1 2014-03-13 Richard Henderson <rth@redhat.com>
4 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
5 (ix86_force_to_memory, ix86_free_from_memory): Remove.
6 * config/i386/i386-protos.h: Likewise.
7 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
8 in the expander instead of a splitter.
9 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
10 any possibility of requiring a memory.
11 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
12 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
13 (fp branch splitters): Update for ix86_split_fp_branch.
14 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
15 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
16 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
17 (*fop_<MODEF>_2_i387): Remove f/r alternative.
18 (*fop_<MODEF>_3_i387): Likewise.
19 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
20 (splitters for the fop_* register patterns): Remove.
21 (fscalexf4_i387): Rename from *fscalexf4_i387.
22 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
24 2014-03-13 Jakub Jelinek <jakub@redhat.com>
26 PR tree-optimization/59779
27 * tree-dfa.c (get_ref_base_and_extent): Use double_int
28 type for bitsize and maxsize instead of HOST_WIDE_INT.
30 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
32 PR rtl-optimization/57320
33 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
34 the CFG after thread_prologue_and_epilogue_insns.
36 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
38 PR rtl-optimization/57189
39 * lra-constraints.c (process_alt_operands): Disfavor spilling
42 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
44 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
47 2014-03-13 Jakub Jelinek <jakub@redhat.com>
49 PR tree-optimization/59025
51 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
52 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
54 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
57 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
58 calls of avr_out_plus_1.
60 2014-03-13 Bin Cheng <bin.cheng@arm.com>
62 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
63 BB's single pred and update the father loop's latch info later.
65 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
67 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
72 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
73 registers, we need to swap double words in little endian mode.
75 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
76 to be a container mode for 128-bit integer operations added in ISA
77 2.07. Unlike TImode and PTImode, the preferred register set is
78 the Altivec/VMX registers for the 128-bit operations.
80 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
82 (rs6000_split_128bit_ok_p): Likewise.
84 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
85 macros for creating ISA 2.07 normal and overloaded builtin
86 functions with 3 arguments.
87 (BU_P8V_OVERLOAD_3): Likewise.
88 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
89 for use as overloaded functions.
90 (VPERM_1TI_UNS): Likewise.
92 (VSEL_1TI_UNS): Likewise.
93 (ST_INTERNAL_1ti): Likewise.
94 (LD_INTERNAL_1ti): Likewise.
95 (XXSEL_1TI): Likewise.
96 (XXSEL_1TI_UNS): Likewise.
97 (VPERM_1TI): Likewise.
98 (VPERM_1TI_UNS): Likewise.
99 (XXPERMDI_1TI): Likewise.
101 (LXVD2X_V1TI): Likewise.
102 (STXVD2X_V1TI): Likewise.
103 (VEC_INIT_V1TI): Likewise.
104 (VEC_SET_V1TI): Likewise.
105 (VEC_EXT_V1TI): Likewise.
106 (EQV_V1TI): Likewise.
107 (NAND_V1TI): Likewise.
108 (ORC_V1TI): Likewise.
109 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
110 added in ISA 2.07. Add both normal 'altivec' builtins, and the
114 (VADDEUQM): Likewise.
115 (VADDECUQ): Likewise.
116 (VSUBEUQM): Likewise.
117 (VSUBECUQ): Likewise.
119 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
120 __int128_t and __uint128_t types.
121 (__uint128_type): Likewise.
122 (altivec_categorize_keyword): Add support for vector __int128_t,
123 vector __uint128_t, vector __int128, and vector unsigned __int128
124 as a container type for TImode operations that need to be done in
125 VSX/Altivec registers.
126 (rs6000_macro_to_expand): Likewise.
127 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
128 to support 128-bit integer instructions vaddcuq, vadduqm,
129 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
130 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
132 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
133 for V1TImode, and set up preferences to use VSX/Altivec registers.
134 Setup VSX reload handlers.
135 (rs6000_debug_reg_global): Likewise.
136 (rs6000_init_hard_regno_mode_ok): Likewise.
137 (rs6000_preferred_simd_mode): Likewise.
138 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
139 (easy_altivec_constant): Likewise.
140 (output_vec_const_move): Likewise.
141 (rs6000_expand_vector_set): Convert V1TImode set and extract to
143 (rs6000_expand_vector_extract): Likewise.
144 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
146 (rs6000_const_vec): Add support for V1TImode.
147 (rs6000_emit_le_vsx_load): Swap double words when loading or
148 storing TImode/V1TImode.
149 (rs6000_emit_le_vsx_store): Likewise.
150 (rs6000_emit_le_vsx_move): Likewise.
151 (rs6000_emit_move): Add support for V1TImode.
152 (altivec_expand_ld_builtin): Likewise.
153 (altivec_expand_st_builtin): Likewise.
154 (altivec_expand_vec_init_builtin): Likewise.
155 (altivec_expand_builtin): Likewise.
156 (rs6000_init_builtins): Add support for V1TImode type. Add
157 support for ISA 2.07 128-bit integer builtins. Define type names
158 for the VSX/Altivec vector types.
159 (altivec_init_builtins): Add support for overloaded vector
160 functions with V1TImode type.
161 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
162 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
164 (rs6000_split_128bit_ok_p): Likewise.
165 (rs6000_handle_altivec_attribute): Create V1TImode from vector
166 __int128_t and vector __uint128_t.
168 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
176 (VS_scalar): Likewise.
177 (VS_double): Likewise.
178 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
180 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
181 we support the ISA 2.07 128-bit integer arithmetic instructions.
182 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
183 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
184 and TImode types for use with the builtin functions.
185 (V1TI_type_node): Likewise.
186 (unsigned_V1TI_type_node): Likewise.
187 (intTI_type_internal_node): Likewise.
188 (uintTI_type_internal_node): Likewise.
190 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
191 128-bit builtin functions.
192 (UNSPEC_VADDEUQM): Likewise.
193 (UNSPEC_VADDECUQ): Likewise.
194 (UNSPEC_VSUBCUQ): Likewise.
195 (UNSPEC_VSUBEUQM): Likewise.
196 (UNSPEC_VSUBECUQ): Likewise.
197 (VM): Add V1TImode to vector mode iterators.
200 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
201 (altivec_vaddcuq): Likewise.
202 (altivec_vsubuqm): Likewise.
203 (altivec_vsubcuq): Likewise.
204 (altivec_vaddeuqm): Likewise.
205 (altivec_vaddecuq): Likewise.
206 (altivec_vsubeuqm): Likewise.
207 (altivec_vsubecuq): Likewise.
209 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
211 (BOOL_128): Likewise.
212 (BOOL_REGS_OUTPUT): Likewise.
213 (BOOL_REGS_OP1): Likewise.
214 (BOOL_REGS_OP2): Likewise.
215 (BOOL_REGS_UNARY): Likewise.
216 (BOOL_REGS_AND_CR0): Likewise.
218 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
219 128-bit integer builtin support.
220 (vec_vadduqm): Likewise.
221 (vec_vaddecuq): Likewise.
222 (vec_vaddeuqm): Likewise.
223 (vec_vsubecuq): Likewise.
224 (vec_vsubeuqm): Likewise.
225 (vec_vsubcuq): Likewise.
226 (vec_vsubuqm): Likewise.
228 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
229 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
230 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
231 128-bit integer add/subtract to ISA 2.07.
233 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
235 * config/arc/arc.c (arc_predicate_delay_insns):
236 Fix third argument passed to conditionalize_nonjump.
238 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
240 * config/aarch64/aarch64-builtins.c
241 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
242 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
243 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
244 instead of __builtin_lfloor.
245 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
247 2014-03-12 Jakub Jelinek <jakub@redhat.com>
249 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
250 (tree_ssa_ifcombine_bb_1): New function.
251 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
252 is an empty forwarder block to then_bb or vice versa and then_bb
253 and else_bb are effectively swapped.
255 2014-03-12 Christian Bruel <christian.bruel@st.com>
258 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
259 REG_CFA_DEF_CFA note.
260 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
261 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
263 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
265 PR tree-optimization/60454
266 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
268 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
270 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
271 Do not define target_cpu_default2 to generic.
272 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
273 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
274 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
276 2014-03-12 Jakub Jelinek <jakub@redhat.com>
277 Marc Glisse <marc.glisse@inria.fr>
279 PR tree-optimization/60502
280 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
281 instead of build_low_bits_mask.
283 2014-03-12 Jakub Jelinek <jakub@redhat.com>
286 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
287 if there are multiple uses, but op doesn't live on E edge.
288 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
289 clobber stmts before __builtin_unreachable.
291 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
293 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
294 hard_frame_pointer_rtx.
295 * cse.c (cse_insn): Remove volatile check.
296 * cselib.c (cselib_process_insn): Likewise.
297 * dse.c (scan_insn): Likewise.
299 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
301 * config/arc/arc.c (conditionalize_nonjump): New function,
303 (arc_ifcvt): ... this.
304 (arc_predicate_delay_insns): Use it.
306 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
308 * config/arc/predicates.md (extend_operand): During/after reload,
309 allow const_int_operand.
310 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
311 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
312 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
314 (umulsi3_highpart_i): Likewise.
316 2014-03-11 Richard Biener <rguenther@suse.de>
318 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
319 Add asserts to guard possible wrong-code bugs.
321 2014-03-11 Richard Biener <rguenther@suse.de>
323 PR tree-optimization/60429
324 PR tree-optimization/60485
325 * tree-ssa-structalias.c (set_union_with_increment): Properly
326 take into account all fields that overlap the shifted vars.
327 (do_sd_constraint): Likewise.
328 (do_ds_constraint): Likewise.
329 (get_constraint_for_ptr_offset): Likewise.
331 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
333 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
334 (nios2_compute_frame_layout):
335 Add calculation of cfun->machine->fp_save_offset.
336 (nios2_expand_prologue): Correct setting of frame pointer register
338 (nios2_expand_epilogue): Update recovery of stack pointer from
339 frame pointer accordingly.
340 (nios2_initial_elimination_offset): Update calculation of offset
341 for eliminating to HARD_FRAME_POINTER_REGNUM.
343 2014-03-10 Jakub Jelinek <jakub@redhat.com>
346 * ipa.c (symtab_remove_unreachable_nodes): Don't call
347 cgraph_get_create_node on VAR_DECLs.
349 2014-03-10 Richard Biener <rguenther@suse.de>
352 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
354 2014-03-08 Douglas B Rupp <rupp@gnat.com>
356 * config/vms/vms.opt (vms_float_format): New variable.
358 2014-03-08 Tobias Burnus <burnus@net-b.de>
360 * doc/invoke.texi (-fcilkplus): Update implementation status.
362 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
363 Richard Biener <rguenther@suse.de>
365 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
366 consistently accross all TUs.
367 (run_gcc): Enable -fshort-double automatically at link at link-time
368 and disallow override.
370 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
373 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
374 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
375 if they can't be used.
377 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
379 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
380 for Solaris 11/x86 ld.
381 * configure: Regenerate.
383 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
385 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
386 (LIB_TLS_SPEC): Save as ld_tls_libs.
387 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
388 (HAVE_AS_IX86_TLSLDM): New test.
389 * configure, config.in: Regenerate.
390 * config/i386/i386.c (legitimize_tls_address): Fall back to
391 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
392 cannot support TLS_MODEL_LOCAL_DYNAMIC.
393 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
394 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
396 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
398 * common.opt (fira-loop-pressure): Mark as optimization.
400 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
402 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
403 an OpenMP mappable type.
405 2014-03-06 Matthias Klose <doko@ubuntu.com>
407 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
408 MULTILIB_OSDIRNAMES is not defined.
410 2014-03-06 Jakub Jelinek <jakub@redhat.com>
411 Meador Inge <meadori@codesourcery.com>
414 * config/arm/arm.c (arm_tls_symbol_p): Remove.
415 (arm_legitimize_address): Call legitimize_tls_address for any
416 arm_tls_referenced_p expression, handle constant addend. Call it
417 before testing for !TARGET_ARM.
418 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
420 2014-03-06 Richard Biener <rguenther@suse.de>
426 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
428 * tree-streamer.c (record_common_node): Assert we don't record
429 nodes with type double.
430 (preload_common_node): Skip type double, complex double and double
431 pointer since it is now frontend dependent due to fshort-double option.
433 2014-03-06 Richard Biener <rguenther@suse.de>
435 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
436 or -fno-lto is specified and the linker has full plugin support.
437 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
438 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
439 * lto-wrapper.c (merge_and_complain): Merge compile-time
441 (run_gcc): And pass it through to the link options.
443 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
447 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
449 * cselib.c (remove_useless_values): Skip to avoid quadratic
450 behavior if the condition moved from...
451 (cselib_process_insn): ... here holds.
453 2014-03-05 Jakub Jelinek <jakub@redhat.com>
456 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
457 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
460 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
461 (TM_H): Add x86-tune.def.
463 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
465 * config/aarch64/aarch64.c (generic_tunings):
466 Use cortexa57_extra_costs.
468 2014-03-05 Jakub Jelinek <jakub@redhat.com>
471 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
472 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
473 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
476 2014-03-04 Heiher <r@hev.cc>
478 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
479 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
481 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
483 * config/i386/predicates.md (const2356_operand): Change to ...
484 (const2367_operand): ... this.
485 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
487 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
488 (*avx512pf_scatterpf<mode>sf): Ditto.
489 (avx512pf_scatterpf<mode>df): Ditto.
490 (*avx512pf_scatterpf<mode>df_mask): Ditto.
491 (*avx512pf_scatterpf<mode>df): Ditto.
492 * config/i386/i386.c (ix86_expand_builtin): Update
493 incorrect hint operand error message.
495 2014-03-04 Richard Biener <rguenther@suse.de>
497 * lto-section-in.c (lto_get_section_data): Fix const cast.
499 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
501 * tree-streamer.c (record_common_node): Assert we don't record
502 nodes with type double.
503 (preload_common_node): Skip type double, complex double and double
504 pointer since it is now frontend dependent due to fshort-double option.
506 2014-03-04 Richard Biener <rguenther@suse.de>
509 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
510 (lto_input_toplevel_asms): Likewise.
511 * lto-section-in.c (lto_get_section_data): Instead do it here
514 2014-03-04 Richard Biener <rguenther@suse.de>
516 PR tree-optimization/60382
517 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
518 dead PHIs a reduction.
520 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
522 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
524 (_mm_prefetch): Move out of GCC target("sse") pragma.
525 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
526 GCC target("prfchw") pragma.
527 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
529 * config/i386/i386.c (ix86_option_override_internal): Enable
530 -mprfchw with -mprefetchwt1.
532 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
534 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
537 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
539 * opts.h (CL_PCH_IGNORE): Define.
540 * targhooks.c (option_affects_pch_p):
541 Return false for options that have CL_PCH_IGNORE set.
542 * opt-functions.awk: Process PchIgnore.
543 * doc/options.texi: Document PchIgnore.
545 * config/arc/arc.opt (misize): Add PchIgnore property.
547 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
549 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
550 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
551 constraint on constants to permit them being loaded into
552 GENERAL_REGS or BASE_REGS.
554 2014-03-03 Nick Clifton <nickc@redhat.com>
556 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
557 anti-cacnonical alternatives.
558 (negandhi3_real): New pattern.
559 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
561 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
563 * config/avr/avr-mcus.def: Remove atxmega16x1.
564 * config/avr/avr-tables.opt: Regenerate.
565 * config/avr/t-multilib: Regenerate.
566 * doc/avr-mmcu.texi: Regenerate.
568 2014-03-03 Tobias Grosser <tobias@grosser.es>
569 Mircea Namolaru <mircea.namolaru@inria.fr>
571 PR tree-optimization/58028
572 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
575 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
577 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
578 not handled by recognizers.
580 2014-03-03 Jakub Jelinek <jakub@redhat.com>
583 * function.c (expand_function_end): Don't emit
584 clobber_return_register sequence if clobber_after is a BARRIER.
585 * cfgexpand.c (construct_exit_block): Append instructions before
586 return_label to prev_bb.
588 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
590 * config/rs6000/constraints.md: Document reserved use of "wc".
592 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
595 * ipa.c (function_and_variable_visibility): When dissolving comdat
596 group, also set all symbols to local.
598 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
603 2013-12-14 Jan Hubicka <jh@suse.cz>
605 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
607 2014-03-02 Jon Beniston <jon@beniston.com>
613 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
614 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
615 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
616 (simple_return, *simple_return): New patterns
617 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
618 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
620 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
622 * dwarf2out.c (gen_subprogram_die): Tidy.
624 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
627 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
628 (*mov_t_msb_neg_negc): ... this new insn.
630 2014-02-28 Jason Merrill <jason@redhat.com>
633 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
636 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
639 * dwarf2out.c (decltype_auto_die): New static.
640 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
641 (gen_type_die_with_usage): Handle 'decltype(auto)'.
642 (is_cxx_auto): Likewise.
644 2014-02-28 Ian Bolton <ian.bolton@arm.com>
646 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
647 we are not using general regs only.
649 2014-02-28 Richard Biener <rguenther@suse.de>
652 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
653 previous fix and only allow to remove trivial pre-headers
654 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
655 (remove_forwarder_block): Properly update the latch of a loop.
657 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
660 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
661 (cselib_preserved_hash_table): New.
662 (preserve_constants_and_equivs): Move preserved vals to it.
663 (cselib_find_slot): Look it up first.
664 (cselib_init): Initialize it.
665 (cselib_finish): Release it.
666 (dump_cselib_table): Dump it.
668 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
671 * cselib.c (remove_useless_values): Skip to avoid quadratic
672 behavior if the condition moved from...
673 (cselib_process_insn): ... here holds.
675 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
678 * var-tracking.c (vt_initialize): Apply the same condition to
679 preserve the CFA base value.
681 2014-02-28 Joey Ye <joey.ye@arm.com>
684 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
685 if reload in progress or completed.
687 2014-02-28 Tobias Burnus <burnus@net-b.de>
690 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
693 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
695 * doc/tm.texi.in (Condition Code Status): Update documention for
696 relative locations of cc0-setter and cc0-user.
698 2014-02-27 Jeff Law <law@redhat.com>
700 PR rtl-optimization/52714
701 * combine.c (try_combine): When splitting an unrecognized PARALLEL
702 into two independent simple sets, if I3 is a jump, ensure the
703 pattern we place into I3 is a (set (pc) ...).
705 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
706 Jeff Law <law@redhat.com>
708 PR rtl-optimization/49847
709 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
710 are in different blocks.
711 * doc/tm.texi (Condition Code Status): Update documention for
712 relative locations of cc0-setter and cc0-user.
714 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
717 * lra.c (lra_emit_add): Check SUBREG too.
719 2014-02-27 Andreas Schwab <schwab@suse.de>
721 * config/m68k/m68k.c (m68k_option_override): Disable
722 -flive-range-shrinkage for classic m68k.
723 (m68k_override_options_after_change): Likewise.
725 2014-02-27 Marek Polacek <polacek@redhat.com>
728 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
729 -Wmaybe-uninitialized.
731 2014-02-27 Alan Modra <amodra@gmail.com>
734 * reload1.c (emit_input_reload_insns): When reload_override_in,
735 set old to rl->in_reg when rl->in_reg is a subreg.
737 2014-02-26 Richard Biener <rguenther@suse.de>
740 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
742 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
744 * common/config/i386/predicates.md (const1256_operand): Remove.
745 (const2356_operand): New.
746 (const_1_to_2_operand): Remove.
747 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
748 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
749 (*avx512pf_gatherpf<mode>sf): Ditto.
750 (avx512pf_gatherpf<mode>df): Ditto.
751 (*avx512pf_gatherpf<mode>df_mask): Ditto.
752 (*avx512pf_gatherpf<mode>df): Ditto.
753 (avx512pf_scatterpf<mode>sf): Ditto.
754 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
755 (*avx512pf_scatterpf<mode>sf): Ditto.
756 (avx512pf_scatterpf<mode>df): Ditto.
757 (*avx512pf_scatterpf<mode>df_mask): Ditto.
758 (*avx512pf_scatterpf<mode>df): Ditto.
759 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
761 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
763 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
764 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
765 (_mm512_mask_testn_epi64_mask): Move to ...
766 * config/i386/avx512cdintrin.h: Here.
767 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
768 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
769 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
770 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
771 TARGET_AVX512F from TARGET_AVX512CD.
773 2014-02-26 Richard Biener <rguenther@suse.de>
776 * ipa.c (walk_polymorphic_call_targets): Properly guard
777 call to inline_update_overall_summary.
779 2014-02-26 Bin Cheng <bin.cheng@arm.com>
782 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
783 and latches only if requested. Fix latch if it is removed.
784 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
785 LOOPS_HAVE_PREHEADERS.
787 2014-02-25 Andrew Pinski <apinski@cavium.com>
789 * builtins.c (expand_builtin_thread_pointer): Create a new target
790 when the target is NULL.
792 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
794 PR rtl-optimization/60317
795 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
796 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
797 * lra-assigns.c: Include params.h.
798 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
799 other reload pseudos considerations.
801 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
803 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
804 to use canonical form for nor<mode>3.
806 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
809 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
812 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
814 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
815 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
816 (ix86_handle_option): Handle OPT_mprefetchwt1.
817 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
818 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
820 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
821 OPTION_MASK_ISA_PREFETCHWT1.
822 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
823 (PTA_PREFETCHWT1): New.
824 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
825 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
826 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
827 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
828 (*prefetch_avx512pf_<mode>_: Change into ...
829 (*prefetch_prefetchwt1_<mode>: This.
830 * config/i386/i386.opt (mprefetchwt1): New.
831 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
832 (_mm_prefetch): Handle intent to write.
833 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
835 2014-02-25 Richard Biener <rguenther@suse.de>
838 * emit-rtl.c (mem_attrs_htab): Remove.
839 (mem_attrs_htab_hash): Likewise.
840 (mem_attrs_htab_eq): Likewise.
841 (set_mem_attrs): Always allocate new mem-attrs when something changed.
842 (init_emit_once): Do not allocate mem_attrs_htab.
844 2014-02-25 Richard Biener <rguenther@suse.de>
847 * lto-opts.c (lto_write_options): Output non-explicit conservative
848 -fwrapv, -fno-trapv and -fno-strict-overflow.
849 * lto-wrapper.c (merge_and_complain): Handle merging those options.
850 (run_gcc): And pass them through.
852 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
854 * sel-sched.c (calculate_new_fences): New parameter ptime.
855 Calculate it as a maximum over all fence cycles.
856 (sel_sched_region_2): Adjust the call to calculate_new_fences.
857 Print the final schedule timing when sched_verbose.
859 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
861 PR rtl-optimization/60292
862 * sel-sched.c (fill_vec_av_set): Do not reset target availability
863 bit fot the fence instruction.
865 2014-02-24 Alangi Derick <alangiderick@gmail.com>
867 * calls.h: Fix typo in comment.
869 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
871 * config/pa/pa.c (pa_output_move_double): Don't valididate when
872 adjusting offsetable addresses.
874 2014-02-24 Guozhi Wei <carrot@google.com>
876 * sparseset.h (sparseset_pop): Fix the wrong index.
878 2014-02-24 Walter Lee <walt@tilera.com>
880 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
881 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
883 * common/config/tilegx/tilegx-common.c
884 (TARGET_DEFAULT_TARGET_FLAGS): Define.
885 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
887 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
888 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
889 (tilegx_gimplify_va_arg_expr): Handle big endian.
890 (tilegx_expand_unaligned_load): Ditto.
891 (tilegx_expand_unaligned_store): Ditto.
892 (TARGET_RETURN_IN_MSB): New.
893 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
894 (TARGET_ENDIAN_DEFAULT): New.
895 (TARGET_BIG_ENDIAN): Handle big endian.
896 (BYTES_BIG_ENDIAN): Ditto.
897 (WORDS_BIG_ENDIAN): Ditto.
898 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
901 * config/tilegx/tilegx.md (extv): Handle big endian.
904 (insn_st<n>_add<bitsuffix>): Ditto.
905 (insn_stnt<n>): Ditto.
906 (insn_stnt<n>_add<bitsuffix>):Ditto.
907 (vec_interleave_highv8qi): Handle big endian.
908 (vec_interleave_highv8qi_be): New.
909 (vec_interleave_highv8qi_le): New.
910 (insn_v1int_h): Handle big endian.
911 (vec_interleave_lowv8qi): Handle big endian.
912 (vec_interleave_lowv8qi_be): New.
913 (vec_interleave_lowv8qi_le): New.
914 (insn_v1int_l): Handle big endian.
915 (vec_interleave_highv4hi): Handle big endian.
916 (vec_interleave_highv4hi_be): New.
917 (vec_interleave_highv4hi_le): New.
918 (insn_v2int_h): Handle big endian.
919 (vec_interleave_lowv4hi): Handle big endian.
920 (vec_interleave_lowv4hi_be): New.
921 (vec_interleave_lowv4hi_le): New.
922 (insn_v2int_l): Handle big endian.
923 (vec_interleave_highv2si): Handle big endian.
924 (vec_interleave_highv2si_be): New.
925 (vec_interleave_highv2si_le): New.
926 (insn_v4int_h): Handle big endian.
927 (vec_interleave_lowv2si): Handle big endian.
928 (vec_interleave_lowv2si_be): New.
929 (vec_interleave_lowv2si_le): New.
930 (insn_v4int_l): Handle big endian.
931 * config/tilegx/tilegx.opt (mbig-endian): New option.
932 (mlittle-endian): New option.
933 * doc/install.texi: Document tilegxbe-linux.
934 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
936 2014-02-24 Martin Jambor <mjambor@suse.cz>
939 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
940 there are no parameter descriptors.
942 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
944 PR rtl-optimization/60268
945 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
946 initialization to ...
947 (sched_rgn_init): ... here.
948 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
950 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
952 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
955 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
957 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
960 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
962 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
963 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
965 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
967 * config/microblaze/predicates.md: Add cmp_op predicate.
968 * config/microblaze/microblaze.md: Add branch_compare instruction
969 which uses cmp_op predicate and emits cmp insn before branch.
970 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
971 to microblaze_expand_conditional_branch and consolidate logic.
972 (microblaze_expand_conditional_branch): emit branch_compare
973 insn instead of handling cmp op separate from branch insn.
975 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
977 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
980 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
982 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
983 define_insn with define_expand and new define_insn
984 *altivec_lve<VI_char>x_internal.
985 (altivec_stve<VI_char>x): Replace define_insn with define_expand
986 and new define_insn *altivec_stve<VI_char>x_internal.
987 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
989 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
991 (altivec_expand_stvex_be): New function.
993 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
995 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
996 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
997 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
998 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1000 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1003 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1004 instead of emit_move_insn.
1006 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1008 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1010 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1011 gen_altivec_vsumsws.
1013 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1015 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1016 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1017 (altivec_lvxl_<mode>): New define_expand incorporating
1018 -maltivec=be semantics where needed.
1019 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1020 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1021 semantics where needed.
1022 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1023 (altivec_stvx_<mode>): New define_expand incorporating
1024 -maltivec=be semantics where needed.
1025 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1026 VM2 iterator instead of V4SI.
1027 (altivec_stvxl_<mode>): New define_expand incorporating
1028 -maltivec=be semantics where needed.
1029 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1030 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1031 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1032 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1033 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1034 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1035 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1036 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1037 ALTIVEC_BUILTIN_STVXL.
1038 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1039 (altivec_expand_stvx_be): Likewise.
1040 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1041 (altivec_expand_lvx_be): Likewise.
1042 (altivec_expand_stvx_be): Likewise.
1043 (altivec_expand_builtin): Add cases for
1044 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1045 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1046 (altivec_init_builtins): Add definitions for
1047 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1048 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1050 2014-02-21 Catherine Moore <clm@codesourcery.com>
1052 * doc/invoke.texi (mvirt, mno-virt): Document.
1053 * config/mips/mips.opt (mvirt): New option.
1054 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1056 2014-02-21 Richard Biener <rguenther@suse.de>
1058 PR tree-optimization/60276
1059 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1060 (STMT_VINFO_MIN_NEG_DIST): New macro.
1061 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1062 STMT_VINFO_MIN_NEG_DIST.
1063 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1064 made for negative dependence distances still hold.
1066 2014-02-21 Richard Biener <rguenther@suse.de>
1069 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1070 DECL_INITIAL for globals not in the current function context.
1072 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1074 PR tree-optimization/56490
1075 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1076 * tree-ssa-uninit.c: Include params.h.
1077 (compute_control_dep_chain): Add num_calls argument, return false
1078 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1079 num_calls to recursive call.
1080 (find_predicates): Change dep_chain into normal array,
1081 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1082 variable and adjust compute_control_dep_chain caller.
1083 (find_def_preds): Likewise.
1085 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1087 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1088 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1090 2014-02-21 Nick Clifton <nickc@redhat.com>
1092 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1093 (pushhi1): Likewise.
1094 (popqi1): Add mode to pre_dec.
1097 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1099 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1100 mode for mask of V8SFmode permutation.
1102 2014-02-20 Richard Henderson <rth@redhat.com>
1105 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1106 a new pseudo for OLDVAL.
1108 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1111 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1112 gen_reg_rtx if d->testing_p.
1113 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1114 if d->testing_p and we will certainly return true.
1115 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1118 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1120 * emit-rtl.c (gen_reg_rtx): Assert that
1121 crtl->emit.regno_pointer_align_length is non-zero.
1123 2014-02-20 Richard Henderson <rth@redhat.com>
1126 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1127 on failure the store back into EXPECT.
1129 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1130 Sandra Loosemore <sandra@codesourcery.com>
1132 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1133 * config/nios2/nios2.c (nios2_function_profiler): Add
1134 -fPIC (flag_pic == 2) support.
1135 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1136 (nios2_large_offset_p): New function.
1137 (nios2_unspec_reloc_p): Move up position, update to use
1138 nios2_large_offset_p.
1139 (nios2_unspec_address): Remove function.
1140 (nios2_unspec_offset): New function.
1141 (nios2_large_got_address): New function.
1142 (nios2_got_address): Add large offset support.
1143 (nios2_legitimize_tls_address): Update usage of removed and new
1145 (nios2_symbol_binds_local_p): New function.
1146 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1147 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1148 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1149 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1150 processing for (const (unspec ...)).
1151 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1153 2014-02-20 Richard Biener <rguenther@suse.de>
1155 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1156 doing the substitution.
1157 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1159 2014-02-20 Martin Jambor <mjambor@suse.cz>
1162 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1163 info when checking whether lattices are bottom.
1165 2014-02-20 Richard Biener <rguenther@suse.de>
1168 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1171 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1174 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1175 parameter specifying the scaling.
1176 (inline_call): Update.
1177 (want_inline_recursively): Guard division by zero.
1178 (recursive_inlining): Update.
1179 * ipa-inline.h (clone_inlined_nodes): Update.
1181 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1184 * config/i386/i386.c (classify_argument): Pass structures of size
1185 64 bytes or less in register.
1187 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1188 Kirill Yukhin <kirill.yukhin@intel.com>
1190 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1191 (_mm_rcp28_round_ss): Ditto.
1192 (_mm_rsqrt28_round_sd): Ditto.
1193 (_mm_rsqrt28_round_ss): Ditto.
1194 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1195 (_mm_rcp14_round_ss): Ditto.
1196 (_mm_rsqrt14_round_sd): Ditto.
1197 (_mm_rsqrt14_round_ss): Ditto.
1198 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1199 the first input operand, get rid of match_dup.
1200 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1202 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1204 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1205 operand as the first input operand, set type attribute.
1206 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1208 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1209 operand as the first input operand, set type attribute.
1211 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1213 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1216 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1219 * config/i386/i386.c (construct_container): Remove TFmode check
1220 for X86_64_INTEGER_CLASS.
1222 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1225 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1226 only when -Wpsabi is enabled.
1228 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1231 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1232 passing arrays in registers are the same as for structs, so remove the
1233 special case for them.
1235 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1237 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1238 destination type, extract only the valid bits if the source type is not
1239 integral and has a different mode.
1241 2014-02-19 Richard Biener <rguenther@suse.de>
1244 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1247 2014-02-19 Richard Biener <rguenther@suse.de>
1250 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1251 (ipa_modify_call_arguments): Emit an argument load explicitely and
1252 preserve virtual SSA form there and for the replacement call.
1253 Do not update SSA form nor free dominance info.
1255 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1257 * ipa.c (function_and_variable_visibility): Also clear WEAK
1258 flag when disolving COMDAT_GROUP.
1260 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1262 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1263 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1265 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1266 do more sanity checks.
1267 (detect_type_change): Return true when giving up early.
1268 (compute_complex_assign_jump_func): Fix type parameter of
1269 ipa_set_ancestor_jf.
1270 (compute_complex_ancestor_jump_func): Likewise.
1271 (update_jump_functions_after_inlining): Fix updating of
1273 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1275 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1277 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1278 inline clones when edge disappears.
1280 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1283 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1284 Split 64-bit moves into 2 patterns. Do not allow the use of
1285 direct move for TDmode in little endian, since the decimal value
1286 has little endian bytes within a word, but the 64-bit pieces are
1287 ordered in a big endian fashion, and normal subreg's of TDmode are
1289 (mov<mode>_64bit_dm): Likewise.
1290 (movtd_64bit_nodm): Likewise.
1292 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1294 PR tree-optimization/60174
1295 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1296 statement of an SSA_NAME that occurs in an abnormal PHI node.
1298 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1301 * final.c (SEEN_BB): Remove.
1302 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1303 (final_scan_insn): Don't force_source_line on second
1304 NOTE_INSN_BASIC_BLOCK.
1306 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1309 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1310 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1311 (type_natural_mode): Warn ABI change when %zmm register is not
1312 available for AVX512F vector value passing.
1314 2014-02-18 Kai Tietz <ktietz@redhat.com>
1317 * config/i386/i386.c (ix86_expand_prologue): Use value in
1318 rax register as displacement when restoring %r10 or %rax.
1319 Fix wrong offset when restoring both registers.
1321 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1323 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1324 assertion with conditional return.
1326 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1327 Uros Bizjak <ubizjak@gmail.com>
1330 * config/i386/driver-i386.c (host_detect_local_cpu): If
1331 YMM state is not saved by the OS, also clear has_f16c. Move
1332 CPUID 0x80000001 handling before YMM state saving checking.
1334 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1336 PR rtl-optimization/58960
1337 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1338 factored out from ...
1339 (sched_init): ... here.
1340 (free_global_sched_pressure_data): New, factored out from ...
1341 (sched_finish): ... here.
1342 * sched-int.h (free_global_sched_pressure_data): Declare.
1343 * sched-rgn.c (nr_regions_initial): New static global.
1344 (haifa_find_rgns): Initialize it.
1345 (schedule_region): Disable sched-pressure for the newly
1348 2014-02-17 Richard Biener <rguenther@suse.de>
1350 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1351 release SSA defs of pattern stmts.
1353 2014-02-17 Richard Biener <rguenther@suse.de>
1355 * tree-inline.c (expand_call_inline): Release the virtual
1356 operand defined by the call we are about to inline.
1358 2014-02-17 Richard Biener <rguenther@suse.de>
1360 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1362 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1363 Ilya Tocar <ilya.tocar@intel.com>
1365 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1366 arguments order in builtin.
1367 (_mm512_permutexvar_epi64): Ditto.
1368 (_mm512_mask_permutexvar_epi64): Ditto
1369 (_mm512_maskz_permutexvar_epi32): Ditto
1370 (_mm512_permutexvar_epi32): Ditto
1371 (_mm512_mask_permutexvar_epi32): Ditto
1373 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1375 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1376 (p8_vmrgow): Likewise.
1378 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1380 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1383 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1386 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1387 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1388 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1389 using direct move instructions on ISA 2.07. Also adjust
1390 instruction length for 64-bit.
1391 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1392 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1394 2014-02-15 Alan Modra <amodra@gmail.com>
1398 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1399 find_replacement on parts of insn rtl that might be reloaded.
1401 2014-02-15 Richard Biener <rguenther@suse.de>
1403 PR tree-optimization/60183
1404 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1405 (tree_ssa_phiprop): Calculate and free post-dominators.
1407 2014-02-14 Jeff Law <law@redhat.com>
1409 PR rtl-optimization/60131
1410 * ree.c (get_extended_src_reg): New function.
1411 (combine_reaching_defs): Use it rather than assuming location of REG.
1412 (find_and_remove_re): Verify first operand of extension is
1413 a REG before adding the insns to the copy list.
1415 2014-02-14 Roland McGrath <mcgrathr@google.com>
1417 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1418 * configure: Regenerated.
1419 * config.in: Regenerated.
1420 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1421 instead of ASM_SHORT.
1423 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1424 Richard Earnshaw <rearnsha@arm.com>
1426 PR rtl-optimization/59535
1427 * lra-constraints.c (process_alt_operands): Encourage alternative
1428 when unassigned pseudo class is superset of the alternative class.
1429 (inherit_reload_reg): Don't inherit when optimizing for code size.
1430 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1431 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1432 modes not less than 4 for Thumb1.
1434 2014-02-14 Kyle McMartin <kyle@redhat.com>
1437 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1439 2014-02-14 Richard Biener <rguenther@suse.de>
1441 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1442 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1443 Do not strip INDIRECT_REFs.
1445 2014-02-14 Richard Biener <rguenther@suse.de>
1448 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1449 DECL_FUNCTION_SPECIFIC_TARGET.
1450 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1451 * tree-streamer-out.c (pack_ts_target_option): Remove.
1452 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1453 (write_ts_function_decl_tree_pointers): Do not stream
1454 DECL_FUNCTION_SPECIFIC_TARGET.
1455 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1456 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1457 (lto_input_ts_function_decl_tree_pointers): Do not stream
1458 DECL_FUNCTION_SPECIFIC_TARGET.
1460 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1462 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1463 (get_initial_def_for_induction, vectorizable_induction): Ignore
1464 debug stmts when looking for exit_phi.
1465 (vectorizable_live_operation): Fix up condition.
1467 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1469 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1470 nreverse() because it changes the content of original tree list.
1472 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1474 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1475 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1477 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1479 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1480 GNU coding standards.
1482 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1485 * dwarf2out.c (gen_subprogram_die): Don't call
1486 add_calling_convention_attribute if subr_die is old_die.
1488 2014-02-13 Sharad Singhai <singhai@google.com>
1490 * doc/optinfo.texi: Fix order of nodes.
1492 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1494 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1495 operands[2], not operands[3].
1497 2014-02-13 Richard Biener <rguenther@suse.de>
1500 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1501 mention the possibility of an in-tree build.
1502 (CLooG): Update recommended version to 0.18.1, mention the
1503 possibility of an in-tree build and clarify that the ISL
1504 bundled with CLooG does not work.
1506 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1509 * expr.c (compress_float_constant): If x is a hard register,
1510 extend into a pseudo and then move to x.
1512 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1514 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1515 caused by bad second argument to warning_at() with -mhotpatch and
1516 nested functions (e.g. with gfortran).
1518 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1520 * opts.c (option_name): Remove "enabled by default" rider.
1522 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1524 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1526 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1527 Uros Bizjak <ubizjak@gmail.com>
1530 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1531 * configure: Regenerated.
1533 2014-02-12 Richard Biener <rguenther@suse.de>
1535 * vec.c (vec_prefix::calculate_allocation): Move as
1536 inline variant to vec.h.
1537 (vec_prefix::calculate_allocation_1): New out-of-line version.
1538 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1539 (vec_prefix::m_has_auto_buf): Rename to ...
1540 (vec_prefix::m_using_auto_storage): ... this.
1541 (vec_prefix::calculate_allocation): Inline the easy cases
1542 and dispatch to calculate_allocation_1 which doesn't need the
1544 (va_heap::reserve): Use gcc_checking_assert.
1545 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1546 m_using_auto_storage.
1547 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1549 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1550 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1551 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1553 2014-02-12 Richard Biener <rguenther@suse.de>
1555 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1556 when we found a dependence.
1558 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1560 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1562 (maybe_fold_stmt): ... into this new function.
1563 * omp-low.c (lower_omp): Update comment.
1565 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1568 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1571 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1573 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1574 identifiers in comments.
1575 (cortexa53_extra_costs): Likewise.
1576 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1577 (cortexa7_extra_costs): Likewise.
1578 (cortexa12_extra_costs): Likewise.
1579 (cortexa15_extra_costs): Likewise.
1580 (v7m_extra_costs): Likewise.
1582 2014-02-12 Richard Biener <rguenther@suse.de>
1585 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1586 of posix_memalign being successful.
1587 (lower_stmt): Restrict lowering of posix_memalign to when
1588 -ftree-bit-ccp is enabled.
1590 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1592 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1594 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1596 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1598 PR rtl-optimization/60116
1599 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1600 other_insn once the combination has been validated.
1602 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1605 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1607 * ipa-devirt.c: Include demangle.h
1608 (odr_violation_reported): New static variable.
1609 (add_type_duplicate): Update odr_violations.
1610 (maybe_record_node): Add completep parameter; update it.
1611 (record_target_from_binfo): Add COMPLETEP parameter;
1612 update it as needed.
1613 (possible_polymorphic_call_targets_1): Likewise.
1614 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1615 rename FINAL to COMPLETE.
1616 (record_targets_from_bases): Sanity check we found the binfo;
1617 fix COMPLETEP updating.
1618 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1619 parameter, fix computing of COMPLETEP.
1620 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1621 at LTO time do demangling.
1622 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1623 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1625 (gimple_get_virt_method_for_binfo): Likewise.
1626 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1627 gimple_get_virt_method_for_vtable): Update prototypes.
1629 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1632 * genautomata.c (add_presence_absence): Fix typo with
1633 {final_}presence_list.
1635 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1638 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1639 for VSX/Altivec vectors that land in GPR registers.
1641 2014-02-11 Richard Henderson <rth@redhat.com>
1642 Jakub Jelinek <jakub@redhat.com>
1645 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1646 around drhs if type conversion to lacc->type is not useless.
1648 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1650 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1652 (cortex-a57.cortex-a53): Likewise.
1653 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1655 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1657 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1660 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1662 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1663 add_options_for_arm_vfp3.
1665 2014-02-11 Jeff Law <law@redhat.com>
1668 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1669 object with an undesirable mode.
1671 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1674 * config/i386/sol2-9.h: New file.
1675 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1676 *-*-solaris2.9*): Use it.
1678 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1680 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1681 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1683 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1685 * config/microblaze/microblaze.c: Extend mcpu version format
1687 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1689 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1691 2014-02-10 Richard Henderson <rth@redhat.com>
1694 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1695 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1696 ms-abi vs -mno-accumulate-outgoing-args.
1697 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1698 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1701 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1704 * cfgexpand.c (expand_asm_operands): Attach source location to
1705 ASM_INPUT rtx objects.
1706 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1708 2014-02-10 Nick Clifton <nickc@redhat.com>
1710 * config/mn10300/mn10300.c (popcount): New function.
1711 (mn10300_expand_prologue): Include saved registers in stack usage
1714 2014-02-10 Jeff Law <law@redhat.com>
1717 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1718 when changing the SET_DEST of a prior insn to avoid an input reload.
1720 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1722 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1723 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1724 -mcall-openbsd, or -mcall-linux.
1725 (CC1_ENDIAN_BIG_SPEC): Remove.
1726 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1727 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1728 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1729 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1730 and %cc1_endian_default.
1731 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1733 2014-02-10 Richard Biener <rguenther@suse.de>
1735 PR tree-optimization/60115
1736 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1737 MEM_REF handling. Properly verify that the accesses are not
1738 out of the objects bound.
1740 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1742 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1745 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1747 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1748 proper constants and fix formatting.
1749 (possible_polymorphic_call_targets): Fix formatting.
1751 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1752 Ilya Tocar <ilya.tocar@intel.com>
1754 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1755 (_mm512_loadu_epi32): Renamed into...
1756 (_mm512_loadu_si512): This.
1757 (_mm512_storeu_epi32): Renamed into...
1758 (_mm512_storeu_si512): This.
1759 (_mm512_maskz_ceil_ps): Removed.
1760 (_mm512_maskz_ceil_pd): Ditto.
1761 (_mm512_maskz_floor_ps): Ditto.
1762 (_mm512_maskz_floor_pd): Ditto.
1763 (_mm512_floor_round_ps): Ditto.
1764 (_mm512_floor_round_pd): Ditto.
1765 (_mm512_ceil_round_ps): Ditto.
1766 (_mm512_ceil_round_pd): Ditto.
1767 (_mm512_mask_floor_round_ps): Ditto.
1768 (_mm512_mask_floor_round_pd): Ditto.
1769 (_mm512_mask_ceil_round_ps): Ditto.
1770 (_mm512_mask_ceil_round_pd): Ditto.
1771 (_mm512_maskz_floor_round_ps): Ditto.
1772 (_mm512_maskz_floor_round_pd): Ditto.
1773 (_mm512_maskz_ceil_round_ps): Ditto.
1774 (_mm512_maskz_ceil_round_pd): Ditto.
1775 (_mm512_expand_pd): Ditto.
1776 (_mm512_expand_ps): Ditto.
1777 * config/i386/i386.c (ix86_builtins): Remove
1778 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1779 (bdesc_args): Ditto.
1780 * config/i386/predicates.md (const1256_operand): New.
1781 (const_1_to_2_operand): Ditto.
1782 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1783 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1784 (*avx512pf_gatherpf<mode>sf): Ditto.
1785 (avx512pf_gatherpf<mode>df): Ditto.
1786 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1787 (*avx512pf_gatherpf<mode>df): Ditto.
1788 (avx512pf_scatterpf<mode>sf): Ditto.
1789 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1790 (*avx512pf_scatterpf<mode>sf): Ditto.
1791 (avx512pf_scatterpf<mode>df): Ditto.
1792 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1793 (*avx512pf_scatterpf<mode>df): Ditto.
1794 (avx512f_expand<mode>): Removed.
1795 (<shift_insn><mode>3<mask_name>): Change predicate type.
1797 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1799 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1800 not at the end of datarefs vector use ordered_remove to avoid
1801 reordering datarefs vector.
1804 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1805 mark local addressable non-static vars as GOVD_PRIVATE
1806 instead of GOVD_LOCAL.
1807 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1808 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1812 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1813 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1814 assume_aligned or alloc_align attributes.
1815 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1816 arguments. Handle also assume_aligned and alloc_align attributes.
1817 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1818 calls to functions with assume_aligned or alloc_align attributes.
1819 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1821 2014-02-08 Terry Guo <terry.guo@arm.com>
1823 * doc/invoke.texi: Document ARM -march=armv7e-m.
1825 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1827 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1828 flag on __cilkrts_rethrow builtin.
1831 * ipa-cp.c (determine_versionability): Fail at -O0
1832 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1833 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1836 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1839 * tree-inline.c (copy_forbidden): Fail for
1840 __attribute__((optimize (0))) functions.
1842 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1844 * varpool.c: Include pointer-set.h.
1845 (varpool_remove_unreferenced_decls): Variables in other partitions
1846 will not be output; be however careful to not lose information
1849 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1851 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1852 lookup in the vtable constructor.
1854 2014-02-07 Jeff Law <law@redhat.com>
1857 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1858 define_insn_and_split.
1860 * ipa-inline.c (inline_small_functions): Fix typos.
1862 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1864 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1865 (s390_can_use_return_insn): Declare.
1866 * config/s390/s390.h (EPILOGUE_USES): Define.
1867 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1869 (s390_chunkify_start): Handle return JUMP_LABELs.
1870 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1871 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1872 (s390_can_use_return_insn): New functions.
1873 (s390_fix_long_loop_prediction): Handle conditional returns.
1874 (TARGET_SET_UP_BY_PROLOGUE): Define.
1875 * config/s390/s390.md (ANY_RETURN): New code iterator.
1876 (*creturn, *csimple_return, return, simple_return): New patterns.
1878 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1880 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1881 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1882 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1883 REG_CFA_RESTORE list when deciding not to restore a register.
1885 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1887 * config/s390/s390.c: Include tree-pass.h and context.h.
1888 (s390_early_mach): New function, split out from...
1889 (s390_emit_prologue): ...here.
1890 (pass_data_s390_early_mach): New pass structure.
1891 (pass_s390_early_mach): New class.
1892 (s390_option_override): Create and register early_mach pass.
1893 Move to end of file.
1895 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1897 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1898 to match for the exit block.
1900 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1902 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1903 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1904 Reject misaligned operands.
1906 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1908 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1910 2014-02-07 Richard Biener <rguenther@suse.de>
1913 * gimple-low.c (lower_builtin_posix_memalign): New function.
1914 (lower_stmt): Call it to lower posix_memalign in a way
1915 to make alignment info accessible.
1917 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1920 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1921 __builtin_setjmp_receiver.
1923 2014-02-07 Richard Biener <rguenther@suse.de>
1926 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1927 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1928 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1929 Handle BUILT_IN_POSIX_MEMALIGN.
1930 (find_func_clobbers): Likewise.
1931 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1932 (call_may_clobber_ref_p_1): Likewise.
1934 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1937 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1940 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1943 * lto-cgraph.c (lto_output_node): Use
1944 symtab_get_symbol_partitioning_class.
1945 (lto_output_varpool_node): likewise.
1946 (symtab_get_symbol_partitioning_class): Move here from
1948 * cgraph.h (symbol_partitioning_class): Likewise.
1949 (symtab_get_symbol_partitioning_class): Declare.
1951 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1953 * ggc.h (ggc_internal_cleared_alloc): New macro.
1954 * vec.h (vec_safe_copy): Handle memory stats.
1955 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1956 * target-globals.c (save_target_globals): Likewise.
1958 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1961 * expr.c (emit_move_resolve_push): Export; be bit more selective
1962 on when to clear alias set.
1963 * expr.h (emit_move_resolve_push): Declare.
1964 * function.h (struct function): Add tail_call_marked.
1965 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1966 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1967 * config/i386/i386.md (TImode move expander): De not call
1969 (FP push expanders): Preserve memory attributes.
1970 * config/i386/sse.md (push<mode>1): Remove.
1971 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1972 (ix86_expand_push): Remove.
1973 * config/i386/mmx.md (push<mode>1): Remove.
1975 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1977 PR rtl-optimization/60030
1978 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1979 lopart with paradoxical subreg before shifting it up by hprec.
1981 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1983 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1984 Remove extra newline at end of file.
1985 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1986 (arm_issue_rate): Handle cortexa57.
1987 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1988 (cortex-a57.cortex-a53): Likewise.
1990 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1993 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1994 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1996 (arm_expand_prologue): Adjust all callers.
1997 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1998 info, registers also at the lowest numbered registers side. Use
1999 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2003 * var-tracking.c (adjust_mems): Before adding a SET to
2004 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2006 2014-02-06 Alan Modra <amodra@gmail.com>
2009 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2010 change SDmode to DDmode when lra_in_progress.
2012 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2015 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2016 free_data_ref on the dr first, and before goto again also set dr
2017 to the next dr. For simd_lane_access, free old datarefs[i] before
2018 overwriting it. For get_vectype_for_scalar_type failure, don't
2019 free_data_ref if simd_lane_access.
2021 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2024 * tree.h (opts_for_fn): New inline function.
2025 (opt_for_fn): Define.
2026 * config/i386/i386.c (ix86_function_regparm): Use
2027 opt_for_fn (decl, optimize) instead of optimize.
2029 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2031 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2032 for SYMBOL_REF in large memory model.
2034 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2036 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2038 (cortex-a57): Likewise.
2039 (cortex-a57.cortex-a53): Likewise.
2041 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2042 Kugan Vivekanandarajah <kuganv@linaro.org>
2044 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2046 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2048 2014-02-06 Richard Biener <rguenther@suse.de>
2050 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2051 set_loop_copy and initialize_original_copy_tables.
2053 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2055 * config/aarch64/aarch64-simd.md
2056 (aarch64_ashr_simddi): Change QI to SI.
2058 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2059 Jakub Jelinek <jakub@redhat.com>
2062 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2065 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2067 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2068 CODE_FOR_altivec_vpku[hw]um to
2069 CODE_FOR_altivec_vpku[hw]um_direct.
2070 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2071 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2072 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2073 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2075 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2077 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2078 generation for -maltivec=be.
2079 (altivec_vsumsws): Simplify redundant test.
2081 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2083 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2084 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2085 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2086 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2087 gen_altivec_vpkuwum.
2088 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2090 (altivec_vpks<VI_char>ss): Likewise.
2091 (altivec_vpks<VI_char>us): Likewise.
2092 (altivec_vpku<VI_char>us): Likewise.
2093 (altivec_vpku<VI_char>um): Likewise.
2094 (altivec_vpku<VI_char>um_direct): New (copy of
2095 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2097 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2098 target is little endian and -maltivec=be is not specified.
2099 (*altivec_vupkhs<VU_char>_direct): New (copy of
2100 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2101 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2102 target is little endian and -maltivec=be is not specified.
2103 (*altivec_vupkls<VU_char>_direct): New (copy of
2104 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2105 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2106 little endian and -maltivec=be is not specified.
2107 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2108 little endian and -maltivec=be is not specified.
2110 2014-02-05 Richard Henderson <rth@redhat.com>
2113 * combine-stack-adj.c: Revert r206943.
2114 * sched-int.h (struct deps_desc): Add last_args_size.
2115 * sched-deps.c (init_deps): Initialize it.
2116 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2117 contain REG_ARGS_SIZE notes.
2119 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2121 * lto-cgraph.c (asm_nodes_output): Make global.
2122 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2123 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2124 (driver_handle_option): Handle OPT_fwpa.
2126 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2129 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2130 a comment typo and formatting issue. If odr_hash hasn't been
2131 created, return vNULL and set *completep to false.
2134 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2135 bb with no successors.
2137 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2140 * doc/invoke.texi (-march): Clarify documentation for ARM.
2144 2014-02-05 Richard Biener <rguenther@suse.de>
2146 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2147 when not vectorizing because of too many alias checks.
2148 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2149 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2151 2014-02-05 Nick Clifton <nickc@redhat.com>
2153 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2154 accept extended registers in any mode when compiling for the MN10300.
2156 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2158 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2159 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2160 sanitization attributes.
2161 (can_inline_edge_p): Likewise.
2162 (sanitize_attrs_match_for_inline_p): New function.
2164 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2166 * ipa-prop.c (detect_type_change): Shor circuit testing of
2167 type changes on THIS pointer.
2169 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2172 * config/pa/pa.c (legitimize_tls_address): Return original address
2173 if not passed a SYMBOL_REF rtx.
2174 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2176 (pa_emit_move_sequence): Simplify TLS source operands.
2177 (pa_legitimate_constant_p): Reject all TLS constants.
2178 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2179 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2181 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2183 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2184 groups when we know they are controlled by LTO.
2185 * varasm.c (default_binds_local_p_1): If object is in other partition,
2186 it will be resolved locally.
2188 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2190 * config/host-linux.c (linux_gt_pch_use_address): Don't
2191 use SSIZE_MAX because it is not always defined.
2193 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2196 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2197 threshold for pseudo splitting.
2198 (update_ebb_live_info): Process call argument hard registers and
2199 hard registers from insn definition too.
2200 (max_small_class_regs_num): New constant.
2201 (inherit_in_ebb): Update live hard regs through EBBs. Update
2202 reloads_num only for small register classes. Don't split for
2205 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2208 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2211 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2213 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2216 2014-02-04 Marek Polacek <polacek@redhat.com>
2218 * gdbinit.in (pel): Define.
2220 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2222 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2225 2014-02-04 Richard Biener <rguenther@suse.de>
2228 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2229 in function context local.
2230 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2231 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2232 similar to LTO_imported_decl_ref.
2234 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2236 PR tree-optimization/60002
2237 * cgraphclones.c (build_function_decl_skip_args): Clear
2240 PR tree-optimization/60023
2241 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2242 false to gsi_replace.
2243 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2244 has been in some EH region and vec_stmt could throw, add
2245 vec_stmt into the same EH region.
2246 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2247 has no lhs, ignore it.
2248 * internal-fn.c (expand_MASK_LOAD): Likewise.
2251 * tree-inline.c (copy_forbidden): Fail for
2252 __attribute__((optimize (0))) functions.
2255 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2256 is set, copy one less argument.
2257 (expand_simd_clones): Don't subtract clone_info->inbranch
2258 from simd_clone_struct_alloc argument.
2260 PR rtl-optimization/57915
2261 * recog.c (simplify_while_replacing): If all unary/binary/relational
2262 operation arguments are constant, attempt to simplify those.
2265 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2266 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2268 2014-02-04 Richard Biener <rguenther@suse.de>
2270 PR tree-optimization/60012
2271 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2272 TBAA disambiguation to all DDRs.
2274 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2277 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2278 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2280 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2283 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2285 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2287 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2288 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2290 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2293 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2294 to figure out targets of polymorphic calls with known decl.
2295 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2296 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2297 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2298 (get_polymorphic_call_info): ... here.
2299 (get_polymorphic_call_info_from_invariant): New function.
2301 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2303 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2304 lookup via vtable pointer; check for type consistency
2305 and turn inconsitent facts into UNREACHABLE.
2306 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2307 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2308 type inconsistent querries; return UNREACHABLE instead.
2310 2014-02-03 Richard Henderson <rth@twiddle.net>
2313 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2314 already processed this node.
2315 (normalize_one_pred_1): Pass along mark_set.
2316 (normalize_one_pred): Create and destroy a pointer_set_t.
2317 (normalize_one_pred_chain): Likewise.
2319 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2321 PR gcov-profile/58602
2322 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2324 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2327 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2328 -fno-devirtualize; try to devirtualize by the knowledge of
2329 virtual table pointer given by aggregate propagation.
2330 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2331 (ipa_print_node_jump_functions): Dump also offset that
2332 is relevant for polymorphic calls.
2333 (determine_known_aggregate_parts): Add arg_type parameter; use it
2334 instead of determining the type from pointer type.
2335 (ipa_compute_jump_functions_for_edge): Update call of
2336 determine_known_aggregate_parts.
2337 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2338 (gimple_get_virt_method_for_binfo): ... here; simplify using
2339 vtable_pointer_value_to_vtable.
2340 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2341 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2342 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2343 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2345 (vtable_pointer_value_to_binfo): ... here.
2346 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2348 2014-02-03 Teresa Johnson <tejohnson@google.com>
2350 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2351 redef of outer loop index variable.
2353 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2357 * doc/extend.texi (Function Attributes): Typo.
2359 2014-02-03 Cong Hou <congh@google.com>
2361 PR tree-optimization/60000
2362 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2363 if the vectorized statement is a store. A store statement can only
2364 appear at the end of pattern statements.
2366 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2368 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2369 (ix86_option_override_internal): Default long double to 64-bit for
2370 32-bit Bionic and to 128-bit for 64-bit Bionic.
2372 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2373 TARGET_LONG_DOUBLE_128 is true.
2374 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2376 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2377 (mlong-double-64): Negate -mlong-double-128.
2378 (mlong-double-128): New option.
2380 * config/i386/i386-c.c (ix86_target_macros): Define
2381 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2383 * doc/invoke.texi: Document -mlong-double-128.
2385 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2387 PR rtl-optimization/60024
2388 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2390 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2392 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2394 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2396 PR rtl-optimization/57662
2397 * sel-sched.c (code_motion_path_driver): Do not mark already not
2398 existing blocks in the visiting bitmap.
2400 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2402 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2403 on the insn being emitted.
2405 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2406 Will Deacon <will.deacon@arm.com>
2408 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2410 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2412 * config/arm/arm-tables.opt: Regenerate.
2414 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2416 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2417 for vector types other than V16QImode.
2418 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2419 define_expand, and call altivec_expand_vec_perm_le when producing
2420 code with little endian element order.
2421 (*altivec_vperm_<mode>_internal): New insn having previous
2422 behavior of altivec_vperm_<mode>.
2423 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2424 altivec_expand_vec_perm_le when producing code with little endian
2426 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2427 behavior of altivec_vperm_<mode>_uns.
2429 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2431 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2432 (altivec_vsumsws): Add handling for -maltivec=be with a little
2434 (altivec_vsumsws_direct): New.
2435 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2436 gen_altivec_vsumsws.
2438 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2440 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2441 vtable_pointer_value_to_binfo): New functions.
2442 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2443 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2445 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2447 * config/nios2/nios2.md (load_got_register): Initialize GOT
2448 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2449 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2451 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2453 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2454 preserverd by passthrough, do not propagate the type.
2456 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2458 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2459 (mips_atomic_assign_expand_fenv): New function.
2460 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2462 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2464 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2465 (__builtin_mips_set_fcsr): Likewise.
2466 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2467 MIPS_USI_FTYPE_VOID.
2468 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2469 (mips16_expand_set_fcsr): Likewise.
2470 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2471 (mips16_set_fcsr_stub): Likewise.
2472 (mips16_get_fcsr_one_only_stub): New class.
2473 (mips16_set_fcsr_one_only_stub): Likewise.
2474 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2475 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2476 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2477 (hard_float): New availability predicate.
2478 (mips_builtins): Add get_fcsr and set_fcsr.
2479 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2480 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2481 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2482 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2483 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2486 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2488 * config/mips/mips.c (mips_one_only_stub): New class.
2489 (mips_need_mips16_rdhwr_p): Replace with...
2490 (mips16_rdhwr_stub): ...this new variable.
2491 (mips16_stub_call_address): New function.
2492 (mips16_rdhwr_one_only_stub): New class.
2493 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2494 (mips_output_mips16_rdhwr): Delete.
2495 (mips_finish_stub): New function.
2496 (mips_code_end): Use it to handle rdhwr stubs.
2498 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2501 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2502 when calculating size of integer atomic types.
2504 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2506 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2508 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2510 PR tree-optimization/60003
2511 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2512 * profile.c (branch_prob): Use gimple_call_builtin_p
2513 to check for BUILT_IN_SETJMP_RECEIVER.
2514 * tree-inline.c (copy_bb): Call notice_special_calls.
2516 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2519 * lra-constraints.c (process_alt_operands): Update reload_sum only
2522 2014-01-31 Richard Henderson <rth@redhat.com>
2525 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2526 until after else_eh is processed.
2528 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2530 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2531 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2532 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2533 in smmintrin.h, remove them.
2534 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2535 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2536 * config/i386/i386.md (ROUND_SAE): Fix value.
2537 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2538 (const48_operand): New.
2539 * config/i386/subst.md (round), (round_expand): Use
2540 const_4_or_8_to_11_operand.
2541 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2543 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2545 * config/i386/constraints.md (Yk): Swap meaning with k.
2546 * config/i386/i386.md (movhi_internal): Change Yk to k.
2547 (movqi_internal): Ditto.
2548 (*k<logic><mode>): Ditto.
2551 (kandn<mode>): Ditto.
2552 (*<code>hi_1): Ditto.
2553 (*<code>qi_1): Ditto.
2554 (kxnor<mode>): Ditto.
2555 (kortestzhi): Ditto.
2556 (kortestchi): Ditto.
2558 (*one_cmplhi2_1): Ditto.
2559 (*one_cmplqi2_1): Ditto.
2560 * config/i386/sse.md (): Change k to Yk.
2561 (avx512f_load<mode>_mask): Ditto.
2562 (avx512f_blendm<mode>): Ditto.
2563 (avx512f_store<mode>_mask): Ditto.
2564 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2565 (avx512f_storedqu<mode>_mask): Ditto.
2566 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2568 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2569 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2570 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2571 (avx512f_maskcmp<mode>3): Ditto.
2572 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2573 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2574 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2575 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2576 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2577 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2578 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2579 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2580 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2581 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2582 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2583 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2584 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2585 (vec_extract_lo_<mode>_maskm): Ditto.
2586 (vec_extract_hi_<mode>_maskm): Ditto.
2587 (avx512f_vternlog<mode>_mask): Ditto.
2588 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2589 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2590 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2591 (avx512f_<code>v8div16qi2_mask): Ditto.
2592 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2593 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2594 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2595 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2596 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2597 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2598 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2599 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2600 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2601 (avx512cd_maskb_vec_dupv8di): Ditto.
2602 (avx512cd_maskw_vec_dupv16si): Ditto.
2603 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2604 (avx512f_vpermi2var<mode>3_mask): Ditto.
2605 (avx512f_vpermi2var<mode>3_mask): Ditto.
2606 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2607 (*avx512f_gathersi<mode>): Ditto.
2608 (*avx512f_gathersi<mode>_2): Ditto.
2609 (*avx512f_gatherdi<mode>): Ditto.
2610 (*avx512f_gatherdi<mode>_2): Ditto.
2611 (*avx512f_scattersi<mode>): Ditto.
2612 (*avx512f_scatterdi<mode>): Ditto.
2613 (avx512f_compress<mode>_mask): Ditto.
2614 (avx512f_compressstore<mode>_mask): Ditto.
2615 (avx512f_expand<mode>_mask): Ditto.
2616 * config/i386/subst.md (mask): Change k to Yk.
2617 (mask_scalar_merge): Ditto.
2620 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2622 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2624 2014-01-31 Richard Biener <rguenther@suse.de>
2627 * builtins.c (fold_builtin_memory_op): Make sure to not
2628 use a floating-point mode or a boolean or enumeral type for
2631 2014-01-30 DJ Delorie <dj@redhat.com>
2633 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2634 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2635 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2636 whenever main() has an epilogue.
2638 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2640 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2641 unused variable "field".
2642 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2643 (vsx_mergeh_<mode>): Likewise.
2644 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2645 (altivec_vmrghh): Likewise.
2646 (altivec_vmrghw): Likewise.
2647 (altivec_vmrglb): Likewise.
2648 (altivec_vmrglh): Likewise.
2649 (altivec_vmrglw): Likewise.
2650 (altivec_vspltb): Add missing uses.
2651 (altivec_vsplth): Likewise.
2652 (altivec_vspltw): Likewise.
2653 (altivec_vspltsf): Likewise.
2655 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2658 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2659 frame related instructions.
2661 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2663 PR rtl-optimization/59959
2664 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2665 any reload of register whose subreg is invalid.
2667 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2669 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2670 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2671 Add missing return type - void.
2673 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2675 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2676 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2677 remove element index adjustment for endian (now handled in vsx.md
2679 (altivec_expand_vec_perm_const): Use
2680 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2681 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2682 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2683 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2684 define_expand and a new define_insn *altivec_vspltb_internal;
2685 adjust for -maltivec=be on a little endian target.
2686 (altivec_vspltb_direct): New.
2687 (altivec_vsplth): Divide into a define_expand and a new
2688 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2689 little endian target.
2690 (altivec_vsplth_direct): New.
2691 (altivec_vspltw): Divide into a define_expand and a new
2692 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2693 little endian target.
2694 (altivec_vspltw_direct): New.
2695 (altivec_vspltsf): Divide into a define_expand and a new
2696 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2697 a little endian target.
2699 2014-01-30 Richard Biener <rguenther@suse.de>
2701 PR tree-optimization/59993
2702 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2703 can propagate form the earlier stmt and avoid the transform
2704 when the intermediate result is needed.
2706 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2708 * README.Portability: Fix typo.
2710 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2712 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2713 comparison_operator with ordered_comparison_operator.
2715 2014-01-30 Nick Clifton <nickc@redhat.com>
2717 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2718 Rename to mn10300_store_multiple_regs.
2719 * config/mn10300/mn10300.c: Likewise.
2720 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2721 store_multiple_regs.
2722 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2723 Call mn10300_store_multiple_regs.
2725 2014-01-30 Nick Clifton <nickc@redhat.com>
2726 DJ Delorie <dj@redhat.com>
2728 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2729 %fp 2 to keep registers after it properly word-aligned.
2730 (rl78_alloc_physical_registers_umul): Handle the case where both
2731 input operands are the same.
2733 2014-01-30 Richard Biener <rguenther@suse.de>
2735 PR tree-optimization/59903
2736 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2739 2014-01-30 Jason Merrill <jason@redhat.com>
2742 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2745 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2747 2014-01-30 Richard Biener <rguenther@suse.de>
2749 PR tree-optimization/59951
2750 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2752 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2755 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2756 SFmode to DFmode case.
2758 2014-01-29 DJ Delorie <dj@redhat.com>
2760 * config/msp430/msp430.opt (-minrt): New.
2761 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2763 (ENDFILE_SPEC): Likewise.
2765 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2767 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2768 (estimate_function_body_sizes): Use it.
2770 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2773 * dwarf2out.c (is_cxx_auto): New.
2774 (is_base_type): Use it.
2775 (gen_type_die_with_usage): Likewise.
2777 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2779 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2780 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2781 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2782 -maltivec=be with LE targets.
2783 (vsx_mergeh_<mode>): Likewise.
2784 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2785 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2786 (altivec_vmrghb): Replace with define_expand and new
2787 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2788 (altivec_vmrghb_direct): New define_insn.
2789 (altivec_vmrghh): Replace with define_expand and new
2790 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2791 (altivec_vmrghh_direct): New define_insn.
2792 (altivec_vmrghw): Replace with define_expand and new
2793 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2794 (altivec_vmrghw_direct): New define_insn.
2795 (*altivec_vmrghsf): Adjust for endianness.
2796 (altivec_vmrglb): Replace with define_expand and new
2797 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2798 (altivec_vmrglb_direct): New define_insn.
2799 (altivec_vmrglh): Replace with define_expand and new
2800 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2801 (altivec_vmrglh_direct): New define_insn.
2802 (altivec_vmrglw): Replace with define_expand and new
2803 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2804 (altivec_vmrglw_direct): New define_insn.
2805 (*altivec_vmrglsf): Adjust for endianness.
2806 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2807 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2808 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2809 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2810 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2811 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2812 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2813 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2815 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2817 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2818 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2821 2014-01-29 Richard Biener <rguenther@suse.de>
2823 PR tree-optimization/58742
2824 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2825 associate_pointerplus_align.
2826 (associate_pointerplus_diff): New function.
2827 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2828 and associate_pointerplus_diff.
2830 2014-01-29 Richard Biener <rguenther@suse.de>
2832 * lto-streamer.h (LTO_major_version): Bump to 3.
2833 (LTO_minor_version): Reset to 0.
2835 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2837 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2838 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2839 (arm_file_start): Generate correct asm header for armv7ve.
2840 * config/arm/bpabi.h: Add multilib support for armv7ve.
2841 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2842 and cortex-a15 to armv7ve.
2843 * config/arm/t-aprofile: Add multilib support for armv7ve.
2844 * doc/invoke.texi: Document -march=armv7ve.
2846 2014-01-29 Richard Biener <rguenther@suse.de>
2848 PR tree-optimization/58742
2849 * tree-ssa-forwprop.c (associate_plusminus): Return true
2850 if we changed sth, defer EH cleanup to ...
2851 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2852 (simplify_mult): New function.
2854 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2857 PR tree-optimization/59920
2858 * tree.c (build_common_builtin_nodes): Remove
2859 __builtin_setjmp_dispatcher initialization.
2860 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2861 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2862 instead of gsi_after_labels + manually skipping debug stmts.
2863 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2864 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2865 * tree-inline.c (copy_edges_for_bb): Remove
2866 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2867 argument. Ignore computed_goto_p stmts. Don't call
2868 make_abnormal_goto_edges. If a call might need abnormal edges
2869 for non-local gotos, see if it already has an edge to
2870 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2871 with true argument, don't do anything then, otherwise add
2872 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2873 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2875 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2876 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2877 (lower_stmt): Don't set data->calls_builtin_setjmp.
2878 (lower_builtin_setjmp): Adjust comment.
2879 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2880 * tree-cfg.c (found_computed_goto): Remove.
2881 (factor_computed_gotos): Remove.
2882 (make_goto_expr_edges): Return bool, true for computed gotos.
2883 Don't call make_abnormal_goto_edges.
2884 (build_gimple_cfg): Don't set found_computed_goto, don't call
2885 factor_computed_gotos.
2886 (computed_goto_p): No longer static.
2887 (make_blocks): Don't set found_computed_goto.
2888 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2889 (make_edges): If make_goto_expr_edges returns true, push bb
2890 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2891 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2892 vector. Record mapping between bbs and OpenMP regions if there
2893 are any, adjust make_gimple_omp_edges caller. Call
2894 handle_abnormal_edges.
2895 (make_abnormal_goto_edges): Remove.
2896 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2897 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2898 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2899 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2900 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2901 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2902 filling *region also set *region_idx to (*region)->entry->index.
2905 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2906 For REGs set ORIGINAL_REGNO.
2908 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2910 * doc/md.texi: Mention that a target shouldn't implement
2911 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2914 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2916 PR tree-optimization/59594
2917 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2918 a copy of the datarefs vector rather than the vector itself.
2920 2014-01-28 Jason Merrill <jason@redhat.com>
2923 * dwarf2out.c (auto_die): New static.
2924 (gen_type_die_with_usage): Handle C++1y 'auto'.
2925 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2928 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2931 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2932 (SPEC_X32): Likewise.
2933 (SPEC_64): Likewise.
2934 * config/i386/i386.c (ix86_option_override_internal): Turn off
2935 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2937 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2938 * config/i386/i386.h (TARGET_16BIT): New macro.
2939 (TARGET_16BIT_P): Likewise.
2940 * config/i386/i386.opt: Add m16.
2941 * doc/invoke.texi: Document -m16.
2943 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2945 PR preprocessor/59935
2946 * input.c (location_get_source_line): Bail out on when line number
2947 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2949 2014-01-28 Richard Biener <rguenther@suse.de>
2951 PR tree-optimization/58742
2952 * tree-ssa-forwprop.c (associate_plusminus): Handle
2953 pointer subtraction of the form (T)(P + A) - (T)P.
2955 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2957 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2960 2014-01-28 Richard Biener <rguenther@suse.de>
2963 2014-01-28 Richard Biener <rguenther@suse.de>
2965 PR rtl-optimization/45364
2966 PR rtl-optimization/59890
2967 * var-tracking.c (local_get_addr_clear_given_value): Handle
2968 already cleared slot.
2969 (val_reset): Handle not allocated local_get_addr_cache.
2970 (vt_find_locations): Use post-order on the inverted CFG.
2972 2014-01-28 Richard Biener <rguenther@suse.de>
2974 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2976 2014-01-28 Richard Biener <rguenther@suse.de>
2978 PR rtl-optimization/45364
2979 PR rtl-optimization/59890
2980 * var-tracking.c (local_get_addr_clear_given_value): Handle
2981 already cleared slot.
2982 (val_reset): Handle not allocated local_get_addr_cache.
2983 (vt_find_locations): Use post-order on the inverted CFG.
2985 2014-01-28 Alan Modra <amodra@gmail.com>
2987 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2988 * configure.ac <recursive call for build != host>: Define
2989 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2990 and LD_FOR_BUILD too.
2991 * configure: Regenerate.
2993 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2995 * config/i386/i386.c (get_builtin_code_for_version): Separate
2996 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2997 Broadwell from Haswell.
2999 2014-01-27 Steve Ellcey <sellcey@mips.com>
3001 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3002 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3003 * config/mips/mips.c (mips_option_override): Change setting
3005 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3006 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3007 Change from Mask to Var.
3009 2014-01-27 Jeff Law <law@redhat.com>
3011 * ipa-inline.c (inline_small_functions): Fix typo.
3013 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3015 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3016 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3017 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3018 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3019 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3020 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3021 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3022 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3023 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3024 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3025 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3026 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3027 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3028 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3029 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3030 (_mm512_storeu_epi64): Ditto.
3031 (_mm512_cmpge_epi32_mask): Ditto.
3032 (_mm512_cmpge_epu32_mask): Ditto.
3033 (_mm512_cmpge_epi64_mask): Ditto.
3034 (_mm512_cmpge_epu64_mask): Ditto.
3035 (_mm512_cmple_epi32_mask): Ditto.
3036 (_mm512_cmple_epu32_mask): Ditto.
3037 (_mm512_cmple_epi64_mask): Ditto.
3038 (_mm512_cmple_epu64_mask): Ditto.
3039 (_mm512_cmplt_epi32_mask): Ditto.
3040 (_mm512_cmplt_epu32_mask): Ditto.
3041 (_mm512_cmplt_epi64_mask): Ditto.
3042 (_mm512_cmplt_epu64_mask): Ditto.
3043 (_mm512_cmpneq_epi32_mask): Ditto.
3044 (_mm512_cmpneq_epu32_mask): Ditto.
3045 (_mm512_cmpneq_epi64_mask): Ditto.
3046 (_mm512_cmpneq_epu64_mask): Ditto.
3047 (_mm512_expand_pd): Ditto.
3048 (_mm512_expand_ps): Ditto.
3049 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3050 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3051 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3052 * config/i386/i386.c (ix86_builtins): Add
3053 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3054 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3055 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3056 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3057 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3058 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3059 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3060 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3061 IX86_BUILTIN_PMOVUSQW512_MEM.
3062 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3063 __builtin_ia32_pmovsqd512mem_mask,
3064 __builtin_ia32_pmovqd512mem_mask,
3065 __builtin_ia32_pmovusqw512mem_mask,
3066 __builtin_ia32_pmovsqw512mem_mask,
3067 __builtin_ia32_pmovqw512mem_mask,
3068 __builtin_ia32_pmovusdw512mem_mask,
3069 __builtin_ia32_pmovsdw512mem_mask,
3070 __builtin_ia32_pmovdw512mem_mask,
3071 __builtin_ia32_pmovqb512mem_mask,
3072 __builtin_ia32_pmovusqb512mem_mask,
3073 __builtin_ia32_pmovsqb512mem_mask,
3074 __builtin_ia32_pmovusdb512mem_mask,
3075 __builtin_ia32_pmovsdb512mem_mask,
3076 __builtin_ia32_pmovdb512mem_mask.
3077 (bdesc_args): Add __builtin_ia32_expanddf512,
3078 __builtin_ia32_expandsf512.
3079 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3080 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3081 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3082 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3083 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3084 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3085 (avx512f_<code>v8div16qi2_mask_store): This.
3086 (avx512f_expand<mode>): New.
3088 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3090 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3092 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3093 (_mm512_prefetch_i32scatter_pd): Ditto.
3094 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3095 (_mm512_prefetch_i64scatter_pd): Ditto.
3096 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3097 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3098 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3099 (_mm512_prefetch_i32scatter_ps): Ditto.
3100 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3101 (_mm512_prefetch_i64scatter_ps): Ditto.
3102 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3103 * config/i386/i386-builtin-types.def: Define
3104 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3105 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3106 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3107 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3108 IX86_BUILTIN_SCATTERPFQPD.
3109 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3110 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3111 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3112 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3113 __builtin_ia32_scatterpfqps.
3114 (ix86_expand_builtin): Expand new built-ins.
3115 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3116 fix memory access data type.
3117 (*avx512pf_gatherpf<mode>_mask): Ditto.
3118 (*avx512pf_gatherpf<mode>): Ditto.
3119 (avx512pf_scatterpf<mode>): Ditto.
3120 (*avx512pf_scatterpf<mode>_mask): Ditto.
3121 (*avx512pf_scatterpf<mode>): Ditto.
3122 (GATHER_SCATTER_SF_MEM_MODE): New.
3123 (avx512pf_gatherpf<mode>df): Ditto.
3124 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3125 (*avx512pf_scatterpf<mode>df): Ditto.
3127 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3130 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3131 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3134 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3136 * common/config/arm/arm-common.c
3137 (arm_rewrite_mcpu): Handle multiple names.
3139 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3141 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3143 * gimple-builder.h (create_gimple_tmp): Delete.
3145 2014-01-27 Christian Bruel <christian.bruel@st.com>
3147 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3150 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3152 * config/pa/pa.md (call): Generate indirect long calls to non-local
3153 functions when outputing 32-bit code.
3154 (call_value): Likewise except for special call to buggy powf function.
3156 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3157 portable runtime and PIC indirect calls.
3158 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3159 and PIC call sequences. Use ldo instead of blr to set return register
3160 in PIC call sequence.
3162 2014-01-25 Walter Lee <walt@tilera.com>
3164 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3165 avoid clobbering a live register.
3167 2014-01-25 Walter Lee <walt@tilera.com>
3169 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3170 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3171 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3172 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3174 2014-01-25 Walter Lee <walt@tilera.com>
3176 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3177 arguments on even registers.
3178 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3180 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3181 (BIGGEST_ALIGNMENT): Ditto.
3182 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3184 2014-01-25 Walter Lee <walt@tilera.com>
3186 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3187 insns before bundling.
3188 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3190 2014-01-25 Walter Lee <walt@tilera.com>
3192 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3193 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3194 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3196 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3198 * config/mips/constraints.md (kl): Delete.
3199 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3200 define expands, using...
3201 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3202 instructions for MIPS16.
3203 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3204 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3206 2014-01-25 Walter Lee <walt@tilera.com>
3208 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3212 2014-01-25 Walter Lee <walt@tilera.com>
3214 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3215 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3217 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3219 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3222 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3224 * print-rtl.c (in_call_function_usage): New var.
3225 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3226 EXPR_LIST mode as mode and not as reg note name.
3229 * cfgloopmanip.c (copy_loop_info): If
3230 loop->warned_aggressive_loop_optimizations, make sure
3231 the flag is set in target loop too.
3233 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3235 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3237 * builtins.def: Likewise.
3238 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3239 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3240 * ira.c (ira_setup_eliminable_regset): Likewise.
3241 * omp-low.c (gate_expand_omp): Likewise.
3242 (execute_lower_omp): Likewise.
3243 (diagnose_sb_0): Likewise.
3244 (gate_diagnose_omp_blocks): Likewise.
3245 (simd_clone_clauses_extract): Likewise.
3248 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3250 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3251 correction for little endian...
3252 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3255 2014-01-24 Jeff Law <law@redhat.com>
3257 PR tree-optimization/59919
3258 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3259 for non-returning calls.
3261 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3263 * common/config/aarch64/aarch64-common.c
3264 (aarch64_rewrite_mcpu): Handle multiple names.
3265 * config/aarch64/aarch64.h
3266 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3268 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3270 * input.c (add_file_to_cache_tab): Handle the case where fopen
3273 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3276 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3277 from push operand if code of push isn't PRE_DEC.
3279 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3282 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3283 -mquad-memory-atomic. Update -mquad-memory documentation to say
3284 it is only used for non-atomic loads/stores.
3286 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3287 -mquad-memory or -mquad-memory-atomic switches.
3289 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3290 -mquad-memory-atomic to ISA 2.07 support.
3292 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3293 to separate support of normal quad word memory operations (ldq, stq)
3294 from the atomic quad word memory operations.
3296 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3297 support to separate non-atomic quad word operations from atomic
3298 quad word operations. Disable non-atomic quad word operations in
3299 little endian mode so that we don't have to swap words after the
3300 load and before the store.
3301 (quad_load_store_p): Add comment about atomic quad word support.
3302 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3303 options printed with -mdebug=reg.
3305 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3306 -mquad-memory-atomic as the test for whether we have quad word
3307 atomic instructions.
3308 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3309 or -mp8-vector are used, allow byte/half-word atomic operations.
3311 * config/rs6000/sync.md (load_lockedti): Insure that the address
3312 is a proper indexed or indirect address for the lqarx instruction.
3313 On little endian systems, swap the hi/lo registers after the lqarx
3315 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3316 insure the address is valid for the lqarx instruction.
3317 (store_conditionalti): Insure that the address is a proper indexed
3318 or indirect address for the stqcrx. instruction. On little endian
3319 systems, swap the hi/lo registers before doing the stqcrx.
3321 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3322 insure the address is valid for the stqcrx. instruction.
3324 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3325 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3326 type of quad memory support is available.
3328 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3331 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3332 there is a danger of looping.
3334 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3336 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3337 force flag_ira_loop_pressure if set via command line.
3339 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3341 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3342 (ashr_simd): New builtin handling DI mode.
3343 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3344 (aarch64_sshr_simddi): New match pattern.
3345 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3346 (vshrd_n_s64): Likewise.
3347 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3349 2014-01-23 Nick Clifton <nickc@redhat.com>
3351 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3352 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3353 favour of mcu specific scripts.
3354 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3357 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3358 Alex Velenko <Alex.Velenko@arm.com>
3360 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3361 (vaddv_s16): Likewise.
3362 (vaddv_s32): Likewise.
3363 (vaddv_u8): Likewise.
3364 (vaddv_u16): Likewise.
3365 (vaddv_u32): Likewise.
3366 (vaddvq_s8): Likewise.
3367 (vaddvq_s16): Likewise.
3368 (vaddvq_s32): Likewise.
3369 (vaddvq_s64): Likewise.
3370 (vaddvq_u8): Likewise.
3371 (vaddvq_u16): Likewise.
3372 (vaddvq_u32): Likewise.
3373 (vaddvq_u64): Likewise.
3374 (vaddv_f32): Likewise.
3375 (vaddvq_f32): Likewise.
3376 (vaddvq_f64): Likewise.
3377 (vmaxv_f32): Likewise.
3378 (vmaxv_s8): Likewise.
3379 (vmaxv_s16): Likewise.
3380 (vmaxv_s32): Likewise.
3381 (vmaxv_u8): Likewise.
3382 (vmaxv_u16): Likewise.
3383 (vmaxv_u32): Likewise.
3384 (vmaxvq_f32): Likewise.
3385 (vmaxvq_f64): Likewise.
3386 (vmaxvq_s8): Likewise.
3387 (vmaxvq_s16): Likewise.
3388 (vmaxvq_s32): Likewise.
3389 (vmaxvq_u8): Likewise.
3390 (vmaxvq_u16): Likewise.
3391 (vmaxvq_u32): Likewise.
3392 (vmaxnmv_f32): Likewise.
3393 (vmaxnmvq_f32): Likewise.
3394 (vmaxnmvq_f64): Likewise.
3395 (vminv_f32): Likewise.
3396 (vminv_s8): Likewise.
3397 (vminv_s16): Likewise.
3398 (vminv_s32): Likewise.
3399 (vminv_u8): Likewise.
3400 (vminv_u16): Likewise.
3401 (vminv_u32): Likewise.
3402 (vminvq_f32): Likewise.
3403 (vminvq_f64): Likewise.
3404 (vminvq_s8): Likewise.
3405 (vminvq_s16): Likewise.
3406 (vminvq_s32): Likewise.
3407 (vminvq_u8): Likewise.
3408 (vminvq_u16): Likewise.
3409 (vminvq_u32): Likewise.
3410 (vminnmv_f32): Likewise.
3411 (vminnmvq_f32): Likewise.
3412 (vminnmvq_f64): Likewise.
3414 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3416 * config/aarch64/aarch64-simd.md
3417 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3418 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3419 (*aarch64_mul3_elt<mode>): Likewise.
3420 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3421 (*aarch64_mul3_elt_to_64v2df): Likewise.
3422 (*aarch64_mla_elt<mode>): Likewise.
3423 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3424 (*aarch64_mls_elt<mode>): Likewise.
3425 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3426 (*aarch64_fma4_elt<mode>): Likewise.
3427 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3428 (*aarch64_fma4_elt_to_64v2df): Likewise.
3429 (*aarch64_fnma4_elt<mode>): Likewise.
3430 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3431 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3432 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3433 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3434 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3435 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3436 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3437 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3438 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3440 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3442 * config/aarch64/aarch64-simd.md
3443 (aarch64_be_checked_get_lane<mode>): New define_expand.
3444 * config/aarch64/aarch64-simd-builtins.def
3445 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3446 New builtin definition.
3447 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3448 Use new safe be builtin.
3450 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3452 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3454 (aarch64_be_st1<mode>): Likewise.
3455 (aarch_ld1<VALL:mode>): Define_expand modified.
3456 (aarch_st1<VALL:mode>): Likewise.
3457 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3458 (UNSPEC_ST1): Likewise.
3460 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3462 * config/microblaze/microblaze.md: Add trap insn and attribute
3464 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3466 PR preprocessor/58580
3467 * input.h (location_get_source_line): Take an additional line_size
3469 (void diagnostics_file_cache_fini): Declare new function.
3470 * input.c (struct fcache): New type.
3471 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3472 New static constants.
3473 (diagnostic_file_cache_init, total_lines_num)
3474 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3475 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3476 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3477 (get_next_line, read_next_line, goto_next_line, read_line_num):
3478 New static function definitions.
3479 (diagnostic_file_cache_fini): New function.
3480 (location_get_source_line): Take an additional output line_len
3481 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3483 * diagnostic.c (diagnostic_finish): Call
3484 diagnostic_file_cache_fini.
3485 (adjust_line): Take an additional input parameter for the length
3486 of the line, rather than calculating it with strlen.
3487 (diagnostic_show_locus): Adjust the use of
3488 location_get_source_line and adjust_line with respect to their new
3489 signature. While displaying a line now, do not stop at the first
3490 null byte. Rather, display the zero byte as a space and keep
3491 going until we reach the size of the line.
3492 * Makefile.in: Add vec.o to OBJS-libcommon
3494 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3495 Ilya Tocar <ilya.tocar@intel.com>
3497 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3498 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3499 (__builtin_ia32_kmov16): Ditto.
3500 * config/i386/i386.md (UNSPEC_KMOV): New.
3503 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3505 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3506 (_mm512_storeu_si512): Ditto.
3508 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3511 * rtl.h (get_referenced_operands): Declare.
3512 * recog.c (get_referenced_operands): New function.
3513 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3514 operands have been referenced when recording LO_SUM references.
3516 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3518 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3520 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3522 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3523 Enable for generic and recent AMD targets.
3525 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3527 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3528 ARG_SIZE note when adjustment was eliminated.
3530 2014-01-22 Jeff Law <law@redhat.com>
3532 PR tree-optimization/59597
3533 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3534 in file. Accept new argument REGISTERING and use it to modify
3535 dump output appropriately.
3536 (register_jump_thread): Corresponding changes.
3537 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3538 thread paths involving joiner blocks. Add code to dump cancelled
3539 jump threading paths.
3541 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3543 PR rtl-optimization/59477
3544 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3545 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3547 2014-01-22 Tom Tromey <tromey@redhat.com>
3549 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3551 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3553 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3555 PR rtl-optimization/59896
3556 * lra-constraints.c (process_alt_operands): Check unused note for
3557 matched operands of insn with no output reloads.
3559 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3561 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3562 (mips_move_from_gpr_cost): Likewise.
3564 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3566 PR rtl-optimization/59858
3567 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3568 ira_class_hard_regs_num.
3569 (process_alt_operands): Increase reject for dying matched operand.
3571 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3574 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3575 smaller than size, perform several stores or loads and stores
3576 at dst + count - size to store or copy all of size bytes, rather
3577 than just last modesize bytes.
3579 2014-01-20 DJ Delorie <dj@redhat.com>
3581 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3582 that CLOBBERs are REGs before propogating their values.
3584 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3587 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3588 (cgraph_inline_failed_type): New function.
3589 * cgraph.h (DEFCIFCODE): Add type.
3590 (cgraph_inline_failed_type_t): New enum.
3591 (cgraph_inline_failed_type): New prototype.
3592 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3593 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3594 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3595 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3596 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3597 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3598 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3599 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3600 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3601 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3602 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3603 OPTIMIZATION_MISMATCH.
3604 * tree-inline.c (expand_call_inline): Emit errors during
3605 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3607 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3610 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3611 mode attribute in insn output.
3613 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3615 * output.h (output_constant): Delete.
3616 * varasm.c (output_constant): Make private.
3618 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3620 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3622 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3625 * tree.h (fold_builtin_strcat): New prototype.
3626 * builtins.c (fold_builtin_strcat): No longer static. Add len
3627 argument, if non-NULL, don't call c_strlen. Optimize
3628 directly into __builtin_memcpy instead of __builtin_strcpy.
3629 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3630 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3632 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3634 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3635 for SImode_address_operand operands, having only a REG argument.
3637 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3639 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3640 loader name using mbig-endian.
3641 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3643 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3645 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3649 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3651 * config/aarch64/aarch64-protos.h
3652 (aarch64_cannot_change_mode_class_ptr): Declare.
3653 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3654 aarch64_cannot_change_mode_class_ptr): New.
3655 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3656 backend hook aarch64_cannot_change_mode_class.
3658 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3660 * common/config/aarch64/aarch64-common.c
3661 (aarch64_handle_option): Don't handle any option order logic here.
3662 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3663 selected_cpu, warn on architecture version mismatch.
3664 (aarch64_override_options): Fix parsing order for option strings.
3666 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3667 Iain Sandoe <iain@codesourcery.com>
3670 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3671 warning. Amend comment to reflect current functionality.
3673 2014-01-20 Richard Biener <rguenther@suse.de>
3676 * builtins.c (fold_builtin_strcat): Remove case better handled
3677 by tree-ssa-strlen.c.
3679 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3681 * config/aarch64/aarch64.opt
3682 (mcpu, march, mtune): Make case-insensitive.
3684 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3687 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3688 if operands[1] is a REG or ZERO_EXTEND of a REG.
3690 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3692 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3694 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3696 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3697 long non-pic millicode calls.
3699 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3701 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3703 2014-01-19 Kito Cheng <kito@0xlab.org>
3705 * builtins.c (expand_movstr): Check movstr expand done or fail.
3707 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3708 H.J. Lu <hongjiu.lu@intel.com>
3711 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3712 to DImode for zero-extended addresses.
3714 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3716 PR rtl-optimization/57763
3717 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3718 on the new indirect jump_insn and increment LABEL_NUSES (label).
3720 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3724 * config.gcc (x86_archs): New variable.
3725 (x86_64_archs): Likewise.
3726 (x86_cpus): Likewise.
3727 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3728 --with-arch/--with-cpu= options.
3729 Support --with-arch=/--with-cpu={nehalem,westmere,
3730 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3732 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3734 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3735 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3737 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3739 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3741 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3744 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3745 clear cpp_get_options (parse_in)->warn_unused_macros for
3746 ix86_target_macros_internal with cpp_define.
3748 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3750 * jump.c (delete_related_insns): Keep (use (insn))s.
3751 * reorg.c (redundant_insn): Check for barriers too.
3753 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3755 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3757 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3759 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3760 call to $$dyncall when TARGET_LONG_CALLS is true.
3762 2014-01-17 Jeff Law <law@redhat.com>
3764 * ree.c (combine_set_extension): Temporarily disable test for
3765 changing number of hard registers.
3767 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3770 * ipa-inline-analysis.c (inline_free_summary):
3771 Do not free summary of aliases.
3773 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3776 * gimplify.c (gimplify_expr): Use create_tmp_var
3777 instead of create_tmp_var_raw. If cond doesn't have
3778 integral type, don't add the IFN_ANNOTATE builtin at all.
3780 2014-01-17 Martin Jambor <mjambor@suse.cz>
3783 * ipa-cp.c (prev_edge_clone): New variable.
3784 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3785 Also resize prev_edge_clone vector.
3786 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3787 (ipcp_edge_removal_hook): New function.
3788 (ipcp_driver): Register ipcp_edge_removal_hook.
3790 2014-01-17 Andrew Pinski <apinski@cavium.com>
3791 Steve Ellcey <sellcey@mips.com>
3794 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3797 2014-01-17 Jeff Law <law@redhat.com>
3800 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3801 so that pass_ccp runs first.
3803 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3805 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3806 (ix86_adjust_cost): Use !TARGET_XXX.
3807 (do_reorder_for_imul): Likewise.
3808 (swap_top_of_ready_list): Likewise.
3809 (ix86_sched_reorder): Likewise.
3811 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3813 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3814 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3815 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3816 (intel_memset): New. Duplicate slm_memset.
3817 (intel_cost): New. Duplicate slm_cost.
3818 (m_INTEL): New macro.
3819 (processor_target_table): Add "intel".
3820 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3821 with PROCESSOR_INTEL for "intel".
3822 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3823 PROCESSOR_SILVERMONT.
3824 (ix86_issue_rate): Likewise.
3825 (ix86_adjust_cost): Likewise.
3826 (ia32_multipass_dfa_lookahead): Likewise.
3827 (swap_top_of_ready_list): Likewise.
3828 (ix86_sched_reorder): Likewise.
3829 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3830 instead of TARGET_OPT_AGU.
3831 * config/i386/i386.h (TARGET_INTEL): New.
3832 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3833 (processor_type): Add PROCESSOR_INTEL.
3834 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3835 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3837 2014-01-17 Marek Polacek <polacek@redhat.com>
3840 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3843 2014-01-17 Richard Biener <rguenther@suse.de>
3845 PR tree-optimization/46590
3846 * opts.c (default_options_table): Add entries for
3847 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3848 all enabled at -O1 but not for -Og.
3849 * common.opt (fbranch-count-reg): Remove Init(1).
3850 (fmove-loop-invariants): Likewise.
3851 (ftree-pta): Likewise.
3853 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3855 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3856 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3857 decls to at least the GCC 4.8 used alignments.
3860 * tree-nested.c (convert_nonlocal_reference_stmt,
3861 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3862 of GIMPLE_BIND stmts, adjust associated decls.
3864 2014-01-17 Richard Biener <rguenther@suse.de>
3866 PR tree-optimization/46590
3867 * vec.h (vec<>::bseach): New member function implementing
3868 binary search according to C89 bsearch.
3869 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3870 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3871 bitmap pointer again. Make accesses_in_loop a flat array.
3872 (mem_ref_obstack): New global.
3873 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3874 (mark_ref_stored): Likewise.
3875 (ref_indep_loop_p_2): Likewise.
3876 (set_ref_stored_in_loop): New helper function.
3877 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3878 (memref_free): Adjust.
3879 (record_mem_ref_loc): Simplify.
3880 (gather_mem_refs_stmt): Adjust.
3881 (sort_locs_in_loop_postorder_cmp): New function.
3882 (analyze_memory_references): Sort accesses_in_loop after
3883 loop postorder number.
3884 (find_ref_loc_in_loop_cmp): New function.
3885 (for_all_locs_in_loop): Find relevant cluster of locs in
3886 accesses_in_loop and iterate without recursion.
3887 (execute_sm): Avoid uninit warning.
3888 (struct ref_always_accessed): Simplify.
3889 (ref_always_accessed::operator ()): Likewise.
3890 (ref_always_accessed_p): Likewise.
3891 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3892 loop postorder numbers here.
3893 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3896 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3899 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3900 on decls for which assemble_alias has been called.
3902 2014-01-17 Nick Clifton <nickc@redhat.com>
3904 * config/msp430/msp430.opt: (mcpu): New option.
3905 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3906 (msp430_option_override): Parse target_cpu. If the MCU name
3907 matches a generic string, clear target_mcu.
3908 (msp430_attr): Allow numeric interrupt values up to 63.
3909 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3910 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3912 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3914 * config/msp430/msp430.md (popm): Use %J rather than %I.
3915 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3916 (addhi_cy_i): Use immediate_operand for operand 2.
3917 * doc/invoke.texi: Document -mcpu option.
3919 2014-01-17 Richard Biener <rguenther@suse.de>
3921 PR rtl-optimization/38518
3922 * df.h (df_analyze_loop): Declare.
3923 * df-core.c: Include cfgloop.h.
3924 (df_analyze_1): Split out main part of df_analyze.
3925 (df_analyze): Adjust.
3926 (loop_inverted_post_order_compute): New function.
3927 (loop_post_order_compute): Likewise.
3928 (df_analyze_loop): New function avoiding whole-function
3930 * loop-invariant.c (find_defs): Use df_analyze_loop.
3931 (find_invariants): Adjust.
3932 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3934 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3936 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3937 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3939 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3941 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3942 traversal when removing references.
3944 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3947 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3949 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3952 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3953 pushing a reload for an autoinc when we had previously reloaded an
3954 inner part of the address.
3956 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3958 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3960 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3961 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3962 when not giving up or versioning for alias only because of
3964 (vect_analyze_data_ref_dependences): Set to true.
3965 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3967 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3968 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3972 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3975 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3976 operand 0 predicate for gathers, use a new pseudo as subtarget.
3978 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3981 * lra-constraints.c (process_alt_operands): Add printing debug info.
3982 Check absence of input/output reloads for matched operands too.
3984 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3986 PR rtl-optimization/59835
3987 * ira.c (ira_init_register_move_cost): Increase cost for
3990 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3992 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3994 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3997 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3998 non-register objects. Use gen_(high/low)part more consistently.
4001 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4004 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4005 endian support, remove tests for WORDS_BIG_ENDIAN.
4006 (p8_mfvsrd_3_<mode>): Likewise.
4007 (reload_gpr_from_vsx<mode>): Likewise.
4008 (reload_gpr_from_vsxsf): Likewise.
4009 (p8_mfvsrd_4_disf): Likewise.
4011 2014-01-16 Richard Biener <rguenther@suse.de>
4013 PR rtl-optimization/46590
4014 * lcm.c (compute_antinout_edge): Use postorder iteration.
4015 (compute_laterin): Use inverted postorder iteration.
4017 2014-01-16 Nick Clifton <nickc@redhat.com>
4020 * varasm.c (output_constant): Return the number of bytes actually
4022 (output_constructor_array_range): Update the field size with the
4023 number of bytes emitted by output_constant.
4024 (output_constructor_regular_field): Likewise. Also do not
4025 complain if the total number of bytes emitted is now greater
4026 than the expected fieldpos.
4027 * output.h (output_constant): Update prototype and descriptive comment.
4029 2014-01-16 Marek Polacek <polacek@redhat.com>
4032 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4033 it is error_mark_node.
4035 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4037 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4038 VALID_AVX256_REG_OR_OI_MODE.
4040 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4042 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4043 current procedure should be profiled.
4045 2014-01-15 Andrew Pinski <apinski@cavium.com>
4047 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4048 of moving from/to the STACK_REG register class.
4050 2014-01-15 Richard Henderson <rth@redhat.com>
4053 * reginfo.c (global_regs_decl): Globalize.
4054 * rtl.h (global_regs_decl): Declare.
4055 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4056 reserved via global_regs.
4058 2014-01-15 Teresa Johnson <tejohnson@google.com>
4060 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4062 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4064 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4065 and vmulosh rather than call gen_vec_widen_smult_*.
4066 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4067 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4068 (vec_widen_smult_even_v16qi): Likewise.
4069 (vec_widen_umult_even_v8hi): Likewise.
4070 (vec_widen_smult_even_v8hi): Likewise.
4071 (vec_widen_umult_odd_v16qi): Likewise.
4072 (vec_widen_smult_odd_v16qi): Likewise.
4073 (vec_widen_umult_odd_v8hi): Likewise.
4074 (vec_widen_smult_odd_v8hi): Likewise.
4075 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4076 vmuloub rather than call gen_vec_widen_umult_*.
4077 (vec_widen_umult_lo_v16qi): Likewise.
4078 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4079 vmulosb rather than call gen_vec_widen_smult_*.
4080 (vec_widen_smult_lo_v16qi): Likewise.
4081 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4082 rather than call gen_vec_widen_umult_*.
4083 (vec_widen_umult_lo_v8hi): Likewise.
4084 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4085 rather than call gen_vec_widen_smult_*.
4086 (vec_widen_smult_lo_v8hi): Likewise.
4088 2014-01-15 Jeff Law <law@redhat.com>
4090 PR tree-optimization/59747
4091 * ree.c (find_and_remove_re): Properly handle case where a second
4092 eliminated extension requires widening a copy created for elimination
4093 of a prior extension.
4094 (combine_set_extension): Ensure that the number of hard regs needed
4095 for a destination register does not change when we widen it.
4097 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4099 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4100 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4101 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4102 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4103 (avr-*-rtems*): Likewise.
4104 (bfin*-rtems*): Likewise.
4105 (moxie-*-rtems*): Likewise.
4106 (h8300-*-rtems*): Likewise.
4107 (i[34567]86-*-rtems*): Likewise.
4108 (lm32-*-rtems*): Likewise.
4109 (m32r-*-rtems*): Likewise.
4110 (m68k-*-rtems*): Likewise.
4111 (microblaze*-*-rtems*): Likewise.
4112 (mips*-*-rtems*): Likewise.
4113 (powerpc-*-rtems*): Likewise.
4114 (sh-*-rtems*): Likewise.
4115 (sparc-*-rtems*): Likewise.
4116 (sparc64-*-rtems*): Likewise.
4117 (v850-*-rtems*): Likewise.
4118 (m32c-*-rtems*): Likewise.
4120 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4122 PR rtl-optimization/59511
4123 * ira.c (ira_init_register_move_cost): Use memory costs for some
4124 cases of register move cost calculations.
4125 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4126 instead of BB frequency.
4127 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4128 * lra-assigns.c (find_hard_regno_for): Ditto.
4130 2014-01-15 Richard Biener <rguenther@suse.de>
4132 PR tree-optimization/59822
4133 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4134 (vectorizable_load): Use it to hoist defs of uses of invariant
4135 loads out of the loop.
4137 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4138 Kugan Vivekanandarajah <kuganv@linaro.org>
4141 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4144 2014-01-15 Richard Biener <rguenther@suse.de>
4146 PR rtl-optimization/59802
4147 * lcm.c (compute_available): Use inverted postorder to seed
4148 the initial worklist.
4150 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4153 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4154 ADDR_REGS for invalid symrefs in non-PIC code.
4156 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4159 * builtins.c (determine_block_size): Initialize *probable_max_size
4160 even if len_rtx is CONST_INT.
4162 2014-01-14 Andrew Pinski <apinski@cavium.com>
4164 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4165 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4166 (cortexa53_tunings): Likewise.
4167 (aarch64_sched_issue_rate): New function.
4168 (TARGET_SCHED_ISSUE_RATE): Define.
4170 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4172 * ira-costs.c (find_costs_and_classes): Add missed
4173 ira_init_register_move_cost_if_necessary.
4175 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4178 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4180 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4183 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4184 to indicate if type is used for function return value. Warn ABI
4185 change if the vector mode isn't available for function return value.
4186 (ix86_function_arg_advance): Pass false to type_natural_mode.
4187 (ix86_function_arg): Likewise.
4188 (ix86_gimplify_va_arg): Likewise.
4189 (function_arg_32): Don't warn ABI change.
4190 (ix86_function_value): Pass true to type_natural_mode.
4191 (ix86_return_in_memory): Likewise.
4192 (ix86_struct_value_rtx): Removed.
4193 (TARGET_STRUCT_VALUE_RTX): Likewise.
4195 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4197 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4198 converting a conditional jump into a conditional return.
4200 2014-01-14 Richard Biener <rguenther@suse.de>
4202 PR tree-optimization/58921
4203 PR tree-optimization/59006
4204 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4205 hoisting invariant stmts.
4206 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4207 invariant loads on the preheader edge if possible.
4209 2014-01-14 Joey Ye <joey.ye@arm.com>
4211 * doc/plugin.texi (Building GCC plugins): Update to C++.
4213 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4215 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4216 (_mm_rcp28_round_ss): Ditto.
4217 (_mm_rsqrt28_round_sd): Ditto.
4218 (_mm_rsqrt28_round_ss): Ditto.
4219 (_mm_rcp28_sd): Ditto.
4220 (_mm_rcp28_ss): Ditto.
4221 (_mm_rsqrt28_sd): Ditto.
4222 (_mm_rsqrt28_ss): Ditto.
4223 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4224 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4225 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4226 (IX86_BUILTIN_RCP28SD): Ditto.
4227 (IX86_BUILTIN_RCP28SS): Ditto.
4228 (IX86_BUILTIN_RSQRT28SD): Ditto.
4229 (IX86_BUILTIN_RSQRT28SS): Ditto.
4230 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4231 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4232 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4233 (ix86_expand_special_args_builtin): Expand new FTYPE.
4234 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4235 (srcp14<mode>): Make insn unary.
4236 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4237 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4238 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4239 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4240 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4241 Fix rounding: make it SAE only.
4242 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4244 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4246 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4247 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4248 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4249 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4250 (round_saeonly_mask_scalar_operand4): Ditto.
4251 (round_saeonly_mask_scalar_op3): Ditto.
4252 (round_saeonly_mask_scalar_op4): Ditto.
4254 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4256 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4257 Implement -maltivec=be for vec_insert and vec_extract.
4259 2014-01-10 DJ Delorie <dj@redhat.com>
4261 * config/msp430/msp430.md (call_internal): Don't allow memory
4262 references with SP as the base register.
4263 (call_value_internal): Likewise.
4264 * config/msp430/constraints.md (Yc): New. For memory references
4265 that don't use SP as a base register.
4267 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4268 "an integer without a # prefix"
4269 * config/msp430/msp430.md (epilogue_helper): Use it.
4271 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4274 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4275 AVX512F gather builtins.
4276 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4277 on gather decls with INTEGER_TYPE masktype.
4278 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4279 directly into the builtin rather than hoisting it before loop.
4281 PR tree-optimization/59387
4282 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4283 (scev_const_prop): If folded_casts and type has undefined overflow,
4284 use force_gimple_operand instead of force_gimple_operand_gsi and
4285 for each added stmt if it is assign with
4286 arith_code_with_undefined_signed_overflow, call
4287 rewrite_to_defined_overflow.
4288 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4289 gimple-fold.h instead.
4290 (arith_code_with_undefined_signed_overflow,
4291 rewrite_to_defined_overflow): Moved to ...
4292 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4293 rewrite_to_defined_overflow): ... here. No longer static.
4294 Include gimplify-me.h.
4295 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4296 rewrite_to_defined_overflow): New prototypes.
4298 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4300 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4302 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4304 * builtins.c (get_object_alignment_2): Minor tweak.
4305 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4307 2014-01-13 Christian Bruel <christian.bruel@st.com>
4309 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4310 optimized non constant lengths.
4312 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4315 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4316 load as __atomic_load_N if possible.
4318 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4320 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4322 (rs6000_expand_builtin): Adjust call.
4324 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4327 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4328 * config/rs6000/rs6000.c: Include target-globals.h.
4329 (rs6000_set_current_function): Instead of doing target_reinit
4330 unconditionally, use save_target_globals_default_opts and
4331 restore_target_globals.
4333 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4335 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4336 (rs6000_expand_builtin): Handle mffs and mtfsf.
4337 (rs6000_init_builtins): Define mffs and mtfsf.
4338 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4339 (rs6000_mffs): New pattern.
4340 (rs6000_mtfsf): New pattern.
4342 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4344 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4345 Start narrowing with START. Apply candidate-use pair
4346 and check overall cost in narrowing.
4347 (iv_ca_prune): Pass new argument.
4349 2014-01-10 Jeff Law <law@redhat.com>
4352 * ree.c (combine_reaching_defs): Ensure the defining statement
4353 occurs before the extension when optimizing extensions with
4354 different source and destination hard registers.
4356 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4359 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4360 vtables into the type inheritance graph.
4362 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4364 PR rtl-optimization/59754
4365 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4366 modes in the REGNO != REGNO case.
4368 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4370 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4372 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4374 PR tree-optimization/59745
4375 * tree-predcom.c (tree_predictive_commoning_loop): Call
4376 free_affine_expand_cache if giving up because components is NULL.
4378 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4379 GC in payload of target_globals struct instead of allocating them on
4380 the heap and the larger structs separately using GC.
4381 * target-globals.h (struct target_globals): Make regs, hard_regs,
4382 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4383 of GTY((skip)) and change type to void *.
4384 (reset_target_globals): Cast loads from those fields to corresponding
4387 2014-01-10 Steve Ellcey <sellcey@mips.com>
4390 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4391 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4392 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4394 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4397 * aarch64-modes.def (CC_Zmode): New flags mode.
4398 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4399 represents an equality.
4400 (aarch64_get_condition_code): Handle CC_Zmode.
4401 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4403 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4405 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4406 extraction in good case.
4408 2014-01-10 Richard Biener <rguenther@suse.de>
4410 PR tree-optimization/59374
4411 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4412 checking after SLP discovery. Mark stmts not participating
4413 in any SLP instance properly.
4415 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4417 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4418 when handling a SET rtx.
4420 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4422 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4423 (cortex-a57): Likewise.
4424 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4426 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4428 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4429 non-iwmmxt builtins.
4431 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4435 * ipa-devirt.c record_target_from_binfo): Take as argument
4436 stack of binfos and lookup matching one for virtual inheritance.
4437 (possible_polymorphic_call_targets_1): Update.
4439 2014-01-10 Huacai Chen <chenhc@lemote.com>
4441 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4442 kernel strings for Loongson-2E/2F/3A.
4444 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4447 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4448 is_gimple_call before calling gimple_call_internal_p.
4450 2014-01-09 Steve Ellcey <sellcey@mips.com>
4452 * Makefile.in (TREE_FLOW_H): Remove.
4453 (TREE_SSA_H): Add file names from tree-flow.h.
4454 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4455 * tree.h: Remove tree-flow.h reference.
4456 * hash-table.h: Remove tree-flow.h reference.
4457 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4458 reference with tree-ssa-loop.h.
4460 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4462 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4463 default element-order behavior for -maltivec.
4464 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4465 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4466 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4467 when targeting big endian, at least for now.
4468 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4470 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4473 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4474 var satisfies use_register_for_decl, just take into account type
4475 alignment, rather than decl alignment.
4477 PR tree-optimization/59622
4478 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4479 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4480 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4481 Don't devirtualize for inplace at all. For targets.length () == 1,
4482 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4484 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4486 * config/i386/i386.md (cpu): Remove the unused btver1.
4488 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4490 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4492 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4495 * tree-core.h (struct target_globals): New forward declaration.
4496 (struct tree_target_option): Add globals field.
4497 * tree.h (TREE_TARGET_GLOBALS): Define.
4498 (prepare_target_option_nodes_for_pch): New prototype.
4499 * target-globals.h (struct target_globals): Define even if
4501 * tree.c (prepare_target_option_node_for_pch,
4502 prepare_target_option_nodes_for_pch): New functions.
4503 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4504 * config/i386/i386.c: Include target-globals.h.
4505 (ix86_set_current_function): Instead of doing target_reinit
4506 unconditionally, use save_target_globals_default_opts and
4507 restore_target_globals.
4509 2014-01-09 Richard Biener <rguenther@suse.de>
4511 PR tree-optimization/59715
4512 * tree-cfg.h (split_critical_edges): Declare.
4513 * tree-cfg.c (split_critical_edges): Export.
4514 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4516 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4518 * cfgexpand.c (expand_stack_vars): Optionally disable
4519 asan stack protection.
4520 (expand_used_vars): Likewise.
4521 (partition_stack_vars): Likewise.
4522 * asan.c (asan_emit_stack_protection): Optionally disable
4523 after return stack usage.
4524 (instrument_derefs): Optionally disable memory access instrumentation.
4525 (instrument_builtin_call): Likewise.
4526 (instrument_strlen_call): Likewise.
4527 (asan_protect_global): Optionally disable global variables protection.
4528 * doc/invoke.texi: Added doc for new options.
4529 * params.def: Added new options.
4530 * params.h: Likewise.
4532 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4534 PR rtl-optimization/59724
4535 * ifcvt.c (cond_exec_process_if_block): Don't call
4536 flow_find_head_matching_sequence with 0 longest_match.
4537 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4538 non-active insns if !stop_after.
4539 (try_head_merge_bb): Revert 2014-01-07 changes.
4541 2014-01-08 Jeff Law <law@redhat.com>
4543 * ree.c (get_sub_rtx): New function, extracted from...
4544 (merge_def_and_ext): Here.
4545 (combine_reaching_defs): Use get_sub_rtx.
4547 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4549 * cgraph.h (varpool_variable_node): Do not choke on null node.
4551 2014-01-08 Catherine Moore <clm@codesourcery.com>
4553 * config/mips/mips.md (simple_return): Attempt to use JRC
4555 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4557 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4559 PR rtl-optimization/59137
4560 * reorg.c (steal_delay_list_from_target): Call update_block for
4562 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4564 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4566 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4567 two duplicate entries.
4569 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4572 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4574 * config/mips/mips.c (mips_truncated_op_cost): New function.
4575 (mips_rtx_costs): Adjust test for BADDU.
4576 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4578 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4580 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4581 (*baddu_si): ...this new pattern.
4583 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4586 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4588 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4591 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4593 (expand_expr, expand_normal): Adjust.
4594 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4595 inner_reference_p. Use inner_reference_p to expand inner references.
4596 (store_expr): Adjust.
4597 * cfgexpand.c (expand_call_stmt): Adjust.
4599 2014-01-08 Rong Xu <xur@google.com>
4601 * gcov-io.c (gcov_var): Move from gcov-io.h.
4602 (gcov_position): Ditto.
4603 (gcov_is_error): Ditto.
4604 (gcov_rewrite): Ditto.
4605 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4606 only part to libgcc/libgcov.h.
4608 2014-01-08 Marek Polacek <polacek@redhat.com>
4611 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4613 2014-01-08 Marek Polacek <polacek@redhat.com>
4616 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4618 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4620 PR rtl-optimization/59649
4621 * stor-layout.c (get_mode_bounds): For BImode return
4622 0 and STORE_FLAG_VALUE.
4624 2014-01-08 Richard Biener <rguenther@suse.de>
4627 * gimple.h (is_gimple_builtin_call): Remove.
4628 (gimple_builtin_call_types_compatible_p): New.
4629 (gimple_call_builtin_p): New overload.
4630 * gimple.c (is_gimple_builtin_call): Remove.
4631 (validate_call): Rename to ...
4632 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4634 (validate_type): New static function.
4635 (gimple_call_builtin_p): New overload and adjust.
4636 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4637 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4638 (gimple_fold_stmt_to_constant_1): Likewise.
4639 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4641 2014-01-08 Richard Biener <rguenther@suse.de>
4644 * gimplify.c (gimplify_expr): Gimplify register-register type
4645 VIEW_CONVERT_EXPRs to separate stmts.
4647 2014-01-07 Jeff Law <law@redhat.com>
4650 * ree.c (combine_set_extension): Handle case where source
4651 and destination registers in an extension insn are different.
4652 (combine_reaching_defs): Allow source and destination registers
4653 in extension to be different under limited circumstances.
4654 (add_removable_extension): Remove restriction that the
4655 source and destination registers in the extension are the same.
4656 (find_and_remove_re): Emit a copy from the extension's
4657 destination to its source after the defining insn if
4658 the source and destination registers are different.
4661 * ifcvt.c (merge_if_block): If we are merging a block with more than
4662 one successor with a block with no successors, remove any BARRIER
4663 after the second block.
4665 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4667 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4669 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4672 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4673 for 14-bit register offsets when INT14_OK_STRICT is false.
4675 2014-01-07 Roland Stigge <stigge@antcom.de>
4676 Michael Meissner <meissner@linux.vnet.ibm.com>
4679 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4680 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4682 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4684 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4687 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4689 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4690 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4691 rtx is const0_rtx or not.
4693 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4696 * target-globals.c (save_target_globals): Remove this_fn_optab
4698 * toplev.c: Include optabs.h.
4699 (target_reinit): Temporarily restore the global options if another
4700 set of options are in force.
4702 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4704 PR rtl-optimization/58668
4705 * cfgcleanup.c (flow_find_cross_jump): Don't count
4706 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4707 to determine what is counted.
4708 (flow_find_head_matching_sequence): Use active_insn_p to determine
4710 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4712 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4713 determine what is counted.
4715 PR tree-optimization/59643
4716 * tree-predcom.c (split_data_refs_to_components): If one dr is
4717 read and one write, determine_offset fails and the write isn't
4718 in the bad component, just put the read into the bad component.
4720 2014-01-07 Mike Stump <mikestump@comcast.net>
4721 Jakub Jelinek <jakub@redhat.com>
4724 * tree-core.h (struct tree_optimization_option): Change optabs
4725 type from unsigned char * to void *.
4726 * optabs.c (init_tree_optimization_optabs): Adjust
4727 TREE_OPTIMIZATION_OPTABS initialization.
4729 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4732 * config/i386/i386.h (struct machine_function): Add
4733 no_drap_save_restore field.
4734 * config/i386/i386.c (ix86_save_reg): Use
4735 !cfun->machine->no_drap_save_restore instead of
4736 crtl->stack_realign_needed.
4737 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4738 this function clears frame_pointer_needed. Set
4739 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4740 and DRAP reg is needed.
4742 2014-01-06 Marek Polacek <polacek@redhat.com>
4745 * doc/implement-c.texi: Mention that other integer types are
4746 permitted as bit-field types in strictly conforming mode.
4748 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4750 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4753 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4755 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4757 2014-01-06 Martin Jambor <mjambor@suse.cz>
4760 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4762 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4764 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4768 * var-tracking.c (add_stores): Preserve the value of the source even if
4769 we don't record the store.
4771 2014-01-06 Terry Guo <terry.guo@arm.com>
4773 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4775 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4778 * config/darwin.c (darwin_function_section): Adjust return values to
4779 correspond to optimisation changes made in r206070.
4781 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4783 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4784 from prefetch_block tune setting.
4785 (nocona_cost): Correct size of prefetch block to 64.
4787 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4789 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4790 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4791 used to save the static chain register in the computation of the offset
4792 from which the FP registers need to be restored.
4794 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4796 PR tree-optimization/59519
4797 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4798 ICE if get_current_def (current_new_name) is already non-NULL, as long
4799 as it is a phi result of some other phi in *new_exit_bb that has
4802 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4803 or vmovdqu* for misaligned_operand.
4804 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4805 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4806 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4807 aligned_mem for AVX512F masked aligned load and store builtins and for
4810 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4812 PR tree-optimization/59651
4813 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4814 Address range for negative step should be added by TYPE_SIZE_UNIT.
4816 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4818 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4819 overlapping registers also for registers other than the stack pointer.
4821 2014-01-03 Marek Polacek <polacek@redhat.com>
4824 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4827 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4830 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4833 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4834 (push splitter): Use <P:MODE_SIZE> instead of
4835 GET_MODE_SIZE (<P:MODE>mode).
4836 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4837 (mov -1, reg peephole2): Likewise.
4838 * config/i386/sse.md (*mov<mode>_internal,
4839 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4840 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4841 *<code><mode>3, *andnot<mode>3<mask_name>,
4842 <mask_codefor><code><mode>3<mask_name>): Likewise.
4843 * config/i386/subst.md (mask_mode512bit_condition,
4844 sd_mask_mode512bit_condition): Likewise.
4846 2014-01-02 Xinliang David Li <davidxl@google.com>
4848 PR tree-optimization/59303
4849 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4850 (dump_predicates): Better output format.
4851 (pred_equal_p): New function.
4852 (is_neq_relop_p): Ditto.
4853 (is_neq_zero_form_p): Ditto.
4854 (pred_expr_equal_p): Ditto.
4855 (pred_neg_p): Ditto.
4856 (simplify_pred): Ditto.
4857 (simplify_preds_2): Ditto.
4858 (simplify_preds_3): Ditto.
4859 (simplify_preds_4): Ditto.
4860 (simplify_preds): Ditto.
4862 (push_to_worklist): Ditto.
4863 (get_pred_info_from_cmp): Ditto.
4864 (is_degenerated_phi): Ditto.
4865 (normalize_one_pred_1): Ditto.
4866 (normalize_one_pred): Ditto.
4867 (normalize_one_pred_chain): Ditto.
4868 (normalize_preds): Ditto.
4869 (normalize_cond_1): Remove function.
4870 (normalize_cond): Ditto.
4871 (is_gcond_subset_of): Ditto.
4872 (is_subset_of_any): Ditto.
4873 (is_or_set_subset_of): Ditto.
4874 (is_and_set_subset_of): Ditto.
4875 (is_norm_cond_subset_of): Ditto.
4876 (pred_chain_length_cmp): Ditto.
4877 (convert_control_dep_chain_into_preds): Type change.
4878 (find_predicates): Ditto.
4879 (find_def_preds): Ditto.
4880 (destroy_predicates_vecs): Ditto.
4881 (find_matching_predicates_in_rest_chains): Ditto.
4882 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4883 (is_pred_expr_subset): Ditto.
4884 (is_pred_chain_subset_of): Ditto.
4885 (is_included_in): Ditto.
4886 (is_superset_of): Ditto.
4888 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4890 Update copyright years.
4892 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4894 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4895 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4896 config/arc/arc.md, config/arc/arc.opt,
4897 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4898 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4899 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4900 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4901 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4902 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4903 vtable-verify.c, vtable-verify.h: Use the standard form for the
4906 2014-01-02 Tobias Burnus <burnus@net-b.de>
4908 * gcc.c (process_command): Update copyright notice dates.
4909 * gcov-dump.c: Ditto.
4911 * doc/cpp.texi: Bump @copying's copyright year.
4912 * doc/cppinternals.texi: Ditto.
4913 * doc/gcc.texi: Ditto.
4914 * doc/gccint.texi: Ditto.
4915 * doc/gcov.texi: Ditto.
4916 * doc/install.texi: Ditto.
4917 * doc/invoke.texi: Ditto.
4919 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4921 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4923 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4925 * config/i386/sse.md (*mov<mode>_internal): Guard
4926 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4928 PR rtl-optimization/59647
4929 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4930 new_rtx into UNSIGNED_FLOAT rtxes.
4932 Copyright (C) 2014 Free Software Foundation, Inc.
4934 Copying and distribution of this file, with or without modification,
4935 are permitted in any medium without royalty provided the copyright
4936 notice and this notice are preserved.