1 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
4 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
5 calls of avr_out_plus_1.
7 2014-03-13 Bin Cheng <bin.cheng@arm.com>
9 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
10 BB's single pred and update the father loop's latch info later.
12 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
19 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
20 registers, we need to swap double words in little endian mode.
22 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
23 to be a container mode for 128-bit integer operations added in ISA
24 2.07. Unlike TImode and PTImode, the preferred register set is
25 the Altivec/VMX registers for the 128-bit operations.
27 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
29 (rs6000_split_128bit_ok_p): Likewise.
31 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
32 macros for creating ISA 2.07 normal and overloaded builtin
33 functions with 3 arguments.
34 (BU_P8V_OVERLOAD_3): Likewise.
35 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
36 for use as overloaded functions.
37 (VPERM_1TI_UNS): Likewise.
39 (VSEL_1TI_UNS): Likewise.
40 (ST_INTERNAL_1ti): Likewise.
41 (LD_INTERNAL_1ti): Likewise.
42 (XXSEL_1TI): Likewise.
43 (XXSEL_1TI_UNS): Likewise.
44 (VPERM_1TI): Likewise.
45 (VPERM_1TI_UNS): Likewise.
46 (XXPERMDI_1TI): Likewise.
48 (LXVD2X_V1TI): Likewise.
49 (STXVD2X_V1TI): Likewise.
50 (VEC_INIT_V1TI): Likewise.
51 (VEC_SET_V1TI): Likewise.
52 (VEC_EXT_V1TI): Likewise.
54 (NAND_V1TI): Likewise.
56 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
57 added in ISA 2.07. Add both normal 'altivec' builtins, and the
66 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
67 __int128_t and __uint128_t types.
68 (__uint128_type): Likewise.
69 (altivec_categorize_keyword): Add support for vector __int128_t,
70 vector __uint128_t, vector __int128, and vector unsigned __int128
71 as a container type for TImode operations that need to be done in
72 VSX/Altivec registers.
73 (rs6000_macro_to_expand): Likewise.
74 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
75 to support 128-bit integer instructions vaddcuq, vadduqm,
76 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
77 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
79 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
80 for V1TImode, and set up preferences to use VSX/Altivec registers.
81 Setup VSX reload handlers.
82 (rs6000_debug_reg_global): Likewise.
83 (rs6000_init_hard_regno_mode_ok): Likewise.
84 (rs6000_preferred_simd_mode): Likewise.
85 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
86 (easy_altivec_constant): Likewise.
87 (output_vec_const_move): Likewise.
88 (rs6000_expand_vector_set): Convert V1TImode set and extract to
90 (rs6000_expand_vector_extract): Likewise.
91 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
93 (rs6000_const_vec): Add support for V1TImode.
94 (rs6000_emit_le_vsx_load): Swap double words when loading or
95 storing TImode/V1TImode.
96 (rs6000_emit_le_vsx_store): Likewise.
97 (rs6000_emit_le_vsx_move): Likewise.
98 (rs6000_emit_move): Add support for V1TImode.
99 (altivec_expand_ld_builtin): Likewise.
100 (altivec_expand_st_builtin): Likewise.
101 (altivec_expand_vec_init_builtin): Likewise.
102 (altivec_expand_builtin): Likewise.
103 (rs6000_init_builtins): Add support for V1TImode type. Add
104 support for ISA 2.07 128-bit integer builtins. Define type names
105 for the VSX/Altivec vector types.
106 (altivec_init_builtins): Add support for overloaded vector
107 functions with V1TImode type.
108 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
109 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
111 (rs6000_split_128bit_ok_p): Likewise.
112 (rs6000_handle_altivec_attribute): Create V1TImode from vector
113 __int128_t and vector __uint128_t.
115 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
123 (VS_scalar): Likewise.
124 (VS_double): Likewise.
125 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
127 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
128 we support the ISA 2.07 128-bit integer arithmetic instructions.
129 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
130 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
131 and TImode types for use with the builtin functions.
132 (V1TI_type_node): Likewise.
133 (unsigned_V1TI_type_node): Likewise.
134 (intTI_type_internal_node): Likewise.
135 (uintTI_type_internal_node): Likewise.
137 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
138 128-bit builtin functions.
139 (UNSPEC_VADDEUQM): Likewise.
140 (UNSPEC_VADDECUQ): Likewise.
141 (UNSPEC_VSUBCUQ): Likewise.
142 (UNSPEC_VSUBEUQM): Likewise.
143 (UNSPEC_VSUBECUQ): Likewise.
144 (VM): Add V1TImode to vector mode iterators.
147 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
148 (altivec_vaddcuq): Likewise.
149 (altivec_vsubuqm): Likewise.
150 (altivec_vsubcuq): Likewise.
151 (altivec_vaddeuqm): Likewise.
152 (altivec_vaddecuq): Likewise.
153 (altivec_vsubeuqm): Likewise.
154 (altivec_vsubecuq): Likewise.
156 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
158 (BOOL_128): Likewise.
159 (BOOL_REGS_OUTPUT): Likewise.
160 (BOOL_REGS_OP1): Likewise.
161 (BOOL_REGS_OP2): Likewise.
162 (BOOL_REGS_UNARY): Likewise.
163 (BOOL_REGS_AND_CR0): Likewise.
165 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
166 128-bit integer builtin support.
167 (vec_vadduqm): Likewise.
168 (vec_vaddecuq): Likewise.
169 (vec_vaddeuqm): Likewise.
170 (vec_vsubecuq): Likewise.
171 (vec_vsubeuqm): Likewise.
172 (vec_vsubcuq): Likewise.
173 (vec_vsubuqm): Likewise.
175 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
176 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
177 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
178 128-bit integer add/subtract to ISA 2.07.
180 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
182 * config/arc/arc.c (arc_predicate_delay_insns):
183 Fix third argument passed to conditionalize_nonjump.
185 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
187 * config/aarch64/aarch64-builtins.c
188 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
189 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
190 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
191 instead of __builtin_lfloor.
192 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
194 2014-03-12 Jakub Jelinek <jakub@redhat.com>
196 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
197 (tree_ssa_ifcombine_bb_1): New function.
198 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
199 is an empty forwarder block to then_bb or vice versa and then_bb
200 and else_bb are effectively swapped.
202 2014-03-12 Christian Bruel <christian.bruel@st.com>
205 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
206 REG_CFA_DEF_CFA note.
207 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
208 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
210 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
212 PR tree-optimization/60454
213 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
215 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
217 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
218 Do not define target_cpu_default2 to generic.
219 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
220 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
221 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
223 2014-03-12 Jakub Jelinek <jakub@redhat.com>
224 Marc Glisse <marc.glisse@inria.fr>
226 PR tree-optimization/60502
227 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
228 instead of build_low_bits_mask.
230 2014-03-12 Jakub Jelinek <jakub@redhat.com>
233 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
234 if there are multiple uses, but op doesn't live on E edge.
235 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
236 clobber stmts before __builtin_unreachable.
238 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
240 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
241 hard_frame_pointer_rtx.
242 * cse.c (cse_insn): Remove volatile check.
243 * cselib.c (cselib_process_insn): Likewise.
244 * dse.c (scan_insn): Likewise.
246 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
248 * config/arc/arc.c (conditionalize_nonjump): New function,
250 (arc_ifcvt): ... this.
251 (arc_predicate_delay_insns): Use it.
253 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
255 * config/arc/predicates.md (extend_operand): During/after reload,
256 allow const_int_operand.
257 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
258 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
259 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
261 (umulsi3_highpart_i): Likewise.
263 2014-03-11 Richard Biener <rguenther@suse.de>
265 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
266 Add asserts to guard possible wrong-code bugs.
268 2014-03-11 Richard Biener <rguenther@suse.de>
270 PR tree-optimization/60429
271 PR tree-optimization/60485
272 * tree-ssa-structalias.c (set_union_with_increment): Properly
273 take into account all fields that overlap the shifted vars.
274 (do_sd_constraint): Likewise.
275 (do_ds_constraint): Likewise.
276 (get_constraint_for_ptr_offset): Likewise.
278 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
280 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
281 (nios2_compute_frame_layout):
282 Add calculation of cfun->machine->fp_save_offset.
283 (nios2_expand_prologue): Correct setting of frame pointer register
285 (nios2_expand_epilogue): Update recovery of stack pointer from
286 frame pointer accordingly.
287 (nios2_initial_elimination_offset): Update calculation of offset
288 for eliminating to HARD_FRAME_POINTER_REGNUM.
290 2014-03-10 Jakub Jelinek <jakub@redhat.com>
293 * ipa.c (symtab_remove_unreachable_nodes): Don't call
294 cgraph_get_create_node on VAR_DECLs.
296 2014-03-10 Richard Biener <rguenther@suse.de>
299 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
301 2014-03-08 Douglas B Rupp <rupp@gnat.com>
303 * config/vms/vms.opt (vms_float_format): New variable.
305 2014-03-08 Tobias Burnus <burnus@net-b.de>
307 * doc/invoke.texi (-fcilkplus): Update implementation status.
309 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
310 Richard Biener <rguenther@suse.de>
312 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
313 consistently accross all TUs.
314 (run_gcc): Enable -fshort-double automatically at link at link-time
315 and disallow override.
317 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
320 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
321 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
322 if they can't be used.
324 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
326 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
327 for Solaris 11/x86 ld.
328 * configure: Regenerate.
330 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
332 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
333 (LIB_TLS_SPEC): Save as ld_tls_libs.
334 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
335 (HAVE_AS_IX86_TLSLDM): New test.
336 * configure, config.in: Regenerate.
337 * config/i386/i386.c (legitimize_tls_address): Fall back to
338 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
339 cannot support TLS_MODEL_LOCAL_DYNAMIC.
340 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
341 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
343 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
345 * common.opt (fira-loop-pressure): Mark as optimization.
347 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
349 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
350 an OpenMP mappable type.
352 2014-03-06 Matthias Klose <doko@ubuntu.com>
354 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
355 MULTILIB_OSDIRNAMES is not defined.
357 2014-03-06 Jakub Jelinek <jakub@redhat.com>
358 Meador Inge <meadori@codesourcery.com>
361 * config/arm/arm.c (arm_tls_symbol_p): Remove.
362 (arm_legitimize_address): Call legitimize_tls_address for any
363 arm_tls_referenced_p expression, handle constant addend. Call it
364 before testing for !TARGET_ARM.
365 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
367 2014-03-06 Richard Biener <rguenther@suse.de>
373 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
375 * tree-streamer.c (record_common_node): Assert we don't record
376 nodes with type double.
377 (preload_common_node): Skip type double, complex double and double
378 pointer since it is now frontend dependent due to fshort-double option.
380 2014-03-06 Richard Biener <rguenther@suse.de>
382 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
383 or -fno-lto is specified and the linker has full plugin support.
384 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
385 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
386 * lto-wrapper.c (merge_and_complain): Merge compile-time
388 (run_gcc): And pass it through to the link options.
390 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
394 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
396 * cselib.c (remove_useless_values): Skip to avoid quadratic
397 behavior if the condition moved from...
398 (cselib_process_insn): ... here holds.
400 2014-03-05 Jakub Jelinek <jakub@redhat.com>
403 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
404 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
407 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
408 (TM_H): Add x86-tune.def.
410 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
412 * config/aarch64/aarch64.c (generic_tunings):
413 Use cortexa57_extra_costs.
415 2014-03-05 Jakub Jelinek <jakub@redhat.com>
418 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
419 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
420 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
423 2014-03-04 Heiher <r@hev.cc>
425 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
426 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
428 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
430 * config/i386/predicates.md (const2356_operand): Change to ...
431 (const2367_operand): ... this.
432 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
434 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
435 (*avx512pf_scatterpf<mode>sf): Ditto.
436 (avx512pf_scatterpf<mode>df): Ditto.
437 (*avx512pf_scatterpf<mode>df_mask): Ditto.
438 (*avx512pf_scatterpf<mode>df): Ditto.
439 * config/i386/i386.c (ix86_expand_builtin): Update
440 incorrect hint operand error message.
442 2014-03-04 Richard Biener <rguenther@suse.de>
444 * lto-section-in.c (lto_get_section_data): Fix const cast.
446 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
448 * tree-streamer.c (record_common_node): Assert we don't record
449 nodes with type double.
450 (preload_common_node): Skip type double, complex double and double
451 pointer since it is now frontend dependent due to fshort-double option.
453 2014-03-04 Richard Biener <rguenther@suse.de>
456 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
457 (lto_input_toplevel_asms): Likewise.
458 * lto-section-in.c (lto_get_section_data): Instead do it here
461 2014-03-04 Richard Biener <rguenther@suse.de>
463 PR tree-optimization/60382
464 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
465 dead PHIs a reduction.
467 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
469 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
471 (_mm_prefetch): Move out of GCC target("sse") pragma.
472 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
473 GCC target("prfchw") pragma.
474 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
476 * config/i386/i386.c (ix86_option_override_internal): Enable
477 -mprfchw with -mprefetchwt1.
479 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
481 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
484 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
486 * opts.h (CL_PCH_IGNORE): Define.
487 * targhooks.c (option_affects_pch_p):
488 Return false for options that have CL_PCH_IGNORE set.
489 * opt-functions.awk: Process PchIgnore.
490 * doc/options.texi: Document PchIgnore.
492 * config/arc/arc.opt (misize): Add PchIgnore property.
494 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
496 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
497 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
498 constraint on constants to permit them being loaded into
499 GENERAL_REGS or BASE_REGS.
501 2014-03-03 Nick Clifton <nickc@redhat.com>
503 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
504 anti-cacnonical alternatives.
505 (negandhi3_real): New pattern.
506 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
508 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
510 * config/avr/avr-mcus.def: Remove atxmega16x1.
511 * config/avr/avr-tables.opt: Regenerate.
512 * config/avr/t-multilib: Regenerate.
513 * doc/avr-mmcu.texi: Regenerate.
515 2014-03-03 Tobias Grosser <tobias@grosser.es>
516 Mircea Namolaru <mircea.namolaru@inria.fr>
518 PR tree-optimization/58028
519 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
522 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
524 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
525 not handled by recognizers.
527 2014-03-03 Jakub Jelinek <jakub@redhat.com>
530 * function.c (expand_function_end): Don't emit
531 clobber_return_register sequence if clobber_after is a BARRIER.
532 * cfgexpand.c (construct_exit_block): Append instructions before
533 return_label to prev_bb.
535 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
537 * config/rs6000/constraints.md: Document reserved use of "wc".
539 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
542 * ipa.c (function_and_variable_visibility): When dissolving comdat
543 group, also set all symbols to local.
545 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
550 2013-12-14 Jan Hubicka <jh@suse.cz>
552 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
554 2014-03-02 Jon Beniston <jon@beniston.com>
560 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
561 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
562 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
563 (simple_return, *simple_return): New patterns
564 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
565 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
567 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
569 * dwarf2out.c (gen_subprogram_die): Tidy.
571 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
574 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
575 (*mov_t_msb_neg_negc): ... this new insn.
577 2014-02-28 Jason Merrill <jason@redhat.com>
580 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
583 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
586 * dwarf2out.c (decltype_auto_die): New static.
587 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
588 (gen_type_die_with_usage): Handle 'decltype(auto)'.
589 (is_cxx_auto): Likewise.
591 2014-02-28 Ian Bolton <ian.bolton@arm.com>
593 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
594 we are not using general regs only.
596 2014-02-28 Richard Biener <rguenther@suse.de>
599 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
600 previous fix and only allow to remove trivial pre-headers
601 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
602 (remove_forwarder_block): Properly update the latch of a loop.
604 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
607 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
608 (cselib_preserved_hash_table): New.
609 (preserve_constants_and_equivs): Move preserved vals to it.
610 (cselib_find_slot): Look it up first.
611 (cselib_init): Initialize it.
612 (cselib_finish): Release it.
613 (dump_cselib_table): Dump it.
615 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
618 * cselib.c (remove_useless_values): Skip to avoid quadratic
619 behavior if the condition moved from...
620 (cselib_process_insn): ... here holds.
622 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
625 * var-tracking.c (vt_initialize): Apply the same condition to
626 preserve the CFA base value.
628 2014-02-28 Joey Ye <joey.ye@arm.com>
631 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
632 if reload in progress or completed.
634 2014-02-28 Tobias Burnus <burnus@net-b.de>
637 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
640 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
642 * doc/tm.texi.in (Condition Code Status): Update documention for
643 relative locations of cc0-setter and cc0-user.
645 2014-02-27 Jeff Law <law@redhat.com>
647 PR rtl-optimization/52714
648 * combine.c (try_combine): When splitting an unrecognized PARALLEL
649 into two independent simple sets, if I3 is a jump, ensure the
650 pattern we place into I3 is a (set (pc) ...).
652 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
653 Jeff Law <law@redhat.com>
655 PR rtl-optimization/49847
656 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
657 are in different blocks.
658 * doc/tm.texi (Condition Code Status): Update documention for
659 relative locations of cc0-setter and cc0-user.
661 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
664 * lra.c (lra_emit_add): Check SUBREG too.
666 2014-02-27 Andreas Schwab <schwab@suse.de>
668 * config/m68k/m68k.c (m68k_option_override): Disable
669 -flive-range-shrinkage for classic m68k.
670 (m68k_override_options_after_change): Likewise.
672 2014-02-27 Marek Polacek <polacek@redhat.com>
675 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
676 -Wmaybe-uninitialized.
678 2014-02-27 Alan Modra <amodra@gmail.com>
681 * reload1.c (emit_input_reload_insns): When reload_override_in,
682 set old to rl->in_reg when rl->in_reg is a subreg.
684 2014-02-26 Richard Biener <rguenther@suse.de>
687 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
689 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
691 * common/config/i386/predicates.md (const1256_operand): Remove.
692 (const2356_operand): New.
693 (const_1_to_2_operand): Remove.
694 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
695 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
696 (*avx512pf_gatherpf<mode>sf): Ditto.
697 (avx512pf_gatherpf<mode>df): Ditto.
698 (*avx512pf_gatherpf<mode>df_mask): Ditto.
699 (*avx512pf_gatherpf<mode>df): Ditto.
700 (avx512pf_scatterpf<mode>sf): Ditto.
701 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
702 (*avx512pf_scatterpf<mode>sf): Ditto.
703 (avx512pf_scatterpf<mode>df): Ditto.
704 (*avx512pf_scatterpf<mode>df_mask): Ditto.
705 (*avx512pf_scatterpf<mode>df): Ditto.
706 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
708 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
710 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
711 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
712 (_mm512_mask_testn_epi64_mask): Move to ...
713 * config/i386/avx512cdintrin.h: Here.
714 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
715 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
716 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
717 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
718 TARGET_AVX512F from TARGET_AVX512CD.
720 2014-02-26 Richard Biener <rguenther@suse.de>
723 * ipa.c (walk_polymorphic_call_targets): Properly guard
724 call to inline_update_overall_summary.
726 2014-02-26 Bin Cheng <bin.cheng@arm.com>
729 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
730 and latches only if requested. Fix latch if it is removed.
731 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
732 LOOPS_HAVE_PREHEADERS.
734 2014-02-25 Andrew Pinski <apinski@cavium.com>
736 * builtins.c (expand_builtin_thread_pointer): Create a new target
737 when the target is NULL.
739 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
741 PR rtl-optimization/60317
742 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
743 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
744 * lra-assigns.c: Include params.h.
745 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
746 other reload pseudos considerations.
748 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
750 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
751 to use canonical form for nor<mode>3.
753 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
756 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
759 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
761 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
762 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
763 (ix86_handle_option): Handle OPT_mprefetchwt1.
764 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
765 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
767 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
768 OPTION_MASK_ISA_PREFETCHWT1.
769 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
770 (PTA_PREFETCHWT1): New.
771 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
772 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
773 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
774 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
775 (*prefetch_avx512pf_<mode>_: Change into ...
776 (*prefetch_prefetchwt1_<mode>: This.
777 * config/i386/i386.opt (mprefetchwt1): New.
778 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
779 (_mm_prefetch): Handle intent to write.
780 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
782 2014-02-25 Richard Biener <rguenther@suse.de>
785 * emit-rtl.c (mem_attrs_htab): Remove.
786 (mem_attrs_htab_hash): Likewise.
787 (mem_attrs_htab_eq): Likewise.
788 (set_mem_attrs): Always allocate new mem-attrs when something changed.
789 (init_emit_once): Do not allocate mem_attrs_htab.
791 2014-02-25 Richard Biener <rguenther@suse.de>
794 * lto-opts.c (lto_write_options): Output non-explicit conservative
795 -fwrapv, -fno-trapv and -fno-strict-overflow.
796 * lto-wrapper.c (merge_and_complain): Handle merging those options.
797 (run_gcc): And pass them through.
799 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
801 * sel-sched.c (calculate_new_fences): New parameter ptime.
802 Calculate it as a maximum over all fence cycles.
803 (sel_sched_region_2): Adjust the call to calculate_new_fences.
804 Print the final schedule timing when sched_verbose.
806 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
808 PR rtl-optimization/60292
809 * sel-sched.c (fill_vec_av_set): Do not reset target availability
810 bit fot the fence instruction.
812 2014-02-24 Alangi Derick <alangiderick@gmail.com>
814 * calls.h: Fix typo in comment.
816 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
818 * config/pa/pa.c (pa_output_move_double): Don't valididate when
819 adjusting offsetable addresses.
821 2014-02-24 Guozhi Wei <carrot@google.com>
823 * sparseset.h (sparseset_pop): Fix the wrong index.
825 2014-02-24 Walter Lee <walt@tilera.com>
827 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
828 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
830 * common/config/tilegx/tilegx-common.c
831 (TARGET_DEFAULT_TARGET_FLAGS): Define.
832 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
834 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
835 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
836 (tilegx_gimplify_va_arg_expr): Handle big endian.
837 (tilegx_expand_unaligned_load): Ditto.
838 (tilegx_expand_unaligned_store): Ditto.
839 (TARGET_RETURN_IN_MSB): New.
840 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
841 (TARGET_ENDIAN_DEFAULT): New.
842 (TARGET_BIG_ENDIAN): Handle big endian.
843 (BYTES_BIG_ENDIAN): Ditto.
844 (WORDS_BIG_ENDIAN): Ditto.
845 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
848 * config/tilegx/tilegx.md (extv): Handle big endian.
851 (insn_st<n>_add<bitsuffix>): Ditto.
852 (insn_stnt<n>): Ditto.
853 (insn_stnt<n>_add<bitsuffix>):Ditto.
854 (vec_interleave_highv8qi): Handle big endian.
855 (vec_interleave_highv8qi_be): New.
856 (vec_interleave_highv8qi_le): New.
857 (insn_v1int_h): Handle big endian.
858 (vec_interleave_lowv8qi): Handle big endian.
859 (vec_interleave_lowv8qi_be): New.
860 (vec_interleave_lowv8qi_le): New.
861 (insn_v1int_l): Handle big endian.
862 (vec_interleave_highv4hi): Handle big endian.
863 (vec_interleave_highv4hi_be): New.
864 (vec_interleave_highv4hi_le): New.
865 (insn_v2int_h): Handle big endian.
866 (vec_interleave_lowv4hi): Handle big endian.
867 (vec_interleave_lowv4hi_be): New.
868 (vec_interleave_lowv4hi_le): New.
869 (insn_v2int_l): Handle big endian.
870 (vec_interleave_highv2si): Handle big endian.
871 (vec_interleave_highv2si_be): New.
872 (vec_interleave_highv2si_le): New.
873 (insn_v4int_h): Handle big endian.
874 (vec_interleave_lowv2si): Handle big endian.
875 (vec_interleave_lowv2si_be): New.
876 (vec_interleave_lowv2si_le): New.
877 (insn_v4int_l): Handle big endian.
878 * config/tilegx/tilegx.opt (mbig-endian): New option.
879 (mlittle-endian): New option.
880 * doc/install.texi: Document tilegxbe-linux.
881 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
883 2014-02-24 Martin Jambor <mjambor@suse.cz>
886 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
887 there are no parameter descriptors.
889 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
891 PR rtl-optimization/60268
892 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
893 initialization to ...
894 (sched_rgn_init): ... here.
895 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
897 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
899 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
902 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
904 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
907 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
909 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
910 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
912 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
914 * config/microblaze/predicates.md: Add cmp_op predicate.
915 * config/microblaze/microblaze.md: Add branch_compare instruction
916 which uses cmp_op predicate and emits cmp insn before branch.
917 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
918 to microblaze_expand_conditional_branch and consolidate logic.
919 (microblaze_expand_conditional_branch): emit branch_compare
920 insn instead of handling cmp op separate from branch insn.
922 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
924 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
927 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
929 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
930 define_insn with define_expand and new define_insn
931 *altivec_lve<VI_char>x_internal.
932 (altivec_stve<VI_char>x): Replace define_insn with define_expand
933 and new define_insn *altivec_stve<VI_char>x_internal.
934 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
936 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
938 (altivec_expand_stvex_be): New function.
940 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
942 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
943 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
944 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
945 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
947 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
950 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
951 instead of emit_move_insn.
953 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
955 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
957 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
960 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
962 * config/rs6000/altivec.md (altivec_lvxl): Rename as
963 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
964 (altivec_lvxl_<mode>): New define_expand incorporating
965 -maltivec=be semantics where needed.
966 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
967 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
968 semantics where needed.
969 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
970 (altivec_stvx_<mode>): New define_expand incorporating
971 -maltivec=be semantics where needed.
972 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
973 VM2 iterator instead of V4SI.
974 (altivec_stvxl_<mode>): New define_expand incorporating
975 -maltivec=be semantics where needed.
976 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
977 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
978 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
979 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
980 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
981 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
982 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
983 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
984 ALTIVEC_BUILTIN_STVXL.
985 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
986 (altivec_expand_stvx_be): Likewise.
987 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
988 (altivec_expand_lvx_be): Likewise.
989 (altivec_expand_stvx_be): Likewise.
990 (altivec_expand_builtin): Add cases for
991 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
992 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
993 (altivec_init_builtins): Add definitions for
994 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
995 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
997 2014-02-21 Catherine Moore <clm@codesourcery.com>
999 * doc/invoke.texi (mvirt, mno-virt): Document.
1000 * config/mips/mips.opt (mvirt): New option.
1001 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1003 2014-02-21 Richard Biener <rguenther@suse.de>
1005 PR tree-optimization/60276
1006 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1007 (STMT_VINFO_MIN_NEG_DIST): New macro.
1008 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1009 STMT_VINFO_MIN_NEG_DIST.
1010 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1011 made for negative dependence distances still hold.
1013 2014-02-21 Richard Biener <rguenther@suse.de>
1016 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1017 DECL_INITIAL for globals not in the current function context.
1019 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1021 PR tree-optimization/56490
1022 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1023 * tree-ssa-uninit.c: Include params.h.
1024 (compute_control_dep_chain): Add num_calls argument, return false
1025 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1026 num_calls to recursive call.
1027 (find_predicates): Change dep_chain into normal array,
1028 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1029 variable and adjust compute_control_dep_chain caller.
1030 (find_def_preds): Likewise.
1032 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1034 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1035 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1037 2014-02-21 Nick Clifton <nickc@redhat.com>
1039 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1040 (pushhi1): Likewise.
1041 (popqi1): Add mode to pre_dec.
1044 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1046 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1047 mode for mask of V8SFmode permutation.
1049 2014-02-20 Richard Henderson <rth@redhat.com>
1052 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1053 a new pseudo for OLDVAL.
1055 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1058 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1059 gen_reg_rtx if d->testing_p.
1060 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1061 if d->testing_p and we will certainly return true.
1062 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1065 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1067 * emit-rtl.c (gen_reg_rtx): Assert that
1068 crtl->emit.regno_pointer_align_length is non-zero.
1070 2014-02-20 Richard Henderson <rth@redhat.com>
1073 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1074 on failure the store back into EXPECT.
1076 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1077 Sandra Loosemore <sandra@codesourcery.com>
1079 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1080 * config/nios2/nios2.c (nios2_function_profiler): Add
1081 -fPIC (flag_pic == 2) support.
1082 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1083 (nios2_large_offset_p): New function.
1084 (nios2_unspec_reloc_p): Move up position, update to use
1085 nios2_large_offset_p.
1086 (nios2_unspec_address): Remove function.
1087 (nios2_unspec_offset): New function.
1088 (nios2_large_got_address): New function.
1089 (nios2_got_address): Add large offset support.
1090 (nios2_legitimize_tls_address): Update usage of removed and new
1092 (nios2_symbol_binds_local_p): New function.
1093 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1094 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1095 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1096 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1097 processing for (const (unspec ...)).
1098 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1100 2014-02-20 Richard Biener <rguenther@suse.de>
1102 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1103 doing the substitution.
1104 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1106 2014-02-20 Martin Jambor <mjambor@suse.cz>
1109 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1110 info when checking whether lattices are bottom.
1112 2014-02-20 Richard Biener <rguenther@suse.de>
1115 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1118 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1121 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1122 parameter specifying the scaling.
1123 (inline_call): Update.
1124 (want_inline_recursively): Guard division by zero.
1125 (recursive_inlining): Update.
1126 * ipa-inline.h (clone_inlined_nodes): Update.
1128 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1131 * config/i386/i386.c (classify_argument): Pass structures of size
1132 64 bytes or less in register.
1134 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1135 Kirill Yukhin <kirill.yukhin@intel.com>
1137 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1138 (_mm_rcp28_round_ss): Ditto.
1139 (_mm_rsqrt28_round_sd): Ditto.
1140 (_mm_rsqrt28_round_ss): Ditto.
1141 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1142 (_mm_rcp14_round_ss): Ditto.
1143 (_mm_rsqrt14_round_sd): Ditto.
1144 (_mm_rsqrt14_round_ss): Ditto.
1145 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1146 the first input operand, get rid of match_dup.
1147 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1149 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1151 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1152 operand as the first input operand, set type attribute.
1153 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1155 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1156 operand as the first input operand, set type attribute.
1158 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1160 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1163 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1166 * config/i386/i386.c (construct_container): Remove TFmode check
1167 for X86_64_INTEGER_CLASS.
1169 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1172 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1173 only when -Wpsabi is enabled.
1175 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1178 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1179 passing arrays in registers are the same as for structs, so remove the
1180 special case for them.
1182 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1184 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1185 destination type, extract only the valid bits if the source type is not
1186 integral and has a different mode.
1188 2014-02-19 Richard Biener <rguenther@suse.de>
1191 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1194 2014-02-19 Richard Biener <rguenther@suse.de>
1197 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1198 (ipa_modify_call_arguments): Emit an argument load explicitely and
1199 preserve virtual SSA form there and for the replacement call.
1200 Do not update SSA form nor free dominance info.
1202 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1204 * ipa.c (function_and_variable_visibility): Also clear WEAK
1205 flag when disolving COMDAT_GROUP.
1207 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1209 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1210 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1212 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1213 do more sanity checks.
1214 (detect_type_change): Return true when giving up early.
1215 (compute_complex_assign_jump_func): Fix type parameter of
1216 ipa_set_ancestor_jf.
1217 (compute_complex_ancestor_jump_func): Likewise.
1218 (update_jump_functions_after_inlining): Fix updating of
1220 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1222 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1224 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1225 inline clones when edge disappears.
1227 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1230 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1231 Split 64-bit moves into 2 patterns. Do not allow the use of
1232 direct move for TDmode in little endian, since the decimal value
1233 has little endian bytes within a word, but the 64-bit pieces are
1234 ordered in a big endian fashion, and normal subreg's of TDmode are
1236 (mov<mode>_64bit_dm): Likewise.
1237 (movtd_64bit_nodm): Likewise.
1239 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1241 PR tree-optimization/60174
1242 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1243 statement of an SSA_NAME that occurs in an abnormal PHI node.
1245 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1248 * final.c (SEEN_BB): Remove.
1249 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1250 (final_scan_insn): Don't force_source_line on second
1251 NOTE_INSN_BASIC_BLOCK.
1253 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1256 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1257 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1258 (type_natural_mode): Warn ABI change when %zmm register is not
1259 available for AVX512F vector value passing.
1261 2014-02-18 Kai Tietz <ktietz@redhat.com>
1264 * config/i386/i386.c (ix86_expand_prologue): Use value in
1265 rax register as displacement when restoring %r10 or %rax.
1266 Fix wrong offset when restoring both registers.
1268 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1270 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1271 assertion with conditional return.
1273 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1274 Uros Bizjak <ubizjak@gmail.com>
1277 * config/i386/driver-i386.c (host_detect_local_cpu): If
1278 YMM state is not saved by the OS, also clear has_f16c. Move
1279 CPUID 0x80000001 handling before YMM state saving checking.
1281 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1283 PR rtl-optimization/58960
1284 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1285 factored out from ...
1286 (sched_init): ... here.
1287 (free_global_sched_pressure_data): New, factored out from ...
1288 (sched_finish): ... here.
1289 * sched-int.h (free_global_sched_pressure_data): Declare.
1290 * sched-rgn.c (nr_regions_initial): New static global.
1291 (haifa_find_rgns): Initialize it.
1292 (schedule_region): Disable sched-pressure for the newly
1295 2014-02-17 Richard Biener <rguenther@suse.de>
1297 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1298 release SSA defs of pattern stmts.
1300 2014-02-17 Richard Biener <rguenther@suse.de>
1302 * tree-inline.c (expand_call_inline): Release the virtual
1303 operand defined by the call we are about to inline.
1305 2014-02-17 Richard Biener <rguenther@suse.de>
1307 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1309 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1310 Ilya Tocar <ilya.tocar@intel.com>
1312 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1313 arguments order in builtin.
1314 (_mm512_permutexvar_epi64): Ditto.
1315 (_mm512_mask_permutexvar_epi64): Ditto
1316 (_mm512_maskz_permutexvar_epi32): Ditto
1317 (_mm512_permutexvar_epi32): Ditto
1318 (_mm512_mask_permutexvar_epi32): Ditto
1320 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1322 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1323 (p8_vmrgow): Likewise.
1325 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1327 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1330 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1333 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1334 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1335 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1336 using direct move instructions on ISA 2.07. Also adjust
1337 instruction length for 64-bit.
1338 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1339 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1341 2014-02-15 Alan Modra <amodra@gmail.com>
1345 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1346 find_replacement on parts of insn rtl that might be reloaded.
1348 2014-02-15 Richard Biener <rguenther@suse.de>
1350 PR tree-optimization/60183
1351 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1352 (tree_ssa_phiprop): Calculate and free post-dominators.
1354 2014-02-14 Jeff Law <law@redhat.com>
1356 PR rtl-optimization/60131
1357 * ree.c (get_extended_src_reg): New function.
1358 (combine_reaching_defs): Use it rather than assuming location of REG.
1359 (find_and_remove_re): Verify first operand of extension is
1360 a REG before adding the insns to the copy list.
1362 2014-02-14 Roland McGrath <mcgrathr@google.com>
1364 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1365 * configure: Regenerated.
1366 * config.in: Regenerated.
1367 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1368 instead of ASM_SHORT.
1370 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1371 Richard Earnshaw <rearnsha@arm.com>
1373 PR rtl-optimization/59535
1374 * lra-constraints.c (process_alt_operands): Encourage alternative
1375 when unassigned pseudo class is superset of the alternative class.
1376 (inherit_reload_reg): Don't inherit when optimizing for code size.
1377 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1378 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1379 modes not less than 4 for Thumb1.
1381 2014-02-14 Kyle McMartin <kyle@redhat.com>
1384 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1386 2014-02-14 Richard Biener <rguenther@suse.de>
1388 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1389 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1390 Do not strip INDIRECT_REFs.
1392 2014-02-14 Richard Biener <rguenther@suse.de>
1395 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1396 DECL_FUNCTION_SPECIFIC_TARGET.
1397 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1398 * tree-streamer-out.c (pack_ts_target_option): Remove.
1399 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1400 (write_ts_function_decl_tree_pointers): Do not stream
1401 DECL_FUNCTION_SPECIFIC_TARGET.
1402 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1403 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1404 (lto_input_ts_function_decl_tree_pointers): Do not stream
1405 DECL_FUNCTION_SPECIFIC_TARGET.
1407 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1409 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1410 (get_initial_def_for_induction, vectorizable_induction): Ignore
1411 debug stmts when looking for exit_phi.
1412 (vectorizable_live_operation): Fix up condition.
1414 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1416 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1417 nreverse() because it changes the content of original tree list.
1419 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1421 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1422 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1424 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1426 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1427 GNU coding standards.
1429 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1432 * dwarf2out.c (gen_subprogram_die): Don't call
1433 add_calling_convention_attribute if subr_die is old_die.
1435 2014-02-13 Sharad Singhai <singhai@google.com>
1437 * doc/optinfo.texi: Fix order of nodes.
1439 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1441 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1442 operands[2], not operands[3].
1444 2014-02-13 Richard Biener <rguenther@suse.de>
1447 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1448 mention the possibility of an in-tree build.
1449 (CLooG): Update recommended version to 0.18.1, mention the
1450 possibility of an in-tree build and clarify that the ISL
1451 bundled with CLooG does not work.
1453 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1456 * expr.c (compress_float_constant): If x is a hard register,
1457 extend into a pseudo and then move to x.
1459 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1461 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1462 caused by bad second argument to warning_at() with -mhotpatch and
1463 nested functions (e.g. with gfortran).
1465 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1467 * opts.c (option_name): Remove "enabled by default" rider.
1469 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1471 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1473 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1474 Uros Bizjak <ubizjak@gmail.com>
1477 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1478 * configure: Regenerated.
1480 2014-02-12 Richard Biener <rguenther@suse.de>
1482 * vec.c (vec_prefix::calculate_allocation): Move as
1483 inline variant to vec.h.
1484 (vec_prefix::calculate_allocation_1): New out-of-line version.
1485 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1486 (vec_prefix::m_has_auto_buf): Rename to ...
1487 (vec_prefix::m_using_auto_storage): ... this.
1488 (vec_prefix::calculate_allocation): Inline the easy cases
1489 and dispatch to calculate_allocation_1 which doesn't need the
1491 (va_heap::reserve): Use gcc_checking_assert.
1492 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1493 m_using_auto_storage.
1494 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1496 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1497 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1498 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1500 2014-02-12 Richard Biener <rguenther@suse.de>
1502 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1503 when we found a dependence.
1505 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1507 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1509 (maybe_fold_stmt): ... into this new function.
1510 * omp-low.c (lower_omp): Update comment.
1512 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1515 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1518 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1520 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1521 identifiers in comments.
1522 (cortexa53_extra_costs): Likewise.
1523 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1524 (cortexa7_extra_costs): Likewise.
1525 (cortexa12_extra_costs): Likewise.
1526 (cortexa15_extra_costs): Likewise.
1527 (v7m_extra_costs): Likewise.
1529 2014-02-12 Richard Biener <rguenther@suse.de>
1532 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1533 of posix_memalign being successful.
1534 (lower_stmt): Restrict lowering of posix_memalign to when
1535 -ftree-bit-ccp is enabled.
1537 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1539 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1541 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1543 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1545 PR rtl-optimization/60116
1546 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1547 other_insn once the combination has been validated.
1549 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1552 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1554 * ipa-devirt.c: Include demangle.h
1555 (odr_violation_reported): New static variable.
1556 (add_type_duplicate): Update odr_violations.
1557 (maybe_record_node): Add completep parameter; update it.
1558 (record_target_from_binfo): Add COMPLETEP parameter;
1559 update it as needed.
1560 (possible_polymorphic_call_targets_1): Likewise.
1561 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1562 rename FINAL to COMPLETE.
1563 (record_targets_from_bases): Sanity check we found the binfo;
1564 fix COMPLETEP updating.
1565 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1566 parameter, fix computing of COMPLETEP.
1567 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1568 at LTO time do demangling.
1569 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1570 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1572 (gimple_get_virt_method_for_binfo): Likewise.
1573 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1574 gimple_get_virt_method_for_vtable): Update prototypes.
1576 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1579 * genautomata.c (add_presence_absence): Fix typo with
1580 {final_}presence_list.
1582 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1585 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1586 for VSX/Altivec vectors that land in GPR registers.
1588 2014-02-11 Richard Henderson <rth@redhat.com>
1589 Jakub Jelinek <jakub@redhat.com>
1592 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1593 around drhs if type conversion to lacc->type is not useless.
1595 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1597 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1599 (cortex-a57.cortex-a53): Likewise.
1600 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1602 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1604 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1607 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1609 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1610 add_options_for_arm_vfp3.
1612 2014-02-11 Jeff Law <law@redhat.com>
1615 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1616 object with an undesirable mode.
1618 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1621 * config/i386/sol2-9.h: New file.
1622 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1623 *-*-solaris2.9*): Use it.
1625 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1627 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1628 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1630 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1632 * config/microblaze/microblaze.c: Extend mcpu version format
1634 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1636 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1638 2014-02-10 Richard Henderson <rth@redhat.com>
1641 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1642 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1643 ms-abi vs -mno-accumulate-outgoing-args.
1644 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1645 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1648 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1651 * cfgexpand.c (expand_asm_operands): Attach source location to
1652 ASM_INPUT rtx objects.
1653 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1655 2014-02-10 Nick Clifton <nickc@redhat.com>
1657 * config/mn10300/mn10300.c (popcount): New function.
1658 (mn10300_expand_prologue): Include saved registers in stack usage
1661 2014-02-10 Jeff Law <law@redhat.com>
1664 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1665 when changing the SET_DEST of a prior insn to avoid an input reload.
1667 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1669 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1670 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1671 -mcall-openbsd, or -mcall-linux.
1672 (CC1_ENDIAN_BIG_SPEC): Remove.
1673 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1674 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1675 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1676 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1677 and %cc1_endian_default.
1678 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1680 2014-02-10 Richard Biener <rguenther@suse.de>
1682 PR tree-optimization/60115
1683 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1684 MEM_REF handling. Properly verify that the accesses are not
1685 out of the objects bound.
1687 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1689 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1692 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1694 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1695 proper constants and fix formatting.
1696 (possible_polymorphic_call_targets): Fix formatting.
1698 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1699 Ilya Tocar <ilya.tocar@intel.com>
1701 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1702 (_mm512_loadu_epi32): Renamed into...
1703 (_mm512_loadu_si512): This.
1704 (_mm512_storeu_epi32): Renamed into...
1705 (_mm512_storeu_si512): This.
1706 (_mm512_maskz_ceil_ps): Removed.
1707 (_mm512_maskz_ceil_pd): Ditto.
1708 (_mm512_maskz_floor_ps): Ditto.
1709 (_mm512_maskz_floor_pd): Ditto.
1710 (_mm512_floor_round_ps): Ditto.
1711 (_mm512_floor_round_pd): Ditto.
1712 (_mm512_ceil_round_ps): Ditto.
1713 (_mm512_ceil_round_pd): Ditto.
1714 (_mm512_mask_floor_round_ps): Ditto.
1715 (_mm512_mask_floor_round_pd): Ditto.
1716 (_mm512_mask_ceil_round_ps): Ditto.
1717 (_mm512_mask_ceil_round_pd): Ditto.
1718 (_mm512_maskz_floor_round_ps): Ditto.
1719 (_mm512_maskz_floor_round_pd): Ditto.
1720 (_mm512_maskz_ceil_round_ps): Ditto.
1721 (_mm512_maskz_ceil_round_pd): Ditto.
1722 (_mm512_expand_pd): Ditto.
1723 (_mm512_expand_ps): Ditto.
1724 * config/i386/i386.c (ix86_builtins): Remove
1725 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1726 (bdesc_args): Ditto.
1727 * config/i386/predicates.md (const1256_operand): New.
1728 (const_1_to_2_operand): Ditto.
1729 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1730 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1731 (*avx512pf_gatherpf<mode>sf): Ditto.
1732 (avx512pf_gatherpf<mode>df): Ditto.
1733 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1734 (*avx512pf_gatherpf<mode>df): Ditto.
1735 (avx512pf_scatterpf<mode>sf): Ditto.
1736 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1737 (*avx512pf_scatterpf<mode>sf): Ditto.
1738 (avx512pf_scatterpf<mode>df): Ditto.
1739 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1740 (*avx512pf_scatterpf<mode>df): Ditto.
1741 (avx512f_expand<mode>): Removed.
1742 (<shift_insn><mode>3<mask_name>): Change predicate type.
1744 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1746 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1747 not at the end of datarefs vector use ordered_remove to avoid
1748 reordering datarefs vector.
1751 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1752 mark local addressable non-static vars as GOVD_PRIVATE
1753 instead of GOVD_LOCAL.
1754 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1755 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1759 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1760 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1761 assume_aligned or alloc_align attributes.
1762 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1763 arguments. Handle also assume_aligned and alloc_align attributes.
1764 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1765 calls to functions with assume_aligned or alloc_align attributes.
1766 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1768 2014-02-08 Terry Guo <terry.guo@arm.com>
1770 * doc/invoke.texi: Document ARM -march=armv7e-m.
1772 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1774 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1775 flag on __cilkrts_rethrow builtin.
1778 * ipa-cp.c (determine_versionability): Fail at -O0
1779 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1780 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1783 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1786 * tree-inline.c (copy_forbidden): Fail for
1787 __attribute__((optimize (0))) functions.
1789 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1791 * varpool.c: Include pointer-set.h.
1792 (varpool_remove_unreferenced_decls): Variables in other partitions
1793 will not be output; be however careful to not lose information
1796 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1798 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1799 lookup in the vtable constructor.
1801 2014-02-07 Jeff Law <law@redhat.com>
1804 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1805 define_insn_and_split.
1807 * ipa-inline.c (inline_small_functions): Fix typos.
1809 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1811 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1812 (s390_can_use_return_insn): Declare.
1813 * config/s390/s390.h (EPILOGUE_USES): Define.
1814 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1816 (s390_chunkify_start): Handle return JUMP_LABELs.
1817 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1818 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1819 (s390_can_use_return_insn): New functions.
1820 (s390_fix_long_loop_prediction): Handle conditional returns.
1821 (TARGET_SET_UP_BY_PROLOGUE): Define.
1822 * config/s390/s390.md (ANY_RETURN): New code iterator.
1823 (*creturn, *csimple_return, return, simple_return): New patterns.
1825 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1827 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1828 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1829 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1830 REG_CFA_RESTORE list when deciding not to restore a register.
1832 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1834 * config/s390/s390.c: Include tree-pass.h and context.h.
1835 (s390_early_mach): New function, split out from...
1836 (s390_emit_prologue): ...here.
1837 (pass_data_s390_early_mach): New pass structure.
1838 (pass_s390_early_mach): New class.
1839 (s390_option_override): Create and register early_mach pass.
1840 Move to end of file.
1842 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1844 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1845 to match for the exit block.
1847 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1849 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1850 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1851 Reject misaligned operands.
1853 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1855 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1857 2014-02-07 Richard Biener <rguenther@suse.de>
1860 * gimple-low.c (lower_builtin_posix_memalign): New function.
1861 (lower_stmt): Call it to lower posix_memalign in a way
1862 to make alignment info accessible.
1864 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1867 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1868 __builtin_setjmp_receiver.
1870 2014-02-07 Richard Biener <rguenther@suse.de>
1873 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1874 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1875 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1876 Handle BUILT_IN_POSIX_MEMALIGN.
1877 (find_func_clobbers): Likewise.
1878 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1879 (call_may_clobber_ref_p_1): Likewise.
1881 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1884 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1887 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1890 * lto-cgraph.c (lto_output_node): Use
1891 symtab_get_symbol_partitioning_class.
1892 (lto_output_varpool_node): likewise.
1893 (symtab_get_symbol_partitioning_class): Move here from
1895 * cgraph.h (symbol_partitioning_class): Likewise.
1896 (symtab_get_symbol_partitioning_class): Declare.
1898 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1900 * ggc.h (ggc_internal_cleared_alloc): New macro.
1901 * vec.h (vec_safe_copy): Handle memory stats.
1902 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1903 * target-globals.c (save_target_globals): Likewise.
1905 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1908 * expr.c (emit_move_resolve_push): Export; be bit more selective
1909 on when to clear alias set.
1910 * expr.h (emit_move_resolve_push): Declare.
1911 * function.h (struct function): Add tail_call_marked.
1912 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1913 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1914 * config/i386/i386.md (TImode move expander): De not call
1916 (FP push expanders): Preserve memory attributes.
1917 * config/i386/sse.md (push<mode>1): Remove.
1918 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1919 (ix86_expand_push): Remove.
1920 * config/i386/mmx.md (push<mode>1): Remove.
1922 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1924 PR rtl-optimization/60030
1925 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1926 lopart with paradoxical subreg before shifting it up by hprec.
1928 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1930 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1931 Remove extra newline at end of file.
1932 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1933 (arm_issue_rate): Handle cortexa57.
1934 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1935 (cortex-a57.cortex-a53): Likewise.
1937 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1940 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1941 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1943 (arm_expand_prologue): Adjust all callers.
1944 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1945 info, registers also at the lowest numbered registers side. Use
1946 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1950 * var-tracking.c (adjust_mems): Before adding a SET to
1951 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1953 2014-02-06 Alan Modra <amodra@gmail.com>
1956 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1957 change SDmode to DDmode when lra_in_progress.
1959 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1962 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1963 free_data_ref on the dr first, and before goto again also set dr
1964 to the next dr. For simd_lane_access, free old datarefs[i] before
1965 overwriting it. For get_vectype_for_scalar_type failure, don't
1966 free_data_ref if simd_lane_access.
1968 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1971 * tree.h (opts_for_fn): New inline function.
1972 (opt_for_fn): Define.
1973 * config/i386/i386.c (ix86_function_regparm): Use
1974 opt_for_fn (decl, optimize) instead of optimize.
1976 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1978 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1979 for SYMBOL_REF in large memory model.
1981 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1983 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1985 (cortex-a57): Likewise.
1986 (cortex-a57.cortex-a53): Likewise.
1988 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1989 Kugan Vivekanandarajah <kuganv@linaro.org>
1991 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1993 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1995 2014-02-06 Richard Biener <rguenther@suse.de>
1997 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1998 set_loop_copy and initialize_original_copy_tables.
2000 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2002 * config/aarch64/aarch64-simd.md
2003 (aarch64_ashr_simddi): Change QI to SI.
2005 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2006 Jakub Jelinek <jakub@redhat.com>
2009 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2012 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2014 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2015 CODE_FOR_altivec_vpku[hw]um to
2016 CODE_FOR_altivec_vpku[hw]um_direct.
2017 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2018 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2019 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2020 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2022 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2024 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2025 generation for -maltivec=be.
2026 (altivec_vsumsws): Simplify redundant test.
2028 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2030 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2031 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2032 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2033 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2034 gen_altivec_vpkuwum.
2035 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2037 (altivec_vpks<VI_char>ss): Likewise.
2038 (altivec_vpks<VI_char>us): Likewise.
2039 (altivec_vpku<VI_char>us): Likewise.
2040 (altivec_vpku<VI_char>um): Likewise.
2041 (altivec_vpku<VI_char>um_direct): New (copy of
2042 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2044 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2045 target is little endian and -maltivec=be is not specified.
2046 (*altivec_vupkhs<VU_char>_direct): New (copy of
2047 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2048 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2049 target is little endian and -maltivec=be is not specified.
2050 (*altivec_vupkls<VU_char>_direct): New (copy of
2051 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2052 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2053 little endian and -maltivec=be is not specified.
2054 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2055 little endian and -maltivec=be is not specified.
2057 2014-02-05 Richard Henderson <rth@redhat.com>
2060 * combine-stack-adj.c: Revert r206943.
2061 * sched-int.h (struct deps_desc): Add last_args_size.
2062 * sched-deps.c (init_deps): Initialize it.
2063 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2064 contain REG_ARGS_SIZE notes.
2066 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2068 * lto-cgraph.c (asm_nodes_output): Make global.
2069 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2070 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2071 (driver_handle_option): Handle OPT_fwpa.
2073 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2076 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2077 a comment typo and formatting issue. If odr_hash hasn't been
2078 created, return vNULL and set *completep to false.
2081 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2082 bb with no successors.
2084 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2087 * doc/invoke.texi (-march): Clarify documentation for ARM.
2091 2014-02-05 Richard Biener <rguenther@suse.de>
2093 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2094 when not vectorizing because of too many alias checks.
2095 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2096 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2098 2014-02-05 Nick Clifton <nickc@redhat.com>
2100 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2101 accept extended registers in any mode when compiling for the MN10300.
2103 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2105 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2106 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2107 sanitization attributes.
2108 (can_inline_edge_p): Likewise.
2109 (sanitize_attrs_match_for_inline_p): New function.
2111 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2113 * ipa-prop.c (detect_type_change): Shor circuit testing of
2114 type changes on THIS pointer.
2116 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2119 * config/pa/pa.c (legitimize_tls_address): Return original address
2120 if not passed a SYMBOL_REF rtx.
2121 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2123 (pa_emit_move_sequence): Simplify TLS source operands.
2124 (pa_legitimate_constant_p): Reject all TLS constants.
2125 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2126 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2128 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2130 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2131 groups when we know they are controlled by LTO.
2132 * varasm.c (default_binds_local_p_1): If object is in other partition,
2133 it will be resolved locally.
2135 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2137 * config/host-linux.c (linux_gt_pch_use_address): Don't
2138 use SSIZE_MAX because it is not always defined.
2140 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2143 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2144 threshold for pseudo splitting.
2145 (update_ebb_live_info): Process call argument hard registers and
2146 hard registers from insn definition too.
2147 (max_small_class_regs_num): New constant.
2148 (inherit_in_ebb): Update live hard regs through EBBs. Update
2149 reloads_num only for small register classes. Don't split for
2152 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2155 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2158 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2160 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2163 2014-02-04 Marek Polacek <polacek@redhat.com>
2165 * gdbinit.in (pel): Define.
2167 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2169 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2172 2014-02-04 Richard Biener <rguenther@suse.de>
2175 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2176 in function context local.
2177 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2178 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2179 similar to LTO_imported_decl_ref.
2181 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2183 PR tree-optimization/60002
2184 * cgraphclones.c (build_function_decl_skip_args): Clear
2187 PR tree-optimization/60023
2188 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2189 false to gsi_replace.
2190 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2191 has been in some EH region and vec_stmt could throw, add
2192 vec_stmt into the same EH region.
2193 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2194 has no lhs, ignore it.
2195 * internal-fn.c (expand_MASK_LOAD): Likewise.
2198 * tree-inline.c (copy_forbidden): Fail for
2199 __attribute__((optimize (0))) functions.
2202 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2203 is set, copy one less argument.
2204 (expand_simd_clones): Don't subtract clone_info->inbranch
2205 from simd_clone_struct_alloc argument.
2207 PR rtl-optimization/57915
2208 * recog.c (simplify_while_replacing): If all unary/binary/relational
2209 operation arguments are constant, attempt to simplify those.
2212 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2213 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2215 2014-02-04 Richard Biener <rguenther@suse.de>
2217 PR tree-optimization/60012
2218 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2219 TBAA disambiguation to all DDRs.
2221 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2224 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2225 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2227 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2230 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2232 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2234 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2235 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2237 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2240 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2241 to figure out targets of polymorphic calls with known decl.
2242 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2243 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2244 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2245 (get_polymorphic_call_info): ... here.
2246 (get_polymorphic_call_info_from_invariant): New function.
2248 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2250 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2251 lookup via vtable pointer; check for type consistency
2252 and turn inconsitent facts into UNREACHABLE.
2253 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2254 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2255 type inconsistent querries; return UNREACHABLE instead.
2257 2014-02-03 Richard Henderson <rth@twiddle.net>
2260 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2261 already processed this node.
2262 (normalize_one_pred_1): Pass along mark_set.
2263 (normalize_one_pred): Create and destroy a pointer_set_t.
2264 (normalize_one_pred_chain): Likewise.
2266 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2268 PR gcov-profile/58602
2269 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2271 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2274 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2275 -fno-devirtualize; try to devirtualize by the knowledge of
2276 virtual table pointer given by aggregate propagation.
2277 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2278 (ipa_print_node_jump_functions): Dump also offset that
2279 is relevant for polymorphic calls.
2280 (determine_known_aggregate_parts): Add arg_type parameter; use it
2281 instead of determining the type from pointer type.
2282 (ipa_compute_jump_functions_for_edge): Update call of
2283 determine_known_aggregate_parts.
2284 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2285 (gimple_get_virt_method_for_binfo): ... here; simplify using
2286 vtable_pointer_value_to_vtable.
2287 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2288 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2289 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2290 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2292 (vtable_pointer_value_to_binfo): ... here.
2293 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2295 2014-02-03 Teresa Johnson <tejohnson@google.com>
2297 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2298 redef of outer loop index variable.
2300 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2304 * doc/extend.texi (Function Attributes): Typo.
2306 2014-02-03 Cong Hou <congh@google.com>
2308 PR tree-optimization/60000
2309 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2310 if the vectorized statement is a store. A store statement can only
2311 appear at the end of pattern statements.
2313 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2315 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2316 (ix86_option_override_internal): Default long double to 64-bit for
2317 32-bit Bionic and to 128-bit for 64-bit Bionic.
2319 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2320 TARGET_LONG_DOUBLE_128 is true.
2321 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2323 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2324 (mlong-double-64): Negate -mlong-double-128.
2325 (mlong-double-128): New option.
2327 * config/i386/i386-c.c (ix86_target_macros): Define
2328 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2330 * doc/invoke.texi: Document -mlong-double-128.
2332 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2334 PR rtl-optimization/60024
2335 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2337 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2339 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2341 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2343 PR rtl-optimization/57662
2344 * sel-sched.c (code_motion_path_driver): Do not mark already not
2345 existing blocks in the visiting bitmap.
2347 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2349 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2350 on the insn being emitted.
2352 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2353 Will Deacon <will.deacon@arm.com>
2355 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2357 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2359 * config/arm/arm-tables.opt: Regenerate.
2361 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2363 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2364 for vector types other than V16QImode.
2365 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2366 define_expand, and call altivec_expand_vec_perm_le when producing
2367 code with little endian element order.
2368 (*altivec_vperm_<mode>_internal): New insn having previous
2369 behavior of altivec_vperm_<mode>.
2370 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2371 altivec_expand_vec_perm_le when producing code with little endian
2373 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2374 behavior of altivec_vperm_<mode>_uns.
2376 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2378 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2379 (altivec_vsumsws): Add handling for -maltivec=be with a little
2381 (altivec_vsumsws_direct): New.
2382 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2383 gen_altivec_vsumsws.
2385 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2387 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2388 vtable_pointer_value_to_binfo): New functions.
2389 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2390 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2392 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2394 * config/nios2/nios2.md (load_got_register): Initialize GOT
2395 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2396 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2398 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2400 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2401 preserverd by passthrough, do not propagate the type.
2403 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2405 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2406 (mips_atomic_assign_expand_fenv): New function.
2407 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2409 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2411 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2412 (__builtin_mips_set_fcsr): Likewise.
2413 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2414 MIPS_USI_FTYPE_VOID.
2415 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2416 (mips16_expand_set_fcsr): Likewise.
2417 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2418 (mips16_set_fcsr_stub): Likewise.
2419 (mips16_get_fcsr_one_only_stub): New class.
2420 (mips16_set_fcsr_one_only_stub): Likewise.
2421 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2422 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2423 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2424 (hard_float): New availability predicate.
2425 (mips_builtins): Add get_fcsr and set_fcsr.
2426 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2427 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2428 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2429 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2430 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2433 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2435 * config/mips/mips.c (mips_one_only_stub): New class.
2436 (mips_need_mips16_rdhwr_p): Replace with...
2437 (mips16_rdhwr_stub): ...this new variable.
2438 (mips16_stub_call_address): New function.
2439 (mips16_rdhwr_one_only_stub): New class.
2440 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2441 (mips_output_mips16_rdhwr): Delete.
2442 (mips_finish_stub): New function.
2443 (mips_code_end): Use it to handle rdhwr stubs.
2445 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2448 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2449 when calculating size of integer atomic types.
2451 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2453 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2455 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2457 PR tree-optimization/60003
2458 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2459 * profile.c (branch_prob): Use gimple_call_builtin_p
2460 to check for BUILT_IN_SETJMP_RECEIVER.
2461 * tree-inline.c (copy_bb): Call notice_special_calls.
2463 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2466 * lra-constraints.c (process_alt_operands): Update reload_sum only
2469 2014-01-31 Richard Henderson <rth@redhat.com>
2472 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2473 until after else_eh is processed.
2475 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2477 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2478 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2479 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2480 in smmintrin.h, remove them.
2481 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2482 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2483 * config/i386/i386.md (ROUND_SAE): Fix value.
2484 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2485 (const48_operand): New.
2486 * config/i386/subst.md (round), (round_expand): Use
2487 const_4_or_8_to_11_operand.
2488 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2490 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2492 * config/i386/constraints.md (Yk): Swap meaning with k.
2493 * config/i386/i386.md (movhi_internal): Change Yk to k.
2494 (movqi_internal): Ditto.
2495 (*k<logic><mode>): Ditto.
2498 (kandn<mode>): Ditto.
2499 (*<code>hi_1): Ditto.
2500 (*<code>qi_1): Ditto.
2501 (kxnor<mode>): Ditto.
2502 (kortestzhi): Ditto.
2503 (kortestchi): Ditto.
2505 (*one_cmplhi2_1): Ditto.
2506 (*one_cmplqi2_1): Ditto.
2507 * config/i386/sse.md (): Change k to Yk.
2508 (avx512f_load<mode>_mask): Ditto.
2509 (avx512f_blendm<mode>): Ditto.
2510 (avx512f_store<mode>_mask): Ditto.
2511 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2512 (avx512f_storedqu<mode>_mask): Ditto.
2513 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2515 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2516 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2517 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2518 (avx512f_maskcmp<mode>3): Ditto.
2519 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2520 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2521 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2522 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2523 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2524 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2525 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2526 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2527 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2528 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2529 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2530 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2531 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2532 (vec_extract_lo_<mode>_maskm): Ditto.
2533 (vec_extract_hi_<mode>_maskm): Ditto.
2534 (avx512f_vternlog<mode>_mask): Ditto.
2535 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2536 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2537 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2538 (avx512f_<code>v8div16qi2_mask): Ditto.
2539 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2540 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2541 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2542 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2543 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2544 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2545 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2546 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2547 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2548 (avx512cd_maskb_vec_dupv8di): Ditto.
2549 (avx512cd_maskw_vec_dupv16si): Ditto.
2550 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2551 (avx512f_vpermi2var<mode>3_mask): Ditto.
2552 (avx512f_vpermi2var<mode>3_mask): Ditto.
2553 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2554 (*avx512f_gathersi<mode>): Ditto.
2555 (*avx512f_gathersi<mode>_2): Ditto.
2556 (*avx512f_gatherdi<mode>): Ditto.
2557 (*avx512f_gatherdi<mode>_2): Ditto.
2558 (*avx512f_scattersi<mode>): Ditto.
2559 (*avx512f_scatterdi<mode>): Ditto.
2560 (avx512f_compress<mode>_mask): Ditto.
2561 (avx512f_compressstore<mode>_mask): Ditto.
2562 (avx512f_expand<mode>_mask): Ditto.
2563 * config/i386/subst.md (mask): Change k to Yk.
2564 (mask_scalar_merge): Ditto.
2567 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2569 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2571 2014-01-31 Richard Biener <rguenther@suse.de>
2574 * builtins.c (fold_builtin_memory_op): Make sure to not
2575 use a floating-point mode or a boolean or enumeral type for
2578 2014-01-30 DJ Delorie <dj@redhat.com>
2580 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2581 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2582 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2583 whenever main() has an epilogue.
2585 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2587 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2588 unused variable "field".
2589 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2590 (vsx_mergeh_<mode>): Likewise.
2591 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2592 (altivec_vmrghh): Likewise.
2593 (altivec_vmrghw): Likewise.
2594 (altivec_vmrglb): Likewise.
2595 (altivec_vmrglh): Likewise.
2596 (altivec_vmrglw): Likewise.
2597 (altivec_vspltb): Add missing uses.
2598 (altivec_vsplth): Likewise.
2599 (altivec_vspltw): Likewise.
2600 (altivec_vspltsf): Likewise.
2602 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2605 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2606 frame related instructions.
2608 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2610 PR rtl-optimization/59959
2611 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2612 any reload of register whose subreg is invalid.
2614 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2616 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2617 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2618 Add missing return type - void.
2620 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2622 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2623 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2624 remove element index adjustment for endian (now handled in vsx.md
2626 (altivec_expand_vec_perm_const): Use
2627 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2628 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2629 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2630 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2631 define_expand and a new define_insn *altivec_vspltb_internal;
2632 adjust for -maltivec=be on a little endian target.
2633 (altivec_vspltb_direct): New.
2634 (altivec_vsplth): Divide into a define_expand and a new
2635 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2636 little endian target.
2637 (altivec_vsplth_direct): New.
2638 (altivec_vspltw): Divide into a define_expand and a new
2639 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2640 little endian target.
2641 (altivec_vspltw_direct): New.
2642 (altivec_vspltsf): Divide into a define_expand and a new
2643 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2644 a little endian target.
2646 2014-01-30 Richard Biener <rguenther@suse.de>
2648 PR tree-optimization/59993
2649 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2650 can propagate form the earlier stmt and avoid the transform
2651 when the intermediate result is needed.
2653 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2655 * README.Portability: Fix typo.
2657 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2659 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2660 comparison_operator with ordered_comparison_operator.
2662 2014-01-30 Nick Clifton <nickc@redhat.com>
2664 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2665 Rename to mn10300_store_multiple_regs.
2666 * config/mn10300/mn10300.c: Likewise.
2667 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2668 store_multiple_regs.
2669 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2670 Call mn10300_store_multiple_regs.
2672 2014-01-30 Nick Clifton <nickc@redhat.com>
2673 DJ Delorie <dj@redhat.com>
2675 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2676 %fp 2 to keep registers after it properly word-aligned.
2677 (rl78_alloc_physical_registers_umul): Handle the case where both
2678 input operands are the same.
2680 2014-01-30 Richard Biener <rguenther@suse.de>
2682 PR tree-optimization/59903
2683 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2686 2014-01-30 Jason Merrill <jason@redhat.com>
2689 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2692 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2694 2014-01-30 Richard Biener <rguenther@suse.de>
2696 PR tree-optimization/59951
2697 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2699 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2702 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2703 SFmode to DFmode case.
2705 2014-01-29 DJ Delorie <dj@redhat.com>
2707 * config/msp430/msp430.opt (-minrt): New.
2708 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2710 (ENDFILE_SPEC): Likewise.
2712 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2714 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2715 (estimate_function_body_sizes): Use it.
2717 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2720 * dwarf2out.c (is_cxx_auto): New.
2721 (is_base_type): Use it.
2722 (gen_type_die_with_usage): Likewise.
2724 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2726 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2727 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2728 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2729 -maltivec=be with LE targets.
2730 (vsx_mergeh_<mode>): Likewise.
2731 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2732 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2733 (altivec_vmrghb): Replace with define_expand and new
2734 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2735 (altivec_vmrghb_direct): New define_insn.
2736 (altivec_vmrghh): Replace with define_expand and new
2737 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2738 (altivec_vmrghh_direct): New define_insn.
2739 (altivec_vmrghw): Replace with define_expand and new
2740 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2741 (altivec_vmrghw_direct): New define_insn.
2742 (*altivec_vmrghsf): Adjust for endianness.
2743 (altivec_vmrglb): Replace with define_expand and new
2744 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2745 (altivec_vmrglb_direct): New define_insn.
2746 (altivec_vmrglh): Replace with define_expand and new
2747 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2748 (altivec_vmrglh_direct): New define_insn.
2749 (altivec_vmrglw): Replace with define_expand and new
2750 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2751 (altivec_vmrglw_direct): New define_insn.
2752 (*altivec_vmrglsf): Adjust for endianness.
2753 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2754 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2755 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2756 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2757 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2758 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2759 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2760 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2762 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2764 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2765 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2768 2014-01-29 Richard Biener <rguenther@suse.de>
2770 PR tree-optimization/58742
2771 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2772 associate_pointerplus_align.
2773 (associate_pointerplus_diff): New function.
2774 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2775 and associate_pointerplus_diff.
2777 2014-01-29 Richard Biener <rguenther@suse.de>
2779 * lto-streamer.h (LTO_major_version): Bump to 3.
2780 (LTO_minor_version): Reset to 0.
2782 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2784 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2785 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2786 (arm_file_start): Generate correct asm header for armv7ve.
2787 * config/arm/bpabi.h: Add multilib support for armv7ve.
2788 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2789 and cortex-a15 to armv7ve.
2790 * config/arm/t-aprofile: Add multilib support for armv7ve.
2791 * doc/invoke.texi: Document -march=armv7ve.
2793 2014-01-29 Richard Biener <rguenther@suse.de>
2795 PR tree-optimization/58742
2796 * tree-ssa-forwprop.c (associate_plusminus): Return true
2797 if we changed sth, defer EH cleanup to ...
2798 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2799 (simplify_mult): New function.
2801 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2804 PR tree-optimization/59920
2805 * tree.c (build_common_builtin_nodes): Remove
2806 __builtin_setjmp_dispatcher initialization.
2807 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2808 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2809 instead of gsi_after_labels + manually skipping debug stmts.
2810 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2811 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2812 * tree-inline.c (copy_edges_for_bb): Remove
2813 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2814 argument. Ignore computed_goto_p stmts. Don't call
2815 make_abnormal_goto_edges. If a call might need abnormal edges
2816 for non-local gotos, see if it already has an edge to
2817 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2818 with true argument, don't do anything then, otherwise add
2819 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2820 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2822 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2823 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2824 (lower_stmt): Don't set data->calls_builtin_setjmp.
2825 (lower_builtin_setjmp): Adjust comment.
2826 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2827 * tree-cfg.c (found_computed_goto): Remove.
2828 (factor_computed_gotos): Remove.
2829 (make_goto_expr_edges): Return bool, true for computed gotos.
2830 Don't call make_abnormal_goto_edges.
2831 (build_gimple_cfg): Don't set found_computed_goto, don't call
2832 factor_computed_gotos.
2833 (computed_goto_p): No longer static.
2834 (make_blocks): Don't set found_computed_goto.
2835 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2836 (make_edges): If make_goto_expr_edges returns true, push bb
2837 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2838 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2839 vector. Record mapping between bbs and OpenMP regions if there
2840 are any, adjust make_gimple_omp_edges caller. Call
2841 handle_abnormal_edges.
2842 (make_abnormal_goto_edges): Remove.
2843 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2844 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2845 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2846 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2847 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2848 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2849 filling *region also set *region_idx to (*region)->entry->index.
2852 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2853 For REGs set ORIGINAL_REGNO.
2855 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2857 * doc/md.texi: Mention that a target shouldn't implement
2858 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2861 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2863 PR tree-optimization/59594
2864 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2865 a copy of the datarefs vector rather than the vector itself.
2867 2014-01-28 Jason Merrill <jason@redhat.com>
2870 * dwarf2out.c (auto_die): New static.
2871 (gen_type_die_with_usage): Handle C++1y 'auto'.
2872 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2875 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2878 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2879 (SPEC_X32): Likewise.
2880 (SPEC_64): Likewise.
2881 * config/i386/i386.c (ix86_option_override_internal): Turn off
2882 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2884 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2885 * config/i386/i386.h (TARGET_16BIT): New macro.
2886 (TARGET_16BIT_P): Likewise.
2887 * config/i386/i386.opt: Add m16.
2888 * doc/invoke.texi: Document -m16.
2890 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2892 PR preprocessor/59935
2893 * input.c (location_get_source_line): Bail out on when line number
2894 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2896 2014-01-28 Richard Biener <rguenther@suse.de>
2898 PR tree-optimization/58742
2899 * tree-ssa-forwprop.c (associate_plusminus): Handle
2900 pointer subtraction of the form (T)(P + A) - (T)P.
2902 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2904 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2907 2014-01-28 Richard Biener <rguenther@suse.de>
2910 2014-01-28 Richard Biener <rguenther@suse.de>
2912 PR rtl-optimization/45364
2913 PR rtl-optimization/59890
2914 * var-tracking.c (local_get_addr_clear_given_value): Handle
2915 already cleared slot.
2916 (val_reset): Handle not allocated local_get_addr_cache.
2917 (vt_find_locations): Use post-order on the inverted CFG.
2919 2014-01-28 Richard Biener <rguenther@suse.de>
2921 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2923 2014-01-28 Richard Biener <rguenther@suse.de>
2925 PR rtl-optimization/45364
2926 PR rtl-optimization/59890
2927 * var-tracking.c (local_get_addr_clear_given_value): Handle
2928 already cleared slot.
2929 (val_reset): Handle not allocated local_get_addr_cache.
2930 (vt_find_locations): Use post-order on the inverted CFG.
2932 2014-01-28 Alan Modra <amodra@gmail.com>
2934 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2935 * configure.ac <recursive call for build != host>: Define
2936 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2937 and LD_FOR_BUILD too.
2938 * configure: Regenerate.
2940 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2942 * config/i386/i386.c (get_builtin_code_for_version): Separate
2943 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2944 Broadwell from Haswell.
2946 2014-01-27 Steve Ellcey <sellcey@mips.com>
2948 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2949 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2950 * config/mips/mips.c (mips_option_override): Change setting
2952 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2953 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2954 Change from Mask to Var.
2956 2014-01-27 Jeff Law <law@redhat.com>
2958 * ipa-inline.c (inline_small_functions): Fix typo.
2960 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2962 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2963 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2964 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2965 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2966 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2967 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2968 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2969 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2970 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2971 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2972 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2973 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2974 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2975 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2976 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2977 (_mm512_storeu_epi64): Ditto.
2978 (_mm512_cmpge_epi32_mask): Ditto.
2979 (_mm512_cmpge_epu32_mask): Ditto.
2980 (_mm512_cmpge_epi64_mask): Ditto.
2981 (_mm512_cmpge_epu64_mask): Ditto.
2982 (_mm512_cmple_epi32_mask): Ditto.
2983 (_mm512_cmple_epu32_mask): Ditto.
2984 (_mm512_cmple_epi64_mask): Ditto.
2985 (_mm512_cmple_epu64_mask): Ditto.
2986 (_mm512_cmplt_epi32_mask): Ditto.
2987 (_mm512_cmplt_epu32_mask): Ditto.
2988 (_mm512_cmplt_epi64_mask): Ditto.
2989 (_mm512_cmplt_epu64_mask): Ditto.
2990 (_mm512_cmpneq_epi32_mask): Ditto.
2991 (_mm512_cmpneq_epu32_mask): Ditto.
2992 (_mm512_cmpneq_epi64_mask): Ditto.
2993 (_mm512_cmpneq_epu64_mask): Ditto.
2994 (_mm512_expand_pd): Ditto.
2995 (_mm512_expand_ps): Ditto.
2996 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2997 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2998 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2999 * config/i386/i386.c (ix86_builtins): Add
3000 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3001 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3002 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3003 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3004 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3005 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3006 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3007 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3008 IX86_BUILTIN_PMOVUSQW512_MEM.
3009 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3010 __builtin_ia32_pmovsqd512mem_mask,
3011 __builtin_ia32_pmovqd512mem_mask,
3012 __builtin_ia32_pmovusqw512mem_mask,
3013 __builtin_ia32_pmovsqw512mem_mask,
3014 __builtin_ia32_pmovqw512mem_mask,
3015 __builtin_ia32_pmovusdw512mem_mask,
3016 __builtin_ia32_pmovsdw512mem_mask,
3017 __builtin_ia32_pmovdw512mem_mask,
3018 __builtin_ia32_pmovqb512mem_mask,
3019 __builtin_ia32_pmovusqb512mem_mask,
3020 __builtin_ia32_pmovsqb512mem_mask,
3021 __builtin_ia32_pmovusdb512mem_mask,
3022 __builtin_ia32_pmovsdb512mem_mask,
3023 __builtin_ia32_pmovdb512mem_mask.
3024 (bdesc_args): Add __builtin_ia32_expanddf512,
3025 __builtin_ia32_expandsf512.
3026 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3027 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3028 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3029 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3030 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3031 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3032 (avx512f_<code>v8div16qi2_mask_store): This.
3033 (avx512f_expand<mode>): New.
3035 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3037 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3039 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3040 (_mm512_prefetch_i32scatter_pd): Ditto.
3041 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3042 (_mm512_prefetch_i64scatter_pd): Ditto.
3043 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3044 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3045 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3046 (_mm512_prefetch_i32scatter_ps): Ditto.
3047 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3048 (_mm512_prefetch_i64scatter_ps): Ditto.
3049 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3050 * config/i386/i386-builtin-types.def: Define
3051 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3052 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3053 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3054 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3055 IX86_BUILTIN_SCATTERPFQPD.
3056 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3057 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3058 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3059 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3060 __builtin_ia32_scatterpfqps.
3061 (ix86_expand_builtin): Expand new built-ins.
3062 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3063 fix memory access data type.
3064 (*avx512pf_gatherpf<mode>_mask): Ditto.
3065 (*avx512pf_gatherpf<mode>): Ditto.
3066 (avx512pf_scatterpf<mode>): Ditto.
3067 (*avx512pf_scatterpf<mode>_mask): Ditto.
3068 (*avx512pf_scatterpf<mode>): Ditto.
3069 (GATHER_SCATTER_SF_MEM_MODE): New.
3070 (avx512pf_gatherpf<mode>df): Ditto.
3071 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3072 (*avx512pf_scatterpf<mode>df): Ditto.
3074 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3077 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3078 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3081 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3083 * common/config/arm/arm-common.c
3084 (arm_rewrite_mcpu): Handle multiple names.
3086 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3088 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3090 * gimple-builder.h (create_gimple_tmp): Delete.
3092 2014-01-27 Christian Bruel <christian.bruel@st.com>
3094 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3097 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3099 * config/pa/pa.md (call): Generate indirect long calls to non-local
3100 functions when outputing 32-bit code.
3101 (call_value): Likewise except for special call to buggy powf function.
3103 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3104 portable runtime and PIC indirect calls.
3105 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3106 and PIC call sequences. Use ldo instead of blr to set return register
3107 in PIC call sequence.
3109 2014-01-25 Walter Lee <walt@tilera.com>
3111 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3112 avoid clobbering a live register.
3114 2014-01-25 Walter Lee <walt@tilera.com>
3116 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3117 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3118 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3119 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3121 2014-01-25 Walter Lee <walt@tilera.com>
3123 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3124 arguments on even registers.
3125 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3127 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3128 (BIGGEST_ALIGNMENT): Ditto.
3129 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3131 2014-01-25 Walter Lee <walt@tilera.com>
3133 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3134 insns before bundling.
3135 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3137 2014-01-25 Walter Lee <walt@tilera.com>
3139 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3140 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3141 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3143 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3145 * config/mips/constraints.md (kl): Delete.
3146 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3147 define expands, using...
3148 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3149 instructions for MIPS16.
3150 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3151 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3153 2014-01-25 Walter Lee <walt@tilera.com>
3155 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3159 2014-01-25 Walter Lee <walt@tilera.com>
3161 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3162 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3164 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3166 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3169 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3171 * print-rtl.c (in_call_function_usage): New var.
3172 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3173 EXPR_LIST mode as mode and not as reg note name.
3176 * cfgloopmanip.c (copy_loop_info): If
3177 loop->warned_aggressive_loop_optimizations, make sure
3178 the flag is set in target loop too.
3180 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3182 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3184 * builtins.def: Likewise.
3185 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3186 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3187 * ira.c (ira_setup_eliminable_regset): Likewise.
3188 * omp-low.c (gate_expand_omp): Likewise.
3189 (execute_lower_omp): Likewise.
3190 (diagnose_sb_0): Likewise.
3191 (gate_diagnose_omp_blocks): Likewise.
3192 (simd_clone_clauses_extract): Likewise.
3195 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3197 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3198 correction for little endian...
3199 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3202 2014-01-24 Jeff Law <law@redhat.com>
3204 PR tree-optimization/59919
3205 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3206 for non-returning calls.
3208 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3210 * common/config/aarch64/aarch64-common.c
3211 (aarch64_rewrite_mcpu): Handle multiple names.
3212 * config/aarch64/aarch64.h
3213 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3215 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3217 * input.c (add_file_to_cache_tab): Handle the case where fopen
3220 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3223 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3224 from push operand if code of push isn't PRE_DEC.
3226 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3229 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3230 -mquad-memory-atomic. Update -mquad-memory documentation to say
3231 it is only used for non-atomic loads/stores.
3233 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3234 -mquad-memory or -mquad-memory-atomic switches.
3236 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3237 -mquad-memory-atomic to ISA 2.07 support.
3239 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3240 to separate support of normal quad word memory operations (ldq, stq)
3241 from the atomic quad word memory operations.
3243 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3244 support to separate non-atomic quad word operations from atomic
3245 quad word operations. Disable non-atomic quad word operations in
3246 little endian mode so that we don't have to swap words after the
3247 load and before the store.
3248 (quad_load_store_p): Add comment about atomic quad word support.
3249 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3250 options printed with -mdebug=reg.
3252 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3253 -mquad-memory-atomic as the test for whether we have quad word
3254 atomic instructions.
3255 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3256 or -mp8-vector are used, allow byte/half-word atomic operations.
3258 * config/rs6000/sync.md (load_lockedti): Insure that the address
3259 is a proper indexed or indirect address for the lqarx instruction.
3260 On little endian systems, swap the hi/lo registers after the lqarx
3262 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3263 insure the address is valid for the lqarx instruction.
3264 (store_conditionalti): Insure that the address is a proper indexed
3265 or indirect address for the stqcrx. instruction. On little endian
3266 systems, swap the hi/lo registers before doing the stqcrx.
3268 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3269 insure the address is valid for the stqcrx. instruction.
3271 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3272 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3273 type of quad memory support is available.
3275 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3278 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3279 there is a danger of looping.
3281 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3283 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3284 force flag_ira_loop_pressure if set via command line.
3286 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3288 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3289 (ashr_simd): New builtin handling DI mode.
3290 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3291 (aarch64_sshr_simddi): New match pattern.
3292 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3293 (vshrd_n_s64): Likewise.
3294 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3296 2014-01-23 Nick Clifton <nickc@redhat.com>
3298 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3299 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3300 favour of mcu specific scripts.
3301 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3304 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3305 Alex Velenko <Alex.Velenko@arm.com>
3307 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3308 (vaddv_s16): Likewise.
3309 (vaddv_s32): Likewise.
3310 (vaddv_u8): Likewise.
3311 (vaddv_u16): Likewise.
3312 (vaddv_u32): Likewise.
3313 (vaddvq_s8): Likewise.
3314 (vaddvq_s16): Likewise.
3315 (vaddvq_s32): Likewise.
3316 (vaddvq_s64): Likewise.
3317 (vaddvq_u8): Likewise.
3318 (vaddvq_u16): Likewise.
3319 (vaddvq_u32): Likewise.
3320 (vaddvq_u64): Likewise.
3321 (vaddv_f32): Likewise.
3322 (vaddvq_f32): Likewise.
3323 (vaddvq_f64): Likewise.
3324 (vmaxv_f32): Likewise.
3325 (vmaxv_s8): Likewise.
3326 (vmaxv_s16): Likewise.
3327 (vmaxv_s32): Likewise.
3328 (vmaxv_u8): Likewise.
3329 (vmaxv_u16): Likewise.
3330 (vmaxv_u32): Likewise.
3331 (vmaxvq_f32): Likewise.
3332 (vmaxvq_f64): Likewise.
3333 (vmaxvq_s8): Likewise.
3334 (vmaxvq_s16): Likewise.
3335 (vmaxvq_s32): Likewise.
3336 (vmaxvq_u8): Likewise.
3337 (vmaxvq_u16): Likewise.
3338 (vmaxvq_u32): Likewise.
3339 (vmaxnmv_f32): Likewise.
3340 (vmaxnmvq_f32): Likewise.
3341 (vmaxnmvq_f64): Likewise.
3342 (vminv_f32): Likewise.
3343 (vminv_s8): Likewise.
3344 (vminv_s16): Likewise.
3345 (vminv_s32): Likewise.
3346 (vminv_u8): Likewise.
3347 (vminv_u16): Likewise.
3348 (vminv_u32): Likewise.
3349 (vminvq_f32): Likewise.
3350 (vminvq_f64): Likewise.
3351 (vminvq_s8): Likewise.
3352 (vminvq_s16): Likewise.
3353 (vminvq_s32): Likewise.
3354 (vminvq_u8): Likewise.
3355 (vminvq_u16): Likewise.
3356 (vminvq_u32): Likewise.
3357 (vminnmv_f32): Likewise.
3358 (vminnmvq_f32): Likewise.
3359 (vminnmvq_f64): Likewise.
3361 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3363 * config/aarch64/aarch64-simd.md
3364 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3365 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3366 (*aarch64_mul3_elt<mode>): Likewise.
3367 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3368 (*aarch64_mul3_elt_to_64v2df): Likewise.
3369 (*aarch64_mla_elt<mode>): Likewise.
3370 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3371 (*aarch64_mls_elt<mode>): Likewise.
3372 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3373 (*aarch64_fma4_elt<mode>): Likewise.
3374 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3375 (*aarch64_fma4_elt_to_64v2df): Likewise.
3376 (*aarch64_fnma4_elt<mode>): Likewise.
3377 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3378 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3379 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3380 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3381 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3382 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3383 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3384 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3385 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3387 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3389 * config/aarch64/aarch64-simd.md
3390 (aarch64_be_checked_get_lane<mode>): New define_expand.
3391 * config/aarch64/aarch64-simd-builtins.def
3392 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3393 New builtin definition.
3394 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3395 Use new safe be builtin.
3397 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3399 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3401 (aarch64_be_st1<mode>): Likewise.
3402 (aarch_ld1<VALL:mode>): Define_expand modified.
3403 (aarch_st1<VALL:mode>): Likewise.
3404 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3405 (UNSPEC_ST1): Likewise.
3407 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3409 * config/microblaze/microblaze.md: Add trap insn and attribute
3411 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3413 PR preprocessor/58580
3414 * input.h (location_get_source_line): Take an additional line_size
3416 (void diagnostics_file_cache_fini): Declare new function.
3417 * input.c (struct fcache): New type.
3418 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3419 New static constants.
3420 (diagnostic_file_cache_init, total_lines_num)
3421 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3422 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3423 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3424 (get_next_line, read_next_line, goto_next_line, read_line_num):
3425 New static function definitions.
3426 (diagnostic_file_cache_fini): New function.
3427 (location_get_source_line): Take an additional output line_len
3428 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3430 * diagnostic.c (diagnostic_finish): Call
3431 diagnostic_file_cache_fini.
3432 (adjust_line): Take an additional input parameter for the length
3433 of the line, rather than calculating it with strlen.
3434 (diagnostic_show_locus): Adjust the use of
3435 location_get_source_line and adjust_line with respect to their new
3436 signature. While displaying a line now, do not stop at the first
3437 null byte. Rather, display the zero byte as a space and keep
3438 going until we reach the size of the line.
3439 * Makefile.in: Add vec.o to OBJS-libcommon
3441 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3442 Ilya Tocar <ilya.tocar@intel.com>
3444 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3445 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3446 (__builtin_ia32_kmov16): Ditto.
3447 * config/i386/i386.md (UNSPEC_KMOV): New.
3450 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3452 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3453 (_mm512_storeu_si512): Ditto.
3455 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3458 * rtl.h (get_referenced_operands): Declare.
3459 * recog.c (get_referenced_operands): New function.
3460 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3461 operands have been referenced when recording LO_SUM references.
3463 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3465 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3467 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3469 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3470 Enable for generic and recent AMD targets.
3472 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3474 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3475 ARG_SIZE note when adjustment was eliminated.
3477 2014-01-22 Jeff Law <law@redhat.com>
3479 PR tree-optimization/59597
3480 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3481 in file. Accept new argument REGISTERING and use it to modify
3482 dump output appropriately.
3483 (register_jump_thread): Corresponding changes.
3484 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3485 thread paths involving joiner blocks. Add code to dump cancelled
3486 jump threading paths.
3488 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3490 PR rtl-optimization/59477
3491 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3492 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3494 2014-01-22 Tom Tromey <tromey@redhat.com>
3496 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3498 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3500 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3502 PR rtl-optimization/59896
3503 * lra-constraints.c (process_alt_operands): Check unused note for
3504 matched operands of insn with no output reloads.
3506 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3508 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3509 (mips_move_from_gpr_cost): Likewise.
3511 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3513 PR rtl-optimization/59858
3514 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3515 ira_class_hard_regs_num.
3516 (process_alt_operands): Increase reject for dying matched operand.
3518 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3521 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3522 smaller than size, perform several stores or loads and stores
3523 at dst + count - size to store or copy all of size bytes, rather
3524 than just last modesize bytes.
3526 2014-01-20 DJ Delorie <dj@redhat.com>
3528 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3529 that CLOBBERs are REGs before propogating their values.
3531 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3534 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3535 (cgraph_inline_failed_type): New function.
3536 * cgraph.h (DEFCIFCODE): Add type.
3537 (cgraph_inline_failed_type_t): New enum.
3538 (cgraph_inline_failed_type): New prototype.
3539 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3540 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3541 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3542 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3543 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3544 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3545 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3546 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3547 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3548 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3549 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3550 OPTIMIZATION_MISMATCH.
3551 * tree-inline.c (expand_call_inline): Emit errors during
3552 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3554 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3557 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3558 mode attribute in insn output.
3560 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3562 * output.h (output_constant): Delete.
3563 * varasm.c (output_constant): Make private.
3565 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3567 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3569 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3572 * tree.h (fold_builtin_strcat): New prototype.
3573 * builtins.c (fold_builtin_strcat): No longer static. Add len
3574 argument, if non-NULL, don't call c_strlen. Optimize
3575 directly into __builtin_memcpy instead of __builtin_strcpy.
3576 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3577 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3579 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3581 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3582 for SImode_address_operand operands, having only a REG argument.
3584 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3586 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3587 loader name using mbig-endian.
3588 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3590 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3592 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3596 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3598 * config/aarch64/aarch64-protos.h
3599 (aarch64_cannot_change_mode_class_ptr): Declare.
3600 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3601 aarch64_cannot_change_mode_class_ptr): New.
3602 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3603 backend hook aarch64_cannot_change_mode_class.
3605 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3607 * common/config/aarch64/aarch64-common.c
3608 (aarch64_handle_option): Don't handle any option order logic here.
3609 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3610 selected_cpu, warn on architecture version mismatch.
3611 (aarch64_override_options): Fix parsing order for option strings.
3613 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3614 Iain Sandoe <iain@codesourcery.com>
3617 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3618 warning. Amend comment to reflect current functionality.
3620 2014-01-20 Richard Biener <rguenther@suse.de>
3623 * builtins.c (fold_builtin_strcat): Remove case better handled
3624 by tree-ssa-strlen.c.
3626 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3628 * config/aarch64/aarch64.opt
3629 (mcpu, march, mtune): Make case-insensitive.
3631 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3634 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3635 if operands[1] is a REG or ZERO_EXTEND of a REG.
3637 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3639 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3641 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3643 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3644 long non-pic millicode calls.
3646 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3648 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3650 2014-01-19 Kito Cheng <kito@0xlab.org>
3652 * builtins.c (expand_movstr): Check movstr expand done or fail.
3654 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3655 H.J. Lu <hongjiu.lu@intel.com>
3658 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3659 to DImode for zero-extended addresses.
3661 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3663 PR rtl-optimization/57763
3664 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3665 on the new indirect jump_insn and increment LABEL_NUSES (label).
3667 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3671 * config.gcc (x86_archs): New variable.
3672 (x86_64_archs): Likewise.
3673 (x86_cpus): Likewise.
3674 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3675 --with-arch/--with-cpu= options.
3676 Support --with-arch=/--with-cpu={nehalem,westmere,
3677 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3679 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3681 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3682 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3684 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3686 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3688 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3691 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3692 clear cpp_get_options (parse_in)->warn_unused_macros for
3693 ix86_target_macros_internal with cpp_define.
3695 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3697 * jump.c (delete_related_insns): Keep (use (insn))s.
3698 * reorg.c (redundant_insn): Check for barriers too.
3700 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3702 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3704 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3706 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3707 call to $$dyncall when TARGET_LONG_CALLS is true.
3709 2014-01-17 Jeff Law <law@redhat.com>
3711 * ree.c (combine_set_extension): Temporarily disable test for
3712 changing number of hard registers.
3714 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3717 * ipa-inline-analysis.c (inline_free_summary):
3718 Do not free summary of aliases.
3720 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3723 * gimplify.c (gimplify_expr): Use create_tmp_var
3724 instead of create_tmp_var_raw. If cond doesn't have
3725 integral type, don't add the IFN_ANNOTATE builtin at all.
3727 2014-01-17 Martin Jambor <mjambor@suse.cz>
3730 * ipa-cp.c (prev_edge_clone): New variable.
3731 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3732 Also resize prev_edge_clone vector.
3733 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3734 (ipcp_edge_removal_hook): New function.
3735 (ipcp_driver): Register ipcp_edge_removal_hook.
3737 2014-01-17 Andrew Pinski <apinski@cavium.com>
3738 Steve Ellcey <sellcey@mips.com>
3741 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3744 2014-01-17 Jeff Law <law@redhat.com>
3747 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3748 so that pass_ccp runs first.
3750 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3752 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3753 (ix86_adjust_cost): Use !TARGET_XXX.
3754 (do_reorder_for_imul): Likewise.
3755 (swap_top_of_ready_list): Likewise.
3756 (ix86_sched_reorder): Likewise.
3758 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3760 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3761 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3762 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3763 (intel_memset): New. Duplicate slm_memset.
3764 (intel_cost): New. Duplicate slm_cost.
3765 (m_INTEL): New macro.
3766 (processor_target_table): Add "intel".
3767 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3768 with PROCESSOR_INTEL for "intel".
3769 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3770 PROCESSOR_SILVERMONT.
3771 (ix86_issue_rate): Likewise.
3772 (ix86_adjust_cost): Likewise.
3773 (ia32_multipass_dfa_lookahead): Likewise.
3774 (swap_top_of_ready_list): Likewise.
3775 (ix86_sched_reorder): Likewise.
3776 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3777 instead of TARGET_OPT_AGU.
3778 * config/i386/i386.h (TARGET_INTEL): New.
3779 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3780 (processor_type): Add PROCESSOR_INTEL.
3781 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3782 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3784 2014-01-17 Marek Polacek <polacek@redhat.com>
3787 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3790 2014-01-17 Richard Biener <rguenther@suse.de>
3792 PR tree-optimization/46590
3793 * opts.c (default_options_table): Add entries for
3794 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3795 all enabled at -O1 but not for -Og.
3796 * common.opt (fbranch-count-reg): Remove Init(1).
3797 (fmove-loop-invariants): Likewise.
3798 (ftree-pta): Likewise.
3800 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3802 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3803 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3804 decls to at least the GCC 4.8 used alignments.
3807 * tree-nested.c (convert_nonlocal_reference_stmt,
3808 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3809 of GIMPLE_BIND stmts, adjust associated decls.
3811 2014-01-17 Richard Biener <rguenther@suse.de>
3813 PR tree-optimization/46590
3814 * vec.h (vec<>::bseach): New member function implementing
3815 binary search according to C89 bsearch.
3816 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3817 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3818 bitmap pointer again. Make accesses_in_loop a flat array.
3819 (mem_ref_obstack): New global.
3820 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3821 (mark_ref_stored): Likewise.
3822 (ref_indep_loop_p_2): Likewise.
3823 (set_ref_stored_in_loop): New helper function.
3824 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3825 (memref_free): Adjust.
3826 (record_mem_ref_loc): Simplify.
3827 (gather_mem_refs_stmt): Adjust.
3828 (sort_locs_in_loop_postorder_cmp): New function.
3829 (analyze_memory_references): Sort accesses_in_loop after
3830 loop postorder number.
3831 (find_ref_loc_in_loop_cmp): New function.
3832 (for_all_locs_in_loop): Find relevant cluster of locs in
3833 accesses_in_loop and iterate without recursion.
3834 (execute_sm): Avoid uninit warning.
3835 (struct ref_always_accessed): Simplify.
3836 (ref_always_accessed::operator ()): Likewise.
3837 (ref_always_accessed_p): Likewise.
3838 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3839 loop postorder numbers here.
3840 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3843 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3846 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3847 on decls for which assemble_alias has been called.
3849 2014-01-17 Nick Clifton <nickc@redhat.com>
3851 * config/msp430/msp430.opt: (mcpu): New option.
3852 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3853 (msp430_option_override): Parse target_cpu. If the MCU name
3854 matches a generic string, clear target_mcu.
3855 (msp430_attr): Allow numeric interrupt values up to 63.
3856 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3857 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3859 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3861 * config/msp430/msp430.md (popm): Use %J rather than %I.
3862 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3863 (addhi_cy_i): Use immediate_operand for operand 2.
3864 * doc/invoke.texi: Document -mcpu option.
3866 2014-01-17 Richard Biener <rguenther@suse.de>
3868 PR rtl-optimization/38518
3869 * df.h (df_analyze_loop): Declare.
3870 * df-core.c: Include cfgloop.h.
3871 (df_analyze_1): Split out main part of df_analyze.
3872 (df_analyze): Adjust.
3873 (loop_inverted_post_order_compute): New function.
3874 (loop_post_order_compute): Likewise.
3875 (df_analyze_loop): New function avoiding whole-function
3877 * loop-invariant.c (find_defs): Use df_analyze_loop.
3878 (find_invariants): Adjust.
3879 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3881 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3883 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3884 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3886 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3888 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3889 traversal when removing references.
3891 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3894 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3896 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3899 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3900 pushing a reload for an autoinc when we had previously reloaded an
3901 inner part of the address.
3903 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3905 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3907 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3908 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3909 when not giving up or versioning for alias only because of
3911 (vect_analyze_data_ref_dependences): Set to true.
3912 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3914 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3915 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3919 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3922 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3923 operand 0 predicate for gathers, use a new pseudo as subtarget.
3925 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3928 * lra-constraints.c (process_alt_operands): Add printing debug info.
3929 Check absence of input/output reloads for matched operands too.
3931 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3933 PR rtl-optimization/59835
3934 * ira.c (ira_init_register_move_cost): Increase cost for
3937 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3939 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3941 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3944 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3945 non-register objects. Use gen_(high/low)part more consistently.
3948 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3951 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3952 endian support, remove tests for WORDS_BIG_ENDIAN.
3953 (p8_mfvsrd_3_<mode>): Likewise.
3954 (reload_gpr_from_vsx<mode>): Likewise.
3955 (reload_gpr_from_vsxsf): Likewise.
3956 (p8_mfvsrd_4_disf): Likewise.
3958 2014-01-16 Richard Biener <rguenther@suse.de>
3960 PR rtl-optimization/46590
3961 * lcm.c (compute_antinout_edge): Use postorder iteration.
3962 (compute_laterin): Use inverted postorder iteration.
3964 2014-01-16 Nick Clifton <nickc@redhat.com>
3967 * varasm.c (output_constant): Return the number of bytes actually
3969 (output_constructor_array_range): Update the field size with the
3970 number of bytes emitted by output_constant.
3971 (output_constructor_regular_field): Likewise. Also do not
3972 complain if the total number of bytes emitted is now greater
3973 than the expected fieldpos.
3974 * output.h (output_constant): Update prototype and descriptive comment.
3976 2014-01-16 Marek Polacek <polacek@redhat.com>
3979 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3980 it is error_mark_node.
3982 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3984 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3985 VALID_AVX256_REG_OR_OI_MODE.
3987 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3989 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3990 current procedure should be profiled.
3992 2014-01-15 Andrew Pinski <apinski@cavium.com>
3994 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3995 of moving from/to the STACK_REG register class.
3997 2014-01-15 Richard Henderson <rth@redhat.com>
4000 * reginfo.c (global_regs_decl): Globalize.
4001 * rtl.h (global_regs_decl): Declare.
4002 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4003 reserved via global_regs.
4005 2014-01-15 Teresa Johnson <tejohnson@google.com>
4007 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4009 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4011 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4012 and vmulosh rather than call gen_vec_widen_smult_*.
4013 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4014 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4015 (vec_widen_smult_even_v16qi): Likewise.
4016 (vec_widen_umult_even_v8hi): Likewise.
4017 (vec_widen_smult_even_v8hi): Likewise.
4018 (vec_widen_umult_odd_v16qi): Likewise.
4019 (vec_widen_smult_odd_v16qi): Likewise.
4020 (vec_widen_umult_odd_v8hi): Likewise.
4021 (vec_widen_smult_odd_v8hi): Likewise.
4022 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4023 vmuloub rather than call gen_vec_widen_umult_*.
4024 (vec_widen_umult_lo_v16qi): Likewise.
4025 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4026 vmulosb rather than call gen_vec_widen_smult_*.
4027 (vec_widen_smult_lo_v16qi): Likewise.
4028 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4029 rather than call gen_vec_widen_umult_*.
4030 (vec_widen_umult_lo_v8hi): Likewise.
4031 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4032 rather than call gen_vec_widen_smult_*.
4033 (vec_widen_smult_lo_v8hi): Likewise.
4035 2014-01-15 Jeff Law <law@redhat.com>
4037 PR tree-optimization/59747
4038 * ree.c (find_and_remove_re): Properly handle case where a second
4039 eliminated extension requires widening a copy created for elimination
4040 of a prior extension.
4041 (combine_set_extension): Ensure that the number of hard regs needed
4042 for a destination register does not change when we widen it.
4044 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4046 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4047 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4048 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4049 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4050 (avr-*-rtems*): Likewise.
4051 (bfin*-rtems*): Likewise.
4052 (moxie-*-rtems*): Likewise.
4053 (h8300-*-rtems*): Likewise.
4054 (i[34567]86-*-rtems*): Likewise.
4055 (lm32-*-rtems*): Likewise.
4056 (m32r-*-rtems*): Likewise.
4057 (m68k-*-rtems*): Likewise.
4058 (microblaze*-*-rtems*): Likewise.
4059 (mips*-*-rtems*): Likewise.
4060 (powerpc-*-rtems*): Likewise.
4061 (sh-*-rtems*): Likewise.
4062 (sparc-*-rtems*): Likewise.
4063 (sparc64-*-rtems*): Likewise.
4064 (v850-*-rtems*): Likewise.
4065 (m32c-*-rtems*): Likewise.
4067 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4069 PR rtl-optimization/59511
4070 * ira.c (ira_init_register_move_cost): Use memory costs for some
4071 cases of register move cost calculations.
4072 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4073 instead of BB frequency.
4074 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4075 * lra-assigns.c (find_hard_regno_for): Ditto.
4077 2014-01-15 Richard Biener <rguenther@suse.de>
4079 PR tree-optimization/59822
4080 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4081 (vectorizable_load): Use it to hoist defs of uses of invariant
4082 loads out of the loop.
4084 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4085 Kugan Vivekanandarajah <kuganv@linaro.org>
4088 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4091 2014-01-15 Richard Biener <rguenther@suse.de>
4093 PR rtl-optimization/59802
4094 * lcm.c (compute_available): Use inverted postorder to seed
4095 the initial worklist.
4097 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4100 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4101 ADDR_REGS for invalid symrefs in non-PIC code.
4103 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4106 * builtins.c (determine_block_size): Initialize *probable_max_size
4107 even if len_rtx is CONST_INT.
4109 2014-01-14 Andrew Pinski <apinski@cavium.com>
4111 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4112 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4113 (cortexa53_tunings): Likewise.
4114 (aarch64_sched_issue_rate): New function.
4115 (TARGET_SCHED_ISSUE_RATE): Define.
4117 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4119 * ira-costs.c (find_costs_and_classes): Add missed
4120 ira_init_register_move_cost_if_necessary.
4122 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4125 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4127 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4130 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4131 to indicate if type is used for function return value. Warn ABI
4132 change if the vector mode isn't available for function return value.
4133 (ix86_function_arg_advance): Pass false to type_natural_mode.
4134 (ix86_function_arg): Likewise.
4135 (ix86_gimplify_va_arg): Likewise.
4136 (function_arg_32): Don't warn ABI change.
4137 (ix86_function_value): Pass true to type_natural_mode.
4138 (ix86_return_in_memory): Likewise.
4139 (ix86_struct_value_rtx): Removed.
4140 (TARGET_STRUCT_VALUE_RTX): Likewise.
4142 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4144 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4145 converting a conditional jump into a conditional return.
4147 2014-01-14 Richard Biener <rguenther@suse.de>
4149 PR tree-optimization/58921
4150 PR tree-optimization/59006
4151 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4152 hoisting invariant stmts.
4153 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4154 invariant loads on the preheader edge if possible.
4156 2014-01-14 Joey Ye <joey.ye@arm.com>
4158 * doc/plugin.texi (Building GCC plugins): Update to C++.
4160 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4162 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4163 (_mm_rcp28_round_ss): Ditto.
4164 (_mm_rsqrt28_round_sd): Ditto.
4165 (_mm_rsqrt28_round_ss): Ditto.
4166 (_mm_rcp28_sd): Ditto.
4167 (_mm_rcp28_ss): Ditto.
4168 (_mm_rsqrt28_sd): Ditto.
4169 (_mm_rsqrt28_ss): Ditto.
4170 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4171 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4172 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4173 (IX86_BUILTIN_RCP28SD): Ditto.
4174 (IX86_BUILTIN_RCP28SS): Ditto.
4175 (IX86_BUILTIN_RSQRT28SD): Ditto.
4176 (IX86_BUILTIN_RSQRT28SS): Ditto.
4177 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4178 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4179 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4180 (ix86_expand_special_args_builtin): Expand new FTYPE.
4181 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4182 (srcp14<mode>): Make insn unary.
4183 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4184 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4185 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4186 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4187 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4188 Fix rounding: make it SAE only.
4189 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4191 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4193 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4194 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4195 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4196 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4197 (round_saeonly_mask_scalar_operand4): Ditto.
4198 (round_saeonly_mask_scalar_op3): Ditto.
4199 (round_saeonly_mask_scalar_op4): Ditto.
4201 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4203 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4204 Implement -maltivec=be for vec_insert and vec_extract.
4206 2014-01-10 DJ Delorie <dj@redhat.com>
4208 * config/msp430/msp430.md (call_internal): Don't allow memory
4209 references with SP as the base register.
4210 (call_value_internal): Likewise.
4211 * config/msp430/constraints.md (Yc): New. For memory references
4212 that don't use SP as a base register.
4214 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4215 "an integer without a # prefix"
4216 * config/msp430/msp430.md (epilogue_helper): Use it.
4218 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4221 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4222 AVX512F gather builtins.
4223 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4224 on gather decls with INTEGER_TYPE masktype.
4225 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4226 directly into the builtin rather than hoisting it before loop.
4228 PR tree-optimization/59387
4229 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4230 (scev_const_prop): If folded_casts and type has undefined overflow,
4231 use force_gimple_operand instead of force_gimple_operand_gsi and
4232 for each added stmt if it is assign with
4233 arith_code_with_undefined_signed_overflow, call
4234 rewrite_to_defined_overflow.
4235 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4236 gimple-fold.h instead.
4237 (arith_code_with_undefined_signed_overflow,
4238 rewrite_to_defined_overflow): Moved to ...
4239 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4240 rewrite_to_defined_overflow): ... here. No longer static.
4241 Include gimplify-me.h.
4242 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4243 rewrite_to_defined_overflow): New prototypes.
4245 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4247 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4249 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4251 * builtins.c (get_object_alignment_2): Minor tweak.
4252 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4254 2014-01-13 Christian Bruel <christian.bruel@st.com>
4256 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4257 optimized non constant lengths.
4259 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4262 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4263 load as __atomic_load_N if possible.
4265 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4267 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4269 (rs6000_expand_builtin): Adjust call.
4271 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4274 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4275 * config/rs6000/rs6000.c: Include target-globals.h.
4276 (rs6000_set_current_function): Instead of doing target_reinit
4277 unconditionally, use save_target_globals_default_opts and
4278 restore_target_globals.
4280 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4282 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4283 (rs6000_expand_builtin): Handle mffs and mtfsf.
4284 (rs6000_init_builtins): Define mffs and mtfsf.
4285 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4286 (rs6000_mffs): New pattern.
4287 (rs6000_mtfsf): New pattern.
4289 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4291 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4292 Start narrowing with START. Apply candidate-use pair
4293 and check overall cost in narrowing.
4294 (iv_ca_prune): Pass new argument.
4296 2014-01-10 Jeff Law <law@redhat.com>
4299 * ree.c (combine_reaching_defs): Ensure the defining statement
4300 occurs before the extension when optimizing extensions with
4301 different source and destination hard registers.
4303 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4306 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4307 vtables into the type inheritance graph.
4309 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4311 PR rtl-optimization/59754
4312 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4313 modes in the REGNO != REGNO case.
4315 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4317 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4319 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4321 PR tree-optimization/59745
4322 * tree-predcom.c (tree_predictive_commoning_loop): Call
4323 free_affine_expand_cache if giving up because components is NULL.
4325 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4326 GC in payload of target_globals struct instead of allocating them on
4327 the heap and the larger structs separately using GC.
4328 * target-globals.h (struct target_globals): Make regs, hard_regs,
4329 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4330 of GTY((skip)) and change type to void *.
4331 (reset_target_globals): Cast loads from those fields to corresponding
4334 2014-01-10 Steve Ellcey <sellcey@mips.com>
4337 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4338 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4339 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4341 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4344 * aarch64-modes.def (CC_Zmode): New flags mode.
4345 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4346 represents an equality.
4347 (aarch64_get_condition_code): Handle CC_Zmode.
4348 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4350 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4352 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4353 extraction in good case.
4355 2014-01-10 Richard Biener <rguenther@suse.de>
4357 PR tree-optimization/59374
4358 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4359 checking after SLP discovery. Mark stmts not participating
4360 in any SLP instance properly.
4362 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4364 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4365 when handling a SET rtx.
4367 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4369 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4370 (cortex-a57): Likewise.
4371 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4373 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4375 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4376 non-iwmmxt builtins.
4378 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4382 * ipa-devirt.c record_target_from_binfo): Take as argument
4383 stack of binfos and lookup matching one for virtual inheritance.
4384 (possible_polymorphic_call_targets_1): Update.
4386 2014-01-10 Huacai Chen <chenhc@lemote.com>
4388 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4389 kernel strings for Loongson-2E/2F/3A.
4391 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4394 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4395 is_gimple_call before calling gimple_call_internal_p.
4397 2014-01-09 Steve Ellcey <sellcey@mips.com>
4399 * Makefile.in (TREE_FLOW_H): Remove.
4400 (TREE_SSA_H): Add file names from tree-flow.h.
4401 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4402 * tree.h: Remove tree-flow.h reference.
4403 * hash-table.h: Remove tree-flow.h reference.
4404 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4405 reference with tree-ssa-loop.h.
4407 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4409 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4410 default element-order behavior for -maltivec.
4411 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4412 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4413 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4414 when targeting big endian, at least for now.
4415 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4417 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4420 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4421 var satisfies use_register_for_decl, just take into account type
4422 alignment, rather than decl alignment.
4424 PR tree-optimization/59622
4425 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4426 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4427 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4428 Don't devirtualize for inplace at all. For targets.length () == 1,
4429 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4431 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4433 * config/i386/i386.md (cpu): Remove the unused btver1.
4435 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4437 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4439 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4442 * tree-core.h (struct target_globals): New forward declaration.
4443 (struct tree_target_option): Add globals field.
4444 * tree.h (TREE_TARGET_GLOBALS): Define.
4445 (prepare_target_option_nodes_for_pch): New prototype.
4446 * target-globals.h (struct target_globals): Define even if
4448 * tree.c (prepare_target_option_node_for_pch,
4449 prepare_target_option_nodes_for_pch): New functions.
4450 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4451 * config/i386/i386.c: Include target-globals.h.
4452 (ix86_set_current_function): Instead of doing target_reinit
4453 unconditionally, use save_target_globals_default_opts and
4454 restore_target_globals.
4456 2014-01-09 Richard Biener <rguenther@suse.de>
4458 PR tree-optimization/59715
4459 * tree-cfg.h (split_critical_edges): Declare.
4460 * tree-cfg.c (split_critical_edges): Export.
4461 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4463 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4465 * cfgexpand.c (expand_stack_vars): Optionally disable
4466 asan stack protection.
4467 (expand_used_vars): Likewise.
4468 (partition_stack_vars): Likewise.
4469 * asan.c (asan_emit_stack_protection): Optionally disable
4470 after return stack usage.
4471 (instrument_derefs): Optionally disable memory access instrumentation.
4472 (instrument_builtin_call): Likewise.
4473 (instrument_strlen_call): Likewise.
4474 (asan_protect_global): Optionally disable global variables protection.
4475 * doc/invoke.texi: Added doc for new options.
4476 * params.def: Added new options.
4477 * params.h: Likewise.
4479 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4481 PR rtl-optimization/59724
4482 * ifcvt.c (cond_exec_process_if_block): Don't call
4483 flow_find_head_matching_sequence with 0 longest_match.
4484 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4485 non-active insns if !stop_after.
4486 (try_head_merge_bb): Revert 2014-01-07 changes.
4488 2014-01-08 Jeff Law <law@redhat.com>
4490 * ree.c (get_sub_rtx): New function, extracted from...
4491 (merge_def_and_ext): Here.
4492 (combine_reaching_defs): Use get_sub_rtx.
4494 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4496 * cgraph.h (varpool_variable_node): Do not choke on null node.
4498 2014-01-08 Catherine Moore <clm@codesourcery.com>
4500 * config/mips/mips.md (simple_return): Attempt to use JRC
4502 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4504 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4506 PR rtl-optimization/59137
4507 * reorg.c (steal_delay_list_from_target): Call update_block for
4509 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4511 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4513 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4514 two duplicate entries.
4516 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4519 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4521 * config/mips/mips.c (mips_truncated_op_cost): New function.
4522 (mips_rtx_costs): Adjust test for BADDU.
4523 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4525 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4527 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4528 (*baddu_si): ...this new pattern.
4530 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4533 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4535 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4538 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4540 (expand_expr, expand_normal): Adjust.
4541 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4542 inner_reference_p. Use inner_reference_p to expand inner references.
4543 (store_expr): Adjust.
4544 * cfgexpand.c (expand_call_stmt): Adjust.
4546 2014-01-08 Rong Xu <xur@google.com>
4548 * gcov-io.c (gcov_var): Move from gcov-io.h.
4549 (gcov_position): Ditto.
4550 (gcov_is_error): Ditto.
4551 (gcov_rewrite): Ditto.
4552 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4553 only part to libgcc/libgcov.h.
4555 2014-01-08 Marek Polacek <polacek@redhat.com>
4558 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4560 2014-01-08 Marek Polacek <polacek@redhat.com>
4563 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4565 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4567 PR rtl-optimization/59649
4568 * stor-layout.c (get_mode_bounds): For BImode return
4569 0 and STORE_FLAG_VALUE.
4571 2014-01-08 Richard Biener <rguenther@suse.de>
4574 * gimple.h (is_gimple_builtin_call): Remove.
4575 (gimple_builtin_call_types_compatible_p): New.
4576 (gimple_call_builtin_p): New overload.
4577 * gimple.c (is_gimple_builtin_call): Remove.
4578 (validate_call): Rename to ...
4579 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4581 (validate_type): New static function.
4582 (gimple_call_builtin_p): New overload and adjust.
4583 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4584 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4585 (gimple_fold_stmt_to_constant_1): Likewise.
4586 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4588 2014-01-08 Richard Biener <rguenther@suse.de>
4591 * gimplify.c (gimplify_expr): Gimplify register-register type
4592 VIEW_CONVERT_EXPRs to separate stmts.
4594 2014-01-07 Jeff Law <law@redhat.com>
4597 * ree.c (combine_set_extension): Handle case where source
4598 and destination registers in an extension insn are different.
4599 (combine_reaching_defs): Allow source and destination registers
4600 in extension to be different under limited circumstances.
4601 (add_removable_extension): Remove restriction that the
4602 source and destination registers in the extension are the same.
4603 (find_and_remove_re): Emit a copy from the extension's
4604 destination to its source after the defining insn if
4605 the source and destination registers are different.
4608 * ifcvt.c (merge_if_block): If we are merging a block with more than
4609 one successor with a block with no successors, remove any BARRIER
4610 after the second block.
4612 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4614 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4616 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4619 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4620 for 14-bit register offsets when INT14_OK_STRICT is false.
4622 2014-01-07 Roland Stigge <stigge@antcom.de>
4623 Michael Meissner <meissner@linux.vnet.ibm.com>
4626 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4627 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4629 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4631 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4634 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4636 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4637 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4638 rtx is const0_rtx or not.
4640 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4643 * target-globals.c (save_target_globals): Remove this_fn_optab
4645 * toplev.c: Include optabs.h.
4646 (target_reinit): Temporarily restore the global options if another
4647 set of options are in force.
4649 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4651 PR rtl-optimization/58668
4652 * cfgcleanup.c (flow_find_cross_jump): Don't count
4653 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4654 to determine what is counted.
4655 (flow_find_head_matching_sequence): Use active_insn_p to determine
4657 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4659 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4660 determine what is counted.
4662 PR tree-optimization/59643
4663 * tree-predcom.c (split_data_refs_to_components): If one dr is
4664 read and one write, determine_offset fails and the write isn't
4665 in the bad component, just put the read into the bad component.
4667 2014-01-07 Mike Stump <mikestump@comcast.net>
4668 Jakub Jelinek <jakub@redhat.com>
4671 * tree-core.h (struct tree_optimization_option): Change optabs
4672 type from unsigned char * to void *.
4673 * optabs.c (init_tree_optimization_optabs): Adjust
4674 TREE_OPTIMIZATION_OPTABS initialization.
4676 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4679 * config/i386/i386.h (struct machine_function): Add
4680 no_drap_save_restore field.
4681 * config/i386/i386.c (ix86_save_reg): Use
4682 !cfun->machine->no_drap_save_restore instead of
4683 crtl->stack_realign_needed.
4684 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4685 this function clears frame_pointer_needed. Set
4686 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4687 and DRAP reg is needed.
4689 2014-01-06 Marek Polacek <polacek@redhat.com>
4692 * doc/implement-c.texi: Mention that other integer types are
4693 permitted as bit-field types in strictly conforming mode.
4695 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4697 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4700 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4702 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4704 2014-01-06 Martin Jambor <mjambor@suse.cz>
4707 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4709 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4711 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4715 * var-tracking.c (add_stores): Preserve the value of the source even if
4716 we don't record the store.
4718 2014-01-06 Terry Guo <terry.guo@arm.com>
4720 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4722 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4725 * config/darwin.c (darwin_function_section): Adjust return values to
4726 correspond to optimisation changes made in r206070.
4728 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4730 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4731 from prefetch_block tune setting.
4732 (nocona_cost): Correct size of prefetch block to 64.
4734 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4736 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4737 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4738 used to save the static chain register in the computation of the offset
4739 from which the FP registers need to be restored.
4741 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4743 PR tree-optimization/59519
4744 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4745 ICE if get_current_def (current_new_name) is already non-NULL, as long
4746 as it is a phi result of some other phi in *new_exit_bb that has
4749 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4750 or vmovdqu* for misaligned_operand.
4751 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4752 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4753 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4754 aligned_mem for AVX512F masked aligned load and store builtins and for
4757 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4759 PR tree-optimization/59651
4760 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4761 Address range for negative step should be added by TYPE_SIZE_UNIT.
4763 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4765 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4766 overlapping registers also for registers other than the stack pointer.
4768 2014-01-03 Marek Polacek <polacek@redhat.com>
4771 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4774 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4777 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4780 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4781 (push splitter): Use <P:MODE_SIZE> instead of
4782 GET_MODE_SIZE (<P:MODE>mode).
4783 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4784 (mov -1, reg peephole2): Likewise.
4785 * config/i386/sse.md (*mov<mode>_internal,
4786 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4787 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4788 *<code><mode>3, *andnot<mode>3<mask_name>,
4789 <mask_codefor><code><mode>3<mask_name>): Likewise.
4790 * config/i386/subst.md (mask_mode512bit_condition,
4791 sd_mask_mode512bit_condition): Likewise.
4793 2014-01-02 Xinliang David Li <davidxl@google.com>
4795 PR tree-optimization/59303
4796 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4797 (dump_predicates): Better output format.
4798 (pred_equal_p): New function.
4799 (is_neq_relop_p): Ditto.
4800 (is_neq_zero_form_p): Ditto.
4801 (pred_expr_equal_p): Ditto.
4802 (pred_neg_p): Ditto.
4803 (simplify_pred): Ditto.
4804 (simplify_preds_2): Ditto.
4805 (simplify_preds_3): Ditto.
4806 (simplify_preds_4): Ditto.
4807 (simplify_preds): Ditto.
4809 (push_to_worklist): Ditto.
4810 (get_pred_info_from_cmp): Ditto.
4811 (is_degenerated_phi): Ditto.
4812 (normalize_one_pred_1): Ditto.
4813 (normalize_one_pred): Ditto.
4814 (normalize_one_pred_chain): Ditto.
4815 (normalize_preds): Ditto.
4816 (normalize_cond_1): Remove function.
4817 (normalize_cond): Ditto.
4818 (is_gcond_subset_of): Ditto.
4819 (is_subset_of_any): Ditto.
4820 (is_or_set_subset_of): Ditto.
4821 (is_and_set_subset_of): Ditto.
4822 (is_norm_cond_subset_of): Ditto.
4823 (pred_chain_length_cmp): Ditto.
4824 (convert_control_dep_chain_into_preds): Type change.
4825 (find_predicates): Ditto.
4826 (find_def_preds): Ditto.
4827 (destroy_predicates_vecs): Ditto.
4828 (find_matching_predicates_in_rest_chains): Ditto.
4829 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4830 (is_pred_expr_subset): Ditto.
4831 (is_pred_chain_subset_of): Ditto.
4832 (is_included_in): Ditto.
4833 (is_superset_of): Ditto.
4835 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4837 Update copyright years.
4839 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4841 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4842 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4843 config/arc/arc.md, config/arc/arc.opt,
4844 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4845 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4846 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4847 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4848 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4849 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4850 vtable-verify.c, vtable-verify.h: Use the standard form for the
4853 2014-01-02 Tobias Burnus <burnus@net-b.de>
4855 * gcc.c (process_command): Update copyright notice dates.
4856 * gcov-dump.c: Ditto.
4858 * doc/cpp.texi: Bump @copying's copyright year.
4859 * doc/cppinternals.texi: Ditto.
4860 * doc/gcc.texi: Ditto.
4861 * doc/gccint.texi: Ditto.
4862 * doc/gcov.texi: Ditto.
4863 * doc/install.texi: Ditto.
4864 * doc/invoke.texi: Ditto.
4866 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4868 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4870 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4872 * config/i386/sse.md (*mov<mode>_internal): Guard
4873 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4875 PR rtl-optimization/59647
4876 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4877 new_rtx into UNSIGNED_FLOAT rtxes.
4879 Copyright (C) 2014 Free Software Foundation, Inc.
4881 Copying and distribution of this file, with or without modification,
4882 are permitted in any medium without royalty provided the copyright
4883 notice and this notice are preserved.