Update.
15916 (rtl_lv_add_condition_to_bb): Update.
15917 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15918 * cgraphunit.c (init_lowered_empty_function): Update.
15919 (cgraph_node::expand_thunk): Update.
15920 * cilk-common.c: Include profile-count.h
15921 * dojump.c (inv): Remove.
15922 (jumpifnot): Update.
15923 (jumpifnot_1): Update.
15924 (do_jump_1): Update.
15926 (do_jump_by_parts_greater_rtx): Update.
15927 (do_compare_rtx_and_jump): Update.
15928 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15929 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15930 * dwarf2cfi.c: Include profile-count.h
15931 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15932 (sjlj_emit_dispatch_table): Likewise.
15933 * explow.c: Include profile-count.h
15934 * expmed.c (emit_store_flag_force): Update.
15935 (do_cmp_and_jump): Update.
15936 * expr.c (compare_by_pieces_d::generate): Update.
15937 (compare_by_pieces_d::finish_mode): Update.
15938 (emit_block_move_via_loop): Update.
15939 (store_expr_with_bounds): Update.
15940 (store_constructor): Update.
15941 (expand_expr_real_2): Update.
15942 (expand_expr_real_1): Update.
15943 * expr.h (try_casesi, try_tablejump): Update prototypes.
15944 * gimple-pretty-print.c (dump_probability): Update.
15945 (dump_profile): New.
15946 (dump_gimple_label): Update.
15947 (dump_gimple_bb_header): Update.
15948 * graph.c (draw_cfg_node_succ_edges): Update.
15949 * hsa-gen.c (convert_switch_statements): Update.
15950 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15951 (find_if_case_1): Update.
15952 (find_if_case_2): Update.
15953 * internal-fn.c (expand_arith_overflow_result_store): Update.
15954 (expand_addsub_overflow): Update.
15955 (expand_neg_overflow): Update.
15956 (expand_mul_overflow): Update.
15957 (expand_vector_ubsan_overflow): Update.
15958 * ipa-cp.c (good_cloning_opportunity_p): Update.
15959 * ipa-split.c (split_function): Use make_single_succ_edge.
15960 * ipa-utils.c (ipa_merge_profiles): Update.
15961 * loop-doloop.c (add_test): Update.
15962 (doloop_modify): Update.
15963 * loop-unroll.c (compare_and_jump_seq): Update.
15964 (unroll_loop_runtime_iterations): Update.
15965 * lra-constraints.c (lra_inheritance): Update.
15966 * lto-streamer-in.c (input_cfg): Update.
15967 * lto-streamer-out.c (output_cfg): Update.
15968 * mcf.c (adjust_cfg_counts): Update.
15969 * modulo-sched.c (sms_schedule): Update.
15970 * omp-expand.c (expand_omp_for_init_counts): Update.
15971 (extract_omp_for_update_vars): Update.
15972 (expand_omp_ordered_sink): Update.
15973 (expand_omp_for_ordered_loops): Update.
15974 (expand_omp_for_generic): Update.
15975 (expand_omp_for_static_nochunk): Update.
15976 (expand_omp_for_static_chunk): Update.
15977 (expand_cilk_for): Update.
15978 (expand_omp_simd): Update.
15979 (expand_omp_taskloop_for_outer): Update.
15980 (expand_omp_taskloop_for_inner): Update.
15981 * omp-simd-clone.c (simd_clone_adjust): Update.
15982 * optabs.c (expand_doubleword_shift): Update.
15983 (expand_abs): Update.
15984 (emit_cmp_and_jump_insn_1): Update.
15985 (expand_compare_and_swap_loop): Update.
15986 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15987 * predict.c (predictable_edge_p): Update.
15988 (edge_probability_reliable_p): Update.
15989 (set_even_probabilities): Update.
15990 (combine_predictions_for_insn): Update.
15991 (combine_predictions_for_bb): Update.
15992 (propagate_freq): Update.
15993 (estimate_bb_frequencies): Update.
15994 (force_edge_cold): Update.
15995 * profile-count.c (profile_count::dump): Add missing space into dump.
15996 (profile_count::debug): Add newline.
15997 (profile_count::differs_from_p): Explicitly convert to unsigned.
15998 (profile_count::stream_in): Update.
15999 (profile_probability::dump): New member function.
16000 (profile_probability::debug): New member function.
16001 (profile_probability::differs_from_p): New member function.
16002 (profile_probability::differs_lot_from_p): New member function.
16003 (profile_probability::stream_in): New member function.
16004 (profile_probability::stream_out): New member function.
16005 * profile-count.h (profile_count_quality): Rename to ...
16006 (profile_quality): ... this one.
16007 (profile_probability): New.
16008 (profile_count): Update.
16009 * profile.c (compute_branch_probabilities): Update.
16010 * recog.c (peep2_attempt): Update.
16011 * sched-ebb.c (schedule_ebbs): Update.
16012 * sched-rgn.c (find_single_block_region): Update.
16013 (compute_dom_prob_ps): Update.
16014 (schedule_region): Update.
16015 * sel-sched-ir.c (compute_succs_info): Update.
16016 * stmt.c (struct case_node): Update.
16017 (do_jump_if_equal): Update.
16018 (get_outgoing_edge_probs): Update.
16019 (conditional_probability): Update.
16020 (emit_case_dispatch_table): Update.
16021 (expand_case): Update.
16022 (expand_sjlj_dispatch_table): Update.
16023 (emit_case_nodes): Update.
16024 * targhooks.c: Update.
16025 * tracer.c (better_p): Update.
16026 (find_best_successor): Update.
16027 * trans-mem.c (expand_transaction): Update.
16028 * tree-call-cdce.c: Update.
16029 * tree-cfg.c (gimple_split_edge): Upate.
16030 (move_sese_region_to_fn): Upate.
16031 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16032 * tree-eh.c (lower_resx): Upate.
16033 (cleanup_empty_eh_move_lp): Upate.
16034 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16035 * tree-inline.c (copy_edges_for_bb): Update.
16036 (copy_cfg_body): Update.
16037 * tree-parloops.c (gen_parallel_loop): Update.
16038 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16039 (gimple_gen_time_profiler): Update.
16040 * tree-ssa-dce.c (remove_dead_stmt): Update.
16041 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16042 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16043 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16044 (unloop_loops): Update.
16045 (try_peel_loop): Update.
16046 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16047 * tree-ssa-loop-split.c (connect_loops): Update.
16048 (split_loop): Update.
16049 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16050 (hoist_guard): Update.
16051 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16052 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16053 (value_replacement): Update.
16054 * tree-ssa-reassoc.c (branch_fixup): Update.
16055 * tree-ssa-tail-merge.c (replace_block_by): Update.
16056 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16057 (create_edge_and_update_destination_phis): Update.
16058 (compute_path_counts): Update.
16059 (recompute_probabilities): Update.
16060 (update_joiner_offpath_counts): Update.
16061 (freqs_to_counts_path): Update.
16062 (duplicate_thread_path): Update.
16063 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16064 (struct switch_conv_info): Update.
16065 (gen_inbound_check): Update.
16066 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16067 (vect_do_peeling): Update.
16068 (vect_loop_versioning): Update.
16069 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16070 (optimize_mask_stores): Update.
16071 * ubsan.c (ubsan_expand_null_ifn): Update.
16072 * value-prof.c (gimple_divmod_fixed_value): Update.
16073 (gimple_divmod_fixed_value_transform): Update.
16074 (gimple_mod_pow2): Update.
16075 (gimple_mod_pow2_value_transform): Update.
16076 (gimple_mod_subtract): Update.
16077 (gimple_mod_subtract_transform): Update.
16078 (gimple_ic): Update.
16079 (gimple_stringop_fixed_value): Update.
16080 (gimple_stringops_transform): Update.
16081 * value-prof.h: Update.
16083 2017-06-29 Carl Love <cel@us.ibm.com>
16085 * config/rs6000/rs6000-c.c: Add support for built-in functions
16086 vector signed int vec_signed (vector float);
16087 vector signed long long vec_signed (vector double);
16088 vector signed int vec_signed2 (vector double, vector double);
16089 vector signed int vec_signede (vector double);
16090 vector signed int vec_signedo (vector double);
16091 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16092 instruction generator.
16093 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16094 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16095 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16097 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16098 vunsignede_v2df): Add define_expands.
16099 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16100 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16101 VEC_UNSIGNEDO): Add definitions.
16102 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16103 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16104 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16105 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16106 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16107 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16108 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16109 vec_unsignede, vec_unsignedo): Add builtin defines.
16110 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16112 * doc/extend.texi: Update the built-in documentation file for the
16113 new built-in functions.
16115 2017-06-29 Richard Biener <rguenther@suse.de>
16117 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16118 reduction chains to LOOP_VINFO_REDUCTIONS.
16119 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16120 SLP reductions after processing reduction chains.
16122 2017-06-29 Nathan Sidwell <nathan@acm.org>
16124 * builtins.c (fold_builtin_FUNCTION): Use
16125 lang_hooks.decl_printable_name.
16127 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16129 PR middle-end/81194
16130 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16131 with only one label.
16132 * stmt.c (expand_case): Assert NCASES is greater than one.
16134 2017-06-29 Richard Biener <rguenther@suse.de>
16136 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16138 (group_case_labels): Likewise.
16139 (find_taken_edge): Push sanity checking on val to workers...
16140 (find_taken_edge_cond_expr): ... here
16141 (find_taken_edge_switch_expr): ... and here, handle cases
16142 with just a default label.
16143 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16144 (group_case_labels): Likewise.
16145 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16146 group_case_labels does anything cleanup the CFG again.
16148 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16150 PR tree-optimization/81196
16151 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16152 exit condition comparing two IVs.
16154 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16156 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16157 profile to the dummy entry at the end of the list of architectures.
16158 * config/arm/arm-cpu-cdata.h: Regenerated.
16160 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16161 Michael Collison <michael.collison@arm.com>
16164 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16166 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16167 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16168 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16169 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16170 of shift when the shift amount is masked with constant equal to
16171 the size of the mode.
16172 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16175 2017-06-29 Martin Liska <mliska@suse.cz>
16177 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16179 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16182 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16185 (edge_growth_cache_entry::edge_growth_cache_entry): New
16187 (reset_edge_growth_cache): Update to use constructor.
16189 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16191 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16192 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16193 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16195 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16197 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16198 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16200 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16202 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16203 (*-linux-uclibc*): Add t-uclibc tmake_file.
16204 * config/t-musl: New.
16205 * config/t-uclibc: New.
16207 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16209 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16211 (gen_comm_data): Emit architectural setting of arch_prof.
16212 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16214 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16215 (armv8-m.base, armv8-m.main): Likewise.
16216 * arm-protos.h (arm_build_target): Add profile field.
16217 (arch_option): Likewise.
16218 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16220 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16221 arm_active_target.profile.
16223 2017-06-28 Richard Biener <rguenther@suse.de>
16225 PR middle-end/81227
16226 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16227 TYPE_OVERFLOW_WRAPS.
16228 * match.pd (negate_expr_p): Likewise.
16229 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16230 fold_build2, not fold_binary.
16232 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16234 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16235 Convert memory address to Pmode.
16236 (aarch64_print_operand): Assert MEM operands are always Pmode.
16238 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16241 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16242 Remove redundant if.
16243 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16244 * config/arm/aarch-common-protos.h
16245 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16246 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16248 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16251 * multiple_target.c (create_dispatcher_calls): Set the default
16252 clone to be static, not public.
16254 2017-06-28 Richard Biener <rguenther@suse.de>
16256 * tree-vect-loop.c (vectorizable_reduction): Move special
16257 cond reduction IV var creation ...
16258 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16259 parameter. Use STMT_VINFO_VECTYPE.
16260 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16263 2017-06-28 Martin Liska <mliska@suse.cz>
16266 * ipa-visibility.c (non_local_p): Handle visibility.
16268 2017-06-28 Martin Liska <mliska@suse.cz>
16271 * common.opt: Add IntegerRange to various options.
16272 * opt-functions.awk (integer_range_info): New function.
16273 * optc-gen.awk: Add integer_range_info to cl_options struct.
16274 * opts-common.c (decode_cmdline_option): Handle
16275 CL_ERR_INT_RANGE_ARG.
16276 (cmdline_handle_error): Likewise.
16277 * opts.c (print_filtered_help): Show valid interval in
16278 when --help is provided.
16279 * opts.h (struct cl_option): Add range_min and range_max fields.
16280 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16282 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16284 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16285 (x * C EQ/NE y * C): New transformation.
16287 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16289 * genmultilib (combination_space): Accept '+' in option names.
16291 2017-06-28 Martin Liska <mliska@suse.cz>
16294 * asan.c (instrument_derefs): Bail out inner references
16295 that are hard register variables.
16297 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16300 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16301 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16303 2017-06-28 Richard Biener <rguenther@suse.de>
16305 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16306 (vect_get_slp_defs): Adjust.
16307 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16309 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16311 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16312 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16313 (vectorizable_reduction): Adjust.
16314 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16316 (vect_get_slp_defs): Likewise.
16317 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16318 (vectorizable_bswap): Adjust.
16319 (vectorizable_call): Likewise.
16320 (vectorizable_conversion): Likewise.
16321 (vectorizable_assignment): Likewise.
16322 (vectorizable_shift): Likewise.
16323 (vectorizable_operation): Likewise.
16324 (vectorizable_store): Likewise.
16325 (vectorizable_condition): Likewise.
16326 (vectorizable_comparison): Likewise.
16328 2017-06-28 Michael Collison <michael.collison@arm.com>
16331 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16333 (arm_gen_movmemqi): Removed unused variable 'i'.
16334 Convert 'for' loop into 'while' loop.
16335 (arm_expand_prologue): Remove last unnecessary set of insn.
16336 (thumb_pop): Remove unused variable 'pushed_words'.
16337 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16339 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16341 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16342 * config/s390/s390-protos.h: Add prototype of
16343 s390_rel_address_ok_p.
16344 * config/s390/s390.c (s390_got_symbol): New function.
16345 (s390_rel_address_ok_p): New function.
16346 (legitimize_pic_address): Use s390_rel_address_ok_p.
16347 (s390_load_got): Use s390_got_symbol.
16348 (s390_option_override): Issue error if
16349 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16350 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16352 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16354 2017-06-27 Andrew Pinski <apinski@cavium.com>
16356 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16357 (X * copysign (1.0, X)): New pattern.
16358 (X * copysign (1.0, -X)): New pattern.
16359 (copysign (-1.0, CST)): New pattern.
16361 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16363 * genmultilib (combination_space): Remove variable.
16364 Validate reuse rules against regular expression for any sequence
16365 of multilib options in any order.
16367 2017-06-27 Michael Collison <michael.collison@arm.com>
16369 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16370 call aarch64_split_simd_combine.
16371 * (aarch64_combine_internal<mode>): Delete pattern.
16372 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16373 Allow register and subreg operands.
16375 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16377 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16378 specific need, just fallback on defaults.
16379 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16381 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16382 Olivier Hainque <hainque@adacore.com>
16384 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16386 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16387 targets. Pick a default if no particular attempt applied.
16388 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16391 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16393 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16394 (x86_64-wrs-vxworks7): Likewise.
16396 2017-06-27 Marek Polacek <polacek@redhat.com>
16399 * ubsan.c (instrument_null): Check get_base_address's result for null.
16401 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16403 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16405 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16407 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16408 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16409 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16410 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16411 New function types.
16412 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16413 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16414 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16415 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16416 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16417 BUILT_IN_FEUPDATEENV): New builtins.
16418 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16419 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16420 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16421 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16423 (builtin_structptr_types): Adjust size.
16424 * tree.c (builtin_structptr_types): Add four entries.
16426 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16427 Olivier Hainque <hainque@adacore.com>
16429 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16430 (TLS_SYM): New local macro, forcing reference to __tls__ on
16431 link command lines for VxWorks 7 RTPs, triggering initialization
16433 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16434 OS features TLS support, true for RTPs on VxWorks 7.
16435 * config/vxworks.c (vxworks_override_options): Setup emutls
16438 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16440 * predict.c (test_prediction_value_range): Use -1U instead of -1
16441 to avoid narrowing conversion warning.
16442 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16443 to avoid narrowing conversion warning.
16444 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16446 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16448 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16450 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16451 64bit configurations.
16452 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16453 (SIZE_TYPE): Likewise.
16454 * config/vxworks.c (vxworks_emutls_var_fields): Use
16455 long_unsigned_type_node instead of unsigned_type_node as the offset
16456 field type, which is "pointer" mode in emutls.c.
16458 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16461 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16463 PR middle-end/81207
16464 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16465 gimple_vuse copying separately from gimple_vdef copying.
16467 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16469 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16472 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16473 Olivier Hainque <hainque@adacore.com>
16475 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16476 for all vxworks7 targets.
16477 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16478 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16479 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16480 variations for VX6/VX7 and 32/64bits later on in ...
16481 (VXWORKS_LIB_SPEC): Leverage new macros.
16482 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16483 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16485 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16487 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16488 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16490 2017-06-26 Carl Love <cel@us.ibm.com>
16492 * config/rs6000/rs6000-c.c: Add support for built-in functions
16493 vector bool char vec_reve (vector bool char);
16494 vector signed char vec_reve (vector signed char);
16495 vector unsigned char vec_reve (vector unsigned char);
16496 vector bool int vec_reve (vector bool int);
16497 vector signed int vec_reve (vector signed int);
16498 vector unsigned int vec_reve (vector unsigned int);
16499 vector bool long long vec_reve (vector bool long long);
16500 vector signed long long vec_reve (vector signed long long);
16501 vector unsigned long long vec_reve (vector unsigned long long);
16502 vector bool short vec_reve (vector bool short);
16503 vector signed short vec_reve (vector signed short);
16504 vector double vec_reve (vector double);
16505 vector float vec_reve (vector float);
16506 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16507 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16508 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16509 (altivec_vreve): New pattern.
16510 * config/rs6000/altivec.h (vec_reve): New define.
16511 * doc/extend.texi (vec_rev): Update the built-in documentation file
16512 for the new built-in functions.
16514 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16516 PR tree-optimization/71815
16517 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
16519 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
16521 (slsr_process_phi): Likewise.
16522 (replace_uncond_cands_and_profitable_phis): Don't replace a
16523 multiply candidate with a stride of 1 (copy or cast).
16524 (phi_incr_cost): Call uses_consumed_by_stmt rather than
16526 (lowest_cost_path): Likewise.
16527 (total_savings): Likewise.
16529 2017-06-26 Richard Biener <rguenther@suse.de>
16532 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
16533 Use def_builtin_pure for all gather builtins.
16535 2017-06-26 Richard Biener <rguenther@suse.de>
16537 PR tree-optimization/81203
16538 * tree-tailcall.c (find_tail_calls): Do not move stmts into
16539 non-dominating BBs.
16541 2017-06-26 Marek Polacek <polacek@redhat.com>
16544 * doc/invoke.texi: Document -Wmultistatement-macros.
16546 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
16548 * doc/sourcebuild.texi (ARM-specific attributes): Document new
16549 arm_neon_ok_no_float_abi effective target.
16551 2017-06-26 Richard Biener <rguenther@suse.de>
16553 PR tree-optimization/80928
16554 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
16555 (copy_bbs): Set BB_DUPLICATED flag early.
16556 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
16558 (execute_on_shrinking_pred): Likewise.
16559 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
16560 BB_DUPLICATED blocks.
16561 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
16562 iterate over all PHIs considering removal of *gsi.
16564 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
16566 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
16569 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16571 * config/rs6000/rs6000-string.c: (expand_block_clear,
16572 do_load_for_compare, select_block_compare_mode,
16573 compute_current_alignment, expand_block_compare,
16574 expand_strncmp_align_check, expand_strn_compare,
16575 expand_block_move, rs6000_output_load_multiple)
16576 Move functions related to string/block move/compare
16577 to a separate file.
16578 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
16579 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
16580 for this function which is now used in two files.
16581 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
16582 * config.gcc: Add rs6000-string.o to extra_objs for
16583 targets powerpc*-*-* and rs6000*-*-*.
16585 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
16588 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
16589 32-bit, since indexed is not valid for DImode.
16590 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
16591 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
16592 (define_peephole2 for Altivec d-form load): Add 32-bit support.
16593 (define_peephole2 for Altivec d-form store): Likewise.
16596 * multiple_target.c (create_dispatcher_calls): Only create the
16597 dispatcher call if the function is the default clone of a
16598 versioned function.
16600 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
16602 PR middle-end/80902
16603 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16604 a call, force the call to not be a tail call.
16606 2017-06-23 Jeff Law <law@redhat.com>
16608 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16611 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16613 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16614 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16615 handling for early expansion of vector shifts (sl,sr,sra,rl).
16616 (builtin_function_type): Add vector shift right instructions
16617 to the unsigned argument list.
16619 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16621 rtl-optimizatoin/79286
16622 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16623 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16624 trap. PIC register plus a const unspec without offset can never trap.
16626 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16628 * tree.h (builtin_structptr_type): New type.
16629 (builtin_structptr_types): Declare new array.
16630 * tree.c (builtin_structptr_types): New array.
16631 (free_lang_data, build_common_tree_nodes): Use it.
16633 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16636 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16639 2017-06-22 Matt Turner <mattst88@gmail.com>
16641 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16642 Lake models to skylake case. Assume skylake for unknown
16643 models with clflushopt.
16645 2017-06-22 Jeff Law <law@redhat.com>
16647 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16648 frame sizes that do not satisfy aarch64_uimm12_shift.
16650 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16652 * profile-count.h (apply_probability,
16653 apply_scale, probability_in): Fix checks for zero.
16655 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16657 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16658 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16660 2016-06-22 Richard Biener <rguenther@suse.de>
16662 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16663 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16664 REDUC_MAX_EXPR support.
16665 (vectorizable_reduction): Likewise.
16666 (vect_create_epilog_for_reduction): Likewise.
16668 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16670 * match.pd (A / (1 << B) -> A >> B): New.
16671 * generic-match-head.c: Include optabs-tree.h.
16672 * gimple-match-head.c: Likewise.
16673 * optabs-tree.h (target_supports_op_p): New.
16674 * optabs-tree.c (target_supports_op_p): New.
16676 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16678 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16679 $gcc_cv_ld --help output.
16680 (gcc_cv_ld_demangle): Likewise.
16681 (gcc_cv_ld_eh_frame_hdr): Likewise.
16682 (gcc_cv_ld_pie): Likewise.
16683 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16684 (gcc_cv_ld_buildid): Likewise.
16685 (gcc_cv_ld_sysroot): Likewise.
16686 (ld_bndplt_support): Likewise.
16687 (ld_pushpopstate_support): Likewise.
16688 * configure: Regenerate.
16689 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16691 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16694 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16695 operands indexes to avoid gap between operands and match_dups.
16697 2017-06-21 Andrew Pinski <apinski@cavium.com>
16699 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16700 Increment Arith_shift and Arith_shift_reg by 1.
16701 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16703 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16704 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16705 (aarch64_strip_extend): Add new argument and test for it.
16706 (aarch64_cheap_mult_shift_p): New function.
16707 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16708 add a cost if it is true.
16709 Update calls to aarch64_strip_extend.
16710 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16712 2017-06-21 Andrew Pinski <apinski@cavium.com>
16714 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16716 (thunderxt88): Likewise.
16717 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16718 (thunderx_prefetch_tune): New variable.
16719 (thunderx2t99_prefetch_tune): Update for the correct values.
16720 (thunderxt88_tunings): New variable.
16721 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16722 generic_prefetch_tune.
16723 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16725 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16727 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16728 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16729 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16730 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16731 (aarch64_atomic_cas<mode>, GPI): Likewise.
16733 2017-06-21 Martin Liska <mliska@suse.cz>
16735 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16736 statements on cold and hot labels.
16737 * predict.c (tree_estimate_probability_bb): Remove the
16738 prediction from this place.
16740 2017-06-21 Martin Liska <mliska@suse.cz>
16742 PR tree-optimization/79489
16743 * gimplify.c (maybe_add_early_return_predict_stmt): New
16745 (gimplify_return_expr): Call the function.
16746 * predict.c (tree_estimate_probability_bb): Remove handling
16748 * predict.def: Update comment about early return predictor.
16749 * gimple-predict.h (is_gimple_predict): New function.
16750 * predict.def: Change default value of early return to 66.
16751 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16753 * passes.def: Put pass_strip_predict_hints to the beginning of
16756 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16758 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16759 FUNCTION_DECL declarations.
16760 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16762 (dwaf2out_decl): Likewise.
16763 * godump.c (go_early_global_decl): Skip call to the real debug hook
16764 for FUNCTION_DECL declarations.
16765 * passes.c (rest_of_decl_compilation): Skip call to the
16766 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16767 -fdump-go-spec is passed.
16769 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16771 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16772 Reorder for compactness.
16773 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16774 (def_builtin_pure, def_builtin_pure2): New functions.
16775 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16777 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16779 * match.pd (nop_convert): New predicate.
16780 ((A +- CST1) +- CST2): Allow some NOP conversions.
16782 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16785 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16786 with ctors/dtors if GOVD_SHARED is set.
16788 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16790 * config/aarch64/aarch64.md (movti_aarch64):
16791 Emit mov rather than orr.
16792 (movtf_aarch64): Likewise.
16793 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16794 Emit mov rather than orr.
16796 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16798 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16799 Swap alternatives, make integer dup more expensive.
16801 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16803 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16804 Return true for non-tls symbols.
16806 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16808 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16809 (cortex-a75): Likewise.
16810 (cortex-a75.cortex-a55): Likewise.
16811 * config/aarch64/aarch64-tune.md: Regenerate.
16812 * doc/invoke.texi (-mtune): Document new values for -mtune.
16814 2017-06-21 Tom de Vries <tom@codesourcery.com>
16816 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16817 stack_size feature.
16818 (Effective-Target Keywords, Other attributes): Suggest using
16819 dg-add-options stack_size feature to get stack limit in stack_size
16820 effective target documentation.
16822 2017-06-21 Julian Brown <julian@codesourcery.com>
16823 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16825 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16826 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16827 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16829 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16830 attribute type list for neon_multiply.
16831 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16832 attribute type list for neon_multiply.
16833 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16834 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16835 attribute type list for neon_multiply.
16836 * config/arm/types.md (crypto_pmull): Add.
16837 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16838 attribute type list.
16840 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16842 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16845 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16847 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16848 to make sure not to dereference a NULL cost_classes_ptr pointer.
16850 2017-06-20 Carl Love <cel@us.ibm.com>
16852 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16853 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16854 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16855 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16856 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16857 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16858 VMULOSW): New enum "unspec" values.
16859 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16860 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16861 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16862 altivec_vmulosw): New patterns.
16863 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16864 VMULOSW): Add definitions.
16866 2017-06-20 Julia Koval <julia.koval@intel.com>
16868 * config/i386/i386.c: Fix rounding expand for new pattern.
16869 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16871 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16873 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16874 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16876 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16878 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16881 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16883 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16884 architecture, then by implementer ID.
16885 * config/aarch64/aarch64-tune.md: Regenerate.
16887 2017-06-20 Richard Biener <rguenther@suse.de>
16889 PR middle-end/81097
16890 * fold-const.c (split_tree): Fold to type before negating.
16892 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16894 * diagnostic-show-locus.c
16895 (selftest::test_fixit_deletion_affecting_newline): New function.
16896 (selftest::diagnostic_show_locus_c_tests): Call it.
16898 2017-06-20 Andreas Schwab <schwab@suse.de>
16901 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16904 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16906 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16907 __ARM_FEATURE_COPROC according to support.
16909 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16911 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16912 Rewritten to avoid overflow for > 32-bit pointers.
16915 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16916 by removing enum keyword.
16917 (ubsan_type_descriptor): Likewise. Formatting fix.
16920 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16921 splitter): Require TARGET_SSE2 in the condition.
16923 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16926 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16927 for doing vector set of SFmode on ISA 3.0.
16928 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16929 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16931 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16932 SFmode value into a V4SF variable that was extracted from another
16933 V4SF variable without converting the element to double precision
16934 and back to single precision vector format.
16935 (vsx_insert_extract_v4sf_p9_2): Likewise.
16937 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16939 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16940 in UWHI to avoid undefined overflow.
16943 * ubsan.h (enum ubsan_encode_value_phase): New.
16944 (ubsan_encode_value): Change second argument to
16945 enum ubsan_encode_value_phase with default value of
16946 UBSAN_ENCODE_VALUE_GENERIC.
16947 * ubsan.c (ubsan_encode_value): Change second argument to
16948 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16949 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16950 create_tmp_var_raw instead of create_tmp_var and use a
16952 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16953 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16954 ubsan_encode_value callers.
16957 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16958 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16959 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16961 2017-06-19 Richard Biener <rguenther@suse.de>
16963 PR middle-end/81118
16964 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16965 estimates if we changed anything.
16967 2017-06-19 Richard Biener <rguenther@suse.de>
16969 PR tree-optimization/80887
16970 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16971 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16972 simplified lookups, then reset mprts_hook.
16973 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16975 (try_to_simplify): Likewise.
16977 2017-06-19 Martin Liska <mliska@suse.cz>
16980 * gimplify.c (gimplify_switch_expr):
16981 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16983 2017-06-19 Martin Liska <mliska@suse.cz>
16985 * doc/install.texi: Document that PGO runs in 4 stages.
16987 2017-06-19 Martin Liska <mliska@suse.cz>
16990 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16991 to dispatcher function name.
16992 * multiple_target.c (replace_function_decl): New function.
16993 (create_dispatcher_calls): Redirect both edges and references.
16995 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16997 * profile-count.c (profile_count::dump): Dump quality.
16998 (profile_count::differs_from_p): Update for unsigned val.
16999 * profile-count.h (profile_count_quality): New enum.
17000 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17002 2017-06-19 Richard Biener <rguenther@suse.de>
17004 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17005 struct function as arg.
17006 (estimate_numbers_of_iterations): Export overload with loop arg.
17007 (free_numbers_of_iterations_estimates_loop): Use an overload of
17008 free_numbers_of_iterations_estimates instead.
17009 * tree-cfg.c (remove_bb): Adjust.
17010 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17011 * tree-parloops.c (gen_parallel_loop): Likewise.
17012 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17014 (tree_unroll_loops_completely): Likewise.
17015 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17016 Use an overload instead and export.
17017 (estimated_loop_iterations): Adjust.
17018 (max_loop_iterations): Likewise.
17019 (likely_max_loop_iterations): Likewise.
17020 (estimate_numbers_of_iterations): Take struct function as arg
17022 (loop_exits_before_overflow): Adjust.
17023 (free_numbers_of_iterations_estimates_loop): Use an overload.
17024 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17025 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17027 2017-06-19 Richard Biener <rguenther@suse.de>
17030 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17031 RANGE_EXPR conservatively.
17033 2017-06-16 Carl Love <cel@us.ibm.com>
17035 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17036 definitions for vec_float, vec_float2, vec_floato,
17037 vec_floate built-ins.
17038 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17039 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17041 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17042 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17043 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17044 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17045 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17046 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17047 vec_floato): Add builtin defines.
17048 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17049 Update the built-in documentation file for the new built-in
17052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17054 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17055 (mthumb): Mark as the negative of -marm.
17057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17059 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17061 (ARM Options, -mtune): Document that this accepts the same
17062 extension options as -mcpu.
17063 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17067 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17068 permitted extensions.
17070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17072 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17073 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17074 (armv8-m.main): Add option +nodsp.
17075 * config/arm/arm-cpu-cdata.h: Regenerated.
17077 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17079 * config/arm/t-fuchsia: New file.
17080 * config.gcc (arm*-*-fuchsia*): Use it.
17082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17084 * config/arm/t-symbian: Rewrite for new option infrastructure.
17086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17088 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17089 (MULTILIB_REQUIRED): Likewise.
17091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17093 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17094 (MULTILIB_RESUE): Likewise.
17095 (MULTILIB_MATCHES): Likewise.
17096 (MULTLIB_REQUIRED): Likewise.
17098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17100 * config/arm/t-rtems: Rewrite for new option framework.
17102 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17104 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17105 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17106 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17107 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17108 * config/arm/t-multilib: ... here.
17109 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17110 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17111 armv7-a and armv8*-a when A-profile libraries have not been built.
17112 * config/arm/t-rmprofile: Rewrite.
17114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17116 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17117 with a backslash. Remove the backslash after substituting unescaped
17119 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17123 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17124 the driver to pass through the default setting of -mfloat-abi.
17125 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17127 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17128 (all_feat_combs): New rule.
17129 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17131 * config/arm/t-aprofile: Rewrite.
17133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17135 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17136 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17137 fpu is not specified by the user/command-line.
17138 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17139 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17140 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17141 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17142 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17143 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17145 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17147 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17148 * config/arm/t-arm-elf: Rewritten.
17150 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17152 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17153 have some floating-point instructions.
17154 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17155 (TARGET_MAYBE_HARD_FLOAT): New macro.
17156 * config/arm/arm-builtins.c (arm_init_builtins): Use
17157 TARGET_MAYBE_HARD_FLOAT.
17158 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17162 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17163 (configargs.h): Include it.
17164 (arm_print_hint_for_fpu_option): New function.
17165 (arm_parse_fpu_option): New function.
17166 (candidate_extension): New class.
17167 (arm_canon_for_multilib): New function.
17168 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17169 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17170 (ARCH_CANONICAL_SPECS): New macro.
17171 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17175 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17176 are set after handling multilib fragments. Set target_cpu_default2
17179 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17181 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17183 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17184 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17186 (check_arch): Likewise.
17187 * config/arm/arm.c (arm_configure_build_target): Handle
17188 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17189 options in the default.
17191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17193 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17194 when an option is an alias of another.
17195 * config/arm/parsecpu.awk (optalias): New parser token.
17196 (gen_comm_data): Mark non-alias options as such. Emit entries
17197 for extension aliases.
17198 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17199 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17200 (armv6kz, armv6zk, armv6t2): Likewise.
17201 (armv7): Make vfpv3-d16 an alias.
17202 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17204 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17205 Sort in canonical order.
17206 (armv8-a): Sort in canonical order.
17207 (armv8.1-a, armv8.2-a): Likewise.
17208 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17210 (cortex-a9): Sort in canonical order.
17211 * config/arm/arm.c (selftests.h): Include it.
17212 (arm_test_cpu_arch_data): New function.
17213 (arm_run_self_tests): New function.
17214 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17215 (targetm): Move declaration to the end of the file.
17216 * arm-cpu-cdata.h: Regenerated.
17218 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17220 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17221 call to target_mode_check describing the type of option passed.
17222 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17223 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17224 arm_parse_cpu_option_name to match parameters against list of
17226 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17227 arm_arch_core_flags data structure.
17228 * config/arm/arm-cpu_cdata.h: Regenerated.
17230 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17232 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17234 (arm_print_hint_for_cpu_option): Likewise.
17235 (arm_print_hint_for_arch_option): Likewise.
17236 (arm_parse_cpu_option_name): Likewise.
17237 (arm_parse_arch_option_name): Likewise.
17238 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17239 of entries in the all_fpus list.
17240 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17241 (arm_parse_cpu_option_name): Declare.
17242 (arm_parse_arch_option_name): Declare.
17243 (arm_parse_option_features): Declare.
17244 (arm_intialize_isa): Declare.
17245 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17247 (gen_comm_data): ... here. Make definitions non-static.
17248 * config/arm/arm-cpu-data.h: Regenerated.
17249 * config/arm/arm-cpu-cdata.h: Regenerated.
17251 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17253 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17254 (cpu_arch_extension): New structure.
17255 (cpu_arch_option, arch_option, cpu_option): New structures.
17256 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17257 architecture types.
17258 (gen_data): Generate new format data tables.
17259 * config/arm/arm.c (cpu_tune): New structure.
17260 (cpu_option, processors): Delete.
17261 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17262 (arm_print_hint_for_cpu_option): ... this and ...
17263 (arm_print_hint_for_arch_option): ... this.
17264 (arm_parse_arch_cpu_name): Delete. Replace with ...
17265 (arm_parse_cpu_option_name): ... this and ...
17266 (arm_parse_arch_option_name): ... this.
17267 (arm_unrecognized_feature): Change type of target parameter to
17269 (arm_parse_arch_cpu_features): Delete. Replace with ...
17270 (arm_parse_option_features): ... this.
17271 (arm_configure_build_target): Rework to use new configuration data
17273 (arm_print_tune_info): Rework for new configuration data tables.
17274 * config/arm/arm-cpu-data.h: Regenerated.
17275 * config/arm/arm-cpu.h: Regenerated.
17277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17279 * Makefile.in (OBJS): Move sbitmap.o from here ...
17280 (OBJS-libcommon): ... to here.
17282 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17284 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17285 (ISA_ALL_CRYPTO): New macro.
17286 (ISA_ALL_SIMD): New macro
17287 (ISA_ALL_FP): New macro.
17288 * config/arm/arm.c (fpu_bitlist): Update initializer.
17289 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17291 (arm9e): Add fpu. Add option for nofp
17292 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17293 (arm926ej-s, arm1026ej-s): Likewise.
17294 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17295 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17296 neon-fp16, neon-vfpv4, nofp and nosimd.
17297 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17298 (cortex-a8): Add fpu. Add option for nofp.
17299 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17300 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17301 (cortex-r4f): Add fpu.
17302 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17303 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17305 (cortex-r8): Likewise.
17306 (cortex-m4): Add fpu. Add option for nofp.
17307 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17308 (cortex-a17.cortex-a7): Likewise.
17309 (cortex-a32): Add fpu. Add options for crypto and nofp.
17310 (cortex-a35, cortex-a53): Likewise.
17311 (cortex-a57): Add fpu. Add option for crypto.
17312 (cortex-a72, cortex-a73): Likewise.
17313 (exynos-m1): Likewise.
17314 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17315 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17316 (cortex-m33): Add fpu. Add option for nofp.
17317 * config/arm/arm-cpu-cdata.h: Regenerated
17318 * config/arm/arm-cpu-data.h: Regenerated.
17320 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17322 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17323 (armv5te, armv5tej): Likewise.
17324 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17325 (armv7): Add options fp and vfpv3-d16.
17326 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17327 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17329 (armv7ve): Likewise.
17330 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17331 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17332 (armv8-a): Add nocrypto option.
17333 (armv8.1-a, armv8.2-a): Likewise.
17334 (armv8-m.main): add options fp, fp.dp and nofp.
17336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17338 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17340 (armv8-a+crc): Delete.
17341 (armv8.1-a): Add options simd, crypto and nofp.
17342 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17343 (armv8.2-a+fp16): Delete.
17344 (armv8-m.main): Add option dsp.
17345 (armv8-m.main+dsp): Delete.
17346 (cortex-a8): Add fpu. Add nofp option.
17347 (cortex-a9): Add fpu. Add nofp and nosimd options.
17348 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17349 link to main cpu and architecture data structures.
17350 (gen_comm_data): Only put isa attributes from the main architecture
17352 (option): New statement for architecture and CPU entries.
17353 * arm.c (struct cpu_option): New structure.
17354 (struct processors): Add entry for options.
17355 (arm_unrecognized_feature): New function.
17356 (arm_parse_arch_cpu_name): Ignore any characters after the first
17358 (arm_parse_arch_cpu_feature): New function.
17359 (arm_configure_build_target): Separate out any CPU and architecture
17360 features and parse separately. Don't error out if -mfpu=auto is
17361 used with only an architecture string.
17362 (arm_print_asm_arch_directives): New function.
17363 (arm_file_start): Call it.
17364 * config/arm/arm-cpu-cdata.h: Regenerated.
17365 * config/arm/arm-cpu-data.h: Likewise.
17366 * config/arm/arm-tables.opt: Likewise.
17368 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17370 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17371 assembler when it is not -mfpu=auto.
17373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17375 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17376 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17377 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17378 (ASM_CPU_SPEC): Rewrite.
17379 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17380 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17381 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17382 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17383 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17384 copied string is NUL-terminated. Also strip any characters prefixed
17386 (arm_rewrite_selected_arch): New function.
17387 (arm_rewrite_march): New function.
17389 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17391 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17392 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17393 (march, mcpu, mtune): Convert to string-based options.
17394 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17395 (arm_parse_arch_cpu_name): New function.
17396 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17397 identify selected architecture or CPU.
17398 (arm_option_save): New function.
17399 (TARGET_OPTION_SAVE): Redefine.
17400 (arm_option_restore): Restore string options.
17401 (arm_option_print): Print string options.
17403 2017-06-16 Martin Sebor <msebor@redhat.com>
17405 PR tree-optimization/80933
17406 PR tree-optimization/80934
17407 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17408 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17409 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17410 (gimple_fold_builtin): Call them.
17412 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17414 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17415 as unlikely; update profile.
17417 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17419 * predict.c (force_edge_cold): Handle declaring edges impossible
17422 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17424 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17426 (try_unroll_loop_completely): Fix reporting.
17428 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17430 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17432 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17435 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17436 if given a non-constant argument for an intrinsic which requires a
17439 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17441 * profile.c (compare_freqs): New function.
17442 (branch_prob): Sort edge list.
17443 (find_spanning_tree): Assume that the list is priority sorted.
17445 2017-06-16 Richard Biener <rguenther@suse.de>
17447 PR tree-optimization/81090
17448 * passes.def (pass_record_bounds): Remove.
17449 * tree-pass.h (make_pass_record_bounds): Likewise.
17450 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17451 make_pass_record_bounds): Likewise.
17452 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17453 not free niter estimates at the beginning but at the end.
17454 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17456 2017-06-16 Richard Biener <rguenther@suse.de>
17458 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17459 initializer to workaround ICE in host GCC 4.8.
17461 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17463 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17465 (clone_inlined_nodes): Update.
17467 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17469 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17470 prefetch settings, and enable prefetching by default at -O3.
17472 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17474 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17475 Set flag_prefetch_loop_arrays according to tuning data.
17477 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17479 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17480 New tune structure.
17481 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17482 [Unrelated to main purpose of the patch] Place the pointer field last
17483 to enable type checking errors when tune structure are wrongly merged.
17484 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17485 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17486 (thunderx2t99_prefetch_tune): New tune constants.
17487 (tune_params *_tunings): Update all tunings (no functional change).
17488 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17489 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17490 from tunings structures.
17492 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17495 * ubsan.c (instrument_null): Add T argument, use it instead
17496 of computing it based on IS_LHS.
17497 (instrument_object_size): Likewise.
17498 (pass_ubsan::execute): Adjust instrument_null and
17499 instrument_object_size callers to pass gimple_get_lhs or
17500 gimple_assign_rhs1 result to it. Use instrument_null instead of
17501 calling get_base_address and instrument_mem_ref. Handle
17502 aggregate call arguments for object-size sanitization.
17504 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17506 PR tree-optimization/81089
17507 * tree-vrp.c (is_masked_range_test): Validate operands of
17510 2017-06-15 Martin Sebor <msebor@redhat.com>
17513 * dumpfile.c (dump_register): Avoid calling memset to initialize
17514 a class with a default ctor.
17515 * gcc.c (struct compiler): Remove const qualification.
17516 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
17517 * hash-table.h: Ditto.
17518 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
17520 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
17521 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
17523 * params.h (struct param_info): Make struct members non-const.
17524 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
17525 with default initialization.
17526 * vec.h (vec_copy_construct, vec_default_construct): New helper
17528 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
17529 with vec_copy_construct.
17530 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
17531 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
17532 * doc/invoke.texi (-Wclass-memaccess): Document.
17534 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17536 * emit-rtl.h (is_leaf): Update comment about local
17537 register allocator.
17539 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17542 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
17543 for a variable to have a section before checking if the section has a
17545 Set section to.persistent if persistent attribute is set.
17546 Warn if .persistent attribute is used on an automatic variable.
17548 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
17550 PR rtl-optimization/80474
17551 * reorg.c (update_block): Do not ignore instructions in a delay slot.
17553 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
17555 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
17558 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
17560 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
17561 (casesi): Emit bounds checking as RTL.
17562 (casesi_internal_mips16_<mode>): Remove bounds checking.
17564 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
17566 * config/xtensa/xtensa.c (xtensa_option_override): Append
17567 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
17568 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
17569 xtensa_doloop_hooks): Define unconditionally.
17570 (xtensa_reorg_loops): Only call reorg_loops in the presence of
17572 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
17573 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
17574 for it in xtensa_option_override.
17575 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
17576 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
17578 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
17580 * doc/cppopts.texi: Document '-' special value to -MF.
17582 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
17584 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
17585 (cortex_a53_fconst): Likewise.
17586 (cortex_a53_fpmul): Likewise.
17587 (cortex_a53_f_load_64): Likewise.
17588 (cortex_a53_f_load_many): Likewise.
17589 (cortex_a53_advsimd_alu): Likewise.
17590 (cortex_a53_advsimd_alu_q): Likewise.
17591 (cortex_a53_advsimd_mul): Likewise.
17592 (cortex_a53_advsimd_mul_q): Likewise.
17593 (fpmac bypass): Add new bypass for fpmac-fpmac case.
17594 Add missing fmul, r2f_cvt and fconst cases.
17596 2017-06-14 Richard Biener <rguenther@suse.de>
17598 PR middle-end/81088
17599 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
17601 (fold_binary_loc): When associating do not treat pre-existing
17602 TREE_OVERFLOW on literal constants as a reason to allow
17603 TREE_OVERFLOW on associated literal constants.
17605 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17607 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17608 (MASK_FEATURES): New macro.
17609 * config/sparc/sparc.c (sparc_option_override): Remove the special
17610 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17612 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17614 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17615 a division of 0 if non-call exceptions are enabled.
17617 2017-06-14 Andrew Pinski <apinski@cavium.com>
17618 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17621 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17622 Improve vector initialization code gen for only variable case.
17624 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17626 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17628 2017-06-14 Richard Biener <rguenther@suse.de>
17630 PR tree-optimization/81083
17631 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17634 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17636 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17637 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17638 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17639 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17640 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17641 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17643 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17645 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17646 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17648 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17650 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17652 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17654 * config/rs6000/t-rtems: Don't handle SPE.
17656 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17658 * config/rs6000/t-linux: Don't handle SPE.
17660 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17662 * config/rs6000/eabispe.h: Delete file.
17664 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17666 * config/rs6000/t-spe: Delete file.
17668 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17670 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17671 (rs6000_legitimate_offset_address_p): Return false for anything in
17672 V2SImode or V2SFmode.
17674 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17676 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17677 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17678 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17680 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17681 (rs6000_legitimate_offset_address_p): Ditto.
17682 (rs6000_emit_move): Ditto.
17683 (rs6000_init_builtins): Remove V4HI_type_node.
17685 2017-06-13 Martin Liska <mliska@suse.cz>
17688 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17689 (gate_asan): Likewise.
17690 * asan.h (asan_no_sanitize_address_p): Remove the function.
17691 (sanitize_flags_p): New function.
17692 * builtins.def: Fix coding style.
17693 * common.opt: Use renamed enum value.
17694 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17695 * doc/extend.texi: Document no_sanitize attribute.
17696 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17697 to SANITIZE_UNDEFINED_NONDEFAULT.
17698 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17699 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17700 Use sanitize_flags_p.
17701 * gimplify.c (gimplify_function_tree): Likewise.
17702 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17703 * opts.c (parse_no_sanitize_attribute): New function.
17704 (common_handle_option): Use renamed enum value.
17705 * opts.h (parse_no_sanitize_attribute): Declare.
17706 * tree.c (sanitize_flags_p): New function.
17707 * tree.h: Declared here.
17708 * tsan.c: Use sanitize_flags_p.
17709 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17710 (instrument_mem_ref): Likewise.
17711 (instrument_bool_enum_load): Likewise.
17712 (do_ubsan_in_current_function): Remove the function.
17713 (pass_ubsan::execute): Use sanitize_flags_p.
17714 * ubsan.h: Remove do_ubsan_in_current_function
17715 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17716 (dump_function_to_file): Use it here.
17718 2017-06-13 Martin Jambor <mjambor@suse.cz>
17720 PR tree-optimization/80803
17721 PR tree-optimization/81063
17722 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17723 (propagate_subaccesses_across_link): Enqueue subtree whenever
17724 necessary instead of relying on the caller.
17726 2017-06-13 Martin Jambor <mjambor@suse.cz>
17728 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17729 that have a first_link.
17730 (sort_and_splice_var_accesses): Do not check first_link before
17732 (subtree_mark_written_and_enqueue): Likewise.
17733 (propagate_all_subaccesses): Likewise and do not stop at first
17734 parent with a first_link.
17736 2017-06-13 Martin Jambor <mjambor@suse.cz>
17738 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17741 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17743 * match.pd: New pattern.
17745 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17747 * tree-vrp.c (is_masked_range_test): New function.
17748 (register_edge_assert_for): Determine ranges for
17751 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17753 PR tree-optimization/67328
17754 * fold-const.c (maskable_range_p): New function.
17755 (build_range_check): Generate bittests if possible.
17757 2017-06-13 Martin Liska <mliska@suse.cz>
17759 * gimple-pretty-print.c (dump_probability): Add new argument.
17760 (dump_edge_probability): Dump both probability and count.
17761 (dump_gimple_label): Likewise.
17762 (dump_gimple_bb_header): Likewise.
17764 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17767 * config/avr/avr-devices.c: Fix indentation.
17768 * config/avr/gen-avr-mmcu-specs.c: Dito.
17770 2017-06-13 Richard Biener <rguenther@suse.de>
17772 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17773 instead get vector type from stmt_info.
17774 (vectorizable_reduction): Adjust. Remove dead code.
17776 2017-06-13 Richard Biener <rguenther@suse.de>
17778 PR middle-end/81065
17779 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17780 case of C * (x * C2 + C3).
17781 (fold_addr_of_array_ref_difference): Properly fold index difference.
17783 2017-06-12 David S. Miller <davem@davemloft.net>
17786 * config/sparc/sparc.md (return expander): Emit frame blockage if
17787 function uses alloca.
17789 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17791 * combine.c (make_field_assignment): Check len rather than the mode
17792 precision when calling force_to_mode.
17794 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17796 Support multilibs and devices that see flash in RAM address range.
17799 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17800 (avr_mcu_t) <flash_pm_offset>: New field.
17801 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17802 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17803 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17804 (AVR_TINY_PM_OFFSET): Remove macro.
17805 * config/avr/avr.opt (-mshort-calls): New option.
17806 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17807 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17808 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17809 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17810 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17811 instead of avr_arch->have_jmp_call.
17812 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17813 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17814 avr_arch->flash_pm_offset to define.
17815 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17816 new field flash_pm_offset. Add entry for avrxmega3.
17817 (avr_texinfo): Add entry for avrxmega3.
17818 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17819 attiny212, attiny214,
17820 attiny412, attiny414, attiny416, attiny417,
17821 attiny814, attiny816, attiny817,
17822 attiny1614, attiny1616, attiny1617,
17823 attiny3214, attiny3216, attiny3217.
17824 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17825 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17826 (avr_print_operand_address) [AVR_TINY]: Same.
17827 (avr_asm_init_sections) <readonly_data_section>: Only patch
17828 callback if avr_arch->flash_pm_offset = 0.
17829 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17830 for rodata if avr_arch->flash_pm_offset != 0.
17831 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17832 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17833 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17834 (m_options): Append opt_rcall.
17835 (m_dirnames): Append dir_rcall.
17836 * config/avr/t-multilib: Regenerate.
17838 * configure.ac [target=avr]: Check whether avrxmega3 default
17839 linker description file works as needed.
17840 * configure: Regenerate.
17841 * doc/avr-mmcu.texi: Regenerate.
17842 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17843 <__AVR_ARCH__>: Document avrxmega3 and 103.
17844 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17845 <__AVR_SHORT_CALLS__>: Document it.
17846 <__AVR_PM_BASE_ADDRESS__>: Document it.
17847 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17848 (AVR Variable Attributes) <progmem>: Document this is
17849 not needed for avrxmega3.
17850 (AVR Named Address Spaces) <__flash>: Dito.
17852 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17854 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17856 2017-06-12 Doug Rupp <rupp@adacore.com>
17858 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17859 Append vxworks-stdint.h to the tm_file list.
17860 * config/vxworks-stdint.h: New file.
17862 2017-06-12 Martin Liska <mliska@suse.cz>
17864 PR tree-optimization/81041
17865 * tree-profile.c (gimple_gen_ic_func_profiler):
17866 Create an extra BB in profile-generate
17867 (gimple_gen_time_profiler): Likewise.
17869 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17871 PR tree-optimization/81003
17872 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17873 (update_range_test): Use it instead of force_gimple_operand_gsi.
17875 2017-06-12 Richard Biener <rguenther@suse.de>
17877 PR tree-optimization/81053
17878 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17879 with backedge value not defined in loop. Simplify def stmt
17882 2017-06-11 Tom de Vries <tom@codesourcery.com>
17885 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17887 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17888 nvptx_cannot_force_const_mem.
17890 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17892 * opts.c (finish_options): Move test for flag_split_stack after
17893 it has been initialized.
17895 2017-06-11 Jason Merrill <jason@redhat.com>
17897 * tree.h (id_equal): New.
17898 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17899 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17900 instead of strcmp of IDENTIFIER_POINTER.
17902 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17904 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17905 (mark_all_inlined_calls_cdtor): Fix formating.
17906 (inline_transform): Rescale profile before inlining.
17908 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17910 * cgraph.h (cgraph_edge::clone): Update prototype.
17911 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17912 (cgraph_node::create_clone): Update.
17913 (cgraph_node::create_version_clone): Update.
17914 * tree-inline.c (copy_bb): Update.
17915 (expand_call_inline): Update.
17917 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17919 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17920 factored out from ...
17921 (rs6000_emit_prologue): ... here.
17923 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17925 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17926 factored out from ...
17927 (rs6000_emit_prologue): ... here.
17929 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17931 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17933 (handle_missing_profiles): Fix computation of tp_first_run.
17934 (counts_to_freqs): Do not touch freqs when count is 0.
17936 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17938 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17941 2017-06-10 Tom de Vries <tom@codesourcery.com>
17943 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17944 attributes): Document signal effective target.
17946 2017-06-10 Tom de Vries <tom@codesourcery.com>
17948 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17949 Document effective target stack_size.
17951 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17953 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17954 to the edit_context if they can be auto-applied.
17956 2017-06-9 Ian Lance Taylor <iant@golang.org>
17958 * opts.c (finish_options): If -fsplit-stack, disable implicit
17959 -forder-blocks-and-partition.
17960 * doc/invoke.texi (Optimize Options): Document that when using
17961 -fsplit-stack -forder-blocks-and-partition is not implicitly
17964 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17966 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17967 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17968 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17969 * builtins.def (abort, trap, unreachable): Declare cold.
17970 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17971 * tree-core.h (ECF_COLD): New.
17972 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17973 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17975 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17977 * predict.c (unlikely_executed_stmt_p): Cleanup.
17979 2017-06-09 Richard Biener <rguenther@suse.de>
17981 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17982 model if the ref is always written to.
17984 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17986 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17988 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17990 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17993 2017-06-09 Tom de Vries <tom@codesourcery.com>
17996 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17997 "target cannot support label values" when encountering LABEL_REF.
17999 2017-06-09 Martin Liska <mliska@suse.cz>
18001 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18002 (gimple_gen_ic_func_profiler): Emit direct comparison
18003 of __gcov_indirect_call_callee with NULL.
18004 (gimple_gen_time_profiler): Change probability from
18005 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18007 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18009 * profile.c (edge_gcov_counts): Turn to pointer.
18010 (compute_branch_probabilities, compute_branch_probabilities): Update.
18011 (branch_prob): Do not clear edge_gcov_count.
18012 * profile.h (edge_gcov_counts): Turn to pointer.
18013 (edge_gcov_count): Update.
18015 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18017 * gimple.h (gimple_check_failed): Mark cold.
18019 2017-06-09 Richard Biener <rguenther@suse.de>
18021 PR tree-optimization/66623
18022 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18023 refactor check_reduction into two parts, properly computing
18024 whether we have to check reduction validity for outer loop
18027 2017-06-09 Richard Biener <rguenther@suse.de>
18029 PR tree-optimization/79483
18030 * graphite-scop-detection.c (order): New global.
18031 (get_order): Compute bb to order mapping that satisfies code
18032 generation constraints.
18033 (cmp_pbbs): New helper.
18034 (build_scops): Start domwalk at entry block, sort generated
18037 2017-06-09 Richard Biener <rguenther@suse.de>
18039 PR middle-end/81007
18040 * ipa-polymorphic-call.c
18041 (ipa_polymorphic_call_context::restrict_to_inner_class):
18042 Skip FIELD_DECLs with error_mark_node type.
18043 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18046 2017-06-09 Martin Liska <mliska@suse.cz>
18048 * predict.c (struct branch_predictor): New struct.
18049 (test_prediction_value_range): New test.
18050 (predict_c_tests): New function.
18051 * selftest-run-tests.c (selftest::run_tests): Run the function.
18052 * selftest.h: Declare new tests.
18054 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18057 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18058 gen_add3_insn did not fail.
18059 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18060 r0, construct that number in a temporary reg and add that reg to r0.
18061 If asked to put the result in r0 as well, fail.
18063 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18065 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18066 for early expansion of vec_eqv.
18068 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18070 PR middle-end/81005
18071 * ubsan.c (instrument_null): Avoid pointless code temporary.
18072 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18074 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18078 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18081 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18084 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18086 * predict.c (unlikely_executed_edge_p): Move ahead.
18087 (probably_never_executed_edge_p): Use it.
18089 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18091 PR middle-end/79988
18092 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18093 gimple_call_builtin_p call.
18095 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18097 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18098 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18099 rtl_check_failed_type2, rtl_check_failed_code1,
18100 rtl_check_failed_code2, rtl_check_failed_code_mode,
18101 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18102 rtvec_check_failed_bounds, rtl_check_failed_flag,
18103 _fatal_insn_not_found, _fatal_insn): Likewise.
18104 * tree.h (tree_contains_struct_check_failed,
18105 tree_check_failed, tree_not_check_failed,
18106 tree_class_check_failed, tree_range_check_failed,
18107 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18108 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18109 tree_operand_check_failed, omp_clause_check_failed,
18110 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18113 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18115 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18116 flag_branch_probabilities.
18117 * ipa-inline.c (edge_badness): Likewise.
18118 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18119 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18120 * predict.c (maybe_hot_frequency_p): Likewise.
18121 (probably_never_executed): Likewise.
18122 * sched-ebb.c (schedule_ebbs): Likewise.
18123 * sched-rgn.c (find_single_block_region): Likewise.
18124 * tracer.c (tail_duplicate): Likewise.
18126 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18128 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18129 longer requires x_flag_profile_use.
18131 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18133 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18134 instead of flag_reorder_blocks_and_partition.
18135 * dbxout.c (dbxout_function_end): Likewise.
18136 * dwarf2out.c (gen_subprogram_die): Likewise.
18137 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18138 * hw-doloop.c (reorg_loops): Likewise.
18139 * varasm.c (assemble_start_function,
18140 assemble_end_function): Likewise.
18141 (decide_function_section): Do not check for
18142 flag_reorder_blocks_and_partition.
18144 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18146 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18148 (chkp_get_hard_register_fake_addr_expr): Ditto.
18149 (chkp_build_addr_expr): Add check for hard reg case.
18150 (chkp_parse_array_and_component_ref): Ditto.
18151 (chkp_find_bounds_1): Ditto.
18152 (chkp_process_stmt): Don't generate bounds store for
18155 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18157 * predict.c (maybe_hot_bb_p): Do not check profile status.
18158 (maybe_hot_edge_p): Likewise.
18159 (probably_never_executed): Check for zero counts even if profile
18161 (unlikely_executed_edge_p): New function.
18162 (unlikely_executed_stmt_p): New function.
18163 (unlikely_executed_bb_p): New function.
18164 (set_even_probabilities): Use unlikely predicates.
18165 (combine_predictions_for_bb): Likewise.
18166 (predict_paths_for_bb): Likewise.
18167 (predict_paths_leading_to_edge): Likewise.
18168 (determine_unlikely_bbs): New function.
18169 (estimate_bb_frequencies): Use it.
18170 (compute_function_frequency): Use zero counts even if profile is
18172 * profile-count.h: Fix typo.
18174 2017-08-08 Julia Koval <julia.koval@intel.com>
18176 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18177 _mm512_mask_cvtsepi16_storeu_epi8,
18178 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18179 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18180 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18181 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18182 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18183 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18184 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18185 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18186 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18187 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18188 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18189 __builtin_ia32_pmovuswb256mem_mask,
18190 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18191 __builtin_ia32_pmovwb512mem_mask): New builtins.
18193 2017-08-08 Julia Koval <julia.koval@intel.com>
18195 PR target/73350,80862
18196 * config/i386/subst.md (round): Fix round pattern.
18197 * config/i386/i386.c (ix86_erase_embedded_rounding):
18198 Fix erasing rounding for the fixed pattern.
18200 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18202 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18204 2017-06-08 Martin Liska <mliska@suse.cz>
18206 PR gcov-profile/80911
18207 * gcov.c (block_info::block_info): New constructor.
18209 2017-06-07 Carl Love <cel@us.ibm.com>
18211 * config/rs6000/rs6000-c: The return type of the following
18212 built-in functions was implemented as int not long long. Fix sign
18213 of return value for the unsigned version of vec_mulo and vec_mule.
18214 vector unsigned long long vec_bperm (vector unsigned long long,
18215 vector unsigned char)
18216 vector signed long long vec_mule (vector signed int,
18218 vector unsigned long long vec_mule (vector unsigned int,
18219 vector unsigned int)
18220 vector signed long long vec_mulo (vector signed int,
18222 vector unsigned long long vec_mulo (vector unsigned int,
18223 vector unsigned int)
18224 * doc/extend.texi: Fix the documentation for the built-in
18227 2017-06-07 Carl Love <cel@us.ibm.com>
18230 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18233 2017-06-07 Carl Love <cel@us.ibm.com>
18235 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18236 support, Generate doublehv for signed int/float for BE case only.
18238 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18240 * doc/invoke.texi (mcx16): Rewrite.
18242 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18244 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18245 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18246 *mov<mode>_softfloat, and an anonymous splitter): Use
18247 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18249 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18251 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18253 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18254 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18255 (rs6000_debug_reg_global): Adjust.
18256 (rs6000_init_hard_regno_mode_ok): Adjust.
18257 (rs6000_dbx_register_number): Adjust.
18258 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18259 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18260 Remove SPE_ACC and SPEFSCR.
18261 (REG_ALLOC_ORDER): Ditto.
18262 (FRAME_POINTER_REGNUM): Change to 111.
18263 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18264 (REG_CLASS_NAMES): Ditto.
18265 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18266 (REGISTER_NAMES): Ditto.
18267 (ADDITIONAL_REG_NAMES): Ditto.
18268 (rs6000_reg_names): Ditto.
18269 * config/rs6000/rs6000.md: Renumber some register number
18272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18274 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18276 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18277 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18279 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18280 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18281 Delete the SPE high registers.
18282 (REG_ALLOC_ORDER): Ditto.
18283 (enum reg_class): Remove SPE_HIGH_REGS.
18284 (REG_CLASS_NAMES): Ditto.
18285 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18286 (REGISTER_NAMES): Ditto.
18287 (rs6000_reg_names): Ditto.
18288 * doc/tm.texi.in: Remove SPE as example.
18289 * doc/tm.texi: Regenerate.
18291 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18293 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18294 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18295 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18296 * config/rs6000/rs6000.md (type): Remove "brinc".
18298 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18300 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18301 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18302 * config/rs6000/linuxspe.h: Delete file.
18303 * config/rs6000/rs6000.md: Don't include spe.md.
18304 * config/rs6000/spe.h: Delete file.
18305 * config/rs6000/spe.md: Delete file.
18306 * config/rs6000/t-rs6000: Remove spe.md.
18308 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18310 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18311 (reg_or_none500mem_operand): Delete.
18312 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18313 instead of reg_or_none500mem_operand.
18315 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18317 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18318 handling of SPE flags.
18319 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18323 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18325 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18326 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18327 paired_divv2sf3): Similar.
18328 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18329 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18330 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18332 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18333 Rename the paired_* instruction patterns.
18334 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18336 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18337 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18338 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18339 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18340 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18341 PAIRED_VECTOR_MODE.
18342 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18343 (spe_func_has_64bit_regs_p): Delete.
18344 (spe_expand_predicate_builtin): Delete.
18345 (spe_expand_evsel_builtin): Delete.
18346 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18347 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18348 (invalid_e500_subreg): Delete.
18349 (rs6000_legitimize_address): Always force_reg op2 as well, for
18350 paired single memory accesses.
18351 (rs6000_member_type_forces_blk): Delete.
18352 (rs6000_spe_function_arg): Delete.
18353 (rs6000_expand_unop_builtin): Delete SPE handling.
18354 (rs6000_expand_binop_builtin): Ditto.
18355 (spe_expand_stv_builtin): Delete.
18356 (bdesc_2arg_spe): Delete.
18357 (spe_expand_builtin): Delete.
18358 (spe_expand_predicate_builtin): Delete.
18359 (spe_expand_evsel_builtin): Delete.
18360 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18361 (spe_init_builtins): Delete.
18362 (spe_func_has_64bit_regs_p): Delete.
18363 (savres_routine_name): Delete "info" parameter. Adjust callers.
18364 (rs6000_emit_stack_reset): Ditto.
18365 (rs6000_dwarf_register_span): Delete.
18366 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18367 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18368 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18370 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18372 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18373 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18374 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18375 mulv2sf3, divv2sf3): Delete expanders.
18377 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18379 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18381 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18383 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18384 * config/rs6000/rs6000.c: Ditto.
18386 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18388 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18389 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18390 comparison_operator.
18392 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18394 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18395 * config/rs6000/rs6000.opt: Ditto.
18396 * config/rs6000/t-rtems: Ditto.
18398 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18400 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18401 TARGET_E500_SINGLE by 0, simplify.
18402 * config/rs6000/rs6000.c: Ditto.
18403 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18404 (spe_build_register_parallel): Delete.
18405 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18406 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18407 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18408 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18409 (E500_CONVERT): Delete.
18410 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18412 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18414 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18415 * config/rs6000/dfp.md: Ditto.
18416 (negdd2, *negdd2_fpr): Merge.
18417 (absdd2, *absdd2_fpr): Merge.
18418 (negtd2, *negtd2_fpr): Merge.
18419 (abstd2, *abstd2_fpr): Merge.
18420 * config/rs6000/e500.h: Delete file.
18421 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18422 TARGET_FPRS by 1 and simplify.
18423 * config/rs6000/rs6000-c.c: Ditto.
18424 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18425 TARGET_DF_SPE by 0.
18426 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18428 * config/rs6000/rs6000.md: Ditto.
18429 (floatdidf2, *floatdidf2_fpr): Merge.
18430 (move_from_CR_gt_bit): Delete.
18431 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18432 (E500_CR_IOR_COMPARE): Delete.
18433 (All patterns that require !TARGET_FPRS): Delete.
18434 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18436 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18438 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18440 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18442 * graphds.c (add_edge): Intitialize edge's attached data.
18443 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18444 pointer parameter. Call pointed function on each edge during
18445 graph traversing. Skip traversing the edge when the function
18447 (graphds_dfs, graphds_scc): Ditto.
18448 (for_each_edge): New parameter. Pass the new parameter to callback
18450 * graphds.h (skip_edge_callback): New function pointer type.
18451 (graphds_dfs, graphds_scc): New function pointer parameter.
18452 (graphds_edge_callback, for_each_edge): New parameter.
18454 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18456 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18457 out code checking if runtime alias check is possible to below ...
18458 Call the new function.
18459 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18460 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18462 2017-06-07 Marek Polacek <polacek@redhat.com>
18465 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18466 TYPE_OVERFLOW_WRAPS check.
18468 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18470 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18471 if versioning is required.
18472 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18473 peeling with the check for versioning.
18475 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18477 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18478 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18479 Set true to new parameter if new ssa variable is defined.
18480 (vect_gen_vector_loop_niters): Refactor. Set range information
18481 for the new vector loop bound variable.
18482 (vect_do_peeling): Ditto.
18484 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18486 * tree-affine.c (ssa.h): Include header file.
18487 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18488 has wrapping overflow behavior.
18490 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18492 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18494 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18496 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18497 (tree_to_aff_combination): ... here.
18499 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18501 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18502 reg_pressure model function.
18503 (ivopts_global_cost_for_size): Delete.
18504 (determine_set_costs, iv_ca_recount_cost): Call new model function
18505 ivopts_estimate_reg_pressure.
18507 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18509 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18510 expensive than udiv. Remove floating point cases from mod.
18512 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18514 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18515 Increase idiv cost.
18517 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18519 * config/aarch64/aarch64.md
18520 (copysignsf3): Fix mask generation.
18522 2017-06-07 Jakub Jelinek <jakub@redhat.com>
18524 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
18526 (class dump_manager): Add register_dumps method.
18527 * dumpfile.c: Include langhooks.h.
18528 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
18529 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
18530 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
18531 (dump_manager::dump_register): Start with 512 entries instead of 32.
18532 (dump_manager::register_dumps): New method.
18533 * toplev.c (general_init): Instead of invoking register_dumps
18534 langhook, invoke register_dumps method on the dump manager.
18535 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
18538 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
18540 * doc/md.texi: Clarify the restrictions on a define_insn condition.
18541 Say that # requires an associated define_split to exist, and that
18542 the define_split must be suitable for use after register allocation.
18544 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18546 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
18547 (compute_outgoing_frequencies): Also initialize zero counts.
18548 (find_many_sub_basic_blocks): Do not produce uninitialized profile
18549 around loops; preserve more of profile when nothing changes.
18551 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
18553 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
18555 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
18556 * config/arm/arm-cpu-cdata.h: Regenerate.
18557 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
18558 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
18559 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
18560 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
18561 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
18563 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
18565 * config/arm/t-rmprofile: Likewise.
18566 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
18568 2017-06-06 David S. Miller <davem@davemloft.net>
18571 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
18572 blockage if function uses alloca.
18574 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18576 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
18577 New "uid" fields to hold pretty-print IDs of group and ref.
18578 Memory references are now identified as <group_id>:<ref_id>
18579 instead of using [random] addresses.
18580 (dump_mem_details): Simplify, no functional change.
18581 (dump_mem_ref): Simplify and make output more concise.
18582 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
18583 (find_or_create_group): Initialize group uid.
18584 (record_ref): Initialize ref uid. Improve debug output.
18585 (prune_group_by_reuse, should_issue_prefetch_p,)
18586 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
18587 (mark_nontemporal_store, determine_loop_nest_reuse):
18588 Improve debug output.
18590 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18592 * dbgcnt.def (prefetch): New debug counter.
18593 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
18594 (schedule_prefetches): Stop issueing prefetches if debug counter
18597 2017-06-06 Tom de Vries <tom@codesourcery.com>
18599 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
18600 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
18602 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18604 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18605 Use aarch64_reg_or_zero predicate for operand 4.
18606 (aarch64_compare_and_swap<mode> define_insn_and_split):
18607 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18608 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18610 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18612 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18613 (arm_compute_save_core_reg_mask): This.
18614 (thumb1_compute_save_reg_mask): Rename into ...
18615 (thumb1_compute_save_core_reg_mask): This.
18616 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18617 (arm_compute_frame_layout): Likewise.
18619 2017-06-06 Richard Biener <rguenther@suse.de>
18621 PR tree-optimization/80974
18622 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18623 keep or clear leaders SSA info.
18625 2017-06-06 Tom de Vries <tom@codesourcery.com>
18627 * config/nvptx/nvptx.c (split_mode_p): New function.
18628 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18630 2017-06-06 Tom de Vries <tom@codesourcery.com>
18632 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18634 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18637 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18640 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18642 * shrink-wrap.c (handle_simple_exit): Update profile.
18643 (try_shrink_wrapping): Upate profile.
18645 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18647 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18648 (tree_guess_outgoing_edge_probabilities): New.
18649 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18650 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18652 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18654 * ipa-split.c (split_function): Initialize return bb profile.
18656 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18658 * profile.c (compute_branch_probabilities): Also initialize
18659 EXIT_BLOCK profile.
18661 2017-06-06 Richard Biener <rguenther@suse.de>
18663 PR tree-optimization/80928
18664 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18665 (vect_analyze_loop_operations): Properly guard analysis for
18667 (vect_transform_loop): Likewise.
18668 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18669 (vect_model_induction_cost): Do not cost for pure SLP.
18670 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18671 of induction in inner loop vectorization.
18672 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18673 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18674 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18676 (vect_analyze_slp_cost_1): Cost induction.
18677 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18678 (vect_get_slp_vect_defs): Likewise.
18679 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18680 (vect_transform_stmt): Handle SLP reductions.
18681 * tree-vectorizer.h (vectorizable_induction): Adjust.
18683 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18685 * config/rs6000/rs6000.c (make_resolver_func): Update
18686 init_lowered_empty_function call.
18688 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18690 * doc/invoke.texi: Document the -fprofile-abs-path option.
18691 * common.opt (fprofile-abs-path): New option.
18692 * gcov-io.h (gcov_write_filename): Declare.
18693 * gcov-io.c (gcov_write_filename): New function.
18694 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18695 * profile.c (output_location): Likewise.
18697 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18699 * shring-wrap.c: Revert accidental commit.
18701 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18703 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18705 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18707 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18709 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18710 profile in callgraph edge.
18711 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18712 (apply_scale): Likewise.
18713 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18714 Also scale profile when inlining function with zero profile.
18715 (initialize_cfun): Update exit block profile even when it is zero.
18716 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18717 when profile is read.
18719 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18721 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18722 (CLONE_*): New constants to define the processors we can generate
18723 code for with the target_clone attribute.
18724 (rs6000_clone_map): New array to identify which clone processors
18725 the current program is running on.
18726 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18727 target_clone attribute.
18728 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18729 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18730 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18731 (cpu_expand_builtin): Add support for target_clone attribute.
18732 (rs6000_valid_attribute_p): Allow "default" attribute.
18733 (get_decl_name): New debug function to simplify printing the
18734 current function name in debugging statements.
18735 (rs6000_clone_priority): New functions to support the target_clone
18736 attribute, and be able to generate code to switch between ISA 2.05
18737 through ISA 3.0 (power6 through power9).
18738 (rs6000_compare_version_priority): Likewise.
18739 (rs6000_get_function_versions_dispatcher): Likewise.
18740 (make_resolver_func): Likewise.
18741 (add_condition_to_bb): Likewise.
18742 (dispatch_function_versions): Likewise.
18743 (rs6000_generate_version_dispatcher_body): Likewise.
18744 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18745 (fusion_gpr_load_p): Fix a spacing issue.
18746 * doc/extend.texi (Common Function Attributes): Document that the
18747 PowerPC supports the target_clone attribute.
18749 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18751 * config/arm/arm.h: explain F symbol found in description of ARM
18752 register allocation in its legend.
18754 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18756 * config/mips/frame-header-opt.c: Include profile-count.h.
18757 * config/riscv/riscv.c: Include profile-count.h
18759 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18761 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18763 (sm_set_flag_if_changed): Add bbs field.
18764 (execute_sm_if_changed_flag_set): Pass BBS.
18765 (execute_sm): Update.
18767 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18769 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18772 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18774 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18775 (peephole2): New peephole2 to emit the above.
18776 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18778 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18780 * config/aarch64/aarch64.c (define_peephole2 above
18781 *sub_<shift>_<mode>): New peephole.
18783 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18785 * config/i386/i386.c (make_resolver_func): Update.
18786 * Makefile.in: Add profile-count.h and profile-count.o
18787 * auto-profile.c (afdo_indirect_call): Update to new API.
18788 (afdo_set_bb_count): Update.
18789 (afdo_propagate_edge): Update.
18790 (afdo_propagate_circuit): Update.
18791 (afdo_calculate_branch_prob): Update.
18792 (afdo_annotate_cfg): Update.
18793 * basic-block.h: Include profile-count.h
18794 (struct edge_def): Turn count to profile_count.
18795 (struct basic_block_def): Likewie.
18796 (REG_BR_PROB_BASE): Move to profile-count.h
18797 (RDIV): Move to profile-count.h
18798 * bb-reorder.c (max_entry_count): Turn to profile_count.
18799 (find_traces): Update.
18800 (rotate_loop):Update.
18801 (connect_traces):Update.
18802 (sanitize_hot_paths):Update.
18803 * bt-load.c (migrate_btr_defs): Update.
18804 * cfg.c (RDIV): Remove.
18805 (init_flow): Use alloc_block.
18806 (alloc_block): Uninitialize count.
18807 (unchecked_make_edge): Uninitialize count.
18808 (check_bb_profile): Update.
18809 (dump_edge_info): Update.
18810 (dump_bb_info): Update.
18811 (update_bb_profile_for_threading): Update.
18812 (scale_bbs_frequencies_int): Update.
18813 (scale_bbs_frequencies_gcov_type): Update.
18814 (scale_bbs_frequencies_profile_count): New.
18815 * cfg.h (update_bb_profile_for_threading): Update.
18816 (scale_bbs_frequencies_profile_count): Declare.
18817 * cfgbuild.c (compute_outgoing_frequencies): Update.
18818 (find_many_sub_basic_blocks): Update.
18819 * cfgcleanup.c (try_forward_edges): Update.
18820 (try_crossjump_to_edge): Update.
18821 * cfgexpand.c (expand_gimple_tailcall): Update.
18822 (construct_exit_block): Update.
18823 * cfghooks.c (verify_flow_info): Update.
18824 (dump_bb_for_graph): Update.
18825 (split_edge): Update.
18826 (make_forwarder_block): Update.
18827 (duplicate_block): Update.
18828 (account_profile_record): Update.
18829 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18830 (get_estimated_loop_iterations): Update.
18831 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18832 (single_likely_exit): Update.
18833 * cfgloopmanip.c (scale_loop_profile): Update.
18835 (set_zero_probability): Update.
18836 (lv_adjust_loop_entry_edge): Update.
18837 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18838 (purge_dead_edges): Update.
18839 (rtl_account_profile_record): Update.
18840 * cgraph.c (cgraph_node::create): Uninitialize count.
18841 (symbol_table::create_edge): Uninitialize count.
18842 (cgraph_update_edges_for_call_stmt_node): Update.
18843 (cgraph_edge::dump_edge_flags): Update.
18844 (cgraph_node::dump): Update.
18845 (cgraph_edge::maybe_hot_p): Update.
18846 * cgraph.h: Include profile-count.h
18847 (create_clone), create_edge, create_indirect_edge): Update.
18848 (cgraph_node): Turn count to profile_count.
18849 (cgraph_edge0: Likewise.
18850 (make_speculative, clone): Update.
18851 (create_edge): Update.
18852 (init_lowered_empty_function): Update.
18853 * cgraphclones.c (cgraph_edge::clone): Update.
18854 (duplicate_thunk_for_node): Update.
18855 (cgraph_node::create_clone): Update.
18856 * cgraphunit.c (cgraph_node::analyze): Update.
18857 (cgraph_node::expand_thunk): Update.
18858 * final.c (dump_basic_block_info): Update.
18859 * gimple-streamer-in.c (input_bb): Update.
18860 * gimple-streamer-out.c (output_bb): Update.
18861 * graphite.c (print_global_statistics): Update.
18862 (print_graphite_scop_statistics): Update.
18863 * hsa-brig.c: Include basic-block.h.
18864 * hsa-dump.c: Include basic-block.h.
18865 * hsa-gen.c (T sum_slice): Update.
18866 (convert_switch_statements):Update.
18867 * hsa-regalloc.c: Include basic-block.h.
18868 * ipa-chkp.c (chkp_produce_thunks): Update.
18869 * ipa-cp.c (struct caller_statistics): Update.
18870 (init_caller_stats): Update.
18871 (gather_caller_stats): Update.
18872 (ipcp_cloning_candidate_p): Update.
18873 (good_cloning_opportunity_p): Update.
18874 (get_info_about_necessary_edges): Update.
18875 (dump_profile_updates): Update.
18876 (update_profiling_info): Update.
18877 (update_specialized_profile): Update.
18878 (perhaps_add_new_callers): Update.
18879 (decide_about_value): Update.
18880 (ipa_cp_c_finalize): Update.
18881 * ipa-devirt.c (struct odr_type_warn_count): Update.
18882 (struct decl_warn_count): Update.
18883 (struct final_warning_record): Update.
18884 (possible_polymorphic_call_targets): Update.
18885 (ipa_devirt): Update.
18886 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18887 * ipa-icf.c (sem_function::merge): Update.
18888 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18889 * ipa-inline.c (compute_uninlined_call_time): Update.
18890 (compute_inlined_call_time): Update.
18891 (want_inline_small_function_p): Update.
18892 (want_inline_self_recursive_call_p): Update.
18893 (edge_badness): Update.
18894 (lookup_recursive_calls): Update.
18895 (recursive_inlining): Update.
18896 (inline_small_functions): Update.
18897 (dump_overall_stats): Update.
18898 (dump_inline_stats): Update.
18899 * ipa-profile.c (ipa_profile_generate_summary): Update.
18900 (ipa_propagate_frequency): Update.
18901 (ipa_profile): Update.
18902 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18903 * ipa-utils.c (ipa_merge_profiles): Update.
18904 * loop-doloop.c (doloop_modify): Update.
18905 * loop-unroll.c (report_unroll): Update.
18906 (unroll_loop_runtime_iterations): Update.
18907 * lto-cgraph.c (lto_output_edge): Update.
18908 (lto_output_node): Update.
18909 (input_node): Update.
18910 (input_edge): Update.
18911 (merge_profile_summaries): Update.
18912 * lto-streamer-in.c (input_cfg): Update.
18913 * lto-streamer-out.c (output_cfg): Update.
18914 * mcf.c (create_fixup_graph): Update.
18915 (adjust_cfg_counts): Update.
18916 (sum_edge_counts): Update.
18917 * modulo-sched.c (sms_schedule): Update.
18918 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18919 * predict.c (maybe_hot_count_p): Update.
18920 (probably_never_executed): Update.
18921 (dump_prediction): Update.
18922 (combine_predictions_for_bb): Update.
18923 (propagate_freq): Update.
18924 (handle_missing_profiles): Update.
18925 (counts_to_freqs): Update.
18926 (rebuild_frequencies): Update.
18927 (force_edge_cold): Update.
18928 * predict.h: Include profile-count.h
18929 (maybe_hot_count_p, counts_to_freqs): UPdate.
18930 * print-rtl-function.c: Do not include cfg.h
18931 * print-rtl.c: Include basic-block.h
18932 * profile-count.c: New file.
18933 * profile-count.h: New file.
18934 * profile.c (is_edge_inconsistent): Update.
18935 (correct_negative_edge_counts): Update.
18936 (is_inconsistent): Update.
18937 (set_bb_counts): Update.
18938 (read_profile_edge_counts): Update.
18939 (compute_frequency_overlap): Update.
18940 (compute_branch_probabilities): Update; Initialize and deinitialize
18942 (branch_prob): Update.
18943 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18944 (edge_gcov_count): New.
18945 (bb_gcov_count): New.
18946 * shrink-wrap.c (try_shrink_wrapping): Update.
18947 * tracer.c (better_p): Update.
18948 * trans-mem.c (expand_transaction): Update.
18949 (ipa_tm_insert_irr_call): Update.
18950 (ipa_tm_insert_gettmclone_call): Update.
18951 * tree-call-cdce.c: Update.
18952 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18953 (gimple_duplicate_sese_tail): Update.
18954 (gimple_account_profile_record): Update.
18955 (execute_fixup_cfg): Update.
18956 * tree-inline.c (copy_bb): Update.
18957 (copy_edges_for_bb): Update.
18958 (initialize_cfun): Update.
18959 (freqs_to_counts): Update.
18960 (copy_cfg_body): Update.
18961 (expand_call_inline): Update.
18962 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18963 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18964 (try_unroll_loop_completely): Update.
18965 (try_peel_loop): Update.
18966 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18967 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18968 * tree-ssa-loop-split.c (connect_loops): Update.
18969 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18970 * tree-ssa-reassoc.c (branch_fixup): Update.
18971 * tree-ssa-tail-merge.c (replace_block_by): Update.
18972 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18973 (compute_path_counts): Update.
18974 (update_profile): Update.
18975 (recompute_probabilities): Update.
18976 (update_joiner_offpath_counts): Update.
18977 (estimated_freqs_path): Update.
18978 (freqs_to_counts_path): Update.
18979 (clear_counts_path): Update.
18980 (ssa_fix_duplicate_block_edges): Update.
18981 (duplicate_thread_path): Update.
18982 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18983 (struct switch_conv_info): Update.
18984 * tree-tailcall.c (decrease_profile): Update.
18985 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18986 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18987 * value-prof.c (check_counter): Update.
18988 (gimple_divmod_fixed_value): Update.
18989 (gimple_mod_pow2): Update.
18990 (gimple_mod_subtract): Update.
18991 (gimple_ic_transform): Update.
18992 (gimple_stringop_fixed_value): Update.
18993 * value-prof.h (gimple_ic): Update.
18995 2017-06-02 Carl Love <cel@us.ibm.com>
18997 * config/rs6000/rs6000-c: Add support for built-in functions
18998 vector double vec_doublee (vector signed int);
18999 vector double vec_doublee (vector unsigned int);
19000 vector double vec_doublee (vector float);
19001 vector double vec_doubleh (vector signed int);
19002 vector double vec_doubleh (vector unsigned int);
19003 vector double vec_doubleh (vector float);
19004 vector double vec_doublel (vector signed int);
19005 vector double vec_doublel (vector unsigned int);
19006 vector double vec_doublel (vector float);
19007 vector double vec_doubleo (vector signed int);
19008 vector double vec_doubleo (vector unsigned int);
19009 vector double vec_doubleo (vector float);.
19010 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19011 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19013 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19014 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19015 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19017 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19018 vec_doublel, vec_doubleh.
19019 * doc/extend.texi: Update the built-in documentation file for the
19020 new built-in functions.
19022 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19025 * ipa-inline-analysis.c (free_growth_caches): Set
19026 edge_removal_hook_holder to NULL after removing it.
19028 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19030 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19031 comparision with zero.
19033 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19034 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19035 for early expansion of vec_min and vec_max builtins.
19036 (builtin_function_type): Add min/max unsigned variants to those
19037 identified as having unsigned arguments.
19039 2017-06-02 Olivier Hainque <hainque@adacore.com>
19041 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19043 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19045 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19046 Use VALL_F16 iterator rather than VALL.
19048 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19050 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19051 Emit CBNZ inside loop when doing a strong exchange and comparing
19052 against zero. Generate the CC flags after the loop.
19054 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19056 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19057 (dl_section_ref): New.
19058 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19059 On AIX, append an expression to subtract the size of the
19060 section length to dl_section_ref.
19062 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19064 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19065 for early expansion of vector absolute builtins.
19067 2017-06-02 Richard Biener <rguenther@suse.de>
19069 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19070 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19072 2017-06-02 Richard Biener <rguenther@suse.de>
19074 PR tree-optimization/80948
19075 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19076 stmt order as well.
19078 2017-06-02 Richard Biener <rguenther@suse.de>
19080 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19082 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19083 for inductions as relevant.
19085 2017-06-02 Richard Biener <rguenther@suse.de>
19087 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19088 (vectorizable_induction): ... this. Remove dead code.
19090 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19092 * builtins. (expand_builtin_alloca): Remove second parameter and
19093 infer its value from the first parameter instead.
19094 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19096 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19098 PR rtl-optimization/80903
19099 * loop-doloop.c (add_test): Unshare sequence.
19101 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19103 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19105 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19107 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19109 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19110 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19111 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19112 (xlogue_layout::get_stub_rtx): Make static.
19113 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19114 (xlogue_layout::compute_stub_managed_regs): Rename to...
19115 (xlogue_layout::count_stub_managed_regs): ...this.
19116 (xlogue_layout::is_stub_managed_reg): New function.
19117 (xlogue_layout::m_stub_names): Rename to...
19118 (xlogue_layout::s_stub_names): ...this, make static.
19119 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19120 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19121 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19122 xlogue_layout::s_stub_names): Instantiate statics.
19123 (stub_managed_regs): Remove.
19124 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19125 (disable_call_ms2sysv_xlogues): Rename to...
19126 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19127 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19129 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19130 change after reload_completed.
19131 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19133 (ix86_expand_prologue): Likewise.
19134 (ix86_expand_epilogue): Likewise.
19135 (ix86_expand_split_stack_prologue): Likewise.
19136 (ix86_compute_frame_layout): Remove frame parameter ...
19137 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19138 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19140 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19141 (ix86_frame): Move from here ...
19142 * config/i386/i386.h (ix86_frame): ... to here.
19143 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19144 complete ix86_frame data structure instead. Remove some_ld_name.
19146 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19148 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19149 symbols that hold a DECL_VALUE_EXPR.
19151 2017-06-01 Martin Jambor <mjambor@suse.cz>
19153 PR tree-optimization/80898
19154 * tree-sra.c (process_subtree_disqualification): Removed.
19155 (disqualify_candidate): Do not acll
19156 process_subtree_disqualification.
19157 (subtree_mark_written_and_enqueue): New function.
19158 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19159 RHS has been disqualified and re-queue LHS if necessary. Apart
19160 from that, ignore disqualified RHS.
19162 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19164 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19165 address fetch for z10 or later.
19167 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19169 * config/arc/arc.md (tst_movb): Add guard when splitting.
19171 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19173 * config/arc/arc.c (arc_can_eliminate): Test against
19174 arc_frame_pointer_needed.
19176 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19178 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19179 to prevent store reordering.
19180 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19181 (type): Add block type.
19182 (stack_tie): Define special instruction to be used in
19185 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19187 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19188 constraint. It is not valid for the pattern.
19189 (noncommutative_binary_comparison): Likewise.
19191 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19193 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19196 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19198 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19199 be used by the reg-alloc.
19201 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19203 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19204 reg-alloc when having mul64 or mul32x16 instructions.
19205 (mulsidi3): Likewise.
19206 (umulsidi3): Likewise.
19207 (mulsi32x16): New pattern.
19208 (mulsi64): Likewise.
19209 (mulsidi64): Likewise.
19210 (umulsidi64): Likewise.
19211 (MUL32x16_REG): Define.
19212 (mul64_600): Use MUL32x16_REG.
19213 (mac64_600): Likewise.
19214 (umul64_600): Likewise.
19215 (umac64_600): Likewise.
19217 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19219 * config/arc/arc.md (mulsi3_700): Make it commutative.
19221 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19223 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19225 (*sign_extendsidi2_insn): Likewise for movstosw.
19227 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19229 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19230 the type of the input discriminant value. Convert the
19231 discriminant value of signedness vary.
19233 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19235 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19236 Add to -Wall section.
19238 2017-06-01 Richard Biener <rguenther@suse.de>
19240 PR middle-end/66313
19241 * fold-const.c (fold_plusminus_mult_expr): If the factored
19242 factor may be zero use a wrapping type for the inner operation.
19243 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19244 and handle moved defs.
19245 (process_assignment): Properly guard the unary op case. Return a
19246 tri-state indicating that moving the stmt before the call may allow
19247 to continue. Pass through to_move.
19248 (find_tail_calls): Handle moving unrelated defs before
19251 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19254 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19255 splitter result in the canonical way.
19257 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19259 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19260 also for 32bit target. Update insn attributes.
19261 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19263 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19265 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19266 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19267 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19268 (_mm_maskz_min_ss): New intrinsics.
19270 2017-05-31 Martin Liska <mliska@suse.cz>
19272 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19273 Change comment style to one we normally use.
19274 (vectorizable_reduction): Likewise.
19275 (vectorizable_induction): Likewise.
19276 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19277 (vectorizable_call): Likewise.
19278 (vectorizable_simd_clone_call): Likewise.
19279 (vectorizable_conversion): Likewise.
19280 (vectorizable_assignment): Likewise.
19281 (vectorizable_shift): Likewise.
19282 (vectorizable_operation): Likewise.
19283 (vectorizable_store): Likewise.
19284 (vectorizable_load): Likewise.
19285 * tree-vectorizer.h: Likewise.
19287 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19289 * passes.c (emergency_dump_function): New.
19290 * tree-pass.h (emergency_dump_function): Declare.
19291 * plugin.c (plugins_internal_error_function): Remove.
19292 * plugin.h (plugins_internal_error_function): Remove declaration.
19293 * toplev.c (internal_error_function): New static function. Use it...
19294 (general_init): ...here.
19296 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19298 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19299 (arc_rtx_costs): Add costs for new patterns.
19300 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19301 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19303 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19305 * tree-ssa-strlen.c (get_next_strinfo): New function.
19306 (get_stridx_plus_constant): Use it.
19307 (zero_length_string): Likewise.
19308 (adjust_related_strinfos): Likewise.
19309 (adjust_last_stmt): Likewise.
19311 2017-05-31 Richard Biener <rguenther@suse.de>
19314 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19315 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19317 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19319 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19320 loop_vinfo argument and use of dependence distance vectors.
19321 Check instead whether the two references differ only in their
19322 initial value and assume that they have the same alignment if the
19323 difference is a multiple of the vector alignment.
19324 (vect_analyze_data_refs_alignment): Update call accordingly.
19326 2017-05-31 Martin Liska <mliska@suse.cz>
19329 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19331 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19333 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19334 (create_intersect_range_checks): Move from ...
19335 * tree-data-ref.c (create_intersect_range_checks_index)
19336 (create_intersect_range_checks): ... to here.
19337 (create_runtime_alias_checks): New function factored from ...
19338 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19339 here. Call above function.
19340 * tree-data-ref.h (create_runtime_alias_checks): New function.
19342 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19344 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19345 segment length for dr_b and compute it in wide_int.
19347 2017-05-31 Richard Biener <rguenther@suse.de>
19349 PR tree-optimization/80906
19350 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19351 and pass through iv_map.
19352 (copy_bb_and_scalar_dependences): Adjust.
19353 (translate_pending_phi_nodes): Likewise.
19354 (copy_loop_close_phi_args): Handle code-generating IVs instead
19357 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19359 * diagnostic-color.c (color_dict): Add "type-diff".
19360 (parse_gcc_colors): Update comment.
19361 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19362 -fdiagnostics-show-template-tree and -fno-elide-type.
19363 (GCC_COLORS): Add type-diff to example.
19365 (-fdiagnostics-show-template-tree): New.
19366 (-fno-elide-type): New.
19367 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19368 the pp_format_decoder callback. Call any m_format_postprocessor's
19370 (pretty_printer::pretty_printer): Initialize
19371 m_format_postprocessor.
19372 (pretty_printer::~pretty_printer): Delete any
19373 m_format_postprocessor.
19374 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19375 (class format_postprocessor): New class.
19376 (struct pretty_printer::format_decoder): Document the new parameters.
19377 (struct pretty_printer::m_format_postprocessor): New field.
19378 * tree-diagnostic.c (default_tree_printer): Update for new
19379 bool and const char ** params.
19380 * tree-diagnostic.h (default_tree_printer): Likewise.
19382 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19384 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19385 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19386 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19387 rs6000_gen_cell_microcode code.
19388 (rs6000_final_prescan_insn): Delete.
19389 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19390 "warn-cell-microcode" entries.
19391 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19392 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19393 throughout. Change cc_reg_not_micro_cr0_operand to
19394 cc_reg_not_cr0_operand throughout.
19395 (*extendhi<mode>2_noload): Delete.
19396 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19397 (mwarn-cell-microcode): Delete.
19398 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19399 -mgen-cell-microcode and -mwarn-cell-microcode.
19401 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19404 * config/i386/constraints.md (Yd): New constraint.
19406 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19407 and (?Yd, r) alternatives. Update insn attributes.
19408 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19409 and (?*Yd, r) alternatives. Update insn attributes.
19410 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19412 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19414 * gimplify.c (gimplify_modify_expr): Don't create a
19415 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19418 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19420 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19422 2017-05-30 Richard Biener <rguenther@suse.de>
19424 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19425 and reduc_def fields.
19426 (STMT_VINFO_REDUC_TYPE): New define.
19427 (STMT_VINFO_REDUC_DEF): Likewise.
19428 (vect_force_simple_reduction): Adjust prototype.
19429 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19430 (vect_is_simple_reduction): Remove check_reduction argument.
19431 (vect_force_simple_reduction): Adjust and set
19432 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19433 (vectorizable_reduction): Do not re-do reduction analysis
19434 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19435 * tree-parloops.c (gather_scalar_reductions): Adjust.
19437 2017-05-30 Richard Biener <rguenther@suse.de>
19439 PR middle-end/80901
19440 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19443 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19445 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19446 Introduce unknown_misalignment parameter and remove vf.
19447 (vect_peeling_hash_get_lowest_cost):
19448 Pass unknown_misalignment parameter.
19449 (vect_enhance_data_refs_alignment):
19450 Fix unsupportable data ref treatment.
19452 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19454 * tree-vect-data-refs.c (vect_get_data_access_cost):
19455 Workaround for SLP handling.
19456 (vect_enhance_data_refs_alignment):
19457 Compute costs for doing no peeling at all, compare to the best
19458 peeling costs so far and avoid peeling if cheaper.
19460 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19462 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19463 Return peeling info and set costs to zero for unlimited cost
19465 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19466 with unknown misalignment. Compute and costs for unknown
19467 misalignment, compare them to the costs for known misalignment
19468 and choose the cheapest for peeling.
19470 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19472 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19473 (vect_get_peeling_costs_all_drs): Create function.
19474 (vect_peeling_hash_get_lowest_cost):
19475 Use vect_get_peeling_costs_all_drs.
19476 (vect_peeling_supportable): Create function.
19477 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19479 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19481 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19482 DR_HAS_NEGATIVE_STEP.
19483 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19484 (vect_enhance_data_refs_alignment): Use.
19485 (vect_duplicate_ssa_name_ptr_info): Use.
19486 * tree-vectorizer.h (dr_misalignment): Use.
19487 (known_alignment_for_access_p): Use.
19489 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19492 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19494 (has_section_name): New function.
19496 2017-05-30 Martin Liska <mliska@suse.cz>
19499 * auto-profile.c (get_function_decl_from_block): Fix
19502 2017-05-30 Richard Biener <rguenther@suse.de>
19504 PR middle-end/80876
19505 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19507 2017-05-30 Martin Liska <mliska@suse.cz>
19509 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19510 * dumpfile.h (struct dump_file_info): Remove ctors.
19512 2017-05-30 Martin Liska <mliska@suse.cz>
19514 * predict.def: Fix GNU coding style.
19516 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19518 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19519 Mark 'to' argument with ATTRIBUTE_UNUSED.
19521 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19523 * config/xtensa/xtensa.c (xtensa_emit_call): Use
19524 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
19525 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
19528 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
19530 * doc/install.texi (Options specification): Restore entry of
19531 --enable-sjlj-exceptions.
19533 2017-05-27 Michael Eager <eager@eagercon.com>
19536 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
19538 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
19540 * config/microblaze/microblaze.h
19541 (FIXED_REGISTERS): Update in macro.
19542 (CALL_USED_REGISTERS): Update in macro.
19544 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
19546 * doc/install.texi: Add links to macOS binary distributions.
19548 2017-05-27 Jakub Jelinek <jakub@redhat.com>
19552 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19554 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19556 2017-05-26 Martin Liska <mliska@suse.cz>
19558 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
19560 2017-05-26 Martin Liska <mliska@suse.cz>
19562 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
19563 always leading ';; '.
19564 (dump_bb_info): Likewise.
19565 (brief_dump_cfg): Likewise.
19566 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
19567 * dumpfile.c: Remove usage of TDF_VERBOSE.
19568 * dumpfile.h (enum dump_kind): Likewise.
19569 (dump_gimple_bb_header): Do not use TDF_COMMENT.
19570 * print-tree.c (debug_verbose): Remove.
19571 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
19572 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
19573 * tree-diagnostic.c (default_tree_printer): Replace
19574 TDF_DIAGNOSTIC with TDF_SLIM.
19576 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19578 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
19579 in parameter loop, rather than loop_vinfo.
19580 (create_intersect_range_checks): Ditto.
19581 (vect_create_cond_for_alias_checks): Update call to above functions.
19583 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19585 PR tree-optimization/80815
19586 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
19587 for merging runtime alias checks. Handle negative DR_STEPs.
19589 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19591 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
19593 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
19594 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
19595 out code pruning runtime alias checks.
19596 * tree-data-ref.c (prune_runtime_alias_test_list): New function
19597 factored out from above.
19598 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19600 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19602 (prune_runtime_alias_test_list): New decalaration.
19604 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19606 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19607 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19608 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19609 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19610 (operator==, comp_dr_with_seg_len_pair): Ditto.
19611 (vect_prune_runtime_alias_test_list): Ditto.
19613 2017-05-26 Martin Liska <mliska@suse.cz>
19616 * params.def: Bound partial-inlining-entry-probability param.
19618 2017-05-26 Marek Polacek <polacek@redhat.com>
19621 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19624 2017-05-26 Richard Biener <rguenther@suse.de>
19626 PR tree-optimization/80842
19627 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19630 2017-05-26 Richard Biener <rguenther@suse.de>
19632 PR tree-optimization/80844
19633 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19635 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19637 * doc/md.texi (Machine Constraints): Update x86 family
19638 machine constraints section to match 'config/i386/constraints.md'.
19640 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19642 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19644 2017-05-25 Nathan Sidwell <nathan@acm.org>
19646 * doc/invoke.texi (--enable-languages): Update documentation.
19648 2017-05-25 Martin Liska <mliska@suse.cz>
19650 * dumpfile.c: Add TDF_FOLDING.
19651 * dumpfile.h (enum dump_kind): Likewise.
19652 * genmatch.c (dt_simplify::gen_1): Use it.
19654 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19656 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19658 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19660 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19661 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19663 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19665 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19666 * match.pd (X == C): Rewrite it here.
19667 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19668 with_certain_nonzero_bits2): New predicates.
19669 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19671 2017-05-24 Nathan Sidwell <nathan@acm.org>
19673 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19676 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19679 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19681 * config/powerpcspe: New port. Files are copied from the rs6000
19682 port, with "rs6000" in filenames replaced by "powerpcspe".
19684 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19686 PR rtl-optimization/80754
19687 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19689 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19691 * config/sparc/sparc.md (length): Return the correct value for -mflat
19692 sibcalls to match output_sibcall.
19694 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19698 * config/rs6000/rs6000.c (struct machine_function): Add new field
19700 (rs6000_get_separate_components): Init that field, use it.
19701 (rs6000_components_for_bb): Use the field.
19703 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19705 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19707 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19709 PR middle-end/80823
19710 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19712 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19715 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19716 address against address_operand predicate.
19717 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19719 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19721 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19722 parameters passed indirectly.
19724 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19726 * config/i386/i386.md (*movdi_internal): Remove SSE4
19727 alternative 18 (?r, *v). Update insn attributes.
19728 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19729 Update insn attributes.
19730 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19731 Update insn attributes.
19732 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19733 alternative 1 (r, v). Remove isa attribute.
19734 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19735 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19736 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19738 2017-05-23 Tom de Vries <tom@codesourcery.com>
19740 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19743 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19745 * cgraphunit.c (symbol_table::process_new_functions): Update.
19746 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19747 (inline_generate_summary): Rename to ...
19748 (ipa_fn_summary_generate): ... this one.
19749 (inline_read_summary): Rename to ...
19750 (ipa_fn_summary_read): ... this one.
19751 (inline_write_summary): Rename to ...
19752 (ipa_fn_summary_write): ... this one.
19753 (inline_free_summary): Rename to ...
19754 (ipa_free_fn_summary): ... this one.
19755 (pass_data_local_fn_summary, pass_local_fn_summary,
19756 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19757 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19758 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19759 make_pass_ipa_fn_summary): New.
19760 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19761 inline_write_summary, inline_free_summary): Remove.
19762 (ipa_free_fn_summary) : New.
19763 * ipa-inline.c (ipa_inline): Update.
19764 (pass_ipa_inline): Do not generate summaries.
19765 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19767 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19768 and add pass_ipa_fn_summary.
19769 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19771 (make_pass_inline_parameters): Remove.
19773 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19775 * omp-low.c (struct omp_context): Remove "default_kind" member.
19778 * omp-offload.c (execute_oacc_device_lower): Remove the
19779 parallelism dimensions function attributes for unparallelized
19780 OpenACC kernels constructs.
19782 2017-05-23 Martin Liska <mliska@suse.cz>
19784 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19786 (cgraph_edge::make_speculative): Likewise.
19787 (cgraph_edge::resolve_speculation): Likewise.
19788 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19789 (cgraph_node::dump): Likewise.
19790 * cgraph.h: Likewise.
19791 * cgraphunit.c (analyze_functions): Likewise.
19792 (symbol_table::compile): Likewise.
19793 * ipa-cp.c (print_all_lattices): Likewise.
19794 (determine_versionability): Likewise.
19795 (initialize_node_lattices): Likewise.
19796 (ipcp_verify_propagated_values): Likewise.
19797 (estimate_local_effects): Likewise.
19798 (update_profiling_info): Likewise.
19799 (create_specialized_node): Likewise.
19800 (perhaps_add_new_callers): Likewise.
19801 (decide_about_value): Likewise.
19802 (decide_whether_version_node): Likewise.
19803 (identify_dead_nodes): Likewise.
19804 (ipcp_store_bits_results): Likewise.
19805 * ipa-devirt.c (dump_targets): Likewise.
19806 (ipa_devirt): Likewise.
19807 * ipa-icf.c (sem_item::dump): Likewise.
19808 (sem_function::equals): Likewise.
19809 (sem_variable::equals): Likewise.
19810 (sem_item_optimizer::read_section): Likewise.
19811 (sem_item_optimizer::execute): Likewise.
19812 (congruence_class::dump): Likewise.
19813 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19814 (dump_inline_summary): Likewise.
19815 (estimate_node_size_and_time): Likewise.
19816 (inline_analyze_function): Likewise.
19817 * ipa-inline-transform.c (inline_call): Likewise.
19818 * ipa-inline.c (report_inline_failed_reason): Likewise.
19819 (want_early_inline_function_p): Likewise.
19820 (edge_badness): Likewise.
19821 (update_edge_key): Likewise.
19822 (inline_small_functions): Likewise.
19823 * ipa-profile.c (ipa_profile): Likewise.
19824 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19825 (ipa_make_edge_direct_to_target): Likewise.
19826 (remove_described_reference): Likewise.
19827 (ipa_impossible_devirt_target): Likewise.
19828 (propagate_controlled_uses): Likewise.
19829 (ipa_print_node_params): Likewise.
19830 (ipcp_transform_function): Likewise.
19831 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19832 (propagate_pure_const): Likewise.
19833 * ipa-reference.c (generate_summary): Likewise.
19834 (read_write_all_from_decl): Likewise.
19835 (propagate): Likewise.
19836 (ipa_reference_read_optimization_summary): Likewise.
19837 * ipa-utils.c (ipa_merge_profiles): Likewise.
19838 * ipa.c (walk_polymorphic_call_targets): Likewise.
19839 (symbol_table::remove_unreachable_nodes): Likewise.
19840 (ipa_single_use): Likewise.
19841 * passes.c (execute_todo): Likewise.
19842 * predict.c (drop_profile): Likewise.
19843 * symtab.c (symtab_node::get_dump_name): New function.
19844 (symtab_node::dump_name): Likewise.
19845 (symtab_node::dump_asm_name): Likewise.
19846 (symtab_node::dump_references): Likewise.
19847 (symtab_node::dump_referring): Likewise.
19848 (symtab_node::dump_base): Likewise.
19849 (symtab_node::debug_symtab): Likewise.
19850 * tree-sra.c (convert_callers_for_node): Likewise.
19851 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19852 * value-prof.c (init_node_map): Likewise.
19854 2017-05-23 Martin Liska <mliska@suse.cz>
19856 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19857 and symtab_node::debug_symtab to symbol_table::debug.
19858 * cgraphunit.c (analyze_functions): Use the renamed function.
19859 (symbol_table::compile): Likewise.
19860 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19861 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19862 * passes.c (execute_todo): Likewise.
19863 * symtab.c (symbol_table::dump): New function.
19864 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19866 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19868 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19869 that nonconst implies exec.
19871 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19873 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19874 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19875 (inline_edge_summary_vec): Turn into ...
19876 (ipa_call_summaries): ... this one.
19877 (redirect_to_unreachable, edge_set_predicate,
19878 evaluate_properties_for_edge, inline_summary_alloc,
19879 reset_ipa_call_summary, reset_inline_summary,
19880 inline_summary_t::duplicate): Update.
19881 (inline_edge_duplication_hook): Turn to ...
19882 (ipa_call_summary_t::duplicate): ... this one.
19883 (inline_edge_removal_hook): Turn to ...
19884 (ipa_call_summary_t::remove): ... this one.
19885 (dump_inline_edge_summary): Turn to ...
19886 (dump_ipa_call_summary): ... this one.
19887 (estimate_function_body_sizes): Update.
19888 (inline_update_callee_summaries): Update.
19889 (remap_edge_change_prob): Update.
19890 (remap_edge_summaries): Update.
19891 (inline_merge_summary): Update.
19892 (do_estimate_edge_time): Update.
19893 (inline_generate_summary): Update.
19894 (inline_read_section): Update.
19895 (inline_read_summary): Update.
19896 (inline_free_summary): Update.
19897 * ipa-inline.c (can_inline_edge_p): Update.
19898 (compute_inlined_call_time): Update.
19899 (want_inline_small_function_p): Update.
19900 (edge_badness): Update.
19901 (early_inliner): Update.
19902 * ipa-inline.h (inline_edge_summary): Turn to ...
19903 (ipa_call_summary): ... this one.
19904 (ipa_call_summary_t): New class.
19905 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19906 (ipa_call_summaries): New.
19907 (inline_edge_summary): Remove.
19908 (estimate_edge_growth): Update.
19909 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19910 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19911 * ipa-split.c (execute_split_functions): Update.
19912 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19914 2017-05-23 Tom de Vries <tom@codesourcery.com>
19916 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19917 attributes): Document rdrand effective target.
19919 2017-05-23 Tom de Vries <tom@codesourcery.com>
19921 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19922 attributes): Sort alphabetically.
19924 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19926 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19928 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19931 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19932 V2DF/V2DI splat into two separate patterns, one that handles
19933 registers, and the other that only handles memory. Drop support
19934 for splatting from a GPR on ISA 2.07 and then splitting the
19935 splat into direct move and splat.
19936 (vsx_splat_<mode>_reg): Likewise.
19937 (vsx_splat_<mode>_mem): Likewise.
19939 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19941 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19943 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19945 PR middle-end/80809
19946 * omp-low.c (finish_taskreg_remap): New function.
19947 (finish_taskreg_scan): If unit size of ctx->record_type
19948 is non-constant, unshare the size expression and replace
19949 decls in it with possible outer var refs.
19951 PR middle-end/80809
19952 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19953 GOVD_SHARED rather than GOVD_PRIVATE with it.
19954 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19955 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19957 PR middle-end/80853
19958 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19959 as last argument to build_outer_var_ref for pointer bases of array
19960 section reductions.
19962 2017-05-19 Martin Sebor <msebor@redhat.com>
19964 * print-tree.c (print_node): Print DECL_READ_P flag.
19966 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19968 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19969 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19970 * cgraph.c: Likewise.
19971 * cgraphunit.c: Likewise.
19972 * gengtype.c: Likewise.
19973 * ipa-cp.c: Likewise.
19974 * ipa-devirt.c: Likewise.
19975 * ipa-icf.c: Likewise.
19976 * ipa-predicate.c: Likewise.
19977 * ipa-profile.c: Likewise.
19978 * ipa-prop.c: Likewise.
19979 * ipa-split.c: Likewise.
19981 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19982 edge_predicate_pool, dump_inline_hints,
19983 inline_summary::account_size_time, redirect_to_unreachable,
19984 edge_set_predicate, set_hint_predicate,
19985 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19986 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19987 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19988 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19989 ipa_call_summary_t::remove, initialize_growth_caches,
19990 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19991 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19992 mark_modified, unmodified_parm_1, unmodified_parm,
19993 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19994 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19995 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19996 will_be_nonconstant_predicate, record_modified_bb_info,
19997 get_minimal_bb, record_modified, param_change_prob,
19998 phi_result_unknown_predicate, predicate_for_phi_result,
19999 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20000 estimate_function_body_sizes, compute_inline_parameters,
20001 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20002 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20003 inline_update_callee_summaries, remap_edge_change_prob,
20004 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20005 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20006 inline_analyze_function, inline_summary_t::insert,
20007 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20008 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20009 inline_free_summary): Move to ipa-fnsummary.h
20010 (predicate_t): Remove.
20011 * ipa-fnsummary.c: New file.
20012 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20013 (enum inline_hints_vals, inline_hints, agg_position_info,
20014 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20015 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20016 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20017 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20018 inline_read_summary, inline_write_summary, inline_free_summary,
20019 inline_analyze_function, initialize_inline_failed,
20020 inline_merge_summary, inline_update_overall_summary,
20021 compute_inline_parameters): Move to ipa-fnsummary.h
20022 * ipa-fnsummary.h: New file.
20023 * ipa-inline-transform.h: Include ipa-inline.h.
20024 * ipa-inline.c: LIkewise.
20026 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20028 * ipa-inline.c (edge_badness): Use inlined_time instead of
20029 inline_summaries->get.
20031 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20033 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20035 2017-05-22 Nathan Sidwell <nathan@acm.org>
20037 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20038 (fdump-lang): Document 'raw' option.
20039 * dumpfile.h (TDI_tu): Delete.
20040 * dumpfile.c (dump_files): Remove translation-unit.
20041 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20043 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20045 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20046 command option from $(AWK) call.
20047 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20048 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20049 [FORMAT]: Remove handling of variable.
20050 * config/avr/t-multilib: Regenerate.
20052 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20054 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20056 (dump_inline_summary): Do not print self_time.
20057 (estimate_function_body_sizes): Do not set self_time.
20058 (compute_inline_parameters): Likewise.
20059 (inline_read_section, inline_write_summary): Do not stream self_time.
20060 * ipa-inline.h (inline_summary): Drop self_time.
20062 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20064 * ipa-inline-analysis.c (account_size_time): Rename to ...
20065 (inline_summary::account_size_time): ... this one.
20066 (reset_ipa_call_summary): Turn to ...
20067 (ipa_call_summary::reset): ... this one.
20068 (reset_inline_summary): Turn to ...
20069 (inline_summary::reset): ... this one.
20070 (inline_summary_t::remove): Update.
20071 (inline_summary_t::duplicate): Update.
20072 (ipa_call_summary_t::remove): Update.
20073 (dump_inline_summary): Update.
20074 (estimate_function_body_sizes): Update.
20075 (compute_inline_parameters): Update.
20076 (estimate_node_size_and_time): Update.
20077 (inline_merge_summary): Update.
20078 (inline_update_overall_summary): Update.
20079 (inline_read_section): Update.
20080 (inline_write_summary): Update.
20081 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20082 add account_size_time and reset member functions.
20083 (ipa_call_summary): Add reset function.
20084 * ipa-predicate.h (predicate::operator &): Constify.
20086 2017-05-22 Richard Biener <rguenther@suse.de>
20088 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20090 2017-05-19 Jason Merrill <jason@redhat.com>
20092 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20094 2017-05-19 Marek Polacek <polacek@redhat.com>
20097 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20098 TYPE_OVERFLOW_WRAPS checks.
20100 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20102 * tree-core.h (enum omp_clause_default_kind): Add
20103 "OMP_CLAUSE_DEFAULT_PRESENT".
20104 * tree-pretty-print.c (dump_omp_clause): Handle it.
20105 * gimplify.c (enum gimplify_omp_var_data): Add
20106 "GOVD_MAP_FORCE_PRESENT".
20107 (gimplify_adjust_omp_clauses_1): Map it to
20108 "GOMP_MAP_FORCE_PRESENT".
20109 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20111 * gimplify.c (oacc_default_clause): Clarify.
20113 2017-05-19 Nathan Sidwell <nathan@acm.org>
20115 LANG_HOOK_REGISTER_DUMPS
20116 * toplev.c (general_init): Call register dump lang hook.
20117 * doc/invoke.texi: Document -fdump-lang option family.
20118 * dumpfile.c (dump_files): Remove class dump here.
20119 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20120 * dumpfile.h (tree_dump_index): Remove TDI_class.
20121 * langhooks-def.h (lhd_register_dumps): Declare.
20122 (LANG_HOOKS_REGISTER_DUMPS): Define.
20123 (LANG_HOOKS_INITIALIZER): Add it.
20124 * langhooks.c (lhd_register_dumps): Define.
20125 * langhooks.h (struct lang_hooks): Add register_dumps.
20127 2017-05-19 Nathan Sidwell <nathan@acm.org>
20129 * context.h (context::set_passes): New.
20130 * context.c (context::context): Do not create pass manager.
20131 * toplev.c (general_init): Create pass manager here.
20133 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20135 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20136 use this splitter if two add or or instructions would also work for
20137 the constant we want to generate.
20139 2017-05-19 Richard Biener <rguenther@suse.de>
20142 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20143 predicate evaluation.
20145 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20147 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20149 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20152 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20154 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20155 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20156 (inline_edge_summary_vec): Turn into ...
20157 (ipa_call_summaries): ... this one.
20158 (redirect_to_unreachable, edge_set_predicate,
20159 evaluate_properties_for_edge, inline_summary_alloc,
20160 reset_ipa_call_summary, reset_inline_summary,
20161 inline_summary_t::duplicate): Update.
20162 (inline_edge_duplication_hook): Turn to ...
20163 (ipa_call_summary_t::duplicate): ... this one.
20164 (inline_edge_removal_hook): Turn to ...
20165 (ipa_call_summary_t::remove): ... this one.
20166 (dump_inline_edge_summary): Turn to ...
20167 (dump_ipa_call_summary): ... this one.
20168 (estimate_function_body_sizes): Update.
20169 (inline_update_callee_summaries): Update.
20170 (remap_edge_change_prob): Update.
20171 (remap_edge_summaries): Update.
20172 (inline_merge_summary): Update.
20173 (do_estimate_edge_time): Update.
20174 (inline_generate_summary): Update.
20175 (inline_read_section): Update.
20176 (inline_read_summary): Update.
20177 (inline_free_summary): Update.
20178 * ipa-inline.c (can_inline_edge_p): Update.
20179 (compute_inlined_call_time): Update.
20180 (want_inline_small_function_p): Update.
20181 (edge_badness): Update.
20182 (early_inliner): Update.
20183 * ipa-inline.h (inline_edge_summary): Turn to ...
20184 (ipa_call_summary): ... this one.
20185 (ipa_call_summary_t): New class.
20186 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20187 (ipa_call_summaries): New.
20188 (inline_edge_summary): Remove.
20189 (estimate_edge_growth): Update.
20190 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20191 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20192 * ipa-split.c (execute_split_functions): Update.
20193 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20195 2017-05-19 Richard Biener <rguenther@suse.de>
20197 PR middle-end/80764
20198 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20200 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20202 * config/rs6000/rs6000.c (struct machine_function): Add field
20203 fpr_is_wrapped_separately.
20204 (rs6000_get_separate_components): Use 64 components. Handle the
20205 new FPR components.
20206 (rs6000_components_for_bb): Handle the FPR components.
20207 (rs6000_emit_prologue_components): Handle the FPR components.
20208 (rs6000_emit_epilogue_components): Handle the FPR components.
20209 (rs6000_set_handled_components): Handle the FPR components.
20210 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20211 that are already separately shrink-wrapped.
20212 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20213 that are already separately shrink-wrapped.
20215 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20218 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20221 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20222 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20223 cases where the register allocator uses a move and an offsettable
20224 memory operation to/from a FPR register on ISA 2.06/2.07.
20225 (define_peephole2 for Altivec d-form store): Likewise.
20227 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20230 * config/i386/mmx.md (*mov<mode>_internal): Enable
20231 alternatives 11, 12, 13 and 14 also for 32bit targets.
20232 Remove alternatives 15, 16, 17 and 18.
20233 * config/i386/sse.md (vec_concatv2di): Change
20234 alternative (!x, *y) to (x, ?!*Yn).
20236 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20238 * dumpfile.h (enum dump_kind): Remove stray comma.
20240 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20242 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20243 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20244 predicate::num_conditions
20245 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20246 (CHANGED): turn into predicate::changed.
20247 (agg_position_info): Move to ipa-predicate.h
20248 (add_condition, predicate::add_clause, predicate::operator &=,
20249 predicate::or_with, predicate::evaluate, predicate::probability,
20250 dump_condition, dump_clause, predicate::dump,
20251 predicate::remap_after_duplication, predicate::remap_after_inlining,
20252 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20253 (evaluate_conditions_for_known_args): Update.
20254 (set_cond_stmt_execution_predicate): Update.
20255 * ipa-inline.h: Include ipa-predicate.h
20256 (condition, inline_param_summary, conditions, agg_position_info,
20257 predicate): Move to ipa-predicate.h
20258 * ipa-predicate.c: New file.
20259 * ipa-predicate.h: New file.
20261 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20263 * final.c (leaf_function_p): Check we are not in a sequence.
20265 2017-05-18 Martin Liska <mliska@suse.cz>
20267 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20268 * dumpfile.c (dump_register): Use new enum dump_kind.
20269 (get_dump_file_name): Likewise.
20270 (dump_enable_all): Likewise.
20271 (dump_switch_p_1): Likewise.
20272 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20273 * dumpfile.h (enum dump_kind): New enum type.
20274 (struct dump_file_info): Create constructor and
20275 format fields and comments.
20276 * passes.c (pass_manager::register_one_dump_file):
20278 * statistics.c (statistics_early_init): Likewise.
20279 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20280 TDF_TREE with TDF_SLIM.
20281 (gather_memory_references_ref): Likewise.
20283 2017-05-18 Martin Liska <mliska@suse.cz>
20285 * vec.h (struct vnull): Use it.
20287 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20289 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20290 (true_predicate, false_predicate, true_predicate_p,
20291 false_predicate_p): Remove.
20292 (single_cond_predicate, not_inlined_predicate): Turn to member function
20294 (add_condition): Update.
20295 (add_clause): Turn to...
20296 (predicate::add_clause): ... this one; update; allow passing NULL
20298 (and_predicates): Turn to ...
20299 (predicate::operator &=): ... this one.
20300 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20301 (or_predicates): Turn to ...
20302 (predicate::or_with): ... this one.
20303 (evaluate_predicate): Turn to ...
20304 (predicate::evaluate): ... this one.
20305 (predicate_probability): Turn to ...
20306 (predicate::probability): ... this one.
20307 (dump_condition): Update.
20308 (dump_predicate): Turn to ...
20309 (predicate::dump): ... this one.
20310 (account_size_time): Update.
20311 (edge_set_predicate): Update.
20312 (set_hint_predicate): UPdate.
20313 (evaluate_conditions_for_known_args): Update.
20314 (evaluate_properties_for_edge): Update.
20315 (remap_predicate_after_duplication): Turn to...
20316 (predicate::remap_after_duplication): ... this one.
20317 (remap_hint_predicate_after_duplication): Update.
20318 (inline_summary_t::duplicate): UPdate.
20319 (dump_inline_edge_summary): Update.
20320 (dump_inline_summary): Update.
20321 (set_cond_stmt_execution_predicate): Update.
20322 (set_switch_stmt_execution_predicate): Update.
20323 (compute_bb_predicates): Update.
20324 (will_be_nonconstant_expr_predicate): Update.
20325 (will_be_nonconstant_predicate): Update.
20326 (phi_result_unknown_predicate): Update.
20327 (predicate_for_phi_result): Update.
20328 (array_index_predicate): Update.
20329 (estimate_function_body_sizes): Update.
20330 (estimate_node_size_and_time): Update.
20331 (estimate_ipcp_clone_size_and_time): Update.
20332 (remap_predicate): Rename to ...
20333 (predicate::remap_after_inlining): ... this one.
20334 (remap_hint_predicate): Update.
20335 (inline_merge_summary): Update.
20336 (inline_update_overall_summary): Update.
20337 (estimate_size_after_inlining): Update.
20338 (read_predicate): Rename to ...
20339 (predicate::stream_in): ... this one.
20340 (read_inline_edge_summary): Update.
20341 (write_predicate): Rename to ...
20342 (predicate::stream_out): ... this one.
20343 (write_inline_edge_summary): Update.
20344 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20345 (clause_t): Turn to uint32_t
20346 (predicate): Turn to class; implement constructor and operators
20348 (size_time_entry): Update.
20349 (inline_summary): Update.
20350 (inline_edge_summary): Update.
20352 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20354 * fold-const.c (fold_binary_loc): Move transformation...
20355 * match.pd (C - X CMP X): ... here.
20357 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20359 * config/sparc/sparc.c (sparc_option_override): Set function
20360 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20361 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20363 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20365 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20367 2017-05-18 Marek Polacek <polacek@redhat.com>
20370 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20371 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20372 gimple_assign_load_p.
20374 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20376 PR middle-end/80692
20377 * real.c (do_compare): Give decimal_do_compare preference over
20378 comparing just the signs.
20380 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20382 * doc/md.texi (Canonicalization of Instructions): Describe the
20383 canonical form of instructions that inherently set a condition
20386 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20388 PR middle-end/80775
20389 * tree-cfg.c: Move deletion of unreachable case statements to after
20390 the merging of consecutive case labels.
20392 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20394 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20395 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20396 restoring of callee-saved registers.
20398 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20400 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20401 * config/visium/visium.c (single_set_and_flags): Likewise.
20402 * config/visium/visium.md (Substitutions): Likewise.
20404 2017-05-17 Martin Liska <mliska@suse.cz>
20406 * cfg.c: Introduce dump_flags_t type and
20407 use it instead of int type.
20409 * cfghooks.c: Likewise.
20410 * cfghooks.h (struct cfg_hooks): Likewise.
20411 * cfgrtl.c: Likewise.
20412 * cfgrtl.h: Likewise.
20413 * cgraph.c (cgraph_node::get_body): Likewise.
20414 * coretypes.h: Likewise.
20415 * domwalk.c: Likewise.
20416 * domwalk.h: Likewise.
20417 * dumpfile.c (struct dump_option_value_info): Likewise.
20418 (dump_enable_all): Likewise.
20419 (dump_switch_p_1): Likewise.
20420 (opt_info_switch_p): Likewise.
20421 * dumpfile.h (enum tree_dump_index): Likewise.
20422 (struct dump_file_info): Likewise.
20423 * genemit.c: Likewise.
20424 * generic-match-head.c: Likewise.
20425 * gengtype.c (open_base_files): Likewise.
20426 * gimple-pretty-print.c: Likewise.
20427 * gimple-pretty-print.h: Likewise.
20428 * graph.c (print_graph_cfg): Likewise.
20429 * graphite-scop-detection.c (dot_all_sese): Likewise.
20430 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20431 * loop-unroll.c (report_unroll): Likewise.
20432 * passes.c (pass_manager::register_one_dump_file): Likewise.
20433 * print-tree.c: Likewise.
20434 * statistics.c: Likewise.
20435 * tree-cfg.c: Likewise.
20436 * tree-cfg.h: Likewise.
20437 * tree-dfa.c: Likewise.
20438 * tree-dfa.h: Likewise.
20439 * tree-dump.c (dump_function): Likewise.
20440 * tree-dump.h (struct dump_info): Likewise.
20441 * tree-pretty-print.c: Likewise.
20442 * tree-pretty-print.h: Likewise.
20443 * tree-ssa-live.c: Likewise.
20444 * tree-ssa-live.h: Likewise.
20445 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20446 * tree-vect-loop.c: Likewise.
20447 * tree-vect-slp.c: Likewise.
20449 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20450 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20452 PR tree-optimization/80457
20453 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20454 of all arguments to a statement as scalar_to_vec operations.
20455 (vectorizable_call): Adjust call to vect_model_simple_cost for
20457 (vectorizable_conversion): Likewise.
20458 (vectorizable_assignment): Likewise.
20459 (vectorizable_shift): Likewise.
20460 (vectorizable_operation): Likewise.
20461 (vectorizable_comparison): Likewise.
20462 (vect_is_simple_cond): Record the def types for operands.
20463 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20464 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20465 for statement argument count.
20467 2017-05-16 Carl Love <cel@us.ibm.com>
20469 * config/rs6000/rs6000-c: Add support for built-in functions
20470 vector unsigned long long vec_bperm (vector unsigned long long,
20471 vector unsigned char)
20472 vector signed long long vec_mule (vector signed int,
20474 vector unsigned long long vec_mule (vector unsigned int,
20475 vector unsigned int)
20476 vector signed long long vec_mulo (vector signed int,
20478 vector unsigned long long vec_mulo (vector unsigned int,
20479 vector unsigned int)
20480 vector signed char vec_sldw (vector signed char,
20481 vector signed char,
20483 vector unsigned char vec_sldw (vector unsigned char,
20484 vector unsigned char,
20486 vector signed short vec_sldw (vector signed short,
20487 vector signed short,
20489 vector unsigned short vec_sldw (vector unsigned short,
20490 vector unsigned short,
20492 vector signed int vec_sldw (vector signed int,
20495 vector unsigned int vec_sldw (vector unsigned int,
20496 vector unsigned int,
20498 vector signed long long vec_sldw (vector signed long long,
20499 vector signed long long,
20501 vector unsigned long long vec_sldw (vector unsigned long long,
20502 vector unsigned long long,
20504 * config/rs6000/rs6000-c: Add support for built-in functions
20505 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20506 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20507 * doc/extend.texi: Update the built-in documentation for the
20508 new built-in functions.
20510 2017-05-16 Marek Polacek <polacek@redhat.com>
20514 * tree.c (save_expr): Don't fold the expression.
20516 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
20518 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
20519 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
20520 and (?*y,m). Update insn attributes.
20522 2017-05-16 Martin Liska <mliska@suse.cz>
20524 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
20525 flags argument of print_gimple_stmt, print_gimple_expr,
20526 print_generic_stmt and print_generic_expr.
20527 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
20528 * coretypes.h: Likewise.
20529 * except.c (dump_eh_tree): Likewise.
20530 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
20531 * gimple-pretty-print.h: Likewise.
20532 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
20533 (backprop::push_to_worklist): Likewise.
20534 (backprop::pop_from_worklist): Likewise.
20535 (backprop::process_use): Likewise.
20536 (backprop::intersect_uses): Likewise.
20537 (note_replacement): Likewise.
20538 * gimple-ssa-store-merging.c
20539 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
20540 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
20541 (pass_store_merging::execute): Likewise.
20542 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
20543 (ssa_base_cand_dump_callback): Likewise.
20544 (dump_incr_vec): Likewise.
20545 (replace_refs): Likewise.
20546 (replace_mult_candidate): Likewise.
20547 (create_add_on_incoming_edge): Likewise.
20548 (create_phi_basis): Likewise.
20549 (insert_initializers): Likewise.
20550 (all_phi_incrs_profitable): Likewise.
20551 (introduce_cast_before_cand): Likewise.
20552 (replace_one_candidate): Likewise.
20553 * gimplify.c (gimplify_expr): Likewise.
20554 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
20555 (set_rename): Likewise.
20556 (rename_uses): Likewise.
20557 (copy_loop_phi_nodes): Likewise.
20558 (add_close_phis_to_merge_points): Likewise.
20559 (copy_loop_close_phi_args): Likewise.
20560 (copy_cond_phi_args): Likewise.
20561 (graphite_copy_stmts_from_block): Likewise.
20562 (translate_pending_phi_nodes): Likewise.
20563 * graphite-poly.c (print_pdr): Likewise.
20564 (dump_gbb_cases): Likewise.
20565 (dump_gbb_conditions): Likewise.
20566 (print_scop_params): Likewise.
20567 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
20568 (build_cross_bb_scalars_use): Likewise.
20569 (gather_bbs::before_dom_children): Likewise.
20570 * hsa-dump.c (dump_hsa_immed): Likewise.
20571 * ipa-cp.c (print_ipcp_constant_value): Likewise.
20572 (get_replacement_map): Likewise.
20573 * ipa-inline-analysis.c (dump_condition): Likewise.
20574 (estimate_function_body_sizes): Likewise.
20575 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
20576 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
20577 * ipa-prop.c (ipa_dump_param): Likewise.
20578 (ipa_print_node_jump_functions_for_edge): Likewise.
20579 (ipa_modify_call_arguments): Likewise.
20580 (ipa_modify_expr): Likewise.
20581 (ipa_dump_param_adjustments): Likewise.
20582 (ipa_dump_agg_replacement_values): Likewise.
20583 (ipcp_modif_dom_walker::before_dom_children): Likewise.
20584 * ipa-pure-const.c (check_stmt): Likewise.
20585 (pass_nothrow::execute): Likewise.
20586 * ipa-split.c (execute_split_functions): Likewise.
20587 * omp-offload.c (dump_oacc_loop_part): Likewise.
20588 (dump_oacc_loop): Likewise.
20589 * trans-mem.c (tm_log_emit): Likewise.
20590 (tm_memopt_accumulate_memops): Likewise.
20591 (dump_tm_memopt_set): Likewise.
20592 (dump_tm_memopt_transform): Likewise.
20593 * tree-cfg.c (gimple_verify_flow_info): Likewise.
20594 (print_loop): Likewise.
20595 * tree-chkp-opt.c (chkp_print_addr): Likewise.
20596 (chkp_gather_checks_info): Likewise.
20597 (chkp_get_check_result): Likewise.
20598 (chkp_remove_check_if_pass): Likewise.
20599 (chkp_use_outer_bounds_if_possible): Likewise.
20600 (chkp_reduce_bounds_lifetime): Likewise.
20601 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
20602 (chkp_mark_completed_bounds): Likewise.
20603 (chkp_register_incomplete_bounds): Likewise.
20604 (chkp_mark_invalid_bounds): Likewise.
20605 (chkp_maybe_copy_and_register_bounds): Likewise.
20606 (chkp_build_returned_bound): Likewise.
20607 (chkp_get_bound_for_parm): Likewise.
20608 (chkp_build_bndldx): Likewise.
20609 (chkp_get_bounds_by_definition): Likewise.
20610 (chkp_generate_extern_var_bounds): Likewise.
20611 (chkp_get_bounds_for_decl_addr): Likewise.
20612 * tree-chrec.c (chrec_apply): Likewise.
20613 * tree-data-ref.c (dump_data_reference): Likewise.
20614 (dump_subscript): Likewise.
20615 (dump_data_dependence_relation): Likewise.
20616 (analyze_overlapping_iterations): Likewise.
20617 * tree-inline.c (expand_call_inline): Likewise.
20618 (tree_function_versioning): Likewise.
20619 * tree-into-ssa.c (dump_defs_stack): Likewise.
20620 (dump_currdefs): Likewise.
20621 (dump_names_replaced_by): Likewise.
20622 (dump_update_ssa): Likewise.
20623 (update_ssa): Likewise.
20624 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20625 * tree-parloops.c (build_new_reduction): Likewise.
20626 (try_create_reduction_list): Likewise.
20627 (ref_conflicts_with_region): Likewise.
20628 (oacc_entry_exit_ok_1): Likewise.
20629 (oacc_entry_exit_single_gang): Likewise.
20630 * tree-pretty-print.h: Likewise.
20631 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20632 (get_scalar_evolution): Likewise.
20633 (add_to_evolution): Likewise.
20634 (get_loop_exit_condition): Likewise.
20635 (analyze_evolution_in_loop): Likewise.
20636 (analyze_initial_condition): Likewise.
20637 (analyze_scalar_evolution): Likewise.
20638 (instantiate_scev): Likewise.
20639 (number_of_latch_executions): Likewise.
20640 (gather_chrec_stats): Likewise.
20641 (final_value_replacement_loop): Likewise.
20642 (scev_const_prop): Likewise.
20643 * tree-sra.c (dump_access): Likewise.
20644 (disqualify_candidate): Likewise.
20645 (create_access): Likewise.
20646 (reject): Likewise.
20647 (maybe_add_sra_candidate): Likewise.
20648 (create_access_replacement): Likewise.
20649 (analyze_access_subtree): Likewise.
20650 (analyze_all_variable_accesses): Likewise.
20651 (sra_modify_assign): Likewise.
20652 (initialize_constant_pool_replacements): Likewise.
20653 (find_param_candidates): Likewise.
20654 (decide_one_param_reduction): Likewise.
20655 (replace_removed_params_ssa_names): Likewise.
20656 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20657 * tree-ssa-copy.c (dump_copy_of): Likewise.
20658 (copy_prop_visit_cond_stmt): Likewise.
20659 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20660 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20661 (record_equivalences_from_stmt): Likewise.
20662 * tree-ssa-dse.c (compute_trims): Likewise.
20663 (delete_dead_call): Likewise.
20664 (delete_dead_assignment): Likewise.
20665 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20666 (forward_propagate_into_cond): Likewise.
20667 (pass_forwprop::execute): Likewise.
20668 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20669 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20671 (move_computations_worker): Likewise.
20672 (execute_sm): Likewise.
20673 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20674 (remove_exits_and_undefined_stmts): Likewise.
20675 (remove_redundant_iv_tests): Likewise.
20676 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20677 (adjust_iv_update_pos): Likewise.
20678 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20679 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20680 (value_replacement): Likewise.
20681 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20682 * tree-ssa-pre.c (print_pre_expr): Likewise.
20683 (get_representative_for): Likewise.
20684 (create_expression_by_pieces): Likewise.
20685 (insert_into_preds_of_block): Likewise.
20686 (eliminate_insert): Likewise.
20687 (eliminate_dom_walker::before_dom_children): Likewise.
20688 (eliminate): Likewise.
20689 (remove_dead_inserted_code): Likewise.
20690 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20691 * tree-ssa-reassoc.c (get_rank): Likewise.
20692 (eliminate_duplicate_pair): Likewise.
20693 (eliminate_plus_minus_pair): Likewise.
20694 (eliminate_not_pairs): Likewise.
20695 (undistribute_ops_list): Likewise.
20696 (eliminate_redundant_comparison): Likewise.
20697 (update_range_test): Likewise.
20698 (optimize_range_tests_var_bound): Likewise.
20699 (optimize_vec_cond_expr): Likewise.
20700 (rewrite_expr_tree): Likewise.
20701 (rewrite_expr_tree_parallel): Likewise.
20702 (linearize_expr): Likewise.
20703 (break_up_subtract): Likewise.
20704 (linearize_expr_tree): Likewise.
20705 (attempt_builtin_powi): Likewise.
20706 (attempt_builtin_copysign): Likewise.
20707 (transform_stmt_to_copy): Likewise.
20708 (transform_stmt_to_multiply): Likewise.
20709 (dump_ops_vector): Likewise.
20710 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20711 (print_scc): Likewise.
20712 (set_ssa_val_to): Likewise.
20713 (visit_reference_op_store): Likewise.
20714 (visit_use): Likewise.
20715 (sccvn_dom_walker::before_dom_children): Likewise.
20716 (run_scc_vn): Likewise.
20717 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20719 (expr_hash_elt::print): Likewise.
20720 (const_and_copies::pop_to_marker): Likewise.
20721 (const_and_copies::record_const_or_copy_raw): Likewise.
20722 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20723 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20724 (dump_predicates): Likewise.
20725 (find_uninit_use): Likewise.
20726 (warn_uninitialized_phi): Likewise.
20727 (pass_late_warn_uninitialized::execute): Likewise.
20728 * tree-ssa.c (verify_vssa): Likewise.
20729 (verify_ssa): Likewise.
20730 (maybe_optimize_var): Likewise.
20731 * tree-vrp.c (dump_value_range): Likewise.
20732 (dump_all_value_ranges): Likewise.
20733 (dump_asserts_for): Likewise.
20734 (register_edge_assert_for_2): Likewise.
20735 (vrp_visit_cond_stmt): Likewise.
20736 (vrp_visit_switch_stmt): Likewise.
20737 (vrp_visit_stmt): Likewise.
20738 (vrp_visit_phi_node): Likewise.
20739 (simplify_cond_using_ranges_1): Likewise.
20740 (fold_predicate_in): Likewise.
20741 (evrp_dom_walker::before_dom_children): Likewise.
20742 (evrp_dom_walker::push_value_range): Likewise.
20743 (evrp_dom_walker::pop_value_range): Likewise.
20744 (execute_early_vrp): Likewise.
20746 2017-05-16 Richard Biener <rguenther@suse.de>
20748 * dwarf2out.c (loc_list_from_tree_1): Do not create
20749 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20751 2017-05-16 Richard Biener <rguenther@suse.de>
20753 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20755 (note_variable_value_in_expr): If we resolved the decl ref
20756 do not push to the stack.
20758 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20760 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20761 operations in fast-math mode.
20762 (vaddq_f16): Likewise.
20763 (vmul_f16): Likewise.
20764 (vmulq_f16): Likewise.
20765 (vsub_f16): Likewise.
20766 (vsubq_f16): Likewise.
20767 * config/arm/neon.md (add<mode>3): New.
20769 (fma:<VH:mode>3): New. Also remove outdated comment.
20772 2017-05-16 Martin Liska <mliska@suse.cz>
20776 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20777 (odr_types_equivalent_p): Likewise.
20779 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20781 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20783 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20786 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20787 non-interunit SSE move alternatives with '?'.
20788 (zero-extendsidi peephole2): New peephole to skip intermediate
20789 general register in SSE zero-extend sequence.
20791 2017-05-15 Jeff Law <law@redhat.com>
20793 * reorg.c (relax_delay_slots): Create a new variable to hold
20794 the temporary target rather than clobbering TARGET_LABEL.
20796 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20797 missing argument to extract_bit_field call.
20798 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20800 2017-05-15 Martin Liska <mliska@suse.cz>
20803 * gcc.c (process_command): Do not allow empty argument of -o option.
20805 2017-05-15 Renlin Li <renlin.li@arm.com>
20807 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20808 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20809 * config/aarch64/constraints.md (Usf): Add long call check.
20810 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20811 (call_value): Likewise.
20812 (sibcall): Likewise.
20813 (sibcall_value): Likewise.
20815 (call_value_insn): New.
20816 (sibcall_insn): Update rtx pattern.
20817 (sibcall_value_insn): Likewise.
20818 (call_internal): Remove.
20819 (call_value_internal): Likewise.
20820 (sibcall_internal): Likewise.
20821 (sibcall_value_internal): Likewise.
20822 (call_reg): Likewise.
20823 (call_symbol): Likewise.
20824 (call_value_reg): Likewise.
20825 (call_value_symbol): Likewise.
20827 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20830 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20832 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20834 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20835 compatible with CCGOCmode and with CCZmode.
20837 2017-05-14 Martin Sebor <msebor@redhat.com>
20839 PR middle-end/77671
20840 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20841 (gimple_fold_builtin_snprintf): Same.
20842 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20843 (gimple_fold_builtin_snprintf): Same.
20844 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20845 of character types.
20846 (is_call_safe): New function.
20847 (try_substitute_return_value): Call it.
20848 (try_simplify_call): New function.
20849 (pass_sprintf_length::handle_gimple_call): Call it.
20851 2017-05-14 Martin Sebor <msebor@redhat.com>
20853 PR middle-end/80669
20854 * builtins.c (expand_builtin_stpncpy): Simplify.
20856 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20858 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20859 * config/i386/i386.h
20860 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20861 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20862 (struct machine_function): Add new members call_ms2sysv,
20863 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20864 (struct machine_frame_state): New fields sp_realigned and
20865 sp_realigned_offset.
20866 * config/i386/i386.c
20867 (enum xlogue_stub): New enum.
20868 (enum xlogue_stub_sets): New enum.
20869 (class xlogue_layout): New class.
20870 (struct ix86_frame): New fields stack_realign_allocate_offset,
20871 stack_realign_offset and outlined_save_offset. Modify comments to
20872 detail stack layout when using out-of-line stubs.
20873 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20874 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20875 -mcall-ms2sysv-xlogues.
20876 (stub_managed_regs): New static variable.
20877 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20878 registers managed by out-of-line stub.
20879 (disable_call_ms2sysv_xlogues): New function.
20880 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20881 m->call_ms2sysv when appropriate and compute frame layout for
20883 (sp_valid_at, fp_valid_at): New inline functions.
20884 (choose_basereg): New function.
20885 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20887 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20888 Use align parameter of choose_baseaddr to generated aligned SSE movs
20890 (pro_epilogue_adjust_stack): Modify to track
20891 machine_frame_state::sp_realigned.
20892 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20893 (ix86_nsaved_sseregs): Likewise.
20894 (ix86_emit_save_regs): Likewise.
20895 (ix86_emit_save_regs_using_mov): Likewise.
20896 (ix86_emit_save_sse_regs_using_mov): Likewise.
20897 (get_scratch_register_on_entry): Likewise.
20898 (gen_frame_set): New function.
20899 (gen_frame_load): Likewise.
20900 (gen_frame_store): Likewise.
20901 (emit_outlined_ms2sysv_save): Likewise.
20902 (emit_outlined_ms2sysv_restore): Likewise.
20903 (ix86_expand_prologue): Modify stack re-alignment code and call
20904 emit_outlined_ms2sysv_save when appropriate.
20905 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20906 parameter rtx_insn *insn, which allows the function to be used to only
20907 generate the notes.
20908 (ix86_expand_epilogue): Modify validity checks of frame and stack
20909 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20910 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20911 * config/i386/predicates.md
20912 (save_multiple): New predicate.
20913 (restore_multiple): Likewise.
20914 * config/i386/sse.md
20915 (save_multiple<mode>): New pattern.
20916 (save_multiple_realign<mode>): Likewise.
20917 (restore_multiple<mode>): Likewise.
20918 (restore_multiple_and_return<mode>): Likewise.
20919 (restore_multiple_leave_return<mode>): Likewise.
20920 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20922 2017-05-14 Julia Koval <julia.koval@intel.com>
20924 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20925 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20926 (__builtin_ia32_xsetbv): New builtins.
20927 * config/i386/i386.c (ix86_expand_special_args_builtin):
20929 (ix86_expand_builtin): Special expand for new intrinsics.
20930 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20931 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20932 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20934 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20936 * cfganal.c (inverted_post_order_compute): Change argument type
20938 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20939 * df-core.c (rest_of_handle_df_initialize): Adjust.
20940 (rest_of_handle_df_finish): Likewise.
20941 (df_analyze_1): Likewise.
20942 (df_analyze): Likewise.
20943 (loop_inverted_post_order_compute): Change argument to be a vec *.
20944 (df_analyze_loop): Adjust.
20945 (df_get_n_blocks): Likewise.
20946 (df_get_postorder): Likewise.
20947 * df.h (struct df_d): Change field to be a vec.
20948 * lcm.c (compute_laterin): Adjust.
20949 (compute_available): Likewise.
20950 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20951 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20952 * tree-ssa-pre.c (compute_antic): Likewise.
20954 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20956 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20957 (depth_first_search::depth_first_search): Change structure init
20958 function to this constructor.
20959 (depth_first_search::add_bb): Rename function to this member.
20960 (depth_first_search::execute): Likewise.
20961 (flow_dfs_compute_reverse_finish): Adjust.
20963 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20965 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20966 (longest_simple_path): Likewise.
20967 * shrink-wrap.c (spread_components): Likewise.
20968 (disqualify_problematic_components): Likewise.
20969 (emit_common_heads_for_components): Likewise.
20970 (emit_common_tails_for_components): Likewise.
20971 (insert_prologue_epilogue_for_components): Likewise.
20973 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20975 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20978 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20980 * df-core.c (df_set_blocks): Start using auto_bitmap.
20981 (df_compact_blocks): Likewise.
20982 * df-problems.c (df_rd_confluence_n): Likewise.
20983 * df-scan.c (df_insn_rescan_all): Likewise.
20984 (df_process_deferred_rescans): Likewise.
20985 (df_update_entry_block_defs): Likewise.
20986 (df_update_exit_block_uses): Likewise.
20987 (df_entry_block_bitmap_verify): Likewise.
20988 (df_exit_block_bitmap_verify): Likewise.
20989 (df_scan_verify): Likewise.
20990 * lra-constraints.c (lra_constraints): Likewise.
20991 (undo_optional_reloads): Likewise.
20992 (lra_undo_inheritance): Likewise.
20993 * lra-remat.c (calculate_gen_cands): Likewise.
20994 (do_remat): Likewise.
20995 * lra-spills.c (assign_spill_hard_regs): Likewise.
20996 (spill_pseudos): Likewise.
20997 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20998 (bitmap_set_subtract_values): Likewise.
21000 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21002 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21003 management with auto_bitmap.
21004 (fix_inter_tick): Likewise.
21005 (fix_recovery_deps): Likewise.
21006 * ira.c (add_store_equivs): Likewise.
21007 (find_moveable_pseudos): Likewise.
21008 (split_live_ranges_for_shrink_wrap): Likewise.
21009 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21010 (rtx_reuse_manager::seen_def_p): Likewise.
21011 (rtx_reuse_manager::set_seen_def): Likewise.
21012 * print-rtl.h (class rtx_reuse_manager): Likewise.
21014 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21016 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21018 (migrate_btr_def): Likewise.
21019 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21020 * df-core.c (loop_post_order_compute): Likewise.
21021 (loop_inverted_post_order_compute): Likewise.
21022 * hsa-common.h: Likewise.
21023 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21024 * init-regs.c (initialize_uninitialized_regs): Likewise.
21025 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21026 (inline_small_functions): Likewise.
21027 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21028 * ira.c (combine_and_move_insns): Likewise.
21029 (build_insn_chain): Likewise.
21030 * loop-invariant.c (find_invariants): Likewise.
21031 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21032 * predict.c (tree_predict_by_opcode): Likewise.
21033 (predict_paths_leading_to): Likewise.
21034 (predict_paths_leading_to_edge): Likewise.
21035 (estimate_loops_at_level): Likewise.
21036 (estimate_loops): Likewise.
21037 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21038 (spread_components): Likewise.
21039 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21040 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21041 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21042 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21043 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21044 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21045 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21046 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21047 (mark_threaded_blocks): Likewise.
21048 (thread_through_all_blocks): Likewise.
21049 * tree-ssa.c (verify_ssa): Likewise.
21050 (execute_update_addresses_taken): Likewise.
21051 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21053 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21055 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21057 (post_order_compute): Likewise.
21058 (inverted_post_order_compute): Likewise.
21059 (pre_and_rev_post_order_compute_fn): Likewise.
21061 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21063 * genrecog.c (int_set::int_set): Explicitly construct our
21064 auto_vec base class.
21065 * vec.h (auto_vec::auto_vec): New constructor.
21067 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21069 * bitmap.h (class auto_bitmap): New constructor taking
21070 bitmap_obstack * argument.
21072 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21074 * bitmap.h (class auto_bitmap): Change type of m_bits to
21075 bitmap_head, and adjust ctor / dtor and member operators.
21077 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21079 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21080 when returned register mode doesn't match original mode.
21082 2017-05-12 Jeff Law <law@redhat.com>
21083 Jakub Jelinek <jakub@redhat.com>
21085 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21086 we look for cc setter after the compare-elim changes.
21087 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21088 within the vector to match what compare-elim now expects.
21089 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21090 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21092 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21093 after the compare-elim changes.
21094 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21095 the vector to match what compare-elim now expects.
21096 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21097 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21098 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21099 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21100 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21102 * config/visium/visium.c (single_set_and_flags): Fix where
21103 we look for cc setter after the compare-elim changes.
21104 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21105 with the vector to match what compare-elim now expects.
21106 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21107 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21108 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21109 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21110 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21111 (neg<mode>2_insn_set_overflow): Likewise.
21113 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21115 PR middle-end/79794
21116 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21117 maybe_expand_insn call, set ops[0].target. If still set after call,
21118 set alt_rtl. Add extra arg to recursive calls.
21119 (extract_bit_field): Add alt_rtl argument. Pass to
21121 * expmed.h (extract_bit_field): Fix prototype.
21122 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21123 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21124 to extract_bit_field_calls.
21125 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21126 Pass alt_rtl to extract_bit_field calls.
21127 * calls.c (store_unaligned_arguments_into_psuedos)
21128 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21129 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21131 * optabs.h (struct expand_operand): Add target bitfield.
21133 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21135 * compare-elim.c (try_eliminate_compare): Canonicalize
21136 operation with embedded compare to
21137 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21138 (set (reg) (operation)].
21140 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21142 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21145 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21146 cost of adding a carry flag for ADC instruction.
21147 [case MINUS]: Ignore the cost of subtracting a carry flag
21148 for SBB instruction.
21150 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21152 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21154 * config/rs6000/bmiintrin.h: New file.
21155 * config/rs6000/bmi2intrin.h: New file.
21156 * config/rs6000/x86intrin.h: New file.
21158 2017-05-12 Jeff Law <law@redhat.com>
21160 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21163 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21165 PR middle-end/80707
21166 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21168 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21170 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21171 early expansion of vector divide builtins.
21172 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21173 builtins identified as having unsigned arguments.
21175 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21177 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21178 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21179 expansion of vector logical operations (and, andc, or, xor,
21182 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21184 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21185 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21187 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21189 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21190 early GIMPLE expansion of vector multiplies.
21192 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21194 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21195 TARGET_HAVE_MOVT conditional.
21196 (movt splitter): Likewise.
21198 2017-05-12 Richard Biener <rguenther@suse.de>
21200 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21201 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21202 Fold all stmts not inplace.
21204 2017-05-12 Richard Biener <rguenther@suse.de>
21206 PR tree-optimization/80713
21207 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21208 inserted_exprs bit for not removed stmts.
21210 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21212 PR middle-end/69921
21213 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21214 parallelized" attribute for parallelized OpenACC kernels.
21215 * omp-offload.c (execute_oacc_device_lower): Use it.
21217 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21218 Set "oacc kernels" attribute.
21219 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21220 parameter. Adjust all users.
21221 (oacc_fn_attrib_kernels_p): Remove function.
21222 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21223 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21224 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21225 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21226 assert "oacc kernels" attribute is set.
21228 2017-05-11 Carl Love <cel@us.ibm.com>
21230 * config/rs6000/rs6000-c: Add support for built-in functions
21231 vector unsigned char vec_popcnt (vector signed char)
21232 vector unsigned char vec_popcnt (vector unsigned char)
21233 vector unsigned short vec_popcnt (vector signed short)
21234 vector unsigned short vec_popcnt (vector unsigned short)
21235 vector unsigned int vec_popcnt (vector signed int)
21236 vector unsigned int vec_popcnt (vector unsigned int)
21237 vector unsigned long long vec_popcnt (vector signed long long)
21238 vector unsigned long long vec_popcnt (vector unsigned long long)
21239 vector signed long long vec_slo (vector signed long long,
21240 vector signed char)
21241 vector signed long long vec_slo (vector signed long long,
21242 vector unsigned char)
21243 vector unsigned long long vec_slo (vector unsigned long long,
21244 vector signed char)
21245 vector unsigned long long vec_slo (vector unsigned long long,
21246 vector unsigned char)
21247 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21248 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21249 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21250 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21251 * doc/extend.texi: Update the built-in documentation file for the
21252 new built-in functions.
21254 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21256 * attribs.h (sorted_attr_string): Move machine independent
21257 functions for target clone support from the i386 port to common
21258 code. Rename ix86_function_versions to common_function_versions.
21259 Rename make_name to make_unique_name.
21260 (common_function_versions): Likewise.
21261 (make_unique_name): Likewise.
21262 (make_dispatcher_decl): Likewise.
21263 (is_function_default_version): Likewise.
21264 * attribs.c (attr_strcmp): Likewise.
21265 (sorted_attr_string): Likewise.
21266 (common_function_versions): Likewise.
21267 (make_unique_name): Likewise.
21268 (make_dispatcher_decl): Likewise.
21269 (is_function_default_version): Likewise.
21270 * config/i386/i386.c (attr_strcmp): Likewise.
21271 (sorted_attr_string): Likewise.
21272 (ix86_function_versions): Likewise.
21273 (make_name): Likewise.
21274 (make_dispatcher_decl): Likewise.
21275 (is_function_default_version): Likewise.
21276 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21278 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21281 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21282 Account for direct move costs for vec_construct of integer
21285 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21288 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21289 (UNSPEC_STX_ATOMIC): Ditto.
21290 (loaddi_via_sse): New insn.
21291 (storedi_via_sse): Ditto.
21292 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21293 Update corresponding peephole2 patterns.
21294 (atomic_storedi_fpu): Ditto.
21296 2017-05-11 Julia Koval <julia.koval@intel.com>
21298 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21299 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21301 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21302 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21303 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21305 2017-05-11 Nathan Sidwell <nathan@acm.org>
21307 * graphite-poly.c: Include dumpfile.h.
21309 * dumpfle.h (dump_function): Declare here ...
21310 * tree-dump.h (dump_function): ... not here.
21311 * dumpfile.c: #include tree-cfg.h.
21312 (dump_function): Move here from ...
21313 * tree-dump.c (dump_function): ... here.
21314 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21315 * graphite-poly.c: Don't include tree-dump.h.
21316 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21317 * print-tree.c: Likewise.
21318 * stor-layout.c: Likewise.
21319 * tree-nested.c: Likewise.
21321 * dumpfile.c (dump_start): Use TDF_FLAGS.
21322 (dump_enable_all): Fix TDF_KIND check thinko.
21324 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21326 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21327 array entries to represent two legal parameterizations of the
21328 overloaded __builtin_cmpb function, as represented by the
21329 P6_OV_BUILTIN_CMPB constant.
21330 (altivec_resolve_overloaded_builtin): Add special case handling
21331 for the __builtin_cmpb function, as represented by the
21332 P6_OV_BUILTIN_CMPB constant.
21333 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21334 (BU_P6_64BIT_2): New macro.
21335 (BU_P6_OVERLOAD_2): New macro
21336 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21337 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21338 (CMPB): Add overload support to represent both 32-bit and 64-bit
21339 compare-bytes function.
21340 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21341 support for TARGET_CMPB.
21342 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21343 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21344 documentation of the __builtin_cmpb overloaded built-in function.
21346 2017-05-11 Richard Biener <rguenther@suse.de>
21348 PR tree-optimization/80705
21349 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21350 bases are not vectorizable.
21352 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21354 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21355 when counting register pressure.
21357 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21359 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21360 (iv_ca_more_deps): Renamed to ...
21361 (iv_ca_compare_deps): ... this.
21362 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21364 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21366 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21368 (determine_group_iv_costs): ... here.
21369 (find_inv_vars_cb): Record inv var if it's not recorded before.
21371 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21373 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21374 (get_shiftadd_cost): Ditto.
21376 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21378 * tree-ssa-address.c: Include header file.
21379 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21381 (add_to_parts): Refactor.
21382 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21383 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21386 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21388 PR tree-optimization/53090
21389 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21391 (extract_cond_operands): Detect condition with IV on both sides
21392 and return COMP_IV_EXPR_2.
21393 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21394 (rewrite_use_compare): Simplify by removing call to function
21395 extract_cond_operands.
21397 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21399 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21400 (extract_cond_operands): Detect condition comparing against non-
21401 invariant bound and return appropriate enum value.
21402 (find_interesting_uses_cond): Update use of extract_cond_operands.
21403 Handle its return value accordingly.
21404 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21406 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21408 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21409 nonlinear iv_use computation in loop invariant sensitive way.
21411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21413 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21414 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21416 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21418 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21419 (dump_cand): Support iv_cand.inv_exprs.
21420 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21422 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21425 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21427 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21429 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21430 as local function. Include necessary header files.
21431 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21433 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21435 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21437 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21439 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21440 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21441 RSHIFT_EXPR and BIT_NOT_EXPR.
21443 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21445 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21446 (adjust_setup_cost): New parameter supporting round up adjustment.
21447 (struct address_cost_data): Delete.
21448 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21449 (split_address_cost, ptr_difference_cost): Delete.
21450 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21451 (struct ainc_cost_data): New struct.
21452 (get_address_cost_ainc): New function.
21453 (get_address_cost, get_computation_cost): Reimplement.
21454 (determine_group_iv_cost_address): Record inv_expr for all uses of
21456 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21457 (iv_ca_has_deps): Reimplemented to ...
21458 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21460 (iv_ca_extend): Call iv_ca_more_deps.
21462 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21464 * tree-ssa-address.c (struct mem_address): Move to header file.
21465 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21466 * tree-ssa-address.h (struct mem_address): Move from C file.
21467 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21469 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21471 * tree-affine.h (aff_combination_type): New interface.
21472 (aff_combination_zero_p): Remove static.
21473 (aff_combination_const_p): New interface.
21474 (aff_combination_singleton_var_p): New interfaces.
21476 2017-05-11 Richard Biener <rguenther@suse.de>
21478 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21479 Skip unreachable blocks and destinations.
21480 (eliminate): Move stmt removal and fixup ...
21481 (fini_eliminate): ... here. Skip inserted exprs.
21482 (pass_pre::execute): Move fini_pre after fini_eliminate.
21483 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21484 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21485 PRE to get rid of dead code that has invalid SSA form and
21486 split critical edges again.
21488 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21490 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21492 2017-05-11 Richard Biener <rguenther@suse.de>
21494 * passes.c (execute_function_todo): Verify loops if they are
21495 said to be up-to-date.
21496 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21497 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21499 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21502 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21503 handle calling assemble_external ourself.
21506 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21507 modes with zero size. Enhance comment.
21509 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21511 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21512 built-ins for vec_xl and vec_xst with short and char pointer
21515 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21517 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
21518 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
21519 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
21520 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
21521 (_mm_maskz_min_round_ss): New intrinsics.
21522 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
21523 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
21524 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
21525 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
21526 (__builtin_ia32_minss_mask_round): New builtins.
21527 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21528 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21529 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
21531 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
21532 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
21534 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
21537 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21539 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
21540 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
21541 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
21542 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
21543 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
21544 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
21545 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
21546 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21547 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21548 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
21549 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
21550 (__builtin_ia32_mulss_mask_round): New builtins.
21551 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21552 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21553 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
21555 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
21556 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21558 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21561 2017-05-10 Julia Koval <julia.koval@intel.com>
21563 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
21564 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
21565 (_mm256_setr_m128i): New intrinsics.
21567 2017-05-10 Julia Koval <julia.koval@intel.com>
21569 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
21570 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
21571 (_mm_maskz_rcp14_ss): New intrinsics.
21572 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
21573 (__builtin_ia32_rcp14ss_mask): New builtins.
21574 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
21576 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
21578 PR tree-optimization/51513
21579 * tree-cfg.c (gimple_seq_unreachable_p): New function.
21580 (assert_unreachable_fallthru_edge_p): Use it.
21581 (group_case_labels_stmt): Likewise.
21582 * tree-cfg.h: Prototype it.
21583 * stmt.c: Include cfghooks.h and tree-cfg.h.
21584 (emit_case_dispatch_table) <gap_label>: New local variable.
21585 Use it to fill dispatch table gaps.
21586 Test for default_label before updating probabilities.
21587 (expand_case) <default_label>: Remove unneeded initialization.
21588 Test for unreachable default case statement and remove its edge.
21589 Set default_label accordingly.
21590 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
21592 2017-05-10 Carl Love <cel@us.ibm.com>
21594 * config/rs6000/rs6000-c: Add support for built-in functions
21595 vector signed char vec_neg (vector signed char)
21596 vector signed short int vec_neg (vector short int)
21597 vector signed int vec_neg (vector signed int)
21598 vector signed long long vec_neg (vector signed long long)
21599 vector float vec_neg (vector float)
21600 vector double vec_neg (vector double)
21601 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21603 * config/rs6000/altivec.h: Add define for vec_neg
21604 * doc/extend.texi: Update the built-in documentation for the
21605 new built-in functions.
21607 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21609 PR tree-optimization/77644
21610 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21612 2017-05-10 Nathan Sidwell <nathan@acm.org>
21614 * dumpfile.h (TDI_lang_all): New.
21615 (TDF_KIND): New. Renumber others
21616 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21618 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21620 (get_dump_file_name): Adjust suffix generation.
21621 (dump_enable_all): Use TDF_KIND.
21622 * doc/invoke.texi (-fdump-lang-all): Document.
21624 * dumpfile.h: Tabify.
21626 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21629 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21630 Move member access before delete.
21632 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21634 * tree-inline.c (expand_call_inline): Split block at stmt
21637 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21640 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21641 are now unused after splitting mov{sf,sd}_hardfloat.
21642 (f32_lr2): Likewise.
21643 (f32_lm): Likewise.
21644 (f32_lm2): Likewise.
21645 (f32_li): Likewise.
21646 (f32_li2): Likewise.
21647 (f32_lv): Likewise.
21648 (f32_sr): Likewise.
21649 (f32_sr2): Likewise.
21650 (f32_sm): Likewise.
21651 (f32_sm2): Likewise.
21652 (f32_si): Likewise.
21653 (f32_si2): Likewise.
21654 (f32_sv): Likewise.
21655 (f32_dm): Likewise.
21656 (f32_vsx): Likewise.
21657 (f32_av): Likewise.
21658 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21659 For movsf, order stores so the VSX stores occur before the GPR
21660 store which encourages the register allocator to use a traditional
21661 FPR instead of a GPR. For movsd, order the stores so that the GPR
21662 store comes before the VSX stores to allow the power6 to work.
21663 This is due to the power6 not having a 32-bit integer store
21664 instruction from a FPR.
21665 (movsf_hardfloat): Likewise.
21666 (movsd_hardfloat): Likewise.
21668 2017-05-09 Martin Sebor <msebor@redhat.com>
21670 PR translation/80280
21671 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21674 PR translation/80280
21675 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21676 data member added in r247778.
21677 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21679 2017-05-09 Nathan Sidwell <nathan@acm.org>
21681 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21683 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21686 2017-05-09 Marek Polacek <polacek@redhat.com>
21688 * doc/invoke.texi: Fix typo.
21690 2017-05-09 Richard Biener <rguenther@suse.de>
21692 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21693 (vrp_val_is_min): Likewise.
21694 (set_value_range_to_value): Likewise.
21695 (set_value_range_to_nonnegative): Likewise.
21696 (gimple_assign_nonzero_p): Likewise.
21697 (gimple_stmt_nonzero_p): Likewise.
21698 (vrp_int_const_binop): Likewise. Remove unreachable case.
21699 (adjust_range_with_scev): Adjust comments.
21700 (compare_range_with_value): Likewise.
21701 (extract_range_from_phi_node): Likewise.
21702 (test_for_singularity): Likewise.
21704 2017-05-09 Richard Biener <rguenther@suse.de>
21706 * tree-vrp.c (get_single_symbol): Add assert that we don't
21707 get overflowed constants as invariant part.
21708 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21709 checks. Use wi::cmp instead of recursing for integer constants.
21710 (compare_values): Just ignore whether we assumed undefined
21711 overflow instead of failing the compare.
21712 (extract_range_for_var_from_comparison_expr): Add comment before the
21713 TREE_NO_WARNING sets.
21714 (test_for_singularity): Likewise.
21715 (extract_range_from_comparison): Do not disable optimization
21716 when we assumed undefined overflow.
21717 (extract_range_basic): Remove init of unused var.
21719 2017-05-09 Richard Biener <rguenther@suse.de>
21721 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21722 (extract_range_from_multiplicative_op_1): Adjust.
21723 (extract_range_from_binary_expr_1): Use int_const_binop.
21725 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21728 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21729 rs6000_store_data_bypass_p in seven define_bypass directives and
21730 in several comments.
21731 * config/rs6000/rs6000-protos.h: Add prototype for
21732 rs6000_store_data_bypass_p function.
21733 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21734 function implements slightly different (rs6000-specific) semantics
21735 than store_data_bypass_p, returning false rather than aborting
21736 with assertion error when arguments do not satisfy the
21737 requirements of store data bypass.
21738 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21739 rs6000_store_data_bypass_p.
21741 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21743 * config/xtensa/xtensa-protos.h
21744 (xtensa_initial_elimination_offset): New declaration.
21745 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21746 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21747 macro definition, add case for FRAME_POINTER_REGNUM when
21748 FRAME_GROWS_DOWNWARD.
21749 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21750 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21751 xtensa_initial_elimination_offset.
21753 2017-05-08 Nathan Sidwell <nathan@acm.org>
21755 * doc/invoke.texi: Alphabetize -fdump options.
21757 2017-05-08 Martin Sebor <msebor@redhat.com>
21759 PR translation/80280
21760 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21762 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21764 * target.def (compute_frame_layout): New optional target hook.
21765 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21766 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21767 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21769 * reload1.c (verify_initial_elim_offsets): Likewise.
21770 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21771 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21772 (arm_get_frame_offsets): Split up into this ...
21773 (arm_compute_frame_layout): ... and this function.
21775 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21777 * config/aarch64/constraints.md (Usa): New constraint.
21778 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21780 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21782 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21783 with_multilib_list after it has been checked.
21785 2017-05-08 Richard Biener <rguenther@suse.de>
21787 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21788 (bitmap_set_subtract_values): Likewise.
21790 2017-05-08 Richard Biener <rguenther@suse.de>
21792 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21793 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21795 (gimple_stmt_nonzero_warnv_p): Rename to ...
21796 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21798 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21799 (extract_range_basic): Adjust, do not disable propagation on
21800 strict overflow sensitive simplification.
21801 (vrp_visit_cond_stmt): Likewise.
21803 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21805 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21806 body size unconditionally.
21808 2017-05-07 Jeff Law <law@redhat.com>
21811 2017-05-06 Jeff Law <law@redhat.com>
21812 PR tree-optimization/78496
21813 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21816 PR tree-optimization/78496
21817 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21818 (simplify_stmt_using_ranges): Call it.
21819 (vrp_dom_walker::before_dom_children): Extract equivalences
21820 from an ASSERT_EXPR with an equality comparison against a
21823 2017-05-06 Jeff Law <law@redhat.com>
21825 PR tree-optimization/78496
21826 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21829 PR tree-optimization/78496
21830 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21831 (simplify_stmt_using_ranges): Call it.
21832 (vrp_dom_walker::before_dom_children): Extract equivalences
21833 from an ASSERT_EXPR with an equality comparison against a
21836 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21838 * lra-constraints.c (lra_copy_reg_equiv): New function.
21839 (split_reg): Use it to copy equivalence information from the
21840 original register to the spill register.
21842 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21844 PR rtl-optimization/75964
21845 * simplify-rtx.c (simplify_const_relational_operation): Remove
21846 invalid handling of comparisons of integer ABS.
21848 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21850 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21851 initialize to zero.
21852 (init_regs): Remove declaration.
21853 (function_arg_advance_32): Initialize error_p as boolean variable.
21855 2017-05-05 Nathan Sidwell <nathan@acm.org>
21857 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21858 lines. Use for (;;).
21860 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21862 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21863 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21864 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21865 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21866 VF=2 that require versioning.
21868 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21870 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21873 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21875 * diagnostic.h (diagnostic_override_option_index): Convert from
21876 macro to inline function.
21878 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21880 * diagnostic.c (last_module_changed_p): New function.
21881 (set_last_module): New function.
21882 (diagnostic_report_current_module): Convert macro usage to
21883 the above functions.
21884 * diagnostic.h (diagnostic_context::last_module): Strengthen
21885 from const line_map * to const line_map_ordinary *.
21886 (diagnostic_last_module_changed): Delete macro.
21887 (diagnostic_set_last_module): Delete macro.
21889 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21891 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21892 with diagnostic_report_diagnostic.
21893 (diagnostic_n_impl_richloc): Likewise.
21894 * diagnostic.h (report_diagnostic): Delete macro.
21895 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21896 with diagnostic_report_diagnostic.
21897 * substring-locations.c (format_warning_va): Likewise.
21899 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21901 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21902 save/restor of format_spec. Move option-printing code to...
21903 (print_option_information): ...this new function, and
21904 reimplement by simply printing to the pretty_printer,
21905 rather than appending to the format string.
21907 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21909 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21910 handling logic into...
21911 (update_effective_level_from_pragmas): ...this new function.
21913 2017-05-04 Andrew Waterman <andrew@sifive.com>
21915 * config/riscv/riscv.opt (mstrict-align): New option.
21916 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21917 (SLOW_UNALIGNED_ACCESS): Define.
21918 (riscv_slow_unaligned_access): Declare.
21919 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21921 (riscv_slow_unaligned_access): New variable.
21922 (rocket_tune_info): Set slow_unaligned_access to true.
21923 (optimize_size_tune_info): Set slow_unaligned_access to false.
21924 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21925 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21926 (riscv_option_override): Set riscv_slow_unaligned_access.
21927 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21929 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21931 * config/riscv/riscv.md: Unify indentation.
21933 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21938 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21940 (extendsi<mode>2): Add support for doing sign extension via
21941 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21942 don't have ISA 3.0 instructions.
21943 (extendsi<mode>2 splitter): Likewise.
21944 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21945 generate the normal insns since SImode can now go in vector
21946 registers. Disallow the special UNSPECs needed for previous
21947 machines to hide SImode being used. Add new insns
21948 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21949 (fix_trunc<mode>si2_stfiwx): Likewise.
21950 (fix_trunc<mode>si2_internal): Likewise.
21951 (fixuns_trunc<mode>si2): Likewise.
21952 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21953 (fctiw<u>z_<mode>_smallint): Likewise.
21954 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21955 of floating point to 32-bit integer from doing a direct move to
21956 the GPR registers to do a store.
21957 (fctiwz_<mode>): Break long line.
21959 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21961 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21962 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21963 (addr_list, addr_offset_valid_p): New.
21964 (split_address_groups): Check offset validity with above function.
21965 (gt-tree-ssa-loop-ivopts.h): Include header file.
21967 2017-05-05 Nathan Sidwell <nathan@acm.org>
21969 * config.gcc (arm*-*-*): Add missing 'fi'.
21971 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21973 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21974 included in -fopt-info does not matter.
21975 * doc/optinfo.texi (-fopt-info): Fix description of default
21976 behavour. Explicitly say order of options included in -fopt-info
21979 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21981 * config.gcc: Allow combinations of aprofile and rmprofile values for
21982 --with-multilib-list.
21983 * config/arm/t-multilib: New file.
21984 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21985 variables. Remove setting of ISA and floating-point ABI in
21986 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21987 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21988 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21989 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21991 * config/arm/t-rmprofile: Likewise except for the matches changes.
21992 * doc/install.texi (--with-multilib-list): Document the combination of
21993 aprofile and rmprofile values and warn about pitfalls in doing that.
21995 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21997 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21998 (movdi_aarch64): Likewise.
22000 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22002 PR tree-optimization/80632
22003 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22005 (build_arrays): Initialize it for virtual phis.
22006 (fix_phi_nodes): Use it for virtual phis.
22008 PR tree-optimization/80558
22009 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22010 [x, y] op z into [x op, y op z] for op & or | if conditions
22013 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22014 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22017 * config/arm/arm.md (use_literal_pool): Remove.
22018 (64-bit immediate split): No longer takes cost into consideration
22019 if arm_disable_literal_pool is enabled.
22020 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22021 used when arm_disable_literal_pool is enabled.
22022 (arm_max_const_double_inline_cost): Remove use of
22023 arm_disable_literal_pool.
22024 (push_minipool_fix): Add assert.
22025 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22026 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22027 (no_literal_pool_sf_immediate): New.
22029 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22031 PR tree-optimization/80613
22032 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22033 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22035 2017-05-05 Richard Biener <rguenther@suse.de>
22037 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22039 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22041 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22042 of this flag from insn conditions due to removal from r247495.
22044 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22046 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22048 (arm_early_store_addr_dep_ptr): Likewise.
22049 * config/arm/aarch-common-protos.h
22050 (arm_early_load_addr_dep_ptr): Add prototype.
22051 (arm_early_store_addr_dep_ptr): Likewise.
22052 * config/arm/cortex-a53.md: Add new bypasses.
22054 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22056 * tree.c (next_type_uid): Change type to unsigned.
22057 (type_hash_canon): Decrement back next_type_uid if
22058 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22059 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22062 2017-05-04 Martin Sebor <msebor@redhat.com>
22064 * builtins.c: Fix a trivial typo in a comment.
22066 PR middle-end/79234
22067 * builtins.c (check_sizes): Adjust to handle reading past the end.
22068 Avoid printing excessive upper bound of ranges. Use %E to print
22069 tree nodes instead of converting them to %wu.
22070 (expand_builtin_memchr): New function.
22071 (compute_dest_size): Rename...
22072 (compute_objsize): ...to this.
22073 (expand_builtin_memcpy): Adjust.
22074 (expand_builtin_mempcpy): Adjust.
22075 (expand_builtin_strcat): Adjust.
22076 (expand_builtin_strcpy): Adjust.
22077 (check_strncat_sizes): Adjust.
22078 (expand_builtin_strncat): Adjust.
22079 (expand_builtin_strncpy): Adjust and simplify.
22080 (expand_builtin_memset): Adjust.
22081 (expand_builtin_bzero): Adjust.
22082 (expand_builtin_memcmp): Adjust.
22083 (expand_builtin): Handle memcmp.
22084 (maybe_emit_chk_warning): Check strncat just once.
22086 2017-05-04 Martin Sebor <msebor@redhat.com>
22088 PR preprocessor/79214
22089 PR middle-end/79222
22090 PR middle-end/79223
22091 * builtins.c (check_sizes): Add inlining context and issue
22092 warnings even when -Wno-system-headers is set.
22093 (check_strncat_sizes): Same.
22094 (expand_builtin_strncat): Same.
22095 (expand_builtin_memmove): New function.
22096 (expand_builtin_stpncpy): Same.
22097 (expand_builtin): Handle memmove and stpncpy.
22099 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22101 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22102 which is not used any more.
22104 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22106 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22108 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22110 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22111 (cortexa53_tunings): Likewise.
22112 (cortexa57_tunings): Likewise.
22113 (cortexa72_tunings): Likewise.
22114 (cortexa73_tunings): Likewise.
22116 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22118 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22119 Set loop alignment to 8.
22121 2017-05-04 Martin Sebor <msebor@redhat.com>
22123 PR translation/80280
22124 * builtins.c (expand_builtin_object_size): Add missing quoting to
22125 %D and like directives.
22126 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22127 (hsa_type_for_tree_type): Same.
22128 (verify_function_arguments): Same.
22129 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22130 * varasm.c (get_section): Same.
22133 2017-05-04 Martin Sebor <msebor@redhat.com>
22135 PR translation/80280
22136 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22138 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22140 * config/aarch64/aarch64.c (generic_addrcost_table):
22141 Change HI/TI mode setting.
22143 2017-05-04 Martin Jambor <mjambor@suse.cz>
22145 PR tree-optimization/80622
22146 * tree-sra.c (comes_initialized_p): New function.
22147 (build_accesses_from_assign): Only set write lazily when
22148 comes_initialized_p is false.
22149 (analyze_access_subtree): Use comes_initialized_p.
22150 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22151 instead of testing for PARM_DECL.
22153 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22155 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22156 constraint on operand 0 to allow more general addressing modes.
22157 Adjust output template.
22158 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22160 * config/aarch64/aarch64-protos.h
22161 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22162 * config/aarch64/constraints.md (Dp): New address constraint.
22163 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22166 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22168 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22169 update use of estimate_ipcp_clone_size_and_time.
22170 (estimate_local_effects): Update use of
22171 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22172 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22173 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22174 Return nonspecialized time.
22176 2017-05-04 Richard Biener <rguenther@suse.de>
22178 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22179 for the last VUSE which def dominates the PHI. Directly call
22181 (get_continuation_for_phi_1): Remove.
22183 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22185 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22186 to explain the use of truncating division. Cap the number of
22187 iterations to the maximum given by nb_iterations_upper_bound,
22190 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22192 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22193 * configure: Regenerate.
22194 * config.in: Regenerate.
22195 * config/i386/driver-mingw32.c: new file.
22196 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22197 * config.host: Link driver-mingw32.o on MinGW host.
22198 * doc/install.texi: Document new --enable-mingw-wildcard configure
22201 2017-05-04 Marek Polacek <polacek@redhat.com>
22203 PR tree-optimization/80612
22204 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22206 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22207 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22209 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22210 (movt splitter): Likewise.
22211 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22212 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22213 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22214 block for Thumb-1 with MOVT.
22215 (thumb2_legitimate_address_p): Move code block ...
22216 (can_avoid_literal_pool_for_label_p): ... into this new function.
22217 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22219 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22220 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22221 "M-profile targets with the MOVT instruction".
22223 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22225 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22226 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22227 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22229 2017-05-04 Martin Liska <mliska@suse.cz>
22231 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22232 variable cond_code.
22234 2017-05-04 Richard Biener <rguenther@suse.de>
22236 * tree.c (array_at_struct_end_p): Handle arrays at struct
22237 end with flexarrays more conservatively. Refactor and treat
22238 arrays of arrays or aggregates more strict. Fix
22239 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22240 * tree.c (array_at_struct_end_p): Adjust prototype.
22241 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22242 * gimple-fold.c (get_range_strlen): Likewise.
22243 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22245 2017-05-04 Richard Biener <rguenther@suse.de>
22247 PR tree-optimization/31130
22248 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22250 (supports_overflow_infinity): Likewise.
22251 (is_negative_overflow_infinity): Likewise.
22252 (is_positive_overflow_infinity): Likewise.
22253 (is_overflow_infinity): Likewise.
22254 (stmt_overflow_infinity): Likewise.
22255 (overflow_infinity_range_p): Likewise.
22256 (usable_range_p): Remove as always returning true.
22257 (make_overflow_infinity): Remove.
22258 (negative_overflow_infinity): Likewise.
22259 (positive_overflow_infinity): Likewise.
22260 (avoid_overflow_infinity): Likewise.
22261 (set_value_range): Adjust accordingly.
22262 (set_value_range_to_nonnegative): Likewise, remove now unused
22263 overflow_infinity arg.
22264 (vrp_operand_equal_p): Adjust.
22265 (update_value_range): Likewise.
22266 (range_int_cst_singleton_p): Likewise.
22267 (operand_less_p): Likewise.
22268 (compare_values_warnv): Likewise.
22269 (extract_range_for_var_from_comparison_expr): Likewise.
22270 (vrp_int_const_binop): Likewise.
22271 (zero_nonzero_bits_from_vr): Likewise.
22272 (extract_range_from_multiplicative_op_1): Likewise.
22273 (extract_range_from_binary_expr_1): Likewise.
22274 (extract_range_from_unary_expr): Likewise.
22275 (extract_range_from_comparison): Likewise.
22276 (extract_range_basic): Likewise.
22277 (adjust_range_with_scev): Likewise.
22278 (compare_ranges): Likewise.
22279 (compare_range_with_value): Likewise.
22280 (dump_value_range): Likewise.
22281 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22283 (simplify_cond_using_ranges): Adjust.
22285 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22287 * brig-builtins.def: Added a builtin for class_f64.
22288 * builtin-types.def: Added a builtin type needed by class_f64.
22290 2017-05-03 Jason Merrill <jason@redhat.com>
22292 * timevar.def: Add TV_CONSTEXPR.
22294 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22296 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22298 2017-05-03 Martin Jambor <mjambor@suse.cz>
22300 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22301 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22302 * ipa-cp.c (ipcp_propagate_stage): Do not call
22303 ipa_update_after_lto_read.
22304 * ipa-inline.c (ipa_inline): Likewise.
22306 2017-05-03 Martin Jambor <mjambor@suse.cz>
22308 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22309 tag. Added a default constructor and a destructor.
22310 (ipa_edge_args_sum_t): New class;
22311 (ipa_edge_args_sum): Declare.
22312 (ipa_edge_args_vector): Remove declaration.
22313 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22314 (ipa_free_edge_args_substructures): Remove declaration.
22315 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22316 (ipa_edge_args_info_available_for_edge_p): Likewise.
22317 * ipa-prop.c (ipa_edge_args_vector): Removed.
22318 (edge_removal_hook_holder): Likewise.
22319 (edge_duplication_hook_holder): Likewise.
22320 (ipa_edge_args_sum): New variable.
22321 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22322 ipa_edge_args_vector.
22323 (ipa_free_edge_args_substructures): Likewise.
22324 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22325 ipa_edge_args_vector.
22326 (ipa_edge_removal_hook): Turned into method
22327 ipa_edge_args_sum_t::remove.
22328 (ipa_edge_duplication_hook): Turned into method
22329 ipa_edge_args_sum_t::duplicate.
22330 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22331 registering edge hooks.
22332 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22333 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22334 ipa_edge_args_sum instead of ipa_edge_args_vector.
22335 * ipa-profile.c (ipa_profile): Likewise.
22337 2017-05-03 Martin Jambor <mjambor@suse.cz>
22339 * symbol-summary.h (function_summary): New method exists.
22340 (function_summary::symtab_removal): Deallocate through release.
22341 (call_summary): New class.
22342 (gt_ggc_mx): New overload.
22343 (gt_pch_nx): Likewise.
22344 (gt_pch_nx): Likewise.
22346 2017-05-03 Jeff Law <law@redhat.com>
22348 PR tree-optimization/78496
22349 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22350 from simplify_cond_using_ranges. Split off code to walk
22351 backwards through casts into ...
22352 (simplify_cond_using_ranges_2): New function.
22353 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22354 (execute_vrp): After identifying jump threads, call
22355 simplify_cond_using_ranges_2.
22357 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22360 * ipa-inline.h (inline_summary): Add ctor.
22361 (create_ggc): Do not use ggc_cleared_alloc.
22363 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22364 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22366 * gcc.c (handle_braces): Support escaping in switch matching
22368 * doc/invoke.texi (Spec Files): Document it.
22369 Remove superfluous @code markup in items.
22371 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22373 * diagnostic-show-locus.c (struct column_range): New struct.
22374 (get_affected_columns): New function.
22375 (get_printed_columns): New function.
22376 (struct correction): New struct.
22377 (correction::ensure_capacity): New function.
22378 (correction::ensure_terminated): New function.
22379 (struct line_corrections): New struct.
22380 (line_corrections::~line_corrections): New dtor.
22381 (line_corrections::add_hint): New function.
22382 (layout::print_trailing_fixits): Reimplement in terms of the new
22384 (selftest::test_overlapped_fixit_printing): New function.
22385 (selftest::diagnostic_show_locus_c_tests): Call it.
22387 2017-05-03 Nathan Sidwell <nathan@acm.org>
22389 Canonicalize canonical type hashing
22390 * tree.h (type_hash_canon_hash): Declare.
22391 * tree.c (type_hash_list, attribute_hash_list): Move into
22392 type_hash_canon_hash.
22393 (build_type_attribute_qual_variant): Break out hash code calc into
22394 type_hash_canon_hash.
22395 (type_hash_canon_hash): New. Generic type hash computation.
22396 (build_range_type_1, build_array_type_1, build_function_type,
22397 build_method_type_directly, build_offset_type, build_complex_type,
22398 make_vector_type): Call it.
22400 2017-05-03 Richard Biener <rguenther@suse.de>
22402 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22403 When all DRs have unknown misaligned do not always peel
22404 when there is a store but apply the same costing model as if
22405 there were only loads.
22407 2017-05-03 Richard Biener <rguenther@suse.de>
22410 PR tree-optimization/80492
22411 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22412 compare_base_decls returning dont-know properly.
22414 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22416 * config/arm/iterators.md (CCSI): New mode iterator.
22417 (arch): New mode attribute.
22418 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22419 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22420 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22421 code iterator for success result mode.
22422 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22423 the corresponding new insn generators.
22425 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22429 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22431 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22433 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22434 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22435 (DDR_A): Wrap DDR argument in brackets.
22436 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22437 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22438 (DDR_REVERSED_P): Likewise.
22440 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22442 PR tree-optimization/79472
22443 * tree-switch-conversion.c (struct switch_conv_info): Add
22444 contiguous_range and default_case_nonstandard fields.
22445 (collect_switch_conv_info): Compute contiguous_range and
22446 default_case_nonstandard fields, don't clear final_bb if
22447 contiguous_range and only the default case doesn't have the required
22449 (check_all_empty_except_final): Set default_case_nonstandard instead
22450 of failing if contiguous_range and the default case doesn't have empty
22452 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22453 and only the default case doesn't have the required constants. Skip
22455 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22456 if default_case_nonstandard.
22457 (build_constructors): Build constant 1 just once. Assert that default
22458 values aren't inserted in between cases if contiguous_range. Skip
22460 (build_arrays): Skip virtual phis.
22461 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22462 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22463 Handle virtual phis.
22464 (gen_inbound_check): Handle default_case_nonstandard case.
22465 (process_switch): Adjust check_final_bb caller. Call
22466 gather_default_values with the first non-default case instead of
22467 default case if default_case_nonstandard.
22469 2017-05-02 Nathan Sidwell <nathan@acm.org>
22471 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22472 check. Fix formatting.
22474 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22476 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22477 errors when comparing specialized and unspecialized times.
22479 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22481 * diagnostic-show-locus.c
22482 (layout::should_print_annotation_line_p): Make private.
22483 (layout::print_annotation_line): Make private.
22484 (layout::annotation_line_showed_range_p): Make private.
22485 (layout::show_ruler): Make private.
22486 (layout::print_source_line): Make private. Pass in line and
22487 line_width, rather than calling location_get_source_line. Drop
22489 (layout::print_leading_fixits): New method.
22490 (layout::print_any_fixits): Rename to...
22491 (layout::print_trailing_fixits): ...this, and make private.
22492 Don't print newline fixits.
22493 (diagnostic_show_locus): Move logic for printing one row into...
22494 (layout::print_line): ...this new function. Move the
22495 location_get_source_line call and error-handling from
22496 print_source_line to here. Call print_leading_fixits, and rename
22497 print_any_fixits to print_trailing_fixits.
22498 (selftest::test_fixit_insert_containing_newline): Update now that
22499 newlines are partially supported.
22500 (selftest::test_fixit_insert_containing_newline_2): New test.
22501 (selftest::test_fixit_replace_containing_newline): Update comments.
22502 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22503 * edit-context.c (class added_line): New class.
22504 (class edited_line): Describe newline handling in comment.
22505 (edited_line::actually_edited_p): New method.
22506 (edited_line::print_content): Delete redundant decl.
22507 (edited_line::m_predecessors): New field.
22508 (edited_file::print_content): Call edited_line::print_content.
22509 (edited_file::print_diff): Update to support newlines.
22510 (edited_file::print_diff_hunk): Likewise.
22511 (edited_file::print_run_of_changed_lines): New function.
22512 (edited_file::print_diff_line): Convert to...
22513 (print_diff_line): ...this.
22514 (edited_file::get_effective_line_count): New function.
22515 (edited_line::edited_line): Initialize new field m_predecessors.
22516 (edited_line::~edited_line): Clean up m_predecessors.
22517 (edited_line::apply_fixit): Handle newlines.
22518 (edited_line::get_effective_line_count): New function.
22519 (edited_line::print_content): New function.
22520 (edited_line::print_diff_lines): New function.
22521 (selftest::test_applying_fixits_insert_containing_newline): New
22523 (selftest::test_applying_fixits_replace_containing_newline): New
22525 (selftest::insert_line): New function.
22526 (selftest::test_applying_fixits_multiple_lines): Add example of
22528 (selftest::edit_context_c_tests): Call the new tests.
22530 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22532 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
22533 parameter cand. Update dump information.
22534 (get_computation_cost): Update uses.
22536 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22538 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
22539 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
22540 (get_computation_at, rewrite_use_address): Update use of
22541 get_computation_aff.
22543 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22545 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
22546 (get_computation): Delete.
22547 (get_computation_cost): Implement like get_computation_cost_at.
22548 Use get_computation_at.
22549 (get_computation_cost_at): Delete.
22550 (rewrite_use_nonlinear_expr): Use get_computation_at.
22551 (rewrite_use_compare, remove_unused_ivs): Ditto.
22553 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22555 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
22557 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22559 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
22560 (ivopts_global_cost_for_size): Rename parameter and update uses.
22561 (iv_ca_recount_cost): Update uses.
22562 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
22563 candidates seperately in n_invs and n_cands.
22564 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
22566 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22568 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
22569 (find_inv_vars_cb): New.
22570 (find_depends): Renamed to ...
22571 (find_inv_vars): ... this.
22572 (add_candidate_1, force_var_cost): Call find_inv_vars.
22573 (split_address_cost, determine_group_iv_cost_cond): Ditto.
22575 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22577 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
22578 inv_vars. Add inv_exprs.
22579 (struct iv_cand): Rename depends_on to inv_vars.
22580 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
22581 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
22582 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
22583 (dump_cand): Dump inv_vars.
22584 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
22585 (record_invariant, find_depends, add_candidate_1): Ditto.
22586 (set_group_iv_cost, force_var_cost): Ditto.
22587 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
22588 (get_computation_cost_at, get_computation_cost): Ditto.
22589 (determine_group_iv_cost_generic): Ditto.
22590 (determine_group_iv_cost_address): Ditto.
22591 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
22592 (determine_group_iv_costs): Ditto.
22593 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
22594 (iv_ca_set_remove_invariants): Renamed to ...
22595 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
22596 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
22597 (iv_ca_set_add_invariants): Renamed to ...
22598 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
22599 (iv_ca_set_cp): Use iv_ca_set_add_invs.
22600 (iv_ca_has_deps): Support inv_vars and inv_exprs.
22601 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
22602 (create_new_ivs): Remove useless dump.
22604 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22606 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22608 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22609 (iv_ca_set_no_cp, create_new_iv): Ditto.
22611 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22613 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22615 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22617 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22618 function tree_check2.
22620 2017-05-02 Martin Liska <mliska@suse.cz>
22622 * doc/gcov.texi: Add missing preposition.
22623 * gcov.c (function_info::function_info): Properly fill up
22624 all member variables.
22626 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22628 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22630 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22632 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22634 2017-05-02 Martin Liska <mliska@suse.cz>
22637 * lto-streamer-in.c (lto_read_tree_1): Remove
22638 LTO_STREAMER_DEBUG.
22639 * lto-streamer.c (struct tree_hash_entry): Likewise.
22640 (struct tree_entry_hasher): Likewise.
22641 (tree_entry_hasher::hash): Likewise.
22642 (tree_entry_hasher::equal): Likewise.
22643 (lto_streamer_init): Likewise.
22644 (lto_orig_address_map): Likewise.
22645 (lto_orig_address_get): Likewise.
22646 (lto_orig_address_remove): Likewise.
22647 * lto-streamer.h: Likewise.
22648 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22649 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22651 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22653 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22654 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22655 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22656 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22657 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22658 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22659 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22660 (mm_maskz_sub_ss): New intrinsics.
22661 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22662 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22663 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22664 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22665 (__builtin_ia32_subss_mask_round): New builtins.
22666 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22667 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22668 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22670 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22671 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22673 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22676 2017-05-02 Martin Jambor <mjambor@suse.cz>
22678 PR tree-optimization/78687
22679 * tree-sra.c (access): New field parent.
22680 (process_subtree_disqualification): New function.
22681 (disqualify_candidate): Call it.
22682 (build_accesses_from_assign): Reset write flag if creating an
22684 (build_access_subtree): Fill in parent field and also prpagate
22685 down grp_write flag.
22686 (create_artificial_child_access): New parameter set_grp_write, set
22687 grp_write to its value.
22688 (propagate_subaccesses_across_link): Also propagate grp_write flag
22690 (propagate_all_subaccesses): Push the closest parent back to work
22691 queue if add_access_to_work_queue returned true.
22693 2017-05-02 Richard Biener <rguenther@suse.de>
22695 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22696 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22697 -fstrict-overflow documentation.
22698 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22699 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22700 flag_strict_overflow.
22701 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22702 * lto-opts.c (lto_write_options): Do not stream it.
22703 * lto-wrapper.c (merge_and_complain): Do not handle it.
22704 * opts.c (default_options_table): Do not set -fstrict-overflow.
22705 (finish_options): Likewise do not clear it when sanitizing.
22706 * simplify-rtx.c (simplify_const_relational_operation): Do not
22707 test flag_strict_overflow.
22709 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22711 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22712 using enabled attribute.
22713 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22714 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22715 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22716 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22717 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22718 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22719 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22720 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22721 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22722 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22724 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22726 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22728 2017-05-02 Richard Biener <rguenther@suse.de>
22730 PR tree-optimization/80591
22732 2017-04-10 Richard Biener <rguenther@suse.de>
22734 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22737 2017-05-02 Richard Biener <rguenther@suse.de>
22739 PR tree-optimization/80549
22740 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22741 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22742 headers if they do not have a preheader.
22744 2017-05-02 Martin Liska <mliska@suse.cz>
22747 * common.opt: Fix typo.
22748 * doc/invoke.texi: Likewise.
22750 2017-05-01 Jan Beulich <jbeulich@suse.com>
22752 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22753 swapping, add (x,x,m,x,n) alternative.
22755 2017-05-01 Nathan Sidwell <nathan@acm.org>
22757 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22758 unnecessary unadjusted_alignment check.
22760 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22763 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22765 * gimplify.c (gimplify_cilk_detach): New function.
22766 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22767 * tree-core.h: Document EXPR_CILK_SPAWN.
22768 * tree.h (EXPR_CILK_SPAWN): Define.
22770 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22772 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22773 to use new fixit_hint representation, using the "replace" logic.
22774 (get_line_span_for_fixit_hint): Likewise.
22775 (layout::print_any_fixits): Likewise.
22776 (selftest::test_one_liner_many_fixits): Rename to...
22777 (selftest::test_one_liner_many_fixits_1): ...this, and update
22778 comment and expected output to reflect that the multiple fix-it
22779 hints are now consolidated into one insertion.
22780 (selftest::test_one_liner_many_fixits_2): New test.
22781 (selftest::test_diagnostic_show_locus_one_liner): Update for
22783 (selftest::test_fixit_consolidation): Update for fix-it API
22785 * diagnostic.c (print_parseable_fixits): Likewise.
22786 * edit-context.c (edited_line::m_line_events): Convert from
22787 auto_vec <line_event *> to auto_vec <line_event>.
22788 (class line_event): Convert from abstract base class to a concrete
22789 class, taking over the role of replace_event.
22790 (class insert_event): Delete.
22791 (class replace_event): Rename to class line_event. Convert to
22793 (edit_context::add_fixits): Reimplement.
22794 (edit_context::apply_insert): Delete.
22795 (edit_context::apply_replace): Rename to...
22796 (edit_context::apply_fixit): ...this. Convert to half-open range.
22797 (edited_file::apply_insert): Delete.
22798 (edited_file::apply_replace): Rename to...
22799 (edited_file::apply_fixit): ...this.
22800 (edited_line::~edited_line): Drop deletion of events.
22801 (edited_line::apply_insert): Delete.
22802 (edited_line::apply_replace): Rename to...
22803 (edited_line::apply_fixit): ...this. Convert to half-open range.
22804 Update for change to type of m_line_events.
22805 * edit-context.h (edit_context::apply_insert): Delete.
22806 (edit_context::apply_replace): Rename to...
22807 (edit_context::apply_fixit): ...this.
22809 2017-05-01 Martin Sebor <msebor@redhat.com>
22811 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22814 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22817 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22818 __get_cpuid_max returns 0.
22819 (__get_cpuid_count): Ditto.
22821 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22823 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22824 replacement expression is another instance of one of its arguments.
22826 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22829 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22830 check for stack push/pop autoinc.
22831 * config/i386/i386.c (ix86_agi_dependent): Return false
22832 if the only reason why modified_in_p returned true is that
22833 addr is SP based and set_insn is a push or pop.
22835 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22837 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22840 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22843 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22844 (account_size_time): Use two predicates - exec_pred and
22846 (evaluate_conditions_for_known_args): Compute both clause and
22848 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22849 (inline_summary_t::duplicate): Update.
22850 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22852 (compute_inline_parameters): Likewise.
22853 (estimate_edge_size_and_time): Update caluclation of time.
22854 (estimate_node_size_and_time): Compute both time and nonspecialized
22856 (estimate_ipcp_clone_size_and_time): Update.
22857 (inline_merge_summary): Update.
22858 (do_estimate_edge_time): Update.
22859 (do_estimate_edge_size): Update.
22860 (do_estimate_edge_hints): Update.
22861 (inline_read_section, inline_write_summary): Stream both new predicates.
22862 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22864 (compute_inlined_call_time): Cleanup.
22865 (big_speedup_p): Update.
22866 (edge_badness): Update.
22867 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22868 (size_time_entry): Replace predicate by exec_predicate and
22869 nonconst_predicate.
22870 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22871 (estimate_edge_time): Return also nonspec_time.
22872 (reset_edge_growth_cache): Update.
22874 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22876 PR rtl-optimization/80491
22877 * ifcvt.c (noce_process_if_block): When looking for x setter
22878 with missing else_bb, don't check only the insn right before
22879 cond_earliest, but look for the last insn that x is modified in
22880 within the same bb.
22882 PR rtl-optimization/80491
22883 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22885 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22887 PR tree-optimization/80487
22888 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22890 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22892 PR tree-optimization/79697
22893 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22894 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22895 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22897 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22898 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22900 2017-04-28 Martin Sebor <msebor@redhat.com>
22902 PR tree-optimization/80523
22903 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22904 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22906 (maybe_warn, format_directive, parse_directive): Use new functions.
22907 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22909 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22911 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22913 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22915 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22916 target_header_dir): Set correctly.
22917 * configure: Regenerated.
22918 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22919 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22920 instead of SYSTEM_HEADER_DIR.
22922 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22924 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22925 (estimate_local_effects): Likewise.
22926 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22927 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22928 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22929 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22930 do_estimate_edge_time, estimate_edge_time): Likewise.
22931 * ipa-inline-analysis.c (estimate_node_size_and_time,
22932 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22933 (estimate_time_after_inlining): Remove.
22935 2017-04-28 Martin Liska <mliska@suse.cz>
22937 * doc/gcov.texi: Enhance documentation of gcov.
22939 2017-04-28 Martin Liska <mliska@suse.cz>
22941 * doc/gcov.texi: Sort options in alphabetic order.
22942 * doc/gcov-dump.texi: Likewise.
22943 * doc/gcov-tool.texi: Likewise.
22944 * gcov.c (print_usage): Likewise.
22945 * gcov-dump.c (print_usage): Likewise.
22946 * gcov-tool.c (print_merge_usage_message): Likewise.
22947 (print_rewrite_usage_message): Likewise.
22948 (print_overlap_usage_message): Likewise.
22950 2017-04-28 Martin Liska <mliska@suse.cz>
22952 PR gcov-profile/53915
22953 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22955 2017-04-28 Martin Liska <mliska@suse.cz>
22957 PR gcov-profile/79891
22958 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22959 is marked by compiler as living on a line.
22960 (get_cycles_count): Remove usage of the union.
22961 (output_intermediate_file): Likewise.
22962 (find_source): Fix GNU coding style.
22963 (accumulate_line_counts): Remove old non-all block mode.
22964 (output_lines): Remove usage of the union.
22965 * profile.c (output_location): Include all BBs, even if
22966 belonging to a same line (and file) as a previous BB.
22968 2017-04-28 Martin Liska <mliska@suse.cz>
22970 * gcov.c (process_args): Handle new argument 'w'.
22971 (read_graph_file): Assign ID to BBs.
22972 (output_branch_count): Display BB # if verbose flag is set.
22973 (output_lines): Likewise for arcs.
22974 (print_usage): Add '--verbose' option help.
22975 * doc/gcov.texi: Document --verbose (-w) option.
22977 2017-04-28 Martin Liska <mliska@suse.cz>
22979 * gcov.c (struct block_location_info): New struct.
22980 (process_file): Fill up the new structure.
22981 (read_graph_file): Replace usage of encoding by the newly added
22983 (add_line_counts): Likewise.
22984 (accumulate_line_counts): Remove usage of the union.
22985 (function_info::function_info): New function.
22986 (function_info::~function_info): Likewise.
22987 (process_file): Call delete instead of release_function.
22988 (release_function): Release the function.
22989 (release_structures): Call delete instead of release_function.
22990 (solve_flow_graph): Replace usage of num_blocks.
22991 (find_exception_blocks): Likewise.
22992 (output_lines): Fix GNU coding style.
22994 2017-04-28 Martin Liska <mliska@suse.cz>
22997 * coverage.c (coverage_remove_note_file): New function.
22998 * coverage.h: Declare the function.
22999 * toplev.c (finalize): Clean if an error has been seen.
23001 2017-04-28 Martin Liska <mliska@suse.cz>
23003 PR gcov-profile/80031
23004 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23005 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23006 * gcov.c (read_graph_file): Read just number of blocks.
23007 * profile.c (branch_prob): Do not stream 0 flags per a basic
23010 2017-04-28 Martin Liska <mliska@suse.cz>
23012 * gcov-dump.c (tag_*): Add new argument to declarations.
23013 (dump_gcov_file): Likewise.
23014 (tag_blocks): Add and use new argument depth.
23015 (tag_arcs): Likewise.
23016 (tag_lines): Likewise.
23017 (tag_counters): Likewise.
23018 (tag_summary): Likewise.
23019 (dump_working_sets): Use depth to do a proper indentation.
23021 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23024 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23025 * symtab.c (symtab_node::debug_symtab): Move definition here.
23027 2017-04-28 Richard Biener <rguenther@suse.de>
23029 * lto-streamer.h (LTO_major_version): Bump to 7.
23031 2017-04-28 Richard Biener <rguenther@suse.de>
23033 * tree-vrp.c (assert_info): New struct.
23034 (add_assert_info): New helper.
23035 (register_edge_assert_for_2): Refactor to add asserts to a vector
23037 (register_edge_assert_for_1): Likewise.
23038 (register_edge_assert_for): Likewise.
23039 (finish_register_edge_assert_for): New helper actually registering
23040 asserts where live on edge.
23041 (find_conditional_asserts): Adjust.
23042 (find_switch_asserts): Likewise.
23043 (evrp_dom_walker::try_find_new_range): Generalize.
23044 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23046 2017-04-27 Marek Polacek <polacek@redhat.com>
23049 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23050 arg10 and arg11 to itype.
23052 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23054 * doc/extend.texi (Object Size Checking): Improve grammar.
23056 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23059 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23060 that the logic for permitting reciprocal estimates matches that
23063 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23066 * tree.c (type_cache_hasher::equal): Only compare
23067 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23068 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23069 non-aggregate element types.
23070 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23071 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23073 2017-04-27 Richard Biener <rguenther@suse.de>
23075 PR middle-end/80533
23076 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23077 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23078 keeping a reference to a trailing array.
23080 2017-04-27 Richard Biener <rguenther@suse.de>
23082 PR middle-end/80539
23083 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23084 being in loop-closed SSA form conservatively.
23085 (chrec_fold_multiply_poly_poly): Likewise.
23087 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23089 PR middle-end/79665
23090 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23091 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23093 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23096 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23097 (aarch64_function_arg_alignment): Return unsigned int again, but still
23098 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23099 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23100 Don't emit -Wpsabi note.
23101 (aarch64_function_arg_boundary): Likewise.
23102 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23105 2017-04-26 Nathan Sidwell <nathan@acm.org>
23107 * tree.h (crc32_unsigned_n): Declare.
23108 (crc32_unsigned, crc32_unsigned): Make inline.
23109 * tree.c (crc32_unsigned_bits): Replace with ...
23110 (crc32_unsigned_n): ... this.
23111 (crc32_unsigned, crc32_byte): Remove.
23112 (crc32_string): Remove unnecessary braces.
23114 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23116 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23117 * ipa-inline-analysis.c (MAX_TIME): Remove.
23118 (account_size_time): Use sreal for time.
23119 (dump_inline_summary): Update.
23120 (estimate_function_body_sizes): Update.
23121 (estimate_edge_size_and_time): Update.
23122 (estimate_calls_size_and_time): Update.
23123 (estimate_node_size_and_time): Update.
23124 (inline_merge_summary): Update.
23125 (inline_update_overall_summary): Update.
23126 (estimate_time_after_inlining): Update.
23127 (inline_read_section): Update.
23128 (inline_write_summary): Update.
23129 * ipa-inline.c (compute_uninlined_call_time): Update.
23130 (compute_inlined_call_time): Update.
23131 (recursive_inlining): Update.
23132 (inline_small_functions): Update.
23133 (dump_overall_stats): Update.
23134 * ipa-inline.h: Include sreal.h.
23135 (size_time_entry): Turn time to sreal.
23136 (inline_summary): Turn self_time nad time to sreal.
23138 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23140 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23142 (sreal::stream_out, sreal::stream_in): New.
23143 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23145 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23147 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23150 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23153 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23155 (dimode_scalar_chain::compute_convert_gain): Ditto.
23156 (dimode_scalar_chain::make_vector_copies): Ditto.
23157 (dimode_scalar_chain::convert_reg): Ditto.
23158 (dimode_scalar_chain::convert_insn): Ditto.
23159 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23160 (VI248_AVX512BW_1): New mode iterator.
23161 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23162 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23165 2017-04-25 Martin Sebor <msebor@redhat.com>
23167 PR tree-optimization/80497
23168 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23169 constants are representable in HOST_WIDE_INT.
23170 (parse_directive): Ditto.
23172 2017-04-25 Martin Sebor <msebor@redhat.com>
23175 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23176 (new_zero_array): Adjust signature.
23177 (dom_info::dom_init): Used unsigned rather that size_t.
23178 (dom_info::dom_info): Same.
23180 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23181 Jakub Jelinek <jakub@redhat.com>
23184 * config/arm/arm.c: Include gimple.h.
23185 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23186 returns negative, increment ncrn only if it returned positive.
23187 (arm_needs_doubleword_align): Return int instead of bool,
23188 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23189 members, but if there is any such non-FIELD_DECL
23190 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23191 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23192 returns negative, increment nregs only if it returned positive.
23193 (arm_setup_incoming_varargs): Likewise.
23194 (arm_function_arg_boundary): Emit -Wpsabi note if
23195 arm_needs_doubleword_align returns negative, return
23196 DOUBLEWORD_ALIGNMENT only if it returned positive.
23198 2017-04-25 Marek Polacek <polacek@redhat.com>
23201 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23202 first argument to type.
23204 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23207 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23208 type checks to test for compatibility instead of equality.
23210 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23211 Jakub Jelinek <jakub@redhat.com>
23214 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23216 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23217 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23218 the alignment computation, but return their maximum in warn_alignment.
23219 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23220 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23222 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23223 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23226 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23228 * config/arc/simdext.md (dmpyh): Fix typo.
23230 2017-04-25 Richard Biener <rguenther@suse.de>
23232 PR tree-optimization/80492
23233 * alias.c (compare_base_decls): Handle registers with asm
23234 specification conservatively.
23235 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23236 compare_base_decls returning dont-know properly.
23238 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23240 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23241 (legitimate_offset_address_p): New function.
23242 (arc_legitimate_address_p): Use above function.
23244 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23246 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23248 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23250 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23251 ACCH registers whenever they are available.
23253 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23255 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23256 double regs fix when not used.
23258 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23260 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23262 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23263 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23265 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23267 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23268 when emitting short ADD instructions.
23270 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23272 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23274 (cmpsi_cc_c_insn): Likewise.
23275 (cbranchsi4_scratch): Compute proper instruction length using
23276 compact_hreg_operand.
23277 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23279 2017-04-25 Richard Biener <rguenther@suse.de>
23281 PR middle-end/80509
23282 * passes.c (pass_manager::pass_manager): Initialize
23283 m_name_to_pass_map.
23285 2017-04-25 Richard Biener <rguenther@suse.de>
23287 PR tree-optimization/79201
23288 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23290 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23293 * config/s390/vector.md: Split MEM->GPR vector moves for
23294 non-s_operand addresses.
23296 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23299 * config/s390/predicates.md (reload_const_wide_int_operand): New
23301 * config/s390/s390.md ("movti"): Remove d/P alternative.
23302 ("movti_bigconst"): New pattern definition.
23304 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23307 * s390-protos.h (s390_expand_cs_hqi): Removed.
23308 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23309 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23310 modes as well as CCZ1mode and CCZmode.
23311 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23312 signature of s390_emit_compare_and_swap.
23313 (s390_expand_cs_hqi): Likewise, make static.
23314 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23315 compare-and-swap, in some cases.
23316 (s390_expand_cs): Wrapper function.
23317 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23319 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23320 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23321 patterns for small and large integers. Forbid symref memory operands.
23322 Move expander to s390.c. Require cc register.
23323 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23324 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23325 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23326 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23327 symref memory operands. Remove CC mode and call s390_match_ccmode
23329 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23331 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23333 * config/s390/s390.md (define_peephole2): New peephole to help
23334 combining the load-and-test pattern with volatile memory.
23336 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23338 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23339 with CCZmode for TARGET_Z196.
23341 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23343 PR rtl-optimization/80501
23344 * combine.c (make_compound_operation_int): Set subreg_code to SET
23345 even for AND with mask of the sign bit of mode.
23347 PR rtl-optimization/80500
23348 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23349 sum's initial value.
23351 2017-04-25 Julian Brown <julian@codesourcery.com>
23352 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23354 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23356 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23358 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23360 2017-04-25 Julian Brown <julian@codesourcery.com>
23361 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23363 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23364 (thunderx2t99_sha): New Reservation.
23366 2017-04-25 Julian Brown <julian@codesourcery.com>
23367 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23369 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23370 type for 1-element load.
23372 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23374 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23376 2017-04-24 Martin Jambor <mjambor@suse.cz>
23378 PR tree-optimization/80293
23379 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23380 char arrays not totally scalarizable if it is false.
23381 (analyze_all_variable_accesses): Pass correct value in the new
23382 parameter. Add a statistics counter.
23384 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23386 PR middle-end/79931
23387 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23389 2017-04-24 Richard Biener <rguenther@suse.de>
23391 PR tree-optimization/80494
23392 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23393 out for complex types.
23395 2017-04-24 Richard Biener <rguenther@suse.de>
23397 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23398 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23399 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23400 (DFS): Adjust and never fail.
23401 (sccvn_dom_walker::fail): Remove.
23402 (sccvn_dom_walker::before_dom_children): Adjust.
23403 (run_scc_vn): Likewise and never fail.
23404 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23405 (pass_fre::execute): Likewise.
23407 2017-04-24 Richard Biener <rguenther@suse.de>
23409 PR tree-optimization/79725
23410 * tree-ssa-sink.c (statement_sink_location): Return whether
23411 failure reason was zero uses. Move that check later.
23412 (sink_code_in_bb): Deal with zero uses by removing the stmt
23415 2017-04-24 Richard Biener <rguenther@suse.de>
23418 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23419 pointer-based references.
23421 2017-04-24 Richard Biener <rguenther@suse.de>
23424 * pass_manager.h (pass_manager::operator new): Remove.
23425 (pass_manager::operator delete): Likewise.
23426 * passes.c (pass_manager::operator new): Remove.
23427 (pass_manager::operator delete): Likewise.
23428 (pass_manager::pass_manager): Zero individual pass members.
23430 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23433 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23434 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23435 Check "XEXP (src, 1)" operand here.
23436 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23437 Check "XEXP (src, 1)" operand here.
23438 (dimode_scalar_chain::make_vector_copies): Detect count register
23439 of a shift instruction. Zero extend count register from QImode
23440 to DImode to satisfy vector shift pattern count operand predicate.
23441 Substitute vector shift count operand with a DImode copy.
23442 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23445 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23447 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23448 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23449 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23450 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23451 (UNSPEC_NOREX_MEM): Remove definition.
23453 2017-04-21 Richard Biener <rguenther@suse.de>
23455 PR tree-optimization/79547
23456 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23457 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23458 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23459 without any constraints.
23461 2017-04-21 Richard Biener <rguenther@suse.de>
23463 PR tree-optimization/78847
23464 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23466 2017-04-21 Richard Biener <rguenther@suse.de>
23468 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23469 (build_distinct_type_copy): Likewise.
23470 (build_variant_type_copy): Likewise.
23471 * tree.c (build_qualified_type): Pass down mem-stat info.
23472 (build_distinct_type_copy): Likewise.
23473 (build_variant_type_copy): Likewise.
23475 2017-04-21 Richard Biener <rguenther@suse.de>
23477 PR tree-optimization/80237
23478 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23480 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23481 for a simplified result.
23483 2016-04-21 Richard Biener <rguenther@suse.de>
23485 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23486 sth as strict as a simple_iv but a chrec without symbols and an
23487 operand defined in the loop we are peeling (and not some subloop).
23488 (propagate_constants_for_unrolling): Propagate all constants.
23490 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23493 * config/i386/i386.c (print_reg): Remove assert for disalowed
23494 regno values, call output_operand_lossage instead.
23496 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23499 * config/i386/constraints.md (Yc): New register constraint.
23500 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23501 Use Yc constraint for alternative 2 of operand 0. Remove
23502 preferred_for_speed attribute.
23504 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23506 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23507 lastprivate clauses in SIMT case.
23509 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23511 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23513 2017-04-20 Richard Biener <rguenther@suse.de>
23515 PR tree-optimization/57796
23516 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
23517 as N scalar stores.
23518 (vect_model_load_cost): Cost gathers as N scalar loads.
23520 2017-04-20 Richard Biener <rguenther@suse.de>
23522 * ggc-page.c (ggc_allocated_p): Rename to ...
23523 (safe_lookup_page_table_entry): ... this and return the lookup
23525 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
23527 2017-04-20 Richard Biener <rguenther@suse.de>
23529 PR tree-optimization/80453
23530 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
23531 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
23532 from the conditions.
23533 (vn_phi_eq): Pass them down.
23534 (vn_phi_lookup): Record them.
23535 (vn_phi_insert): Likewise.
23537 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
23539 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
23540 uninitialized variable warning to avoid buffer overrun.
23542 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23545 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
23546 is suppressed for '{ 0 }' in C.
23548 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23550 * BASE-VER: Set to 8.0.0.
23552 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
23554 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
23555 priority .init_array and .fini_array section with SECTION_NOTYPE
23558 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23560 PR middle-end/80423
23561 * tree.h (build_array_type): Add typeless_storage default argument.
23562 * tree.c (type_cache_hasher::equal): Also compare
23563 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
23564 (build_array_type): Add typeless_storage argument, set
23565 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
23567 (build_nonshared_array_type): Adjust build_array_type_1 caller.
23568 (build_array_type): Likewise. Add typeless_storage argument.
23570 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23571 Jakub Jelinek <jakub@redhat.com>
23573 PR tree-optimization/80426
23574 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
23575 operation on symbolic operands, also compute the overflow for the
23576 invariant part when the operation degenerates into a negation.
23578 2017-04-19 Jakub Jelinek <jakub@redhat.com>
23581 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
23582 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
23585 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
23587 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
23590 * config/avr/avr.c (tree.h): Include it.
23591 (cgraph.h): Include it.
23592 (avr_encode_section_info): Don't warn for uninitialized progmem
23593 variable if it's just an alias.
23595 2017-04-19 Richard Biener <rguenther@suse.de>
23598 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
23599 when needed by AutoPGO.
23601 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23604 * doc/lto.texi: Remove an extra 'that'.
23606 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23608 PR rtl-optimization/80429
23609 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23610 are only used in debug insns.
23612 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23613 Vladimir Makarov <vmakarov@redhat.com>
23615 * config/sparc/predicates.md (input_operand): Add comment. Return
23616 true for any memory operand when LRA is in progress.
23617 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23619 2017-04-18 Jeff Law <law@redhat.com>
23622 * mips.md ({return,simple_return}_internal): Do not overwrite
23625 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23627 PR tree-optimization/80443
23628 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23629 instead of adding 1, subtract -1 and similarly instead of subtracting
23632 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23634 PR rtl-optimization/80357
23635 * haifa-sched.c (tmp_bitmap): New variable.
23636 (model_recompute): Handle duplicate use records.
23637 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23638 (free_global_sched_pressure_data): Free it.
23640 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23643 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23644 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23645 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23646 instead of SYSTEM_HEADER_DIR.
23648 2017-04-18 Jeff Law <law@redhat.com>
23650 PR middle-end/80422
23651 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23652 predecessors after walking up the insn chain.
23654 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23657 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23658 sizetype type into debug info.
23660 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23663 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23664 unneeded test for TARGET_UPPER_REGS_SF.
23665 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23667 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23670 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23671 instead of gsi_after_labels.
23673 2017-04-18 Jeff Law <law@redhat.com>
23675 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23679 2017-04-13 Jeff Law <law@redhat.com>
23680 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23681 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23683 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23686 * config/avr/avr.c (intl.h): Include it.
23687 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23689 2017-04-18 Martin Liska <mliska@suse.cz>
23691 PR gcov-profile/78783
23692 * gcov-tool.c (gcov_output_files): Validate that destination
23693 file is either removed by the tool or by a user.
23695 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23696 Guy Benyei <guybe@mellanox.com>
23698 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23699 block, and do not negate it, the stored id is already negative.
23701 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23703 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23705 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23708 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23709 masks of options that should be turned off if the VSX vector
23710 options are turned off.
23711 (OTHER_P8_VECTOR_MASKS): Likewise.
23712 (OTHER_VSX_VECTOR_MASKS): Likewise.
23713 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23714 rs6000_disable_incompatible_switches to validate no type switches
23716 (rs6000_incompatible_switch): New function to disallow turning on
23717 other vector options if -mno-vsx, -mno-power8-vector, or
23718 -mno-power9-vector are specified.
23720 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23722 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23724 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23726 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23727 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23728 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23729 (ARG_POINTER_CFA_OFFSET): Likewise.
23731 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23733 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23734 conditions to take advantage of various optimizations.
23736 2017-04-13 Jeff Law <law@redhat.com>
23738 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23739 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23740 (zero_extendsidi2_dext): Likewise.
23742 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23745 * fold-const.c (fold_ternary_loc): Revert
23746 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23749 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23751 PR rtl-optimization/80343
23752 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23753 new scratch pseudo.
23755 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23758 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23759 to ubsan_encode_value.
23761 2017-04-13 Jeff Law <law@redhat.com>
23763 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23764 appearing in DEBUG_INSNs.
23766 2017-04-13 Martin Liska <mliska@suse.cz>
23768 PR gcov-profile/80413
23769 * gcov-io.c (gcov_write_string): Copy to buffer just when
23770 allocated size is greater than zero.
23772 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23775 * dwarf2out.c (decls_for_scope): Ignore declarations of
23776 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23778 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23781 * ipa-visibility.c (non_local_p): Fix typos.
23782 (localize_node): When localizing symbol in same comdat group,
23783 dissolve the group only when we know external symbols are going
23785 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23787 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23789 PR tree-optimization/79390
23790 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23791 order does not result in usable sequence, retry with reversed operand
23797 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23798 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23799 op0 instead of fold_convert_loc (loc, type, arg0).
23801 2017-04-12 Jeff Law <law@redhat.com>
23803 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23804 has a delay slot in the generated code.
23806 * config/cris/cris.md (cris_preferred_reload_class): Return
23807 GENNONACR_REGS rather than GENERAL_REGS.
23809 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23812 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23813 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23814 signedness of the result type.
23816 2017-04-12 Richard Biener <rguenther@suse.de>
23817 Jeff Law <law@redhat.com>
23819 PR tree-optimization/80359
23820 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23821 trim stores to TARGET_MEM_REFs.
23823 2017-04-12 Richard Biener <rguenther@suse.de>
23825 PR tree-optimization/79390
23826 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23827 threading case even more.
23829 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23832 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23833 for quad_address_p for TImode, instead of just not indexed_address.
23835 2017-04-12 Richard Biener <rguenther@suse.de>
23836 Bernd Edlinger <bernd.edlinger@hotmail.de>
23838 PR middle-end/79671
23839 * alias.c (component_uses_parent_alias_set_from): Handle
23840 TYPE_TYPELESS_STORAGE.
23841 (get_alias_set): Likewise.
23842 * tree-core.h (tree_type_common): Add typeless_storage flag.
23843 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23844 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23845 for types containing members with TYPE_TYPELESS_STORAGE.
23846 (place_field): Likewise.
23847 (layout_type): Likewise for ARRAY_TYPE.
23848 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23849 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23850 TYPE_TYPELESS_STORAGE.
23851 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23853 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23856 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23857 first argument to type.
23859 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23863 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23864 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23865 (rs6000_expand_binop_builtin): Likewise.
23866 (rs6000_expand_ternop_builtin): Likewise; also add missing
23867 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23868 vshasigma built-ins.
23869 * doc/extend.texi: Document that vec_xxpermdi's third argument
23870 must be a constant.
23872 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23874 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23875 Use shift_const cost parameter when calculating gain of STV shifts.
23877 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23879 PR rtl-optimization/70478
23880 * lra-constraints.c (process_alt_operands): Check memory for
23881 disfavoring memory insn operand.
23883 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23885 PR middle-end/80100
23886 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23887 left shift in unsigned HOST_WIDE_INT type.
23889 PR rtl-optimization/80385
23890 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23891 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23894 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23895 if they have any depend clauses.
23897 2017-04-11 Martin Liska <mliska@suse.cz>
23900 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23901 * ipa-split.c (split_function): Create a local comdat symbol
23902 if caller is in a comdat group.
23904 2017-04-11 Martin Liska <mliska@suse.cz>
23907 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23910 2017-04-11 Martin Sebor <msebor@redhat.com>
23912 PR middle-end/80364
23913 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23914 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23916 (directive::set_width, directive::set_precision, format_character):
23918 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23921 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23924 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23925 conflict, set target->arch_name instead of target->cpu_name.
23927 2017-04-11 Richard Biener <rguenther@suse.de>
23929 PR tree-optimization/80374
23930 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23931 build_zero_cst, remove fold_convertible_p check again.
23933 2017-04-11 Martin Liska <mliska@suse.cz>
23936 * ubsan.c (instrument_object_size): Do not instrument register
23939 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23942 * config/i386/i386-builtin-types.def
23943 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23944 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23945 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23946 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23947 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23948 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23949 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23950 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23951 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23952 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23953 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23954 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23955 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23956 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23957 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23958 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23959 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23960 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23961 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23962 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23963 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23964 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23965 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23966 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23967 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23968 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23969 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23970 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23971 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23972 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23973 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23974 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23975 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23976 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23977 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23978 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23979 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23980 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23981 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23982 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23983 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23984 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23985 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23986 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23987 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23988 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23990 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23991 flag to second_arg_count, handle 4 argument function type _COUNT
23992 aliases, handle second_arg_count on second argument rather than last.
23994 2017-04-10 Jeff Law <law@redhat.com>
23996 PR tree-optimization/80374
23997 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23998 record anything if we can not convert integer_zero_node to the
24001 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24004 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24005 Enhance special handling given to the TARGET_P9_MINMAX option in
24006 relation to certain other options.
24008 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24010 PR tree-optimization/80153
24011 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24012 remove POINTER_PLUS_EXPR's base part directly, rather than through
24015 2017-04-10 Richard Biener <rguenther@suse.de>
24016 Bin Cheng <bin.cheng@arm.com>
24018 PR tree-optimization/80153
24019 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24020 the first element of pointer type aff_tree. Build result expr in
24022 (add_elt_to_tree): Convert to type unconditionally. Remove other
24023 fold_convert calls.
24024 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24025 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24027 2017-04-10 Richard Biener <rguenther@suse.de>
24029 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24032 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24034 PR rtl-optimization/70478
24035 * lra-constraints.c (curr_small_class_check): New.
24036 (update_and_check_small_class_inputs): New.
24037 (process_alt_operands): Update curr_small_class_check. Disfavor
24038 alternative insn memory operands. Check available regs for small
24041 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24044 * config/mips/mips.opt (-mvirt): Update description.
24045 * doc/invoke.texi (-mvirt): Likewise.
24047 2017-04-10 Richard Biener <rguenther@suse.de>
24049 PR middle-end/80362
24050 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24051 looking for NEGATE_EXPR in -A / -B to A / B folding.
24053 2017-04-10 Martin Liska <mliska@suse.cz>
24055 PR gcov-profile/80224
24056 * gcov.c (print_usage): Fix usage string.
24057 (get_gcov_intermediate_filename): Remove.
24058 (output_gcov_file): Use both for normal and intermediate format.
24059 (generate_results): Do not initialize special file for
24060 intermediate format.
24062 2017-04-10 Richard Biener <rguenther@suse.de>
24064 PR tree-optimization/80304
24065 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24068 2017-04-10 Nathan Sidwell <nathan@acm.org>
24071 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24072 (rs6000_init_builtins): Use it.
24074 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24076 * config/arm/arm.md (<mrc>): Add mode to SET source.
24077 (<mrrc>): Likewise.
24079 2017-04-10 Richard Biener <rguenther@suse.de>
24081 PR middle-end/80344
24082 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24084 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24087 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24088 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24089 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24090 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24091 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24092 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24093 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24094 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24095 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24096 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24097 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24098 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24099 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24100 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24101 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24102 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24103 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24104 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24105 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24106 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24107 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24108 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24109 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24111 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24113 PR rtl-optimization/70478
24114 * lra-constraints.c: Reverse the last patch.
24116 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24118 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24119 Add comment for WCHAR_T.
24121 2017-04-08 Martin Liska <mliska@suse.cz>
24124 2017-04-07 Martin Liska <mliska@suse.cz>
24127 * ipa-split.c (split_function): Add function part to a same comdat
24130 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24133 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24135 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24137 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24138 (vec_pairing): ...to this.
24139 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24140 (rs6000_sched_init): Adjust for name change.
24141 (struct rs6000_sched_context): Likewise.
24142 (rs6000_init_sched_context): Likewise.
24143 (rs6000_set_sched_context): Likewise.
24145 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24151 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24153 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24154 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24155 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24157 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24159 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24161 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24163 PR rtl-optimization/70703
24164 * ira-color.c (update_conflict_hard_regno_costs): Use
24165 int64_t instead of HOST_WIDE_INT.
24167 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24169 PR rtl-optimization/70478
24170 * lra-constraints.c (process_alt_operands): Disfavor alternative
24171 insn memory operands.
24173 2017-04-07 Jeff Law <law@redhat.com>
24175 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24176 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24178 2017-04-07 Martin Liska <mliska@suse.cz>
24181 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24182 Show error message instead of an ICE.
24184 2017-04-07 Martin Liska <mliska@suse.cz>
24187 * ipa-split.c (split_function): Add function part to a same comdat
24190 2017-04-07 Richard Biener <rguenther@suse.de>
24192 PR middle-end/80341
24193 * tree.c (get_unwidened): Also handle ! for_type case for
24195 * convert.c (do_narrow): Split out from ...
24196 (convert_to_integer_1): ... here. Do not pass final truncation
24197 type to get_unwidened for TRUNC_DIV_EXPR.
24199 2017-04-07 Richard Biener <rguenther@suse.de>
24201 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24203 (aff_combination_const): Adjust.
24204 (aff_combination_scale): Likewise.
24205 (aff_combination_add_elt): Likewise.
24206 (aff_combination_add_cst): Likewise.
24207 (aff_combination_convert): Likewise.
24208 (add_elt_to_tree): Likewise. Remove unused argument.
24209 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24211 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24213 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24215 * config/arm/arm.c (arm_default_short_enums): Use
24216 ARM_DEFAULT_SHORT_ENUMS.
24217 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24219 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24222 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24223 members with redundant out-of-class redeclaration.
24225 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24228 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24229 * config/i386/i386.md (*zero_extendsidi2):
24230 Add (?*x,*x) and (?*v,*v) alternatives.
24232 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24235 * config/i386/i386.c (ix86_expand_builtin)
24236 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24237 mode from insn data. Convert operands to insn operand mode.
24238 Copy operands that don't satisfy insn predicate to a register.
24240 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24242 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24245 2017-04-06 Richard Biener <rguenther@suse.de>
24247 PR tree-optimization/80334
24248 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24249 preserve alignment of accesses.
24251 2017-04-06 Richard Biener <rguenther@suse.de>
24253 PR tree-optimization/80262
24254 * tree-sra.c (build_ref_for_offset): Preserve address-space
24256 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24257 Drop useless address-space information on MEM_REF offsets.
24259 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24261 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24263 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24265 PR rtl-optimization/70703
24266 * ira-color.c (update_conflict_hard_regno_costs): Use
24267 HOST_WIDE_INT instead of long.
24269 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24272 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24273 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24275 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24276 for x86_64 target. Handle -m3dnowa option.
24278 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24280 PR rtl-optimization/70703
24281 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24282 (update_conflict_hard_regno_costs): Use long instead of unsigned
24283 arithmetic for cost calculation.
24285 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24286 Bernd Edlinger <bernd.edlinger@hotmail.de>
24289 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24292 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24295 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24296 ptr_mode with Pmode throughout.
24297 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24298 into probe_stack_range and use DImode.
24300 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24303 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24304 call_eh_return is true.
24306 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24308 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24309 Initialize last_match_fntype_index.
24311 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24314 * tree-nvr.c: Include internal-fn.h.
24315 (pass_return_slot::execute): Ignore internal calls without
24318 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24319 Richard Biener <rguenther@suse.de>
24322 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24323 captures used multiple times, except for the last use.
24324 * generic-match-head.c: Include gimplify.h.
24326 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24328 PR tree-optimization/79390
24329 * target.h (struct noce_if_info): Declare.
24330 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24331 * target.def (noce_conversion_profitable_p): New target hook.
24332 * ifcvt.h (struct noce_if_info): New type, moved from ...
24333 * ifcvt.c (struct noce_if_info): ... here.
24334 (noce_conversion_profitable_p): Renamed to ...
24335 (default_noce_conversion_profitable_p): ... this. No longer
24337 (noce_try_store_flag_constants, noce_try_addcc,
24338 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24339 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24340 instead of noce_conversion_profitable_p.
24341 * config/i386/i386.c: Include ifcvt.h.
24342 (ix86_option_override_internal): Don't override
24343 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24344 (ix86_noce_conversion_profitable_p): New function.
24345 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24346 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24347 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24348 * doc/tm.texi: Regenerated.
24350 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24352 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24355 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24358 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24359 instructions for small multiply cores.
24361 2017-04-04 Jeff Law <law@redhat.com>
24363 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24365 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24366 that are not set by the loop over the elements.
24368 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24371 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24372 int mode, convert_modes it to mode as unsigned, otherwise use
24373 lowpart_subreg to mode rather than SImode.
24374 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24375 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24376 Use DImode instead of SImode for the shift count operand.
24377 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24380 2017-04-04 Richard Biener <rguenther@suse.de>
24382 PR middle-end/80281
24383 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24384 arithmetic done for the negate or the plus. Simplify.
24385 (A - (-B) -> A + B): Likewise.
24386 * fold-const.c (split_tree): Make sure to not negate pointers.
24388 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24390 PR rtl-optimization/60818
24391 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24392 a compare of comparisons with the thing compared if this results
24393 in a different machine mode.
24395 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24397 * alias.c (base_alias_check): Fix typo in comment.
24398 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24399 * cgraphunit.c (symbol_table::compile): Likewise.
24400 * collect2.c (maybe_run_lto_and_relink): Likewise.
24401 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24402 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24403 * config/avr/avr.c (avr_map_op_t): Likewise.
24404 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24405 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24406 * config/epiphany/epiphany.md (movcc): Likewise.
24407 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24408 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24410 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24411 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24412 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24413 * config/sh/sh.c (sh_rtx_costs): Likewise.
24414 * fold-const.c (fold_truth_andor): Likewise.
24415 * genautomata.c (collapse_flag): Likewise.
24416 * gengtype.h (struct type::u::s): Likewise.
24417 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24418 * input.c (FORMAT_AMOUNT): Likewise.
24419 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24420 (known_aggs_to_agg_replacement_list): Likewise.
24421 * ipa-inline-analysis.c: Likewise.
24422 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24423 * ipa-polymorphic-call.c
24424 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24425 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24426 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24428 * modulo-sched.c (apply_reg_moves): Likewise.
24429 * omp-expand.c (build_omp_regions_1): Likewise.
24430 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24431 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24432 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24433 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24434 * value-prof.c: Likewise.
24435 * var-tracking.c (val_reset): Likewise.
24437 2017-04-03 Richard Biener <rguenther@suse.de>
24439 PR tree-optimization/80275
24440 * fold-const.c (split_address_to_core_and_offset): Handle
24443 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24445 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24446 descriptors is at least equal to that of functions.
24448 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24450 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24452 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24455 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24456 (mov<IMOD4:mode>): New expander.
24457 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24459 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24461 PR rtl-optimization/79405
24462 * fwprop.c (propagations_left): New variable.
24463 (forward_propagate_into): Decrement it.
24464 (fwprop_init): Initialize it.
24465 (fw_prop): If the variable has reached zero, stop propagating.
24466 (fwprop_addr): Ditto.
24468 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24471 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24472 a FUNCTION_DECL, pass it as decl instead of origin to
24475 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24477 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24480 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24483 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24484 TARGET_VSX_SMALL_INTEGER.
24486 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24488 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24489 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24491 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24493 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24494 extraction from odd-numbered MSA register.
24496 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24498 PR middle-end/80173
24499 * expmed.c (store_bit_field_1): Don't attempt to create
24500 a word subreg out of hard registers wider than word if they
24501 have HARD_REGNO_NREGS of 1 for their mode.
24503 PR middle-end/80163
24504 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24505 conversions to integer types wider than word and pointer.
24508 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24509 (rtx_equal_for_cselib_p): Pass 0 to it.
24510 * cselib.c (cselib_hasher::equal): Likewise.
24511 (rtx_equal_for_cselib_1): Add depth argument. If depth
24512 is 128, don't look up VALUE locs and punt. Increment
24513 depth in recursive calls when walking VALUE locs.
24515 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
24517 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
24518 (make_gcov_file_name): Use the canonical path name for generating
24520 (read_line): Fix handling of files with ascii null bytes.
24522 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
24524 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
24525 to initialise a vector register instead
24526 of using a const_int.
24528 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24530 PR translation/80189
24531 * gimplify.c (omp_default_clause): Use %qs instead of %s in
24532 diagnostic messages.
24534 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
24537 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
24538 (dfp_diex_<mode>): Update mode of operand 1.
24539 * doc/extend.texi (dxex, dxexq): Document change to return type.
24540 (diex, diexq): Document change to argument type.
24542 2017-03-30 Martin Jambor <mjambor@suse.cz>
24545 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
24546 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
24547 it reflects the signature changes performed at the callee side.
24548 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
24549 to cgraph_build_function_type_skip_args.
24550 (build_function_decl_skip_args): Adjust call to the above function.
24552 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24555 * config/i386/sse.md
24556 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
24557 register as dest whenever it is a MEM not rtx_equal_p to the
24558 corresponding dup operand, and when forcing into reg move the
24559 reg into the memory afterwards.
24560 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
24561 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
24562 for the force_reg mode.
24563 (avx512vl_vextractf128<mode>): Use register as dest either
24564 always when a MEM, or when it is a MEM not rtx_equal_p to the
24565 corresponding dup operand, or even not when it is a CONST_VECTOR
24566 depending on the mode and lo vs. hi.
24567 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
24569 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
24570 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
24571 Likewise. Require that operands[2] is even.
24572 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
24573 Remove extraneous parens. Require that operands[2] is a multiple
24575 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
24576 operands[0] is a MEM if <mask_applied>, the predicates/constraints
24577 disallow memory then.
24579 2017-03-30 Richard Biener <rguenther@suse.de>
24581 PR tree-optimization/77498
24582 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
24583 to non-constants over backedges.
24585 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
24587 PR rtl-optimization/80233
24588 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
24589 as last_combined_insn. Do not test for BARRIER_P separately.
24591 2017-03-29 Andreas Schwab <schwab@suse.de>
24594 * calls.c (prepare_call_address): Convert funexp to Pmode before
24595 copying to temp reg.
24597 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24599 PR tree-optimization/80158
24600 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
24601 Handle possible future case of more than one alternate
24603 (replace_rhs_if_not_dup): Likewise.
24604 (replace_one_candidate): Likewise.
24606 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24608 PR rtl-optimization/80193
24609 * ira.c (ira): Do not check allocation for LRA.
24611 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24613 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24614 (nvptx_output_simt_exit): Declare.
24615 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24616 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24617 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24618 (nvptx_declare_function_name): ...here. Emit declaration of local
24619 memory space buffer for omp_simt_enter insn.
24620 (nvptx_output_unisimt_switch): New.
24621 (nvptx_output_softstack_switch): New.
24622 (nvptx_output_simt_enter): New.
24623 (nvptx_output_simt_exit): New.
24624 * config/nvptx/nvptx.h (struct machine_function): New fields
24625 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24626 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24627 (UNSPECV_SIMT_EXIT): Ditto.
24628 (omp_simt_enter_insn): New insn.
24629 (omp_simt_enter): New expansion.
24630 (omp_simt_exit): New insn.
24631 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24633 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24634 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24635 (expand_GOMP_SIMT_EXIT): New.
24636 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24637 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24638 (GOMP_SIMT_EXIT): Ditto.
24639 * target-insns.def (omp_simt_enter): New insn.
24640 (omp_simt_exit): Ditto.
24641 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24643 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24644 (lower_rec_input_clauses): Likewise.
24645 (lower_lastprivate_clauses): Handle SIMT privatization.
24647 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24648 (ompdevlow_adjust_simt_enter): New.
24649 (find_simtpriv_var_op): New.
24650 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24651 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24653 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24654 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24655 (copy_decl_for_dup_finish): Ditto.
24657 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24659 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24662 * config/i386/i386.c (ix86_option_override_internal): Always
24663 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24665 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24667 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24669 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24671 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24672 mark new edge's irreducible flag accordign to it.
24673 (vect_do_peeling): Check loop preheader edge's irreducible flag
24674 and pass it to function slpeel_add_loop_guard.
24676 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24678 PR tree-optimization/80218
24679 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24680 Update block frequencies and counts.
24682 2017-03-28 Richard Biener <rguenther@suse.de>
24684 PR tree-optimization/78644
24685 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24686 of a simplification result we may not use it at all.
24688 2017-03-28 Richard Biener <rguenther@suse.de>
24691 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24692 without arguments, generate default definition of a SSA name.
24694 2017-03-28 Richard Biener <rguenther@suse.de>
24696 PR middle-end/80222
24697 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24698 TYPE_REF_CAN_ALIAS_ALL references.
24699 * fold-const.c (fold_indirect_ref_1): Likewise.
24701 2017-03-28 Martin Liska <mliska@suse.cz>
24704 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24705 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24707 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24708 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24710 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24711 (EXTRA_SPECS): Define.
24712 (SUBTARGET_EXTRA_SPECS): Likewise.
24713 (SUBTARGET_CPP_SPEC): Likewise.
24714 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24715 SUBTARGET_EXTRA_SPECS.
24716 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24718 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24720 * config/arc/simdext.md (vst64_insn): Update pattern.
24721 (vld32wh_insn): Likewise.
24722 (vld32wl_insn): Likewise.
24723 (vld64_insn): Likewise.
24724 (vld32_insn): Likewise.
24726 2017-03-28 Marek Polacek <polacek@redhat.com>
24729 * fold-const.c (fold_comparison): Use protected_set_expr_location
24730 instead of SET_EXPR_LOCATION.
24732 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24734 * tree.c (add_expr): Avoid name lookup warning.
24736 2017-03-27 Jeff Law <law@redhat.com>
24738 PR tree-optimization/80216
24739 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24740 function name. Limit recursion depth.
24741 (record_temporary_equivalences): Corresponding changes.
24743 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24745 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24748 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24751 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24753 * cfgcleanup.c (reg_note_cfa_p): New array.
24754 (insns_have_identical_cfa_notes): New function.
24755 (old_insns_match_p): Don't cross-jump in between /f
24756 and non-/f instructions. If both i1 and i2 are frame related,
24757 verify all CFA notes, their order and content.
24759 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24762 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24763 HImode and SImode with zero extend to DImode to one insn.
24764 (bswap<mode>2_extenddi): Likewise.
24765 (bswapsi2_extenddi): Likewise.
24766 (bswaphi2_extendsi): Likewise.
24767 (bswaphi2): Combine bswap HImode and SImode into one insn.
24768 Separate memory insns from swapping register.
24769 (bswapsi2): Likewise.
24770 (bswap<mode>2): Likewise.
24771 (bswaphi2_internal): Delete, no longer used.
24772 (bswapsi2_internal): Likewise.
24773 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24774 store, and gpr<-gpr swap insns.
24775 (bswap<mode>2_store): Likewise.
24776 (bswaphi2_reg): Register only splitter, combine with the splitter.
24777 (bswaphi2 splitter): Likewise.
24778 (bswapsi2_reg): Likewise.
24779 (bswapsi2 splitter): Likewise.
24780 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24781 the insns into load, store, and register/register insns.
24782 (bswapdi2_ldbrx): Likewise.
24783 (bswapdi2_load): Likewise.
24784 (bswapdi2_store): Likewise.
24785 (bswapdi2_reg): Likewise.
24787 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24789 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24790 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24792 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24795 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24797 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24798 special handling for target option conflicts between dform
24799 options (-mpower9-dform, -mpower9-dform-vector,
24800 -mpower9-dform-scalar) and -mno-direct-move.
24802 2017-03-27 Richard Biener <rguenther@suse.de>
24804 PR tree-optimization/80181
24805 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24807 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24809 * config/arc/predicates.md (move_double_src_operand): Replace the
24810 call to move_double_src_operand with a call to address_operand.
24812 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24814 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24815 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24816 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24818 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24820 * config/arc/predicates.md (long_immediate_loadstore_operand):
24821 Consider scaled addresses cases.
24823 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24825 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24826 restored when in interrupt.
24827 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24828 doesn't have delay slot.
24830 2017-03-27 Richard Biener <rguenther@suse.de>
24833 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24834 inlined thunk clones.
24836 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24839 * asan.c (instrument_derefs): Copy over last operand from
24840 original COMPONENT_REF to the new COMPONENT_REF with
24841 DECL_BIT_FIELD_REPRESENTATIVE.
24842 * ubsan.c (instrument_object_size): Likewise.
24844 2017-03-27 Richard Biener <rguenther@suse.de>
24846 PR tree-optimization/80170
24847 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24848 sure DR/SCEV didnt fold in constants we do not see when looking
24849 at the reference base alignment.
24851 2017-03-27 Richard Biener <rguenther@suse.de>
24853 PR middle-end/80171
24854 * gimple-fold.c (fold_ctor_reference): Properly guard against
24855 NULL return value from canonicalize_constructor_val.
24857 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24860 * config/i386/i386.c (ix86_expand_builtin)
24861 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24862 flags reg setting and flags reg using instructions.
24863 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24864 clobbering instructions to zero extend op2.
24866 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24868 * doc/install.texi (Configuration) <--with-aix-soname>:
24869 Update link to AIX ld.
24871 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24873 PR rtl-optimization/80160
24874 PR rtl-optimization/80159
24875 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24876 reg_alternate_class into account.
24878 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24881 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24882 to consider in curr_insn_transform.
24884 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24886 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24887 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24888 and emit_mode_inner.
24890 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24892 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24893 argument to the overloaded builtin variants. Use the new flag to
24894 deprecate certain builtin variants.
24895 * config/s390/s390-builtin-types.def: Add new builtin types.
24896 * config/s390/s390-builtins.h: Support new flags field for
24897 overloaded builtins.
24898 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24899 (s390_macro_to_expand): Enable vector float data type.
24900 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24901 builtins by incrementing the __VEC__ version number.
24902 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24904 (s390_resolve_overloaded_builtin): Emit error messages depending
24905 on the builtin flags.
24906 * config/s390/s390.c (s390_expand_builtin): Support additional
24907 flags argument. Change error message to match the messages
24908 emitted in s390-c.c.
24909 * config/s390/s390.md: New UNSPEC_* constants.
24910 (op_type): Add new instruction types.
24911 * config/s390/vecintrin.h: Add new builtins and test data class
24913 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24914 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24915 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24916 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24917 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24918 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24920 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24921 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24922 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24923 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24925 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24926 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24927 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24928 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24930 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24931 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24932 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24933 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24934 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24935 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24936 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24938 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24939 ("vec_scatter_element<V_HW_4:mode>_DI")
24940 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24941 ("vec_fpint<mode>", "vflls")
24942 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24943 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24944 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24945 ("*vec_cmphe<mode>_cc"): ... these.
24947 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24948 mode constant instead of magic value.
24950 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24952 * config/s390/s390.c (s390_expand_vec_compare): Support other
24953 vector floating point modes than just V2DF.
24954 (s390_expand_vcond): Likewise.
24955 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24956 (s390_cannot_change_mode_class): Prevent mode changes between TF
24957 and V1TF in vector registers.
24958 * config/s390/s390.md (DF, SF): New mode attributes.
24959 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24960 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24961 SFmode support for VRs.
24962 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24964 (VFT, VF_HW): New mode iterators.
24965 (vw, sdx): New mode attributes.
24966 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24967 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24968 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24969 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24970 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24971 also the new vector floating point modes. Renaming to ...
24973 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24974 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24975 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24976 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24977 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24978 ("vec_unordered<mode>"): ... these.
24980 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24981 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24982 ("*vec_extendv2df"): New insn definitions.
24984 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24986 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24987 ("mulditi3_2", "*muldi3_sign"): New patterns.
24988 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24989 rename the pattern definition.
24991 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24993 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24995 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24997 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24999 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25000 instruction if possible.
25001 * config/s390/vector.md (vec_halfnumelts): New mode
25003 ("*vec_vllezlf<mode>"): New pattern.
25005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25007 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25008 ("popcountv4si2", "popcountv2di2"): Rename to ...
25009 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25010 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25012 ("popcount<mode>2_vxe"): New pattern.
25014 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25016 * common/config/s390/s390-common.c (processor_flags_table): Add
25018 * config.gcc: Add arch12.
25019 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25020 Default to arch12 for unknown CPU model numbers.
25021 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25022 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25023 PROCESSOR_max sanity check.
25024 * config/s390/s390-opts.h (enum processor_type): Add
25026 * config/s390/s390.c (processor_table): Add arch12.
25027 (s390_expand_builtin): Add check for B_VXE flag.
25028 (s390_issue_rate): Add PROCESSOR_ARCH12.
25029 (s390_get_sched_attrmask): Likewise.
25030 (s390_get_unit_mask): Likewise.
25031 (s390_sched_score): Enable z13 scheduling for arch12.
25032 (s390_sched_reorder): Likewise.
25033 (s390_sched_variable_issue): Likewise.
25034 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25036 (s390_tune_attr): Use z13 scheduling also for arch12.
25037 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25038 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25039 (TARGET_VXE_P): New macros.
25040 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25041 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25042 * config/s390/s390.opt: Add arch12 as processor_type.
25044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25046 * config/s390/s390.md
25047 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25048 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25049 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25051 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25052 Rename expanders to ...
25054 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25055 ("fixuns_truncdddi2_emu"): ... these.
25057 ("fixuns_trunc<mode>si2_emu"): New expander.
25059 ("*fixuns_truncdfdi2_z13"): Rename to ...
25060 ("*fixuns_truncdfdi2_vx"): ... this.
25062 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25064 * config/s390/2964.md: Remove the single element vector compare
25065 instructions which are no longer used.
25066 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25068 (s390_canonicalize_comparison): Remove handling of DFmode
25070 (s390_expand_vec_compare_scalar): Remove function.
25071 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25072 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25074 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25076 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25078 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25080 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25081 will anyway by matched by mov<mode>_64dfp.
25083 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25085 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25086 vlef/vstef. Add missing operand to vleif.
25088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25090 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25091 pair for all vector types with 64 bit elements.
25092 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25093 * config/s390/vector.md (V_HW_64): ... here.
25094 (V_128_NOSINGLE): New mode iterator.
25095 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25096 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25097 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25098 ("*vec_load_pairv2di"): Change to ...
25099 ("*vec_load_pair<mode>"): ... this one.
25101 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25103 * config/s390/constraints.md: Add comments.
25104 (jKK): Reject element sizes > 8 bytes.
25105 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25107 * config/s390/s390.md: Add the s_operand checks formerly in
25108 s390_split_ok_p to various splitters where they are still
25110 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25111 for 128 bit vectors. Plus two splitters.
25113 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25115 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25118 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25121 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25122 error if the boundary argument is not constant.
25124 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25126 PR rtl-optimization/80112
25127 * loop-doloop.c (doloop_condition_get): Don't check condition
25128 if cmp isn't SET with IF_THEN_ELSE src.
25130 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25132 PR tree-optimization/80158
25133 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25134 replacing a candidate statement, also replace it for the
25135 candidate's alternate interpretation.
25136 (replace_rhs_if_not_dup): Likewise.
25137 (replace_one_candidate): Likewise.
25139 2017-03-24 Richard Biener <rguenther@suse.de>
25141 PR tree-optimization/80167
25142 * graphite-isl-ast-to-gimple.c
25143 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25145 (translate_isl_ast_to_gimple::get_rename): Likewise.
25147 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25149 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25150 handling of certain combinations of target options, including the
25151 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25152 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25154 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25157 * config/arm/arm.md (*load_multiple): Add reload_completed to
25158 matching condition.
25160 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25161 Richard Biener <rguenth@suse.de>
25163 PR tree-optimization/79908
25164 PR tree-optimization/80136
25165 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25166 been cast away, gimplify_and_add suffices.
25168 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25170 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25172 2017-03-23 Richard Biener <rguenther@suse.de>
25174 PR tree-optimization/80032
25175 * gimplify.c (gimple_push_cleanup): Forced unconditional
25176 cleanups still have to go to the conditional_cleanups
25179 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25181 PR tree-optimization/80072
25182 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25184 (next_operand_entry_id): Change type to unsigned int.
25185 (sort_by_operand_rank): Make sure to return the right return value
25186 even if unsigned fields are bigger than INT_MAX.
25187 (struct oecount): Change cnt and id type to unsigned int.
25188 (oecount_hasher::equal): Formatting fix.
25189 (oecount_cmp): Make sure to return the right return value
25190 even if unsigned fields are bigger than INT_MAX.
25191 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25194 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25195 TREE_READONLY on result if writing it more than once.
25198 * doc/invoke.texi (-fsanitize=thread): Document that with
25199 -fnon-call-exceptions atomics are not able to throw
25203 * tsan.c: Include tree-eh.h.
25204 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25205 maybe_clean_or_replace_eh_stmt where needed.
25206 (instrument_memory_accesses): Add cfg_changed argument.
25207 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25208 if it returned true.
25209 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25211 PR rtl-optimization/63191
25212 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25213 wrapper function, moved the whole old content into ...
25214 (ix86_delegitimize_address_1): ... this. New inline function.
25215 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25216 true as last argument instead of ix86_delegitimize_address.
25218 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25220 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25221 cortexa57_branch_cost.
25223 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25225 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25227 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25230 * doc/md.texi (Constraints): Document wA constraint.
25231 * config/rs6000/constraints.md (wA): New.
25232 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25233 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25234 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25235 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25237 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25240 * gimplify.c (is_oacc_declared): New function.
25241 (oacc_default_clause): Use it to set default flags for acc declared
25242 variables inside parallel regions.
25243 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25244 declared variables.
25245 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25246 declare attribute to any decl as necessary.
25248 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25251 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25252 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25253 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25254 (arm_arch_lpae): This.
25255 * config/arm/arm.c (arm_arch7ve): Rename into ...
25256 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25257 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25260 2017-03-22 Martin Liska <mliska@suse.cz>
25263 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25264 error message instead of an ICE.
25266 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25268 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25270 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25272 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25274 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25277 2017-03-21 Martin Sebor <msebor@redhat.com>
25279 * doc/extend.texi: Use "cannot" instead of "can't."
25280 * doc/hostconfig.texi: Same.
25281 * doc/install.texi: Same.
25282 * doc/invoke.texi: Same.
25283 * doc/loop.texi: Same.
25284 * doc/md.texi: Same.
25285 * doc/objc.texi: Same.
25286 * doc/rtl.texi: Same.
25287 * doc/tm.texi: Same.
25288 * doc/tm.texi.in: Same.
25289 * doc/trouble.texi: Same.
25291 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25294 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25295 (collect_checksum_attributes): Set it.
25296 (die_checksum_ordered): Use it.
25298 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25300 PR tree-optimization/79908
25301 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25302 change: For a VA_ARG whose LHS has been cast away, use
25303 force_gimple_operand to construct the side effects.
25305 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25307 PR translation/80001
25308 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25309 more amenable to translation.
25310 (oacc_loop_auto_partitions): Likewise.
25312 2017-03-21 Marek Polacek <polacek@redhat.com>
25313 Martin Sebor <msebor@redhat.com>
25315 PR tree-optimization/80109
25316 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25317 on INTEGRAL_TYPE_P.
25319 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25320 Segher Boessenkool <segher@kernel.crashing.org>
25323 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25324 check reg_used_between_p between insn and one of succ or succ2
25325 depending on if succ is artificial insn not inserted into insn
25328 2017-03-21 Martin Liska <mliska@suse.cz>
25330 PR gcov-profile/80081
25331 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25332 * doc/gcc.texi: Include gcov-dump stuff.
25333 * doc/gcov-dump.texi: New file.
25335 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25337 PR rtl-optimization/79150
25338 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25339 conditional jump, if the jump is the last insn of the loop.
25341 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25342 Richard Biener <rguenth@suse.de>
25344 PR tree-optimization/79908
25345 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25346 been cast away, use force_gimple_operand to construct the side
25349 2017-03-21 Martin Liska <mliska@suse.cz>
25351 PR libfortran/79956
25352 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25355 2017-03-21 Brad Spengler <spender@grsecurity.net>
25358 * plugin.c (htab_hash_plugin): New function.
25359 (add_new_plugin): Use it and adjust.
25360 (parse_plugin_arg_opt): Adjust.
25361 (init_one_plugin): Likewise.
25363 2017-03-21 Richard Biener <rguenther@suse.de>
25365 PR tree-optimization/80032
25366 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25367 if set force the cleanup to happen unconditionally.
25368 (gimplify_target_expr): Push inserted clobbers with force_uncond
25369 to avoid them being removed by control-dependent DCE.
25371 2017-03-21 Richard Biener <rguenther@suse.de>
25373 PR tree-optimization/80122
25374 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25375 va_arg_pack_len when the inlined call stmt requires pack
25377 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25379 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25382 * tsan.c (instrument_builtin_call): If the memory model argument
25383 is not a constant, assume it is valid.
25386 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25389 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25391 PR rtl-optimization/79910
25392 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25393 if its dest is used by an insn before I2 (other than the combined
25394 insns themselves, which are properly handled already).
25396 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25399 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25401 * combine.c (record_used_regs): New static function.
25402 (try_combine): Handle situations where there is an additional
25403 instruction between I2 and I3 which needs to have a LOG_LINK
25407 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25409 * combine.c (try_combine): Delete redundant i1 test. Call
25410 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25412 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25415 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25416 alternatives 13/14.
25418 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25420 PR tree-optimization/80054
25421 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25422 the optimization if a PHI or any of its arguments is not dominated
25423 by the candidate's basis. Use gphi* rather than gimple* as
25425 (replace_profitable_candidates): Clean up a gimple* variable that
25426 should be a gphi* variable.
25428 2017-03-20 Martin Sebor <msebor@redhat.com>
25431 * doc/extend.texi (attribute constructor): Document present limitation.
25433 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25436 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25437 __POWER9_VECTOR__ #ifdef control, change template definition to
25438 use Power9-specific built-in function.
25439 (vec_any_eq): Likewise.
25440 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25441 to control outcomes from this test.
25442 (vector_ae_<mode>p): For VEC_F modes, likewise.
25444 2017-03-20 Ian Lance Taylor <iant@google.com>
25446 * config/i386/i386.c (ix86_function_regparm): Save an extra
25447 register for -fsplit-stack with DECL_STATIC_CHAIN.
25449 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25452 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25453 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25455 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25457 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25459 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25462 2017-03-20 Marek Polacek <polacek@redhat.com>
25465 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25467 2017-03-20 Richard Biener <rguenther@suse.de>
25469 PR tree-optimization/80113
25470 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25471 allocate extra SSA name for PHI def.
25472 (add_close_phis_to_outer_loops): Likewise.
25473 (add_close_phis_to_merge_points): Likewise.
25474 (copy_loop_close_phi_args): Likewise.
25475 (copy_cond_phi_nodes): Likewise.
25477 2017-03-20 Martin Liska <mliska@suse.cz>
25479 PR middle-end/79753
25480 * tree-chkp.c (chkp_build_returned_bound): Do not build
25481 returned bounds for a LHS that's not a BOUNDED_P type.
25483 2017-03-20 Martin Liska <mliska@suse.cz>
25487 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25488 COMPLEX_CST and VECTOR_CST.
25490 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25493 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25494 target operand. A new splitter adds the clobber statement in case
25495 the target operand is dead anyway.
25497 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25499 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25500 to age-old versions of binutils and glibc.
25502 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25504 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25506 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25508 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25510 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25512 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25513 requirement for binutils 2.13.
25515 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25517 * combine.c (try_combine): Delete redundant i1 test. Call
25518 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25520 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
25522 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
25523 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
25525 <riscv64-*-elf>: Re-arrange section
25526 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
25527 <riscv32-*-linux>: Likewise.
25528 <riscv64-*-elf>: Likewise
25529 <riscv64-*-linux>: Likewise.
25531 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
25534 * aarch64.opt(verbose-cost-dump): Fix typo.
25536 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
25539 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
25540 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
25542 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25544 * reload.c (find_reloads): When reloading a nonoffsettable address,
25545 use RELOAD_OTHER for it and its address reloads.
25547 PR rtl-optimization/79910
25548 * combine.c (record_used_regs): New static function.
25549 (try_combine): Handle situations where there is an additional
25550 instruction between I2 and I3 which needs to have a LOG_LINK
25553 2017-03-17 Jeff Law <law@redhat.com>
25555 PR tree-optimization/71437
25556 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
25557 conditional in the hash table first.
25558 (vrp_dom_walker::before_dom_children): Extract condition from
25559 ASSERT_EXPR. Record condition, its inverion and any implied
25560 conditions as well.
25562 2017-03-17 Marek Polacek <polacek@redhat.com>
25563 Markus Trippelsdorf <markus@trippelsdorf.de>
25565 PR tree-optimization/80079
25566 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
25569 2017-03-17 Richard Biener <rguenther@suse.de>
25571 PR middle-end/80075
25572 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
25573 Properly verify the LHS before the RHS possibly claims to be
25575 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
25578 2017-03-17 Martin Jambor <mjambor@suse.cz>
25580 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
25581 (List of -O2 options): Likewise.
25582 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
25585 2017-03-17 Tom de Vries <tom@codesourcery.com>
25587 * gcov-dump.c (print_usage): Print bug_report_url.
25589 2017-03-17 Richard Biener <rguenther@suse.de>
25591 PR middle-end/80050
25592 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
25593 (parser::peek): Likewise.
25595 2017-03-17 Richard Biener <rguenther@suse.de>
25597 PR tree-optimization/80048
25598 * sese.c (free_sese_info): Properly release rename_map and
25599 copied_bb_map elements.
25601 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25603 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25604 Add linked-list forward and backlinks. Insert on
25605 construction, remove on destruction.
25606 (class pass_store_merging): Add m_stores_head field.
25607 (pass_store_merging::terminate_and_process_all_chains):
25608 Iterate over m_stores_head list.
25609 (pass_store_merging::terminate_all_aliasing_chains):
25611 (pass_store_merging::execute): Check for debug stmts first.
25612 Push new chains onto the m_stores_head stack.
25614 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25617 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25618 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25619 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25621 2017-03-16 Jeff Law <law@redhat.com>
25623 PR tree-optimization/71437
25624 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25625 member function. Implementation moved into after_dom_children
25626 member function and into the threader's thread_outgoing_edges
25628 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25629 some code into new thread_outgoing_edges.
25630 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25631 definition. Simplify marker handling (do it here). Assume we always
25632 have the available expression and the const/copies tables.
25633 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25635 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25636 * tree-vrp.c (equiv_stack): No longer file scoped.
25637 (vrp_dom_walker): New class.
25638 (vrp_dom_walker::before_dom_children): New member function.
25639 (vrp_dom_walker::after_dom_children): Likewise.
25640 (identify_jump_threads): Setup domwalker. Use it rather than
25641 walking edges in a random order by hand. Simplify setup/finalization.
25642 (finalize_jump_threads): Remove.
25643 (vrp_finalize): Do not call identify_jump_threads here.
25644 (execute_vrp): Do it here instead and call thread_through_all_blocks
25647 PR tree-optimization/71437
25648 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25650 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25652 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25653 (dom_opt_dom_walker::thread_across_edge): Remove
25654 handle_dominating_asserts argument. All callers changed.
25655 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25656 changes. Remove calls to lhs_of_dominating_assert. Other
25657 uses of handle_dominating_asserts turn into unconditional code
25658 (simplify_control_stmt_condition_1): Likewise.
25659 (simplify_control_stmt_condition): Likewise.
25660 (thread_through_normal_block, thread_across_edge): Likewise.
25661 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25662 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25663 object if it is not an SSA_NAME.
25664 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25665 before calling into the VRP specific simplifiers.
25666 (identify_jump_threads): Remove handle_dominating_asserts
25669 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25672 * tree-diagnostic.c (default_tree_printer): No longer static.
25673 * tree-diagnostic.h (default_tree_printer): New prototype.
25675 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25677 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25678 Change ins into fmov.
25680 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25682 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25683 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25684 Use h_con constraint for operand 1.
25685 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25686 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25688 2017-03-15 Jeff Law <law@redhat.com>
25690 PR tree-optimization/71437
25691 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25692 (record_temporary_equivalences): Use it.
25694 PR tree-optimization/71437
25695 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25696 tree-ssa-scopedtables.
25697 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25698 (record_conditions, record_cond, vuse_eq): Likewise.
25699 (record_edge_info): Adjust to API tweak of record_conditions.
25700 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25701 (record_temporary_equivalences, optimize_stmt): Likewise.
25702 (eliminate_redundant_computations): Likewise.
25703 (record_equivalences_from_stmt): Likewise.
25704 * tree-ssa-scopedtables.c: Include options.h and params.h.
25705 (vuse_eq): New function, moved from tree-ssa-dom.c
25706 (build_and_record_new_cond): Likewise.
25707 (record_conditions): Likewise. Accept vector of conditions rather
25708 than edge_equivalence structure for first argument.
25709 for the first argument.
25710 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25711 from tree-ssa-dom.c.
25712 (avail_exprs_stack::record_cond): Likewise.
25713 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25714 from tree-ssa-dom.c.
25715 (avail_exprs_stack): Add new member functions lookup_avail_expr
25717 (record_conditions): Declare.
25719 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25722 * lra-constraints.c (process_alt_operands): Increase reject for
25723 reloading an input/output operand.
25725 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25728 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25729 insns to convert from signed/unsigned char/short to IEEE 128-bit
25731 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25733 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25736 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25737 subreg of inner mode for values already in registers.
25739 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25741 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25742 iteration reg is used after the loop.
25744 2017-03-14 Martin Sebor <msebor@redhat.com>
25746 PR tree-optimization/79800
25747 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25748 precision in negative-positive range.
25749 (format_floating): Call non-const overload with adjusted precision.
25751 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25754 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25757 2017-03-14 Martin Sebor <msebor@redhat.com>
25759 PR middle-end/80020
25760 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25761 * builtins.def (aligned_alloc): Use it.
25764 * Makefile.in (GTFILES): Add calls.c.
25765 * calls.c: Include "gt-calls.h".
25767 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25769 PR rtl-optimization/79728
25770 * regs.h (struct target_regs): New field
25771 x_contains_allocatable_regs_of_mode.
25772 (contains_allocatable_regs_of_mode): New macro.
25773 * reginfo.c (init_reg_sets_1): Initialize it, and change
25774 contains_reg_of_mode so it includes global regs as well.
25775 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25776 rather than contains_regs_of_mode.
25778 2017-03-14 Martin Liska <mliska@suse.cz>
25780 * doc/invoke.texi: Document options that can't be combined with
25781 -fcheck-pointer-bounds.
25783 2017-03-14 Martin Liska <mliska@suse.cz>
25785 PR middle-end/79831
25786 * doc/invoke.texi (-Wchkp): Document the option.
25788 2017-03-14 Martin Liska <mliska@suse.cz>
25790 * Makefile.in: Install gcov-dump.
25792 2017-03-14 Martin Liska <mliska@suse.cz>
25794 * multiple_target.c (expand_target_clones): Bail out for
25795 an invalid attribute.
25797 2017-03-14 Richard Biener <rguenther@suse.de>
25799 * alias.c (struct alias_set_entry): Pack properly.
25800 * cfgloop.h (struct loop): Likewise.
25801 * cse.c (struct set): Likewise.
25802 * ipa-utils.c (struct searchc_env): Likewise.
25803 * loop-invariant.c (struct invariant): Likewise.
25804 * lra-remat.c (struct cand): Likewise.
25805 * recog.c (struct change_t): Likewise.
25806 * rtl.h (struct address_info): Likewise.
25807 * symbol-summary.h (function_summary): Likewise.
25808 * tree-loop-distribution.c (struct partition): Likewise.
25809 * tree-object-size.c (struct object_size_info): Likewise.
25810 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25811 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25812 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25813 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25814 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25815 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25816 (struct _stmt_vec_info): Likewise.
25818 2017-03-14 Martin Liska <mliska@suse.cz>
25821 * multiple_target.c (create_dispatcher_calls): Check that
25822 a target can create a function dispatcher.
25824 2017-03-14 Martin Liska <mliska@suse.cz>
25827 * multiple_target.c (expand_target_clones): Drop local.local
25828 flag for default implementation.
25830 2017-03-14 Richard Biener <rguenther@suse.de>
25832 PR tree-optimization/80030
25833 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25835 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25837 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25838 gcc_fallthrough() instead of __attribute__((fallthrough));
25840 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25842 * doc/gcc.texi: Remove "up" link to (DIR).
25843 * doc/gccint.texi: Ditto.
25845 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25847 * doc/install.texi (Specific) <avr>: Remove reference to
25850 2017-03-13 Jeff Law <law@redhat.com>
25852 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25853 attribute rather than comments.
25855 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25856 match_scratch operand is highest.
25858 2017-03-13 Martin Liska <mliska@suse.cz>
25860 PR middle-end/78339
25861 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25862 is a CHKP clone, use original declaration.
25864 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25866 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25867 (arc_conditional_register_usage): Use a different allocation order
25868 when optimizing for size.
25869 * common/config/arc/arc-common.c (arc_option_optimization_table):
25870 Section anchors default on when optimizing for size.
25872 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25874 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25876 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25878 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25879 * config/arc/arc.md (cpu_facility): Add cd variant.
25880 (*movqi_insn): Add code density variant.
25881 (*movhi_insn): Likewise.
25882 (*movqi_insn): Likewise.
25883 (*addsi3_mixed): Likewise.
25884 (subsi3_insn): Likewise.
25886 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25888 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25890 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25892 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25893 expressions with MINUS and UNARY ops.
25895 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25898 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25900 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25901 between vec_select and vector argument.
25902 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25903 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25904 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25905 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25906 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25907 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25909 2017-03-13 Richard Biener <rguenther@suse.de>
25912 * params.def (vect-max-peeling-for-alignment): Fix typo.
25914 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25916 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25917 issue that only occurred with binutils below 2.18.
25919 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25921 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25922 refer to binutils 2.11/2.12 minimum.
25924 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25926 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25927 ftp.kernel.org and simplify binutils requirement.
25929 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25931 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25933 (Optimize Options): Ditto. Also remove redundancy.
25935 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25937 PR translation/79848
25938 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25940 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25941 to G_ to avoid double translation.
25943 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25945 PR translation/79923
25946 * auto-profile.c (get_combined_location): Convert leading
25947 character of diagnostics to lower case and remove trailing period.
25948 (read_profile): Likewise for various diagnostics.
25949 * config/arm/arm.c (arm_option_override): Remove trailing period
25950 from various diagnostics.
25951 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25952 (msp430_expand_delay_cycles): Likewise.
25954 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25957 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25958 full command-line argument, rather than just "str".
25959 (aarch64_validate_march): Likewise.
25960 (aarch64_validate_mtune): Likewise.
25962 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25964 PR rtl-optimization/78911
25965 * lra-assigns.c (must_not_spill_p): New function.
25966 (spill_for): Use it.
25968 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25970 PR tree-optimization/79981
25971 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25972 ATOMIC_COMPARE_EXCHANGE ifn result.
25973 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25974 IFN_ATOMIC_COMPARE_EXCHANGE.
25976 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25979 * opts.c (parse_sanitizer_options): Add missing question mark to
25980 "did you mean" message.
25982 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25984 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25986 (VMULEUH_UNS): Likewise.
25987 (VMULOUB_UNS): Likewise.
25988 (VMULOUH_UNS): Likewise.
25989 * config/rs6000/rs6000.c (builtin_function_type): Remove
25990 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25992 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25995 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25996 x with result of extra_parsing_for_operand_code_0.
25997 (function_reader::extra_parsing_for_operand_code_0): Convert
25998 return type from void to rtx, returning x. When reading
25999 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26000 larger size containing struct block_symbol.
26002 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26004 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26005 -mfloat128-hardware without -m64.
26007 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26010 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26011 entries to the case statement that marks unsigned arguments to
26012 overloaded functions.
26014 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26017 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26019 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26022 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26023 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26025 2017-03-10 Martin Liska <mliska@suse.cz>
26029 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26030 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26033 2017-03-10 Olivier Hainque <hainque@adacore.com>
26035 * tree-switch-conversion (array_value_type): Start by resetting
26036 candidate type to it's main variant.
26038 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26040 PR rtl-optimization/79909
26041 * combine.c (try_combine): Use simplify_replace_rtx on individual
26042 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26043 of the whole CALL_INSN_FUNCTION_USAGE.
26045 PR tree-optimization/79972
26046 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26047 get_range_info on SSA_NAMEs. Formatting fixes.
26049 2017-03-10 Richard Biener <rguenther@suse.de>
26050 Jakub Jelinek <jakub@redhat.com>
26052 PR tree-optimization/77975
26053 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26054 edge to be constant.
26055 (get_val_for): For constant x return it. Formatting fix.
26056 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26057 would use the same bases as the current one.
26059 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26061 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26062 instead of vec_select for V1TImode.
26063 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26065 (VSX_LE_128): Add V1TI to this mode iterator.
26066 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26067 (*vsx_le_perm_store_<mode>): Likewise.
26068 (pre-reload splitter for VSX stores): Likewise.
26069 (post-reload splitter for VSX stores): Likewise.
26070 (*vsx_xxpermdi2_le_<mode>): Likewise.
26071 (*vsx_lxvd2x2_le_<mode>): Likewise.
26072 (*vsx_stxvd2x2_le_<mode>): Likewise.
26074 2017-03-09 Michael Eager <eager@eagercon.com>
26076 Correct failures with --enable-checking=yes,rtl.
26078 * config/microblaze/microblaze.c (microblaze_expand_shift):
26079 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26080 test for const0_rtx.
26081 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26082 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26084 2017-03-09 Richard Biener <rguenther@suse.de>
26086 PR tree-optimization/79977
26087 * graphite-scop-detection.c (scop_detection::merge_sese):
26088 Handle the case of extra exits to blocks dominating the entry.
26090 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26092 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26095 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26097 PR rtl-optimization/79949
26098 * lra-constraints.c (process_alt_operands): Check memory when
26099 trying to predict a cycle. Print about the overall increase.
26101 2017-03-09 Richard Biener <rguenther@suse.de>
26103 PR middle-end/79971
26104 * gimple-expr.c (useless_type_conversion_p): Preserve
26105 TYPE_SATURATING for fixed-point types.
26107 2017-03-09 Richard Biener <rguenther@suse.de>
26110 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26111 alignment of BLKmode params.
26113 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26116 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26117 (VALL_NO_V2Q): Likewise.
26119 * config/aarch64/aarch64-simd.md
26120 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26122 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26123 VALL_NO_V2Q mode iterator.
26124 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26126 2017-03-09 Martin Liska <mliska@suse.cz>
26128 PR tree-optimization/79631
26129 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26130 tree_int_cst_sign_bit just for INTEGER constants.
26132 2017-03-09 Martin Liska <mliska@suse.cz>
26136 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26139 2017-03-09 Marek Polacek <polacek@redhat.com>
26142 * tree.c (inchash::add_expr): Handle TREE_VEC.
26144 2017-03-09 Martin Liska <mliska@suse.cz>
26147 (chkp_narrow_size_and_offset): New function.
26148 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26149 (void chkp_parse_bit_field_ref): New function.
26150 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26151 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26153 2017-03-09 Martin Liska <mliska@suse.cz>
26156 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26157 (chkp_find_bounds_1): Remove gcc_unreachable.
26159 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26162 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26163 BUILT_IN_SYNC*, determine the access type from the size suffix and
26164 always build a MEM_REF with that type. Handle forgotten
26165 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26168 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26169 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26170 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26171 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26172 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26173 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26174 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26175 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26176 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26177 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26178 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26179 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26180 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26181 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26182 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26183 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26184 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26185 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26186 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26187 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26188 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26189 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26190 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26191 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26192 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26193 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26194 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26195 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26196 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26197 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26198 definitions outside of __OPTIMIZE__ guarded section.
26201 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26202 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26203 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26204 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26207 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26209 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26210 ("vfenez<mode>"): Add missing constraints.
26212 2017-03-08 Martin Sebor <msebor@redhat.com>
26215 * config/nds32/nds32.c (nds32_option_override):
26216 Fix misspelled diagnostic.
26218 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26221 * gimplify.c (gimplify_omp_for): Replace index var in outer
26222 taskloop statement with an artificial variable and add
26223 OMP_CLAUSE_PRIVATE clause for it.
26225 2017-03-08 Richard Biener <rguenther@suse.de>
26227 PR tree-optimization/79955
26228 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26229 for accesses that are completely outside of the variable.
26231 2017-03-08 Andrew Haley <aph@redhat.com>
26233 PR tree-optimization/79943
26234 * tree-ssa-loop-split.c (compute_new_first_bound): When
26235 calculating the new upper bound, (END-BEG) should be added, not
26238 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26240 * config/avr/avr.md (setmemhi): Make sure match_dup
26241 operand number comes before match_scratch.
26243 2017-03-08 Richard Biener <rguenther@suse.de>
26245 PR tree-optimization/79920
26246 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26247 with ncopies == 1 to ...
26248 (vect_transform_slp_perm_load): ... here. Properly compute
26249 all element loads by iterating VF times over the group. Do
26250 not handle ncopies (computed in a broken way) in
26251 vect_create_mask_and_perm.
26253 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26256 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26257 is a uniform vector, use uniform_vector_p return value instead of
26258 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26260 2017-03-07 Marek Polacek <polacek@redhat.com>
26262 PR middle-end/79809
26263 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26264 (alloca_call_type): Likewise.
26266 2017-03-07 Martin Liska <mliska@suse.cz>
26268 * gcov.c (process_args): Put comment to correct location.
26270 2017-03-07 Martin Liska <mliska@suse.cz>
26272 PR middle-end/68270
26273 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26274 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26275 (chkp_narrow_bounds_for_field): Likewise.
26276 (chkp_parse_array_and_component_ref): Pass one more argument to
26279 2017-03-07 Richard Biener <rguenther@suse.de>
26281 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26284 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26286 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26287 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26289 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26292 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26293 to end of description.
26294 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26296 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26298 PR rtl-optimization/79901
26299 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26301 (*avx512f_<code><mode>3<mask_name>): ... this.
26302 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26303 iterator instead of VI8_AVX2_AVX512BW.
26305 PR rtl-optimization/79901
26306 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26307 min/max expander, expand it using expand_vec_cond_expr.
26310 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26313 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26316 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26317 to void * for PCH reasons.
26318 * dwarf2out.c (output_loc_operands, output_die): Cast
26319 v.val_vec.array to unsigned char *.
26321 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26324 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26327 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26329 PR rtl-optimization/79571
26330 * lra-constraints.c (process_alt_operands): Calculate static
26331 reject and subtract it from overall when only addresses will be
26334 2017-03-06 Julia Koval <julia.koval@intel.com>
26337 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26338 incoming stack boundary to 128 for 64-bit targets.
26340 2017-03-06 Richard Biener <rguenther@suse.de>
26342 PR tree-optimization/79894
26343 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26344 to NULL after folding it.
26346 2017-03-06 Richard Biener <rguenther@suse.de>
26348 PR tree-optimization/79824
26349 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26350 check disabling peeling for gaps.
26352 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26354 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26355 attributes): Document gettimeofday.
26357 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26359 * config/s390/s390.c (s390_option_override_internal): Set
26360 PARAM_MIN_VECT_LOOP_BOUND
26362 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26364 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26365 * config/s390/s390.md: Likewise.
26367 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26370 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26371 (<avx2_avx512>_perm<mode>): Rename to ...
26372 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26374 (<avx512>_perm<mode>_mask): Rename to ...
26375 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26377 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26378 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26379 instead of VI8F_256_512.
26380 (avx512f_perm<mode>): New define_expand.
26381 (avx512f_perm<mode>_mask): Likewise.
26382 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26383 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26385 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26387 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26388 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26390 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26392 2017-03-06 Martin Liska <mliska@suse.cz>
26395 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26396 when having a SSA NAME w/o VAR_DECL assigned to it.
26398 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26400 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26401 msa_dpsub_<su>_d): Fix MODE for vec_select.
26403 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26405 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26407 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26409 2017-03-06 Richard Biener <rguenther@suse.de>
26411 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26412 * plugin.c (register_plugin_info): Likewise.
26413 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26415 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26417 * config/i386/sse.md (sse_storehps, sse_storelps,
26418 avx_<castmode><avxsizesuffix>_<castmode>,
26419 avx512f_<castmode><avxsizesuffix>_<castmode>,
26420 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26421 in condition that at least one operand is not a MEM.
26423 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26425 PR middle-end/79805
26426 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26427 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26429 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26430 gimple_call_nothrow_p flag based on whether original builtin can throw.
26431 If it can, emit following stmts on the fallthrough edge.
26432 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26433 don't create new bb if inserting just debug stmts on the edge, try to
26434 insert them on the fallthru bb or just reset debug stmts.
26436 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26439 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26440 restore recog_data (including the operand rtxes inside it) around
26441 the call to get_insn_template.
26443 2017-03-03 Martin Sebor <msebor@redhat.com>
26445 PR tree-optimization/79699
26446 * context.c (context::~context): Free MPFR caches to avoid
26447 a memory leak on program exit.
26449 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26451 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26452 Use wide_int::ulow () instead of .elt (0).
26454 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26456 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26457 (*pushxf): Limit oF constraint to 32bit targets and add oC
26458 constraint for 64bit targets.
26459 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26460 (*pushdf): Change rmF constraint to rmC.
26462 2017-03-03 Martin Liska <mliska@suse.cz>
26464 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26465 Remove unused variable.
26467 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26470 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26471 is a memory operand, increase num_memory.
26472 (ix86_expand_args_builtin): Likewise.
26474 2017-03-03 Jan Hubicka <jh@suse.cz>
26477 * ipa-devirt.c (maybe_record_node): Properly handle
26478 __cxa_pure_virtual visibility.
26480 2017-03-03 Martin Liska <mliska@suse.cz>
26482 PR tree-optimization/79803
26483 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26485 (pass_loop_prefetch::execute): Disabled optimization if an
26486 assumption about L1 cache size is not met.
26488 2017-03-03 Martin Liska <mliska@suse.cz>
26490 PR rtl-optimization/79574
26491 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26492 (hash_scan_set): Likewise.
26493 (dump_hash_table): Likewise.
26494 (hoist_code): Likewise.
26496 2017-03-03 Richard Biener <rguenther@suse.de>
26498 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26499 in place of uhigh ().
26500 (fixed_convert_from_real): Likewise.
26502 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26505 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26507 2017-03-03 Richard Biener <rguenther@suse.de>
26509 PR middle-end/79818
26510 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26511 TYPE_OVERFLOW_UNDEFINED check.
26513 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26515 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
26517 (vector_ae_<mode>_p): Likewise.
26518 (vector_nez_<mode>_p): Likewise.
26519 (vector_ne_v2di_p): Likewise.
26520 (vector_ae_v2di_p): Likewise.
26521 (vector_ne_<mode>_p): Likewise.
26522 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
26524 (vsx_tsqrt<mode>2_fe): Likewise.
26526 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
26529 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
26531 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26533 PR rtl-optimization/79780
26534 * cprop.c (one_cprop_pass): When second and further conditional trap
26535 in a single basic block is turned into an unconditional trap, turn it
26536 into a deleted note to avoid RTL verification failures.
26538 2017-03-02 Richard Biener <rguenther@suse.de>
26540 * fold-const.c (const_binop): Use ulow () instead of elt (0).
26542 2017-03-02 Richard Biener <rguenther@suse.de>
26544 PR tree-optimization/79345
26546 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
26547 param and abort the walk, returning -1 if it is hit.
26548 (walk_aliased_vdefs): Take a limit param and pass it on.
26549 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
26550 defaulting to 0 and return a signed int.
26551 * tree-ssa-uninit.c (struct check_defs_data): New struct.
26552 (check_defs): New helper.
26553 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
26554 about uninitialized memory.
26555 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
26556 bogus uninitialized warning.
26557 (fixed_convert_from_real): Likewise.
26559 2017-03-02 Bin Cheng <bin.cheng@arm.com>
26561 PR tree-optimization/66768
26562 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
26563 iv_use if base object can't be determined.
26565 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26567 PR tree-optimization/79345
26568 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
26569 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
26570 (get_pattern_stats): Initialize it.
26571 * genemit.c (gen_expand): Verify match_scratch numbers come after
26572 match_operand/match_dup numbers.
26573 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
26574 match_scratch numbers.
26575 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
26577 * config/s390/s390.md (trunctdsd2): Likewise.
26579 2017-03-02 Richard Biener <rguenther@suse.de>
26581 * wide-int.h (wide_int_storage::operator=): Implement in terms
26584 2017-03-02 Richard Biener <rguenther@suse.de>
26586 PR tree-optimization/79777
26587 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
26588 the to insert expression to sth existing.
26590 2017-03-01 Martin Sebor <msebor@redhat.com>
26592 PR middle-end/79692
26593 * gimple-ssa-sprintf.c
26594 (directive::known_width_and_precision): New function.
26595 (format_integer): Use it.
26596 (get_mpfr_format_length): Consider the full range of precision
26597 when computing %g output with the # flag. Set the likely byte
26598 count to 3 rather than 1 when precision is indeterminate.
26599 (format_floating): Correct the lower bound of precision.
26601 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26603 * doc/invoke.texi: Document default code model for 64-bit Linux.
26605 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26608 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26609 udiv rather than div since input pattern is unsigned.
26611 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26613 * config/i386/i386.c (print_reg): Warn for values of
26614 unsupported size in integer register.
26616 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26619 * config/rs6000/predicates.md (current_file_function_operand): Do
26620 not allow self calls to be local if the function is replaceable.
26622 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26625 * config/rs6000/altivec.h (vec_ctz and others): Change the
26626 preprocessor macro that controls conditional compilation from
26627 _ARCH_PWR9 to __POWER9_VECTOR__.
26628 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26629 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26630 control (instead of _ARCH_PWR9 control) so that template
26631 definition uses power9-specific function.
26632 (vec_any_eq): Likewise.
26633 (vec_all_ne): Change macro definition to use a power9-specific
26634 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26635 _ARCH_PWR9 control).
26636 (vec_any_eq) Likewise.
26637 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26638 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26639 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26640 support for xvcmpnedp instruction.
26641 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26642 macro expansion so that Power9 implementation of vec_all_ne does
26643 not use the AltiVec predicate framework.
26644 (VCMPNEH_P): Likewise.
26645 (VCMPNEW_P): Likewise.
26646 (VCMPNED_P): Likewise.
26647 (VCMPNEFP_P): Likewise.
26648 (VCMPNEDP_P): Likewise.
26649 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26650 implementation of vec_any_eq to not use AltiVec predicate
26652 (VCMPAEH_P): Likewise.
26653 (VCMPAEW_P): Likewise.
26654 (VCMPAED_P): Likewise.
26655 (VCMPAEFP_P): Likewise.
26656 (VCMPAEDP_P): Likewise.
26657 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26658 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26659 not use the AltiVec predicate framework.
26660 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26661 of vec_any_eq to not use AltiVec predicate framework.
26662 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26663 support for predefined __POWER9_VECTOR__ macro to indicate that
26664 Power9 instruction selection is enabled.
26665 (altivec_overloaded_builtins): Remove extraneous
26666 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26667 function argument types RS6000_BTI_bool_V16QI and
26668 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26669 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26670 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26671 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26672 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26673 Power9 for implementations of vec_cmpne. Change the signature for
26674 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26675 (representing vec_all_ne) to remove the previously described first
26676 argument of type RS6000_BTI_INTSI, as this was an artifact of
26677 reliance on the AltiVec predicate framework, which is no longer
26678 used in the implementation of these functions. Add
26679 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26680 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26681 since, unlike the AltiVec predicate framework implementation, we
26682 do not share function descriptors between vec_alle and vec_anyeq.
26683 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26684 set of modes that receive special treatment even when
26685 TARGET_P9_VECTOR is true. The special treatment emits code that
26686 does not depend on Power9 instructions.
26687 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26688 define_expand to not rely on AltiVec predicate framework.
26689 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26691 (vector_ne_v2di_p): Change this define_expand to not rely on
26692 AltiVec predicate framework.
26693 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26695 (vector_ne_<mode>_p): Change this define_expand to not rely on
26696 AltiVec predicate framework.
26697 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26699 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26700 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26701 define_insn pattern.
26702 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26703 define_insn pattern because the xvcmpne<VSs>. instruction is not
26705 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26706 instruction is not supported.
26708 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26710 * config/nvptx/nvptx.c: Include intl.h.
26712 2017-03-01 Martin Jambor <mjambor@suse.cz>
26715 * ipa-prop.h (ipa_bits): Removed field known.
26716 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26717 to pointers. Adjusted their comments to warn about their sharing.
26718 (ipcp_transformation_summary): Change bits to a vector of pointers.
26719 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26720 (ipa_get_ipa_bits_for_value): Declare.
26721 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26722 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26723 (ipa_bits_hash_table): Likewise.
26724 (ipa_vr_ggc_hash_traits): Likewise.
26725 (ipa_vr_hash_table): Likewise.
26726 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26727 being pointers and vr_known being removed.
26728 (ipa_set_jf_unknown): Likewise.
26729 (ipa_get_ipa_bits_for_value): New function.
26730 (ipa_set_jfunc_bits): Likewise.
26731 (ipa_get_value_range): New overloaded functions.
26732 (ipa_set_jfunc_vr): Likewise.
26733 (ipa_compute_jump_functions_for_edge): Use the above functions to
26734 construct bits and vr parts of jump functions.
26735 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26736 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26738 (ipcp_grow_transformations_if_necessary): Also allocate
26739 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26741 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26742 them. Fix too long lines.
26743 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26744 vr_known being removed.
26745 (ipa_read_jump_function): Use new setter functions to construct bits
26746 and vr parts of jump functions or set them to NULL.
26747 (write_ipcp_transformation_info): Adjust for bits being pointers.
26748 (read_ipcp_transformation_info): Likewise.
26749 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26751 Include gt-ipa-prop.h.
26752 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26754 (ipcp_store_bits_results): Likewise.
26755 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26756 Do not write to existing jump functions but use a temporary instead.
26758 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26761 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26762 attempt to use its first operand as BIT_FIELD_REF base.
26764 2017-03-01 Richard Biener <rguenther@suse.de>
26766 PR middle-end/79721
26767 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26768 interpolating formula in wrapping arithmetic.
26769 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26771 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26773 PR tree-optimization/79734
26774 * tree-vect-generic.c (expand_vector_condition): Optimize
26775 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26776 Handle VEC_COND_EXPR where comparison has different inner width from
26777 type's inner width.
26779 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26781 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26782 markup, and similar issues. Remove @opindex entries for things
26783 that aren't options. Add missing -mmpy-option entries.
26785 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26787 PR tree-optimization/79737
26788 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26789 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26790 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26791 instead of byte_size. Formatting fix.
26792 (shift_bytes_in_array_right): Formatting fix.
26794 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26797 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26798 condition on optimize for the leaf function test.
26800 2017-02-28 Martin Liska <mliska@suse.cz>
26803 * read-rtl-function.c (function_reader::handle_unknown_directive):
26804 Bail out when one uses -flto.
26806 2017-02-28 Martin Liska <mliska@suse.cz>
26808 * common.opt: Replace space with tabular for options of <number>
26810 * config/i386/i386.opt: Show <number> value for
26811 -mlarge-data-threshold.
26812 * opts.c (print_filtered_help): Do not display number in hexadecimal
26815 2017-02-28 Martin Liska <mliska@suse.cz>
26817 * common.opt: Fix --help=option -Q for options which are of
26820 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26822 * config/i386/i386.c (print_reg): Error out for values
26823 of 8-bit size in invalid integer register.
26825 2017-02-28 Martin Sebor <msebor@redhat.com>
26827 PR tree-optimization/79691
26828 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26830 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26833 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26834 gcc_unreachable with output_operand_lossage.
26836 2017-02-28 Richard Biener <rguenther@suse.de>
26838 PR tree-optimization/79740
26839 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26841 (visit_nary_op): Insert the nary into the hashtable if we
26842 pattern-matched sth.
26843 * tree-ssa-pre.c (eliminate_insert): Robustify.
26845 2017-02-28 Richard Biener <rguenther@suse.de>
26847 PR middle-end/79731
26848 * fold-const.c (decode_field_reference): Reject out-of-bound
26851 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26853 * config/i386/i386.c: Include intl.h.
26854 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26855 instead of just cond ? "..." : "...".
26856 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26857 * coverage.c (read_counts_file): Likewise.
26858 * omp-offload.c: Include intl.h.
26859 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26860 of just cond ? "..." : "...".
26861 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26862 of just cond ? "..." : "...".
26864 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26867 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26869 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26870 'tune for' CPU name.
26871 * config/arm/arm-cpu-data.h: Regenerated.
26873 2017-02-28 Richard Biener <rguenther@suse.de>
26875 PR tree-optimization/79732
26876 * tree-inline.c (expand_call_inline): Do not shadow var.
26878 2017-02-28 Richard Biener <rguenther@suse.de>
26880 PR tree-optimization/79723
26881 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26882 address-space properly.
26884 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26886 * doc/optinfo.texi (Optimization groups): Fix option used for
26888 * doc/invoke.texi (-fopt-info): Document "omp".
26889 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26890 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26891 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26892 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26893 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26895 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26897 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26898 OPTGROUP_OMP with "omp".
26900 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26903 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26904 for arithmetic shift of unsigned V2DI.
26906 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26908 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26909 arc/linux.h headers.
26910 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26911 (LINK_SPEC): Likewise.
26912 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26913 (EXTRA_SPECS): Likewise.
26914 (STARTFILE_SPEC): Likewise.
26915 (ENDFILE_SPEC): Likewise.
26916 (LIB_SPEC): Likewise.
26917 (TARGET_SDATA_DEFAULT): Likewise.
26918 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26919 (MULTILIB_DEFAULTS): Likewise.
26920 (DWARF2_UNWIND_INFO): Likewise.
26921 * config/arc/big.h: New file.
26922 * config/arc/elf.h: Likewise.
26923 * config/arc/linux.h: Likewise.
26924 * config/arc/t-uClibc: Remove.
26926 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26928 PR tree-optimization/77536
26929 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26930 (tree_transform_and_unroll_loop): Use above function to compute the
26931 estimated niter of unrolled loop and use it when scaling profile.
26932 Also use count info rather than frequency if it's non-zero.
26933 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26934 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26935 (vect_transform_loop): Call above function.
26937 2017-02-27 Richard Biener <rguenther@suse.de>
26939 PR tree-optimization/45397
26940 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26941 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26942 (visit_nary_op): Add pattern matching for CSEing sign-changed
26943 or truncated operations with wider ones.
26945 2017-02-27 Richard Biener <rguenther@suse.de>
26947 PR tree-optimization/79690
26948 * tree-vect-stmts.c (vectorizable_store): Use vector type
26949 built from the DR with address-space.
26951 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26953 * doc/invoke.texi (Optimize Options): Refine the description
26954 of asan-use-after-return.
26956 2017-02-25 Alan Modra <amodra@gmail.com>
26958 PR rtl-optimization/79584
26959 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26960 base, not ad->base_term, the reg within base. Remove assertion
26961 that ad->base == ad->base_term. Replace gen_int_mode using
26962 bogus mode with const0_rtx.
26964 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26966 PR middle-end/79396
26967 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26968 FMA_EXPR like tcc_binary or tcc_unary.
26970 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26973 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26975 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26976 (output_loc_operands): Handle DW_OP_call_ref and
26977 DW_OP_GNU_variable_value.
26978 (struct variable_value_struct): New type.
26979 (struct variable_value_hasher): Likewise.
26980 (variable_value_hash): New variable.
26981 (string_types): Remove.
26982 (copy_loc_descr): New function.
26983 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26984 (prepend_loc_descr_to_each): New function.
26985 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26986 instead of add_loc_descr_to_each if the first argument is single
26987 location list and the second has multiple.
26988 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26989 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26990 when looking for variable value which doesn't have other location info.
26991 (loc_list_from_tree): Formatting fix.
26992 (gen_array_type_die): Simplify DW_AT_string_length handling.
26993 (adjust_string_types): Remove.
26994 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26995 string_types. Call resolve_variable_values.
26996 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26997 (resolve_addr_in_expr): Likewise. Add A argument.
26998 (copy_deref_exprloc): Remove deref argument. Adjust for the
26999 original expression being DW_OP_GNU_variable_value with optionally
27000 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27001 optionally after it.
27002 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27003 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27004 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27005 (variable_value_hasher::hash, variable_value_hasher::equal): New
27007 (resolve_variable_value_in_expr, resolve_variable_value,
27008 resolve_variable_values, note_variable_value_in_expr,
27009 note_variable_value): New functions.
27010 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27013 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27016 * opts.h (handle_generated_option): Add GENERATED_P argument.
27017 * opts-common.c (handle_option): Adjust function comment.
27018 (handle_generated_option): Add GENERATED_P argument, pass it to
27020 (control_warning_option): Pass false to handle_generated_option
27022 * opts.c (maybe_default_option): Pass true to handle_generated_option
27024 * optc-gen.awk: Likewise.
27026 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27028 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27029 a REG, look at the REG it is a SUBREG of.
27030 (splitter for cmpeqsi_t): Ditto.
27032 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27034 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27035 the special USEs with the pattern of the insn, not the insn itself.
27037 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27040 * doc/invoke.texi: Document -mload-store-pairs.
27042 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27043 Sandra Loosemore <sandra@codesourcery.com>
27045 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27046 argument isn't a CONST_INT.
27047 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27048 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27049 (nios2_validate_compare): Bypass alternate compare logic if *op2
27050 is not a CONST_INT.
27051 (ldstwm_operation_p): Return false if first_base is not a REG or
27052 if first_offset is not a CONST_INT.
27054 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27056 * config/cris/cris.md: Use correct operand in a define_peephole2.
27058 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27060 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27062 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27064 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27065 this_insn if it is an INSN or JUMP_INSN.
27066 (force_offsettable): Look at base, not at addr.
27067 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27068 on things that aren't necessarily CONST_INTs.
27070 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27072 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27073 -mfpmath=sse is the default also for x86-32 targets with SSE2
27074 instruction set when @option{-ffast-math} is enabled
27076 2017-02-24 Jeff Law <law@redhat.com>
27078 PR rtl-optimizatoin/79286
27079 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27082 2017-02-24 Richard Biener <rguenther@suse.de>
27084 PR tree-optimization/79389
27085 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27088 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27090 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27091 function comment to reflect reality.
27092 (loop_exits_before_overflow): Fix typo in function description.
27094 2017-02-24 Richard Biener <rguenther@suse.de>
27096 PR tree-optimization/79389
27097 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27098 properly that a threading opportunity exists. Detect conditional
27099 copy/constant propagation opportunities.
27101 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27103 * config/visium/visium.md (type): Add trap.
27104 (b): New mode attribute.
27105 (*btst): Rename into...
27106 (*btst<mode>): ...this and adjust.
27107 (*cbranchsi4_btst_insn): Rename into...
27108 (*cbranch<mode>4_btst_insn): ...this and adjust.
27109 (trap): New define_insn.
27111 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27113 PR tree-optimization/79389
27114 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27115 (noce_reversed_cond_code): New function.
27116 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27117 reversed_comparison_code. Formatting fix.
27118 (noce_try_store_flag): Test rev_cond != NULL in addition to
27119 reversed_comparison_code.
27120 (noce_try_store_flag_constants): Likewise.
27121 (noce_try_store_flag_mask): Likewise.
27122 (noce_try_addcc): Use rev_cond if non-NULL instead of
27123 reversed_comparison_code.
27124 (noce_try_cmove_arith): Likewise. Formatting fixes.
27125 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27126 (noce_find_if_block): Initialize rev_cond.
27127 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27128 instead of false as last argument never attempt to reverse it
27131 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27133 PR tree-optimization/79663
27134 * tree-predcom.c (combine_chains): Process refs in reverse order
27135 only for ZERO length chains, and add explaining comment.
27137 2017-02-23 Jeff Law <law@redhat.com>
27139 PR tree-optimization/79578
27140 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27141 in call to operand_equal_p.
27143 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27146 * config/i386/cpuid.h: Fix another undefined behavior.
27148 2017-02-23 Richard Biener <rguenther@suse.de>
27150 PR tree-optimization/79683
27151 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27152 vector types for data-refs.
27154 2017-02-23 Martin Liska <mliska@suse.cz>
27156 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27158 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27160 PR middle-end/79665
27161 * internal-fn.c (get_range_pos_neg): Moved to ...
27162 * tree.c (get_range_pos_neg): ... here. No longer static.
27163 * tree.h (get_range_pos_neg): New prototype.
27164 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27165 are known to be in between 0 and signed maximum inclusive, try to
27166 expand both unsigned and signed divmod and use the cheaper one from
27169 2017-02-22 Jeff Law <law@redhat.com>
27171 PR tree-optimization/79578
27172 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27173 to compare base operands.
27175 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27178 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27179 gpc_reg_operand instead of fpr_reg_operand.
27181 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27183 * config/mips/mips.c (mips_return_in_memory): Force FP
27184 vector types to be returned in memory for o32 ABI.
27186 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27188 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27189 instead of DW_TAG_member for static data member declarations and don't
27190 set no_linkage_name for static inline data members.
27191 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27194 2017-02-22 Martin Liska <mliska@suse.cz>
27196 * doc/invoke.texi: Replace inequality signs with square brackets
27199 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27201 PR tree-optimization/68644
27202 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27204 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27207 * lra-constraints.c (simplify_operand_subreg): Handle
27208 WORD_REGISTER_OPERATIONS targets.
27210 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27213 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27214 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27215 elimination by swapping fld*.
27217 2017-02-22 Richard Biener <rguenther@suse.de>
27219 PR tree-optimization/79673
27220 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27221 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27222 irrelevant address-space qualifiers and avoiding a
27223 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27225 2017-02-22 Richard Biener <rguenther@suse.de>
27227 PR tree-optimization/79666
27228 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27229 to not symbolically negate if that may introduce undefined
27232 2017-02-22 Martin Liska <mliska@suse.cz>
27235 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27236 * data-streamer-out.c (streamer_write_gcov_count_stream):
27238 * value-prof.c (stream_out_histogram_value): Make assert more
27239 precise based on type of counter.
27241 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27244 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27245 Use nonimmediate_operand instead of memory_operand for operand 1.
27246 (float-extend standard_x87sse_constant_load splitter): Ditto.
27248 2017-02-21 Jeff Law <law@redhat.com>
27250 PR tree-optimization/79621
27251 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27252 blocks with edges to themselves.
27254 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27257 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27258 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27259 Use gimple_call_builtin_p.
27262 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27263 on temporarily removed DEBUG_INSNs.
27265 PR tree-optimization/79649
27266 * tree-loop-distribution.c (classify_partition): Give up on
27267 non-generic address space loads/stores.
27269 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27271 * doc/loop.texi (Loop manipulation): Remove nonexistent
27272 tree_ssa_loop_version from the documentation.
27273 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27275 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27278 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27279 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27280 * config/rs6000/rs6000.c: Include except.h.
27281 (rs6000_expand_split_stack_prologue): Call
27282 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27284 2017-02-21 Martin Jambor <mjambor@suse.cz>
27287 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27288 have been analyzed.
27290 2017-02-21 Martin Jambor <mjambor@suse.cz>
27292 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27293 for backward compatibility only.
27294 * doc/invoke.texi (Option Summary): Remove all references to
27295 -fipa-cp-alignment.
27297 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27301 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27303 * lra-constraints.c (curr_insn_transform): Handle
27304 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27306 2017-02-21 Martin Liska <mliska@suse.cz>
27308 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27310 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27312 PR translation/79638
27313 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27315 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27318 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27319 (arm_function_ok_for_sibcall): Return false for an indirect call by
27320 descriptor if all the argument registers are used.
27321 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27322 alignment of the function.
27324 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27326 PR tree-optimization/61441
27327 * simplify-rtx.c (simplify_const_unary_operation): For
27328 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27329 the sNaN unmodified.
27331 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27333 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27334 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27335 instead of SYSTEM_HEADER_DIR.
27337 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27338 Martin Liška <mliska@suse.cz>
27340 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27341 Fix typos and grammar, use active voice, and clarify.
27343 2017-02-20 Marek Polacek <polacek@redhat.com>
27345 PR middle-end/79537
27346 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27349 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27351 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27354 * config/i386/i386.c (ix86_expand_builtin): Handle
27355 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27356 ix86_builtins_isa[fcode].isa as a requirement of those
27357 flags and any other flag in the bitmask.
27358 (ix86_init_mmx_sse_builtins): Use 0 instead of
27359 ~OPTION_MASK_ISA_64BIT as mask.
27360 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27361 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27362 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27363 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27365 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27368 * lra-constraints.c (split_reg): Check requested split mode
27369 is supported by the register.
27371 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27373 * lra-constraints.c (simplify_operand_subreg): Remove early
27376 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27379 * lra-constraints.c (curr_insn_transform): Tighten condition
27380 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27382 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27385 * lra-constraints.c (curr_insn_transform): Handle
27386 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27388 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27391 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27393 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27395 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27398 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27401 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27403 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27406 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27408 * final.c (last_columnnum, override_columnnum): New variables.
27409 (final_start_function): Set last_columnnum, pass it to begin_prologue
27410 hook and pass 0 to dwarf2out_begin_prologue.
27411 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27412 to source_line debug hook.
27413 (notice_source_line): Compute last_columnnum and for debug_column_info
27414 return true on column changes.
27415 * debug.h (struct gcc_debug_hooks): Add column argument to
27416 source_line and begin_prologue hooks.
27417 (debug_nothing_int_charstar_int_bool): Remove prototype.
27418 (debug_nothing_int_int_charstar,
27419 debug_nothing_int_int_charstar_int_bool): New prototypes.
27420 (dwarf2out_begin_prologue): Add column argument.
27421 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27422 begin_prologue hooks.
27423 (debug_nothing_int_charstar_int_bool): Remove.
27424 (debug_nothing_int_int_charstar,
27425 debug_nothing_int_int_charstar_int_bool): New functions.
27426 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27427 through to dwarf2out_source_line.
27428 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27429 (dwarf2out_source_line): Add column argument, emit it if requested.
27430 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27432 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27433 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27434 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27435 through to dwarf2out_begin_prologue.
27436 (vmsdbgout_source_line): Add column argument, pass it through to
27437 dwarf2out_source_line.
27438 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27439 dbxout_source_line caller.
27440 (dbxout_source_line): Add column argument.
27442 * common.opt (gno-column-info, gcolumn-info): New options.
27443 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27444 (check_die): Also test for multiple DW_AT_decl_column attributes.
27445 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27446 DW_AT_decl_column if requested.
27447 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27449 (gen_variable_die): Likewise.
27450 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27451 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27454 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27455 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27456 (ix86_handle_option): Handle OPT_m3dnowa.
27457 * doc/invoke.texi (-m3dnowa): Document.
27458 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27459 -m3dnowa instead of -m3dnow -march=athlon.
27462 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27463 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27465 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27468 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27469 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27470 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27471 generator for vsx_xxpermdi_<mode>_be.
27472 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27473 force big-endian semantics.
27474 (vsx_xxpermdi_<mode>_be): New define_expand with same
27475 implementation as previous version of vsx_xxpermdi_<mode>.
27477 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27479 PR tree-optimization/79327
27480 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27481 variable, its initialization and use.
27483 2017-02-17 Julia Koval <julia.koval@intel.com>
27485 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27486 (OPTION_MASK_ISA_PKU_UNSET): New.
27487 (ix86_handle_option): Handle -mrdpid.
27488 * config/i386/cpuid.h (bit_RDPID): New.
27489 * config/i386/driver-i386.c (host_detect_local_cpu):
27490 Detect RDPID feature.
27491 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27492 * config/i386/i386-c.c (ix86_target_macros_internal):
27494 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27495 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27496 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27497 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27498 * config/i386/i386.md (define_insn "rdpid"): New.
27499 * config/i386/i386.opt Add -mrdpid.
27500 * config/i386/immintrin.h (_rdpid_u32): New.
27502 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27504 PR rtl-optimization/79541
27505 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27506 instead of transforming it into USE.
27508 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27510 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27511 If HONOR_SNANS (SFmode) force the input to a register.
27512 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27513 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27514 an frsp or similar insn.
27516 2017-02-17 Martin Liska <mliska@suse.cz>
27518 PR rtl-optimization/79577
27519 * params.def (selsched-max-sched-times): Increase minimum to 1.
27521 2017-02-17 Martin Liska <mliska@suse.cz>
27523 PR rtl-optimization/79574
27524 * gcse.c (want_to_gcse_p): Prevent integer overflow.
27526 2017-02-17 Martin Liska <mliska@suse.cz>
27528 PR tree-optimization/79529
27529 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
27530 ssa_defined_default_def_p to handle cases which are implicitly
27532 * tree-ssa.c (ssa_defined_default_def_p): New function.
27533 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
27534 which are implicitly defined.
27535 * tree-ssa.h (ssa_defined_default_def_p): Declare.
27537 2017-02-17 Richard Biener <rguenther@suse.de>
27539 PR middle-end/79576
27540 * params.def (max-ssa-name-query-depth): Limit to 10.
27542 2017-02-17 Richard Biener <rguenther@suse.de>
27544 PR tree-optimization/79552
27545 * tree-ssa-structalias.c (visit_loadstore): Properly verify
27548 2017-02-17 Richard Biener <rguenther@suse.de>
27551 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
27553 2017-02-17 Marek Polacek <polacek@redhat.com>
27555 PR middle-end/79536
27556 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
27557 (fold_negate_expr): New wrapper.
27559 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
27561 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
27562 Correct terminology and de-emphasize pre-standard behavior.
27564 2017-02-16 Alan Modra <amodra@gmail.com>
27566 PR rtl-optimization/79286
27567 * ira.c (def_dominates_uses): New function.
27568 (update_equiv_regs): Don't create an equivalence for insns that
27569 may trap where the register def does not dominate the use.
27571 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
27573 PR rtl-optimization/78127
27574 * lra.c (lra): Call lra_eliminate before finish the loop after
27577 2017-02-16 Richard Biener <rguenther@suse.de>
27579 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
27581 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
27582 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
27583 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
27584 (isl_val_int_from_wi): New function.
27585 (extract_affine_gmp): Rename to ...
27586 (extract_affine_wi): ... this, take a widest_int.
27587 (extract_affine_int): Just wrap extract_affine_wi.
27588 (add_param_constraints): Use isl_val_int_from_wi.
27589 (add_loop_constraints): Likewise, and extract_affine_wi.
27591 2017-02-15 Jeff Law <law@redhat.com>
27593 PR middle-end/79521
27594 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
27595 ira_init_register_move_cost_if_necessary.
27597 2017-02-15 Martin Sebor <msebor@redhat.com>
27599 PR middle-end/32003
27600 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
27601 removed in a prior commit.
27603 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27605 PR tree-optimization/79347
27606 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27607 counters during peeling.
27609 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27611 * Makefile.in (site.exp): Remove "set ISLVER".
27613 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27616 * real.c (real_from_integer): Call real_convert even for decimal.
27618 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27621 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27623 2017-02-14 Andrew Pinski <apinski@cavium.com>
27625 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27626 cores and change the partno/implementer to be correct.
27627 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27628 the 'B" as the implementer.
27629 * config/aarch64/aarch64-tune.md: Regenerate.
27631 2017-02-14 Carl Love <cel@us.ibm.com>
27633 * config/rs6000/rs6000.c: Add case statement entry to make the
27634 xvcvuxdsp built-in argument unsigned.
27635 * config/rs6000/vsx.md: Fix the source and return operand types so they
27636 match the instruction definitions from the ISA document. Fix typo
27637 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27640 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27643 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27644 member early_clobber_alts.
27645 * lra-lives.c (reg_early_clobber_p): New.
27646 (process_bb_lives): Use it.
27647 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27648 (debug_operand_data): Initialize early_clobber_alts.
27649 (setup_operand_alternative): Set up early_clobber_alts.
27650 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27651 alternatives to new_insn_reg.
27652 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27654 (lra_update_insn_regno_info): Pass the new arg.
27656 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27658 PR middle-end/79505
27659 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27660 (new_oacc_loop_raw): Don't clear already cleared fields.
27663 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27664 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27665 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27667 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27670 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27672 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27675 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27676 the extra instruction to the right place to store 128-bit constant
27679 2017-02-14 Martin Sebor <msebor@redhat.com>
27681 PR middle-end/79448
27682 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27683 warning for strings of unknown length.
27685 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27687 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27689 2017-02-14 Jeff Law <law@redhat.com>
27692 * ira-costs.c (scan_one_insn): Initialize register move costs
27693 for pseudos seen in USE/CLOBBER insns.
27695 PR tree-optimization/79095
27696 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27697 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27698 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27699 if the operands are known to be not equal, then the resulting range
27701 (intersect_ranges): If the new range is ~[0,0] and the old range is
27702 wide, then prefer ~[0,0].
27703 * tree-vrp.c (overflow_comparison_p_1): New function.
27704 (overflow_comparison_p): New function.
27705 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27706 if NAME is used in an overflow test.
27707 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27708 overflow check that can be expressed as an equality test, then adjust
27709 ops to be that equality test.
27711 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27713 * config/s390/s390-builtin-types.def: Remove flags argument.
27714 * config/s390/s390.c (s390_init_builtins): Likewise.
27716 2017-02-14 Martin Liska <mliska@suse.cz>
27718 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27719 vector. Fix trailing white spaces.
27721 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27723 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27726 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27728 PR rtl-optimization/68664
27729 * config/arm/arm.c (arm_sched_can_speculate_insn):
27730 New function. Declare prototype.
27731 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27733 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27735 PR rtl-optimization/68664
27736 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27738 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27740 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27742 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27743 max skip bytes for function, loop and jump.
27745 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27747 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27748 ABS_EXPR for gimple dump.
27750 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27753 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27755 PR tree-optimization/79408
27756 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27757 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27758 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27759 also if rhs1 is INTEGER_CST.
27761 2017-02-14 Richard Biener <rguenther@suse.de>
27763 PR middle-end/79432
27764 * tree-into-ssa.c (insert_phi_nodes): When the function can
27765 have abnormal edges rewrite SSA names with broken use-def
27766 dominance out of SSA and register them for PHI insertion.
27768 2017-02-13 Martin Sebor <msebor@redhat.com>
27770 PR middle-end/79496
27771 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27772 clearing info.nowrite flag when snprintf size argument is a range.
27774 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27776 * cprop.c (cprop_jump): Add missing space in string literal.
27777 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27778 (get_constraint_for_component_ref): Likewise.
27779 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27780 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27781 * lra-constraints.c (process_alt_operands): Likewise.
27782 * ipa-inline.c (inline_small_functions): Likewise.
27783 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27784 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27785 * trans-mem.c (diagnose_tm_1_op): Likewise.
27786 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27787 (grid_parallel_clauses_gridifiable): Likewise.
27789 * config/nvptx/mkoffload.c (process): Add space in between
27792 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27793 "MOD4_SSE_REGS" and "ALL_REGS".
27795 * spellcheck.c (test_data): Add , in between "foo" and "food".
27797 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27800 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27801 boundary crossing check and subsequent code generation agree.
27803 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27805 * config/aarch64/aarch64.c (has_memory_op): Delete.
27806 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27809 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27811 PR rtl-optimization/79388
27812 PR rtl-optimization/79450
27813 * combine.c (distribute_notes): When removing TEM_INSN for which
27814 corresponding dest has last value recorded, invalidate that last
27817 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27819 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27820 of explicit '@'. Add missing assembly comment marker on branch costs
27823 2017-02-13 Nathan Sidwell <nathan@acm.org>
27825 * gengtype-lex.l (<in_struct>): Add '/'.
27827 2017-02-13 Martin Liska <mliska@suse.cz>
27830 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27832 2017-02-13 Richard Biener <rguenther@suse.de>
27834 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27836 * configure: Re-generate.
27837 * config.in: Likewise.
27838 * graphite-dependences.c: Simplify as if
27839 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27840 * graphite-isl-ast-to-gimple.c: Likewise.
27841 * graphite-optimize-isl.c: Likewise.
27842 * graphite-poly.c: Likewise.
27843 * graphite-sese-to-poly.c: Likewise.
27844 * graphite.h: Likewise.
27845 * toplev.c: Include isl/version.h and use isl_version () for
27846 printing the ISL version.
27847 * doc/install.texi: Update ISL requirement.
27849 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27851 * doc/standards.texi (Standards): Update reference to
27854 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27856 * doc/extend.texi (Named Address Spaces): sourceware.org now
27858 * doc/install.texi (Binaries): Ditto.
27861 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27863 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27864 standard terminology "stringize"/"stringizing" throughout.
27865 * doc/cppinternals.texi: Likewise.
27867 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27869 * doc/extend.texi: Fix some spelling mistakes and typos.
27870 * doc/invoke.texi: Likewise.
27872 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27875 * params.def (inline-min-speedup) Change from 10 to 8.
27877 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27879 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27882 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27885 * ipa-inline-analysis.c (get_minimal_bb): New function.
27886 (record_modified): Use it.
27887 (remap_edge_change_prob): Handle also ancestor functions.
27889 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27891 * doc/contrib.texi (Contributors): Remove broken link into
27892 the Mauve CVS repository.
27894 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27896 PR middle-end/79454
27897 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27898 result computation whenever lhs doesn't have vector mode, not
27899 just when it has BLKmode.
27901 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27903 * doc/makefile.texi (profiledbootstrap): Refer to the
27904 installation instructions only in textual form.
27906 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27909 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27911 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27913 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27914 (Specific): Update mingw-w64 reference.
27916 (Specific): Remove broken link to Renesas RX processor.
27918 2017-02-10 Richard Biener <rguenther@suse.de>
27920 * toplev.c (process_options): Do not mention obsolete graphite
27921 options when printing sorry message about missing graphite support.
27922 Mention -floop-nest-optimize.
27924 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27926 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27927 (vtst_p16): Likewise.
27928 (vtstq_p8): Likewise.
27929 (vtstq_p16): Likewise.
27931 (vtstq_p64): Likewise.
27932 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27933 (vset_lane_p64): New.
27934 (vsetq_lane_p64): New.
27936 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27938 PR tree-optimization/79411
27939 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27940 stmt operands are SSA_NAMEs used in abnormal phis.
27941 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27944 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27947 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27950 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27952 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27954 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27956 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27960 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27961 not arbitrary TREE_CONSTANT.
27964 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27965 "omp declare target link" attribute unless is_global_var.
27966 * omp-offload.c (find_link_var_op): Likewise.
27968 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27969 Chung-Lin Tang <cltang@codesourcery.com>
27971 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27973 (gimplify_adjust_omp_clauses): Don't delete TILE.
27974 (gimplify_omp_for): Deal with TILE.
27975 * internal-fn.c (expand_GOACC_TILE): New function.
27976 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27978 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27979 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27981 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27982 avoid DIV for outermost collapse var.
27983 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27984 Remove out of date comments, fix whitespace.
27985 * omp-general.c (omp_extract_for_data): Deal with tiling.
27986 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27987 adjust OLF_DIM_BASE value.
27988 (struct omp_for_data): Add tiling field.
27989 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27990 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27991 for auto loops. Remove default auto determining, moved to
27992 oacc_loop_fixed_partitions.
27993 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27994 stmts, add e_mask field.
27995 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27996 (oacc_thread_numbers): Use oacc_dim_call.
27997 (oacc_xform_tile): New.
27998 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27999 (finish_oacc_loop): Adjust for ifns vector.
28000 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28001 add case for GOACC_TILE fns.
28002 (oacc_loop_xform_loop): Delete.
28003 (oacc_loop_process): Iterate over call list directly, and add
28004 handling for GOACC_TILE fns.
28005 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28007 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28008 vector partitioning to outer loops. Assign 2 partitions to loops
28009 when available. Add TILE handling.
28010 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28011 (execite_oacc_device_lower): Process GOACC_TILE fns,
28012 ignore unknown specs.
28013 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28014 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28015 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28017 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28019 * configure.ac (ACX_BUGURL): Update.
28020 * configure: Regenerate.
28022 2017-02-09 Richard Biener <rguenther@suse.de>
28024 PR tree-optimization/69823
28025 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28026 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28028 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28030 * config/arc/arc-c.def: Add __NPS400__ definition.
28031 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28032 (TARGET_NPS400): Define.
28034 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28036 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28038 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28039 pointer, arch_info.
28040 (arc_cpu_types): Fill the arch_info field with a pointer into the
28041 arc_arch_types table.
28042 (arc_selected_cpu): Declare.
28043 * config/arc/arc.c (arc_selected_cpu): Make global.
28044 (arc_selected_arch): Delete.
28045 (arc_base_cpu): Delete.
28046 (arc_override_options): Remove references to deleted variables,
28047 update access to arch information.
28048 (ARC_OPT): Update access to arch information.
28049 (ARC_OPTX): Likewise.
28050 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28051 (TARGET_ARC600): Update access to arch information.
28052 (TARGET_ARC601): Likewise.
28053 (TARGET_ARC700): Likewise.
28054 (TARGET_EM): Likewise.
28055 (TARGET_HS): Likewise.
28056 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28059 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28062 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28063 condition/operands for integer GE/LE/GEU/LEU operations.
28065 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28067 PR translation/79397
28068 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28071 2017-02-08 Martin Jambor <mjambor@suse.cz>
28074 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28075 whether allocation happened.
28076 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28077 nothing was allocated.
28079 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28081 PR tree-optimization/79408
28082 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28083 constant, but SSA_NAME with a known integer range, use the minimum
28084 of that range instead of op1 to determine if modulo can be replaced
28085 with its first operand.
28087 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28089 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28091 2017-02-08 Richard Biener <rguenther@suse.de>
28093 PR tree-optimization/71824
28094 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28095 Check all loops contained in the merged region.
28097 2017-02-07 Andrew Pinski <apinski@cavium.com>
28099 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28101 2017-02-07 Andrew Pinski <apinski@cavium.com>
28103 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28104 (thunderxt88): Likewise.
28105 (thunderxt81): Disable LSE and change v8.1 to v8.
28106 (thunderxt83): Likewise.
28108 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28109 Richard Biener <rguenther@suse.de>
28111 PR middle-end/79399
28112 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28113 type from int to size_t.
28114 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28116 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28118 PR rtl-optimization/79386
28119 * cprop.c (bypass_conditional_jumps): Initialize
28120 bypass_last_basic_block already before splitting bbs after
28121 unconditional traps...
28122 (bypass_conditional_jumps): ... rather than here.
28125 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28126 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28127 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28128 fix -masm=intel patterns.
28130 2017-02-07 Richard Biener <rguenther@suse.de>
28132 PR tree-optimization/79256
28133 PR middle-end/79278
28134 * builtins.c (get_object_alignment_2): Use min_align_of_type
28135 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28136 and ADJUST_FIELD_ALIGN.
28138 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28140 * doc/tm.texi: Regenerate.
28141 * stor-layout.c (layout_decl): Adjust.
28142 (update_alignment_for_field): Likewise.
28143 (place_field): Likewise.
28144 (min_align_of_type): Likewise.
28145 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28146 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28147 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28148 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28149 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28150 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28151 * config/i386/i386.c (x86_field_alignment): Likewise.
28152 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28153 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28154 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28155 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28156 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28157 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28161 2017-01-30 Richard Biener <rguenther@suse.de>
28163 PR tree-optimization/79256
28164 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28165 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28168 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28170 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28171 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28172 builtins to SImode and emit a zero-extend, if necessary.
28174 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28176 * docs/invoke.texi (RISC-V Options): Alphabetize.
28178 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28180 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28183 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28185 * config/riscv/riscv.c: New file.
28186 * common/config/riscv/riscv-common.c: Likewise.
28187 * config.gcc: Likewise.
28188 * config/riscv/constraints.md: Likewise.
28189 * config/riscv/elf.h: Likewise.
28190 * config/riscv/generic.md: Likewise.
28191 * config/riscv/linux.h: Likewise.
28192 * config/riscv/multilib-generator: Likewise.
28193 * config/riscv/peephole.md: Likewise.
28194 * config/riscv/pic.md: Likewise.
28195 * config/riscv/predicates.md: Likewise.
28196 * config/riscv/riscv-builtins.c: Likewise.
28197 * config/riscv/riscv-c.c: Likewise.
28198 * config/riscv/riscv-ftypes.def: Likewise.
28199 * config/riscv/riscv-modes.def: Likewise.
28200 * config/riscv/riscv-opts.h: Likewise.
28201 * config/riscv/riscv-protos.h: Likewise.
28202 * config/riscv/riscv.h: Likewise.
28203 * config/riscv/riscv.md: Likewise.
28204 * config/riscv/riscv.opt: Likewise.
28205 * config/riscv/sync.md: Likewise.
28206 * config/riscv/t-elf-multilib: Likewise.
28207 * config/riscv/t-linux: Likewise.
28208 * config/riscv/t-linux-multilib: Likewise.
28209 * config/riscv/t-riscv: Likewise.
28210 * configure.ac: Likewise.
28211 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28212 Waterman as RISC-V maintainers.
28213 * doc/install.texi: Add RISC-V entries.
28214 * doc/invoke.texi: Add RISC-V options section.
28215 * doc/md.texi: Add RISC-V constraints section.
28216 * configure: Regenerated.
28218 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28221 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28222 false values to be constant vectors with all 0 or all 1 bits set.
28223 (vcondu<mode><mode>): Likewise.
28224 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28226 (fpmask_comparison_operator): Update comment.
28227 (vecint_comparison_operator): New predicate.
28228 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28229 vector conditionals when the true and false values are constant
28230 vectors with all 0 bits or all 1 bits set.
28232 2017-02-06 Martin Sebor <msebor@redhat.com>
28234 PR tree-optimization/79376
28235 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28237 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28239 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28240 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28241 to simplify split condition.
28243 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28245 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28246 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28249 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28251 PR rtl-optimization/68664
28252 * target.def (can_speculate_insn): New hook.
28253 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28254 * doc/tm.texi: Regenerate.
28255 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28256 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28257 (rs6000_sched_can_speculate_insn): New function.
28259 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28261 PR tree-optimization/79284
28262 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28263 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28264 vectorizable_mask_load_store, vectorizable_operation,
28265 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28266 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28267 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28268 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28269 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28270 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28271 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28272 is_gimple_assign (stmt). Replace another such test with
28273 is_gimple_assign (stmt).
28275 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28278 * config/avr/avr.c (rtl-iter.h): Include it.
28279 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28280 (avr_legitimate_combined_insn): ...and implementation.
28282 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28284 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28285 * config/s390/s390.c (s390_const_operand_ok)
28286 (s390_canonicalize_comparison, s390_extract_part)
28287 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28288 (s390_contiguous_bitmask_p, s390_rtx_costs)
28289 (legitimize_pic_address): Likewise.
28290 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28291 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28292 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28293 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28294 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28296 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28298 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28299 REGNO($0) == REGNO($1).
28301 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28303 * config/s390/linux.h(SIZE_TYPE): Add comment.
28305 2017-02-06 Julian Brown <julian@codesourcery.com>
28306 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28307 Virendra Pathak <virendra.pathak@broadcom.com>
28309 * config/aarch64/aarch64-cores.def: Change the scheduler
28311 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28312 * config/aarch64/thunderx2t99.md: New file.
28314 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28316 * doc/standards.texi (Go Language): Update link to language
28319 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28321 * tree-eh.c (lower_resx): Sanitize profile.
28322 (cleanup_empty_eh_move_lp): Likewise.
28324 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28327 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28329 * cfgloopmanip.h (loop_version): Update prototype.
28330 * modulo-sched.c (sms_schedule): Update call of loop_version.
28331 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28332 * tree-parloops.c (gen_parallel_loop): Likewise.
28333 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28334 * tree-ssa-loop-split.c (split_loop): Likewise.
28335 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28336 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28338 2017-02-05 Martin Liska <mliska@suse.cz>
28341 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28343 (print_operand_address): Initialize a struct to zero.
28345 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28347 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28348 garbage collector only in textual form.
28350 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28352 * doc/extend.texi (x86 specific memory model extensions for
28353 transactional memory): Simplify a phrase.
28355 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28358 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28359 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28360 (atomic_storedi_1): Likewise.
28362 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28364 PR tree-optimization/79338
28365 * tree-parloops.c (gather_scalar_reductions): Don't call
28366 vect_analyze_loop_form for loop->inner before destroying loop's
28369 2017-02-03 Martin Sebor <msebor@redhat.com>
28371 PR tree-optimization/79327
28372 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28373 when precision has resulted in leading zeros.
28374 (format_integer): Adjust the likely counter to assume an unknown
28375 argument that may be zero is non-zero.
28377 2017-02-03 Jason Merrill <jason@redhat.com>
28380 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28381 avoid copying non-taken branch.
28383 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28385 PR tree-optimization/79340
28386 * tree-vect-loop.c (vectorizable_reduction): Release
28387 vec_defs elements after safe_splicing them into other vectors.
28390 PR tree-optimization/79327
28391 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28392 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28394 (format_integer): Use wide_int_to_tree instead of build_int_cst
28395 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28396 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28397 of shortest and longest sequence.
28399 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28401 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28402 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28404 2017-02-03 Walter Lee <walt@tilera.com>
28407 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28408 after initial stackframe link reg save.
28409 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28411 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28414 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28415 wu for stxssp alternative.
28417 2017-02-03 Martin Sebor <msebor@redhat.com>
28419 PR tree-optimization/79352
28420 * gimple-fold.c (get_range_strlen): Add argument.
28421 (get_range_strlen): Change return type to bool.
28422 (get_maxval_strlen): Pass in a dummy argument.
28423 * gimple-fold.h (get_range_strlen): Change return type to bool.
28424 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28425 * tree.h (array_at_struct_end_p): Add argument.
28426 * tree.c (array_at_struct_end_p): Handle it.
28428 2017-02-03 Martin Liska <mliska@suse.cz>
28431 * multiple_target.c (create_dispatcher_calls): Redirect edge
28432 from a caller of a dispatcher.
28433 (expand_target_clones): Make the clones local.
28434 (ipa_target_clone): Do both target clones and resolvers.
28435 (ipa_dispatcher_calls): Remove the pass.
28436 (pass_dispatcher_calls::gate): Likewise.
28437 (make_pass_dispatcher_calls): Likewise.
28438 * passes.def (pass_target_clone): Put as very first IPA early
28441 2017-02-03 Martin Liska <mliska@suse.cz>
28443 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28444 in case of a function with ifunc attribute.
28446 2017-02-03 Martin Liska <mliska@suse.cz>
28448 * cgraph.c (cgraph_node::dump): Dump function version info.
28449 * symtab.c (symtab_node::dump_base): Add missing new line.
28451 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28453 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28454 (ifcombine_ifandif): Use it.
28456 2017-02-03 Martin Liska <mliska@suse.cz>
28458 * doc/invoke.texi: Document default value for
28459 use-after-scope-direct-emission-threshold.
28461 2017-02-03 Martin Liska <mliska@suse.cz>
28463 PR tree-optimization/79339
28464 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28465 (format_floating): Likewise.
28467 2017-02-03 Martin Liska <mliska@suse.cz>
28470 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28472 (ipa_node_params_t::remove): Likewise.
28473 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28474 initialization from removed ipa_node_params_t::insert.
28475 (ipa_node_params::~ipa_node_params): Move from removed
28476 ipa_node_params_t::release.
28477 * symbol-summary.h (symbol_summary::m_released): New member.
28478 Do not release a summary twice. Do not allow to call finalizer
28479 for types of a summary that live in GGC memory.
28481 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28483 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28486 2017-02-02 Martin Sebor <msebor@redhat.com>
28488 PR middle-end/79275
28489 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28490 (format_string): Tighten up the range of output for non-constant
28491 strings and correct the expected range for wide non-constant strings.
28493 2017-02-02 Martin Sebor <msebor@redhat.com>
28495 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28497 PR middle-end/32003
28498 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28500 (-fdump-tree-@var): Add to index and document how to come up
28501 with pass-specific option and dump file names.
28502 (-fdump-passes): Clarify where to look for output.
28504 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28506 PR middle-end/77445
28507 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28508 statistics of the analyzed path; allow threading for speed when
28509 any of BBs along the path are optimized for speed.
28511 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28513 PR middle-end/78468
28514 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28515 settings of the virtual registers.
28518 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
28520 * explow.c (get_dynamic_stack_size): Take known alignment of stack
28521 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
28524 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28526 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
28527 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
28529 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28531 * config/s390/s390.md: Add missing comments with the expanded
28533 * config/s390/vector.md: Likewise.
28534 * config/s390/vx-builtins.md: Likewise.
28536 2017-02-02 Jakub Jelinek <jakub@redhat.com>
28539 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
28540 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
28541 conditions on a single line.
28543 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28545 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28546 __S390_VX__ to __VX__.
28548 2017-02-01 Andrew Pinski <apinski@cavium.com>
28550 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
28551 stmt_info to record_stmt_cost.
28552 (vect_get_known_peeling_cost): Pass stmt_info if known to
28554 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
28555 cpu_vector_cost field into
28556 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
28557 field into vec_int_stmt_cost and vec_fp_stmt_cost.
28558 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
28559 splitting of scalar_stmt_cost and vec_stmt_cost.
28560 (thunderx_vector_cost): Likewise.
28561 (cortexa57_vector_cost): LIkewise.
28562 (exynosm1_vector_cost): Likewise.
28563 (xgene1_vector_cost): Likewise.
28564 (thunderx2t99_vector_cost): Improve after the splitting of the two
28566 (aarch64_builtin_vectorization_cost): Update for the splitting of
28567 scalar_stmt_cost and vec_stmt_cost.
28569 2017-02-01 Torvald Riegel <triegel@redhat.com>
28570 Richard Henderson <rth@redhat.com>
28572 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
28573 conditional on existance of a fast atomic load.
28574 * optabs-query.c (can_atomic_load_p): New function.
28575 * optabs-query.h (can_atomic_load_p): Declare it.
28576 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
28577 no fast atomic load is available for the particular size of access.
28578 (expand_atomic_compare_and_swap): Likewise.
28579 (expand_atomic_load): Likewise.
28580 (expand_atomic_store): Likewise.
28581 (expand_atomic_fetch_op): Likewise.
28582 * testsuite/lib/target-supports.exp
28583 (check_effective_target_sync_int_128): Remove x86 because it provides
28584 no fast atomic load.
28585 (check_effective_target_sync_int_128_runtime): Likewise.
28587 2017-02-01 Richard Biener <rguenther@suse.de>
28589 * graphite.c: Include tree-vectorizer.h for find_loop_location.
28590 (graphite_transform_loops): Provide opt-info for optimized nests.
28591 * tree-parloop.c (parallelize_loops): Provide opt-info for
28592 parallelized loops.
28594 2017-02-01 Richard Biener <rguenther@suse.de>
28596 PR middle-end/79315
28597 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
28598 was not set before.
28600 2017-02-01 Richard Biener <rguenther@suse.de>
28602 PR tree-optimization/71824
28603 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28604 Verify the loops are valid in the merged SESE region.
28605 (scop_detection::can_represent_loop_1): Check analyzing the
28606 evolution of the number of iterations in the region succeeds.
28608 2017-01-31 Ian Lance Taylor <iant@golang.org>
28610 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28611 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28613 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28615 PR preprocessor/79210
28616 * input.c (get_substring_ranges_for_loc): Replace line_width
28617 assertion with error-handling.
28619 2017-01-31 Richard Biener <rguenther@suse.de>
28621 PR tree-optimization/77318
28622 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28623 (create_pw_aff_from_tree): Take loop parameter.
28624 (add_condition_to_pbb): Pass loop of the condition to
28625 create_pw_aff_from_tree.
28627 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28629 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28630 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28632 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28636 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28638 (convert_int_to_float128): Likewise.
28639 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28640 (convert_int_to_float128): Likewise.
28641 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28642 (UNSPEC_IEEE128_CONVERT): Likewise.
28643 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28644 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28645 Use local variables for IBM extended format.
28646 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28647 (fix_trunc<mode>si2_fprs): Likewise.
28648 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28649 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28650 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28651 to know that we can now have integers of all sizes in vector
28653 (fix<uns>_<mode>di2_hw): Likewise.
28654 (float<uns>_<mode>si2_hw): Likewise.
28655 (fix_<mode>si2_hw): Likewise.
28656 (fixuns_<mode>si2_hw): Likewise.
28657 (float<uns>_<mode>di2_hw): Likewise.
28658 (float_<mode>di2_hw): Likewise.
28659 (float_<mode>si2_hw): Likewise.
28660 (floatuns_<mode>di2_hw): Likewise.
28661 (floatuns_<mode>si2_hw): Likewise.
28662 (xscvqp<su>wz_<mode>): Delete, no longer used.
28663 (xscvqp<su>dz_<mode>): Likewise.
28664 (xscv<su>dqp_<mode>): Likewise.
28665 (ieee128_mfvsrd_64bit): Likewise.
28666 (ieee128_mfvsrd_32bit): Likewise.
28667 (ieee128_mfvsrwz): Likewise.
28668 (ieee128_mtvsrw): Likewise.
28669 (ieee128_mtvsrd_64bit): Likewise.
28670 (ieee128_mtvsrd_32bit): Likewise.
28672 2017-01-31 Martin Liska <mliska@suse.cz>
28675 * ipa-prop.c (ipa_free_all_node_params): Call release method
28676 instead of ~sumbol_summary to not to trigger double times
28679 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28681 PR tree-optimization/71691
28682 * bitmap.h (class auto_bitmap): New.
28683 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28684 is_maybe_undefined instead of ssa_undefined_value_p.
28686 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28688 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28689 __S390_ARCH_LEVEL__ to __ARCH__.
28691 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28693 PR tree-optimization/79267
28694 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28695 if should_remove_lhs_p is true.
28697 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28700 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28701 (add_alignment_attribute): New.
28702 (base_type_die): Add alignment attribute.
28703 (subrange_type_die): Likewise.
28704 (modified_type_die): Likewise.
28705 (gen_array_type_die): Likewise.
28706 (gen_descr_array_type_die: Likewise.
28707 (gen_enumeration_type_die): Likewise.
28708 (gen_subprogram_die): Likewise.
28709 (gen_variable_die): Likewise.
28710 (gen_field_die): Likewise.
28711 (gen_ptr_to_mbr_type_die): Likewise.
28712 (gen_struct_or_union_type_die): Likewise.
28713 (gen_subroutine_type_die): Likewise.
28714 (gen_typedef_die): Likewise.
28715 (base_type_cmp): Compare alignment attribute.
28717 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28720 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28721 (setb_unsigned) New pattern for setb with CCUNS.
28722 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28723 subfc./subfe sequence to avoid overflow problems. Generate a
28724 shorter sequence with cmpld/setb for power9.
28725 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28726 for generating subfc. instruction.
28727 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28728 now uses this instruction.
28730 2017-01-30 Ian Lance Taylor <iant@google.com>
28733 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28734 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28736 2017-01-30 Martin Sebor <msebor@redhat.com>
28738 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28739 Move constant to the right of a relational operator.
28740 (get_mpfr_format_length, format_character, format_string): Ditto.
28741 (should_warn_p, maybe_warn): Same.
28743 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28745 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28748 * asan.c (get_translation_unit_decl): Remove function.
28749 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28751 2017-01-30 Martin Liska <mliska@suse.cz>
28753 PR gcov-profile/79259
28754 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28755 -fprofile-generate.
28757 2017-01-30 Martin Liska <mliska@suse.cz>
28760 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28761 Initialize variables with NULL value.
28763 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28766 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28768 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28770 2017-01-30 Richard Biener <rguenther@suse.de>
28772 PR tree-optimization/79276
28773 * tree-vrp.c (process_assert_insertions): Properly adjust common
28774 when removing a duplicate.
28776 * gcc.dg/torture/pr79276.c: New testcase.
28778 2017-01-30 Richard Biener <rguenther@suse.de>
28780 PR tree-optimization/79256
28781 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28782 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28784 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28786 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28789 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28790 ("*r<noxa>sbg_<mode>_sll_bitmask")
28791 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28792 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28793 Use contiguous_bitmask_nowrap_operand.
28795 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28798 * config/rs6000/altivec.h (vec_xl): Revise #define.
28799 (vec_xst): Likewise.
28801 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28803 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28805 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28807 PR rtl-optimization/79194
28808 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28809 traps before call to bypass_conditional_jumps.
28811 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28813 PR tree-optimization/71374
28814 * lra-constraints.c (check_conflict_input_operands): New.
28815 (match_reload): Use it.
28817 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28820 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28821 account to calculate conflict_set.
28823 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28825 PR rtl-optimization/78559
28826 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28827 other_insn in combine.
28829 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28831 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28832 uint16_type_node for BT_UINT16.
28834 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28836 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28837 "RTL Tests" to menu.
28838 (GIMPLE Tests): New node.
28839 (RTL Tests): New node.
28841 2017-01-27 Richard Biener <rguenther@suse.de>
28843 PR tree-optimization/79245
28844 * tree-loop-distribution.c (distribute_loop): Apply cost
28845 modeling also to detected patterns.
28847 2017-01-27 Richard Biener <rguenther@suse.de>
28849 PR tree-optimization/71433
28850 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28851 (compare_assert_loc): New function.
28852 (process_assert_insertions): Sort and optimize assert locations
28853 to remove duplicates and push down identical assertions on
28854 edges to their destination block.
28856 2017-01-27 Richard Biener <rguenther@suse.de>
28858 PR tree-optimization/79244
28859 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28860 out SSA names even if abnormal.
28862 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28864 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28865 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28866 instead of MPFR_RNDN.
28868 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28871 * arm.c (arm_option_override): Don't call build_target_option_node
28872 until after doing all option overrides.
28873 (arm_valid_target_attribute_tree): Likewise.
28875 2017-01-27 Martin Liska <mliska@suse.cz>
28877 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28878 for {cd}tors and C++ {cd}tors.
28880 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28882 * config/s390/s390.md ("*setmem_long_and")
28883 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28885 2017-01-26 Martin Sebor <msebor@redhat.com>
28887 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28890 2017-01-26 Martin Sebor <msebor@redhat.com>
28892 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28893 HAVE_DFmode before using XFmode or DFmode.
28894 (parse_directive): Avoid using the z length modifier to avoid
28895 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28897 PR middle-end/78703
28898 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28899 to accept adjustment as an array.
28900 (get_int_range): New function.
28901 (struct directive): Make width and prec arrays.
28902 (directive::set_width, directive::set_precision): Call get_int_range.
28903 (format_integer, format_floating): Handle width and precision ranges.
28904 (format_string, parse_directive): Same.
28906 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28909 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28910 comdat_type_p set, just clone them, but keep the children in the
28914 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28915 which have direct callers with -fvar-tracking-assignments enabled
28917 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28918 inside of type units.
28920 2017-01-26 Martin Sebor <msebor@redhat.com>
28922 PR middle-end/78703
28923 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28925 (struct format_result): Replace number_chars, number_chars_min,
28926 and number_chars_max with a single member of struct result_range.
28928 (format_result::operator+=): Adjust.
28929 (struct fmtresult): Remove bounded. Handle likely and unlikely
28931 (fmtresult::adjust_for_width_or_precision): New function.
28932 (fmtresult:type_max_digits): New function.
28933 (bytes_remaining): Handle likely and unlikely counters.
28934 (min_bytes_remaining): Remove.
28935 (format_percent): Simplify.
28936 (format_integer, format_floating): Set likely and unlikely counters.
28937 (get_string_length, format_character, format_string): Same.
28938 (format_plain, should_warn_p): New function.
28939 (maybe_warn): Call should_warn_p. Update diagnostic messages
28940 and handle those for all directives, including plain strings.
28941 (format_directive): Handle likely and unlikely counters.
28942 Remove unnecessary quoting from diagnostics. Add an informational
28944 (add_bytes): Remove.
28945 (pass_sprintf_length::compute_format_length): Simplify.
28946 (try_substitute_return_value): Handle likely and unlikely counters.
28948 2017-01-26 Carl Love <cel@us.ibm.com>
28950 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28951 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28953 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28956 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28957 endianess for subregs into account.
28958 * lra-constraints.c (lra_constraints): Do risky transformations
28959 always on the first iteration.
28960 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28961 last_call_used_reg_set.
28962 (process_bb_lives): Define and use last_call_used_reg_set.
28963 * lra.c (lra): Always continue after lra_constraints on the first
28966 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28968 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28970 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28972 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28974 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28975 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28976 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28977 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28978 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28979 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28980 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28981 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28982 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28984 2017-01-26 Marek Polacek <polacek@redhat.com>
28987 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28988 for the third operand.
28990 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28992 PR middle-end/79236
28993 * omp-low.c (struct omp_context): Add simt_stmt field.
28994 (scan_omp_for): Return omp_context *.
28995 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28996 context to the _simt_ SIMD stmt.
28997 (lower_omp_for): For combined SIMD with sibling _simt_
28998 SIMD, make sure to use the same decls in _looptemp_
28999 clauses as in the sibling.
29001 2017-01-26 David Sherwood <david.sherwood@arm.com>
29003 PR middle-end/79212
29004 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29007 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29010 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29011 emit fld b; fld a; if possible.
29013 * brig-builtins.def: Update copyright years.
29014 * config/arm/arm_acle_builtins.def: Update copyright years.
29016 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29019 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29020 constraint instead of o for the stxsd instruction.
29022 2017-01-25 Carl Love <cel@us.ibm.com>
29024 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29025 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29027 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29029 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29031 2017-01-25 Richard Biener <rguenther@suse.de>
29033 PR tree-optimization/69264
29034 * target.def (vector_alignment_reachable): Improve documentation.
29035 * doc/tm.texi: Regenerate.
29036 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29038 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29039 earlier changes with respect to TYPE_USER_ALIGN.
29040 (vector_alignment_reachable_p): Likewise. Improve dumping.
29042 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29045 * config/arm/arm.md (xordi3): Force constant operand into a register
29048 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29050 * doc/invoke.texi (-fstore-merging): Correct default optimization
29051 levels at which it is enabled.
29052 (-O): Move -fstore-merging from list to...
29055 2017-01-25 Richard Biener <rguenther@suse.de>
29058 * omp-expand.c: Include debug.h.
29059 (expand_omp_taskreg): Make sure to generate early debug before
29060 outlining anything from a function.
29061 (expand_omp_target): Likewise.
29062 (grid_expand_target_grid_body): Likewise.
29064 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29067 * asan.c (get_translation_unit_decl): New function.
29068 (asan_add_global): Extract modules file name from globals
29069 TRANSLATION_UNIT_DECL name.
29071 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29074 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29075 for long calls with APCS frame and VFP.
29077 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29079 * cfg.c (original_copy_tables_initialized_p): New function.
29080 * cfg.h (original_copy_tables_initialized_p): New decl.
29081 * cfgrtl.c (relink_block_chain): Guard the call to
29082 free_original_copy_tables with a call to
29083 original_copy_tables_initialized_p.
29084 * cgraph.h (symtab_node::native_rtl_p): New decl.
29085 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29086 (symtab_node::needed_p): Don't assert for early assembly output
29087 for __RTL functions.
29088 (cgraph_node::finalize_function): Set "force_output" for __RTL
29090 (cgraph_node::analyze): Bail out early for __RTL functions.
29091 (analyze_functions): Update assertion to support __RTL functions.
29092 (cgraph_node::expand): Bail out early for __RTL functions.
29093 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29095 * function.h (struct function): Update comment for field
29097 * gimple-expr.c: Include "tree-pass.h".
29098 (gimple_has_body_p): Return false for __RTL functions.
29099 * Makefile.in (OBJS): Add run-rtl-passes.o.
29100 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29102 (gcc::pass_manager::get_clean_slate): New accessor.
29103 * passes.c: Include "insn-addr.h".
29104 (should_skip_pass_p): Add logging. Update logic for running
29105 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29106 property-provider override so it is only done for gimple passes.
29108 (skip_pass): New function.
29109 (execute_one_pass): Call skip_pass when skipping passes.
29110 * read-md.c (md_reader::read_char): Support filtering
29111 the input to a subset of line numbers.
29112 (md_reader::md_reader): Initialize fields
29113 m_first_line and m_last_line.
29114 (md_reader::read_file_fragment): New function.
29115 * read-md.h (md_reader::read_file_fragment): New decl.
29116 (md_reader::m_first_line): New field.
29117 (md_reader::m_last_line): New field.
29118 * read-rtl-function.c (function_reader::create_function): Only
29119 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29120 curr_properties. Set DECL_INITIAL to a dummy block.
29121 (read_rtl_function_body_from_file_range): New function.
29122 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29124 * run-rtl-passes.c: New file.
29125 * run-rtl-passes.h: New file.
29127 2017-01-24 Jeff Law <law@redhat.com>
29129 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29132 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29134 PR tree-optimization/79159
29135 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29136 (record_nonwrapping_iv): Improve boundary using above function if no
29137 value range information.
29139 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29140 Martin Jambor <mjambor@suse.cz>
29142 * brig-builtins.def: New file.
29143 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29144 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29145 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29146 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29147 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29148 * builtin-types.def (BT_INT8): New.
29149 (BT_INT16): Likewise.
29150 (BT_UINT8): Likewise.
29151 (BT_UINT16): Likewise.
29152 (BT_FN_ULONG): Likewise.
29153 (BT_FN_UINT_INT): Likewise.
29154 (BT_FN_UINT_ULONG): Likewise.
29155 (BT_FN_UINT_LONG): Likewise.
29156 (BT_FN_UINT_PTR): Likewise.
29157 (BT_FN_ULONG_PTR): Likewise.
29158 (BT_FN_INT8_FLOAT): Likewise.
29159 (BT_FN_INT16_FLOAT): Likewise.
29160 (BT_FN_UINT32_FLOAT): Likewise.
29161 (BT_FN_UINT16_FLOAT): Likewise.
29162 (BT_FN_UINT8_FLOAT): Likewise.
29163 (BT_FN_UINT64_FLOAT): Likewise.
29164 (BT_FN_UINT16_UINT32): Likewise.
29165 (BT_FN_UINT32_UINT16): Likewise.
29166 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29167 (BT_FN_INT_PTR_INT): Likewise.
29168 (BT_FN_UINT_PTR_UINT): Likewise.
29169 (BT_FN_LONG_PTR_LONG): Likewise.
29170 (BT_FN_ULONG_PTR_ULONG): Likewise.
29171 (BT_FN_VOID_UINT64_UINT64): Likewise.
29172 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29173 (BT_FN_INT8_INT8_INT8): Likewise.
29174 (BT_FN_INT16_INT16_INT16): Likewise.
29175 (BT_FN_INT_INT_INT): Likewise.
29176 (BT_FN_UINT_FLOAT_UINT): Likewise.
29177 (BT_FN_FLOAT_UINT_UINT): Likewise.
29178 (BT_FN_ULONG_UINT_UINT): Likewise.
29179 (BT_FN_ULONG_UINT_PTR): Likewise.
29180 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29181 (BT_FN_UINT_UINT_UINT): Likewise.
29182 (BT_FN_VOID_UINT_PTR): Likewise.
29183 (BT_FN_UINT_UINT_PTR: Likewise.
29184 (BT_FN_UINT32_UINT64_PTR): Likewise.
29185 (BT_FN_INT_INT_UINT_UINT): Likewise.
29186 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29187 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29188 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29189 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29190 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29191 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29192 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29193 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29194 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29195 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29196 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29197 * doc/frontends.texi: List BRIG FE.
29198 * doc/install.texi (Testing): Add BRIG tesring requirements.
29199 * doc/invoke.texi (Overall Options): Mention BRIG.
29200 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29202 2017-01-24 Richard Biener <rguenther@suse.de>
29204 PR translation/79208
29205 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29207 2017-01-24 Martin Jambor <mjambor@suse.cz>
29210 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29211 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29212 and known_contexts.
29214 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29216 PR middle-end/79123
29217 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29218 casts from signed to unsigned really don't have a range.
29220 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29222 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29223 GMP_RNDx for compatiblity.
29225 2017-01-24 Martin Liska <mliska@suse.cz>
29228 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29229 that would prevent us to call alloca with -1 as argument.
29231 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29233 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29234 Avoid -Wformat-security warning.
29236 2017-01-23 Andrew Pinski <apinski@cavium.com>
29238 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29241 2017-01-23 Martin Sebor <msebor@redhat.com>
29243 PR middle-end/78703
29244 * gimple-ssa-sprintf.c (warn_level): New global.
29245 (format_integer): Use it here and throughout the rest of the file.
29246 Use the same switch to compute sign as base.
29247 (maybe_warn): New function.
29248 (format_directive): Factor out warnings into maybe_warn.
29249 Add debugging output. Use warn_level.
29250 (add_bytes): Use warn_level.
29251 (pass_sprintf_length::compute_format_length): Add debugging output.
29252 (try_substitute_return_value): Same.
29253 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29255 PR middle-end/78703
29256 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29257 (struct fmtresult, format_integer, format_floating): Adjust.
29258 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29259 (get_string_length, format_string,format_directive): Same.
29260 (pass_sprintf_length::compute_format_length): Same.
29261 (try_substitute_return_value): Simplify slightly.
29263 PR middle-end/78703
29264 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29265 (fmtresult::operator+=): Outlined.
29266 (struct fmtresult): Add ctors.
29267 (struct conversion_spec): Rename...
29268 (struct directive): ...to this. Add and remove data members.
29269 (directive::set_width, directive::set_precision): New functions.
29270 (format_percent): Use fmtresult ctor.
29271 (get_width_and_precision): Remove.
29272 (format_integer): Make naming changes. Avoid computing width and
29274 (format_floating): Same. Adjust indentation.
29275 (format_character, format_none): New functions.
29276 (format_string): Moved character handling to format_character.
29277 (format_directive): Remove arguments, change return type.
29278 (parse_directive): New function.
29279 (pass_sprintf_length::compute_format_length): Move directive
29280 parsing to parse_directive.
29282 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29284 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29285 (assign_assembler_name_if_needed): ... this.
29286 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29287 (assign_assembler_name_if_needed): ... this.
29288 (free_lang_data_in_cgraph): Adjust callers.
29289 * cgraphunit.c (cgraph_node::analyze): Likewise.
29290 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29293 2017-01-23 Richard Biener <rguenther@suse.de>
29295 PR tree-optimization/79088
29296 PR tree-optimization/79188
29297 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29298 resetting loop bounds after last path deletion. Reset loop
29299 bounds of the target loop, make code match the comments.
29300 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29301 Make sure loops need no fixups.
29303 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29305 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29306 exponent support with double type for first argument.
29307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29308 type returned by __builtin_vec_extract_sig,
29309 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29310 functions from "vector int" to "vector unsigned int" or from
29311 "vector long long int" to "vector unsigned long long int".
29312 Changed type returned by __builtin_vec_extract_exp,
29313 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29314 functions from "vector int" to "vector unsigned int" or from
29315 "vector long long int" to "vector unsigned long long int".
29316 Changed return type of __builtin_vec_test_data_class,
29317 __builtin_vec_test_data_class_sp, and
29318 __builtin_vec_test_data_class_dp from "vector int" to
29319 "vector bool int" or from "vector long long int" to "vector bool
29320 long long int" and changed second argument type from "unsigned
29321 int" to "int". Added new overloaded function forms "vector float
29322 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29323 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29324 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29325 double, vector unsigned long long int)" and "vector double
29326 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29327 long int)". Changed return type of
29328 __builtin_scalar_test_data_class and
29329 __builtin_scalar_test_data_class_sp and
29330 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29331 int" and changed second argument from "unsigned int" to "int".
29332 Changed type returned by __builtin_scalar_test_neg,
29333 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29334 from "int" to "bool int". Added new overloaded function form
29335 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29336 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29337 exponent double-precision with floating point first argument.
29338 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29339 documentation of scalar_test_data_class, scalar_test_neg,
29340 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29341 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29342 vec_test_data_class built-in functions to reflect refinements in
29343 their type signatures.
29345 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29347 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29349 (aarch64_elf_asm_destructor): Likewise.
29351 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29353 PR rtl-optimization/78634
29354 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29355 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29356 * ifcvt.c (noce_try_cmove): Add missing cost check.
29358 PR rtl-optimization/71724
29359 * combine.c (if_then_else_cond): Look for situations where it is
29360 beneficial to undo the work of one of the recursive calls.
29362 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29364 PR tree-optimization/70754
29365 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29366 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29367 combined stmt before it if not NULL.
29368 (combine_chains): Process refs reversely and compute dominance point
29371 2017-01-23 Martin Liska <mliska@suse.cz>
29373 PR tree-optimization/79196
29374 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29375 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29377 (strlen_optimize_stmt): Call the renamed function.
29379 2017-01-23 Michael Matz <matz@suse.de>
29381 PR tree-optimization/78384
29382 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29384 2017-01-23 Richard Biener <rguenther@suse.de>
29386 PR tree-optimization/79186
29387 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29388 both incoming edges before moving an assert.
29390 2017-01-23 Martin Jambor <mjambor@suse.cz>
29392 * ipa-prop.c (load_from_param_1): Removed.
29393 (load_from_unmodified_param): Bits from load_from_param_1 put back
29395 (load_from_param): Removed.
29396 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29397 with stmt. Reverted back to use of load_from_unmodified_param.
29399 2017-01-23 Martin Jambor <mjambor@suse.cz>
29402 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29403 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29404 field a pointer to garbage collected vector, mark lattices and
29405 ipcp_orig_node with GTY((skip)).
29406 (ipa_get_param_count): Adjust to descriptors being a pointer.
29407 (ipa_get_param): Likewise.
29408 (ipa_get_type): Likewise.
29409 (ipa_get_param_move_cost): Likewise.
29410 (ipa_set_param_used): Likewise.
29411 (ipa_get_controlled_uses): Likewise.
29412 (ipa_set_controlled_uses): Likewise.
29413 (ipa_is_param_used): Likewise.
29414 (ipa_node_params_t): Move into garbage collector. New methods insert
29416 (ipa_node_params_sum): Annotate wth GTY(()).
29417 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29419 (ipa_load_from_parm_agg): Adjust declaration.
29420 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29421 * ipa-profile.c (ipa_profile): Likewise.
29422 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29423 (ipa_populate_param_decls): Make descriptors parameter garbage
29425 (ipa_dump_param): Adjust to descriptors being a pointer.
29426 (ipa_alloc_node_params): Likewise.
29427 (ipa_initialize_node_params): Likewise.
29428 (load_from_param_1): Make descriptors parameter garbage collected.
29429 (load_from_unmodified_param): Likewise.
29430 (load_from_param): Likewise.
29431 (ipa_load_from_parm_agg): Likewise.
29432 (ipa_node_params::~ipa_node_params): Removed.
29433 (ipa_free_all_node_params): Remove call to delete operator.
29434 (ipa_node_params_t::insert): New.
29435 (ipa_node_params_t::remove): Likewise.
29436 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29437 copy known_csts and known_contexts vectors.
29438 (ipa_read_node_info): Adjust to descriptors being a pointer.
29439 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29441 (ipcp_transform_function): Make descriptors variable garbage
29444 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29446 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29447 * config/i386/avx512dqintrin.h: Ditto.
29448 * config/i386/avx512fintrin.h: Ditto.
29449 * config/i386/i386.c: Handle new builtins.
29450 * config/i386/i386-builtin.def: Add new builtins.
29451 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29452 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29454 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29455 Martin Liska <mliska@suse.cz>
29457 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29458 * asan.c (asan_expand_poison_ifn): Support stores and use
29459 appropriate ASAN report function.
29460 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29461 * internal-fn.def (ASAN_POISON_USE): Declare.
29462 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29463 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29464 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29465 ASAN_POISON calls w/o LHS.
29466 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29467 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29468 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29469 * gimplify.c (asan_poison_variables): Add attribute
29470 use_after_scope_memory to variables that really needs to live
29472 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29473 having the attribute.
29475 2017-01-23 Martin Liska <mliska@suse.cz>
29477 * asan.c (create_asan_shadow_var): New function.
29478 (asan_expand_poison_ifn): Likewise.
29479 * asan.h (asan_expand_poison_ifn): New declaration.
29480 * internal-fn.c (expand_ASAN_POISON): Likewise.
29481 * internal-fn.def (ASAN_POISON): New builtin.
29482 * sanopt.c (pass_sanopt::execute): Expand
29483 asan_expand_poison_ifn.
29484 * tree-inline.c (copy_decl_for_dup_finish): Make function
29486 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29487 * tree-ssa.c (is_asan_mark_p): New function.
29488 (execute_update_addresses_taken): Rewrite local variables
29489 (identified just by use-after-scope as addressable) into SSA.
29491 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29493 * doc/install.texi (Specific): opensource.apple.com uses https
29494 now. Remove trailing slash.
29496 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29498 * README.Portability: Remove note on an Irix compatibility issue.
29500 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29502 * gcov.c (INCLUDE_ALGORITHM): Define.
29503 (INCLUDE_VECTOR): Define.
29504 No longer include <vector> and <algorithm> directly.
29506 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29508 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29510 * doc/invoke.texi (Code Gen Options): Ditto.
29512 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29515 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
29517 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
29519 rtl-optimization/79125
29520 * cprop.c (local_cprop_pass): Handle cases where we make an
29521 unconditional trap.
29523 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
29527 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
29528 read from, for big endian.
29530 2017-01-20 Jiong Wang <jiong.wang@arm.com>
29532 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
29533 register pauth builtins for LP64 only.
29535 2017-01-20 Marek Polacek <polacek@redhat.com>
29538 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
29541 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29543 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
29545 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
29546 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
29547 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
29549 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29552 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
29553 in big-endian mode when they are not a single duplicated value.
29555 2017-01-20 Richard Biener <rguenther@suse.de>
29557 * BASE-VER: Bump to 7.0.1.
29559 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29561 * omp-low.c (omplow_simd_context): New struct. Use it...
29562 (lower_rec_simd_input_clauses): ...here and...
29563 (lower_rec_input_clauses): ...here to hold common data. Adjust all
29564 references to idx, lane, max_vf, is_simt.
29566 2017-01-20 Graham Markall <graham.markall@embecosm.com>
29568 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
29571 2017-01-20 Martin Jambor <mjambor@suse.cz>
29573 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
29574 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
29576 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
29577 (GTFILES): Rename hsa.c to hsa-common.c.
29578 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
29579 * hsa-dump.c: Likewise.
29580 * hsa-gen.c: Likewise.
29581 * hsa-regalloc.c: Likewise.
29582 * ipa-hsa.c: Likewise.
29583 * omp-expand.c: Likewise.
29584 * omp-low.c: Likewise.
29585 * toplev.c: Likewise.
29587 2017-01-20 Marek Polacek <polacek@redhat.com>
29590 * doc/invoke.texi: Document -Wduplicated-branches.
29591 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
29592 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
29593 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
29594 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
29595 return 0 only when not OEP_LEXICOGRAPHIC.
29596 (fold_build_cleanup_point_expr): Use the expression
29597 location when building CLEANUP_POINT_EXPR.
29598 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
29599 * tree.c (add_expr): Handle error_mark_node.
29601 2017-01-20 Martin Liska <mliska@suse.cz>
29604 * tree-profile.c (init_ic_make_global_vars): Do not call
29606 (gimple_init_gcov_profiler): Likewise.
29608 2017-01-20 Martin Liska <mliska@suse.cz>
29611 * cgraph.h (maybe_create_reference): Remove argument and
29613 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29615 * ipa-cp.c (create_specialized_node): Likewise.
29616 * symtab.c (symtab_node::maybe_create_reference): Handle
29617 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29619 2017-01-20 Martin Liska <mliska@suse.cz>
29621 * read-rtl-function.c (function_reader::create_function): Use
29622 build_decl instread of build_decl_stat.
29624 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29626 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29627 * config/i386/avx512dqintrin.h: Ditto.
29628 * config/i386/avx512fintrin.h: Ditto.
29629 * config/i386/i386-builtin-types.def: Add new types.
29630 * config/i386/i386.c: Handle new types.
29631 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29632 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29633 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29634 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29635 (__builtin_ia32_kshiftridi): New.
29636 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29638 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29642 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29643 define to rs6000_init_stack_protect_guard.
29644 (rs6000_init_stack_protect_guard): New function.
29646 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29647 Yunqiang Su <yunqiang.su@imgtec.com>
29649 * config.gcc (supported_defaults): Add madd4.
29650 (with_madd4): Add validation.
29651 (all_defaults): Add madd4.
29652 * config/mips/mips.opt (mmadd4): New option.
29653 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29655 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29657 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29658 (ISA_HAS_FUSED_MADD4): Likewise.
29659 * doc/invoke.texi (-mmadd4): Document the new option.
29660 * doc/install.texi (--with-madd4): Document the new option.
29662 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29664 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29665 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29666 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29667 (aarch64_init_pauth_hint_builtins): New.
29668 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29669 (aarch64_expand_builtin): Expand new builtins.
29671 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29673 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29674 * combine-stack-adj.c (no_unhandled_cfa): Handle
29675 REG_CFA_TOGGLE_RA_MANGLE.
29676 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29677 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29678 info for return address signing.
29679 (aarch64_expand_epilogue): Likewise.
29681 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29683 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29684 * config/aarch64/aarch64-protos.h
29685 (aarch64_return_address_signing_enabled): New declaration.
29686 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29688 (aarch64_expand_prologue): Sign return address before it's pushed onto
29690 (aarch64_expand_epilogue): Authenticate return address fetched from
29692 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29693 (aarch64_attributes): New function attributes for "sign-return-address".
29694 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29695 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29696 ("*do_return"): Generate combined instructions according to key index.
29697 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29698 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29700 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29701 * config/aarch64/aarch64.opt (msign-return-address=): New.
29702 * doc/extend.texi (AArch64 Function Attributes): Documents
29703 "sign-return-address=".
29704 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29706 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29708 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29709 overall option summary.
29711 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29713 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29714 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29715 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29716 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29718 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29720 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29721 -mpower9-minmax by default for -mcpu=power9.
29722 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29723 128-bit floating point.
29725 2017-01-20 Alan Modra <amodra@gmail.com>
29727 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29728 optimizing for size.
29730 2017-01-20 Alan Modra <amodra@gmail.com>
29733 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29734 for strcmp and strncmp from corresponding builtin decl.
29736 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29738 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29739 instead of i386/rtems-64.h.
29740 * config/i386/rtems-64.h: Remove.
29742 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29746 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29748 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29750 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29752 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29753 Change int to HOST_WIDE_INT.
29754 * config/aarch64/aarch64-protos.h
29755 (aarch64_simd_gen_const_vector_dup): Likewise.
29756 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29758 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29760 * langhooks-def.h (lhd_type_for_size): New decl.
29761 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29762 * langhooks.c (lhd_type_for_size): New function, taken from
29765 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29767 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29768 define_bypass for CR latency.
29769 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29770 (power9-alu2): Add define_bypass for CR latency.
29772 (power9-mul): Update insn latency.
29773 (power9-mul-compare): Update insn latency, bypass latency and remove
29776 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29778 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29780 * config/aarch64/aarch64.md
29781 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29782 aarch64_nopcrelative_literal_loads.
29783 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29785 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29787 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29788 TARGET_LOONGSON_3A.
29789 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29791 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29794 * config.gcc (supported_defaults): Add lxc1-sxc1.
29795 (with_lxc1_sxc1): Add validation.
29796 (all_defaults): Add lxc1-sxc1.
29797 * config/mips/mips.opt (mlxc1-sxc1): New option.
29798 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29800 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29801 __mips_no_lxc1_sxc1.
29802 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29803 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29804 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29806 2017-01-19 Richard Biener <rguenther@suse.de>
29808 PR tree-optimization/72488
29809 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29810 sure to restore SSA info.
29811 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29813 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29815 PR rtl-optimization/79121
29816 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29817 of the inner type when shifting an extended value.
29819 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29822 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29823 interposable aliases.
29825 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29828 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29829 Use the evmergelohi instruction.
29830 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29831 (mov_sitf_e500_subreg8_2_be): Likewise.
29832 (mov_sitf_e500_subreg12_2_le): Likewise.
29833 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29834 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29835 (mov_sitf_e500_subreg8_2_le): Likewise.
29836 (mov_sitf_e500_subreg12_2_be): Likewise.
29838 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29840 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29841 attribute from vecsimple to vecperm.
29842 (altivec_vbpermq2): Likewise.
29844 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29847 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29849 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29850 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29851 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29852 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29853 case where N arg is SIZE_MAX.
29854 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29855 (cmpstrsi): Add pattern.
29857 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29859 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29860 __builtin_vec_revb builtins.
29861 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29862 built-in functions to support generation of the ISA 3.0 XXBR<x>
29863 vector byte reverse instructions.
29864 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29865 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29866 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29867 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29868 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29869 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29870 (P9V_BUILTIN_VEC_REVB): Likewise.
29871 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29872 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29873 (p9_xxbrq_v16qi): Likewise.
29874 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29875 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29876 (p9_xxbrh_v8hi): Likewise.
29877 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29878 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29879 vec_revb built-in functions.
29881 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29883 PR rtl-optimization/78952
29884 * config/i386/i386.md (any_extract): New code iterator.
29885 (*insvqi_2): Use any_extract for source operand.
29886 (*insvqi_3): Use any_shiftrt for source operand.
29888 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29890 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29892 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29894 2017-01-18 Matthias Klose <doko@ubuntu.com>
29896 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29898 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29900 * config/rs6000/altivec.h (vec_bperm): Change #define.
29901 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29902 (altivec_vbpermq2): New define_insn.
29903 (altivec_vbpermd): Likewise.
29904 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29905 function interface.
29906 (VBPERMD): Likewise.
29907 (VBPERM): New polymorphic function interface.
29908 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29909 Add entries for P9V_BUILTIN_VEC_VBPERM.
29910 * doc/extend.texi: Add interfaces for vec_bperm.
29912 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29914 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29915 first letter of error messages.
29916 (s390_resolve_overloaded_builtin): Likewise.
29917 * config/s390/s390.c (s390_expand_builtin): Likewise.
29918 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29919 (s390_valid_target_attribute_inner_p): Likewise.
29920 * config/s390/s390.md ("tabort"): Likewise.
29922 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29924 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29925 (ISA_AVOID_DIV_HILO): New macro.
29926 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29927 (ISA_HAS_DDIV): Likewise.
29929 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29931 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29933 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29935 * doc/invoke.texi (fabi-version): Spelling fix.
29937 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29940 * doc/invoke.texi (fabi-version): Mention mangling fix for
29943 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29946 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29948 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29951 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29954 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29956 (stack_protect_test): Ditto.
29957 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29958 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29960 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29961 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29962 -mstack-protector-guard-offset=.
29963 (RS/6000 and PowerPC Options): Ditto.
29965 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29967 * config/i386/i386.h (MASK_CLASS_P): New define.
29968 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29969 there are no registers from different register sets also when
29970 mask registers are used. Update function comment.
29971 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29972 to (*k/*r) and (*k/*km) alternatives.
29974 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29976 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29977 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29978 (EH_RETURN_HANDLER_RTX): New define.
29979 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29980 Force frame pointer in EH return functions.
29981 (aarch64_expand_epilogue): Add barrier for eh_return.
29982 (aarch64_final_eh_return_addr): Remove.
29983 (aarch64_eh_return_handler_rtx): New function.
29984 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29986 (aarch64_eh_return_handler_rtx): New prototype.
29988 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29990 * config/rs6000/altivec.h (vec_rlmi): New #define.
29991 (vec_vrlnm): Likewise.
29992 (vec_rlnm): Likewise.
29993 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29994 (UNSPEC_VRLNM): Likewise.
29995 (VIlong): New mode iterator.
29996 (altivec_vrl<VI_char>mi): New define_insn.
29997 (altivec_vrl<VI_char>nm): Likewise.
29998 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30000 (VRLDNM): Likewise.
30001 (RLNM): New polymorphic function entry.
30002 (VRLWMI): New monomorphic function entry.
30003 (VRLDMI): Likewise.
30004 (RLMI): New polymorphic function entry.
30005 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30006 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30007 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30010 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30013 * dwarf2out.c (field_byte_offset): Restore the
30014 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30015 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30016 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30019 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30022 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30024 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30027 * dwarf2out.c (add_data_member_location_attribute): For constant
30028 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30029 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30030 DW_AT_byte_size attributes.
30032 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30034 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30035 after forcing to constant memory when the code model is medium.
30037 2017-01-17 Julia Koval <julia.koval@intel.com>
30040 * config/i386/avx512fintrin.h
30041 (_mm512_i32gather_ps): Change __addr type to void const*.
30042 (_mm512_mask_i32gather_ps): Ditto.
30043 (_mm512_i32gather_pd): Ditto.
30044 (_mm512_mask_i32gather_pd): Ditto.
30045 (_mm512_i64gather_ps): Ditto.
30046 (_mm512_mask_i64gather_ps): Ditto.
30047 (_mm512_i64gather_pd): Ditto.
30048 (_mm512_mask_i64gather_pd): Ditto.
30049 (_mm512_i32gather_epi32): Ditto.
30050 (_mm512_mask_i32gather_epi32): Ditto.
30051 (_mm512_i32gather_epi64): Ditto.
30052 (_mm512_mask_i32gather_epi64): Ditto.
30053 (_mm512_i64gather_epi32): Ditto.
30054 (_mm512_mask_i64gather_epi32): Ditto.
30055 (_mm512_i64gather_epi64): Ditto.
30056 (_mm512_mask_i64gather_epi64): Ditto.
30057 (_mm512_i32scatter_ps): Change __addr type to void*.
30058 (_mm512_mask_i32scatter_ps): Ditto.
30059 (_mm512_i32scatter_pd): Ditto.
30060 (_mm512_mask_i32scatter_pd): Ditto.
30061 (_mm512_i64scatter_ps): Ditto.
30062 (_mm512_mask_i64scatter_ps): Ditto.
30063 (_mm512_i64scatter_pd): Ditto.
30064 (_mm512_mask_i64scatter_pd): Ditto.
30065 (_mm512_i32scatter_epi32): Ditto.
30066 (_mm512_mask_i32scatter_epi32): Ditto.
30067 (_mm512_i32scatter_epi64): Ditto.
30068 (_mm512_mask_i32scatter_epi64): Ditto.
30069 (_mm512_i64scatter_epi32): Ditto.
30070 (_mm512_mask_i64scatter_epi32): Ditto.
30071 (_mm512_i64scatter_epi64): Ditto.
30072 (_mm512_mask_i64scatter_epi64): Ditto.
30073 * config/i386/avx512pfintrin.h
30074 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30075 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30076 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30077 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30078 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30079 (_mm512_prefetch_i32scatter_ps): Ditto.
30080 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30081 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30082 (_mm512_prefetch_i64scatter_pd): Ditto.
30083 (_mm512_prefetch_i64scatter_ps): Ditto.
30084 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30085 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30086 * config/i386/avx512vlintrin.h
30087 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30088 (_mm_mmask_i32gather_ps): Ditto.
30089 (_mm256_mmask_i32gather_pd): Ditto.
30090 (_mm_mmask_i32gather_pd): Ditto.
30091 (_mm256_mmask_i64gather_ps): Ditto.
30092 (_mm_mmask_i64gather_ps): Ditto.
30093 (_mm256_mmask_i64gather_pd): Ditto.
30094 (_mm_mmask_i64gather_pd): Ditto.
30095 (_mm256_mmask_i32gather_epi32): Ditto.
30096 (_mm_mmask_i32gather_epi32): Ditto.
30097 (_mm256_mmask_i32gather_epi64): Ditto.
30098 (_mm_mmask_i32gather_epi64): Ditto.
30099 (_mm256_mmask_i64gather_epi32): Ditto.
30100 (_mm_mmask_i64gather_epi32): Ditto.
30101 (_mm256_mmask_i64gather_epi64): Ditto.
30102 (_mm_mmask_i64gather_epi64): Ditto.
30103 (_mm256_i32scatter_ps): Change __addr type to void*.
30104 (_mm256_mask_i32scatter_ps): Ditto.
30105 (_mm_i32scatter_ps): Ditto.
30106 (_mm_mask_i32scatter_ps): Ditto.
30107 (_mm256_i32scatter_pd): Ditto.
30108 (_mm256_mask_i32scatter_pd): Ditto.
30109 (_mm_i32scatter_pd): Ditto.
30110 (_mm_mask_i32scatter_pd): Ditto.
30111 (_mm256_i64scatter_ps): Ditto.
30112 (_mm256_mask_i64scatter_ps): Ditto.
30113 (_mm_i64scatter_ps): Ditto.
30114 (_mm_mask_i64scatter_ps): Ditto.
30115 (_mm256_i64scatter_pd): Ditto.
30116 (_mm256_mask_i64scatter_pd): Ditto.
30117 (_mm_i64scatter_pd): Ditto.
30118 (_mm_mask_i64scatter_pd): Ditto.
30119 (_mm256_i32scatter_epi32): Ditto.
30120 (_mm256_mask_i32scatter_epi32): Ditto.
30121 (_mm_i32scatter_epi32): Ditto.
30122 (_mm_mask_i32scatter_epi32): Ditto.
30123 (_mm256_i32scatter_epi64): Ditto.
30124 (_mm256_mask_i32scatter_epi64): Ditto.
30125 (_mm_i32scatter_epi64): Ditto.
30126 (_mm_mask_i32scatter_epi64): Ditto.
30127 (_mm256_i64scatter_epi32): Ditto.
30128 (_mm256_mask_i64scatter_epi32): Ditto.
30129 (_mm_i64scatter_epi32): Ditto.
30130 (_mm_mask_i64scatter_epi32): Ditto.
30131 (_mm256_i64scatter_epi64): Ditto.
30132 (_mm256_mask_i64scatter_epi64): Ditto.
30133 (_mm_i64scatter_epi64): Ditto.
30134 (_mm_mask_i64scatter_epi64): Ditto.
30135 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30136 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30137 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30138 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30139 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30140 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30141 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30142 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30143 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30144 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30145 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30146 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30147 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30148 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30149 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30150 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30151 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30152 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30153 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30154 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30155 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30156 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30157 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30158 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30159 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30160 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30161 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30162 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30163 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30164 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30165 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30166 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30167 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30168 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30169 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30170 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30171 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30172 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30173 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30174 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30175 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30176 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30177 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30178 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30179 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30180 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30181 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30182 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30183 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30184 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30185 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30186 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30187 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30188 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30189 definitions accordingly.
30191 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30192 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30195 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30198 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30201 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30202 hard regs for inner regno.
30204 2017-01-17 Martin Liska <mliska@suse.cz>
30207 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30208 assumption and add comment.
30210 2017-01-17 Nathan Sidwell <nathan@acm.org>
30212 * ipa-visibility.c (localize_node): New function, broken out of ...
30213 (function_and_variable_visibility): ... here. Call it.
30215 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30217 PR middle-end/77445
30218 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30219 correctly set frequency of oudgoing edge.
30220 (duplicate_thread_path): Fix profile updating.
30222 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30225 * configure.ac: Add GCC_BASE_VER.
30226 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30227 version from BASE-VER file.
30228 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30229 (gcc.o): Depend on $(BASEVER).
30230 * common.opt (dumpfullversion): New option.
30231 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30232 * doc/invoke.texi: Document -dumpfullversion.
30233 * doc/install.texi: Document --with-gcc-major-version-only.
30234 * configure: Regenerated.
30236 2017-01-17 Richard Biener <rguenther@suse.de>
30238 PR tree-optimization/71433
30239 * tree-vrp.c (register_new_assert_for): Merge same asserts
30240 on all incoming edges.
30241 (process_assert_insertions_for): Handle insertions at the
30244 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30246 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30247 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30249 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30252 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30255 2017-01-17 Alan Modra <amodra@gmail.com>
30258 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30259 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30260 symbolic stack limit when pic.
30262 2017-01-16 Martin Sebor <msebor@redhat.com>
30264 PR tree-optimization/78608
30265 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30267 2017-01-16 Jeff Law <law@redhat.com>
30270 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30271 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30272 for several include directories that may be relative to sysroot.
30273 * config/i386/x-mingw32 (gplus_includedir): Define.
30274 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30275 (native_system_includedir): Likewise.
30276 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30277 override if TARGET_SYSTEM_ROOT is defined.
30278 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30280 PR tree-optimization/79090
30281 PR tree-optimization/33562
30282 PR tree-optimization/61912
30283 PR tree-optimization/77485
30284 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30285 and computed trims into the dump file.
30287 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30289 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30291 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30294 * gimplify.c (gimplify_init_constructor): If want_value and
30295 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30299 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30300 sequence. Formatting fixes.
30301 (doloop_optimize): Formatting fixes.
30304 * gcc.c (debug_level_greater_than_spec_func): New function.
30305 (static_spec_functions): Add debug-level-gt spec function.
30306 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30308 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30309 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30310 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30311 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30312 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30313 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30315 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30317 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30318 QImode fixups to general and mask registers only.
30320 2017-01-16 Carl Love <cel@us.ibm.com>
30322 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30323 for built-in functions
30324 vector signed char vec_nabs (vector signed char)
30325 vector signed short vec_nabs (vector signed short)
30326 vector signed int vec_nabs (vector signed int)
30327 vector signed long long vec_nabs (vector signed long long)
30328 vector float vec_nabs (vector float)
30329 vector double vec_nabs (vector double)
30330 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30332 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30333 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30334 * doc/extend.texi: Update the documentation file for the new built-in
30337 2017-01-16 Martin Sebor <msebor@redhat.com>
30339 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30342 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30344 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30345 UNSPEC_VSX__XXSPLTD to require special splat handling.
30347 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30350 * system.h: Poison strndup.
30352 2017-01-16 Alan Modra <amodra@gmail.com>
30355 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30358 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30360 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30362 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30364 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30365 call recog here. Assert that INSN_CODE (insn) is non-negative.
30367 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30370 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30372 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30373 in the currently scheduled RTL fragment.
30375 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30377 PR rtl-optimization/78751
30378 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30381 2017-01-14 Jeff Law <law@redhat.com>
30383 PR tree-optimization/79090
30384 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30385 variable length stores.
30386 (compute_trims): Delete dead assignment to *trim_tail.
30387 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30390 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30392 PR rtl-optimization/78626
30393 PR rtl-optimization/78727
30394 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30395 of a block, and split such blocks after everything else is finished.
30397 2017-01-14 Alan Modra <amodra@gmail.com>
30400 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30401 target legitimate_combined_insn.
30402 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30403 (rs6000_legitimate_combined_insn): New function.
30404 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30406 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30407 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30408 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30410 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30412 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30414 2017-01-13 Jeff Law <law@redhat.com>
30416 PR tree-optimization/33562
30417 PR tree-optimization/61912
30418 PR tree-optimization/77485
30419 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30421 (delete_dead_assignment): Likewise.
30422 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30423 statement to delete_dead_call and delete_dead_assignment.
30425 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30428 * substring-locations.c (format_warning_va): Strengthen case 1 so
30429 that both endpoints of the substring must be within the format
30430 range for just the substring to be printed.
30432 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30434 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30435 * config/i386/i386.c (ix86_target_string): Add missing options
30436 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30437 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30438 flags_other and ix86_target_other to flags2_other. Display unknown
30440 (ix86_valid_target_attribute_inner_p): Add missing options and
30441 reorder options by implied ISAs, as in ix86_target_string.
30443 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30445 * hash-table.h (hash_table::too_empty_p): New function.
30446 (hash_table::expand): Use it.
30447 (hash_table::traverse): Likewise.
30448 (hash_table::empty_slot): Use sizeof (value_type) instead of
30449 sizeof (PTR) to convert bytes to elements. Shrink the table
30450 if the current size is excessive for the current number of
30453 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30455 * ira-costs.c (record_reg_classes): Break from the inner loop
30456 early once alt_fail is known to be true. Update outer loop
30457 handling accordingly.
30459 2017-01-13 Jeff Law <law@redhat.com>
30461 * tree-ssa-dse.c (decrement_count): New function.
30462 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30463 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30464 when we know the partially dead statement is a mem* function.
30466 PR tree-optimization/61912
30467 PR tree-optimization/77485
30468 * tree-ssa-dse.c: Include expr.h.
30469 (maybe_trim_constructor_store): New function.
30470 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30472 PR tree-optimization/33562
30473 PR tree-optimization/61912
30474 PR tree-optimization/77485
30475 * doc/invoke.texi: Document new dse-max-object-size param.
30476 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30477 * tree-ssa-dse.c: Include params.h.
30478 (dse_store_status): New enum.
30479 (initialize_ao_ref_for_dse): New, partially extracted from
30481 (valid_ao_ref_for_dse, normalize_ref): New.
30482 (setup_live_bytes_from_ref, compute_trims): Likewise.
30483 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30484 (maybe_trim_partially_dead_store): Likewise.
30485 (maybe_trim_complex_store): Likewise.
30486 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30487 Track what bytes live from the original store. Return tri-state
30488 for dead, partially dead or live.
30489 (dse_dom_walker): Add constructor, destructor and new private members.
30490 (delete_dead_call, delete_dead_assignment): New extracted from
30492 (dse_optimize_stmt): Make a member of dse_dom_walker.
30493 Use initialize_ao_ref_for_dse.
30495 PR tree-optimization/33562
30496 PR tree-optimization/61912
30497 PR tree-optimization/77485
30498 * sbitmap.h (bitmap_count_bits): Prototype.
30499 (bitmap_clear_range, bitmap_set_range): Likewise.
30500 * sbitmap.c (bitmap_clear_range): New function.
30501 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30503 2017-01-13 Martin Liska <mliska@suse.cz>
30506 * function.c (set_cfun): Add new argument force.
30507 * function.h (set_cfun): Likewise.
30508 * ipa-inline-transform.c (inline_call): Use the function when
30509 strict alising from is dropped for function we inline to.
30511 2017-01-13 Richard Biener <rguenther@suse.de>
30513 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30514 for dumping GIMPLE INTEGER_CSTs.
30516 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30518 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
30519 to 201112L since C++17.
30521 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
30524 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
30525 if -fsanitize=kernel-address is present.
30527 2017-01-13 Richard Biener <rguenther@suse.de>
30529 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
30530 as _Literal ( type ) number in case usual suffixes do not
30531 preserve all information.
30533 2017-01-13 Richard Biener <rguenther@suse.de>
30535 PR tree-optimization/77283
30536 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
30537 and ssa-iterators.h.
30538 (is_feasible_trace): Implement a cost model based on joiner
30541 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
30544 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
30545 char or short to __float128/_Float128 directly.
30547 2017-01-12 Martin Sebor <msebor@redhat.com>
30549 to -Wformat-overflow.
30550 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
30551 (min_bytes_remaining): Same.
30552 (get_string_length): Same.
30553 (format_string): Same.
30554 (format_directive): Same.
30556 (pass_sprintf_length::handle_gimple_call): Same.
30558 2017-01-12 Jakub Jelinek <jakub@redhat.com>
30560 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
30561 info.nowrite calls with no lhs that can't throw. Return bool
30562 whether gsi_remove has been called or not.
30563 (pass_sprintf_length::handle_gimple_call): Return bool whether
30564 try_substitute_return_value called gsi_remove. Formatting fix.
30565 (pass_sprintf_length::execute): Don't use gsi_remove if
30566 handle_gimple_call returned true.
30569 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
30570 be removed due to side-effects, don't remove following barrier nor
30571 turn the successor edge into fallthru edge.
30573 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30576 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
30577 element-reversing loads and stores as not swappable.
30579 2017-01-12 Nathan Sidwell <nathan@acm.org>
30580 Nicolai Stange <nicstange@gmail.com>
30582 * combine.c (try_combine): Don't ignore result of overlap checking
30583 loop. Combine overlap & asm check into single loop.
30585 2017-01-12 Richard Biener <rguenther@suse.de>
30587 * tree-pretty-print.c (dump_generic_node): Provide -gimple
30588 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
30590 2017-01-12 Richard Biener <rguenther@suse.de>
30592 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
30593 and TS_TARGET_OPTION directly derive from TS_BASE.
30594 * tree-core.h (tree_optimization_option): Derive from tree_base.
30595 (tree_target_option): Likewise.
30597 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
30599 * config/i386/i386.c (memory_address_length): Increase len
30600 only when rip_relative_addr_p returns false.
30602 2017-01-11 Julia Koval <julia.koval@intel.com>
30604 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30605 (OPTION_MASK_ISA_SGX_SET): New.
30606 (ix86_handle_option): Handle OPT_msgx.
30607 * config.gcc: Added sgxintrin.h.
30608 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30609 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30610 * config/i386/i386.c (ix86_target_string): Add -msgx.
30612 (ix86_option_override_internal): Handle new options.
30613 (ix86_valid_target_attribute_inner_p): Add sgx.
30614 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30615 * config/i386/i386.opt: Add msgx.
30616 * config/i386/sgxintrin.h: New file.
30617 * config/i386/x86intrin.h: Add sgxintrin.h.
30619 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30622 * fold-const.c (maybe_nonzero_address): Return 1 for function
30624 (tree_single_nonzero_warnv_p): Don't handle function local objects
30628 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30631 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30634 * opts.c: Include "spellcheck.h"
30635 (struct string_fragment): New struct.
30636 (struct edit_distance_traits<const string_fragment &>): New
30638 (get_closest_sanitizer_option): New function.
30639 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30641 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30643 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30645 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30646 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30647 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30648 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30649 for initial die_offset if dwarf_split_debug_info.
30650 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30651 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30653 (output_skeleton_debug_sections): Formatting fix. Use
30654 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30655 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30657 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30659 * config/arm/cortex-a53.md: Add bypasses for
30660 cortex_a53_r2f_cvt.
30661 (cortex_a53_r2f): Only use for transfers.
30662 (cortex_a53_f2r): Likewise.
30663 (cortex_a53_r2f_cvt): Add reservation for conversions.
30664 (cortex_a53_f2r_cvt): Likewise.
30666 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30668 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30669 to all inlined functions, change static to extern.
30671 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30674 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30676 (arm_assemble_integer): Likewise.
30678 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30680 * config.gcc: Use new awk script to check CPU, FPU and architecture
30681 parameters for --with-... options.
30682 * config/arm/parsecpu.awk: New file
30683 * config/arm/arm-cpus.in: New file.
30684 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30686 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30688 * config/arm/t-arm: Update dependency rules.
30689 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30690 of processing .def files.
30691 * config/arm/genopt.sh: Deleted.
30692 * config/arm/gentune.sh: Deleted.
30693 * config/arm/arm-cores.def: Deleted.
30694 * config/arm/arm-arches.def: Deleted.
30695 * config/arm/arm-fpus.def: Deleted.
30696 * config/arm/arm-tune.md: Regenerated.
30697 * config/arm/arm-tables.opt: Regenerated.
30698 * config/arm/arm-cpu.h: New generated file.
30699 * config/arm/arm-cpu-data.h: New generated file.
30700 * config/arm/arm-cpu-cdata.h: New generated file.
30702 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30705 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30707 (input_varpool_node): Unpack dynamically_initialized bit.
30709 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30711 PR rtl-optimization/79032
30712 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30713 the alignment of the adjusted memory reference against that of MODE,
30714 instead of the alignment of the original memory reference.
30716 2017-01-11 Martin Jambor <mjambor@suse.cz>
30718 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30720 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30721 decorated functions.
30723 2017-01-11 Richard Biener <rguenther@suse.de>
30725 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30726 set range/nonnull info for PHI results. Do not set it on
30727 stmts marked for removal.
30729 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30731 * expr.c (store_field): In the bitfield case, fetch the return value
30732 from the registers before applying a single big-endian adjustment.
30733 Always do a final load for a BLKmode value not larger than a word.
30735 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30738 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30739 that we correctly handle column numbers greater than
30740 LINE_MAP_MAX_COLUMN_NUMBER.
30742 2017-01-10 Martin Sebor <msebor@redhat.com>
30744 PR middle-end/78245
30745 * gimple-ssa-sprintf.c (get_destination_size): Call
30746 {init,fini}object_sizes.
30747 * tree-object-size.c (addr_object_size): Adjust.
30748 (pass_through_call): Adjust.
30749 (pass_object_sizes::execute): Adjust.
30750 * tree-object-size.h (fini_object_sizes): Declare.
30752 2017-01-10 Martin Sebor <msebor@redhat.com>
30754 PR tree-optimization/78775
30755 * builtins.c (get_size_range): Move...
30756 * calls.c: ...to here.
30757 (alloc_max_size): Accept zero argument.
30758 (operand_signed_p): Remove.
30759 (maybe_warn_alloc_args_overflow): Call get_size_range.
30760 * calls.h (get_size_range): Declare.
30762 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30764 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30765 from TI's devices.csv file as of September 2016.
30766 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30768 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30770 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30771 * doc/invoke.texi: Likewise.
30772 * doc/md.texi: Likewise.
30773 * doc/objc.texi: Likewise.
30775 2017-01-10 Joshua Conner <joshconner@google.com>
30777 * config/arm/fuchsia-elf.h: New file.
30778 * config/fuchsia.h: New file.
30779 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30780 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30782 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30784 2016-01-10 Richard Biener <rguenther@suse.de>
30786 PR tree-optimization/79034
30787 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30788 Propagate out degenerate PHIs in the joiner.
30790 2017-01-10 Martin Liska <mliska@suse.cz>
30792 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30793 (sort_congruence_classes_by_decl_uid): Likewise.
30794 (sort_congruence_class_groups_by_decl_uid): Likewise.
30795 (sem_item_optimizer::merge_classes): Sort class, groups in these
30796 classes and members in the groups by DECL_UID of declarations.
30797 This would make merge operations stable.
30799 2017-01-10 Martin Liska <mliska@suse.cz>
30801 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30802 usage of m_classes_vec.
30803 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30804 (sem_item_optimizer::get_group_by_hash): Likewise.
30805 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30806 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30807 (sem_item_optimizer::verify_classes): Likewise.
30808 (sem_item_optimizer::process_cong_reduction): Likewise.
30809 (sem_item_optimizer::dump_cong_classes): Likewise.
30810 (sem_item_optimizer::merge_classes): Likewise.
30811 * ipa-icf.h (congruence_class_hash): Rename from
30812 congruence_class_group_hash. Remove declaration of m_classes_vec.
30814 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30816 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30817 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30818 * config.gcc: Add avx512vpopcntdqintrin.h.
30819 * config/i386/avx512vpopcntdqintrin.h: New.
30820 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30821 * config/i386/i386-builtin-types.def: Add new types.
30822 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30823 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30824 __builtin_ia32_vpopcountq_v8di_mask): New.
30825 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30826 __AVX512VPOPCNTDQ__.
30827 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30828 (PTA_AVX512VPOPCNTDQ): Define.
30829 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30830 TARGET_AVX512VPOPCNTDQ_P): Define.
30831 * config/i386/i386.opt: Add mavx512vpopcntdq.
30832 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30833 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30835 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30837 PR middle-end/77484
30838 * predict.def (PRED_CALL): Set to 67.
30840 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30842 * expr.c (store_field): In the bitfield case, if the value comes from
30843 a function call and is of an aggregate type returned in registers, do
30844 not modify the field mode; extract the value in all cases if the mode
30845 is BLKmode and the size is not larger than a word.
30847 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30850 * config/i386/cpuid.h: Fix undefined behavior.
30852 2017-01-04 Jeff Law <law@redhat.com>
30854 PR tree-optimization/79007
30855 PR tree-optimization/67955
30856 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30857 conservative for pt.null when flag_non_call_exceptions is on.
30859 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30861 PR translation/79019
30862 PR translation/79020
30863 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30864 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30865 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30867 * config/avr/avr.opt (maccumulate-args): Likewise.
30868 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30869 * common.opt (freport-bug): Likewise.
30870 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30871 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30872 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30873 translatable string.
30874 * config/i386/i386.c (function_value_32): Likewise.
30875 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30876 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30878 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30879 * common/config/msp430/msp430-common.c (msp430_handle_option):
30881 * symtab.c (symtab_node::verify_base): Likewise.
30882 * opts.c (set_debug_level): Likewise.
30883 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30884 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30885 missing whitespace to translatable strings.
30886 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30887 * config/sh/superh.h: Likewise.
30888 * config/i386/xopintrin.h: Likewise.
30889 * config/i386/znver1.md: Likewise.
30890 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30891 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30892 * double-int.h (struct double_int): Likewise.
30893 * double-int.c (div_and_round_double): Likewise.
30894 * wide-int.cc: Likewise.
30895 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30896 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30897 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30898 (crossjumps_occurred): ... this.
30899 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30902 PR tree-optimization/78899
30903 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30904 returning bool return struct loop *, NULL for failure and the new
30906 (versionable_outer_loop_p): Don't version outer loop if it has
30907 dont_vectorized bit set.
30908 (tree_if_conversion): When versioning outer loop, ensure
30909 tree_if_conversion is performed also on the inner loop of the
30910 non-vectorizable outer loop copy.
30911 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30912 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30913 prevent vectorization of it.
30914 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30915 the outer loop vectorization of the non-scalar version is attempted
30916 before vectorization of the inner loop in scalar version. If
30917 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30918 vectorization of its inner loop.
30919 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30920 has 2 inner loops, rename also on edges from bb whose single pred
30921 is outer_loop->header. Fix typo in function comment.
30923 2017-01-09 Martin Sebor <msebor@redhat.com>
30926 * asan.c (asan_emit_stack_protection): Increase local buffer size
30927 to avoid snprintf truncation warning.
30929 2017-01-09 Andrew Pinski <apinski@cavium.com>
30931 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30932 to reference thunderx2t99 for the tuning structure
30933 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30935 (thunderx2t99_extra_costs): This.
30936 * config/aarch64/aarch64-tune.md: Regenerate.
30937 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30938 (vulcan_addrcost_table): This.
30939 (vulcan_regmove_cost): Rename to ...
30940 (thunderx2t99_regmove_cost): This.
30941 (vulcan_vector_cost): Rename to ...
30942 (thunderx2t99_vector_cost): this.
30943 (vulcan_branch_cost): Rename to ...
30944 (thunderx2t99_branch_cost): This.
30945 (vulcan_tunings): Rename to ...
30946 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30947 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30949 2017-01-09 Martin Jambor <mjambor@suse.cz>
30953 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30954 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30955 (propagate_vr_accross_jump_function): Use the above function for all
30956 value range computations for pass-through jump functions and type
30957 converasion from explicit value range values.
30958 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30959 parameters from TYPE_ARG_TYPES.
30960 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30961 (ipa_write_node_info): Stream type of the actual argument.
30962 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30964 2017-01-09 Martin Liska <mliska@suse.cz>
30967 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30968 (lookup_compiler): Do not show error message with have_E.
30970 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30972 PR tree-optimization/78938
30973 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30974 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30975 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30976 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30979 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30981 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30984 2017-01-09 Richard Biener <rguenther@suse.de>
30986 PR tree-optimization/78997
30987 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30988 name condition properly.
30990 2017-01-09 Richard Biener <rguenther@suse.de>
30993 * dwarf2out.c (is_cxx): New overload with context.
30994 (is_naming_typedef_decl): Use it.
30996 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30998 * invoke.texi (Option Summary): Correct spacing in option lists
30999 and add line breaks to fix over-long lines.
31001 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31003 PR middle-end/17660
31005 * extend.texi (Common Variable Attributes): Add xref to GCC
31006 Internals manual to explain mode attribute keywords.
31008 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31011 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31012 and Preprocessor Options.
31013 (Options for Linking): Document -pthread here....
31014 (RS/6000 and PowerPC Options): ...not here.
31015 (Solaris 2 Options): ...or here.
31016 * doc/cppopts.texi: Document -pthread.
31018 2017-01-08 Martin Sebor <msebor@redhat.com>
31020 PR middle-end/77708
31021 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31022 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31023 New member functions.
31024 (format_directive): Used them.
31026 (pass_sprintf_length::handle_gimple_call): Same.
31027 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31028 to avoid truncation for any argument.
31029 (extract_affine_mul): Same.
31030 * tree.c (get_file_function_name): Same.
31032 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31034 PR middle-end/77484
31035 * predict.def (PRED_INDIR_CALL): Set to 86.
31037 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31039 PR preprocessor/54124
31040 * doc/cppopts.texi: Reformat -d subtable to list the full name
31041 of the options. Add cross-reference to the docs for the general
31042 compiler -d options.
31043 * doc/invoke.texi (Developer Options): Add cross-reference to the
31044 preprocessor-specific -d option documentation.
31046 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31048 PR preprocessor/13498
31049 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31050 redudant material, and reflect new command-line options.
31051 (System Headers): Likewise.
31053 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31055 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31056 -isystem, and -idirafter. Copy-edit.
31057 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31058 default for -ftrack-macro-expansion. Delete obsolete and
31059 badly-formatted implementation details about -fdebug-cpp output.
31060 * doc/cppwarnopts.texi: Copy-edit.
31062 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31065 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31066 that the transition from a max line width >= 1<<10 to narrower
31067 lines works correctly.
31069 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31071 * doc/options.texi (PerFunction): New.
31072 * opt-functions.awk (switch_flags): Map both Optimization and
31073 PerFunction to CL_OPTIMIZATION.
31074 * opth-gen.awk: Test for PerFunction flag along with
31076 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31077 it only when the latter is present. Skip those that don't in
31078 the hash function generator.
31079 * common.opt (fvar-tracking): Mark as PerFunction instead of
31081 (fvar-tracking-assignments): Likewise.
31082 (fvar-tracking-assignments-toggle): Likewise.
31083 (fvar-tracking-uninit): Likewise.
31085 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31087 PR translation/79018
31088 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31091 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31094 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31095 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31096 TARGET_LONG_JUMP_TABLE_OFFSETS.
31097 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31098 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31099 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31100 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31101 * config/m68k/m68k.md (tablejump expander): Likewise.
31102 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31103 TARGET_LONG_JUMP_TABLE_OFFSETS.
31104 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31105 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31107 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31108 David Holsgrove <david.holsgrove@xilinx.com>
31110 * common/config/microblaze/microblaze-common.c
31111 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31112 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31114 * config/microblaze/microblaze.c (microblaze_must_save_register)
31115 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31117 (microblaze_eh_return): New function.
31118 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31119 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31120 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31121 * config/microblaze/microblaze.md (eh_return): New pattern.
31123 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31125 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31126 GCC_DIAGNOSTIC_STRINGIFY): Define.
31128 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31130 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31132 * config/arm/arm.md (<mcrr>): New.
31134 * config/arm/arm.c (arm_arch5te): New.
31135 (arm_option_override): Set arm_arch5te.
31136 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31138 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31139 (arm_mcrr_qualifiers): ... this. New.
31140 (MRRC_QUALIFIERS): Define to...
31141 (arm_mrrc_qualifiers): ... this. New.
31142 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31144 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31145 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31146 (MRRCI, mrrc, MRRC): New.
31147 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31148 VUNSPEC_MRRC2): New.
31150 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31152 * config/arm/arm.md (<mcr>): New.
31154 * config/arm/arm.c (arm_coproc_builtin_available): Add
31155 support for mcr, mrc, mcr2 and mrc2.
31156 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31157 (arm_mcr_qualifiers): ... this. New.
31158 (MRC_QUALIFIERS): Define to ...
31159 (arm_mrc_qualifiers): ... this. New.
31160 (MCR_QUALIFIERS): Define to ...
31161 (arm_mcr_qualifiers): ... this. New.
31162 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31164 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31165 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31166 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31167 VUNSPEC_MRC2): New.
31169 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31171 * config/arm/arm.md (*ldc): New.
31175 * config/arm/arm.c (arm_coproc_builtin_available): Add
31176 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31177 (arm_coproc_ldc_stc_legitimate_address): New.
31178 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31179 'qualifier_const_pointer'.
31180 (LDC_QUALIFIERS): Define to...
31181 (arm_ldc_qualifiers): ... this. New.
31182 (STC_QUALIFIERS): Define to...
31183 (arm_stc_qualifiers): ... this. New.
31184 * config/arm/arm-protos.h
31185 (arm_coproc_ldc_stc_legitimate_address): New.
31186 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31187 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31188 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31189 stc2, stcl, stc2l): New.
31190 * config/arm/constraints.md (Uz): New.
31191 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31192 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31193 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31194 VUNSPEC_STC2L): New.
31196 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31198 * config/arm/arm.md (<cdp>): New.
31199 * config/arm/arm.c (neon_const_bounds): Rename this ...
31200 (arm_const_bounds): ... this.
31201 (arm_coproc_builtin_available): New.
31202 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31203 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31204 (CDP_QUALIFIERS): Define to...
31205 (arm_cdp_qualifiers): ... this. New.
31207 (arm_expand_builtin_args): Add case for 6 arguments.
31208 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31209 (arm_const_bounds): ... this.
31210 (arm_coproc_builtin_available): New.
31211 * config/arm/arm_acle.h (__arm_cdp): New.
31213 * config/arm/arm_acle_builtins.def (cdp): New.
31215 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31216 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31217 'arm_const_bounds'.
31218 * config/arm/types.md (coproc): New.
31219 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31220 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31221 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31222 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31224 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31226 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31227 (UBINOP_QUALIFIERS): New.
31229 (acle_builtin_data): New. Change comment.
31230 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31231 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31232 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31233 arm_acle_builtins.def.
31234 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31235 (arm_init_acle_builtins): New.
31236 (CRC32_BUILTIN): Remove.
31237 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31238 crc32cb, crc32ch and crc32cw.
31239 (arm_init_crc32_builtins): Remove.
31240 (arm_init_builtins): Use arm_init_acle_builtins rather
31241 than arm_init_crc32_builtins.
31242 (arm_expand_acle_builtin): New.
31243 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31244 * config/arm/arm_acle_builtins.def: New.
31246 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31248 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31249 (arm_builtin_datum): ... this.
31250 (arm_init_neon_builtin): Rename to ...
31251 (arm_init_builtin): ... this. Add a new parameters PREFIX
31252 and USE_SIG_IN_NAME.
31253 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31254 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31255 'arm_builtin_datum'.
31256 (arm_init_vfp_builtins): Likewise.
31257 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31258 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31259 (arm_expand_neon_args): Rename to ...
31260 (arm_expand_builtin_args): ... this. Rename builtin_arg
31261 enum values and differentiate between ARG_BUILTIN_MEMORY
31262 and ARG_BUILTIN_NEON_MEMORY.
31263 (arm_expand_neon_builtin_1): Rename to ...
31264 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31265 values, arm_expand_builtin_args and add bool parameter NEON.
31266 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31267 (arm_expand_vfp_builtin): Likewise.
31268 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31270 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31272 PR middle-end/77484
31273 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31274 * predict.c (tree_estimate_probability_bb): Reverse direction of
31275 polymorphic call predictor.
31277 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31279 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31280 (determine_pass_name_match): ...this new function and...
31281 (should_skip_pass_p): ...this new function.
31283 2017-01-06 Nathan Sidwell <nathan@acm.org>
31285 * ipa-visibility.c (function_and_variable_visibility): Reformat
31286 comments and long lines. Remove extrneous if.
31287 * symtab.c (symtab_node::make_decl_local): Fix code format.
31288 (symtab_node::set_section_for_node): Fix comment typo.
31290 2017-01-06 Martin Liska <mliska@suse.cz>
31293 * lra-constraints.c: Rename invariant to lra_invariant.
31294 * predict.c (set_even_probabilities): Initialize e to NULL.
31296 2017-01-05 Martin Sebor <msebor@redhat.com>
31298 PR tree-optimization/78910
31299 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31300 (format_integer): Correct off-by-one error in the handling
31301 of precision with negative numbers in signed conversions..
31303 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31305 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31307 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31309 PR tree-optimization/71016
31310 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31311 factor_out_conditional_conversion. Formatting fix.
31312 (factor_out_conditional_conversion): Add cond_stmt argument.
31313 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31314 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31317 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31319 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31320 read-rtl-function.o, and selftest-rtl.o.
31321 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31322 (selftest::aarch64_test_loading_full_dump): New function.
31323 (selftest::aarch64_run_selftests): New function.
31324 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31325 selftest::aarch64_run_selftests.
31326 * config/i386/i386.c
31327 (selftest::ix86_test_loading_dump_fragment_1): New function.
31328 (selftest::ix86_test_loading_call_insn): New function.
31329 (selftest::ix86_test_loading_full_dump): New function.
31330 (selftest::ix86_test_loading_unspec): New function.
31331 (selftest::ix86_run_selftests): Call the new functions.
31332 * emit-rtl.c (maybe_set_max_label_num): New function.
31333 * emit-rtl.h (maybe_set_max_label_num): New decl.
31334 * function.c (instantiate_decls): Guard call to
31335 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31336 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31338 * gensupport.c (gen_reader::gen_reader): Pass "false"
31339 for new "compact" param of rtx_reader.
31340 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31341 rather than an empty string for NULL strings.
31342 * read-md.c: Potentially include config.h rather than bconfig.h.
31343 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31344 (have_error): New global, copied from errors.c.
31345 (md_reader::read_name): Rename to...
31346 (md_reader::read_name_1): ...this, adding "out_loc" param,
31347 and converting "missing name or number" to returning false, rather
31349 (md_reader::read_name): Reimplement in terms of read_name_1.
31350 (md_reader::read_name_or_nil): New function.
31351 (md_reader::read_string): Handle "(nil)" by returning NULL.
31352 (md_reader::md_reader): Add new param "compact".
31353 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31354 (md_reader::read_file): New method.
31355 * read-md.h (md_reader::md_reader): Add new param "compact".
31356 (md_reader::read_file): New method.
31357 (md_reader::is_compact): New accessor.
31358 (md_reader::read_name): Convert return type from void to file_location.
31359 (md_reader::read_name_or_nil): New decl.
31360 (md_reader::read_name_1): New decl.
31361 (md_reader::m_compact): New field.
31362 (noop_reader::noop_reader): Pass "false" for new "compact" param
31364 (rtx_reader::rtx_reader): Add new "compact" param.
31365 (rtx_reader::read_rtx_operand): Make virtual and convert return
31366 type from void to rtx.
31367 (rtx_reader::read_until): New decl.
31368 (rtx_reader::handle_any_trailing_information): New virtual function.
31369 (rtx_reader::postprocess): New virtual function.
31370 (rtx_reader::finalize_string): New virtual function.
31371 (rtx_reader::m_in_call_function_usage): New field.
31372 (rtx_reader::m_reuse_rtx_by_id): New field.
31373 * read-rtl-function.c: New file.
31374 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31375 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31376 (selftest::verify_three_block_rtl_cfg): New decl.
31377 * read-rtl-function.h: New file.
31378 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31379 For host, include function.h, memmodel.h, and emit-rtl.h.
31380 (one_time_initialization): New function.
31381 (struct compact_insn_name): New struct.
31382 (compact_insn_names): New array.
31383 (find_code): Handle insn codes in compact dumps.
31384 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31385 (bind_subst_iter_and_attr): Likewise.
31386 (add_condition_to_string): Likewise.
31387 (add_condition_to_rtx): Likewise.
31388 (apply_attribute_uses): Likewise.
31389 (add_current_iterators): Likewise.
31390 (apply_iterators): Likewise.
31391 (initialize_iterators): Guard usage of apply_subst_iterator with
31392 #ifdef GENERATOR_FILE.
31393 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31394 (md_reader::read_mapping): Likewise.
31395 (add_define_attr_for_define_subst): Likewise.
31396 (add_define_subst_attr): Likewise.
31397 (read_subst_mapping): Likewise.
31398 (check_code_iterator): Likewise.
31399 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31401 (one_time_initialization): New function.
31402 (rtx_reader::read_until): New method.
31403 (read_flags): New function.
31404 (parse_reg_note_name): New function.
31405 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31406 Handle reuse_rtx ids.
31407 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31408 Add parsing support for RTL dumps, mirroring the special-cases in
31409 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31410 values, and calling handle_any_trailing_information.
31411 (rtx_reader::read_rtx_operand): Convert return type from void
31412 to rtx, returning return_rtx. Handle case 'e'. Call
31413 finalize_string on XSTR and XTMPL fields.
31414 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31415 "(nil)" values were omitted. Call the postprocess vfunc on the
31417 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31418 class ctor. Initialize m_in_call_function_usage. Call
31419 one_time_initialization.
31420 * rtl-tests.c (selftest::test_uncond_jump): Call
31421 set_new_first_and_last_insn.
31422 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31423 * selftest-rtl.c: New file.
31424 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31425 (selftest::get_insn_by_uid): New decl.
31426 * selftest-run-tests.c (selftest::run_tests): Call
31427 read_rtl_function_c_tests.
31428 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31429 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31432 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31434 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31435 operands in a special way. Assert that pos+len <= mode precision.
31437 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31439 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31440 3 argument Alias with unlimited for the negative form.
31441 (fno-vect-cost-model): Removed.
31443 2017-01-05 Martin Liska <mliska@suse.cz>
31445 * hsa-gen.c (gen_hsa_divmod): New function.
31446 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31448 2017-01-05 Martin Liska <mliska@suse.cz>
31451 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31454 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31456 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31457 small constant length operands.
31459 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31461 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31462 between loop iterations.
31464 2017-01-05 Martin Liska <mliska@suse.cz>
31467 * gimplify.c (gimplify_decl_expr): Compare to
31468 asan_poisoned_variables instread of checking flags.
31469 (gimplify_target_expr): Likewise.
31470 (gimplify_expr): Likewise.
31471 (gimplify_function_tree): Conditionally initialize
31472 asan_poisoned_variables.
31474 2017-01-04 Jeff Law <law@redhat.com>
31476 PR tree-optimizatin/78812
31477 * rtl.h (contains_mem_rtx_p): Prototype.
31478 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31479 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31480 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31481 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31482 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31484 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31486 * input.c (assert_char_at_range): Default-initialize actual_range.
31488 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31490 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31491 to match the caller.
31493 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31495 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31496 insns after final jump in test to emit dummy move.
31498 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31500 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31501 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31503 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31505 * multiple_target.c (create_dispatcher_calls): Init e_next.
31506 * tree-ssa-loop-split.c (split_loop): Init border.
31507 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31510 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31515 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31516 (altivec_register_operand): Do not return true if the operand
31517 contains a SUBREG mixing SImode and SFmode.
31518 (vsx_register_operand): Likewise.
31519 (vsx_reg_sfsubreg_ok): New predicate.
31520 (vfloat_operand): Do not return true if the operand contains a
31521 SUBREG mixing SImode and SFmode.
31522 (vint_operand): Likewise.
31523 (vlogical_operand): Likewise.
31524 (gpc_reg_operand): Likewise.
31525 (int_reg_operand): Likewise.
31526 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
31527 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
31528 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
31530 (rs6000_emit_move_si_sf_subreg): New helper function.
31531 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
31532 fixup SUBREGs involving SImode and SFmode.
31533 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
31534 numbers for the new peephole2 optimization.
31535 (peephole2 for SFmode unions): New peephole2 to optimize cases in
31536 the GLIBC math library that do AND/IOR/XOR operations on single
31537 precision floating point.
31538 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
31539 target macros to say whether we need to avoid SUBREGs mixing
31541 (TARGET_ALLOW_SF_SUBREG): Likewise.
31542 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
31543 (UNSPEC_SI_FROM_SF): Likewise.
31544 (iorxor): Change spacing.
31545 (and_ior_xor): New iterator for AND, IOR, and XOR.
31546 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
31547 (movdi_from_sf_zero_ext): Likewise.
31548 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
31549 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
31550 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
31551 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
31552 (fms<mode>4): Likewise.
31553 (fnma<mode>4): Likewise.
31554 (fnms<mode>4): Likewise.
31555 (nfma<mode>4): Likewise.
31556 (nfms<mode>4): Likewise.
31558 2017-01-04 Marek Polacek <polacek@redhat.com>
31561 * doc/invoke.texi: Document -Wpointer-compare.
31563 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31565 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
31568 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
31569 descriptions for -gdwarf-5 and emit them as uleb128 instead of
31572 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
31575 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
31576 documentation of the powerpc_popcntb_ok attribute.
31577 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31578 code to issue warning messages if a requested CPU configuration is
31579 not supported by the binary (assembler and loader) toolchain.
31580 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
31581 made to define a built-in function that has been disabled.
31582 (paired_init_builtins): Add assertion to prevent ICE if attempt is
31583 made to define a built-in function that has been disabled.
31584 (altivec_init_builtins): Add comment explaining why definition
31585 of the DST built-in functions is not preceded by an assertion
31586 check. Add assertions to prevent ICE if attempts are made to
31587 define an altivec predicate or an abs* built-in function that has
31589 (htm_init_builtins): Add comment explaining why definition of the
31590 htm built-in functions is not preceded by an assertion check.
31592 2017-01-04 Jeff Law <law@redhat.com>
31594 PR tree-optimizatin/67955
31595 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
31596 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
31597 the points-to solution does not include pt_null. Use DECL_PT_UID
31600 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
31602 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31603 Use gen_int_mode instead of gen_lopwart for const_int operands.
31605 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31607 PR tree-optimization/71563
31608 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31609 out of range value - has low bits known to be zero.
31611 2017-01-04 Alan Modra <amodra@gmail.com>
31613 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31614 * configure: Regenerate.
31615 * config.in: Regenerate.
31617 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31620 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31621 a substring of the message, but strcmp with the whole message. Ifdef
31622 ENABLE_NLS, translate the message first using dgettext.
31624 2017-01-03 Jeff Law <law@redhat.com>
31626 PR tree-optimizatin/78856
31627 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31628 (mark_threaded_blocks): Remove code to truncate thread paths that
31629 cross multiple loop headers. Instead invalidate the cached loop
31630 iteration information and handle case of a thread path walking
31631 into an irreducible region.
31633 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31636 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31637 assertions. Add support for doing the signbit if the IEEE 128-bit
31638 floating point value is in a GPR.
31639 * config/rs6000/rs6000.md (Fsignbit): Delete.
31640 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31641 Update the length attribute if the value is in a GPR.
31642 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31643 the sign or zero extension instruction, since the value is always 0/1.
31644 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31647 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31648 extracting SImode to a GPR register so that we can generate a
31649 store, limit the vector to be in a traditional Altivec register
31650 for the vextuwrx instruction.
31652 2017-01-03 Ian Lance Taylor <iant@google.com>
31654 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31656 2017-01-03 Martin Sebor <msebor@redhat.com>
31658 PR tree-optimization/78696
31659 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31660 precision. Use MPFR for %f for greater fidelity. Correct handling
31662 (pass_sprintf_length::compute_format_length): Set width and precision
31663 specified by asrerisk to void_node for vararg functions.
31664 (try_substitute_return_value): Adjust dump output.
31666 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31668 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31670 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31672 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31673 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31674 -mlra/-mno-lra was passed to the compiler.
31676 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31678 PR rtl-optimization/65618
31679 * emit-rtl.c (try_split): Move initialization of "before" and
31680 "after" to just before the call to emit_insn_after_setloc.
31682 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31684 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31686 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31688 * dwarf2out.c (gen_enumeration_type_die): When
31689 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31691 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31693 PR tree-optimization/78965
31694 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31695 Change first argument from const call_info & to call_info &. For %n
31696 set info.nowrite to false.
31698 PR middle-end/78901
31699 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31700 possibly throwing calls.
31702 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31703 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31704 and fns handling, rather than in a separate case SSA_NAME.
31706 2017-01-02 Jeff Law <law@redhat.com>
31708 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31709 fixes for first_period and second_period variables.
31711 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31714 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31715 (*insvqi_1): New insn pattern.
31716 (*insvqi_1_mem_rex64): Ditto.
31717 (*insvqi_2): Ditto.
31718 (*insvqi_3): Rename from *insvqi.
31720 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31722 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31724 * doc/cfg.texi (Edges): Remove reference to Java.
31725 (Maintaining the CFG): Ditto.
31727 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31729 PR middle-end/77674
31730 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31731 transparent aliases.
31733 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31735 PR middle-end/77484
31736 * predict.def (PRED_CALL): Update hitrate.
31737 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31738 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31739 into direct/indirect/polymorphic variants.
31741 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31743 Update copyright years.
31745 * gcc.c (process_command): Update copyright notice dates.
31746 * gcov-dump.c (print_version): Ditto.
31747 * gcov.c (print_version): Ditto.
31748 * gcov-tool.c (print_version): Ditto.
31749 * gengtype.c (create_file): Ditto.
31750 * doc/cpp.texi: Bump @copying's copyright year.
31751 * doc/cppinternals.texi: Ditto.
31752 * doc/gcc.texi: Ditto.
31753 * doc/gccint.texi: Ditto.
31754 * doc/gcov.texi: Ditto.
31755 * doc/install.texi: Ditto.
31756 * doc/invoke.texi: Ditto.
31758 Copyright (C) 2017 Free Software Foundation, Inc.
31760 Copying and distribution of this file, with or without modification,
31761 are permitted in any medium without royalty provided the copyright
31762 notice and this notice are preserved.