gcc/
[official-gcc.git] / gcc / ChangeLog
blob41c1e8076832c43bcb9a3a552142a72f9c6f423c
1 2014-05-12  Georg-Johann Lay  <avr@gjlay.de>
3         * config/arm/arm.h (License): Add GCC Runtime Library Exception.
4         * config/arm/aout.h (License): Same.
5         * config/arm/bpabi.h (License): Same.
6         * config/arm/elf.h (License): Same.
7         * config/arm/linux-elf.h (License): Same.
8         * config/arm/linux-gas.h (License): Same.
9         * config/arm/netbsd-elf.h (License): Same.
10         * config/arm/uclinux-eabi.h (License): Same.
11         * config/arm/uclinux-elf.h (License): Same.
12         * config/arm/vxworks.h (License): Same.
14 2014-05-11  Jakub Jelinek  <jakub@redhat.com>
16         * tree.h (OMP_CLAUSE_LINEAR_STMT): Define.
17         * tree.c (omp_clause_num_ops): Increase OMP_CLAUSE_LINEAR
18         number of operands to 3.
19         (walk_tree_1): Walk all operands of OMP_CLAUSE_LINEAR.
20         * tree-nested.c (convert_nonlocal_omp_clauses,
21         convert_local_omp_clauses): Handle OMP_CLAUSE_DEPEND.
22         * gimplify.c (gimplify_scan_omp_clauses): Handle
23         OMP_CLAUSE_LINEAR_STMT.
24         * omp-low.c (lower_rec_input_clauses): Fix typo.
25         (maybe_add_implicit_barrier_cancel, lower_omp_1): Add
26         cast between Fortran boolean_type_node and C _Bool if
27         needed.
29 2014-05-11  Richard Sandiford  <rdsandiford@googlemail.com>
31         PR tree-optimization/61136
32         * wide-int.h (multiple_of_p): Define a version that doesn't return
33         the quotient.
34         * fold-const.c (extract_muldiv_1): Use wi::multiple_of_p instead of an
35         integer_zerop/const_binop pair.
36         (multiple_of_p): Likewise, converting both operands to widest_int
37         precision.
39 2014-05-09  Teresa Johnson  <tejohnson@google.com>
41         * cgraphunit.c (analyze_functions): Use correct dump file.
43 2014-05-09  Florian Weimer  <fweimer@redhat.com>
45         * cfgexpand.c (stack_protect_decl_p): New function, extracted from
46         expand_used_vars.
47         (stack_protect_return_slot_p): New function.
48         (expand_used_vars): Call stack_protect_decl_p and
49         stack_protect_return_slot_p for -fstack-protector-strong.
51 2014-05-09  Kenneth Zadeck  <zadeck@naturalbridge.com>
53         PR middle-end/61111
54         * fold-const.c (fold_binary_loc): Changed width of mask.
56 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
58         * config/avr/avr-fixed.md (round<mode>3): Use -1U instead of -1 in
59         unsigned int initializers for regno_in, regno_out.
61 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
63         PR target/61055
64         * config/avr/avr.md (cc): Add new attribute set_vzn.
65         (addqi3, addqq3, adduqq3, subqi3, subqq3, subuqq3, negqi2) [cc]:
66         Set cc insn attribute to set_vzn instead of set_zn for alternatives
67         with INC, DEC or NEG.
68         * config/avr/avr.c (avr_notice_update_cc): Handle SET_VZN.
69         (avr_out_plus_1): ADIW sets cc0 to CC_SET_CZN.
70         INC, DEC and ADD+ADC set cc0 to CC_CLOBBER.
72 2014-05-09  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
74         Revert:
75         2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
77         * wide-int.cc (UTItype): Define.
78         (UDWtype): Define for appropriate W_TYPE_SIZE.
80 2014-05-09  Richard Biener  <rguenther@suse.de>
82         * Makefile.in (GTFILES): Remove tree-ssa-propagate.c.
83         * tree-ssa-propagate.c: Do not include gt-tree-ssa-propagate.h.
84         (interesting_ssa_edges, varying_ssa_edges): Move out of GC space.
85         (add_ssa_edge, process_ssa_edge_worklist, ssa_prop_init,
86         ssa_propagate): Adjust.
88 2014-05-08  Jeff Law  <law@redhat.com>
90         PR tree-optimization/61009
91         * tree-ssa-threadedge.c (thread_through_normal_block): Return a
92         tri-state rather than a boolean.  When a block is too big to
93         thread through, inform caller via negative return value.
94         (thread_across_edge): If a block was too big for normal threading,
95         then it's too big for a joiner too, so remove temporary equivalences
96         and return immediately.
98 2014-05-08  Manuel López-Ibáñez  <manu@gcc.gnu.org>
99             Matthias Klose  <doko@ubuntu.com>
101         PR driver/61106
102         * optc-gen.awk: Fix option handling for -Wunused-parameter.
104 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
106         PR target/59952
107         * config/i386/i386.c (PTA_HASWELL): Remove PTA_RTM.
109 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
111         PR target/61092
112         * config/alpha/alpha.c: Include gimple-iterator.h.
113         (alpha_gimple_fold_builtin): New function.  Move
114         ALPHA_BUILTIN_UMULH folding from ...
115         (alpha_fold_builtin): ... here.
116         (TARGET_GIMPLE_FOLD_BUILTIN): New define.
118 2014-05-08  Wei Mi  <wmi@google.com>
120         PR target/58066
121         * config/i386/i386.c (ix86_compute_frame_layout): Update
122         preferred_stack_boundary for call, expanded from tls descriptor.
123         * config/i386/i386.md (*tls_global_dynamic_32_gnu): Update RTX
124         to depend on SP register.
125         (*tls_local_dynamic_base_32_gnu): Ditto.
126         (*tls_local_dynamic_32_once): Ditto.
127         (tls_global_dynamic_64_<mode>): Set
128         ix86_tls_descriptor_calls_expanded_in_cfun.
129         (tls_local_dynamic_base_64_<mode>): Ditto.
130         (tls_global_dynamic_32): Set
131         ix86_tls_descriptor_calls_expanded_in_cfun. Update RTX
132         to depend on SP register.
133         (tls_local_dynamic_base_32): Ditto.
135 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
137         * config/arm/arm_neon.h: Update comment.
138         * config/arm/neon-docgen.ml: Delete.
139         * config/arm/neon-gen.ml: Delete.
140         * doc/arm-neon-intrinsics.texi: Update comment.
142 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
144         * config/arm/arm_neon_builtins.def (vadd, vsub): Only define the v2sf
145         and v4sf versions.
146         (vand, vorr, veor, vorn, vbic): Remove.
147         * config/arm/neon.md (neon_vadd, neon_vsub, neon_vadd_unspec): Adjust
148         iterator.
149         (neon_vsub_unspec): Likewise.
150         (neon_vorr, neon_vand, neon_vbic, neon_veor, neon_vorn): Remove.
152 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
154         * config/arm/arm_neon.h (vadd_s8): GNU C implementation
155         (vadd_s16): Likewise.
156         (vadd_s32): Likewise.
157         (vadd_f32): Likewise.
158         (vadd_u8): Likewise.
159         (vadd_u16): Likewise.
160         (vadd_u32): Likewise.
161         (vadd_s64): Likewise.
162         (vadd_u64): Likewise.
163         (vaddq_s8): Likewise.
164         (vaddq_s16): Likewise.
165         (vaddq_s32): Likewise.
166         (vaddq_s64): Likewise.
167         (vaddq_f32): Likewise.
168         (vaddq_u8): Likewise.
169         (vaddq_u16): Likewise.
170         (vaddq_u32): Likewise.
171         (vaddq_u64): Likewise.
172         (vmul_s8): Likewise.
173         (vmul_s16): Likewise.
174         (vmul_s32): Likewise.
175         (vmul_f32): Likewise.
176         (vmul_u8): Likewise.
177         (vmul_u16): Likewise.
178         (vmul_u32): Likewise.
179         (vmul_p8): Likewise.
180         (vmulq_s8): Likewise.
181         (vmulq_s16): Likewise.
182         (vmulq_s32): Likewise.
183         (vmulq_f32): Likewise.
184         (vmulq_u8): Likewise.
185         (vmulq_u16): Likewise.
186         (vmulq_u32): Likewise.
187         (vsub_s8): Likewise.
188         (vsub_s16): Likewise.
189         (vsub_s32): Likewise.
190         (vsub_f32): Likewise.
191         (vsub_u8): Likewise.
192         (vsub_u16): Likewise.
193         (vsub_u32): Likewise.
194         (vsub_s64): Likewise.
195         (vsub_u64): Likewise.
196         (vsubq_s8): Likewise.
197         (vsubq_s16): Likewise.
198         (vsubq_s32): Likewise.
199         (vsubq_s64): Likewise.
200         (vsubq_f32): Likewise.
201         (vsubq_u8): Likewise.
202         (vsubq_u16): Likewise.
203         (vsubq_u32): Likewise.
204         (vsubq_u64): Likewise.
205         (vand_s8): Likewise.
206         (vand_s16): Likewise.
207         (vand_s32): Likewise.
208         (vand_u8): Likewise.
209         (vand_u16): Likewise.
210         (vand_u32): Likewise.
211         (vand_s64): Likewise.
212         (vand_u64): Likewise.
213         (vandq_s8): Likewise.
214         (vandq_s16): Likewise.
215         (vandq_s32): Likewise.
216         (vandq_s64): Likewise.
217         (vandq_u8): Likewise.
218         (vandq_u16): Likewise.
219         (vandq_u32): Likewise.
220         (vandq_u64): Likewise.
221         (vorr_s8): Likewise.
222         (vorr_s16): Likewise.
223         (vorr_s32): Likewise.
224         (vorr_u8): Likewise.
225         (vorr_u16): Likewise.
226         (vorr_u32): Likewise.
227         (vorr_s64): Likewise.
228         (vorr_u64): Likewise.
229         (vorrq_s8): Likewise.
230         (vorrq_s16): Likewise.
231         (vorrq_s32): Likewise.
232         (vorrq_s64): Likewise.
233         (vorrq_u8): Likewise.
234         (vorrq_u16): Likewise.
235         (vorrq_u32): Likewise.
236         (vorrq_u64): Likewise.
237         (veor_s8): Likewise.
238         (veor_s16): Likewise.
239         (veor_s32): Likewise.
240         (veor_u8): Likewise.
241         (veor_u16): Likewise.
242         (veor_u32): Likewise.
243         (veor_s64): Likewise.
244         (veor_u64): Likewise.
245         (veorq_s8): Likewise.
246         (veorq_s16): Likewise.
247         (veorq_s32): Likewise.
248         (veorq_s64): Likewise.
249         (veorq_u8): Likewise.
250         (veorq_u16): Likewise.
251         (veorq_u32): Likewise.
252         (veorq_u64): Likewise.
253         (vbic_s8): Likewise.
254         (vbic_s16): Likewise.
255         (vbic_s32): Likewise.
256         (vbic_u8): Likewise.
257         (vbic_u16): Likewise.
258         (vbic_u32): Likewise.
259         (vbic_s64): Likewise.
260         (vbic_u64): Likewise.
261         (vbicq_s8): Likewise.
262         (vbicq_s16): Likewise.
263         (vbicq_s32): Likewise.
264         (vbicq_s64): Likewise.
265         (vbicq_u8): Likewise.
266         (vbicq_u16): Likewise.
267         (vbicq_u32): Likewise.
268         (vbicq_u64): Likewise.
269         (vorn_s8): Likewise.
270         (vorn_s16): Likewise.
271         (vorn_s32): Likewise.
272         (vorn_u8): Likewise.
273         (vorn_u16): Likewise.
274         (vorn_u32): Likewise.
275         (vorn_s64): Likewise.
276         (vorn_u64): Likewise.
277         (vornq_s8): Likewise.
278         (vornq_s16): Likewise.
279         (vornq_s32): Likewise.
280         (vornq_s64): Likewise.
281         (vornq_u8): Likewise.
282         (vornq_u16): Likewise.
283         (vornq_u32): Likewise.
284         (vornq_u64): Likewise.
286 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
288         * wide-int.cc (UTItype): Define.
289         (UDWtype): Define for appropriate W_TYPE_SIZE.
291 2014-05-08  Marc Glisse  <marc.glisse@inria.fr>
293         PR tree-optimization/59100
294         * tree-ssa-phiopt.c: Include tree-inline.h.
295         (neutral_element_p, absorbing_element_p): New functions.
296         (value_replacement): Handle conditional binary operations with a
297         neutral or absorbing element.
299 2014-05-08  Richard Biener  <rguenther@suse.de>
301         * tree-ssa-sccvn.c (vn_get_expr_for): Valueize operands before
302         folding the expression.
303         (valueize_expr): Remove.
304         (visit_reference_op_load): Do not valueize the result of
305         vn_get_expr_for.
306         (simplify_binary_expression): Likewise.
307         (simplify_unary_expression): Likewise.
309 2014-05-08  Richard Biener  <rguenther@suse.de>
311         * gimplify.c (gimplify_call_expr): Use saved fnptrtype for
312         looking at TYPE_ARG_TYPES.
314 2014-05-08  Richard Biener  <rguenther@suse.de>
316         * gimple-fold.c (gimple_fold_stmt_to_constant_1): Remove
317         pointer propagation special-case.
319 2014-05-08  Bin Cheng  <bin.cheng@arm.com>
321         * tree-affine.c (tree_to_aff_combination): Handle MEM_REF for
322         core part of address expressions.
324 2014-05-08  Alan Modra  <amodra@gmail.com>
326         PR target/60737
327         * config/rs6000/rs6000.c (expand_block_move): Allow 64-bit
328         loads and stores when -mno-strict-align at any alignment.
329         (expand_block_clear): Similarly.  Also correct calculation of
330         instruction count.
332 2014-05-07  Thomas Preud'homme  <thomas.preudhomme@arm.com>
334         PR middle-end/39246
335         * tree-complex.c (expand_complex_move): Keep line info when expanding
336         complex move.
337         * tree-ssa-uninit.c (warn_uninit): New argument. Ignore assignment
338         of complex expression. Use new argument to display correct location
339         for values coming from phi statement.
340         (warn_uninitialized_vars): Adapt to new signature of warn_uninit.
341         (warn_uninitialized_phi): Pass location of phi argument to
342         warn_uninit.
343         * tree-ssa.c (ssa_undefined_value_p): For SSA_NAME initialized by a
344         COMPLEX_EXPR, recurse on each part of the COMPLEX_EXPR.
346 2014-05-07  Segher Boessenkool  <segher@kernel.crashing.org>
348         * config/rs6000/predicates.md (indexed_address_mem): New.
349         * config/rs6000/rs6000.md (type): Remove load_ext, load_ext_u,
350         load_ext_ux, load_ux, load_u, store_ux, store_u, fpload_ux, fpload_u,
351         fpstore_ux, fpstore_u.
352         (sign_extend, indexed, update): New.
353         (cell_micro): Adjust.
354         (*zero_extend<mode>di2_internal1, *zero_extendsidi2_lfiwzx,
355         *extendsidi2_lfiwax, *extendsidi2_nocell, *extendsfdf2_fpr,
356         *movsi_internal1, *movsi_internal1_single, *movhi_internal,
357         *movqi_internal, *movcc_internal1, mov<mode>_hardfloat,
358         *mov<mode>_softfloat, *mov<mode>_hardfloat32, *mov<mode>_hardfloat64,
359         *mov<mode>_softfloat64, *movdi_internal32, *movdi_internal64,
360         *mov<mode>_string, *ldmsi8, *ldmsi7, *ldmsi6, *ldmsi5, *ldmsi4,
361         *ldmsi3, *stmsi8, *stmsi7, *stmsi6, *stmsi5, *stmsi4, *stmsi3,
362         *movdi_update1, movdi_<mode>_update, movdi_<mode>_update_stack,
363         *movsi_update1, *movsi_update2, movsi_update, movsi_update_stack,
364         *movhi_update1, *movhi_update2, *movhi_update3, *movhi_update4,
365         *movqi_update1, *movqi_update2, *movqi_update3, *movsf_update1,
366         *movsf_update2, *movsf_update3, *movsf_update4, *movdf_update1,
367         *movdf_update2, load_toc_aix_si, load_toc_aix_di, probe_stack_<mode>,
368         *stmw, *lmw, as well as 10 anonymous patterns): Adjust.
370         * config/rs6000/dfp.md (movsd_store, movsd_load): Adjust.
371         * config/rs6000/vsx.md (*vsx_movti_32bit, *vsx_extract_<mode>_load,
372         *vsx_extract_<mode>_store): Adjust.
373         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_microcoded_insn,
374         is_cracked_insn, insn_must_be_first_in_group,
375         insn_must_be_last_in_group): Adjust.
377         * config/rs6000/40x.md (ppc403-load, ppc403-store, ppc405-float):
378         Adjust.
379         * config/rs6000/440.md (ppc440-load, ppc440-store, ppc440-fpload,
380         ppc440-fpstore): Adjust.
381         * config/rs6000/476.md (ppc476-load, ppc476-store, ppc476-fpload,
382         ppc476-fpstore): Adjust.
383         * config/rs6000/601.md (ppc601-load, ppc601-store, ppc601-fpload,
384         ppc601-fpstore): Adjust.
385         * config/rs6000/603.md (ppc603-load, ppc603-store, ppc603-fpload):
386         Adjust.
387         * config/rs6000/6xx.md (ppc604-load, ppc604-store, ppc604-fpload):
388         Adjust.
389         * config/rs6000/7450.md (ppc7450-load, ppc7450-store, ppc7450-fpload,
390         ppc7450-fpstore): Adjust.
391         * config/rs6000/7xx.md (ppc750-load, ppc750-store): Adjust.
392         * config/rs6000/8540.md (ppc8540_load, ppc8540_store): Adjust.
393         * config/rs6000/a2.md (ppca2-load, ppca2-fp-load, ppca2-fp-store):
394         Adjust.
395         * config/rs6000/cell.md (cell-load, cell-load-ux, cell-load-ext,
396         cell-fpload, cell-fpload-update, cell-store, cell-store-update,
397         cell-fpstore, cell-fpstore-update): Adjust.
398         * config/rs6000/e300c2c3.md (ppce300c3_load, ppce300c3_fpload,
399         ppce300c3_store, ppce300c3_fpstore): Adjust.
400         * config/rs6000/e500mc.md (e500mc_load, e500mc_fpload, e500mc_store,
401         e500mc_fpstore): Adjust.
402         * config/rs6000/e500mc64.md (e500mc64_load, e500mc64_fpload,
403         e500mc64_store, e500mc64_fpstore): Adjust.
404         * config/rs6000/e5500.md (e5500_load, e5500_fpload, e5500_store,
405         e5500_fpstore): Adjust.
406         * config/rs6000/e6500.md (e6500_load, e6500_fpload, e6500_store,
407         e6500_fpstore): Adjust.
408         * config/rs6000/mpc.md (mpccore-load, mpccore-store, mpccore-fpload):
409         Adjust.
410         * config/rs6000/power4.md (power4-load, power4-load-ext,
411         power4-load-ext-update, power4-load-ext-update-indexed,
412         power4-load-update-indexed, power4-load-update, power4-fpload,
413         power4-fpload-update, power4-store, power4-store-update,
414         power4-store-update-indexed, power4-fpstore, power4-fpstore-update):
415         Adjust.
416         * config/rs6000/power5.md (power5-load, power5-load-ext,
417         power5-load-ext-update, power5-load-ext-update-indexed,
418         power5-load-update-indexed, power5-load-update, power5-fpload,
419         power5-fpload-update, power5-store, power5-store-update,
420         power5-store-update-indexed, power5-fpstore, power5-fpstore-update):
421         Adjust.
422         * config/rs6000/power6.md (power6-load, power6-load-ext,
423         power6-load-update, power6-load-update-indexed,
424         power6-load-ext-update, power6-load-ext-update-indexed, power6-fpload,
425         power6-fpload-update, power6-store, power6-store-update,
426         power6-store-update-indexed, power6-fpstore, power6-fpstore-update):
427         Adjust.
428         * config/rs6000/power7.md (power7-load, power7-load-ext,
429         power7-load-update, power7-load-update-indexed,
430         power7-load-ext-update, power7-load-ext-update-indexed, power7-fpload,
431         power7-fpload-update, power7-store, power7-store-update,
432         power7-store-update-indexed, power7-fpstore, power7-fpstore-update):
433         Adjust.
434         * config/rs6000/power8.md (power8-load, power8-load-update,
435         power8-load-ext, power8-load-ext-update, power8-fpload,
436         power8-fpload-update, power8-store, power8-store-update-indexed,
437         power8-fpstore, power8-fpstore-update): Adjust.
438         * config/rs6000/rs64.md (rs64a-load, rs64a-store, rs64a-fpload):
439         Adjust.
440         * config/rs6000/titan.md (titan_lsu_load, titan_lsu_fpload,
441         titan_lsu_store, titan_lsu_fpstore): Adjust.
442         * config/rs6000/xfpu.md (fp-load, fp-store): Adjust.
444 2014-05-07  Oleg Endo  <olegendo@gcc.gnu.org>
446         PR target/60884
447         * config/sh/sh-mem.cc (sh_expand_strlen): Use loop when emitting
448         unrolled byte insns.  Emit address increments after move insns.
450 2014-05-07  David Malcolm  <dmalcolm@redhat.com>
452         * gimple.h (gimple_builtin_call_types_compatible_p): Accept a
453         const_gimple, rather than a gimple.
454         (gimple_call_builtin_p): Likewise, for the three variants.
456         * gimple.c (gimple_builtin_call_types_compatible_p): Likewise.
457         (gimple_call_builtin_p): Likewise, for the three variants.
459 2014-05-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
461         PR tree-optimization/61095
462         * tree-ssanames.c (get_nonzero_bits): Fix type extension in wi::shwi.
464 2014-05-07  Richard Biener  <rguenther@suse.de>
466         PR tree-optimization/61034
467         * tree-ssa-alias.c (call_may_clobber_ref_p_1): Export.
468         (maybe_skip_until): Use translate to take into account
469         lattices when trying to do disambiguations.
470         (get_continuation_for_phi_1): Likewise.
471         (get_continuation_for_phi): Adjust for added translate arguments.
472         (walk_non_aliased_vuses): Likewise.
473         * tree-ssa-alias.h (get_continuation_for_phi): Adjust prototype.
474         (walk_non_aliased_vuses): Likewise.
475         (call_may_clobber_ref_p_1): Declare.
476         * tree-ssa-sccvn.c (vn_reference_lookup_3): Also disambiguate against
477         calls.  Stop early if we are only supposed to disambiguate.
478         * tree-ssa-pre.c (translate_vuse_through_block): Adjust.
480 2014-05-07  Joern Rennecke  <joern.rennecke@embecosm.com>
482         * config/epiphany/epiphany.c (epiphany_handle_interrupt_attribute):
483         Emit an error when the function has arguments.
485 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
487         * cfgloop.h (unswitch_loops): Remove.
488         * doc/passes.texi: Remove references to loop-unswitch.c
489         * timevar.def (TV_LOOP_UNSWITCH): Remove.
491 2014-05-07  Evgeny Stupachenko  <evstupac@gmail.com>
493         * tree-vect-data-refs.c (vect_grouped_load_supported): New
494         check for loads group of length 3.
495         (vect_permute_load_chain): New permutations for loads group of
496         length 3.
497         * tree-vect-stmts.c (vect_model_load_cost): Change cost
498         of vec_perm_shuffle for the new permutations.
500 2014-05-07  Alan Lawrence  <alan.lawrence@arm.com>
502         * config/aarch64/arm_neon.h (vtrn1_f32, vtrn1_p8, vtrn1_p16, vtrn1_s8,
503         vtrn1_s16, vtrn1_s32, vtrn1_u8, vtrn1_u16, vtrn1_u32, vtrn1q_f32,
504         vtrn1q_f64, vtrn1q_p8, vtrn1q_p16, vtrn1q_s8, vtrn1q_s16, vtrn1q_s32,
505         vtrn1q_s64, vtrn1q_u8, vtrn1q_u16, vtrn1q_u32, vtrn1q_u64, vtrn2_f32,
506         vtrn2_p8, vtrn2_p16, vtrn2_s8, vtrn2_s16, vtrn2_s32, vtrn2_u8,
507         vtrn2_u16, vtrn2_u32, vtrn2q_f32, vtrn2q_f64, vtrn2q_p8, vtrn2q_p16,
508         vtrn2q_s8, vtrn2q_s16, vtrn2q_s32, vtrn2q_s64, vtrn2q_u8, vtrn2q_u16,
509         vtrn2q_u32, vtrn2q_u64): Replace temporary asm with __builtin_shuffle.
511 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
513         * loop-unswitch.c: Delete.
515 2014-05-07  Richard Biener  <rguenther@suse.de>
517         * config.gcc: Always set need_64bit_hwint to yes.
519 2014-05-07  Chung-Ju Wu  <jasonwucj@gmail.com>
521         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Have it in favor
522         of using optimize_size.
524 2014-05-06  Mike Stump  <mikestump@comcast.net>
526         * wide-int.h (wi::int_traits <HOST_WIDE_INT>): Always define.
528 2014-05-06  Joseph Myers  <joseph@codesourcery.com>
530         * config/i386/sse.md (*mov<mode>_internal)
531         (*<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>)
532         (*<sse2_avx_avx512f>_loaddqu<mode><mask_name>)
533         (<sse>_andnot<mode>3, <code><mode>3, *andnot<mode>3)
534         (*<code><mode>3, *andnot<mode>3<mask_name>)
535         (<mask_codefor><code><mode>3<mask_name>): Only consider
536         TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL for modes of size 16.
538 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
540         Revert:
541         2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
543         * lra-constraints.c (valid_address_p): Move earlier in file.
544         Add a constraint argument to the address_info version.
545         (satisfies_memory_constraint_p): New function.
546         (satisfies_address_constraint_p): Likewise.
547         (process_alt_operands, curr_insn_transform): Use them.
548         (process_address): Pass the constraint to valid_address_p when
549         checking address operands.
551 2014-05-06  Richard Sandiford  <r.sandiford@uk.ibm.com>
553         * lto-cgraph.c (compute_ltrans_boundary): Make node variables local
554         to their respective blocks.  Fix inadvertent use of "node".
556 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
558         * emit-rtl.c (init_derived_machine_modes): New functionm, split
559         out from...
560         (init_emit_once): ...here.
561         * rtl.h (init_derived_machine_modes): Declare.
562         * toplev.c (do_compile): Call it even if no_backend.
564 2014-05-06  Kenneth Zadeck  <zadeck@naturalbridge.com>
565             Mike Stump  <mikestump@comcast.net>
566             Richard Sandiford  <rdsandiford@googlemail.com>
567             Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
569         * alias.c (ao_ref_from_mem): Use wide-int interfaces.
570         (rtx_equal_for_memref_p): Update comment.
571         (adjust_offset_for_component_ref): Use wide-int interfaces.
572         * builtins.c (get_object_alignment_2): Likewise.
573         (c_readstr): Likewise.
574         (target_char_cast): Add comment.
575         (determine_block_size): Use wide-int interfaces.
576         (expand_builtin_signbit): Likewise.
577         (fold_builtin_int_roundingfn): Likewise.
578         (fold_builtin_bitop): Likewise.
579         (fold_builtin_bswap): Likewise.
580         (fold_builtin_logarithm): Use signop.
581         (fold_builtin_pow): Likewise.
582         (fold_builtin_memory_op): Use wide-int interfaces.
583         (fold_builtin_object_size): Likewise.
584         * cfgloop.c (alloc_loop): Initialize nb_iterations_upper_bound and
585         nb_iterations_estimate.
586         (record_niter_bound): Use wide-int interfaces.
587         (get_estimated_loop_iterations_int): Likewise.
588         (get_estimated_loop_iterations): Likewise.
589         (get_max_loop_iterations): Likewise.
590         * cfgloop.h: Include wide-int.h.
591         (struct nb_iter_bound): Change bound to widest_int.
592         (struct loop): Change nb_iterations_upper_bound and
593         nb_iterations_estimate to widest_int.
594         (record_niter_bound): Switch to use widest_int.
595         (get_estimated_loop_iterations): Likewise.
596         (get_max_loop_iterations): Likewise.
597         (gcov_type_to_double_int): Rename to gcov_type_to_wide_int and
598         update for wide-int.
599         * cgraph.c (cgraph_add_thunk): Use wide-int interfaces.
600         * combine.c (try_combine): Likewise.
601         (subst): Use CONST_SCALAR_INT_P rather than CONST_INT_P.
602         * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Use wide-int
603         interfaces.
604         (aarch64_float_const_representable_p): Likewise.
605         * config/arc/arc.c: Include wide-int.h.
606         (arc_can_use_doloop_p): Use wide-int interfaces.
607         * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
608         (vfp3_const_double_index): Likewise.
609         * config/avr/avr.c (avr_out_round): Likewise.
610         (avr_fold_builtin): Likewise.
611         * config/bfin/bfin.c (bfin_local_alignment): Likewise.
612         (bfin_can_use_doloop_p): Likewise.
613         * config/darwin.c (darwin_mergeable_constant_section): Likewise.
614         (machopic_select_rtx_section): Update to handle CONST_WIDE_INT.
615         * config/i386/i386.c: Include wide-int.h.
616         (ix86_data_alignment): Use wide-int interfaces.
617         (ix86_local_alignment): Likewise.
618         (ix86_emit_swsqrtsf): Update real_from_integer.
619         * config/msp430/msp430.c (msp430_attr): Use wide-int interfaces.
620         * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
621         * config/rs6000/predicates.md (any_operand): Add const_wide_int.
622         (zero_constant): Likewise.
623         (input_operand): Likewise.
624         (splat_input_operand): Likewise.
625         (non_logical_cint_operand): Change const_double to const_wide_int.
626         * config/rs6000/rs6000.c (num_insns_constant): Handle CONST_WIDE_INT.
627         (easy_altivec_constant): Remove comment.
628         (paired_expand_vector_init): Use CONSTANT_P.
629         (rs6000_legitimize_address): Handle CONST_WIDE_INT.
630         (rs6000_emit_move): Update checks.
631         (rs6000_aggregate_candidate): Use wide-int interfaces.
632         (rs6000_expand_ternop_builtin): Likewise.
633         (rs6000_output_move_128bit): Handle CONST_WIDE_INT.
634         (rs6000_assemble_integer): Likewise.
635         (rs6000_hash_constant): Likewise.
636         (output_toc): Likewise.
637         (rs6000_rtx_costs): Likewise.
638         (rs6000_emit_swrsqrt); Update call to real_from_integer.
639         * config/rs6000/rs6000-c.c: Include wide-int.h.
640         (altivec_resolve_overloaded_builtin): Use wide-int interfaces.
641         * config/rs6000/rs6000.h (TARGET_SUPPORTS_WIDE_INT): New.
642         * config/rs6000/rs6000.md: Use const_scalar_int_operand.
643         Handle CONST_WIDE_INT.
644         * config/sol2-c.c (solaris_pragma_align): Change low to unsigned HWI.
645         Use tree_fits_uhwi_p.
646         * config/sparc/sparc.c: Include wide-int.h.
647         (sparc_fold_builtin): Use wide-int interfaces.
648         * config/vax/vax.c: Include wide-int.h.
649         (vax_float_literal): Use real_from_integer.
650         * coretypes.h (struct hwivec_def): New.
651         (hwivec): New.
652         (const_hwivec): New.
653         * cse.c (hash_rtx_cb): Handle CONST_WIDE_INT.
654         (equiv_constant): Handle CONST_WIDE_INT.
655         * cselib.c (rtx_equal_for_cselib_1): Use CASE_CONST_UNIQUE.
656         (cselib_hash_rtx): Handle CONST_WIDE_INT.
657         * dbxout.c (stabstr_U): Use wide-int interfaces.
658         (dbxout_type): Update to use cst_fits_shwi_p.
659         * defaults.h (LOG2_BITS_PER_UNIT): Define.
660         (TARGET_SUPPORTS_WIDE_INT): Add default.
661         * dfp.c: Include wide-int.h.
662         (decimal_real_to_integer2): Use wide-int interfaces and rename to
663         decimal_real_to_integer.
664         * dfp.h (decimal_real_to_integer2): Return a wide_int and rename to
665         decimal_real_to_integer.
666         * doc/generic.texi (Constant expressions): Update for wide_int.
667         * doc/rtl.texi (const_double): Likewise.
668         (const_wide_int, CONST_WIDE_INT, CONST_WIDE_INT_VEC): New.
669         (CONST_WIDE_INT_NUNITS, CONST_WIDE_INT_ELT): New.
670         * doc/tm.texi.in (REAL_VALUE_TO_INT): Remove.
671         (REAL_VALUE_FROM_INT): Remove.
672         (TARGET_SUPPORTS_WIDE_INT): New.
673         * doc/tm.texi: Regenerate.
674         * dojump.c (prefer_and_bit_test): Use wide-int interfaces.
675         * double-int.h: Include wide-int.h.
676         (struct wi::int_traits): New.
677         * dwarf2out.c (get_full_len): New.
678         (dw_val_equal_p): Add case dw_val_class_wide_int.
679         (size_of_loc_descr): Likewise.
680         (output_loc_operands): Likewise.
681         (insert_double): Remove.
682         (insert_wide_int): New.
683         (add_AT_wide): New.
684         (print_die): Add case dw_val_class_wide_int.
685         (attr_checksum): Likewise.
686         (attr_checksum_ordered): Likewise.
687         (same_dw_val_p): Likewise.
688         (size_of_die): Likewise.
689         (value_format): Likewise.
690         (output_die): Likewise.
691         (double_int_type_size_in_bits): Rename to offset_int_type_size_in_bits.
692         Use wide-int.
693         (clz_loc_descriptor): Use wide-int interfaces.
694         (mem_loc_descriptor): Likewise.  Handle CONST_WIDE_INT.
695         (loc_descriptor): Use wide-int interfaces.  Handle CONST_WIDE_INT.
696         (round_up_to_align): Use wide-int interfaces.
697         (field_byte_offset): Likewise.
698         (insert_double): Rename to insert_wide_int.  Use wide-int interfaces.
699         (add_const_value_attribute): Handle CONST_WIDE_INT.  Update
700         CONST_DOUBLE handling.  Use wide-int interfaces.
701         (add_bound_info): Use tree_fits_uhwi_p.  Use wide-int interfaces.
702         (gen_enumeration_type_die): Use add_AT_wide.
703         (hash_loc_operands): Add case dw_val_class_wide_int.
704         (compare_loc_operands): Likewise.
705         * dwarf2out.h: Include wide-int.h.
706         (wide_int_ptr): New.
707         (enum dw_val_class): Add dw_val_class_wide_int.
708         (struct dw_val_struct): Add val_wide.
709         * emit-rtl.c (const_wide_int_htab): New.
710         (const_wide_int_htab_hash): New.
711         (const_wide_int_htab_eq): New.
712         (lookup_const_wide_int): New.
713         (const_double_htab_hash): Use wide-int interfaces.
714         (const_double_htab_eq): Likewise.
715         (rtx_to_double_int): Conditionally compile for wide-int.
716         (immed_double_int_const): Rename to immed_wide_int_const and
717         update for wide-int.
718         (immed_double_const): Conditionally compile for wide-int.
719         (init_emit_once): Use wide-int interfaces.
720         * explow.c (plus_constant): Likewise.
721         * expmed.c (mask_rtx): Move further up file.  Use wide-int interfaces.
722         (lshift_value): Use wide-int interfaces.
723         (expand_mult): Likewise.
724         (choose_multiplier): Likewise.
725         (expand_smod_pow2): Likewise.
726         (make_tree): Likewise.
727         * expr.c (convert_modes): Consolidate handling of constants.
728         Use wide-int interfaces.
729         (emit_group_load_1): Add note.
730         (store_expr): Update comment.
731         (get_inner_reference): Use wide-int interfaces.
732         (expand_constructor): Update comment.
733         (expand_expr_real_2): Use wide-int interfaces.
734         (expand_expr_real_1): Likewise.
735         (reduce_to_bit_field_precision): Likewise.
736         (const_vector_from_tree): Likewise.
737         * final.c: Include wide-int-print.h.
738         (output_addr_const): Handle CONST_WIDE_INT.  Use CONST_DOUBLE_AS_INT_P.
739         * fixed-value.c: Include wide-int.h.
740         (fixed_from_string): Use wide-int interfaces.
741         (fixed_to_decimal): Likewise.
742         (fixed_convert_from_real): Likewise.
743         (real_convert_from_fixed): Likewise.
744         * fold-const.h (mem_ref_offset): Return an offset_int.
745         (div_if_zero_remainder): Remove code parameter.
746         * fold-const.c (div_if_zero_remainder): Remove code parameter.
747         Use wide-int interfaces.
748         (may_negate_without_overflow_p): Use wide-int interfaces.
749         (negate_expr_p): Likewise.
750         (fold_negate_expr): Likewise.
751         (int_const_binop_1): Likewise.
752         (const_binop): Likewise.
753         (fold_convert_const_int_from_int): Likewise.
754         (fold_convert_const_int_from_real): Likewise.
755         (fold_convert_const_int_from_fixed): Likewise.
756         (fold_convert_const_fixed_from_int): Likewise.
757         (all_ones_mask_p): Take an unsigned size.  Use wide-int interfaces.
758         (sign_bit_p): Use wide-int interfaces.
759         (make_range_step): Likewise.
760         (build_range_check): Likewise.  Pass an integer of the correct type
761         instead of using integer_one_node.
762         (range_predecessor): Pass an integer of the correct type instead
763         of using integer_one_node.
764         (range_successor): Likewise.
765         (merge_ranges): Likewise.
766         (unextend): Use wide-int interfaces.
767         (extract_muldiv_1): Likewise.
768         (fold_div_compare): Likewise.
769         (fold_single_bit_test): Likewise.
770         (fold_sign_changed_comparison): Likewise.
771         (try_move_mult_to_index): Update calls to div_if_zero_remainder.
772         (fold_plusminus_mult_expr): Use wide-int interfaces.
773         (native_encode_int): Likewise.
774         (native_interpret_int): Likewise.
775         (fold_unary_loc): Likewise.
776         (pointer_may_wrap_p): Likewise.
777         (size_low_cst): Likewise.
778         (mask_with_tz): Likewise.
779         (fold_binary_loc): Likewise.
780         (fold_ternary_loc): Likewise.
781         (multiple_of_p): Likewise.
782         (tree_call_nonnegative_warnv_p): Update calls to
783         tree_int_cst_min_precision and real_from_integer.
784         (fold_negate_const): Use wide-int interfaces.
785         (fold_abs_const): Likewise.
786         (fold_relational_const): Use tree_int_cst_lt.
787         (round_up_loc): Use wide-int interfaces.
788         * genemit.c (gen_exp): Add CONST_WIDE_INT case.
789         * gengenrtl.c (excluded_rtx): Add CONST_WIDE_INT case.
790         * gengtype.c: Remove include of double-int.h.
791         (do_typedef): Use wide-int interfaces.
792         (open_base_files): Add wide-int.h.
793         (main): Add offset_int and widest_int typedefs.
794         * gengtype-lex.l: Handle "^".
795         (CXX_KEYWORD): Add "static".
796         * gengtype-parse.c (require3): New.
797         (require_template_declaration): Handle constant template arguments
798         and nested templates.
799         * gengtype-state.c: Don't include "double-int.h".
800         * genpreds.c (write_one_predicate_function): Update comment.
801         (write_tm_constrs_h): Add check for hval and lval use in
802         CONST_WIDE_INT.
803         * genrecog.c (validate_pattern): Add CONST_WIDE_INT case.
804         (add_to_sequence): Likewise.
805         * gensupport.c (struct std_pred_table): Add const_scalar_int_operand
806         and const_double_operand.
807         * gimple.c (preprocess_case_label_vec_for_gimple): Use wide-int
808         interfaces.
809         * gimple-fold.c (get_base_constructor): Likewise.
810         (fold_array_ctor_reference): Likewise.
811         (fold_nonarray_ctor_reference): Likewise.
812         (fold_const_aggregate_ref_1): Likewise.
813         (gimple_val_nonnegative_real_p): Likewise.
814         (gimple_fold_indirect_ref): Likewise.
815         * gimple-pretty-print.c (dump_ssaname_info): Likewise.
816         * gimple-ssa-strength-reduction.c: Include wide-int-print.h.
817         (struct slsr_cand_d): Change index to be widest_int.
818         (struct incr_info_d): Change incr to be widest_int.
819         (alloc_cand_and_find_basis): Use wide-int interfaces.
820         (slsr_process_phi): Likewise.
821         (backtrace_base_for_ref): Likewise.  Return a widest_int.
822         (restructure_reference): Take a widest_int instead of a double_int.
823         (slsr_process_ref): Use wide-int interfaces.
824         (create_mul_ssa_cand): Likewise.
825         (create_mul_imm_cand): Likewise.
826         (create_add_ssa_cand): Likewise.
827         (create_add_imm_cand): Take a widest_int instead of a double_int.
828         (slsr_process_add): Use wide-int interfaces.
829         (slsr_process_cast): Likewise.
830         (slsr_process_copy): Likewise.
831         (dump_candidate): Likewise.
832         (dump_incr_vec): Likewise.
833         (replace_ref): Likewise.
834         (cand_increment): Likewise.  Return a widest_int.
835         (cand_abs_increment): Likewise.
836         (replace_mult_candidate): Take a widest_int instead of a double_int.
837         (replace_unconditional_candidate): Use wide-int interfaces.
838         (incr_vec_index): Take a widest_int instead of a double_int.
839         (create_add_on_incoming_edge): Likewise.
840         (create_phi_basis): Use wide-int interfaces.
841         (replace_conditional_candidate): Likewise.
842         (record_increment): Take a widest_int instead of a double_int.
843         (record_phi_increments): Use wide-int interfaces.
844         (phi_incr_cost): Take a widest_int instead of a double_int.
845         (lowest_cost_path): Likewise.
846         (total_savings): Likewise.
847         (analyze_increments): Use wide-int interfaces.
848         (ncd_with_phi): Take a widest_int instead of a double_int.
849         (ncd_of_cand_and_phis): Likewise.
850         (nearest_common_dominator_for_cands): Likewise.
851         (insert_initializers): Use wide-int interfaces.
852         (all_phi_incrs_profitable): Likewise.
853         (replace_one_candidate): Likewise.
854         (replace_profitable_candidates): Likewise.
855         * godump.c: Include wide-int-print.h.
856         (go_output_typedef): Use wide-int interfaces.
857         * graphite-clast-to-gimple.c (gmp_cst_to_tree): Likewise.
858         * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
859         (build_loop_iteration_domains): Likewise.
860         * hooks.h: Include wide-int.h rather than double-int.h.
861         (hook_bool_dint_dint_uint_bool_true): Delete.
862         (hook_bool_wint_wint_uint_bool_true): Declare.
863         * hooks.c (hook_bool_dint_dint_uint_bool_true): Removed.
864         (hook_bool_wint_wint_uint_bool_true): New.
865         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): Use wide-int
866         interfaces.
867         (ubsan_expand_si_overflow_mul_check): Likewise.
868         * ipa-devirt.c (get_polymorphic_call_info): Likewise.
869         * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
870         (get_ancestor_addr_info): Likewise.
871         (ipa_modify_call_arguments): Likewise.
872         * loop-doloop.c (doloop_modify): Likewise.
873         (doloop_optimize): Likewise.
874         * loop-iv.c (iv_number_of_iterations): Likewise.
875         * loop-unroll.c (decide_unroll_constant_iterations): Likewise.
876         (unroll_loop_constant_iterations): Likewise.
877         (decide_unroll_runtime_iterations): Likewise.
878         (unroll_loop_runtime_iterations): Likewise.
879         (decide_peel_simple): Likewise.
880         (decide_unroll_stupid): Likewise.
881         * lto-streamer-in.c (streamer_read_wi): Add.
882         (input_cfg): Use wide-int interfaces.
883         (lto_input_tree_1): Likewise.
884         * lto-streamer-out.c (streamer_write_wi): Add.
885         (hash_tree): Use wide-int interfaces.
886         (output_cfg): Likewise.
887         * Makefile.in (OBJS): Add wide-int.o and wide-int-print.o.
888         (GTFILES): Add wide-int.h and signop.h.
889         (TAGS): Look for .cc files too.
890         * omp-low.c (scan_omp_1_op): Use wide-int interfaces.
891         * optabs.c (expand_subword_shift): Likewise.
892         (expand_doubleword_shift): Likewise.
893         (expand_absneg_bit): Likewise.
894         (expand_copysign_absneg): Likewise.
895         (expand_copysign_bit): Likewise.
896         * postreload.c (reload_cse_simplify_set): Likewise.
897         * predict.c (predict_iv_comparison): Likewise.
898         * pretty-print.h: Include wide-int-print.h.
899         (pp_wide_int) New.
900         * print-rtl.c (print_rtx): Add CONST_WIDE_INT case.
901         * print-tree.c: Include wide-int-print.h.
902         (print_node_brief): Use wide-int interfaces.
903         (print_node): Likewise.
904         * read-rtl.c (validate_const_wide_int): New.
905         (read_rtx_code): Add CONST_WIDE_INT case.
906         * real.c: Include wide-int.h.
907         (real_to_integer2): Delete.
908         (real_to_integer): New function, returning a wide_int.
909         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
910         (ten_to_ptwo): Update call to real_from_integer.
911         (real_digit): Likewise.
912         * real.h: Include signop.h, wide-int.h and insn-modes.h.
913         (real_to_integer2, REAL_VALUE_FROM_INT, REAL_VALUE_FROM_UNSIGNED_INT)
914         (REAL_VALUE_TO_INT): Delete.
915         (real_to_integer): Declare a wide-int form.
916         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
917         * recog.c (const_int_operand): Improve comment.
918         (const_scalar_int_operand): New.
919         (const_double_operand): Add a separate definition for CONST_WIDE_INT.
920         * rtlanal.c (commutative_operand_precedence): Handle CONST_WIDE_INT.
921         (split_double): Likewise.
922         * rtl.c (DEF_RTL_EXPR): Handle CONST_WIDE_INT.
923         (rtx_size): Likewise.
924         (rtx_alloc_stat_v): New.
925         (rtx_alloc_stat): Now calls rtx_alloc_stat_v.
926         (cwi_output_hex): New.
927         (iterative_hash_rtx): Handle CONST_WIDE_INT.
928         (cwi_check_failed_bounds): New.
929         * rtl.def (CONST_WIDE_INT): New.
930         * rtl.h: Include <utility> and wide-int.h.
931         (struct hwivec_def): New.
932         (CWI_GET_NUM_ELEM): New.
933         (CWI_PUT_NUM_ELEM): New.
934         (struct rtx_def): Add num_elem and hwiv.
935         (CASE_CONST_SCALAR_INT): Modify for TARGET_SUPPORTS_WIDE_INT.
936         (CASE_CONST_UNIQUE): Likewise.
937         (CASE_CONST_ANY): Likewise.
938         (CONST_SCALAR_INT_P): Likewise.
939         (CONST_WIDE_INT_P): New.
940         (CWI_ELT): New.
941         (HWIVEC_CHECK): New.
942         (cwi_check_failed_bounds): New.
943         (CWI_ELT): New.
944         (HWIVEC_CHECK): New.
945         (CONST_WIDE_INT_VEC) New.
946         (CONST_WIDE_INT_NUNITS) New.
947         (CONST_WIDE_INT_ELT) New.
948         (rtx_mode_t): New type.
949         (wi::int_traits <rtx_mode_t>): New.
950         (wi::shwi): New.
951         (wi::min_value): New.
952         (wi::max_value): New.
953         (rtx_alloc_v) New.
954         (const_wide_int_alloc): New.
955         (immed_wide_int_const): New.
956         * sched-vis.c (print_value): Handle CONST_WIDE_INT.
957         * sel-sched-ir.c (lhs_and_rhs_separable_p): Update comment.
958         * signop.h: New file.
959         * simplify-rtx.c (mode_signbit_p): Handle CONST_WIDE_INT.
960         (simplify_const_unary_operation): Use wide-int interfaces.
961         (simplify_binary_operation_1): Likewise.
962         (simplify_const_binary_operation): Likewise.
963         (simplify_const_relational_operation): Likewise.
964         (simplify_immed_subreg): Likewise.
965         * stmt.c (expand_case): Likewise.
966         * stor-layout.h (set_min_and_max_values_for_integral_type): Take a
967         signop rather than a bool.
968         * stor-layout.c (layout_type): Use wide-int interfaces.
969         (initialize_sizetypes): Update calls to
970         set_min_and_max_values_for_integral_type.
971         (set_min_and_max_values_for_integral_type): Take a signop rather
972         than a bool.  Use wide-int interfaces.
973         (fixup_signed_type): Update accordingly.  Remove
974         HOST_BITS_PER_DOUBLE_INT limit.
975         (fixup_unsigned_type): Likewise.
976         * system.h (STATIC_CONSTANT_P): New.
977         (STATIC_ASSERT): New.
978         * target.def (can_use_doloop_p): Take widest_ints rather than
979         double_ints.
980         * target.h: Include wide-int.h rather than double-int.h.
981         * targhooks.h (can_use_doloop_if_innermost): Take widest_ints rather
982         than double_ints.
983         * targhooks.c (default_cxx_get_cookie_size): Use tree_int_cst_lt
984         rather than INT_CST_LT_UNSIGNED.
985         (can_use_doloop_if_innermost): Take widest_ints rather than
986         double_ints.
987         * tree-affine.c: Include wide-int-print.h.
988         (double_int_ext_for_comb): Delete.
989         (wide_int_ext_for_comb): New.
990         (aff_combination_zero): Use wide-int interfaces.
991         (aff_combination_const): Take a widest_int instead of a double_int.
992         (aff_combination_elt): Use wide-int interfaces.
993         (aff_combination_scale): Take a widest_int instead of a double_int.
994         (aff_combination_add_elt): Likewise.
995         (aff_combination_add_cst): Likewise.
996         (aff_combination_add): Use wide-int interfaces.
997         (aff_combination_convert): Likewise.
998         (tree_to_aff_combination): Likewise.
999         (add_elt_to_tree): Take a widest_int instead of a double_int.
1000         (aff_combination_to_tree): Use wide-int interfaces.
1001         (aff_combination_remove_elt): Likewise.
1002         (aff_combination_add_product): Take a widest_int instead of
1003         a double_int.
1004         (aff_combination_mult): Use wide-int interfaces.
1005         (aff_combination_expand): Likewise.
1006         (double_int_constant_multiple_p): Delete.
1007         (wide_int_constant_multiple_p): New.
1008         (aff_combination_constant_multiple_p): Take a widest_int pointer
1009         instead of a double_int pointer.
1010         (print_aff): Use wide-int interfaces.
1011         (get_inner_reference_aff): Take a widest_int pointer
1012         instead of a double_int pointer.
1013         (aff_comb_cannot_overlap_p): Take widest_ints instead of double_ints.
1014         * tree-affine.h: Include wide-int.h.
1015         (struct aff_comb_elt): Change type of coef to widest_int.
1016         (struct affine_tree_combination): Change type of offset to widest_int.
1017         (double_int_ext_for_comb): Delete.
1018         (wide_int_ext_for_comb): New.
1019         (aff_combination_const): Use widest_int instead of double_int.
1020         (aff_combination_scale): Likewise.
1021         (aff_combination_add_elt): Likewise.
1022         (aff_combination_constant_multiple_p): Likewise.
1023         (get_inner_reference_aff): Likewise.
1024         (aff_comb_cannot_overlap_p): Likewise.
1025         (aff_combination_zero_p): Use wide-int interfaces.
1026         * tree.c: Include tree.h.
1027         (init_ttree): Use make_int_cst.
1028         (tree_code_size): Removed code for INTEGER_CST case.
1029         (tree_size): Add INTEGER_CST case.
1030         (make_node_stat): Update comment.
1031         (get_int_cst_ext_nunits, build_new_int_cst, build_int_cstu): New.
1032         (build_int_cst_type): Use wide-int interfaces.
1033         (double_int_to_tree): Likewise.
1034         (double_int_fits_to_tree_p): Delete.
1035         (force_fit_type_double): Delete.
1036         (force_fit_type): New.
1037         (int_cst_hash_hash): Use wide-int interfaces.
1038         (int_cst_hash_eq): Likewise.
1039         (build_int_cst_wide): Delete.
1040         (wide_int_to_tree): New.
1041         (cache_integer_cst): Use wide-int interfaces.
1042         (build_low_bits_mask): Likewise.
1043         (cst_and_fits_in_hwi): Likewise.
1044         (real_value_from_int_cst): Likewise.
1045         (make_int_cst_stat): New.
1046         (integer_zerop): Use wide_int interfaces.
1047         (integer_onep): Likewise.
1048         (integer_all_onesp): Likewise.
1049         (integer_pow2p): Likewise.
1050         (integer_nonzerop): Likewise.
1051         (tree_log2): Likewise.
1052         (tree_floor_log2): Likewise.
1053         (tree_ctz): Likewise.
1054         (int_size_in_bytes): Likewise.
1055         (mem_ref_offset): Return an offset_int rather than a double_int.
1056         (build_type_attribute_qual_variant): Use wide_int interfaces.
1057         (type_hash_eq): Likewise
1058         (tree_int_cst_equal): Likewise.
1059         (tree_int_cst_lt): Delete.
1060         (tree_int_cst_compare): Likewise.
1061         (tree_fits_shwi_p): Use wide_int interfaces.
1062         (tree_fits_uhwi_p): Likewise.
1063         (tree_int_cst_sign_bit): Likewise.
1064         (tree_int_cst_sgn): Likewise.
1065         (tree_int_cst_min_precision): Take a signop rather than a bool.
1066         (simple_cst_equal): Use wide_int interfaces.
1067         (compare_tree_int): Likewise.
1068         (iterative_hash_expr): Likewise.
1069         (int_fits_type_p): Likewise.  Use tree_int_cst_lt rather than
1070         INT_CST_LT.
1071         (get_type_static_bounds): Use wide_int interfaces.
1072         (tree_int_cst_elt_check_failed): New.
1073         (build_common_tree_nodes): Reordered to set prec before filling in
1074         value.
1075         (int_cst_value): Check cst_and_fits_in_hwi.
1076         (widest_int_cst_value): Use wide_int interfaces.
1077         (upper_bound_in_type): Likewise.
1078         (lower_bound_in_type): Likewise.
1079         (num_ending_zeros): Likewise.
1080         (drop_tree_overflow): Likewise.
1081         * tree-call-cdce.c (check_pow): Update call to real_from_integer.
1082         (gen_conditions_for_pow_cst_base): Likewise.
1083         * tree-cfg.c: Include wide-int.h and wide-int-print.h.
1084         (group_case_labels_stmt): Use wide-int interfaces.
1085         (verify_gimple_assign_binary): Likewise.
1086         (print_loop): Likewise.
1087         * tree-chrec.c (tree_fold_binomial): Likewise.
1088         * tree-core.h (struct tree_base): Add int_length.
1089         (struct tree_int_cst): Change rep of value.
1090         * tree-data-ref.c (dr_analyze_innermost): Use wide-int interfaces.
1091         (dr_may_alias_p): Likewise.
1092         (max_stmt_executions_tree): Likewise.
1093         * tree.def (INTEGER_CST): Update comment.
1094         * tree-dfa.c (get_ref_base_and_extent): Use wide-int interfaces.
1095         * tree-dfa.h (get_addr_base_and_unit_offset_1): Likewise.
1096         * tree-dump.c: Include wide-int.h and wide-int-print.h.
1097         (dequeue_and_dump): Use wide-int interfaces.
1098         * tree.h: Include wide-int.h.
1099         (NULL_TREE): Moved to earlier loc in file.
1100         (TREE_INT_CST_ELT_CHECK): New.
1101         (tree_int_cst_elt_check_failed): New.
1102         (TYPE_SIGN): New.
1103         (TREE_INT_CST): Delete.
1104         (TREE_INT_CST_LOW): Use wide-int interfaces.
1105         (TREE_INT_CST_HIGH): Delete.
1106         (TREE_INT_CST_NUNITS): New.
1107         (TREE_INT_CST_EXT_NUNITS): Likewise.
1108         (TREE_INT_CST_OFFSET_NUNITS): Likewise.
1109         (TREE_INT_CST_ELT): Likewise.
1110         (INT_CST_LT): Delete.
1111         (tree_int_cst_elt_check): New (two forms).
1112         (type_code_size): Update comment.
1113         (make_int_cst_stat, make_int_cst): New.
1114         (tree_to_double_int): Delete.
1115         (double_int_fits_to_tree_p): Delete.
1116         (force_fit_type_double): Delete.
1117         (build_int_cstu): Replace with out-of-line function.
1118         (build_int_cst_wide): Delete.
1119         (tree_int_cst_lt): Define inline.
1120         (tree_int_cst_le): New.
1121         (tree_int_cst_compare): Define inline.
1122         (tree_int_cst_min_precision): Take a signop rather than a bool.
1123         (wi::int_traits <const_tree>): New.
1124         (wi::int_traits <tree>): New.
1125         (wi::extended_tree): New.
1126         (wi::int_traits <wi::extended_tree>): New.
1127         (wi::to_widest): New.
1128         (wi::to_offset): New.
1129         (wi::fits_to_tree_p): New.
1130         (wi::min_value): New.
1131         (wi::max_value): New.
1132         * tree-inline.c (remap_gimple_op_r): Use wide-int interfaces.
1133         (copy_tree_body_r): Likewise.
1134         * tree-object-size.c (compute_object_offset): Likewise.
1135         (addr_object_size): Likewise.
1136         * tree-predcom.c: Include wide-int-print.h.
1137         (struct dref_d): Change type of offset to widest_int.
1138         (dump_dref): Call wide-int printer.
1139         (aff_combination_dr_offset): Use wide-int interfaces.
1140         (determine_offset): Take a widest_int pointer rather than a
1141         double_int pointer.
1142         (split_data_refs_to_components): Use wide-int interfaces.
1143         (suitable_component_p): Likewise.
1144         (order_drefs): Likewise.
1145         (add_ref_to_chain): Likewise.
1146         (valid_initializer_p): Likewise.
1147         (determine_roots_comp): Likewise.
1148         * tree-pretty-print.c: Include wide-int-print.h.
1149         (dump_generic_node): Use wide-int interfaces.
1150         * tree-sra.c (sra_ipa_modify_expr): Likewise.
1151         * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1152         (move_fixed_address_to_symbol): Likewise.
1153         (move_hint_to_base): Likewise.
1154         (move_pointer_to_base): Likewise.
1155         (move_variant_to_index): Likewise.
1156         (most_expensive_mult_to_index): Likewise.
1157         (addr_to_parts): Likewise.
1158         (copy_ref_info): Likewise.
1159         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1160         (indirect_refs_may_alias_p): Likewise.
1161         (stmt_kills_ref_p_1): Likewise.
1162         * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
1163         * tree-ssa-ccp.c: Update comment at top of file.  Include
1164         wide-int-print.h.
1165         (struct prop_value_d): Change type of mask to widest_int.
1166         (extend_mask): New function.
1167         (dump_lattice_value): Use wide-int interfaces.
1168         (get_default_value): Likewise.
1169         (set_constant_value): Likewise.
1170         (set_value_varying): Likewise.
1171         (valid_lattice_transition): Likewise.
1172         (set_lattice_value): Likewise.
1173         (value_to_double_int): Delete.
1174         (value_to_wide_int): New.
1175         (get_value_from_alignment): Use wide-int interfaces.
1176         (get_value_for_expr): Likewise.
1177         (do_dbg_cnt): Likewise.
1178         (ccp_finalize): Likewise.
1179         (ccp_lattice_meet): Likewise.
1180         (bit_value_unop_1): Use widest_ints rather than double_ints.
1181         (bit_value_binop_1): Likewise.
1182         (bit_value_unop): Use wide-int interfaces.
1183         (bit_value_binop): Likewise.
1184         (bit_value_assume_aligned): Likewise.
1185         (evaluate_stmt): Likewise.
1186         (ccp_fold_stmt): Likewise.
1187         (visit_cond_stmt): Likewise.
1188         (ccp_visit_stmt): Likewise.
1189         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1190         (constant_pointer_difference): Likewise.
1191         (associate_pointerplus): Likewise.
1192         (combine_conversions): Likewise.
1193         * tree-ssa-loop.h: Include wide-int.h.
1194         (struct tree_niter_desc): Change type of max to widest_int.
1195         * tree-ssa-loop-im.c (mem_refs_may_alias_p): Use wide-int interfaces.
1196         * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Likewise.
1197         (remove_redundant_iv_tests): Likewise.
1198         (canonicalize_loop_induction_variables): Likewise.
1199         * tree-ssa-loop-ivopts.c (alloc_iv): Likewise.
1200         (constant_multiple_of): Take a widest_int pointer instead of
1201         a double_int pointer.
1202         (get_computation_aff): Use wide-int interfaces.
1203         (ptr_difference_cost): Likewise.
1204         (difference_cost): Likewise.
1205         (get_loop_invariant_expr_id): Likewise.
1206         (get_computation_cost_at): Likewise.
1207         (iv_elimination_compare_lt): Likewise.
1208         (may_eliminate_iv): Likewise.
1209         * tree-ssa-loop-niter.h (estimated_loop_iterations): Use widest_int
1210         instead of double_int.
1211         (max_loop_iterations): Likewise.
1212         (max_stmt_executions): Likewise.
1213         (estimated_stmt_executions): Likewise.
1214         * tree-ssa-loop-niter.c: Include wide-int-print.h.
1215         (split_to_var_and_offset): Use wide-int interfaces.
1216         (determine_value_range): Likewise.
1217         (bound_difference_of_offsetted_base): Likewise.
1218         (bounds_add): Take a widest_int instead of a double_int.
1219         (number_of_iterations_ne_max): Use wide-int interfaces.
1220         (number_of_iterations_ne): Likewise.
1221         (number_of_iterations_lt_to_ne): Likewise.
1222         (assert_loop_rolls_lt): Likewise.
1223         (number_of_iterations_lt): Likewise.
1224         (number_of_iterations_le): Likewise.
1225         (number_of_iterations_cond): Likewise.
1226         (number_of_iterations_exit): Likewise.
1227         (finite_loop_p): Likewise.
1228         (derive_constant_upper_bound_assign): Likewise.
1229         (derive_constant_upper_bound): Return a widest_int.
1230         (derive_constant_upper_bound_ops): Likewise.
1231         (do_warn_aggressive_loop_optimizations): Use wide-int interfaces.
1232         (record_estimate): Take a widest_int rather than a double_int.
1233         (record_nonwrapping_iv): Use wide-int interfaces.
1234         (double_int_cmp): Delete.
1235         (wide_int_cmp): New.
1236         (bound_index): Take a widest_int rather than a double_int.
1237         (discover_iteration_bound_by_body_walk): Use wide-int interfaces.
1238         (maybe_lower_iteration_bound): Likewise.
1239         (estimate_numbers_of_iterations_loop): Likewise.
1240         (estimated_loop_iterations): Take a widest_int pointer than than
1241         a double_int pointer.
1242         (estimated_loop_iterations_int): Use wide-int interfaces.
1243         (max_loop_iterations): Take a widest_int pointer than than
1244         a double_int pointer.
1245         (max_loop_iterations_int): Use wide-int interfaces.
1246         (max_stmt_executions): Take a widest_int pointer than than
1247         a double_int pointer.
1248         (estimated_stmt_executions): Likewise.
1249         (n_of_executions_at_most): Use wide-int interfaces.
1250         (scev_probably_wraps_p): Likewise.
1251         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Update calls
1252         to real_to_integer.
1253         * tree-scalar-evolution.c (simplify_peeled_chrec): Use wide-int
1254         interfaces.
1255         * tree-ssanames.c (set_range_info): Use wide_int_refs rather than
1256         double_ints.  Adjust for trailing_wide_ints <3> representation.
1257         (set_nonzero_bits): Likewise.
1258         (get_range_info): Return wide_ints rather than double_ints.
1259         Adjust for trailing_wide_ints <3> representation.
1260         (get_nonzero_bits): Likewise.
1261         (duplicate_ssa_name_range_info): Adjust for trailing_wide_ints <3>
1262         representation.
1263         * tree-ssanames.h (struct range_info_def): Replace min, max and
1264         nonzero_bits with a trailing_wide_ints <3>.
1265         (set_range_info): Use wide_int_refs rather than double_ints.
1266         (set_nonzero_bits): Likewise.
1267         (get_range_info): Return wide_ints rather than double_ints.
1268         (get_nonzero_bits): Likewise.
1269         * tree-ssa-phiopt.c (jump_function_from_stmt): Use wide-int interfaces.
1270         * tree-ssa-pre.c (phi_translate_1): Likewise.
1271         * tree-ssa-reassoc.c (decrement_power): Use calls to real_from_integer.
1272         (acceptable_pow_call): Likewise.
1273         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Use wide-int
1274         interfaces.
1275         (vn_reference_fold_indirect): Likewise.
1276         (vn_reference_maybe_forwprop_address): Likewise.
1277         (valueize_refs_1): Likewise.
1278         * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1279         * tree-ssa-uninit.c (is_value_included_in): Use wide-int interfaces,
1280         tree_int_cst_lt and tree_int_cst_le.
1281         * tree-streamer-in.c (unpack_ts_base_value_fields): Use wide-int
1282         interfaces.
1283         (streamer_alloc_tree): Likewise.
1284         * tree-streamer-out.c (pack_ts_int_cst_value_fields): Likewise.
1285         (streamer_write_tree_header): Likewise.
1286         (streamer_write_integer_cst): Likewise.
1287         * tree-switch-conversion.c (emit_case_bit_tests): Likewise.
1288         (build_constructors): Likewise.
1289         (array_value_type): Likewise.
1290         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Likewise.
1291         (vect_check_gather): Likewise.
1292         * tree-vect-generic.c (build_replicated_const): Likewise.
1293         (expand_vector_divmod): Likewise.
1294         * tree-vect-loop.c (vect_transform_loop): Likewise.
1295         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
1296         (vect_do_peeling_for_alignment): Likewise.
1297         * tree-vect-patterns.c (vect_recog_divmod_pattern): Likewise.
1298         * tree-vrp.c: Include wide-int.h.
1299         (operand_less_p): Use wide-int interfaces and tree_int_cst_lt.
1300         (extract_range_from_assert): Use wide-int interfaces.
1301         (vrp_int_const_binop): Likewise.
1302         (zero_nonzero_bits_from_vr): Take wide_int pointers rather than
1303         double_int pointers.
1304         (ranges_from_anti_range): Use wide-int interfaces.
1305         (quad_int_cmp): Delete.
1306         (quad_int_pair_sort): Likewise.
1307         (extract_range_from_binary_expr_1): Use wide-int interfaces.
1308         (extract_range_from_unary_expr_1): Likewise.
1309         (adjust_range_with_scev): Likewise.
1310         (masked_increment): Take and return wide_ints rather than double_ints.
1311         (register_edge_assert_for_2): Use wide-int interfaces.
1312         (check_array_ref): Likewise.
1313         (search_for_addr_array): Likewise.
1314         (maybe_set_nonzero_bits): Likewise.
1315         (union_ranges): Pass an integer of the correct type instead of
1316         using integer_one_node.
1317         (intersect_ranges): Likewise.
1318         (simplify_truth_ops_using_ranges): Likewise.
1319         (simplify_bit_ops_using_ranges): Use wide-int interfaces.
1320         (range_fits_type_p): Likewise.
1321         (simplify_cond_using_ranges): Likewise.  Take a signop rather than
1322         a bool.
1323         (simplify_conversion_using_ranges): Use wide-int interfaces.
1324         (simplify_float_conversion_using_ranges): Likewise.
1325         (vrp_finalize): Likewise.
1326         * value-prof.c (gimple_divmod_fixed_value_transform): Likewise.
1327         (gimple_stringops_transform): Likewise.
1328         * varasm.c (decode_addr_const): Likewise.
1329         (const_hash_1): Likewise.
1330         (const_rtx_hash_1): Likewise
1331         (output_constant): Likewise.
1332         (array_size_for_constructor): Likewise.
1333         (output_constructor_regular_field): Likewise.
1334         (output_constructor_bitfield): Likewise.
1335         * var-tracking.c (loc_cmp): Handle CONST_WIDE_INT.
1336         * mkconfig.sh: Include machmode.h to pick up BITS_PER_UNIT for
1337         GENERATOR_FILEs.
1338         * gencheck.c: Define BITS_PER_UNIT.
1339         * wide-int.cc: New.
1340         * wide-int.h: New.
1341         * wide-int-print.cc: New.
1342         * wide-int-print.h: New.
1344 2014-05-06  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1346         * config/avr/avr.c (avr_can_eliminate): Mark unused argument.
1348 2014-05-06  Richard Biener  <rguenther@suse.de>
1350         * tree-pass.h (TODO_verify_ssa, TODO_verify_flow,
1351         TODO_verify_stmts, TODO_verify_rtl_sharing): Remove.
1352         (TODO_verify_all): Adjust.
1353         * asan.c: Remove references to TODO_verify_ssa, TODO_verify_flow,
1354         TODO_verify_stmts and TODO_verify_rtl_sharing.
1355         * bb-reorder.c: Likewise.
1356         * cfgexpand.c: Likewise.
1357         * cprop.c: Likewise.
1358         * cse.c: Likewise.
1359         * function.c: Likewise.
1360         * fwprop.c: Likewise.
1361         * gcse.c: Likewise.
1362         * gimple-ssa-isolate-paths.c: Likewise.
1363         * gimple-ssa-strength-reduction.c: Likewise.
1364         * ipa-split.c: Likewise.
1365         * loop-init.c: Likewise.
1366         * loop-unroll.c: Likewise.
1367         * lower-subreg.c: Likewise.
1368         * modulo-sched.c: Likewise.
1369         * postreload-gcse.c: Likewise.
1370         * predict.c: Likewise.
1371         * recog.c: Likewise.
1372         * sched-rgn.c: Likewise.
1373         * store-motion.c: Likewise.
1374         * tracer.c: Likewise.
1375         * trans-mem.c: Likewise.
1376         * tree-call-cdce.c: Likewise.
1377         * tree-cfg.c: Likewise.
1378         * tree-cfgcleanup.c: Likewise.
1379         * tree-complex.c: Likewise.
1380         * tree-eh.c: Likewise.
1381         * tree-emutls.c: Likewise.
1382         * tree-if-conv.c: Likewise.
1383         * tree-into-ssa.c: Likewise.
1384         * tree-loop-distribution.c: Likewise.
1385         * tree-object-size.c: Likewise.
1386         * tree-parloops.c: Likewise.
1387         * tree-pass.h: Likewise.
1388         * tree-sra.c: Likewise.
1389         * tree-ssa-ccp.c: Likewise.
1390         * tree-ssa-copy.c: Likewise.
1391         * tree-ssa-copyrename.c: Likewise.
1392         * tree-ssa-dce.c: Likewise.
1393         * tree-ssa-dom.c: Likewise.
1394         * tree-ssa-dse.c: Likewise.
1395         * tree-ssa-forwprop.c: Likewise.
1396         * tree-ssa-ifcombine.c: Likewise.
1397         * tree-ssa-loop-ch.c: Likewise.
1398         * tree-ssa-loop-ivcanon.c: Likewise.
1399         * tree-ssa-loop.c: Likewise.
1400         * tree-ssa-math-opts.c: Likewise.
1401         * tree-ssa-phiopt.c: Likewise.
1402         * tree-ssa-phiprop.c: Likewise.
1403         * tree-ssa-pre.c: Likewise.
1404         * tree-ssa-reassoc.c: Likewise.
1405         * tree-ssa-sink.c: Likewise.
1406         * tree-ssa-strlen.c: Likewise.
1407         * tree-ssa-tail-merge.c: Likewise.
1408         * tree-ssa-uncprop.c: Likewise.
1409         * tree-switch-conversion.c: Likewise.
1410         * tree-tailcall.c: Likewise.
1411         * tree-vect-generic.c: Likewise.
1412         * tree-vectorizer.c: Likewise.
1413         * tree-vrp.c: Likewise.
1414         * tsan.c: Likewise.
1415         * var-tracking.c: Likewise.
1416         * bt-load.c: Likewise.
1417         * cfgcleanup.c: Likewise.
1418         * combine-stack-adj.c: Likewise.
1419         * combine.c: Likewise.
1420         * compare-elim.c: Likewise.
1421         * config/epiphany/resolve-sw-modes.c: Likewise.
1422         * config/i386/i386.c: Likewise.
1423         * config/mips/mips.c: Likewise.
1424         * config/s390/s390.c: Likewise.
1425         * config/sh/sh_treg_combine.cc: Likewise.
1426         * config/sparc/sparc.c: Likewise.
1427         * dce.c: Likewise.
1428         * dse.c: Likewise.
1429         * final.c: Likewise.
1430         * ifcvt.c: Likewise.
1431         * mode-switching.c: Likewise.
1432         * passes.c: Likewise.
1433         * postreload.c: Likewise.
1434         * ree.c: Likewise.
1435         * reg-stack.c: Likewise.
1436         * regcprop.c: Likewise.
1437         * regrename.c: Likewise.
1438         * web.c: Likewise.
1440 2014-05-06  Richard Biener  <rguenther@suse.de>
1442         PR middle-end/61070
1443         * bitmap.c (debug_bitmap): Dump to stderr, not stdout.
1444         * tree-ssa-structalias.c (dump_solution_for_var): Likewise.
1446 2014-05-05  Jan Hubicka  <hubicka@ucw.cz>
1448         PR ipa/60965
1449         * ipa-devirt.c (get_class_context): Allow POD to change to non-POD.
1451 2014-05-05  Radovan Obradovic  <robradovic@mips.com>
1452             Tom de Vries  <tom@codesourcery.com>
1454         * target.def (call_fusage_contains_non_callee_clobbers): New
1455         DEFHOOKPOD.
1456         * doc/tm.texi.in (@node Stack and Calling): Add Miscellaneous Register
1457         Hooks to @menu.
1458         (@node Miscellaneous Register Hooks): New node.
1459         (@hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): New hook.
1460         * doc/tm.texi: Regenerate.
1462 2014-05-05  Marek Polacek  <polacek@redhat.com>
1464         PR driver/61065
1465         * opts.c (common_handle_option): Call error_at instead of warning_at.
1467 2014-05-05  Richard Biener  <rguenther@suse.de>
1469         * passes.c (execute_function_todo): Don't reset TODO_verify_ssa
1470         from last_verified if update_ssa ran.  Move TODO_verify_rtl_sharing
1471         under the TODO_verify_il umbrella.
1473 2014-05-05  Richard Biener  <rguenther@suse.de>
1475         * passes.c (execute_function_todo): Move TODO_verify_flow under
1476         the TODO_verify_ul umbrella.
1478 2014-05-05  Richard Biener  <rguenther@suse.de>
1480         PR middle-end/61010
1481         * fold-const.c (fold_binary_loc): Consistently avoid canonicalizing
1482         X & CST away from a CST that is the mask of a mode.
1484 2014-05-05  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1486         * config/picochip/picochip-protos.h (picochip_regno_nregs): Change
1487         int argument to enum machine_mode.
1488         (picochip_class_max_nregs): Ditto.
1489         * config/picochip/picochip.c (picochip_regno_nregs): Ditto.
1490         (picochip_class_max_nregs): Ditto.
1492 2014-05-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1494         * target.def: Add new target hook.
1495         * doc/tm.texi: Regenerate.
1496         * targhooks.h (default_keep_leaf_when_profiled): Add prototype.
1497         * targhooks.c (default_keep_leaf_when_profiled): New function.
1499         * config/s390/s390.c (s390_keep_leaf_when_profiled): New function.
1500         (TARGET_KEEP_LEAF_WHEN_PROFILED): Define.
1502 2014-05-05  Bin Cheng  <bin.cheng@arm.com>
1504         PR tree-optimization/60363
1505         * gcc/tree-ssa-threadupdate.c (get_value_locus_in_path): New.
1506         (copy_phi_args): New parameters.  Call get_value_locus_in_path.
1507         (update_destination_phis): New parameter.
1508         (create_edge_and_update_destination_phis): Ditto.
1509         (ssa_fix_duplicate_block_edges): Pass new arguments.
1510         (thread_single_edge): Ditto.
1512 2014-05-04  Peter Bergner  <bergner@vnet.ibm.com>
1514         * config/rs6000/rs6000.h (RS6000_BTM_HARD_FLOAT): New define.
1515         (RS6000_BTM_COMMON): Add RS6000_BTM_HARD_FLOAT.
1516         (TARGET_EXTRA_BUILTINS): Add TARGET_HARD_FLOAT.
1517         * config/rs6000/rs6000-builtin.def (BU_MISC_1):
1518         Use RS6000_BTM_HARD_FLOAT.
1519         (BU_MISC_2): Likewise.
1520         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
1521         RS6000_BTM_HARD_FLOAT.
1522         (rs6000_option_override_internal): Enforce -mhard-float if -mhard-dfp
1523         is explicitly used.
1524         (rs6000_invalid_builtin): Add hard floating builtin support.
1525         (rs6000_expand_builtin): Relax the gcc_assert to allow the new
1526         hard float builtins.
1527         (rs6000_builtin_mask_names): Add RS6000_BTM_HARD_FLOAT.
1529 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1531         * config/sh/sh_optimize_sett_clrt.cc (sh_optimize_sett_clrt::execute):
1532         Add missing function* argument.
1534 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1536         * lra-constraints.c (valid_address_p): Move earlier in file.
1537         Add a constraint argument to the address_info version.
1538         (satisfies_memory_constraint_p): New function.
1539         (satisfies_address_constraint_p): Likewise.
1540         (process_alt_operands, curr_insn_transform): Use them.
1541         (process_address): Pass the constraint to valid_address_p when
1542         checking address operands.
1544 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
1546         * config/mips/mips.c (mips_isa_rev): New variable.
1547         (mips_set_architecture): Set it.
1548         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
1549         from mips_isa_rev.
1550         (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
1551         (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
1552         (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
1553         (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
1554         (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
1555         conditions in terms of mips_isa_rev.
1556         (mips_isa_rev): Declare.
1558 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1560         * config/sh/sh-mem.cc: Use tabs instead of spaces.
1561         (prob_unlikely, prob_likely): Make variables const.
1563 2014-05-03  Denis Chertykov  <chertykov@gmail.com>
1565         * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
1567 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1569         * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
1571 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1573         * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
1574         (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
1575         * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
1576         functions.
1577         (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
1578         sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
1579         sh_pass_in_reg_p.
1580         Replace usage of ROUND_REG with sh_round_reg.
1581         Use CEIL instead of ROUND_ADVANCE.
1583 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
1585         PR target/61026
1586         * config/sh/sh.c: Include stdlib headers before everything else.
1588 2014-05-02  Jakub Jelinek  <jakub@redhat.com>
1590         * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
1591         GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
1592         (gimplify_adjust_omp_clauses): Simd region is never
1593         directly nested in combined parallel.  Instead, for linear
1594         with copyin/copyout, if in combined for simd loop, make decl
1595         firstprivate/lastprivate on OMP_FOR.
1596         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
1597         expand_omp_for_static_chunk): When setting endvar, also set
1598         fd->loop.v to the same value.
1600 2014-05-02  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1602         * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
1604 2014-05-02  Alan Lawrence  <alan.lawrence@arm.com>
1606         * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
1607         expression.
1609 2014-05-02  Marek Polacek  <polacek@redhat.com>
1611         * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
1613 2014-05-02  Kito Cheng  <kito@0xlab.org>
1615         * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
1616         to a C expression marco.
1617         * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
1618         * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
1619         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
1620         * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
1621         HONOR_REG_ALLOC_ORDER.
1622         * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
1624 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1626         * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
1628 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1630         * config/arc/arc.c (arc_select_cc_mode): Fix typo.
1632 2014-05-01  Yuri Rumyantsev  <ysrumyan@gmail.com>
1634         * tree-if-conv.c (is_cond_scalar_reduction): New function.
1635         (convert_scalar_cond_reduction): Likewise.
1636         (predicate_scalar_phi): Add recognition and transformation
1637         of simple conditioanl reduction to be vectorizable.
1639 2014-05-01  Marek Polacek  <polacek@redhat.com>
1641         PR c/43245
1642         * doc/invoke.texi: Document -Wdiscarded-qualifiers.
1644 2014-04-30  Alan Lawrence  <alan.lawrence@arm.com>
1646         * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
1647         vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
1648         vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
1649         vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
1650         vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
1651         vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
1652         vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
1653         vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
1655 2014-04-30  Joern Rennecke  <joern.rennecke@embecosm.com>
1657         * config/arc/arc.opt (mlra): Move comment above option name
1658         to avoid mis-parsing as language options.
1660 2014-04-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1662         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
1663         * config/sol2.h: ... here.
1664         * config/sol2-10.h: Remove.
1666         * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
1667         (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
1668         (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
1669         (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
1670         (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
1671         * config/sol2.h: ... here.
1672         (SECTION_NAME_FORMAT): Don't redefine.
1673         (STARTFILE_ARCH32_SPEC): Rename to ...
1674         (STARTFILE_ARCH_SPEC): ... this.
1675         (ASM_OUTPUT_ALIGNED_COMMON): Move ...
1676         * config/sparc/sol2.h: ... here.
1677         (SECTION_NAME_FORMAT): Don't undef.
1678         * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
1679         (SUBTARGET_EXTRA_SPECS): Remove.
1680         * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
1682         * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
1683         (MD_STARTFILE_PREFIX): Remove.
1684         (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
1685         (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
1686         (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
1687         (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
1688         (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
1689         (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
1690         * config/i386/sol2.h: ... here.
1691         (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
1692         * config/i386/sol2-bi.h: Remove.
1693         * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
1694         (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
1696         * config/i386/t-sol2-64: Rename to ...
1697         * config/i386/t-sol2: ... this.
1698         * config/sparc/t-sol2-64: Rename to ...
1699         * config/sparc/t-sol2: ... this.
1701         * config.gcc (*-*-solaris2*): Split sol2_tm_file into
1702         sol2_tm_file_head, sol2_tm_file_tail.
1703         Include ${cpu_type}/sol2.h before sol2.h.
1704         Remove sol2-10.h.
1705         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
1706         i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
1707         Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
1708         Reflect i386/t-sol2-64 renaming.
1709         (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
1710         Reflect sparc/t-sol2-64 renaming.
1712 2014-04-30  Richard Biener  <rguenther@suse.de>
1714         * passes.c (execute_function_todo): Move TODO_verify_stmts
1715         and TODO_verify_ssa under the TODO_verify_il umbrella.
1716         * tree-ssa.h (verify_ssa): Adjust prototype.
1717         * tree-ssa.c (verify_ssa): Add parameter to tell whether
1718         we should verify SSA operands.
1719         * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
1720         * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
1721         whether we should verify whether not throwing stmts have EH info.
1722         * graphite-scop-detection.c (create_sese_edges): Adjust.
1723         * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
1724         * tree-eh.c (lower_try_finally_switch): Do not add the
1725         default case label twice.
1727 2014-04-30  Marek Polacek  <polacek@redhat.com>
1729         * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
1730         * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
1731         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
1732         * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
1734 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
1736         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
1737         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
1738         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
1739         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
1740         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
1741         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
1742         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
1743         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
1745 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
1747         * tree-cfg.c (dump_function_to_file): Dump the return type of
1748         functions, in a line to itself before the function body, mimicking
1749         the layout of a C function.
1751 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
1753         PR tree-optimization/60971
1754         * tree-tailcall.c (process_assignment): Reject conversions which
1755         reduce precision.
1757 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
1759         * calls.c (initialize_argument_information): Always treat
1760         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
1761         (expand_call): Likewise.
1762         (emit_library_call_calue_1): Likewise.
1763         * expr.c (PUSH_ARGS_REVERSED): Do not define.
1764         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
1765         code accordingly.
1767 2014-04-29  Nick Clifton  <nickc@redhat.com>
1769         * config/msp430/msp430.md (umulsidi): Fix typo.
1770         (mulhisi3): Enable even inside interrupt handlers.
1771         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
1772         bigger return address pushed in large mode.
1774 2014-04-29  Nick Clifton  <nickc@redhat.com>
1776         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
1777         (arc_init_reg_tables): Use a machine_mode enum to iterate over
1778         available modes.
1779         * config/m32r/m32r.c (init_reg_tables): Likewise.
1780         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
1781         enum to hold the modes.
1783 2014-04-29  Richard Biener  <rguenther@suse.de>
1785         * dominance.c (free_dominance_info): Add overload with
1786         function parameter.
1787         (dom_info_state): Likewise.
1788         (dom_info_available_p): Likewise.
1789         * basic-block.h (free_dominance_info, dom_info_state,
1790         dom_info_available_p): Declare overloads.
1791         * passes.c (execute_function_todo): Verify that verifiers
1792         don't change dominator info state.  Drop dominator info
1793         for IPA pass invocations.
1794         * cgraph.c (release_function_body): Restore asserts that
1795         dominator information is released.
1797 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
1799         * doc/invoke.texi: Fix typo.
1800         * tree-vrp.c: Fix typos.
1801         * gimple.c (infer_nonnull_range): Reorder operands of an && condition.
1803 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1805         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
1807 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
1809         * config/aarch64/aarch64-builtins.c
1810         (aarch64_types_storestruct_lane_qualifiers): New.
1811         (TYPES_STORESTRUCT_LANE): Likewise.
1812         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
1813         (st3_lane): Likewise.
1814         (st4_lane): Likewise.
1815         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
1816         (vec_store_lanesci_lane<mode>): Likewise.
1817         (vec_store_lanesxi_lane<mode>): Likewise.
1818         (aarch64_st2_lane<VQ:mode>): Likewise.
1819         (aarch64_st3_lane<VQ:mode>): Likewise.
1820         (aarch64_st4_lane<VQ:mode>): Likewise.
1821         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
1822         * config/aarch64/arm_neon.h
1823         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
1824         use new macro arguments.
1825         (__ST3_LANE_FUNC): Likewise.
1826         (__ST4_LANE_FUNC): Likewise.
1827         * config/aarch64/iterators.md (V_TWO_ELEM): New.
1828         (V_THREE_ELEM): Likewise.
1829         (V_FOUR_ELEM): Likewise.
1831 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
1833         * doc/gimple.texi: Replace the description of the now-defunct
1834         union gimple_statement_d with a diagram showing the
1835         gimple_statement_base class hierarchy and its relationships to
1836         the GSS_ and GIMPLE_ enums.
1838 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
1840         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
1841         * config/aarch64/aarch64.c
1842         (aarch64_cannot_change_mode_class): Weaken conditions.
1843         (aarch64_modes_tieable_p): New.
1844         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
1846 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
1848         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1849         (loadsync_<mode>): Change mode.
1850         (load_quadpti, store_quadpti): New.
1851         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1852         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1854 2014-04-28  Martin Jambor  <mjambor@suse.cz>
1856         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
1857         same alias type as the original statement.
1858         (subreplacement_assignment_data): New type.
1859         (handle_unscalarized_data_in_subtree): New type of parameter,
1860         generate new memory accesses with same alias type as the original
1861         statement.
1862         (load_assign_lhs_subreplacements): Likewise.
1863         (sra_modify_constructor_assign): Generate new memory accesses with
1864         same alias type as the original statement.
1866 2014-04-28  Richard Biener  <rguenther@suse.de>
1868         * tree-pass.h (TODO_verify_il): Define.
1869         (TODO_verify_all): Complete properly.
1870         * passes.c (execute_function_todo): Move existing loop-closed
1871         SSA verification under TODO_verify_il.
1872         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
1873         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
1874         Fix tree sharing issue.
1876 2014-04-28  Richard Biener  <rguenther@suse.de>
1878         PR middle-end/60092
1879         * builtins.def (DEF_C11_BUILTIN): Add.
1880         (BUILT_IN_ALIGNED_ALLOC): Likewise.
1881         * coretypes.h (enum function_class): Add function_c11_misc.
1882         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
1883         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
1884         (call_may_clobber_ref_p_1): Likewise.
1885         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1886         (mark_all_reaching_defs_necessary_1): Likewise.
1887         (propagate_necessity): Likewise.
1888         (eliminate_unnecessary_stmts): Likewise.
1889         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
1891 2014-04-28  Richard Biener  <rguenther@suse.de>
1893         * tree-vrp.c (vrp_var_may_overflow): Remove.
1894         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
1895         with overflow immediately bump to one before that value and
1896         let iteration figure out overflow status.
1898 2014-04-28  Richard Biener  <rguenther@suse.de>
1900         * configure.ac: Do valgrind header checks unconditionally.
1901         Add --enable-valgrind-annotations.
1902         * system.h: Guard valgrind header inclusion with
1903         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
1904         * alloc-pool.c (pool_alloc, pool_free): Use
1905         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
1906         to guard possibly dead code.
1907         * config.in: Regenerated.
1908         * configure: Likewise.
1910 2014-04-28  Jeff Law  <law@redhat.com>
1912         PR tree-optimization/60902
1913         * tree-ssa-threadedge.c
1914         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
1915         over real defs when invalidating outputs from statements that do not
1916         produce useful outputs for threading.
1918 2014-04-28  Richard Biener  <rguenther@suse.de>
1920         PR tree-optimization/60979
1921         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
1922         SCOPs that end in a block with a successor with abnormal
1923         predecessors.
1925 2014-04-28  Richard Biener  <rguenther@suse.de>
1927         * tree-pass.h (execute_pass_list): Adjust prototype.
1928         * passes.c (pass_manager::execute_early_local_passes): Adjust.
1929         (do_per_function): Change callback signature, push all actual
1930         work to the callbals.
1931         (do_per_function_toporder): Likewise.
1932         (execute_function_dump): Adjust.
1933         (execute_function_todo): Likewise.
1934         (clear_last_verified): Likewise.
1935         (verify_curr_properties): Likewise.
1936         (update_properties_after_pass): Likewise.
1937         (execute_pass_list_1): Split out from ...
1938         (execute_pass_list): ... here.  Adjust.
1939         (execute_ipa_pass_list): Likewise.
1940         * cgraphunit.c (cgraph_add_new_function): Adjust.
1941         (analyze_function): Likewise.
1942         (expand_function): Likewise.
1943         * cgraph.c (release_function_body): Free dominance info
1944         here instead of asserting it was magically freed elsewhere.
1946 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
1948         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
1949         * configure: Regenerate.
1950         * config/sparc/sparc.opt (muser-mode): New option.
1951         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
1952         for LEON3.
1953         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
1954         * doc/invoke.texi (SPARC options): Document -muser-mode.
1956 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
1958         * cselib.c (find_slot_memmode): Delete.
1959         (cselib_hasher): Change compare_type to a struct.
1960         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
1961         constants.
1962         (preserve_constants_and_equivs): Adjust for new compare_type.
1963         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
1964         (wrap_constant): Delete.
1965         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
1967 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
1969         * doc/install.texi (Building with profile feedback): Remove
1970         outdated sentence.
1972 2014-04-26  Tom de Vries  <tom@codesourcery.com>
1974         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
1975         array accesses.
1977 2014-04-25  Cary Coutant  <ccoutant@google.com>
1979         PR debug/60929
1980         * dwarf2out.c (should_move_die_to_comdat): A type definition
1981         can contain a subprogram definition, but don't move it to a
1982         comdat unit.
1983         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
1984         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
1985         from original DIE.
1986         (clone_tree_hash): Rename to...
1987         (clone_tree_partial): ...this; change callers.  Copy
1988         DW_TAG_subprogram DIEs as declarations.
1989         (copy_decls_walk): Don't copy children of a declaration into a
1990         type unit.
1992 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
1994         PR target/60969
1995         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
1996         alternative 12.
1998 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
2000         * config/arm/predicates.md (call_insn_operand): Add long_call check.
2001         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
2002         reg for long_call.
2003         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
2004         restriction.
2006 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2008         * config/arm/arm.c (arm_cortex_a8_tune): Initialise T16-related fields.
2010 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2012         PR tree-optimization/60930
2013         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
2014         creating a multiply candidate by folding two constant
2015         multiplicands when the result overflows.
2017 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
2019         PR tree-optimization/60960
2020         * tree-vect-generic.c (expand_vector_operation): Only call
2021         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
2023 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2025         * expr.c (clobber_reg_mode): New function.
2026         * expr.h (clobber_reg): New function.
2028 2014-04-25  Tom de Vries  <tom@codesourcery.com>
2030         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
2031         clobbers.
2033 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
2034             Tom de Vries  <tom@codesourcery.com>
2036         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
2037         handle.
2038         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
2039         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
2040         new argument to find_all_hard_reg_sets call.
2042 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2044         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
2045         Use HOST_WIDE_INT_C for mask literal.
2046         (aarch_rev16_shleft_mask_imm_p): Likewise.
2048 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
2050         PR target/60941
2051         * config/sparc/sparc.md (ashlsi3_extend): Delete.
2053 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
2055         PR preprocessor/56540
2056         * config/i386/i386-c.c (ix86_target_macros): Define
2057         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
2059 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2061         * configure.ac (tga_func): Remove.
2062         (LIB_TLS_SPEC): Remove.
2063         * configure: Regenerate.
2064         * config.in: Regenerate.
2065         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
2067 2014-04-25  Richard Biener  <rguenther@suse.de>
2069         PR ipa/60912
2070         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
2071         call stmt use/clobber sets during stmt walk instead of
2072         walking the possibly incomplete set of caller edges.
2074 2014-04-25  Richard Biener  <rguenther@suse.de>
2076         PR ipa/60911
2077         * passes.c (apply_ipa_transforms): Inline into only caller ...
2078         (execute_one_pass): ... here.  Properly bring in function
2079         bodies for nodes we want to apply IPA transforms to.
2081 2014-04-24  Cong Hou  <congh@google.com>
2083         PR tree-optimization/60896
2084         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
2085         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
2086         (vect_mark_pattern_stmts): Set the def type of all statements in
2087         PATTERN_DEF_SEQ as vect_internal_def.
2089 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
2091         * doc/extend.texi (PowerPC Built-in Functions): Document new
2092         powerpc extended divide, bcd, pack/unpack 128-bit, builtin functions.
2093         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
2095         * config/rs6000/predicates.md (const_0_to_3_operand): New
2096         predicate to match 0..3 integer constants.
2098         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
2099         to support adding miscellaneous builtin functions.
2100         (BU_DFP_MISC_2): Likewise.
2101         (BU_P7_MISC_1): Likewise.
2102         (BU_P7_MISC_2): Likewise.
2103         (BU_P8V_MISC_3): Likewise.
2104         (BU_MISC_1): Likewise.
2105         (BU_MISC_2): Likewise.
2106         (DIVWE): Add extended divide builtin functions.
2107         (DIVWEO): Likewise.
2108         (DIVWEU): Likewise.
2109         (DIVWEUO): Likewise.
2110         (DIVDE): Likewise.
2111         (DIVDEO): Likewise.
2112         (DIVDEU): Likewise.
2113         (DIVDEUO): Likewise.
2114         (DXEX): Add decimal floating-point builtin functions.
2115         (DXEXQ): Likewise.
2116         (DDEDPD): Likewise.
2117         (DDEDPDQ): Likewise.
2118         (DENBCD): Likewise.
2119         (DENBCDQ): Likewise.
2120         (DIEX): Likewise.
2121         (DIEXQ): Likewise.
2122         (DSCLI): Likewise.
2123         (DSCLIQ): Likewise.
2124         (DSCRI): Likewise.
2125         (DSCRIQ): Likewise.
2126         (CDTBCD): Add new BCD builtin functions.
2127         (CBCDTD): Likewise.
2128         (ADDG6S): Likewise.
2129         (BCDADD): Likewise.
2130         (BCDADD_LT): Likewise.
2131         (BCDADD_EQ): Likewise.
2132         (BCDADD_GT): Likewise.
2133         (BCDADD_OV): Likewise.
2134         (BCDSUB): Likewise.
2135         (BCDSUB_LT): Likewise.
2136         (BCDSUB_EQ): Likewise.
2137         (BCDSUB_GT): Likewise.
2138         (BCDSUB_OV): Likewise.
2139         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
2140         (UNPACK_TD): Likewise.
2141         (PACK_TF): Likewise.
2142         (UNPACK_TF): Likewise.
2143         (UNPACK_TF_0): Likewise.
2144         (UNPACK_TF_1): Likewise.
2145         (PACK_V1TI): Likewise.
2146         (UNPACK_V1TI): Likewise.
2148         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2149         support for decimal floating point builtin functions.
2150         (rs6000_expand_ternop_builtin): Add checks for the new builtin
2151         functions that take constant arguments.
2152         (rs6000_invalid_builtin): Add decimal floating point builtin support.
2153         (rs6000_init_builtins): Setup long double, _Decimal64, and
2154         _Decimal128 types for new builtin functions.
2155         (builtin_function_type): Set the unsigned flags appropriately for
2156         the new builtin functions.
2157         (rs6000_opt_masks): Add support for decimal floating point builtin
2158         functions.
2160         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
2161         floating point builtin functions.
2162         (RS6000_BTM_COMMON): Likewise.
2163         (RS6000_BTI_long_double): Likewise.
2164         (RS6000_BTI_dfloat64): Likewise.
2165         (RS6000_BTI_dfloat128): Likewise.
2166         (long_double_type_internal_node): Likewise.
2167         (dfloat64_type_internal_node): Likewise.
2168         (dfloat128_type_internal_node): Likewise.
2170         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
2171         2.07 bcd arithmetic instructions.
2172         (UNSPEC_BCDSUB): Likewise.
2173         (UNSPEC_BCD_OVERFLOW): Likewise.
2174         (UNSPEC_BCD_ADD_SUB): Likewise.
2175         (bcd_add_sub): Likewise.
2176         (BCD_TEST): Likewise.
2177         (bcd<bcd_add_sub>): Likewise.
2178         (bcd<bcd_add_sub>_test): Likewise.
2179         (bcd<bcd_add_sub>_test2): Likewise.
2180         (bcd<bcd_add_sub>_<code>): Likewise.
2181         (peephole2 for combined bcd ops): Likewise.
2183         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
2184         decimal floating point builtin functions.
2185         (UNSPEC_DENBCD): Likewise.
2186         (UNSPEC_DXEX): Likewise.
2187         (UNSPEC_DIEX): Likewise.
2188         (UNSPEC_DSCLI): Likewise.
2189         (UNSPEC_DSCRI): Likewise.
2190         (D64_D128): Likewise.
2191         (dfp_suffix): Likewise.
2192         (dfp_ddedpd_<mode>): Likewise.
2193         (dfp_denbcd_<mode>): Likewise.
2194         (dfp_dxex_<mode>): Likewise.
2195         (dfp_diex_<mode>): Likewise.
2196         (dfp_dscli_<mode>): Likewise.
2197         (dfp_dscri_<mode>): Likewise.
2199         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
2200         builtin functions.
2201         (UNSPEC_CDTBCD): Likewise.
2202         (UNSPEC_CBCDTD): Likewise.
2203         (UNSPEC_DIVE): Add support for new extended divide builtin functions.
2204         (UNSPEC_DIVEO): Likewise.
2205         (UNSPEC_DIVEU): Likewise.
2206         (UNSPEC_DIVEUO): Likewise.
2207         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
2208         pack/unpack 128-bit types.
2209         (UNSPEC_PACK_128BIT): Likewise.
2210         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
2211         (udiv<mode>3): Use idiv_ldiv mode attribute.
2212         (div<mode>3): Likewise.
2213         (addg6s): Add new BCD builtin functions.
2214         (cdtbcd): Likewise.
2215         (cbcdtd): Likewise.
2216         (UNSPEC_DIV_EXTEND): Add support for new extended divide instructions.
2217         (div_extend): Likewise.
2218         (div<div_extend>_<mode>"): Likewise.
2219         (FP128_64): Add support for new builtin functions to pack/unpack
2220         128-bit types.
2221         (unpack<mode>): Likewise.
2222         (unpacktf_0): Likewise.
2223         (unpacktf_1): Likewise.
2224         (unpack<mode>_dm): Likewise.
2225         (unpack<mode>_nodm): Likewise.
2226         (pack<mode>): Likewise.
2227         (unpackv1ti): Likewise.
2228         (packv1ti): Likewise.
2230 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
2232         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
2233         is disabled.
2235 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
2237         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
2238         * gimplify.c (omp_is_private): Change last argument's type to int.
2239         Only diagnose lastprivate if the simd argument is 1, only diagnose
2240         linear if the simd argument is 2.
2241         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
2242         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
2243         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
2244         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
2245         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
2246         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2247         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
2248         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
2249         * tree-nested.c (convert_nonlocal_omp_clauses,
2250         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
2252 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
2254         PR target/60822
2255         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
2256         operand 1.
2258 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
2260         * flag-types.h (enum ivar_visibility): Add.
2262 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
2264         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
2265         function * argument.
2267 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
2269         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
2271 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
2272             Tom de Vries  <tom@codesourcery.com>
2274         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
2275         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
2276         reg-note.
2277         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
2278         * emit-rtl.c (try_split): Same.
2280 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
2281             Tom de Vries  <tom@codesourcery.com>
2283         * common.opt (fuse-caller-save): New option.
2285 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
2287         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of
2288         elements for big-endian.
2290 2014-04-24  Richard Biener  <rguenther@suse.de>
2292         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
2293         during TER and instead use the sepops interface for expanding
2294         non-GIMPLE_SINGLE_RHS.
2296 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2298         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
2299         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
2301 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2303         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
2304         assembler 64-bit option.
2305         * configure: Regenerate.
2307 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2309         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
2310         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
2311         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
2312         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
2313         (TARGET_CRYPTO): Take TARGET_SIMD into account.
2315 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2317         * config/aarch64/aarch64-builtins.c
2318         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
2319         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
2320         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
2321         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
2322         builtins.
2323         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
2324         (Vrevsuff): New mode attribute.
2326 2014-04-24  Terry Guo  <terry.guo@arm.com>
2328         * config/arm/arm.h (machine_function): Define variable
2329         after_arm_reorg here.
2330         * config/arm/arm.c (after_arm_reorg): Remove the definition.
2331         (arm_split_constant): Update the way to access variable
2332         after_arm_reorg.
2333         (arm_reorg): Ditto.
2334         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
2336 2014-04-23  Tom de Vries  <tom@codesourcery.com>
2338         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
2340 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
2342         * is-a.h: Update comments to reflect the following changes to the
2343         "pointerness" of the API, making the template parameter match the
2344         return type, allowing use of is-a.h with typedefs of pointers.
2345         (is_a_helper::cast): Return a T rather then a pointer to a T, so
2346         that the return type matches the parameter to the is_a_helper.
2347         (as_a): Likewise.
2348         (dyn_cast): Likewise.
2350         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
2351         pointer from the is-a.h API.
2353         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
2354         (is_a_helper <cgraph_node *>::test): ...this, matching change to
2355         is-a.h API.
2356         (is_a_helper <varpool_node>::test): Likewise, convert to...
2357         (is_a_helper <varpool_node *>::test): ...this.
2359         (varpool_first_variable): Update for removal of implicit pointer
2360         from the is-a.h API.
2361         (varpool_next_variable): Likewise.
2362         (varpool_first_static_initializer): Likewise.
2363         (varpool_next_static_initializer): Likewise.
2364         (varpool_first_defined_variable): Likewise.
2365         (varpool_next_defined_variable): Likewise.
2366         (cgraph_first_defined_function): Likewise.
2367         (cgraph_next_defined_function): Likewise.
2368         (cgraph_first_function): Likewise.
2369         (cgraph_next_function): Likewise.
2370         (cgraph_first_function_with_gimple_body): Likewise.
2371         (cgraph_next_function_with_gimple_body): Likewise.
2372         (cgraph_alias_target): Likewise.
2373         (varpool_alias_target): Likewise.
2374         (cgraph_function_or_thunk_node): Likewise.
2375         (varpool_variable_node): Likewise.
2376         (symtab_real_symbol_p): Likewise.
2377         * cgraphunit.c (referred_to_p): Likewise.
2378         (analyze_functions): Likewise.
2379         (handle_alias_pairs): Likewise.
2380         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
2381         * gimple-ssa.h (gimple_vuse_op): Likewise.
2382         (gimple_vdef_op): Likewise.
2383         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
2384         * gimple.c (gimple_build_asm_1): Likewise.
2385         (gimple_build_try): Likewise.
2386         (gimple_build_resx): Likewise.
2387         (gimple_build_eh_dispatch): Likewise.
2388         (gimple_build_omp_for): Likewise.
2389         (gimple_omp_for_set_clauses): Likewise.
2391         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
2392         (is_a_helper <gimple_statement_asm *>::test): ...this.
2393         (is_a_helper <gimple_statement_bind>::test): Convert to...
2394         (is_a_helper <gimple_statement_bind *>::test): ...this.
2395         (is_a_helper <gimple_statement_call>::test): Convert to...
2396         (is_a_helper <gimple_statement_call *>::test): ...this.
2397         (is_a_helper <gimple_statement_catch>::test): Convert to...
2398         (is_a_helper <gimple_statement_catch *>::test): ...this.
2399         (is_a_helper <gimple_statement_resx>::test): Convert to...
2400         (is_a_helper <gimple_statement_resx *>::test): ...this.
2401         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
2402         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
2403         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
2404         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
2405         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
2406         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
2407         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
2408         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
2409         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
2410         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
2411         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
2412         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
2413         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
2414         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
2415         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
2416         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
2417         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
2418         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
2419         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
2420         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
2421         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
2422         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
2423         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
2424         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
2425         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
2426         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
2427         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
2428         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
2429         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
2430         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
2431         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
2432         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
2433         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
2434         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
2435         (is_a_helper <gimple_statement_phi>::test): Convert to...
2436         (is_a_helper <gimple_statement_phi *>::test): ...this.
2437         (is_a_helper <gimple_statement_transaction>::test): Convert to...
2438         (is_a_helper <gimple_statement_transaction *>::test): ...this.
2439         (is_a_helper <gimple_statement_try>::test): Convert to...
2440         (is_a_helper <gimple_statement_try *>::test): ...this.
2441         (is_a_helper <gimple_statement_wce>::test): Convert to...
2442         (is_a_helper <gimple_statement_wce *>::test): ...this.
2443         (is_a_helper <const gimple_statement_asm>::test): Convert to...
2444         (is_a_helper <const gimple_statement_asm *>::test): ...this.
2445         (is_a_helper <const gimple_statement_bind>::test): Convert to...
2446         (is_a_helper <const gimple_statement_bind *>::test): ...this.
2447         (is_a_helper <const gimple_statement_call>::test): Convert to...
2448         (is_a_helper <const gimple_statement_call *>::test): ...this.
2449         (is_a_helper <const gimple_statement_catch>::test): Convert to...
2450         (is_a_helper <const gimple_statement_catch *>::test): ...this.
2451         (is_a_helper <const gimple_statement_resx>::test): Convert to...
2452         (is_a_helper <const gimple_statement_resx *>::test): ...this.
2453         (is_a_helper <const gimple_statement_eh_dispatch>::test): Convert to...
2454         (is_a_helper <const gimple_statement_eh_dispatch *>::test): ...this.
2455         (is_a_helper <const gimple_statement_eh_filter>::test): Convert to...
2456         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
2457         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
2458         Convert to...
2459         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
2460         ...this.
2461         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
2462         Convert to...
2463         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
2464         ...this.
2465         (is_a_helper <const gimple_statement_omp_return>::test): Convert to...
2466         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
2467         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
2468         to...
2469         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
2470         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
2471         to...
2472         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
2473         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
2474         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
2475         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert to...
2476         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
2477         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
2478         to...
2479         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
2480         (is_a_helper <const gimple_statement_omp_target>::test): Convert to...
2481         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
2482         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
2483         to...
2484         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
2485         (is_a_helper <const gimple_statement_omp_single>::test): Convert to...
2486         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
2487         (is_a_helper <const gimple_statement_omp_teams>::test): Convert to...
2488         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
2489         (is_a_helper <const gimple_statement_omp_task>::test): Convert to...
2490         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
2491         (is_a_helper <const gimple_statement_phi>::test): Convert to...
2492         (is_a_helper <const gimple_statement_phi *>::test): ...this.
2493         (is_a_helper <const gimple_statement_transaction>::test): Convert to...
2494         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
2495         (is_a_helper <const gimple_statement_with_ops>::test): Convert to...
2496         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
2497         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
2498         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
2499         (is_a_helper <const gimple_statement_with_memory_ops>::test): Convert
2500         to...
2501         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
2502         ...this.
2503         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert to...
2504         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
2506         (gimple_use_ops): Update for removal of implicit pointer from the
2507         is-a.h API.
2508         (gimple_set_use_ops): Likewise.
2509         (gimple_vuse): Likewise.
2510         (gimple_vdef): Likewise.
2511         (gimple_vuse_ptr): Likewise.
2512         (gimple_vdef_ptr): Likewise.
2513         (gimple_set_vuse): Likewise.
2514         (gimple_set_vdef): Likewise.
2515         (gimple_omp_return_set_lhs): Likewise.
2516         (gimple_omp_return_lhs): Likewise.
2517         (gimple_omp_return_lhs_ptr): Likewise.
2518         (gimple_call_fntype): Likewise.
2519         (gimple_call_set_fntype): Likewise.
2520         (gimple_call_set_internal_fn): Likewise.
2521         (gimple_call_use_set): Likewise.
2522         (gimple_call_clobber_set): Likewise.
2523         (gimple_bind_vars): Likewise.
2524         (gimple_bind_set_vars): Likewise.
2525         (gimple_bind_body_ptr): Likewise.
2526         (gimple_bind_set_body): Likewise.
2527         (gimple_bind_add_stmt): Likewise.
2528         (gimple_bind_block): Likewise.
2529         (gimple_bind_set_block): Likewise.
2530         (gimple_asm_ninputs): Likewise.
2531         (gimple_asm_noutputs): Likewise.
2532         (gimple_asm_nclobbers): Likewise.
2533         (gimple_asm_nlabels): Likewise.
2534         (gimple_asm_input_op): Likewise.
2535         (gimple_asm_input_op_ptr): Likewise.
2536         (gimple_asm_output_op): Likewise.
2537         (gimple_asm_output_op_ptr): Likewise.
2538         (gimple_asm_set_output_op): Likewise.
2539         (gimple_asm_clobber_op): Likewise.
2540         (gimple_asm_set_clobber_op): Likewise.
2541         (gimple_asm_label_op): Likewise.
2542         (gimple_asm_set_label_op): Likewise.
2543         (gimple_asm_string): Likewise.
2544         (gimple_catch_types): Likewise.
2545         (gimple_catch_types_ptr): Likewise.
2546         (gimple_catch_handler_ptr): Likewise.
2547         (gimple_catch_set_types): Likewise.
2548         (gimple_catch_set_handler): Likewise.
2549         (gimple_eh_filter_types): Likewise.
2550         (gimple_eh_filter_types_ptr): Likewise.
2551         (gimple_eh_filter_failure_ptr): Likewise.
2552         (gimple_eh_filter_set_types): Likewise.
2553         (gimple_eh_filter_set_failure): Likewise.
2554         (gimple_eh_must_not_throw_fndecl): Likewise.
2555         (gimple_eh_must_not_throw_set_fndecl): Likewise.
2556         (gimple_eh_else_n_body_ptr): Likewise.
2557         (gimple_eh_else_e_body_ptr): Likewise.
2558         (gimple_eh_else_set_n_body): Likewise.
2559         (gimple_eh_else_set_e_body): Likewise.
2560         (gimple_try_eval_ptr): Likewise.
2561         (gimple_try_cleanup_ptr): Likewise.
2562         (gimple_try_set_eval): Likewise.
2563         (gimple_try_set_cleanup): Likewise.
2564         (gimple_wce_cleanup_ptr): Likewise.
2565         (gimple_wce_set_cleanup): Likewise.
2566         (gimple_phi_capacity): Likewise.
2567         (gimple_phi_num_args): Likewise.
2568         (gimple_phi_result): Likewise.
2569         (gimple_phi_result_ptr): Likewise.
2570         (gimple_phi_set_result): Likewise.
2571         (gimple_phi_arg): Likewise.
2572         (gimple_phi_set_arg): Likewise.
2573         (gimple_resx_region): Likewise.
2574         (gimple_resx_set_region): Likewise.
2575         (gimple_eh_dispatch_region): Likewise.
2576         (gimple_eh_dispatch_set_region): Likewise.
2577         (gimple_omp_critical_name): Likewise.
2578         (gimple_omp_critical_name_ptr): Likewise.
2579         (gimple_omp_critical_set_name): Likewise.
2580         (gimple_omp_for_clauses): Likewise.
2581         (gimple_omp_for_clauses_ptr): Likewise.
2582         (gimple_omp_for_set_clauses): Likewise.
2583         (gimple_omp_for_collapse): Likewise.
2584         (gimple_omp_for_index): Likewise.
2585         (gimple_omp_for_index_ptr): Likewise.
2586         (gimple_omp_for_set_index): Likewise.
2587         (gimple_omp_for_initial): Likewise.
2588         (gimple_omp_for_initial_ptr): Likewise.
2589         (gimple_omp_for_set_initial): Likewise.
2590         (gimple_omp_for_final): Likewise.
2591         (gimple_omp_for_final_ptr): Likewise.
2592         (gimple_omp_for_set_final): Likewise.
2593         (gimple_omp_for_incr): Likewise.
2594         (gimple_omp_for_incr_ptr): Likewise.
2595         (gimple_omp_for_set_incr): Likewise.
2596         (gimple_omp_for_pre_body_ptr): Likewise.
2597         (gimple_omp_for_set_pre_body): Likewise.
2598         (gimple_omp_parallel_clauses): Likewise.
2599         (gimple_omp_parallel_clauses_ptr): Likewise.
2600         (gimple_omp_parallel_set_clauses): Likewise.
2601         (gimple_omp_parallel_child_fn): Likewise.
2602         (gimple_omp_parallel_child_fn_ptr): Likewise.
2603         (gimple_omp_parallel_set_child_fn): Likewise.
2604         (gimple_omp_parallel_data_arg): Likewise.
2605         (gimple_omp_parallel_data_arg_ptr): Likewise.
2606         (gimple_omp_parallel_set_data_arg): Likewise.
2607         (gimple_omp_task_clauses): Likewise.
2608         (gimple_omp_task_clauses_ptr): Likewise.
2609         (gimple_omp_task_set_clauses): Likewise.
2610         (gimple_omp_task_child_fn): Likewise.
2611         (gimple_omp_task_child_fn_ptr): Likewise.
2612         (gimple_omp_task_set_child_fn): Likewise.
2613         (gimple_omp_task_data_arg): Likewise.
2614         (gimple_omp_task_data_arg_ptr): Likewise.
2615         (gimple_omp_task_set_data_arg): Likewise.
2616         (gimple_omp_taskreg_clauses): Likewise.
2617         (gimple_omp_taskreg_clauses_ptr): Likewise.
2618         (gimple_omp_taskreg_set_clauses): Likewise.
2619         (gimple_omp_taskreg_child_fn): Likewise.
2620         (gimple_omp_taskreg_child_fn_ptr): Likewise.
2621         (gimple_omp_taskreg_set_child_fn): Likewise.
2622         (gimple_omp_taskreg_data_arg): Likewise.
2623         (gimple_omp_taskreg_data_arg_ptr): Likewise.
2624         (gimple_omp_taskreg_set_data_arg): Likewise.
2625         (gimple_omp_task_copy_fn): Likewise.
2626         (gimple_omp_task_copy_fn_ptr): Likewise.
2627         (gimple_omp_task_set_copy_fn): Likewise.
2628         (gimple_omp_task_arg_size): Likewise.
2629         (gimple_omp_task_arg_size_ptr): Likewise.
2630         (gimple_omp_task_set_arg_size): Likewise.
2631         (gimple_omp_task_arg_align): Likewise.
2632         (gimple_omp_task_arg_align_ptr): Likewise.
2633         (gimple_omp_task_set_arg_align): Likewise.
2634         (gimple_omp_single_clauses): Likewise.
2635         (gimple_omp_single_clauses_ptr): Likewise.
2636         (gimple_omp_single_set_clauses): Likewise.
2637         (gimple_omp_target_clauses): Likewise.
2638         (gimple_omp_target_clauses_ptr): Likewise.
2639         (gimple_omp_target_set_clauses): Likewise.
2640         (gimple_omp_target_child_fn): Likewise.
2641         (gimple_omp_target_child_fn_ptr): Likewise.
2642         (gimple_omp_target_set_child_fn): Likewise.
2643         (gimple_omp_target_data_arg): Likewise.
2644         (gimple_omp_target_data_arg_ptr): Likewise.
2645         (gimple_omp_target_set_data_arg): Likewise.
2646         (gimple_omp_teams_clauses): Likewise.
2647         (gimple_omp_teams_clauses_ptr): Likewise.
2648         (gimple_omp_teams_set_clauses): Likewise.
2649         (gimple_omp_sections_clauses): Likewise.
2650         (gimple_omp_sections_clauses_ptr): Likewise.
2651         (gimple_omp_sections_set_clauses): Likewise.
2652         (gimple_omp_sections_control): Likewise.
2653         (gimple_omp_sections_control_ptr): Likewise.
2654         (gimple_omp_sections_set_control): Likewise.
2655         (gimple_omp_for_set_cond): Likewise.
2656         (gimple_omp_for_cond): Likewise.
2657         (gimple_omp_atomic_store_set_val): Likewise.
2658         (gimple_omp_atomic_store_val): Likewise.
2659         (gimple_omp_atomic_store_val_ptr): Likewise.
2660         (gimple_omp_atomic_load_set_lhs): Likewise.
2661         (gimple_omp_atomic_load_lhs): Likewise.
2662         (gimple_omp_atomic_load_lhs_ptr): Likewise.
2663         (gimple_omp_atomic_load_set_rhs): Likewise.
2664         (gimple_omp_atomic_load_rhs): Likewise.
2665         (gimple_omp_atomic_load_rhs_ptr): Likewise.
2666         (gimple_omp_continue_control_def): Likewise.
2667         (gimple_omp_continue_control_def_ptr): Likewise.
2668         (gimple_omp_continue_set_control_def): Likewise.
2669         (gimple_omp_continue_control_use): Likewise.
2670         (gimple_omp_continue_control_use_ptr): Likewise.
2671         (gimple_omp_continue_set_control_use): Likewise.
2672         (gimple_transaction_body_ptr): Likewise.
2673         (gimple_transaction_label): Likewise.
2674         (gimple_transaction_label_ptr): Likewise.
2675         (gimple_transaction_set_body): Likewise.
2676         (gimple_transaction_set_label): Likewise.
2678         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
2679         * ipa-inline-analysis.c (inline_write_summary): Likewise.
2680         * ipa-ref.c (ipa_record_reference): Likewise.
2681         * ipa-reference.c (analyze_function): Likewise.
2682         (ipa_reference_write_optimization_summary): Likewise.
2683         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
2684         (address_taken_from_non_vtable_p): Likewise.
2685         (comdat_can_be_unshared_p_1): Likewise.
2686         * lto-cgraph.c (lto_output_ref): Likewise.
2687         (add_references): Likewise.
2688         (compute_ltrans_boundary): Likewise.
2689         (output_symtab): Likewise.
2690         (input_ref): Likewise.
2691         (input_cgraph_1): Likewise.
2692         (output_cgraph_opt_summary): Likewise.
2693         * lto-streamer-out.c (lto_output): Likewise.
2694         (output_symbol_p): Likewise.
2695         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
2696         (lsei_start_function_in_partition): Likewise.
2697         (lsei_next_variable_in_partition): Likewise.
2698         (lsei_start_variable_in_partition): Likewise.
2699         * symtab.c (insert_to_assembler_name_hash): Likewise.
2700         (unlink_from_assembler_name_hash): Likewise.
2701         (symtab_unregister_node): Likewise.
2702         (symtab_remove_node): Likewise.
2703         (dump_symtab_node): Likewise.
2704         (verify_symtab_base): Likewise.
2705         (verify_symtab_node): Likewise.
2706         (symtab_make_decl_local): Likewise.
2707         (symtab_alias_ultimate_target): Likewise.
2708         (symtab_resolve_alias): Likewise.
2709         (symtab_get_symbol_partitioning_class): Likewise.
2710         * tree-phinodes.c (allocate_phi_node): Likewise.
2711         (reserve_phi_args_for_new_edge): Likewise.
2712         (remove_phi_args): Likewise.
2713         * varpool.c (varpool_node_for_asm): Likewise.
2714         (varpool_remove_unreferenced_decls): Likewise.
2716 2014-04-23  Jeff Law  <law@redhat.com>
2718         PR tree-optimization/60902
2719         * tree-ssa-threadedge.c
2720         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
2721         invalidate outputs from statements that do not produce useful
2722         outputs for threading.
2724 2014-04-23  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
2726         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
2727         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
2728         machine descriptions for Stack Smashing Protector.
2730 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
2732         * aarch64.md (<optab>_rol<mode>3): New pattern.
2733         (<optab>_rolsi3_uxtw): Likewise.
2734         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
2736 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
2738         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
2739         (arm_cortex_a12_tune): Likewise.
2741 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2743         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
2745 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2747         * config/arm/arm.md (arm_rev16si2): New pattern.
2748         (arm_rev16si2_alt): Likewise.
2749         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
2751 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2753         * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
2754         (rev16<mode>2_alt): Likewise.
2755         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
2756         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
2757         (aarch_rev16_shleft_mask_imm_p): Likewise.
2758         (aarch_rev16_p_1): Likewise.
2759         (aarch_rev16_p): Likewise.
2760         * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
2761         (aarch_rev16_shright_mask_imm_p): Likewise.
2762         (aarch_rev16_shleft_mask_imm_p): Likewise.
2764 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2766         * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
2767         * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
2768         rev cost.
2769         (cortex_a53_extra_costs): Likewise.
2770         (cortex_a57_extra_costs): Likewise.
2771         * config/arm/arm.c (cortexa9_extra_costs): Likewise.
2772         (cortexa7_extra_costs): Likewise.
2773         (cortexa8_extra_costs): Likewise.
2774         (cortexa12_extra_costs): Likewise.
2775         (cortexa15_extra_costs): Likewise.
2776         (v7m_extra_costs): Likewise.
2777         (arm_new_rtx_costs): Handle BSWAP.
2779 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2781         * config/arm/arm.c (cortexa8_extra_costs): New table.
2782         (arm_cortex_a8_tune): New tuning struct.
2783         * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
2785 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2787         * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
2789 2014-04-23  Richard Biener  <rguenther@suse.de>
2791         * Makefile.in (OBJS): Remove loop-unswitch.o.
2792         * tree-pass.h (make_pass_rtl_unswitch): Remove.
2793         * passes.def (pass_rtl_unswitch): Likewise.
2794         * loop-init.c (gate_rtl_unswitch): Likewise.
2795         (rtl_unswitch): Likewise.
2796         (pass_data_rtl_unswitch): Likewise.
2797         (pass_rtl_unswitch): Likewise.
2798         (make_pass_rtl_unswitch): Likewise.
2799         * rtl.h (reversed_condition): Likewise.
2800         (compare_and_jump_seq): Likewise.
2801         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
2802         and make static.
2803         * loop-unroll.c (compare_and_jump_seq): Likewise.
2805 2014-04-23  Richard Biener  <rguenther@suse.de>
2807         PR tree-optimization/60903
2808         * tree-ssa-loop-im.c (analyze_memory_references): Remove
2809         commented code block.
2810         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
2811         loop flags to newly created BBs and edges.
2813 2014-04-23  Nick Clifton  <nickc@redhat.com>
2815         * config/msp430/msp430.c (msp430_handle_option): Move function
2816         to msp430-common.c
2817         (msp430_option_override): Simplify mcu and mcpu option handling.
2818         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
2819         support for -mhwmult command line option.
2820         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
2821         -mhwmult command line option.
2822         (msp430_hwmult_enabled): Delete.
2823         (msp43o_output_labelref): Add support for -mhwmult command line option.
2824         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
2825         (umulsidi3): Likewise.
2826         * config/msp430/msp430.opt (mmcu): Add Report attribute.
2827         (mcpu, mlarge, msmall): Likewise.
2828         (mhwmult): New option.
2829         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
2830         prototype.
2831         (msp430_is_f5_mcu): Remove prototype.
2832         (msp430_use_f5_series_hwmult): Add prototype.
2833         * config/msp430/msp430-opts.h: New file.
2834         * common/config/msp430: New directory.
2835         * common/config/msp430/msp430-common.c: New file.
2836         * config.gcc (msp430): Remove target_has_targetm_common.
2837         * doc/invoke.texi: Document -mhwmult command line option.
2839 2014-04-23  Nick Clifton  <nickc@redhat.com>
2841         * config/i386/cygwin.h (ENDFILE_SPEC): Include
2842         default-manifest.o if it can be found in the search path.
2843         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
2845 2014-04-23  Terry Guo  <terry.guo@arm.com>
2847         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
2849 2014-04-23  Richard Biener  <rguenther@suse.de>
2851         PR middle-end/60895
2852         * tree-inline.c (declare_return_variable): Use mark_addressable.
2854 2014-04-23  Richard Biener  <rguenther@suse.de>
2856         PR middle-end/60891
2857         * loop-init.c (loop_optimizer_init): Make sure to apply
2858         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
2860 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
2862         PR sanitizer/60275
2863         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
2864         New options.
2865         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
2866         if flag_sanitize_undefined_trap_on_error.
2867         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
2868         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
2869         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
2870         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2871         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
2872         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
2873         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
2874         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
2875         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
2876         * ubsan.c (ubsan_instrument_unreachable): Return
2877         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
2878         (ubsan_expand_null_ifn): Emit __builtin_trap ()
2879         if flag_sanitize_undefined_trap_on_error and
2880         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
2881         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
2882         instrument_bool_enum_load): Emit __builtin_trap () if
2883         flag_sanitize_undefined_trap_on_error and
2884         __builtin_handle_*_abort () if !flag_sanitize_recover.
2885         * doc/invoke.texi (-fsanitize-recover,
2886         -fsanitize-undefined-trap-on-error): Document.
2888 2014-04-22  Christian Bruel  <christian.bruel@st.com>
2890         * config/sh/sh.md (mov<mode>): Replace movQIHI.
2891          Force immediates to SImode.
2893 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
2895         * config/nios2/nios2.md (UNSPEC_ROUND): New.
2896         (lroundsfsi2): New.
2897         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
2898         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
2899         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
2900         (nios2_fpu_insn): Add entry for round.
2901         (N2FPU_NO_ERRNO_P): Define.
2902         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
2903         flag_errno_math.
2904         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
2906 2014-04-22  Richard Henderson  <rth@redhat.com>
2908         * config/aarch64/aarch64 (addti3, subti3): New expanders.
2909         (add<GPI>3_compare0): Remove leading * from name.
2910         (add<GPI>3_carryin): Likewise.
2911         (sub<GPI>3_compare0): Likewise.
2912         (sub<GPI>3_carryin): Likewise.
2913         (<su_optab>mulditi3): New expander.
2914         (multi3): New expander.
2915         (madd<GPI>): Remove leading * from name.
2917 2014-04-22  Martin Jambor  <mjambor@suse.cz>
2919         * cgraphclones.c (cgraph_function_versioning): Copy
2920         ipa_transforms_to_apply instead of asserting it is empty.
2922 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
2924         PR target/60868
2925         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
2926         on count_exp to get mode.
2928 2014-04-22  Andrew Pinski  <apinski@cavium.com>
2930         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2931         Handle TLS for ILP32.
2932         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
2933         (tlsie_small_<mode>): this and handle PTR.
2934         (tlsie_small_sidi): New pattern.
2935         (tlsle_small): Change to an expand to handle ILP32.
2936         (tlsle_small_<mode>): New pattern.
2937         (tlsdesc_small): Rename to ...
2938         (tlsdesc_small_<mode>): this and handle PTR.
2940 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2942         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
2944 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
2946         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
2947         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
2948         (aarch64_types_signed_poly_qualifiers): Likewise.
2949         (aarch64_types_unsigned_signed_qualifiers): Likewise.
2950         (aarch64_types_poly_signed_qualifiers): Likewise.
2951         (TYPES_REINTERP_SS): Type macro added.
2952         (TYPES_REINTERP_SU): Likewise.
2953         (TYPES_REINTERP_SP): Likewise.
2954         (TYPES_REINTERP_US): Likewise.
2955         (TYPES_REINTERP_PS): Likewise.
2956         (aarch64_fold_builtin): New expression folding added.
2957         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
2958         Declarations removed.
2959         (REINTERP_SS): Declarations added.
2960         (REINTERP_US): Likewise.
2961         (REINTERP_PS): Likewise.
2962         (REINTERP_SU): Likewise.
2963         (REINTERP_SP): Likewise.
2964         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
2965         (vreinterpretq_p8_f64): Likewise.
2966         (vreinterpret_p16_f64): Likewise.
2967         (vreinterpretq_p16_f64): Likewise.
2968         (vreinterpret_f32_f64): Likewise.
2969         (vreinterpretq_f32_f64): Likewise.
2970         (vreinterpret_f64_f32): Likewise.
2971         (vreinterpret_f64_p8): Likewise.
2972         (vreinterpret_f64_p16): Likewise.
2973         (vreinterpret_f64_s8): Likewise.
2974         (vreinterpret_f64_s16): Likewise.
2975         (vreinterpret_f64_s32): Likewise.
2976         (vreinterpret_f64_s64): Likewise.
2977         (vreinterpret_f64_u8): Likewise.
2978         (vreinterpret_f64_u16): Likewise.
2979         (vreinterpret_f64_u32): Likewise.
2980         (vreinterpret_f64_u64): Likewise.
2981         (vreinterpretq_f64_f32): Likewise.
2982         (vreinterpretq_f64_p8): Likewise.
2983         (vreinterpretq_f64_p16): Likewise.
2984         (vreinterpretq_f64_s8): Likewise.
2985         (vreinterpretq_f64_s16): Likewise.
2986         (vreinterpretq_f64_s32): Likewise.
2987         (vreinterpretq_f64_s64): Likewise.
2988         (vreinterpretq_f64_u8): Likewise.
2989         (vreinterpretq_f64_u16): Likewise.
2990         (vreinterpretq_f64_u32): Likewise.
2991         (vreinterpretq_f64_u64): Likewise.
2992         (vreinterpret_s64_f64): Likewise.
2993         (vreinterpretq_s64_f64): Likewise.
2994         (vreinterpret_u64_f64): Likewise.
2995         (vreinterpretq_u64_f64): Likewise.
2996         (vreinterpret_s8_f64): Likewise.
2997         (vreinterpretq_s8_f64): Likewise.
2998         (vreinterpret_s16_f64): Likewise.
2999         (vreinterpretq_s16_f64): Likewise.
3000         (vreinterpret_s32_f64): Likewise.
3001         (vreinterpretq_s32_f64): Likewise.
3002         (vreinterpret_u8_f64): Likewise.
3003         (vreinterpretq_u8_f64): Likewise.
3004         (vreinterpret_u16_f64): Likewise.
3005         (vreinterpretq_u16_f64): Likewise.
3006         (vreinterpret_u32_f64): Likewise.
3007         (vreinterpretq_u32_f64): Likewise.
3009 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3011         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
3012         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
3013         (vreinterpret_p8_s8): Likewise.
3014         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
3015         (vreinterpret_p8_s16): Likewise.
3016         (vreinterpret_p8_s32): Likewise.
3017         (vreinterpret_p8_s64): Likewise.
3018         (vreinterpret_p8_f32): Likewise.
3019         (vreinterpret_p8_u8): Likewise.
3020         (vreinterpret_p8_u16): Likewise.
3021         (vreinterpret_p8_u32): Likewise.
3022         (vreinterpret_p8_u64): Likewise.
3023         (vreinterpret_p8_p16): Likewise.
3024         (vreinterpretq_p8_s8): Likewise.
3025         (vreinterpretq_p8_s16): Likewise.
3026         (vreinterpretq_p8_s32): Likewise.
3027         (vreinterpretq_p8_s64): Likewise.
3028         (vreinterpretq_p8_f32): Likewise.
3029         (vreinterpretq_p8_u8): Likewise.
3030         (vreinterpretq_p8_u16): Likewise.
3031         (vreinterpretq_p8_u32): Likewise.
3032         (vreinterpretq_p8_u64): Likewise.
3033         (vreinterpretq_p8_p16): Likewise.
3034         (vreinterpret_p16_s8): Likewise.
3035         (vreinterpret_p16_s16): Likewise.
3036         (vreinterpret_p16_s32): Likewise.
3037         (vreinterpret_p16_s64): Likewise.
3038         (vreinterpret_p16_f32): Likewise.
3039         (vreinterpret_p16_u8): Likewise.
3040         (vreinterpret_p16_u16): Likewise.
3041         (vreinterpret_p16_u32): Likewise.
3042         (vreinterpret_p16_u64): Likewise.
3043         (vreinterpret_p16_p8): Likewise.
3044         (vreinterpretq_p16_s8): Likewise.
3045         (vreinterpretq_p16_s16): Likewise.
3046         (vreinterpretq_p16_s32): Likewise.
3047         (vreinterpretq_p16_s64): Likewise.
3048         (vreinterpretq_p16_f32): Likewise.
3049         (vreinterpretq_p16_u8): Likewise.
3050         (vreinterpretq_p16_u16): Likewise.
3051         (vreinterpretq_p16_u32): Likewise.
3052         (vreinterpretq_p16_u64): Likewise.
3053         (vreinterpretq_p16_p8): Likewise.
3054         (vreinterpret_f32_s8): Likewise.
3055         (vreinterpret_f32_s16): Likewise.
3056         (vreinterpret_f32_s32): Likewise.
3057         (vreinterpret_f32_s64): Likewise.
3058         (vreinterpret_f32_u8): Likewise.
3059         (vreinterpret_f32_u16): Likewise.
3060         (vreinterpret_f32_u32): Likewise.
3061         (vreinterpret_f32_u64): Likewise.
3062         (vreinterpret_f32_p8): Likewise.
3063         (vreinterpret_f32_p16): Likewise.
3064         (vreinterpretq_f32_s8): Likewise.
3065         (vreinterpretq_f32_s16): Likewise.
3066         (vreinterpretq_f32_s32): Likewise.
3067         (vreinterpretq_f32_s64): Likewise.
3068         (vreinterpretq_f32_u8): Likewise.
3069         (vreinterpretq_f32_u16): Likewise.
3070         (vreinterpretq_f32_u32): Likewise.
3071         (vreinterpretq_f32_u64): Likewise.
3072         (vreinterpretq_f32_p8): Likewise.
3073         (vreinterpretq_f32_p16): Likewise.
3074         (vreinterpret_s64_s8): Likewise.
3075         (vreinterpret_s64_s16): Likewise.
3076         (vreinterpret_s64_s32): Likewise.
3077         (vreinterpret_s64_f32): Likewise.
3078         (vreinterpret_s64_u8): Likewise.
3079         (vreinterpret_s64_u16): Likewise.
3080         (vreinterpret_s64_u32): Likewise.
3081         (vreinterpret_s64_u64): Likewise.
3082         (vreinterpret_s64_p8): Likewise.
3083         (vreinterpret_s64_p16): Likewise.
3084         (vreinterpretq_s64_s8): Likewise.
3085         (vreinterpretq_s64_s16): Likewise.
3086         (vreinterpretq_s64_s32): Likewise.
3087         (vreinterpretq_s64_f32): Likewise.
3088         (vreinterpretq_s64_u8): Likewise.
3089         (vreinterpretq_s64_u16): Likewise.
3090         (vreinterpretq_s64_u32): Likewise.
3091         (vreinterpretq_s64_u64): Likewise.
3092         (vreinterpretq_s64_p8): Likewise.
3093         (vreinterpretq_s64_p16): Likewise.
3094         (vreinterpret_u64_s8): Likewise.
3095         (vreinterpret_u64_s16): Likewise.
3096         (vreinterpret_u64_s32): Likewise.
3097         (vreinterpret_u64_s64): Likewise.
3098         (vreinterpret_u64_f32): Likewise.
3099         (vreinterpret_u64_u8): Likewise.
3100         (vreinterpret_u64_u16): Likewise.
3101         (vreinterpret_u64_u32): Likewise.
3102         (vreinterpret_u64_p8): Likewise.
3103         (vreinterpret_u64_p16): Likewise.
3104         (vreinterpretq_u64_s8): Likewise.
3105         (vreinterpretq_u64_s16): Likewise.
3106         (vreinterpretq_u64_s32): Likewise.
3107         (vreinterpretq_u64_s64): Likewise.
3108         (vreinterpretq_u64_f32): Likewise.
3109         (vreinterpretq_u64_u8): Likewise.
3110         (vreinterpretq_u64_u16): Likewise.
3111         (vreinterpretq_u64_u32): Likewise.
3112         (vreinterpretq_u64_p8): Likewise.
3113         (vreinterpretq_u64_p16): Likewise.
3114         (vreinterpret_s8_s16): Likewise.
3115         (vreinterpret_s8_s32): Likewise.
3116         (vreinterpret_s8_s64): Likewise.
3117         (vreinterpret_s8_f32): Likewise.
3118         (vreinterpret_s8_u8): Likewise.
3119         (vreinterpret_s8_u16): Likewise.
3120         (vreinterpret_s8_u32): Likewise.
3121         (vreinterpret_s8_u64): Likewise.
3122         (vreinterpret_s8_p8): Likewise.
3123         (vreinterpret_s8_p16): Likewise.
3124         (vreinterpretq_s8_s16): Likewise.
3125         (vreinterpretq_s8_s32): Likewise.
3126         (vreinterpretq_s8_s64): Likewise.
3127         (vreinterpretq_s8_f32): Likewise.
3128         (vreinterpretq_s8_u8): Likewise.
3129         (vreinterpretq_s8_u16): Likewise.
3130         (vreinterpretq_s8_u32): Likewise.
3131         (vreinterpretq_s8_u64): Likewise.
3132         (vreinterpretq_s8_p8): Likewise.
3133         (vreinterpretq_s8_p16): Likewise.
3134         (vreinterpret_s16_s8): Likewise.
3135         (vreinterpret_s16_s32): Likewise.
3136         (vreinterpret_s16_s64): Likewise.
3137         (vreinterpret_s16_f32): Likewise.
3138         (vreinterpret_s16_u8): Likewise.
3139         (vreinterpret_s16_u16): Likewise.
3140         (vreinterpret_s16_u32): Likewise.
3141         (vreinterpret_s16_u64): Likewise.
3142         (vreinterpret_s16_p8): Likewise.
3143         (vreinterpret_s16_p16): Likewise.
3144         (vreinterpretq_s16_s8): Likewise.
3145         (vreinterpretq_s16_s32): Likewise.
3146         (vreinterpretq_s16_s64): Likewise.
3147         (vreinterpretq_s16_f32): Likewise.
3148         (vreinterpretq_s16_u8): Likewise.
3149         (vreinterpretq_s16_u16): Likewise.
3150         (vreinterpretq_s16_u32): Likewise.
3151         (vreinterpretq_s16_u64): Likewise.
3152         (vreinterpretq_s16_p8): Likewise.
3153         (vreinterpretq_s16_p16): Likewise.
3154         (vreinterpret_s32_s8): Likewise.
3155         (vreinterpret_s32_s16): Likewise.
3156         (vreinterpret_s32_s64): Likewise.
3157         (vreinterpret_s32_f32): Likewise.
3158         (vreinterpret_s32_u8): Likewise.
3159         (vreinterpret_s32_u16): Likewise.
3160         (vreinterpret_s32_u32): Likewise.
3161         (vreinterpret_s32_u64): Likewise.
3162         (vreinterpret_s32_p8): Likewise.
3163         (vreinterpret_s32_p16): Likewise.
3164         (vreinterpretq_s32_s8): Likewise.
3165         (vreinterpretq_s32_s16): Likewise.
3166         (vreinterpretq_s32_s64): Likewise.
3167         (vreinterpretq_s32_f32): Likewise.
3168         (vreinterpretq_s32_u8): Likewise.
3169         (vreinterpretq_s32_u16): Likewise.
3170         (vreinterpretq_s32_u32): Likewise.
3171         (vreinterpretq_s32_u64): Likewise.
3172         (vreinterpretq_s32_p8): Likewise.
3173         (vreinterpretq_s32_p16): Likewise.
3174         (vreinterpret_u8_s8): Likewise.
3175         (vreinterpret_u8_s16): Likewise.
3176         (vreinterpret_u8_s32): Likewise.
3177         (vreinterpret_u8_s64): Likewise.
3178         (vreinterpret_u8_f32): Likewise.
3179         (vreinterpret_u8_u16): Likewise.
3180         (vreinterpret_u8_u32): Likewise.
3181         (vreinterpret_u8_u64): Likewise.
3182         (vreinterpret_u8_p8): Likewise.
3183         (vreinterpret_u8_p16): Likewise.
3184         (vreinterpretq_u8_s8): Likewise.
3185         (vreinterpretq_u8_s16): Likewise.
3186         (vreinterpretq_u8_s32): Likewise.
3187         (vreinterpretq_u8_s64): Likewise.
3188         (vreinterpretq_u8_f32): Likewise.
3189         (vreinterpretq_u8_u16): Likewise.
3190         (vreinterpretq_u8_u32): Likewise.
3191         (vreinterpretq_u8_u64): Likewise.
3192         (vreinterpretq_u8_p8): Likewise.
3193         (vreinterpretq_u8_p16): Likewise.
3194         (vreinterpret_u16_s8): Likewise.
3195         (vreinterpret_u16_s16): Likewise.
3196         (vreinterpret_u16_s32): Likewise.
3197         (vreinterpret_u16_s64): Likewise.
3198         (vreinterpret_u16_f32): Likewise.
3199         (vreinterpret_u16_u8): Likewise.
3200         (vreinterpret_u16_u32): Likewise.
3201         (vreinterpret_u16_u64): Likewise.
3202         (vreinterpret_u16_p8): Likewise.
3203         (vreinterpret_u16_p16): Likewise.
3204         (vreinterpretq_u16_s8): Likewise.
3205         (vreinterpretq_u16_s16): Likewise.
3206         (vreinterpretq_u16_s32): Likewise.
3207         (vreinterpretq_u16_s64): Likewise.
3208         (vreinterpretq_u16_f32): Likewise.
3209         (vreinterpretq_u16_u8): Likewise.
3210         (vreinterpretq_u16_u32): Likewise.
3211         (vreinterpretq_u16_u64): Likewise.
3212         (vreinterpretq_u16_p8): Likewise.
3213         (vreinterpretq_u16_p16): Likewise.
3214         (vreinterpret_u32_s8): Likewise.
3215         (vreinterpret_u32_s16): Likewise.
3216         (vreinterpret_u32_s32): Likewise.
3217         (vreinterpret_u32_s64): Likewise.
3218         (vreinterpret_u32_f32): Likewise.
3219         (vreinterpret_u32_u8): Likewise.
3220         (vreinterpret_u32_u16): Likewise.
3221         (vreinterpret_u32_u64): Likewise.
3222         (vreinterpret_u32_p8): Likewise.
3223         (vreinterpret_u32_p16): Likewise.
3224         (vreinterpretq_u32_s8): Likewise.
3225         (vreinterpretq_u32_s16): Likewise.
3226         (vreinterpretq_u32_s32): Likewise.
3227         (vreinterpretq_u32_s64): Likewise.
3228         (vreinterpretq_u32_f32): Likewise.
3229         (vreinterpretq_u32_u8): Likewise.
3230         (vreinterpretq_u32_u16): Likewise.
3231         (vreinterpretq_u32_u64): Likewise.
3232         (vreinterpretq_u32_p8): Likewise.
3233         (vreinterpretq_u32_p16): Likewise.
3235 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3237         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
3238         Pattern extended.
3239         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator extended.
3240         (sqabs): Likewise.
3241         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
3242         (vqnegd_s64): Likewise.
3243         (vqabs_s64): Likewise.
3244         (vqabsd_s64): Likewise.
3246 2014-04-22  Richard Henderson  <rth@redhat.com>
3248         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
3249         computation to the top of the loop.
3251 2014-04-22  Renlin  <renlin.li@arm.com>
3252             Jiong Wang  <jiong.wang@arm.com>
3254         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
3255         * config/aarch64/aarch64.c (aarch64_layout_frame)
3256         (aarch64_initial_elimination_offset): Likewise.
3258 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3260         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
3261         Fix indentation.
3263 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
3265         * machmode.h (bitwise_mode_for_mode): Declare.
3266         * stor-layout.h (bitwise_type_for_mode): Likewise.
3267         * stor-layout.c (bitwise_mode_for_mode): New function.
3268         (bitwise_type_for_mode): Likewise.
3269         * builtins.c (fold_builtin_memory_op): Use it instead of
3270         int_mode_for_mode and build_nonstandard_integer_type.
3272 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3274         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
3275         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
3276         (*-*-solaris2*): Simplify.
3277         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
3278         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
3279         *-*-solaris2.9* handling.
3281         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
3282         as bug.
3283         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
3284         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
3285         handling, simplify.
3286         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
3287         * configure: Regenerate.
3289         * config/i386/sol2-9.h: Remove.
3291         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
3292         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
3293         Remove Solaris 9 references.
3295 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
3297         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
3298         (floatuns<GPI:mode><GPF:mode>2): Remove.
3299         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
3300         and floatuns conversions.
3301         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
3302         and floatuns conversions.
3303         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
3304         (w1,w2): New mode attributes for inequal width conversions.
3306 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
3308         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
3309         the output asm format.
3311 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
3313         * config/aarch64/aarch64-simd.md
3314         (aarch64_cm<optab>di): Always split.
3315         (*aarch64_cm<optab>di): New.
3316         (aarch64_cmtstdi): Always split.
3317         (*aarch64_cmtstdi): New.
3319 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3321         PR tree-optimization/60823
3322         * omp-low.c (ipa_simd_modify_function_body): Go through
3323         all SSA_NAMEs and for those refering to vector arguments
3324         which are going to be replaced adjust SSA_NAME_VAR and,
3325         if it is a default definition, change it into a non-default
3326         definition assigned at the beginning of function from new_decl.
3327         (ipa_simd_modify_stmt_ops): Rewritten.
3328         * tree-dfa.c (set_ssa_default_def): When removing default def,
3329         check for NULL loc instead of NULL *loc.
3331 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3333         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
3334         restrictions on core registers for DImode values in Thumb2.
3336 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3338         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
3339         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
3341 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3343         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
3344         (*iordi_notzesidi_di): Likewise.
3345         (*iordi_notsesidi_di): Likewise.
3347 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
3349         * config/arm/arm-protos.h (tune_params): New struct members.
3350         * config/arm/arm.c: Initialise tune_params per processor.
3351         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
3352         for speed, based on new tune_params.
3354 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
3356         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro added.
3357         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added macro.
3358         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment corrected.
3359         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
3360         * config/aarch64/arm_neon.h (vrnd_f64): Added.
3361         (vrnda_f64): Likewise.
3362         (vrndi_f64): Likewise.
3363         (vrndm_f64): Likewise.
3364         (vrndn_f64): Likewise.
3365         (vrndp_f64): Likewise.
3366         (vrndx_f64): Likewise.
3368 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3370         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
3371         GET_MODE_SIZE argument is enum machine_mode.
3373 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
3375         PR target/60910
3376         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
3377         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
3379 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
3381         PR middle-end/60281
3382         * asan.c (asan_emit_stack_protection): Force the base to align to
3383         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
3384         appropriate bits if STRICT_ALIGNMENT.
3385         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
3386         when asan is on.
3387         (expand_used_vars): Leave a space in the stack frame for alignment
3388         if STRICT_ALIGNMENT.
3390 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
3392         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
3393         than a gimple.
3394         (gimple_store_p): Likewise.
3395         (gimple_assign_load_p): Likewise.
3396         (gimple_assign_cast_p): Likewise.
3397         (gimple_clobber_p): Likewise.
3399         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
3400         rather than a gimple.
3401         (gimple_assign_cast_p): Likewise.
3403 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
3405         PR target/60735
3406         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
3407         If mode is DDmode and TARGET_E500_DOUBLE allow move.
3409         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
3410         more debug information for E500 if -mdebug=reg.
3412 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
3414         PR target/60909
3415         * config/i386/i386.c (ix86_expand_builtin)
3416         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
3417         register for target RTX.
3418         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
3420 2014-04-18  Cong Hou  <congh@google.com>
3422         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
3423         the widen-mult pattern by handling two operands with different sizes,
3424         and operands whose size is smaller than half of the result type.
3426 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
3428         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
3429         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
3430         (do_estimate_edge_time): Compute it.
3431         * ipa-inline.c (want_inline_small_function_p): Bypass
3432         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
3434 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
3436         * ipa-inline.c (spec_rem): New static variable.
3437         (dump_overall_stats): New function.
3438         (dump_inline_stats): New function.
3440 2014-04-18  Richard Henderson  <rth@redhat.com>
3442         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
3443         to GET_MODE_SIZE, not a reg_class_t.
3445 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3447         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
3448         (vsx_xxmrglw_<mode>): Likewise.
3450 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3452         PR target/60876
3453         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
3454         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
3455         (rs6000_init_hard_regno_mode_ok): Likewise.
3457 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
3459         * ipa-inline.c (inline_small_functions): Account only non-cold
3460         functions.
3461         * doc/invoke.texi (inline-unit-growth): Update documentation.
3463 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
3465         * config/rs6000/rs6000.md (addti3, subti3): New.
3467 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
3469         PR target/60863
3470         * config/i386/i386.c (ix86_expand_clear): Remove outdated
3471         comment.  Check optimize_insn_for_size_p instead of
3472         optimize_insn_for_speed_p.
3474 2014-04-17  Martin Jambor  <mjambor@suse.cz>
3476         * gimple-iterator.c (gsi_start_edge): New function.
3477         * gimple-iterator.h (gsi_start_edge): Declare.
3478         * tree-sra.c (single_non_eh_succ): New function.
3479         (disqualify_ops_if_throwing_stmt): Renamed to
3480         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
3481         having one non-EH successor BB.
3482         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
3483         generate loads into replacements.
3484         (sra_modify_assign): Likewise and and also use the simple path for
3485         such statements.
3486         (sra_modify_function_body): Commit statements on edges.
3488 2014-04-17  Richard Biener  <rguenther@suse.de>
3490         PR middle-end/60849
3491         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
3492         comparison results and add clarifying comment.
3494 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
3496         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
3497         (blank_mode): Initialize it.
3498         (emit_mode_size_inline, emit_mode_nunits_inline,
3499         emit_mode_inner_inline): New functions.
3500         (emit_insn_modes_h): Call them and surround their output with
3501         #if GCC_VERSION >= 4001 ... #endif.
3502         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
3503         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
3504         mode_* arrays if the argument is __builtin_constant_p.
3505         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
3506         is enum machine_mode.
3508 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3510         * passes.c (opt_pass::execute): Adjust.
3511         (pass_manager::execute_pass_mode_switching): Likewise.
3512         (early_local_passes::execute): Likewise.
3513         (execute_one_pass): Pass cfun to the pass's execute method.
3514         * tree-pass.h (opt_pass::execute): Add function * argument.
3515         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
3516         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
3517         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
3518         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3519         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
3520         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
3521         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
3522         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
3523         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
3524         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
3525         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
3526         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
3527         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
3528         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
3529         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
3530         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
3531         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
3532         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
3533         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
3534         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3535         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3536         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3537         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3538         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3539         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3540         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3541         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
3542         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
3543         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
3544         Adjust.
3546 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3548         * passes.c (opt_pass::gate): Take function * argument.
3549         (gate_all_early_local_passes): Merge into
3550         (early_local_passes::gate): this.
3551         (gate_all_early_optimizations): Merge into
3552         (all_early_optimizations::gate): this.
3553         (gate_all_optimizations): Mege into
3554         (all_optimizations::gate): this.
3555         (gate_all_optimizations_g): Merge into
3556         (all_optimizations_g::gate): this.
3557         (gate_rest_of_compilation): Mege into
3558         (rest_of_compilation::gate): this.
3559         (gate_postreload): Merge into
3560         (postreload::gate): this.
3561         (dump_one_pass): Pass cfun to the pass's gate method.
3562         (execute_ipa_summary_passes): Likewise.
3563         (execute_one_pass): Likewise.
3564         (ipa_write_summaries_2): Likewise.
3565         (ipa_write_optimization_summaries_1): Likewise.
3566         (ipa_read_summaries_1): Likewise.
3567         (ipa_read_optimization_summaries_1): Likewise.
3568         (execute_ipa_stmt_fixups): Likewise.
3569         * tree-pass.h (opt_pass::gate): Add function * argument.
3570         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
3571         combine-stack-adj.c, combine.c, compare-elim.c,
3572         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3573         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
3574         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
3575         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
3576         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
3577         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
3578         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
3579         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
3580         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
3581         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
3582         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
3583         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
3584         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
3585         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
3586         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3587         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3588         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3589         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3590         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3591         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3592         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3593         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3594         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3595         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
3596         var-tracking.c, vtable-verify.c, web.c: Adjust.
3598 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3600         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
3601         * configure: Regenerate.
3603 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3605         * passes.c (dump_one_pass): don't check pass->has_gate.
3606         (execute_ipa_summary_passes): Likewise.
3607         (execute_one_pass): Likewise.
3608         (ipa_write_summaries_2): Likewise.
3609         (ipa_write_optimization_summaries_1): Likewise.
3610         (ipa_read_optimization_summaries_1): Likewise.
3611         (execute_ipa_stmt_fixups): Likewise.
3612         * tree-pass.h (pass_data::has_gate): Remove.
3613         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
3614         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
3615         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
3616         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
3617         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
3618         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
3619         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
3620         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
3621         gimple-low.c, gimple-ssa-isolate-paths.c,
3622         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
3623         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3624         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
3625         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
3626         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
3627         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
3628         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
3629         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
3630         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
3631         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
3632         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
3633         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3634         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3635         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
3636         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
3637         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
3638         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3639         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3640         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
3641         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
3642         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
3643         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
3644         Adjust.
3646 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3648         * pass_manager.h (pass_manager::register_dump_files_1): Remove
3649         declaration.
3650         * passes.c (pass_manager::register_dump_files_1): Merge into
3651         (pass_manager::register_dump_files): this, and remove its handling of
3652         properties since the pass always has the properties anyway.
3653         (pass_manager::pass_manager): Adjust.
3655 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
3657         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
3658         * passes.c (pass_manager::register_dump_files_1): Remove dead code
3659         dealing with properties.
3660         (pass_manager::register_dump_files): Adjust.
3662 2014-03-20  Mark Wielaard  <mjw@redhat.com>
3664         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
3665         then represent the bound as normal constant value.
3667 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
3669         PR target/60847
3670         Forward port from 4.8 branch
3671         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
3673         * config/i386/bmiintrin.h (_blsi_u32): New.
3674         (_blsi_u64): Ditto.
3675         (_blsr_u32): Ditto.
3676         (_blsr_u64): Ditto.
3677         (_blsmsk_u32): Ditto.
3678         (_blsmsk_u64): Ditto.
3679         (_tzcnt_u32): Ditto.
3680         (_tzcnt_u64): Ditto.
3682 2014-04-17  Kito Cheng  <kito@0xlab.org>
3684         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
3686 2014-04-17  Richard Biener  <rguenther@suse.de>
3688         PR middle-end/60849
3689         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
3690         boolean results for comparisons.
3692 2014-04-17  Richard Biener  <rguenther@suse.de>
3694         PR tree-optimization/60836
3695         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
3696         initial PHI args to be gimple values.
3698 2014-04-17   Richard Biener  <rguenther@suse.de>
3700         PR tree-optimization/60841
3701         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
3702         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
3703         of stmts to SLP build.
3704         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3705         (vect_analyze_slp): Likewise.
3706         (vect_analyze_slp_instance): Likewise.
3707         (vect_build_slp_tree): Limit overall SLP tree growth.
3708         * tree-vectorizer.h (vect_analyze_data_refs,
3709         vect_analyze_slp): Adjust prototypes.
3711 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3713         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
3714         Silvermont.
3716 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3718         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
3719         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
3720         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
3721         for TARGET_SLOW_PSHUFB
3723 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
3725         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
3726         * config/i386/i386.c (intel_cost): Ditto.
3728 2014-04-17  Joey Ye  <joey.ye@arm.com>
3730         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
3732 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3734         * opts.c (common_handle_option): Disable -fipa-reference coorectly
3735         with -fuse-profile.
3737 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3739         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
3740         (type_all_derivations_known_p): New predicate.
3741         (type_all_ctors_visible_p): New predicate.
3742         (type_possibly_instantiated_p): New predicate.
3743         (get_odr_type): Compute all_derivations_known.
3744         (dump_odr_type): Dump the flag.
3745         (maybe_record_type): Cleanup.
3746         (record_target_from_binfo): Add bases_to_consider array;
3747         record bases for types w/o instances and skip CXX destructor.
3748         (possible_polymorphic_call_targets_1): Add bases_to_consider
3749         and consider_construction parameters; check if type may have instance.
3750         (get_polymorphic_call_info): Set maybe_in_construction to true
3751         when we know nothing.
3752         (record_targets_from_bases): Skip CXX destructors; they are
3753         never called for types in construction.
3754         (possible_polymorphic_call_targets): Do not record target when
3755         type may not have instance.
3757 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
3759         PR ipa/60854
3760         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
3761         external aliases alive, too.
3763 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
3765         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
3766         definition.
3768 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
3770         * final.c (compute_alignments): Do not apply loop alignment to a block
3771         falling through to the exit.
3773 2014-04-16  Catherine Moore  <clm@codesourcery.com>
3775         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
3776         Adjust constraints for microMIPS store patterns.
3778 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
3780         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
3782 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
3784         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
3785         (append_use): Run at -O0.
3786         (append_vdef): Likewise.
3787         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
3788         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
3790 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
3792         PR tree-optimization/60844
3793         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
3794         (propagate_op_to_single_use, remove_visited_stmt_chain,
3795         linearize_expr, repropagate_negates, reassociate_bb): Use it
3796         instead of gsi_remove.
3798 2014-04-16  Martin Jambor  <mjambor@suse.cz>
3800         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
3801         ipa_transforms_to_apply.
3802         (cgraph_function_versioning): Assert that old_node has empty
3803         ipa_transforms_to_apply.
3804         * trans-mem.c (ipa_tm_create_version): Likewise.
3805         * tree-inline.c (tree_function_versioning): Do not duplicate
3806         ipa_transforms_to_apply.
3808 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3810         PR target/60817
3811         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
3812         x86_64-*-* cases.
3813         Pass necessary as flags on 64-bit Solaris/x86.
3814         Use lowercase relocs for x86_64-*-*.
3815         * configure: Regenerate.
3817 2014-04-15  Jan Hubicka  <jh@suse.cz>
3819         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
3820         (maybe_record_node, likely_target_p): Use it.
3822 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3824         PR target/60839
3825         Revert following patch
3827         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3829         PR target/60735
3830         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
3831         software floating point or no floating point registers, do not
3832         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
3833         in GPRs that occurs after we tested for GPRs that would never be
3834         true.
3836         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
3837         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
3838         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
3839         specifically allow DDmode, since that does not use the SPE SIMD
3840         instructions.
3842 2014-03-21  Mark Wielaard  <mjw@redhat.com>
3844         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
3845         as unsigned or int depending on type and value used.
3847 2014-04-15  Richard Biener  <rguenther@suse.de>
3849         PR rtl-optimization/56965
3850         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
3851         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
3852         ... here.
3853         * alias.c (true_dependence_1): Do not call
3854         nonoverlapping_component_refs_p.
3855         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
3856         nonoverlapping_component_refs_p.
3857         (indirect_refs_may_alias_p): Likewise.
3859 2014-04-15  Teresa Johnson  <tejohnson@google.com>
3861         * cfg.c (dump_bb_info): Fix flags check.
3862         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
3864 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3866         PR rtl-optimization/60663
3867         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
3868         avoid 0 cost.
3870 2014-04-15  Richard Biener  <rguenther@suse.de>
3872         * lto-streamer.h (LTO_major_version): Bump to 4.
3874 2014-04-15  Richard Biener  <rguenther@suse.de>
3876         * common.opt (lto_partition_model): New enum.
3877         (flto-partition=): Merge separate options with a single with argument,
3878         add -flto-partition=one support.
3879         * flag-types.h (enum lto_partition_model): Declare.
3880         * opts.c (finish_options): Remove duplicate -flto-partition=
3881         option check.
3882         * lto-wrapper.c (run_gcc): Adjust.
3884 2014-04-15  Richard Biener  <rguenther@suse.de>
3886         * alias.c (ncr_compar): New function.
3887         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
3889 2014-04-15  Richard Biener  <rguenther@suse.de>
3891         * alias.c (record_component_aliases): Do not walk BINFOs.
3893 2014-04-15  Richard Biener  <rguenther@suse.de>
3895         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3896         Add struct function argument and adjust.
3897         (find_func_aliases_for_call): Likewise.
3898         (find_func_aliases): Likewise.
3899         (find_func_clobbers): Likewise.
3900         (intra_create_variable_infos): Likewise.
3901         (compute_points_to_sets): Likewise.
3902         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
3904 2014-04-15  Richard Biener  <rguenther@suse.de>
3906         * tree.c (iterative_hash_expr): Use enum tree_code_class
3907         to store TREE_CODE_CLASS.
3908         (tree_block): Likewise.
3909         (tree_set_block): Likewise.
3910         * tree.h (fold_build_pointer_plus_loc): Use
3911         convert_to_ptrofftype_loc.
3913 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
3915         PR plugins/59335
3916         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
3917         added in 4.9.
3919 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
3921         * cfgloop.h (struct loop): Move force_vectorize down.
3922         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
3923         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
3924         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
3925         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
3926         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
3927         * tree-core.h (enum annot_expr_kind): Add new kind values.
3928         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
3929         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
3930         kinds.
3931         * tree.def (ANNOTATE_EXPR): Tweak comment.
3933 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
3935         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
3936         cxa_pure_virtual).
3938 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
3940         * tree.h (TYPE_IDENTIFIER): Declare.
3941         * tree.c (subrange_type_for_debug_p): Use it.
3942         * godump.c (go_format_type): Likewise.
3943         * dwarf2out.c (is_cxx_auto, modified_type_die,
3944         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
3945         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
3947 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
3949         PR lto/60820
3950         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
3952 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
3954         * config/i386/i386.c (examine_argument): Return bool.  Return true if
3955         parameter should be passed in memory.
3956         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
3957         (construct_container): Update calls to examine_argument.
3958         (function_arg_advance_64): Ditto.
3959         (return_in_memory_32): Merge with ix86_return_in_memory.
3960         (return_in_memory_64): Ditto.
3961         (return_in_memory_ms_64): Ditto.
3963 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
3965         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
3966         * coverage.c (coverage_compute_profile_id): Handle externally visible
3967         symbols.
3969 2014-04-14  Martin Jambor  <mjambor@suse.cz>
3971         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
3972         DECL_DISREGARD_INLINE_LIMITS functions.
3974 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
3976         PR target/60827
3977         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
3979 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
3981         PR target/60827
3982         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
3983         optimize_insn_for_speed_p instead of
3984         optimize_function_for_speed_p.
3986 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
3988         * doc/invoke.texi (free): Document AArch64.
3990 2014-04-14  Richard Biener  <rguenther@suse.de>
3992         PR tree-optimization/60042
3993         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
3994         (insert_into_preds_of_block): Do not prevent PHI insertion
3995         for REFERENCE exprs here ...
3996         (eliminate_dom_walker::before_dom_children): ... but prevent
3997         their use here under similar conditions when applied to the
3998         IL after PRE optimizations.
4000 2014-04-14  Richard Biener  <rguenther@suse.de>
4002         * passes.def: Move early points-to after early SRA.
4004 2014-04-14  Richard Biener  <rguenther@suse.de>
4006         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
4007         check for which sign-changes we allow when forwarding
4008         a converted value into a switch.
4010 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4012         * stor-layout.c (place_field): Finalize non-constant offset for the
4013         field, if any.
4015 2014-04-14  Richard Biener  <rguenther@suse.de>
4017         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
4018         as argument.
4019         (expand_switch_using_bit_tests_p): Likewise.
4020         (process_switch): Compute and pass on speed_p based on the
4021         switch stmt.
4022         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
4023         optimize_bb_for_speed_p.
4025 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
4027         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
4028         * function.h (struct function): Rename has_force_vect_loops into
4029         has_force_vectorize_loops.
4030         * lto-streamer-in.c (input_cfg): Adjust for renaming.
4031         (input_struct_function_base): Likewise.
4032         * lto-streamer-out.c (output_cfg): Likewise.
4033         (output_struct_function_base): Likewise.
4034         * omp-low.c (expand_omp_simd): Likewise.
4035         * tree-cfg.c (move_sese_region_to_fn): Likewise.
4036         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4037         (version_loop_for_if_conversion): Likewise.
4038         (tree_if_conversion): Likewise.
4039         (main_tree_if_conversion): Likewise.
4040         (gate_tree_if_conversion): Likewise.
4041         * tree-inline.c (copy_loops): Likewise.
4042         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
4043         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
4044         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
4045         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
4046         * tree-vectorizer.c (vectorize_loops): Likewise.
4047         * tree-vectorizer.h (unlimited_cost_model): Likewise.
4049 2014-04-14  Richard Biener  <rguenther@suse.de>
4051         PR lto/60720
4052         * lto-streamer-out.c (wrap_refs): New function.
4053         (lto_output): Wrap symbol references in global initializes in
4054         type-preserving MEM_REFs.
4056 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4058         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
4060 2014-04-14  Christian Bruel  <christian.bruel@st.com>
4062         * config/sh/sh.md (setmemqi): New expand pattern.
4063         * config/sh/sh.h (CLEAR_RATIO): Define.
4064         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
4065         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
4067 2014-04-14  Richard Biener  <rguenther@suse.de>
4069         PR middle-end/55022
4070         * fold-const.c (negate_expr_p): Don't negate directional rounding
4071         division.
4072         (fold_negate_expr): Likewise.
4074 2014-04-14  Richard Biener  <rguenther@suse.de>
4076         PR tree-optimization/59817
4077         PR tree-optimization/60453
4078         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
4079         recursion to catch all CHRECs in the scalar evolution and restrict
4080         the predicate for the remains appropriately.
4082 2014-04-12  Catherine Moore  <clm@codesourcery.com>
4084         * config/mips/constraints.md: Add new register constraint "kb".
4085         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
4086         (*movhi_internal): Likewise.
4087         (*movqi_internal): Likewise.
4088         * config/mips/mips.h (M16_STORE_REGS): New register class.
4089         (REG_CLASS_NAMES): Add M16_STORE_REGS.
4090         (REG_CLASS_CONTENTS): Likewise.
4091         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
4093 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4095         PR c/60194
4096         * doc/invoke.texi (-Wformat-signedness): Document it.
4097         (Wformat=2): Mention that this enables -Wformat-signedness.
4099 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4101         * common/config/epiphany/epiphany-common.c
4102         (epiphany_option_optimization_table): Enable section anchors by
4103         default at -O1 or higher.
4104         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
4105         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
4106         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
4107         carries no extra cost.
4108         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
4109         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
4110         * config/epiphany/predicates.md (memclob_operand): New predicate.
4111         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
4112         Use memclob_operand predicate and X constraint for operand 3.
4114 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4116         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
4117         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
4118         its operands.
4120 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4122         PR rtl-optimization/60651
4123         * mode-switching.c (optimize_mode_switching): Make sure to emit
4124         sets of a lower numbered entity before sets of a higher numbered
4125         entity to a mode of the same or lower priority.
4126         When creating a seginfo for a basic block that starts with a code
4127         label, move the insertion point past the code label.
4128         (new_seginfo): Document and enforce requirement that
4129         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
4130         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
4131         * doc/tm.texi: Regenerate.
4133 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4135         PR target/60811
4136         * config/arc/arc.c (arc_save_restore): Fix assert typo.
4138 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
4140         * BASE-VER: Set to 4.10.0.
4142 2014-04-11  Tobias Burnus  <burnus@net-b.de>
4144         PR other/59055
4145         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
4146         * doc/gcc.texi (Service): Update description in the @menu
4147         * doc/invoke.texi (Option Summary): Remove misplaced and
4148         duplicated @menu.
4150 2014-04-11  Steve Ellcey  <sellcey@mips.com>
4151             Jakub Jelinek  <jakub@redhat.com>
4153         PR middle-end/60556
4154         * expr.c (convert_move): Use emit_store_flag_force instead of
4155         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
4156         argument to it.
4158 2014-04-11  Richard Biener  <rguenther@suse.de>
4160         PR middle-end/60797
4161         * varasm.c (assemble_alias): Avoid endless error reporting
4162         recursion by setting TREE_ASM_WRITTEN.
4164 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4166         * config/s390/s390.md: Add a splitter for NOT rtx.
4168 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
4170         PR rtl-optimization/60663
4171         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
4173 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
4174             Jakub Jelinek  <jakub@redhat.com>
4176         PR lto/60567
4177         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
4178         flag from decl_node to node.
4180 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4182         PR debug/60655
4183         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
4184         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
4185         ameliorating the cases where it can be.
4187 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
4189         Revert
4190         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4192         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4193         (loadsync_<mode>): Change mode.
4194         (load_quadpti, store_quadpti): New.
4195         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4196         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4197         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
4199 2014-04-09  Cong Hou  <congh@google.com>
4201         PR testsuite/60773
4202         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
4203         documentation.
4205 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4207         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
4208         instead of vnor to exploit possible fusion opportunity in the
4209         future.
4210         (altivec_expand_vec_perm_const_le): Likewise.
4212 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
4214         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4215         (loadsync_<mode>): Change mode.
4216         (load_quadpti, store_quadpti): New.
4217         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
4218         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
4220 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
4222         PR target/60763
4223         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
4224         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
4225         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
4227 2014-04-08  Richard Biener  <rguenther@suse.de>
4229         PR middle-end/60706
4230         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
4231         a 64bit widest int print double-int similar to on HWI64 hosts.
4233 2014-04-08  Richard Biener  <rguenther@suse.de>
4235         PR tree-optimization/60785
4236         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
4237         default defs properly.
4239 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
4241         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
4242         (Weffc++): Likewise.
4244 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
4246         * ipa-devirt.c (maybe_record_node): When node is not recorded,
4247         set completep to false rather than true.
4249 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
4251         PR target/60504
4252         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
4253         ARM_TARGET2_DWARF_FORMAT.
4255 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
4257         PR target/60609
4258         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
4259         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
4260         ADDR_DIFF_VEC.
4262 2014-04-07  Richard Biener  <rguenther@suse.de>
4264         PR tree-optimization/60766
4265         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
4266         (may_eliminate_iv): Convert cand_value_at result to desired type.
4268 2014-04-07  Jason Merrill  <jason@redhat.com>
4270         PR c++/60731
4271         * common.opt (-fno-gnu-unique): Add.
4272         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
4274 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4276         * haifa-sched.c: Fix outdated function reference and minor
4277         grammar errors in introductory comment.
4279 2014-04-07  Richard Biener  <rguenther@suse.de>
4281         PR middle-end/60750
4282         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
4283         for noreturn calls.
4284         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
4286 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
4288         PR debug/55794
4289         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
4290         size accounting for thunks.
4291         (pa_asm_output_mi_thunk): Use final_start_function() and
4292         final_end_function() to output function start and end directives.
4294 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
4296         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
4297         device specific ISA/ feature information. Remove short_sp and
4298         errata_skip ds.  Add avr_device_specific_features enum to have device
4299         specific info.
4300         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
4301         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
4302         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
4303         updated device specific info.
4304         * config/avr/avr-mcus.def: Merge device specific details to
4305         dev_attribute field.
4306         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
4307         errata_skip.
4308         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
4309         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
4310         assembler if RMW isa supported by current device.
4311         * config/avr/genmultilib.awk: Update as device info structure changed.
4312         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
4314 2014-04-04  Cong Hou  <congh@google.com>
4316         PR tree-optimization/60656
4317         * tree-vect-stmts.c (supportable_widening_operation):
4318         Fix a bug that elements in a vector with vect_used_by_reduction
4319         property are incorrectly reordered when the operation on it is not
4320         consistant with the one in reduction operation.
4322 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
4324         PR rtl-optimization/60155
4325         * gcse.c (record_set_data): New function.
4326         (single_set_gcse): New function.
4327         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
4328         (hoist_code): Likewise.
4329         (get_pressure_class_and_nregs): Likewise.
4331 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
4333         * explow.c (probe_stack_range): Emit a final optimization blockage.
4335 2014-04-04  Anthony Green  <green@moxielogic.com>
4337         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
4338         typos.
4340 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
4342         PR ipa/59626
4343         * lto-cgraph.c (input_overwrite_node): Check that partitioning
4344         flags are set only during streaming.
4345         * ipa.c (process_references, walk_polymorphic_call_targets,
4346         symtab_remove_unreachable_nodes): Drop bodies of always inline
4347         after early inlining.
4348         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
4350 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
4351         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4353         PR debug/60655
4354         * dwarf2out.c (const_ok_for_output_1): Reject expressions
4355         containing a NOT.
4357 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4359         PR bootstrap/60743
4360         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
4361         duration.
4362         (cortex_a53_fdivd): Likewise.
4364 2014-04-04  Martin Jambor  <mjambor@suse.cz>
4366         PR ipa/60640
4367         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
4368         Adjust all callers.
4369         * cgraph.c (clone_of_p): Also return true if thunks match.
4370         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
4371         cgraph_function_or_thunk_node and an obsolete comment.
4372         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
4373         file.
4374         (build_function_decl_skip_args): Likewise.
4375         (set_new_clone_decl_and_node_flags): New function.
4376         (duplicate_thunk_for_node): Likewise.
4377         (redirect_edge_duplicating_thunks): Likewise.
4378         (cgraph_clone_node): New parameter args_to_skip, pass it to
4379         redirect_edge_duplicating_thunks which is called instead of
4380         cgraph_redirect_edge_callee.
4381         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
4382         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
4384 2014-04-04  Jeff Law  <law@redhat.com>
4386         PR target/60657
4387         * config/arm/predicates.md (const_int_I_operand): New predicate.
4388         (const_int_M_operand): Similarly.
4389         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
4390         const_int_operand.
4391         (insv_t2, extv_reg, extzv_t2): Likewise.
4392         (load_multiple_with_writeback): Similarly for const_int_I_operand.
4393         (pop_multiple_with_writeback_and_return): Likewise.
4394         (vfp_pop_multiple_with_writeback): Likewise
4396 2014-04-04  Richard Biener  <rguenther@suse.de>
4398         PR ipa/60746
4399         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
4400         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
4401         non-GIMPLE_LABELs.
4402         * gimplify.h (gimple_add_tmp_var_fn): Declare.
4403         * gimplify.c (gimple_add_tmp_var_fn): New function.
4404         * gimple-expr.h (create_tmp_reg_fn): Declare.
4405         * gimple-expr.c (create_tmp_reg_fn): New function.
4406         * gimple-low.c (record_vars_into): Don't change cfun.
4407         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
4408         code generation without cfun.
4410 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
4412         PR bootstrap/60719
4413         * Makefile.in (install-driver): Fix shell scripting.
4415 2014-04-03  Cong Hou  <congh@google.com>
4417         PR tree-optimization/60505
4418         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
4419         threshold of number of iterations below which no vectorization
4420         will be done.
4421         * tree-vect-loop.c (new_loop_vec_info):
4422         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
4423         * tree-vect-loop.c (vect_analyze_loop_operations):
4424         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
4425         * tree-vect-loop.c (vect_transform_loop):
4426         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
4427         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
4428         of iterations of the loop and see if we should build the epilogue.
4430 2014-04-03  Richard Biener  <rguenther@suse.de>
4432         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
4433         (streamer_tree_cache_create): Adjust.
4434         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
4435         to allow optional nodes array.
4436         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
4437         (streamer_tree_cache_append): Likewise.
4438         (streamer_tree_cache_create): Create nodes array optionally
4439         as specified by parameter.
4440         * lto-streamer-out.c (create_output_block): Avoid maintaining
4441         the node array in the writer cache.
4442         (DFS_write_tree): Remove assertion.
4443         (produce_asm_for_decls): Free the out decl state hash table early.
4444         * lto-streamer-in.c (lto_data_in_create): Adjust for
4445         streamer_tree_cache_create prototype change.
4447 2014-04-03  Richard Biener  <rguenther@suse.de>
4449         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
4450         set TREE_CHAIN to NULL_TREE.
4452 2014-04-03  Richard Biener  <rguenther@suse.de>
4454         PR tree-optimization/60740
4455         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
4456         over all GIMPLE_COND operands.
4458 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
4460         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
4461         (Weffc++): Remove Scott's numbering, merge lists and reference
4462         Wnon-virtual-dtor.
4464 2014-04-03  Nick Clifton  <nickc@redhat.com>
4466         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
4467         properly.
4469 2014-04-03  Martin Jambor  <mjambor@suse.cz>
4471         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
4472         mention gcc_unreachable before failing.
4473         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
4474         removed symbols.
4476 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
4478         PR ipa/60659
4479         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
4480         inconsistent code and instead mark the context inconsistent.
4481         (possible_polymorphic_call_targets): For inconsistent contexts
4482         return empty complete list.
4484 2014-04-02  Anthony Green  <green@moxielogic.com>
4486         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
4487         (extendqisi2, extendhisi2): Define.
4488         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
4489         (WCHAR_TYPE): Change to unsigned int.
4491 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4493         PR tree-optimization/60733
4494         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
4495         insertion point for PHI candidates to be the end of the feeding
4496         block for the PHI argument.
4498 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
4500         PR rtl-optimization/60650
4501         * lra-constraints.c (process_alt_operands): Decrease reject for
4502         earlyclobber matching.
4504 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4506         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
4508 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4510         * config/spu/spu.c (pad_bb): Do not crash when the last
4511         insn is CODE_FOR_blockage.
4513 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4515         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
4516         lies outside the target mode.
4518 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
4520         PR target/60735
4521         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
4522         software floating point or no floating point registers, do not
4523         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
4524         in GPRs that occurs after we tested for GPRs that would never be
4525         true.
4527         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
4528         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
4529         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
4530         specifically allow DDmode, since that does not use the SPE SIMD
4531         instructions.
4533 2014-04-02  Richard Biener  <rguenther@suse.de>
4535         PR middle-end/60729
4536         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
4537         MODE_INTs.  Properly use negv_optab.
4538         (expand_abs): Likewise.
4540 2014-04-02  Richard Biener  <rguenther@suse.de>
4542         PR bootstrap/60719
4543         * Makefile.in (install-driver): Guard extra installs with special
4544         names properly.
4546 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
4548         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4549         Document vec_vgbbd.
4551 2014-04-01  Richard Henderson  <rth@redhat.com>
4553         PR target/60704
4554         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
4555         alternative enabled before register allocation.
4557 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
4559         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
4560         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
4561         typo.
4562         (nios2_large_got_address): Remove unneeded 'sym' parameter.
4563         (nios2_got_address): Update nios2_large_got_address call site.
4564         (nios2_delegitimize_address): New function.
4565         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
4566         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
4567         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
4569 2014-04-01  Martin Husemann  <martin@duskware.de>
4571         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
4572         for -mabi=32.
4574 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
4576         PR rtl-optimization/60604
4577         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
4578         check from register_operand.
4579         (register_operand): Redefine in terms of general_operand.
4580         (nonmemory_operand): Use register_operand for the non-constant cases.
4582 2014-04-01  Richard Biener  <rguenther@suse.de>
4584         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
4586 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4588         * doc/invoke.texi (mapp-regs): Clarify.
4590 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
4592         * config/i386/avx512fintrin.h (__v32hi): Define type.
4593         (__v64qi): Likewise.
4594         (_mm512_set1_epi8): Define.
4595         (_mm512_set1_epi16): Define.
4596         (_mm512_set4_epi32): Define.
4597         (_mm512_set4_epi64): Define.
4598         (_mm512_set4_pd): Define.
4599         (_mm512_set4_ps): Define.
4600         (_mm512_setr4_epi64): Define.
4601         (_mm512_setr4_epi32): Define.
4602         (_mm512_setr4_pd): Define.
4603         (_mm512_setr4_ps): Define.
4604         (_mm512_setzero_epi32): Define.
4606 2014-03-31  Martin Jambor  <mjambor@suse.cz>
4608         PR middle-end/60647
4609         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
4610         callsite_arguments_match_p.  Updated all callers.  Also check types of
4611         corresponding formal parameters and actual arguments.
4612         (not_all_callers_have_enough_arguments_p) Renamed to
4613         some_callers_have_mismatched_arguments_p.
4615 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
4617         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
4619 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
4621         PR target/60034
4622         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
4623         section anchor.
4625 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
4627         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
4628         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
4629         Split out
4630         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
4631         Use FMAMODE_NOVF512 mode iterator.
4632         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
4633         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
4634         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
4635         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
4636         Split out
4637         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
4638         Use VF_128_256 mode iterator.
4639         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
4640         Ditto.
4642 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4644         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
4645         static chain if needed.
4647 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
4649         PR target/60697
4650         * lra-constraints.c (index_part_to_reg): New.
4651         (process_address): Use it.
4653 2014-03-27  Jeff Law  <law@redhat.com>
4654             Jakub Jelinek  <jakub@redhat.com>
4656         PR target/60648
4657         * expr.c (do_tablejump): Use simplify_gen_binary rather than
4658         gen_rtx_{PLUS,MULT} to build up the address expression.
4660         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
4661         creating non-canonical RTL.
4663 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4665         PR ipa/60243
4666         * ipa-inline.c (want_inline_small_function_p): Short circuit large
4667         functions; reorganize to make cheap checks first.
4668         (inline_small_functions): Do not estimate growth when dumping;
4669         it is expensive.
4670         * ipa-inline.h (inline_summary): Add min_size.
4671         (growth_likely_positive): New function.
4672         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
4673         (set_cond_stmt_execution_predicate): Cleanup.
4674         (estimate_edge_size_and_time): Compute min_size.
4675         (estimate_calls_size_and_time): Likewise.
4676         (estimate_node_size_and_time): Likewise.
4677         (inline_update_overall_summary): Update min_size.
4678         (do_estimate_edge_time): Likewise.
4679         (do_estimate_edge_size): Update.
4680         (do_estimate_edge_hints): Update.
4681         (growth_likely_positive): New function.
4683 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
4685         PR target/60693
4686         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
4687         also if addr has VOIDmode.
4689 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4691         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
4692         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
4693         Declare extern.
4694         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
4695         instructions as well as AdvancedSIMD loads.
4697 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4699         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
4700         Use crypto_aese type.
4701         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
4702         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
4703         crypto_aese, crypto_aesmc.  Move to types.md.
4704         * config/arm/types.md (crypto_aes): Split into crypto_aese,
4705         crypto_aesmc.
4706         * config/arm/iterators.md (crypto_type): Likewise.
4708 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
4710         * cgraph.c: Include expr.h and tree-dfa.h.
4711         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
4712         remove LHS.
4714 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
4716         PR target/60675
4717         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
4718         regs from checking multi-reg pseudos.
4720 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4722         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
4724 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4726         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
4727         if it would clobber the stack pointer, even temporarily.
4729 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
4731         * mode-switching.c: Make small adjustments to the top comment.
4733 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
4735         * config/rs6000/constraints.md (wD constraint): New constraint to
4736         match the constant integer to get the top DImode/DFmode out of a
4737         vector in a VSX register.
4739         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
4740         match the constant integer to get the top DImode/DFmode out of a
4741         vector in a VSX register.
4743         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
4744         for ISA 2.07.
4746         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4747         vbpermq builtins.
4749         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
4750         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
4752         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
4753         Optimize vec_extract of 64-bit values, where the value being
4754         extracted is in the top word, where we can use scalar
4755         instructions.  Add direct move and store support.  Combine the big
4756         endian/little endian vector select load support into a single insn.
4757         (vsx_extract_<mode>_internal1): Likewise.
4758         (vsx_extract_<mode>_internal2): Likewise.
4759         (vsx_extract_<mode>_load): Likewise.
4760         (vsx_extract_<mode>_store): Likewise.
4761         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
4762         combined into vsx_extract_<mode>_load.
4763         (vsx_extract_<mode>_one_le): Likewise.
4765         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
4766         define the top 64-bit vector element.
4768         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
4769         constraint.
4771         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4772         Document vec_vbpermq builtin.
4774         PR target/60672
4775         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
4776         enable use of xxsldwi and xxpermdi builtin functions.
4777         (vec_xxpermdi): Likewise.
4779         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
4780         Document use of vec_xxsldwi and vec_xxpermdi builtins.
4782 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
4784         PR rtl-optimization/60650
4785         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
4786         first_p.  Use it.
4787         (find_spills_for): New.
4788         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
4789         Spill all pseudos on the second iteration.
4791 2014-03-27  Marek Polacek  <polacek@redhat.com>
4793         PR c/50347
4794         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
4795         types.
4797 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4799         * config/s390/s390.c (s390_can_use_return_insn): Check for
4800         call-saved FPRs on 31 bit.
4802 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
4804         PR middle-end/60682
4805         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
4806         if they need regimplification, just drop them instead of
4807         calling gimple_regimplify_operands on them.
4809 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4811         PR target/60580
4812         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
4813         (aarch64_frame_pointer_required): Adjust logic.
4814         (aarch64_can_eliminate): Adjust logic.
4815         (aarch64_override_options_after_change): Adjust logic.
4817 2014-03-27  Dehao Chen  <dehao@google.com>
4819         * ipa-inline.c (early_inliner): Update node's inline info.
4821 2014-03-26  Dehao Chen  <dehao@google.com>
4823         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
4824         compiler inserted conditional jumps for NAN float check.
4826 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
4828         * ubsan.h (ubsan_create_data): Change second argument's type
4829         to const location_t *.
4830         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
4831         _("<unknown>").
4832         (ubsan_create_data): Change second argument to const location_t *PLOC.
4833         Create Loc field whenever PLOC is non-NULL.
4834         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
4835         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
4836         callers.
4838         PR other/59545
4839         * real.c (real_to_integer2): Change type of low to UHWI.
4841 2014-03-26  Tobias Burnus  <burnus@net-b.de>
4843         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
4844         (CILK_SELF_SPECS): New define.
4845         (driver_self_specs): Use it.
4847 2014-03-26  Richard Biener  <rguenther@suse.de>
4849         * tree-pretty-print.c (percent_K_format): Implement special
4850         case for LTO and its stripped down BLOCK tree.
4852 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
4854         PR sanitizer/60636
4855         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
4857         * tree-vrp.c (simplify_internal_call_using_ranges): If only
4858         one range is range_int_cst_p, but not both, at least optimize
4859         addition/subtraction of 0 and multiplication by 0 or 1.
4860         * gimple-fold.c (gimple_fold_call): Fold
4861         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
4862         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
4863         INTEGER_CSTs, try to fold at least x * 0 and y - y.
4865 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
4867         PR rtl-optimization/60452
4868         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
4869         <case REG>: Return 1 for invalid offsets from the frame pointer.
4871 2014-03-26  Marek Polacek  <polacek@redhat.com>
4873         PR c/37428
4874         * doc/extend.texi (C Extensions): Mention variable-length arrays in
4875         a structure/union.
4877 2014-03-26  Marek Polacek  <polacek@redhat.com>
4879         PR c/39525
4880         * doc/extend.texi (Designated Inits): Describe what happens to omitted
4881         field members.
4883 2014-03-26  Marek Polacek  <polacek@redhat.com>
4885         PR other/59545
4886         * ira-color.c (update_conflict_hard_regno_costs): Perform the
4887         multiplication in unsigned type.
4889 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
4891         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
4893 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
4895         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
4897 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
4899         PR ipa/60315
4900         * cif-code.def (UNREACHABLE) New code.
4901         * ipa-inline.c (inline_small_functions): Skip edges to
4902         __builtlin_unreachable.
4903         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
4904         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
4905         predicate to __bulitin_unreachable.
4906         (set_cond_stmt_execution_predicate): Fix issue when
4907         invert_tree_comparison returns ERROR_MARK.
4908         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
4909         propagate to inline clones.
4910         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
4911         to unreachable.
4912         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
4913         * cgraphclones.c (cgraph_clone_node): If call destination is already
4914         ureachable, do not redirect it back.
4915         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
4916         unreachable.
4918 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
4920         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
4921         Do not modify inline clones.
4923 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
4925         * config/i386/i386.md (general_sext_operand): New mode attr.
4926         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
4927         don't generate (sign_extend (const_int)).
4928         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
4929         operands[2].  Use We constraint instead of <i> and
4930         <general_sext_operand> predicate instead of <general_operand>.
4931         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
4932         * config/i386/constraints.md (We): New constraint.
4933         * config/i386/predicates.md (x86_64_sext_operand,
4934         sext_operand): New predicates.
4936 2014-03-25  Martin Jambor  <mjambor@suse.cz>
4938         PR ipa/60600
4939         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
4940         inconsistent devirtualizations to __builtin_unreachable.
4942 2014-03-25  Marek Polacek  <polacek@redhat.com>
4944         PR c/35449
4945         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
4947 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
4949         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
4950         order of elements for big-endian.
4952 2014-03-25  Richard Biener  <rguenther@suse.de>
4954         PR middle-end/60635
4955         * gimplify-me.c (gimple_regimplify_operands): Update the
4956         re-gimplifed stmt.
4958 2014-03-25  Martin Jambor  <mjambor@suse.cz>
4960         PR ipa/59176
4961         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
4962         (lto_output_varpool_node): Likewise.
4963         (input_overwrite_node): Likewise.
4964         (input_varpool_node): Likewise.
4966 2014-03-25  Richard Biener  <rguenther@suse.de>
4968         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
4969         (run_gcc): Likewise.
4971 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
4973         * combine.c (simplify_compare_const): Add MODE argument.
4974         Handle mode_width 0 as very large mode_width.
4975         (try_combine, simplify_comparison): Adjust callers.
4977         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
4978         type to avoid signed integer overflow.
4979         * explow.c (plus_constant): Likewise.
4981 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4983         * doc/generic.texi: Correct typos.
4985 2014-03-24  Tobias Burnus  <burnus@net-b.de>
4987         * doc/invoke.texi (-flto): Expand section about
4988         using static libraries with LTO.
4990 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4992         PR rtl-optimization/60501
4993         * optabs.def (addptr3_optab): New optab.
4994         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
4995         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
4996         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
4998         * lra.c (emit_add3_insn): Use the addptr pattern if available.
5000         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
5002 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
5004         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
5005         _mm512_set1_pd.
5007         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
5008         (_mm256_undefined_ps): Define.
5009         (_mm256_undefined_pd): Define.
5010         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
5011         (_mm_undefined_pd): Define.
5012         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
5013         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
5014         (_mm512_undefined_ps): Define.
5015         (_mm512_undefined_pd): Define.
5016         Use _mm*_undefined_*.
5017         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
5019 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
5021         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
5022         (lshr_simd): DI mode added.
5023         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
5024         (aarch64_ushr_simddi): Likewise.
5025         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
5026         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
5027         (vshrd_n_u64): Likewise.
5029 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5031         * Makefile.in (s-macro_list): Depend on cc1.
5033 2014-03-23  Teresa Johnson  <tejohnson@google.com>
5035         * ipa-utils.c (ipa_print_order): Use specified dump file.
5037 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
5039         PR rtl-optimization/60601
5040         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
5042         * gcc.c (eval_spec_function): Initialize save_growing_value.
5044 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
5046         PR sanitizer/60613
5047         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
5048         code == MINUS_EXPR, never swap op0 with op1.
5050         * toplev.c (init_local_tick): Avoid signed integer multiplication
5051         overflow.
5052         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
5053         shift by first operand's bitsize.
5055 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
5057         PR target/60610
5058         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
5059         redefine to 1 or 0.
5060         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
5061         TARGET_ISA_64BIT_P(x).
5063 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5065         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
5066         pattern for vector nor instead of subtract from splat(-1).
5067         (altivec_expand_vec_perm_const_le): Likewise.
5069 2014-03-21  Richard Henderson  <rth@twiddle.net>
5071         PR target/60598
5072         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
5073         related insns after epilogue_completed.
5075 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5077         PR ipa/59176
5078         * cgraph.h (symtab_node): New flag body_removed.
5079         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
5080         when removing bodies.
5081         * symtab.c (dump_symtab_base): Dump body_removed flag.
5082         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
5083         had their bodies removed.
5085 2014-03-21  Martin Jambor  <mjambor@suse.cz>
5087         PR ipa/60419
5088         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
5089         in the border.
5091 2014-03-21  Richard Biener  <rguenther@suse.de>
5093         PR tree-optimization/60577
5094         * tree-core.h (struct tree_base): Document nothrow_flag use
5095         in VAR_DECL_NONALIASED.
5096         * tree.h (VAR_DECL_NONALIASED): New.
5097         (may_be_aliased): Adjust.
5098         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
5100 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5102         * expr.c (expand_expr_real_1): Remove outdated comment.
5104 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
5106         PR middle-end/60597
5107         * ira.c (adjust_cleared_regs): Call copy_rtx on
5108         *reg_equiv[REGNO (loc)].src_p before passing it to
5109         simplify_replace_fn_rtx.
5111         PR target/60568
5112         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
5113         into CONST, put pic register as first operand of PLUS.  Use
5114         gen_const_mem for both 32-bit and 64-bit PIC got loads.
5116 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5118         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
5120 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5122         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
5123         around for store forwarding issue in the FPU on the UT699.
5124         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
5125         loads and operations if -mfix-ut699 is specified.
5126         (divtf3_hq): Tweak attribute.
5127         (sqrttf2_hq): Likewise.
5129 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
5131         * calls.c (store_one_arg): Remove incorrect const qualification on the
5132         type of the temporary.
5133         * cfgexpand.c (expand_return): Likewise.
5134         * expr.c (expand_constructor): Likewise.
5135         (expand_expr_real_1): Likewise.
5137 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5139         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
5140         of parts.
5142 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
5144         PR target/60039
5145         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
5147 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
5149         * config/arm/aarch-common-protos.h
5150         (alu_cost_table): Fix spelling of "extend".
5151         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
5153 2014-03-19  Richard Biener  <rguenther@suse.de>
5155         PR middle-end/60553
5156         * tree-core.h (tree_type_common): Re-order pointer members
5157         to reduce recursion depth during GC walks.
5159 2014-03-19  Marek Polacek  <polacek@redhat.com>
5161         PR sanitizer/60569
5162         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
5163         before accessing it.
5165 2014-03-19  Richard Biener  <rguenther@suse.de>
5167         PR lto/59543
5168         * lto-streamer-in.c (input_function): In WPA stage do not drop
5169         debug stmts.
5171 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
5173         PR tree-optimization/60559
5174         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
5175         with build_zero_cst assignment.
5177 2014-03-18  Kai Tietz  <ktietz@redhat.com>
5179         PR rtl-optimization/56356
5180         * sdbout.c (sdbout_parms): Verify that parms'
5181         incoming argument is valid.
5182         (sdbout_reg_parms): Likewise.
5184 2014-03-18  Richard Henderson  <rth@redhat.com>
5186         PR target/60562
5187         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
5188         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
5189         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
5191 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
5193         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
5194         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
5195         Italicize plugin event names in description.  Explain that
5196         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
5197         Remind that no GCC functions should be called after PLUGIN_FINISH.
5198         Explain what pragmas with expansion are.
5200 2014-03-18  Martin Liska  <mliska@suse.cz>
5202         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
5203         gimple call statement is update.
5204         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
5205         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
5207 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5209         PR sanitizer/60557
5210         * ubsan.c (ubsan_instrument_unreachable): Call
5211         initialize_sanitizer_builtins.
5212         (ubsan_pass): Likewise.
5214         PR sanitizer/60535
5215         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
5216         varpool_finalize_decl instead of rest_of_decl_compilation.
5218 2014-03-18  Richard Biener  <rguenther@suse.de>
5220         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
5221         by using bitmap_and_compl instead of bitmap_and_compl_into.
5222         (df_rd_transfer_function): Likewise.
5224 2014-03-18  Richard Biener  <rguenther@suse.de>
5226         * doc/lto.texi (fresolution): Fix typo.
5228 2014-03-18  Richard Biener  <rguenther@suse.de>
5230         * doc/invoke.texi (flto): Update for changes in 4.9.
5232 2014-03-18  Richard Biener  <rguenther@suse.de>
5234         * doc/loop.texi: Remove section on the removed lambda framework.
5235         Update loop docs with recent changes in preserving loop structure.
5237 2014-03-18  Richard Biener  <rguenther@suse.de>
5239         * doc/lto.texi (-fresolution): Document.
5241 2014-03-18  Richard Biener  <rguenther@suse.de>
5243         * doc/contrib.texi: Adjust my name.
5245 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
5247         PR ipa/58721
5248         * internal-fn.c: Include diagnostic-core.h.
5249         (expand_BUILTIN_EXPECT): New function.
5250         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
5251         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
5252         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
5253         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
5254         IFN_BUILTIN_EXPECT.
5255         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
5256         Revert 3 argument __builtin_expect code.
5257         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
5258         * gimple-fold.c (gimple_fold_call): Likewise.
5259         * tree.h (fold_builtin_expect): New prototype.
5260         * builtins.c (build_builtin_expect_predicate): Add predictor
5261         argument, if non-NULL, create 3 argument __builtin_expect.
5262         (fold_builtin_expect): No longer static.  Add ARG2 argument,
5263         pass it through to build_builtin_expect_predicate.
5264         (fold_builtin_2): Adjust caller.
5265         (fold_builtin_3): Handle BUILT_IN_EXPECT.
5266         * internal-fn.def (BUILTIN_EXPECT): New.
5268 2014-03-18  Tobias Burnus  <burnus@net-b.de>
5270         PR ipa/58721
5271         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
5272         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
5273         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
5275 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
5277         PR ipa/58721
5278         * predict.c (combine_predictions_for_bb): Fix up formatting.
5279         (expr_expected_value_1, expr_expected_value): Add predictor argument,
5280         fill what it points to if non-NULL.
5281         (tree_predict_by_opcode): Adjust caller, use the predictor.
5282         * predict.def (PRED_COMPARE_AND_SWAP): Add.
5284 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
5286         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
5287         proper constant for the store mode.
5289 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
5291         * symtab.c (change_decl_assembler_name): Fix transparent alias
5292         chain construction.
5294 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
5296         * config/aarch64/aarch64.c: Correct the comments about the
5297         aarch64 stack layout.
5299 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
5301         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
5302         check for GF_OMP_FOR_KIND_FOR.
5304 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
5306         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
5307         ymm and zmm register names.
5309 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
5311         PR target/60516
5312         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
5313         note creation for the 2010-08-31 changes.
5315 2014-03-17  Marek Polacek  <polacek@redhat.com>
5317         PR middle-end/60534
5318         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
5319         as -fno-tree-loop-vectorize.
5320         (expand_omp_simd): Likewise.
5322 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
5324         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
5325         (eligible_for_call_delay): New prototype.
5326         * config/sparc/sparc.c (tls_call_delay): Rename into...
5327         (eligible_for_call_delay): ...this.  Return false if the instruction
5328         cannot be put in the delay slot of a branch.
5329         (eligible_for_restore_insn): Simplify.
5330         (eligible_for_return_delay): Return false if the instruction cannot be
5331         put in the delay slot of a branch and simplify.
5332         (eligible_for_sibcall_delay): Return false if the instruction cannot be
5333         put in the delay slot of a branch.
5334         * config/sparc/sparc.md (fix_ut699): New attribute.
5335         (tls_call_delay): Delete.
5336         (in_call_delay): Reimplement.
5337         (eligible_for_sibcall_delay): Rename into...
5338         (in_sibcall_delay): ...this.
5339         (eligible_for_return_delay): Rename into...
5340         (in_return_delay): ...this.
5341         (in_branch_delay): Reimplement.
5342         (in_uncond_branch_delay): Delete.
5343         (in_annul_branch_delay): Delete.
5345 2014-03-14  Richard Henderson  <rth@redhat.com>
5347         PR target/60525
5348         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
5349         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
5350         (*floathi<X87MODEF>2_i387_with_temp): Remove.
5351         (floathi splitters): Remove.
5352         (float<SWI48x>xf2): New pattern.
5353         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
5354         code that tried to handle DImode for 32-bit, but which was excluded
5355         by the pattern's condition.  Drop allocation of stack temporary.
5356         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
5357         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
5358         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
5359         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
5360         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
5361         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
5362         (*float<SWI48><MODEF>2_sse_interunit): Remove.
5363         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
5364         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
5365         (*float<SWI48x><X87MODEF>2_i387): Remove.
5366         (all float _with_temp splitters): Remove.
5367         (*float<SWI48x><MODEF>2_i387): New pattern.
5368         (*float<SWI48><MODEF>2_sse): New pattern.
5369         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
5370         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
5372 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
5373             Marek Polacek  <polacek@redhat.com>
5375         PR middle-end/60484
5376         * common.opt (dump_base_name_prefixed): New Variable.
5377         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
5378         if x_dump_base_name_prefixed is already set, set it at the end.
5380 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
5382         PR rtl-optimization/60508
5383         * lra-constraints.c (get_reload_reg): Add new parameter
5384         in_subreg_p.
5385         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
5386         Pass the new parameter values.
5388 2014-03-14  Richard Biener  <rguenther@suse.de>
5390         * common.opt: Revert unintented changes from r205065.
5391         * opts.c: Likewise.
5393 2014-03-14  Richard Biener  <rguenther@suse.de>
5395         PR middle-end/60518
5396         * cfghooks.c (split_block): Properly adjust all loops the
5397         block was a latch of.
5399 2014-03-14  Martin Jambor  <mjambor@suse.cz>
5401         PR lto/60461
5402         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
5403         and simplify it.
5405 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
5407         PR target/59396
5408         * config/avr/avr.c (avr_set_current_function): Pass function name
5409         through default_strip_name_encoding before sanity checking instead
5410         of skipping the first char of the assembler name.
5412 2014-03-13  Richard Henderson  <rth@redhat.com>
5414         PR debug/60438
5415         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
5416         (ix86_force_to_memory, ix86_free_from_memory): Remove.
5417         * config/i386/i386-protos.h: Likewise.
5418         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
5419         in the expander instead of a splitter.
5420         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
5421         any possibility of requiring a memory.
5422         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
5423         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
5424         (fp branch splitters): Update for ix86_split_fp_branch.
5425         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
5426         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
5427         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
5428         (*fop_<MODEF>_2_i387): Remove f/r alternative.
5429         (*fop_<MODEF>_3_i387): Likewise.
5430         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
5431         (splitters for the fop_* register patterns): Remove.
5432         (fscalexf4_i387): Rename from *fscalexf4_i387.
5433         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
5435 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
5437         PR tree-optimization/59779
5438         * tree-dfa.c (get_ref_base_and_extent): Use double_int
5439         type for bitsize and maxsize instead of HOST_WIDE_INT.
5441 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
5443         PR rtl-optimization/57320
5444         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
5445         the CFG after thread_prologue_and_epilogue_insns.
5447 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
5449         PR rtl-optimization/57189
5450         * lra-constraints.c (process_alt_operands): Disfavor spilling
5451         vector pseudos.
5453 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
5455         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
5457 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
5459         PR tree-optimization/59025
5460         PR middle-end/60418
5461         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
5462         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
5464 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
5466         PR target/60486
5467         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
5468         calls of avr_out_plus_1.
5470 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
5472         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
5473         BB's single pred and update the father loop's latch info later.
5475 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
5477         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
5478         (VEC_M): Likewise.
5479         (VEC_N): Likewise.
5480         (VEC_R): Likewise.
5481         (VEC_base): Likewise.
5482         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
5483         registers, we need to swap double words in little endian mode.
5485         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
5486         to be a container mode for 128-bit integer operations added in ISA
5487         2.07.  Unlike TImode and PTImode, the preferred register set is
5488         the Altivec/VMX registers for the 128-bit operations.
5490         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
5491         declarations.
5492         (rs6000_split_128bit_ok_p): Likewise.
5494         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
5495         macros for creating ISA 2.07 normal and overloaded builtin
5496         functions with 3 arguments.
5497         (BU_P8V_OVERLOAD_3): Likewise.
5498         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
5499         for use as overloaded functions.
5500         (VPERM_1TI_UNS): Likewise.
5501         (VSEL_1TI): Likewise.
5502         (VSEL_1TI_UNS): Likewise.
5503         (ST_INTERNAL_1ti): Likewise.
5504         (LD_INTERNAL_1ti): Likewise.
5505         (XXSEL_1TI): Likewise.
5506         (XXSEL_1TI_UNS): Likewise.
5507         (VPERM_1TI): Likewise.
5508         (VPERM_1TI_UNS): Likewise.
5509         (XXPERMDI_1TI): Likewise.
5510         (SET_1TI): Likewise.
5511         (LXVD2X_V1TI): Likewise.
5512         (STXVD2X_V1TI): Likewise.
5513         (VEC_INIT_V1TI): Likewise.
5514         (VEC_SET_V1TI): Likewise.
5515         (VEC_EXT_V1TI): Likewise.
5516         (EQV_V1TI): Likewise.
5517         (NAND_V1TI): Likewise.
5518         (ORC_V1TI): Likewise.
5519         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
5520         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
5521         overloaded builtin.
5522         (VADDUQM): Likewise.
5523         (VSUBCUQ): Likewise.
5524         (VADDEUQM): Likewise.
5525         (VADDECUQ): Likewise.
5526         (VSUBEUQM): Likewise.
5527         (VSUBECUQ): Likewise.
5529         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
5530         __int128_t and __uint128_t types.
5531         (__uint128_type): Likewise.
5532         (altivec_categorize_keyword): Add support for vector __int128_t,
5533         vector __uint128_t, vector __int128, and vector unsigned __int128
5534         as a container type for TImode operations that need to be done in
5535         VSX/Altivec registers.
5536         (rs6000_macro_to_expand): Likewise.
5537         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
5538         to support 128-bit integer instructions vaddcuq, vadduqm,
5539         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
5540         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
5542         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
5543         for V1TImode, and set up preferences to use VSX/Altivec registers.
5544         Setup VSX reload handlers.
5545         (rs6000_debug_reg_global): Likewise.
5546         (rs6000_init_hard_regno_mode_ok): Likewise.
5547         (rs6000_preferred_simd_mode): Likewise.
5548         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
5549         (easy_altivec_constant): Likewise.
5550         (output_vec_const_move): Likewise.
5551         (rs6000_expand_vector_set): Convert V1TImode set and extract to
5552         simple move.
5553         (rs6000_expand_vector_extract): Likewise.
5554         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
5555         addressing.
5556         (rs6000_const_vec): Add support for V1TImode.
5557         (rs6000_emit_le_vsx_load): Swap double words when loading or
5558         storing TImode/V1TImode.
5559         (rs6000_emit_le_vsx_store): Likewise.
5560         (rs6000_emit_le_vsx_move): Likewise.
5561         (rs6000_emit_move): Add support for V1TImode.
5562         (altivec_expand_ld_builtin): Likewise.
5563         (altivec_expand_st_builtin): Likewise.
5564         (altivec_expand_vec_init_builtin): Likewise.
5565         (altivec_expand_builtin): Likewise.
5566         (rs6000_init_builtins): Add support for V1TImode type.  Add
5567         support for ISA 2.07 128-bit integer builtins.  Define type names
5568         for the VSX/Altivec vector types.
5569         (altivec_init_builtins): Add support for overloaded vector
5570         functions with V1TImode type.
5571         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
5572         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
5573         external function.
5574         (rs6000_split_128bit_ok_p): Likewise.
5575         (rs6000_handle_altivec_attribute): Create V1TImode from vector
5576         __int128_t and vector __uint128_t.
5578         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
5579         and mode attributes.
5580         (VSX_M): Likewise.
5581         (VSX_M2): Likewise.
5582         (VSm): Likewise.
5583         (VSs): Likewise.
5584         (VSr): Likewise.
5585         (VSv): Likewise.
5586         (VS_scalar): Likewise.
5587         (VS_double): Likewise.
5588         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
5590         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
5591         we support the ISA 2.07 128-bit integer arithmetic instructions.
5592         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
5593         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
5594         and TImode types for use with the builtin functions.
5595         (V1TI_type_node): Likewise.
5596         (unsigned_V1TI_type_node): Likewise.
5597         (intTI_type_internal_node): Likewise.
5598         (uintTI_type_internal_node): Likewise.
5600         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
5601         128-bit builtin functions.
5602         (UNSPEC_VADDEUQM): Likewise.
5603         (UNSPEC_VADDECUQ): Likewise.
5604         (UNSPEC_VSUBCUQ): Likewise.
5605         (UNSPEC_VSUBEUQM): Likewise.
5606         (UNSPEC_VSUBECUQ): Likewise.
5607         (VM): Add V1TImode to vector mode iterators.
5608         (VM2): Likewise.
5609         (VI_unit): Likewise.
5610         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
5611         (altivec_vaddcuq): Likewise.
5612         (altivec_vsubuqm): Likewise.
5613         (altivec_vsubcuq): Likewise.
5614         (altivec_vaddeuqm): Likewise.
5615         (altivec_vaddecuq): Likewise.
5616         (altivec_vsubeuqm): Likewise.
5617         (altivec_vsubecuq): Likewise.
5619         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
5620         mode iterators.
5621         (BOOL_128): Likewise.
5622         (BOOL_REGS_OUTPUT): Likewise.
5623         (BOOL_REGS_OP1): Likewise.
5624         (BOOL_REGS_OP2): Likewise.
5625         (BOOL_REGS_UNARY): Likewise.
5626         (BOOL_REGS_AND_CR0): Likewise.
5628         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
5629         128-bit integer builtin support.
5630         (vec_vadduqm): Likewise.
5631         (vec_vaddecuq): Likewise.
5632         (vec_vaddeuqm): Likewise.
5633         (vec_vsubecuq): Likewise.
5634         (vec_vsubeuqm): Likewise.
5635         (vec_vsubcuq): Likewise.
5636         (vec_vsubuqm): Likewise.
5638         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5639         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
5640         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
5641         128-bit integer add/subtract to ISA 2.07.
5643 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
5645         * config/arc/arc.c (arc_predicate_delay_insns):
5646         Fix third argument passed to conditionalize_nonjump.
5648 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
5650         * config/aarch64/aarch64-builtins.c
5651         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
5652         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
5653         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
5654         instead of __builtin_lfloor.
5655         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
5657 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5659         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
5660         (tree_ssa_ifcombine_bb_1): New function.
5661         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
5662         is an empty forwarder block to then_bb or vice versa and then_bb
5663         and else_bb are effectively swapped.
5665 2014-03-12  Christian Bruel  <christian.bruel@st.com>
5667         PR target/60264
5668         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
5669         REG_CFA_DEF_CFA note.
5670         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
5671         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
5673 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
5675         PR tree-optimization/60454
5676         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
5678 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5680         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
5681         Do not define target_cpu_default2 to generic.
5682         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
5683         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
5684         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
5686 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5687             Marc Glisse  <marc.glisse@inria.fr>
5689         PR tree-optimization/60502
5690         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
5691         instead of build_low_bits_mask.
5693 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
5695         PR middle-end/60482
5696         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
5697         if there are multiple uses, but op doesn't live on E edge.
5698         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
5699         clobber stmts before __builtin_unreachable.
5701 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
5703         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
5704         hard_frame_pointer_rtx.
5705         * cse.c (cse_insn): Remove volatile check.
5706         * cselib.c (cselib_process_insn): Likewise.
5707         * dse.c (scan_insn): Likewise.
5709 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
5711         * config/arc/arc.c (conditionalize_nonjump): New function,
5712         broken out of ...
5713         (arc_ifcvt): ... this.
5714         (arc_predicate_delay_insns): Use it.
5716 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
5718         * config/arc/predicates.md (extend_operand): During/after reload,
5719         allow const_int_operand.
5720         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
5721         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
5722         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
5723         to "i".
5724         (umulsi3_highpart_i): Likewise.
5726 2014-03-11  Richard Biener  <rguenther@suse.de>
5728         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
5729         Add asserts to guard possible wrong-code bugs.
5731 2014-03-11  Richard Biener  <rguenther@suse.de>
5733         PR tree-optimization/60429
5734         PR tree-optimization/60485
5735         * tree-ssa-structalias.c (set_union_with_increment): Properly
5736         take into account all fields that overlap the shifted vars.
5737         (do_sd_constraint): Likewise.
5738         (do_ds_constraint): Likewise.
5739         (get_constraint_for_ptr_offset): Likewise.
5741 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
5743         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
5744         (nios2_compute_frame_layout):
5745         Add calculation of cfun->machine->fp_save_offset.
5746         (nios2_expand_prologue): Correct setting of frame pointer register
5747         in prologue.
5748         (nios2_expand_epilogue): Update recovery of stack pointer from
5749         frame pointer accordingly.
5750         (nios2_initial_elimination_offset): Update calculation of offset
5751         for eliminating to HARD_FRAME_POINTER_REGNUM.
5753 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
5755         PR ipa/60457
5756         * ipa.c (symtab_remove_unreachable_nodes): Don't call
5757         cgraph_get_create_node on VAR_DECLs.
5759 2014-03-10  Richard Biener  <rguenther@suse.de>
5761         PR middle-end/60474
5762         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
5764 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
5766         * config/vms/vms.opt (vms_float_format): New variable.
5768 2014-03-08  Tobias Burnus  <burnus@net-b.de>
5770         * doc/invoke.texi (-fcilkplus): Update implementation status.
5772 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
5773             Richard Biener  <rguenther@suse.de>
5775         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
5776         consistently accross all TUs.
5777         (run_gcc): Enable -fshort-double automatically at link at link-time
5778         and disallow override.
5780 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
5782         PR target/58271
5783         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
5784         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
5785         if they can't be used.
5787 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5789         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
5790         for Solaris 11/x86 ld.
5791         * configure: Regenerate.
5793 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5795         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
5796         (LIB_TLS_SPEC): Save as ld_tls_libs.
5797         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
5798         (HAVE_AS_IX86_TLSLDM): New test.
5799         * configure, config.in: Regenerate.
5800         * config/i386/i386.c (legitimize_tls_address): Fall back to
5801         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
5802         cannot support TLS_MODEL_LOCAL_DYNAMIC.
5803         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
5804         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
5806 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
5808         * common.opt (fira-loop-pressure): Mark as optimization.
5810 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
5812         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
5813         an OpenMP mappable type.
5815 2014-03-06  Matthias Klose  <doko@ubuntu.com>
5817         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
5818         MULTILIB_OSDIRNAMES is not defined.
5820 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
5821             Meador Inge  <meadori@codesourcery.com>
5823         PR target/58595
5824         * config/arm/arm.c (arm_tls_symbol_p): Remove.
5825         (arm_legitimize_address): Call legitimize_tls_address for any
5826         arm_tls_referenced_p expression, handle constant addend.  Call it
5827         before testing for !TARGET_ARM.
5828         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
5830 2014-03-06  Richard Biener  <rguenther@suse.de>
5832         PR middle-end/60445
5833         PR lto/60424
5834         PR lto/60427
5835         Revert
5836         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
5838         * tree-streamer.c (record_common_node): Assert we don't record
5839         nodes with type double.
5840         (preload_common_node): Skip type double, complex double and double
5841         pointer since it is now frontend dependent due to fshort-double option.
5843 2014-03-06  Richard Biener  <rguenther@suse.de>
5845         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
5846         or -fno-lto is specified and the linker has full plugin support.
5847         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
5848         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
5849         * lto-wrapper.c (merge_and_complain): Merge compile-time
5850         optimization levels.
5851         (run_gcc): And pass it through to the link options.
5853 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
5855         PR debug/60381
5856         Revert:
5857         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
5858         PR debug/59992
5859         * cselib.c (remove_useless_values): Skip to avoid quadratic
5860         behavior if the condition moved from...
5861         (cselib_process_insn): ... here holds.
5863 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
5865         PR plugins/59335
5866         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
5867         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
5869         PR plugins/59335
5870         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
5871         (TM_H): Add x86-tune.def.
5873 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5875         * config/aarch64/aarch64.c (generic_tunings):
5876         Use cortexa57_extra_costs.
5878 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
5880         PR lto/60404
5881         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
5882         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
5883         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
5884         cost for in_lto_p.
5886 2014-03-04  Heiher  <r@hev.cc>
5888         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
5889         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
5891 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
5893         * config/i386/predicates.md (const2356_operand): Change to ...
5894         (const2367_operand): ... this.
5895         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
5896         const2367_operand.
5897         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5898         (*avx512pf_scatterpf<mode>sf): Ditto.
5899         (avx512pf_scatterpf<mode>df): Ditto.
5900         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5901         (*avx512pf_scatterpf<mode>df): Ditto.
5902         * config/i386/i386.c (ix86_expand_builtin): Update
5903         incorrect hint operand error message.
5905 2014-03-04  Richard Biener  <rguenther@suse.de>
5907         * lto-section-in.c (lto_get_section_data): Fix const cast.
5909 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
5911         * tree-streamer.c (record_common_node): Assert we don't record
5912         nodes with type double.
5913         (preload_common_node): Skip type double, complex double and double
5914         pointer since it is now frontend dependent due to fshort-double option.
5916 2014-03-04  Richard Biener  <rguenther@suse.de>
5918         PR lto/60405
5919         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
5920         (lto_input_toplevel_asms): Likewise.
5921         * lto-section-in.c (lto_get_section_data): Instead do it here
5922         for every section.
5924 2014-03-04  Richard Biener  <rguenther@suse.de>
5926         PR tree-optimization/60382
5927         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
5928         dead PHIs a reduction.
5930 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
5932         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
5933         hint value.
5934         (_mm_prefetch): Move out of GCC target("sse") pragma.
5935         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
5936         GCC target("prfchw") pragma.
5937         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
5938         for locality <= 2.
5939         * config/i386/i386.c (ix86_option_override_internal): Enable
5940         -mprfchw with -mprefetchwt1.
5942 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
5944         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
5945         Mark as varying.
5947 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
5949         * opts.h (CL_PCH_IGNORE): Define.
5950         * targhooks.c (option_affects_pch_p):
5951         Return false for options that have CL_PCH_IGNORE set.
5952         * opt-functions.awk: Process PchIgnore.
5953         * doc/options.texi: Document PchIgnore.
5955         * config/arc/arc.opt (misize): Add PchIgnore property.
5957 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5959         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
5960         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
5961         constraint on constants to permit them being loaded into
5962         GENERAL_REGS or BASE_REGS.
5964 2014-03-03  Nick Clifton  <nickc@redhat.com>
5966         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
5967         anti-cacnonical alternatives.
5968         (negandhi3_real): New pattern.
5969         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
5971 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
5973         * config/avr/avr-mcus.def: Remove atxmega16x1.
5974         * config/avr/avr-tables.opt: Regenerate.
5975         * config/avr/t-multilib: Regenerate.
5976         * doc/avr-mmcu.texi: Regenerate.
5978 2014-03-03  Tobias Grosser  <tobias@grosser.es>
5979             Mircea Namolaru  <mircea.namolaru@inria.fr>
5981         PR tree-optimization/58028
5982         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
5983         scalar dimensions.
5985 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5987         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
5988         not handled by recognizers.
5990 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
5992         PR middle-end/60175
5993         * function.c (expand_function_end): Don't emit
5994         clobber_return_register sequence if clobber_after is a BARRIER.
5995         * cfgexpand.c (construct_exit_block): Append instructions before
5996         return_label to prev_bb.
5998 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6000         * config/rs6000/constraints.md: Document reserved use of "wc".
6002 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6004         PR ipa/60150
6005         * ipa.c (function_and_variable_visibility): When dissolving comdat
6006         group, also set all symbols to local.
6008 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
6010         PR ipa/60306
6012         Revert:
6013         2013-12-14   Jan Hubicka  <jh@suse.cz>
6014         PR middle-end/58477
6015         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
6017 2014-03-02  Jon Beniston  <jon@beniston.com>
6019         PR bootstrap/48230
6020         PR bootstrap/50927
6021         PR bootstrap/52466
6022         PR target/46898
6023         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
6024           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
6025         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
6026         (simple_return, *simple_return): New patterns
6027         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
6028         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
6030 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
6032         * dwarf2out.c (gen_subprogram_die): Tidy.
6034 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
6036         PR target/60071
6037         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
6038         (*mov_t_msb_neg_negc): ... this new insn.
6040 2014-02-28  Jason Merrill  <jason@redhat.com>
6042         PR c++/58678
6043         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
6044         function.
6046 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
6048         PR c++/60314
6049         * dwarf2out.c (decltype_auto_die): New static.
6050         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
6051         (gen_type_die_with_usage): Handle 'decltype(auto)'.
6052         (is_cxx_auto): Likewise.
6054 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
6056         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
6057         we are not using general regs only.
6059 2014-02-28  Richard Biener  <rguenther@suse.de>
6061         PR target/60280
6062         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
6063         previous fix and only allow to remove trivial pre-headers
6064         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
6065         (remove_forwarder_block): Properly update the latch of a loop.
6067 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6069         PR debug/59992
6070         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
6071         (cselib_preserved_hash_table): New.
6072         (preserve_constants_and_equivs): Move preserved vals to it.
6073         (cselib_find_slot): Look it up first.
6074         (cselib_init): Initialize it.
6075         (cselib_finish): Release it.
6076         (dump_cselib_table): Dump it.
6078 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6080         PR debug/59992
6081         * cselib.c (remove_useless_values): Skip to avoid quadratic
6082         behavior if the condition moved from...
6083         (cselib_process_insn): ... here holds.
6085 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
6087         PR debug/57232
6088         * var-tracking.c (vt_initialize): Apply the same condition to
6089         preserve the CFA base value.
6091 2014-02-28  Joey Ye  <joey.ye@arm.com>
6093         PR target/PR60169
6094         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
6095         if reload in progress or completed.
6097 2014-02-28  Tobias Burnus  <burnus@net-b.de>
6099         PR middle-end/60147
6100         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
6101         NAMELIST_DECL.
6103 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
6105         * doc/tm.texi.in (Condition Code Status): Update documention for
6106         relative locations of cc0-setter and cc0-user.
6108 2014-02-27  Jeff Law  <law@redhat.com>
6110         PR rtl-optimization/52714
6111         * combine.c (try_combine): When splitting an unrecognized PARALLEL
6112         into two independent simple sets, if I3 is a jump, ensure the
6113         pattern we place into I3 is a (set (pc) ...).
6115 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
6116             Jeff Law  <law@redhat.com>
6118         PR rtl-optimization/49847
6119         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
6120         are in different blocks.
6121         * doc/tm.texi (Condition Code Status): Update documention for
6122         relative locations of cc0-setter and cc0-user.
6124 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
6126         PR target/59222
6127         * lra.c (lra_emit_add): Check SUBREG too.
6129 2014-02-27  Andreas Schwab  <schwab@suse.de>
6131         * config/m68k/m68k.c (m68k_option_override): Disable
6132         -flive-range-shrinkage for classic m68k.
6133         (m68k_override_options_after_change): Likewise.
6135 2014-02-27  Marek Polacek  <polacek@redhat.com>
6137         PR middle-end/59223
6138         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
6139         -Wmaybe-uninitialized.
6141 2014-02-27  Alan Modra  <amodra@gmail.com>
6143         PR target/57936
6144         * reload1.c (emit_input_reload_insns): When reload_override_in,
6145         set old to rl->in_reg when rl->in_reg is a subreg.
6147 2014-02-26  Richard Biener  <rguenther@suse.de>
6149         PR bootstrap/60343
6150         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
6152 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6154         * common/config/i386/predicates.md (const1256_operand): Remove.
6155         (const2356_operand): New.
6156         (const_1_to_2_operand): Remove.
6157         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
6158         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6159         (*avx512pf_gatherpf<mode>sf): Ditto.
6160         (avx512pf_gatherpf<mode>df): Ditto.
6161         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6162         (*avx512pf_gatherpf<mode>df): Ditto.
6163         (avx512pf_scatterpf<mode>sf): Ditto.
6164         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6165         (*avx512pf_scatterpf<mode>sf): Ditto.
6166         (avx512pf_scatterpf<mode>df): Ditto.
6167         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6168         (*avx512pf_scatterpf<mode>df): Ditto.
6169         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
6171 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
6173         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
6174         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
6175         (_mm512_mask_testn_epi64_mask): Move to ...
6176         * config/i386/avx512cdintrin.h: Here.
6177         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
6178         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
6179         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
6180         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
6181         TARGET_AVX512F from TARGET_AVX512CD.
6183 2014-02-26  Richard Biener  <rguenther@suse.de>
6185         PR ipa/60327
6186         * ipa.c (walk_polymorphic_call_targets): Properly guard
6187         call to inline_update_overall_summary.
6189 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
6191         PR target/60280
6192         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
6193         and latches only if requested.  Fix latch if it is removed.
6194         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
6195         LOOPS_HAVE_PREHEADERS.
6197 2014-02-25  Andrew Pinski  <apinski@cavium.com>
6199         * builtins.c (expand_builtin_thread_pointer): Create a new target
6200         when the target is NULL.
6202 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
6204         PR rtl-optimization/60317
6205         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6206         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
6207         * lra-assigns.c: Include params.h.
6208         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
6209         other reload pseudos considerations.
6211 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6213         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
6214         to use canonical form for nor<mode>3.
6216 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6218         PR target/55426
6219         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
6220         conversions.
6222 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
6224         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
6225         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
6226         (ix86_handle_option): Handle OPT_mprefetchwt1.
6227         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
6228         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
6229         PREFETCHWT1 CPUID.
6230         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6231         OPTION_MASK_ISA_PREFETCHWT1.
6232         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
6233         (PTA_PREFETCHWT1): New.
6234         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
6235         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
6236         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
6237         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
6238         (*prefetch_avx512pf_<mode>_: Change into ...
6239         (*prefetch_prefetchwt1_<mode>: This.
6240         * config/i386/i386.opt (mprefetchwt1): New.
6241         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
6242         (_mm_prefetch): Handle intent to write.
6243         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
6245 2014-02-25  Richard Biener  <rguenther@suse.de>
6247         PR middle-end/60291
6248         * emit-rtl.c (mem_attrs_htab): Remove.
6249         (mem_attrs_htab_hash): Likewise.
6250         (mem_attrs_htab_eq): Likewise.
6251         (set_mem_attrs): Always allocate new mem-attrs when something changed.
6252         (init_emit_once): Do not allocate mem_attrs_htab.
6254 2014-02-25  Richard Biener  <rguenther@suse.de>
6256         PR lto/60319
6257         * lto-opts.c (lto_write_options): Output non-explicit conservative
6258         -fwrapv, -fno-trapv and -fno-strict-overflow.
6259         * lto-wrapper.c (merge_and_complain): Handle merging those options.
6260         (run_gcc): And pass them through.
6262 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
6264         * sel-sched.c (calculate_new_fences): New parameter ptime.
6265         Calculate it as a maximum over all fence cycles.
6266         (sel_sched_region_2): Adjust the call to calculate_new_fences.
6267         Print the final schedule timing when sched_verbose.
6269 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
6271         PR rtl-optimization/60292
6272         * sel-sched.c (fill_vec_av_set): Do not reset target availability
6273         bit fot the fence instruction.
6275 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
6277         * calls.h: Fix typo in comment.
6279 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
6281         * config/pa/pa.c (pa_output_move_double): Don't valididate when
6282         adjusting offsetable addresses.
6284 2014-02-24  Guozhi Wei  <carrot@google.com>
6286         * sparseset.h (sparseset_pop): Fix the wrong index.
6288 2014-02-24  Walter Lee  <walt@tilera.com>
6290         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
6291         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
6292         triplet.
6293         * common/config/tilegx/tilegx-common.c
6294         (TARGET_DEFAULT_TARGET_FLAGS): Define.
6295         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
6296         (LINK_SPEC): Ditto.
6297         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
6298         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
6299         (tilegx_gimplify_va_arg_expr): Handle big endian.
6300         (tilegx_expand_unaligned_load): Ditto.
6301         (tilegx_expand_unaligned_store): Ditto.
6302         (TARGET_RETURN_IN_MSB): New.
6303         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
6304         (TARGET_ENDIAN_DEFAULT): New.
6305         (TARGET_BIG_ENDIAN): Handle big endian.
6306         (BYTES_BIG_ENDIAN): Ditto.
6307         (WORDS_BIG_ENDIAN): Ditto.
6308         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
6309         (ENDIAN_SPEC): New.
6310         (EXTRA_SPECS): New.
6311         * config/tilegx/tilegx.md (extv): Handle big endian.
6312         (extzv): Ditto.
6313         (insn_st<n>): Ditto.
6314         (insn_st<n>_add<bitsuffix>): Ditto.
6315         (insn_stnt<n>): Ditto.
6316         (insn_stnt<n>_add<bitsuffix>):Ditto.
6317         (vec_interleave_highv8qi): Handle big endian.
6318         (vec_interleave_highv8qi_be): New.
6319         (vec_interleave_highv8qi_le): New.
6320         (insn_v1int_h): Handle big endian.
6321         (vec_interleave_lowv8qi): Handle big endian.
6322         (vec_interleave_lowv8qi_be): New.
6323         (vec_interleave_lowv8qi_le): New.
6324         (insn_v1int_l): Handle big endian.
6325         (vec_interleave_highv4hi): Handle big endian.
6326         (vec_interleave_highv4hi_be): New.
6327         (vec_interleave_highv4hi_le): New.
6328         (insn_v2int_h): Handle big endian.
6329         (vec_interleave_lowv4hi): Handle big endian.
6330         (vec_interleave_lowv4hi_be): New.
6331         (vec_interleave_lowv4hi_le): New.
6332         (insn_v2int_l): Handle big endian.
6333         (vec_interleave_highv2si): Handle big endian.
6334         (vec_interleave_highv2si_be): New.
6335         (vec_interleave_highv2si_le): New.
6336         (insn_v4int_h): Handle big endian.
6337         (vec_interleave_lowv2si): Handle big endian.
6338         (vec_interleave_lowv2si_be): New.
6339         (vec_interleave_lowv2si_le): New.
6340         (insn_v4int_l): Handle big endian.
6341         * config/tilegx/tilegx.opt (mbig-endian): New option.
6342         (mlittle-endian): New option.
6343         * doc/install.texi: Document tilegxbe-linux.
6344         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
6346 2014-02-24  Martin Jambor  <mjambor@suse.cz>
6348         PR ipa/60266
6349         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
6350         there are no parameter descriptors.
6352 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
6354         PR rtl-optimization/60268
6355         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
6356         initialization to ...
6357         (sched_rgn_init): ... here.
6358         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
6360 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
6362         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
6363         names.
6365 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
6367         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
6368         definition.
6370 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
6372         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
6373         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
6375 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
6377         * config/microblaze/predicates.md: Add cmp_op predicate.
6378         * config/microblaze/microblaze.md: Add branch_compare instruction
6379         which uses cmp_op predicate and emits cmp insn before branch.
6380         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
6381         to microblaze_expand_conditional_branch and consolidate logic.
6382         (microblaze_expand_conditional_branch): emit branch_compare
6383         insn instead of handling cmp op separate from branch insn.
6385 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6387         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
6388         to permit subregs.
6390 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6392         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
6393         define_insn with define_expand and new define_insn
6394         *altivec_lve<VI_char>x_internal.
6395         (altivec_stve<VI_char>x): Replace define_insn with define_expand
6396         and new define_insn *altivec_stve<VI_char>x_internal.
6397         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
6398         prototype.
6399         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
6400         lve*x built-ins.
6401         (altivec_expand_stvex_be): New function.
6403 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
6405         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
6406         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
6407         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
6408         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
6410 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
6412         PR target/60298
6413         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
6414         instead of emit_move_insn.
6416 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6418         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
6419         vspltw with vsldoi.
6420         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
6421         gen_altivec_vsumsws.
6423 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6425         * config/rs6000/altivec.md (altivec_lvxl): Rename as
6426         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
6427         (altivec_lvxl_<mode>): New define_expand incorporating
6428         -maltivec=be semantics where needed.
6429         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
6430         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
6431         semantics where needed.
6432         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
6433         (altivec_stvx_<mode>): New define_expand incorporating
6434         -maltivec=be semantics where needed.
6435         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
6436         VM2 iterator instead of V4SI.
6437         (altivec_stvxl_<mode>): New define_expand incorporating
6438         -maltivec=be semantics where needed.
6439         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
6440         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
6441         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
6442         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
6443         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
6444         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
6445         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
6446         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
6447         ALTIVEC_BUILTIN_STVXL.
6448         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
6449         (altivec_expand_stvx_be): Likewise.
6450         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
6451         (altivec_expand_lvx_be): Likewise.
6452         (altivec_expand_stvx_be): Likewise.
6453         (altivec_expand_builtin): Add cases for
6454         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
6455         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
6456         (altivec_init_builtins): Add definitions for
6457         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
6458         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
6460 2014-02-21  Catherine Moore  <clm@codesourcery.com>
6462         * doc/invoke.texi (mvirt, mno-virt): Document.
6463         * config/mips/mips.opt (mvirt): New option.
6464         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
6466 2014-02-21  Richard Biener  <rguenther@suse.de>
6468         PR tree-optimization/60276
6469         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
6470         (STMT_VINFO_MIN_NEG_DIST): New macro.
6471         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
6472         STMT_VINFO_MIN_NEG_DIST.
6473         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
6474         made for negative dependence distances still hold.
6476 2014-02-21  Richard Biener  <rguenther@suse.de>
6478         PR middle-end/60291
6479         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
6480         DECL_INITIAL for globals not in the current function context.
6482 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
6484         PR tree-optimization/56490
6485         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
6486         * tree-ssa-uninit.c: Include params.h.
6487         (compute_control_dep_chain): Add num_calls argument, return false
6488         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
6489         num_calls to recursive call.
6490         (find_predicates): Change dep_chain into normal array,
6491         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
6492         variable and adjust compute_control_dep_chain caller.
6493         (find_def_preds): Likewise.
6495 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
6497         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
6498         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
6500 2014-02-21  Nick Clifton  <nickc@redhat.com>
6502         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
6503         (pushhi1): Likewise.
6504         (popqi1): Add mode to pre_dec.
6505         (pophi1): Likewise.
6507 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
6509         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
6510         mode for mask of V8SFmode permutation.
6512 2014-02-20  Richard Henderson  <rth@redhat.com>
6514         PR c++/60272
6515         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
6516         a new pseudo for OLDVAL.
6518 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
6520         PR target/57896
6521         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
6522         gen_reg_rtx if d->testing_p.
6523         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
6524         if d->testing_p and we will certainly return true.
6525         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
6526         if d->testing_p.
6528 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
6530         * emit-rtl.c (gen_reg_rtx): Assert that
6531         crtl->emit.regno_pointer_align_length is non-zero.
6533 2014-02-20  Richard Henderson  <rth@redhat.com>
6535         PR c++/60272
6536         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
6537         on failure the store back into EXPECT.
6539 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
6540             Sandra Loosemore  <sandra@codesourcery.com>
6542         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
6543         * config/nios2/nios2.c (nios2_function_profiler): Add
6544         -fPIC (flag_pic == 2) support.
6545         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
6546         (nios2_large_offset_p): New function.
6547         (nios2_unspec_reloc_p): Move up position, update to use
6548         nios2_large_offset_p.
6549         (nios2_unspec_address): Remove function.
6550         (nios2_unspec_offset): New function.
6551         (nios2_large_got_address): New function.
6552         (nios2_got_address): Add large offset support.
6553         (nios2_legitimize_tls_address): Update usage of removed and new
6554         functions.
6555         (nios2_symbol_binds_local_p): New function.
6556         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
6557         (nios2_legitimize_address): Update to use nios2_large_offset_p.
6558         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
6559         (nios2_print_operand): Merge H/L processing, add hiadj/lo
6560         processing for (const (unspec ...)).
6561         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
6563 2014-02-20  Richard Biener  <rguenther@suse.de>
6565         * tree-cfg.c (replace_uses_by): Mark altered BBs before
6566         doing the substitution.
6567         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
6569 2014-02-20  Martin Jambor  <mjambor@suse.cz>
6571         PR ipa/55260
6572         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
6573         info when checking whether lattices are bottom.
6575 2014-02-20  Richard Biener  <rguenther@suse.de>
6577         PR middle-end/60221
6578         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
6579         regions at -O0.
6581 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
6583         PR ipa/58555
6584         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
6585         parameter specifying the scaling.
6586         (inline_call): Update.
6587         (want_inline_recursively): Guard division by zero.
6588         (recursive_inlining): Update.
6589         * ipa-inline.h (clone_inlined_nodes): Update.
6591 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
6593         PR target/60204
6594         * config/i386/i386.c (classify_argument): Pass structures of size
6595         64 bytes or less in register.
6597 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
6598             Kirill Yukhin  <kirill.yukhin@intel.com>
6600         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
6601         (_mm_rcp28_round_ss): Ditto.
6602         (_mm_rsqrt28_round_sd): Ditto.
6603         (_mm_rsqrt28_round_ss): Ditto.
6604         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
6605         (_mm_rcp14_round_ss): Ditto.
6606         (_mm_rsqrt14_round_sd): Ditto.
6607         (_mm_rsqrt14_round_ss): Ditto.
6608         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
6609         the first input operand, get rid of match_dup.
6610         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
6611         attribute to sse.
6612         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6613         Ditto.
6614         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
6615         operand as the first input operand, set type attribute.
6616         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6617         Set type attribute.
6618         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
6619         operand as the first input operand, set type attribute.
6621 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6623         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
6624         bit of zero.
6626 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
6628         PR target/60207
6629         * config/i386/i386.c (construct_container): Remove TFmode check
6630         for X86_64_INTEGER_CLASS.
6632 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
6634         PR target/59794
6635         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
6636         only when -Wpsabi is enabled.
6638 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
6640          PR target/59799
6641         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
6642         passing arrays in registers are the same as for structs, so remove the
6643         special case for them.
6645 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
6647         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
6648         destination type, extract only the valid bits if the source type is not
6649         integral and has a different mode.
6651 2014-02-19  Richard Biener  <rguenther@suse.de>
6653         PR ipa/60243
6654         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
6655         for all calls.
6657 2014-02-19  Richard Biener  <rguenther@suse.de>
6659         PR ipa/60243
6660         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
6661         (ipa_modify_call_arguments): Emit an argument load explicitely and
6662         preserve virtual SSA form there and for the replacement call.
6663         Do not update SSA form nor free dominance info.
6665 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6667         * ipa.c (function_and_variable_visibility): Also clear WEAK
6668         flag when disolving COMDAT_GROUP.
6670 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6672         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
6673         * ipa-prop.c (ipa_set_jf_known_type): Return early when
6674         not devirtualizing.
6675         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
6676         do more sanity checks.
6677         (detect_type_change): Return true when giving up early.
6678         (compute_complex_assign_jump_func): Fix type parameter of
6679         ipa_set_ancestor_jf.
6680         (compute_complex_ancestor_jump_func): Likewise.
6681         (update_jump_functions_after_inlining): Fix updating of
6682         ancestor function.
6683         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
6685 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
6687         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
6688         inline clones when edge disappears.
6690 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
6692         PR target/60203
6693         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
6694         Split 64-bit moves into 2 patterns.  Do not allow the use of
6695         direct move for TDmode in little endian, since the decimal value
6696         has little endian bytes within a word, but the 64-bit pieces are
6697         ordered in a big endian fashion, and normal subreg's of TDmode are
6698         not allowed.
6699         (mov<mode>_64bit_dm): Likewise.
6700         (movtd_64bit_nodm): Likewise.
6702 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
6704         PR tree-optimization/60174
6705         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
6706         statement of an SSA_NAME that occurs in an abnormal PHI node.
6708 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
6710         PR sanitizer/60142
6711         * final.c (SEEN_BB): Remove.
6712         (SEEN_NOTE, SEEN_EMITTED): Renumber.
6713         (final_scan_insn): Don't force_source_line on second
6714         NOTE_INSN_BASIC_BLOCK.
6716 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
6718         PR target/60205
6719         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
6720         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
6721         (type_natural_mode): Warn ABI change when %zmm register is not
6722         available for AVX512F vector value passing.
6724 2014-02-18  Kai Tietz  <ktietz@redhat.com>
6726         PR target/60193
6727         * config/i386/i386.c (ix86_expand_prologue): Use value in
6728         rax register as displacement when restoring %r10 or %rax.
6729         Fix wrong offset when restoring both registers.
6731 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
6733         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
6734         assertion with conditional return.
6736 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
6737             Uros Bizjak  <ubizjak@gmail.com>
6739         PR driver/60233
6740         * config/i386/driver-i386.c (host_detect_local_cpu): If
6741         YMM state is not saved by the OS, also clear has_f16c.  Move
6742         CPUID 0x80000001 handling before YMM state saving checking.
6744 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
6746         PR rtl-optimization/58960
6747         * haifa-sched.c (alloc_global_sched_pressure_data): New,
6748         factored out from ...
6749         (sched_init): ... here.
6750         (free_global_sched_pressure_data): New, factored out from ...
6751         (sched_finish): ... here.
6752         * sched-int.h (free_global_sched_pressure_data): Declare.
6753         * sched-rgn.c (nr_regions_initial): New static global.
6754         (haifa_find_rgns): Initialize it.
6755         (schedule_region): Disable sched-pressure for the newly
6756         generated regions.
6758 2014-02-17  Richard Biener  <rguenther@suse.de>
6760         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
6761         release SSA defs of pattern stmts.
6763 2014-02-17  Richard Biener  <rguenther@suse.de>
6765         * tree-inline.c (expand_call_inline): Release the virtual
6766         operand defined by the call we are about to inline.
6768 2014-02-17  Richard Biener  <rguenther@suse.de>
6770         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
6772 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
6773             Ilya Tocar  <ilya.tocar@intel.com>
6775         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
6776         arguments order in builtin.
6777         (_mm512_permutexvar_epi64): Ditto.
6778         (_mm512_mask_permutexvar_epi64): Ditto
6779         (_mm512_maskz_permutexvar_epi32): Ditto
6780         (_mm512_permutexvar_epi32): Ditto
6781         (_mm512_mask_permutexvar_epi32): Ditto
6783 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6785         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
6786         (p8_vmrgow): Likewise.
6788 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6790         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
6791         endian targets.
6793 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
6795         PR target/60203
6796         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
6797         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
6798         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
6799         using direct move instructions on ISA 2.07.  Also adjust
6800         instruction length for 64-bit.
6801         (mov<mode>_64bit, TFmode/TDmode): Likewise.
6802         (mov<mode>_32bit, TFmode/TDmode): Likewise.
6804 2014-02-15  Alan Modra  <amodra@gmail.com>
6806         PR target/58675
6807         PR target/57935
6808         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
6809         find_replacement on parts of insn rtl that might be reloaded.
6811 2014-02-15  Richard Biener  <rguenther@suse.de>
6813         PR tree-optimization/60183
6814         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
6815         (tree_ssa_phiprop): Calculate and free post-dominators.
6817 2014-02-14  Jeff Law  <law@redhat.com>
6819         PR rtl-optimization/60131
6820         * ree.c (get_extended_src_reg): New function.
6821         (combine_reaching_defs): Use it rather than assuming location of REG.
6822         (find_and_remove_re): Verify first operand of extension is
6823         a REG before adding the insns to the copy list.
6825 2014-02-14  Roland McGrath  <mcgrathr@google.com>
6827         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
6828         * configure: Regenerated.
6829         * config.in: Regenerated.
6830         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
6831         instead of ASM_SHORT.
6833 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
6834             Richard Earnshaw  <rearnsha@arm.com>
6836         PR rtl-optimization/59535
6837         * lra-constraints.c (process_alt_operands): Encourage alternative
6838         when unassigned pseudo class is superset of the alternative class.
6839         (inherit_reload_reg): Don't inherit when optimizing for code size.
6840         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
6841         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
6842         modes not less than 4 for Thumb1.
6844 2014-02-14  Kyle McMartin  <kyle@redhat.com>
6846         PR pch/60010
6847         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
6849 2014-02-14  Richard Biener  <rguenther@suse.de>
6851         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
6852         (get_frame_arg): Drop the assert with langhook types_compatible_p.
6853         Do not strip INDIRECT_REFs.
6855 2014-02-14  Richard Biener  <rguenther@suse.de>
6857         PR lto/60179
6858         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
6859         DECL_FUNCTION_SPECIFIC_TARGET.
6860         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
6861         * tree-streamer-out.c (pack_ts_target_option): Remove.
6862         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
6863         (write_ts_function_decl_tree_pointers): Do not stream
6864         DECL_FUNCTION_SPECIFIC_TARGET.
6865         * tree-streamer-in.c (unpack_ts_target_option): Remove.
6866         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
6867         (lto_input_ts_function_decl_tree_pointers): Do not stream
6868         DECL_FUNCTION_SPECIFIC_TARGET.
6870 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
6872         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
6873         (get_initial_def_for_induction, vectorizable_induction): Ignore
6874         debug stmts when looking for exit_phi.
6875         (vectorizable_live_operation): Fix up condition.
6877 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
6879         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
6880         nreverse() because it changes the content of original tree list.
6882 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
6884         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
6885         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
6887 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
6889         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
6890         GNU coding standards.
6892 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
6894         PR debug/60152
6895         * dwarf2out.c (gen_subprogram_die): Don't call
6896         add_calling_convention_attribute if subr_die is old_die.
6898 2014-02-13  Sharad Singhai  <singhai@google.com>
6900         * doc/optinfo.texi: Fix order of nodes.
6902 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
6904         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
6905         operands[2], not operands[3].
6907 2014-02-13  Richard Biener  <rguenther@suse.de>
6909         PR bootstrap/59878
6910         * doc/install.texi (ISL): Update recommended version to 0.12.2,
6911         mention the possibility of an in-tree build.
6912         (CLooG): Update recommended version to 0.18.1, mention the
6913         possibility of an in-tree build and clarify that the ISL
6914         bundled with CLooG does not work.
6916 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
6918         PR target/43546
6919         * expr.c (compress_float_constant): If x is a hard register,
6920         extend into a pseudo and then move to x.
6922 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6924         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
6925         caused by bad second argument to warning_at() with -mhotpatch and
6926         nested functions (e.g. with gfortran).
6928 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
6930         * opts.c (option_name): Remove "enabled by default" rider.
6932 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
6934         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
6936 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
6937             Uros Bizjak  <ubizjak@gmail.com>
6939         PR target/60151
6940         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
6941         * configure: Regenerated.
6943 2014-02-12  Richard Biener  <rguenther@suse.de>
6945         * vec.c (vec_prefix::calculate_allocation): Move as
6946         inline variant to vec.h.
6947         (vec_prefix::calculate_allocation_1): New out-of-line version.
6948         * vec.h (vec_prefix::calculate_allocation_1): Declare.
6949         (vec_prefix::m_has_auto_buf): Rename to ...
6950         (vec_prefix::m_using_auto_storage): ... this.
6951         (vec_prefix::calculate_allocation): Inline the easy cases
6952         and dispatch to calculate_allocation_1 which doesn't need the
6953         prefix address.
6954         (va_heap::reserve): Use gcc_checking_assert.
6955         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
6956         m_using_auto_storage.
6957         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
6958         member and adjust.
6959         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
6960         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
6961         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
6963 2014-02-12  Richard Biener  <rguenther@suse.de>
6965         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
6966         when we found a dependence.
6968 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
6970         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
6971         common code...
6972         (maybe_fold_stmt): ... into this new function.
6973         * omp-low.c (lower_omp): Update comment.
6975         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
6976         last use.
6978         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
6979         dereference.
6981 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
6983         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
6984         identifiers in comments.
6985         (cortexa53_extra_costs): Likewise.
6986         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
6987         (cortexa7_extra_costs): Likewise.
6988         (cortexa12_extra_costs): Likewise.
6989         (cortexa15_extra_costs): Likewise.
6990         (v7m_extra_costs): Likewise.
6992 2014-02-12  Richard Biener  <rguenther@suse.de>
6994         PR middle-end/60092
6995         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
6996         of posix_memalign being successful.
6997         (lower_stmt): Restrict lowering of posix_memalign to when
6998         -ftree-bit-ccp is enabled.
7000 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
7002         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
7003         arg_loc.
7004         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
7006 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
7008         PR rtl-optimization/60116
7009         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
7010         other_insn once the combination has been validated.
7012 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
7014         PR lto/59468
7015         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
7016         and wrapper.
7017         * ipa-devirt.c: Include demangle.h
7018         (odr_violation_reported): New static variable.
7019         (add_type_duplicate): Update odr_violations.
7020         (maybe_record_node): Add completep parameter; update it.
7021         (record_target_from_binfo): Add COMPLETEP parameter;
7022         update it as needed.
7023         (possible_polymorphic_call_targets_1): Likewise.
7024         (struct polymorphic_call_target_d): Add nonconstruction_targets;
7025         rename FINAL to COMPLETE.
7026         (record_targets_from_bases): Sanity check we found the binfo;
7027         fix COMPLETEP updating.
7028         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
7029         parameter, fix computing of COMPLETEP.
7030         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
7031         at LTO time do demangling.
7032         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
7033         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
7034         parameter.
7035         (gimple_get_virt_method_for_binfo): Likewise.
7036         * gimple-fold.h (gimple_get_virt_method_for_binfo,
7037         gimple_get_virt_method_for_vtable): Update prototypes.
7039 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
7041         PR target/49008
7042         * genautomata.c (add_presence_absence): Fix typo with
7043         {final_}presence_list.
7045 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7047         PR target/60137
7048         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
7049         for VSX/Altivec vectors that land in GPR registers.
7051 2014-02-11  Richard Henderson  <rth@redhat.com>
7052             Jakub Jelinek  <jakub@redhat.com>
7054         PR debug/59776
7055         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
7056         around drhs if type conversion to lacc->type is not useless.
7058 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7060         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
7061         tuning struct.
7062         (cortex-a57.cortex-a53): Likewise.
7063         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
7065 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7067         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
7068         arm_restrict_it.
7070 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
7072         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
7073         add_options_for_arm_vfp3.
7075 2014-02-11  Jeff Law  <law@redhat.com>
7077         PR middle-end/54041
7078         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
7079         object with an undesirable mode.
7081 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7083         PR libgomp/60107
7084         * config/i386/sol2-9.h: New file.
7085         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
7086         *-*-solaris2.9*): Use it.
7088 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7090         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
7091         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
7093 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
7095         * config/microblaze/microblaze.c: Extend mcpu version format
7097 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
7099         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
7101 2014-02-10  Richard Henderson  <rth@redhat.com>
7103         PR target/59927
7104         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
7105         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
7106         ms-abi vs -mno-accumulate-outgoing-args.
7107         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
7108         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
7109         respect to ms-abi.
7111 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7113         PR middle-end/60080
7114         * cfgexpand.c (expand_asm_operands): Attach source location to
7115         ASM_INPUT rtx objects.
7116         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
7118 2014-02-10  Nick Clifton  <nickc@redhat.com>
7120         * config/mn10300/mn10300.c (popcount): New function.
7121         (mn10300_expand_prologue): Include saved registers in stack usage
7122         count.
7124 2014-02-10  Jeff Law  <law@redhat.com>
7126         PR middle-end/52306
7127         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
7128         when changing the SET_DEST of a prior insn to avoid an input reload.
7130 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7132         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
7133         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
7134         -mcall-openbsd, or -mcall-linux.
7135         (CC1_ENDIAN_BIG_SPEC): Remove.
7136         (CC1_ENDIAN_LITTLE_SPEC): Remove.
7137         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7138         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
7139         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
7140         and %cc1_endian_default.
7141         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
7143 2014-02-10  Richard Biener  <rguenther@suse.de>
7145         PR tree-optimization/60115
7146         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
7147         MEM_REF handling.  Properly verify that the accesses are not
7148         out of the objects bound.
7150 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7152         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
7153         coretex to cortex.
7155 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
7157         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
7158         proper constants and fix formatting.
7159         (possible_polymorphic_call_targets): Fix formatting.
7161 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
7162             Ilya Tocar  <ilya.tocar@intel.com>
7164         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
7165         (_mm512_loadu_epi32): Renamed into...
7166         (_mm512_loadu_si512): This.
7167         (_mm512_storeu_epi32): Renamed into...
7168         (_mm512_storeu_si512): This.
7169         (_mm512_maskz_ceil_ps): Removed.
7170         (_mm512_maskz_ceil_pd): Ditto.
7171         (_mm512_maskz_floor_ps): Ditto.
7172         (_mm512_maskz_floor_pd): Ditto.
7173         (_mm512_floor_round_ps): Ditto.
7174         (_mm512_floor_round_pd): Ditto.
7175         (_mm512_ceil_round_ps): Ditto.
7176         (_mm512_ceil_round_pd): Ditto.
7177         (_mm512_mask_floor_round_ps): Ditto.
7178         (_mm512_mask_floor_round_pd): Ditto.
7179         (_mm512_mask_ceil_round_ps): Ditto.
7180         (_mm512_mask_ceil_round_pd): Ditto.
7181         (_mm512_maskz_floor_round_ps): Ditto.
7182         (_mm512_maskz_floor_round_pd): Ditto.
7183         (_mm512_maskz_ceil_round_ps): Ditto.
7184         (_mm512_maskz_ceil_round_pd): Ditto.
7185         (_mm512_expand_pd): Ditto.
7186         (_mm512_expand_ps): Ditto.
7187         * config/i386/i386.c (ix86_builtins): Remove
7188         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
7189         (bdesc_args): Ditto.
7190         * config/i386/predicates.md (const1256_operand): New.
7191         (const_1_to_2_operand): Ditto.
7192         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
7193         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
7194         (*avx512pf_gatherpf<mode>sf): Ditto.
7195         (avx512pf_gatherpf<mode>df): Ditto.
7196         (*avx512pf_gatherpf<mode>df_mask): Ditto.
7197         (*avx512pf_gatherpf<mode>df): Ditto.
7198         (avx512pf_scatterpf<mode>sf): Ditto.
7199         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
7200         (*avx512pf_scatterpf<mode>sf): Ditto.
7201         (avx512pf_scatterpf<mode>df): Ditto.
7202         (*avx512pf_scatterpf<mode>df_mask): Ditto.
7203         (*avx512pf_scatterpf<mode>df): Ditto.
7204         (avx512f_expand<mode>): Removed.
7205         (<shift_insn><mode>3<mask_name>): Change predicate type.
7207 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7209         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
7210         not at the end of datarefs vector use ordered_remove to avoid
7211         reordering datarefs vector.
7213         PR c/59984
7214         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
7215         mark local addressable non-static vars as GOVD_PRIVATE
7216         instead of GOVD_LOCAL.
7217         * omp-low.c (lower_omp_for): Move gimple_bind_vars
7218         and BLOCK_VARS of gimple_bind_block to new_stmt rather
7219         than copying them.
7221         PR middle-end/60092
7222         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
7223         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
7224         assume_aligned or alloc_align attributes.
7225         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
7226         arguments.  Handle also assume_aligned and alloc_align attributes.
7227         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
7228         calls to functions with assume_aligned or alloc_align attributes.
7229         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
7231 2014-02-08  Terry Guo  <terry.guo@arm.com>
7233         * doc/invoke.texi: Document ARM -march=armv7e-m.
7235 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
7237         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
7238         flag on __cilkrts_rethrow builtin.
7240         PR ipa/60026
7241         * ipa-cp.c (determine_versionability): Fail at -O0
7242         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
7243         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
7245         Revert:
7246         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
7248         PR ipa/60026
7249         * tree-inline.c (copy_forbidden): Fail for
7250         __attribute__((optimize (0))) functions.
7252 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7254         * varpool.c: Include pointer-set.h.
7255         (varpool_remove_unreferenced_decls): Variables in other partitions
7256         will not be output; be however careful to not lose information
7257         about partitioning.
7259 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
7261         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
7262         lookup in the vtable constructor.
7264 2014-02-07  Jeff Law  <law@redhat.com>
7266         PR target/40977
7267         * config/m68k/m68k.md (ashldi_extsi): Turn into a
7268         define_insn_and_split.
7270         * ipa-inline.c (inline_small_functions): Fix typos.
7272 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7274         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
7275         (s390_can_use_return_insn): Declare.
7276         * config/s390/s390.h (EPILOGUE_USES): Define.
7277         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
7278         instructions.
7279         (s390_chunkify_start): Handle return JUMP_LABELs.
7280         (s390_early_mach): Emit a main_pool instruction on the entry edge.
7281         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
7282         (s390_can_use_return_insn): New functions.
7283         (s390_fix_long_loop_prediction): Handle conditional returns.
7284         (TARGET_SET_UP_BY_PROLOGUE): Define.
7285         * config/s390/s390.md (ANY_RETURN): New code iterator.
7286         (*creturn, *csimple_return, return, simple_return): New patterns.
7288 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7290         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
7291         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
7292         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
7293         REG_CFA_RESTORE list when deciding not to restore a register.
7295 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7297         * config/s390/s390.c: Include tree-pass.h and context.h.
7298         (s390_early_mach): New function, split out from...
7299         (s390_emit_prologue): ...here.
7300         (pass_data_s390_early_mach): New pass structure.
7301         (pass_s390_early_mach): New class.
7302         (s390_option_override): Create and register early_mach pass.
7303         Move to end of file.
7305 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7307         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
7308         to match for the exit block.
7310 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7312         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
7313         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
7314         Reject misaligned operands.
7316 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7318         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
7320 2014-02-07  Richard Biener  <rguenther@suse.de>
7322         PR middle-end/60092
7323         * gimple-low.c (lower_builtin_posix_memalign): New function.
7324         (lower_stmt): Call it to lower posix_memalign in a way
7325         to make alignment info accessible.
7327 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
7329         PR c++/60082
7330         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
7331         __builtin_setjmp_receiver.
7333 2014-02-07  Richard Biener  <rguenther@suse.de>
7335         PR middle-end/60092
7336         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
7337         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
7338         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
7339         Handle BUILT_IN_POSIX_MEMALIGN.
7340         (find_func_clobbers): Likewise.
7341         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7342         (call_may_clobber_ref_p_1): Likewise.
7344 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7346         PR ipa/59918
7347         * ipa-devirt.c (record_target_from_binfo): Remove overactive
7348         sanity check.
7350 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7352         PR ipa/59469
7353         * lto-cgraph.c (lto_output_node): Use
7354         symtab_get_symbol_partitioning_class.
7355         (lto_output_varpool_node): likewise.
7356         (symtab_get_symbol_partitioning_class): Move here from
7357         lto/lto-partition.c
7358         * cgraph.h (symbol_partitioning_class): Likewise.
7359         (symtab_get_symbol_partitioning_class): Declare.
7361 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7363         * ggc.h (ggc_internal_cleared_alloc): New macro.
7364         * vec.h (vec_safe_copy): Handle memory stats.
7365         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
7366         * target-globals.c (save_target_globals): Likewise.
7368 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
7370         PR target/60077
7371         * expr.c (emit_move_resolve_push): Export; be bit more selective
7372         on when to clear alias set.
7373         * expr.h (emit_move_resolve_push): Declare.
7374         * function.h (struct function): Add tail_call_marked.
7375         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
7376         * config/i386/i386-protos.h (ix86_expand_push): Remove.
7377         * config/i386/i386.md (TImode move expander): De not call
7378         ix86_expand_push.
7379         (FP push expanders): Preserve memory attributes.
7380         * config/i386/sse.md (push<mode>1): Remove.
7381         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
7382         (ix86_expand_push): Remove.
7383         * config/i386/mmx.md (push<mode>1): Remove.
7385 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
7387         PR rtl-optimization/60030
7388         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
7389         lopart with paradoxical subreg before shifting it up by hprec.
7391 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7393         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
7394         Remove extra newline at end of file.
7395         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
7396         (arm_issue_rate): Handle cortexa57.
7397         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
7398         (cortex-a57.cortex-a53): Likewise.
7400 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
7402         PR target/59575
7403         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
7404         don't record in REG_FRAME_RELATED_EXPR registers not set in that
7405         bitmask.
7406         (arm_expand_prologue): Adjust all callers.
7407         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
7408         info, registers also at the lowest numbered registers side.  Use
7409         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
7410         XEXP.
7412         PR debug/59992
7413         * var-tracking.c (adjust_mems): Before adding a SET to
7414         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
7416 2014-02-06  Alan Modra  <amodra@gmail.com>
7418         PR target/60032
7419         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
7420         change SDmode to DDmode when lra_in_progress.
7422 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
7424         PR middle-end/59150
7425         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
7426         free_data_ref on the dr first, and before goto again also set dr
7427         to the next dr.  For simd_lane_access, free old datarefs[i] before
7428         overwriting it.  For get_vectype_for_scalar_type failure, don't
7429         free_data_ref if simd_lane_access.
7431         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
7433         PR target/60062
7434         * tree.h (opts_for_fn): New inline function.
7435         (opt_for_fn): Define.
7436         * config/i386/i386.c (ix86_function_regparm): Use
7437         opt_for_fn (decl, optimize) instead of optimize.
7439 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7441         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
7442         for SYMBOL_REF in large memory model.
7444 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7446         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
7447         and crypto support.
7448         (cortex-a57): Likewise.
7449         (cortex-a57.cortex-a53): Likewise.
7451 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
7452             Kugan Vivekanandarajah  <kuganv@linaro.org>
7454         * config/arm/arm.c (arm_vector_alignment_reachable): Check
7455         unaligned_access.
7456         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
7458 2014-02-06  Richard Biener  <rguenther@suse.de>
7460         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
7461         set_loop_copy and initialize_original_copy_tables.
7463 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
7465         * config/aarch64/aarch64-simd.md
7466         (aarch64_ashr_simddi): Change QI to SI.
7468 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
7469             Jakub Jelinek  <jakub@redhat.com>
7471         PR middle-end/60013
7472         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
7473         of the dataflow.
7475 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7477         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
7478         CODE_FOR_altivec_vpku[hw]um to
7479         CODE_FOR_altivec_vpku[hw]um_direct.
7480         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
7481         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
7482         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
7483         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
7485 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7487         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
7488         generation for -maltivec=be.
7489         (altivec_vsumsws): Simplify redundant test.
7491 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7493         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
7494         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
7495         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
7496         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
7497         gen_altivec_vpkuwum.
7498         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
7499         BYTES_BIG_ENDIAN.
7500         (altivec_vpks<VI_char>ss): Likewise.
7501         (altivec_vpks<VI_char>us): Likewise.
7502         (altivec_vpku<VI_char>us): Likewise.
7503         (altivec_vpku<VI_char>um): Likewise.
7504         (altivec_vpku<VI_char>um_direct): New (copy of
7505         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
7506         internal use).
7507         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
7508         target is little endian and -maltivec=be is not specified.
7509         (*altivec_vupkhs<VU_char>_direct): New (copy of
7510         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
7511         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
7512         target is little endian and -maltivec=be is not specified.
7513         (*altivec_vupkls<VU_char>_direct): New (copy of
7514         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
7515         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
7516         little endian and -maltivec=be is not specified.
7517         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
7518         little endian and -maltivec=be is not specified.
7520 2014-02-05  Richard Henderson  <rth@redhat.com>
7522         PR debug/52727
7523         * combine-stack-adj.c: Revert r206943.
7524         * sched-int.h (struct deps_desc): Add last_args_size.
7525         * sched-deps.c (init_deps): Initialize it.
7526         (sched_analyze_insn): Add OUTPUT dependencies between insns that
7527         contain REG_ARGS_SIZE notes.
7529 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
7531         * lto-cgraph.c (asm_nodes_output): Make global.
7532         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
7533         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
7534         (driver_handle_option): Handle OPT_fwpa.
7536 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
7538         PR ipa/59947
7539         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
7540         a comment typo and formatting issue.  If odr_hash hasn't been
7541         created, return vNULL and set *completep to false.
7543         PR middle-end/57499
7544         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
7545         bb with no successors.
7547 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
7549         PR target/59718
7550         * doc/invoke.texi (-march): Clarify documentation for ARM.
7551         (-mtune): Likewise.
7552         (-mcpu): Likewise.
7554 2014-02-05  Richard Biener  <rguenther@suse.de>
7556         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
7557         when not vectorizing because of too many alias checks.
7558         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
7559         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
7561 2014-02-05  Nick Clifton  <nickc@redhat.com>
7563         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
7564         accept extended registers in any mode when compiling for the MN10300.
7566 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
7568         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
7569         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
7570         sanitization attributes.
7571         (can_inline_edge_p): Likewise.
7572         (sanitize_attrs_match_for_inline_p): New function.
7574 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7576         * ipa-prop.c (detect_type_change): Shor circuit testing of
7577         type changes on THIS pointer.
7579 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
7581         PR target/59777
7582         * config/pa/pa.c (legitimize_tls_address): Return original address
7583         if not passed a SYMBOL_REF rtx.
7584         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
7585         addresses.
7586         (pa_emit_move_sequence): Simplify TLS source operands.
7587         (pa_legitimate_constant_p): Reject all TLS constants.
7588         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
7589         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
7591 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7593         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
7594         groups when we know they are controlled by LTO.
7595         * varasm.c (default_binds_local_p_1): If object is in other partition,
7596         it will be resolved locally.
7598 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7600         * config/host-linux.c (linux_gt_pch_use_address): Don't
7601         use SSIZE_MAX because it is not always defined.
7603 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
7605         PR bootstrap/59913
7606         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
7607         threshold for pseudo splitting.
7608         (update_ebb_live_info): Process call argument hard registers and
7609         hard registers from insn definition too.
7610         (max_small_class_regs_num): New constant.
7611         (inherit_in_ebb): Update live hard regs through EBBs.  Update
7612         reloads_num only for small register classes.  Don't split for
7613         outputs of jumps.
7615 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
7617         PR ipa/60058
7618         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
7619         is non-null.
7621 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
7623         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
7624         visibility is safe.
7626 2014-02-04  Marek Polacek  <polacek@redhat.com>
7628         * gdbinit.in (pel): Define.
7630 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7632         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
7633         behavior.
7635 2014-02-04  Richard Biener  <rguenther@suse.de>
7637         PR lto/59723
7638         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
7639         in function context local.
7640         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
7641         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
7642         similar to LTO_imported_decl_ref.
7644 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
7646         PR tree-optimization/60002
7647         * cgraphclones.c (build_function_decl_skip_args): Clear
7648         DECL_LANG_SPECIFIC.
7650         PR tree-optimization/60023
7651         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
7652         false to gsi_replace.
7653         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
7654         has been in some EH region and vec_stmt could throw, add
7655         vec_stmt into the same EH region.
7656         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
7657         has no lhs, ignore it.
7658         * internal-fn.c (expand_MASK_LOAD): Likewise.
7660         PR ipa/60026
7661         * tree-inline.c (copy_forbidden): Fail for
7662         __attribute__((optimize (0))) functions.
7664         PR other/58712
7665         * omp-low.c (simd_clone_struct_copy): If from->inbranch
7666         is set, copy one less argument.
7667         (expand_simd_clones): Don't subtract clone_info->inbranch
7668         from simd_clone_struct_alloc argument.
7670         PR rtl-optimization/57915
7671         * recog.c (simplify_while_replacing): If all unary/binary/relational
7672         operation arguments are constant, attempt to simplify those.
7674         PR middle-end/59261
7675         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
7676         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
7678 2014-02-04  Richard Biener  <rguenther@suse.de>
7680         PR tree-optimization/60012
7681         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
7682         TBAA disambiguation to all DDRs.
7684 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7686         PR target/59788
7687         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
7688         (LINK_SPEC): Use it for -shared, -shared-libgcc.
7690 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7692         PR ipa/59882
7693         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
7695 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7697         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
7698         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
7700 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7702         PR ipa/59831
7703         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
7704         to figure out targets of polymorphic calls with known decl.
7705         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7706         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
7707         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
7708         (get_polymorphic_call_info): ... here.
7709         (get_polymorphic_call_info_from_invariant): New function.
7711 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7713         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
7714         lookup via vtable pointer; check for type consistency
7715         and turn inconsitent facts into UNREACHABLE.
7716         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7717         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
7718         type inconsistent querries; return UNREACHABLE instead.
7720 2014-02-03  Richard Henderson  <rth@twiddle.net>
7722         PR tree-opt/59924
7723         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
7724         already processed this node.
7725         (normalize_one_pred_1): Pass along mark_set.
7726         (normalize_one_pred): Create and destroy a pointer_set_t.
7727         (normalize_one_pred_chain): Likewise.
7729 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
7731         PR gcov-profile/58602
7732         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
7734 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
7736         PR ipa/59831
7737         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
7738         -fno-devirtualize; try to devirtualize by the knowledge of
7739         virtual table pointer given by aggregate propagation.
7740         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
7741         (ipa_print_node_jump_functions): Dump also offset that
7742         is relevant for polymorphic calls.
7743         (determine_known_aggregate_parts): Add arg_type parameter; use it
7744         instead of determining the type from pointer type.
7745         (ipa_compute_jump_functions_for_edge): Update call of
7746         determine_known_aggregate_parts.
7747         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
7748         (gimple_get_virt_method_for_binfo): ... here; simplify using
7749         vtable_pointer_value_to_vtable.
7750         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
7751         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
7752         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
7753         (vtable_pointer_value_to_vtable): Break out from ...; handle also
7754         POINTER_PLUS_EXPR.
7755         (vtable_pointer_value_to_binfo): ... here.
7756         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
7758 2014-02-03  Teresa Johnson  <tejohnson@google.com>
7760         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
7761         redef of outer loop index variable.
7763 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
7765         PR c++/53017
7766         PR c++/59211
7767         * doc/extend.texi (Function Attributes): Typo.
7769 2014-02-03  Cong Hou  <congh@google.com>
7771         PR tree-optimization/60000
7772         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
7773         if the vectorized statement is a store.  A store statement can only
7774         appear at the end of pattern statements.
7776 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
7778         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
7779         (ix86_option_override_internal): Default long double to 64-bit for
7780         32-bit Bionic and to 128-bit for 64-bit Bionic.
7782         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
7783         TARGET_LONG_DOUBLE_128 is true.
7784         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
7786         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
7787         (mlong-double-64): Negate -mlong-double-128.
7788         (mlong-double-128): New option.
7790         * config/i386/i386-c.c (ix86_target_macros): Define
7791         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
7793         * doc/invoke.texi: Document -mlong-double-128.
7795 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
7797         PR rtl-optimization/60024
7798         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
7800 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
7802         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
7804 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
7806         PR rtl-optimization/57662
7807         * sel-sched.c (code_motion_path_driver): Do not mark already not
7808         existing blocks in the visiting bitmap.
7810 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
7812         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
7813         on the insn being emitted.
7815 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
7816             Will Deacon  <will.deacon@arm.com>
7818         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
7820 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7822         * config/arm/arm-tables.opt: Regenerate.
7824 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7826         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
7827         for vector types other than V16QImode.
7828         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
7829         define_expand, and call altivec_expand_vec_perm_le when producing
7830         code with little endian element order.
7831         (*altivec_vperm_<mode>_internal): New insn having previous
7832         behavior of altivec_vperm_<mode>.
7833         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
7834         altivec_expand_vec_perm_le when producing code with little endian
7835         element order.
7836         (*altivec_vperm_<mode>_uns_internal): New insn having previous
7837         behavior of altivec_vperm_<mode>_uns.
7839 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7841         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
7842         (altivec_vsumsws): Add handling for -maltivec=be with a little
7843         endian target.
7844         (altivec_vsumsws_direct): New.
7845         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
7846         gen_altivec_vsumsws.
7848 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
7850         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
7851         vtable_pointer_value_to_binfo): New functions.
7852         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
7853         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
7855 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
7857         * config/nios2/nios2.md (load_got_register): Initialize GOT
7858         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
7859         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
7861 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
7863         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
7864         preserverd by passthrough, do not propagate the type.
7866 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
7868         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
7869         (mips_atomic_assign_expand_fenv): New function.
7870         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
7872 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
7874         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
7875         (__builtin_mips_set_fcsr): Likewise.
7876         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
7877         MIPS_USI_FTYPE_VOID.
7878         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
7879         (mips16_expand_set_fcsr): Likewise.
7880         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
7881         (mips16_set_fcsr_stub): Likewise.
7882         (mips16_get_fcsr_one_only_stub): New class.
7883         (mips16_set_fcsr_one_only_stub): Likewise.
7884         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
7885         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
7886         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
7887         (hard_float): New availability predicate.
7888         (mips_builtins): Add get_fcsr and set_fcsr.
7889         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
7890         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
7891         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
7892         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
7893         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
7894         patterns.
7896 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
7898         * config/mips/mips.c (mips_one_only_stub): New class.
7899         (mips_need_mips16_rdhwr_p): Replace with...
7900         (mips16_rdhwr_stub): ...this new variable.
7901         (mips16_stub_call_address): New function.
7902         (mips16_rdhwr_one_only_stub): New class.
7903         (mips_expand_thread_pointer): Use mips16_stub_call_address.
7904         (mips_output_mips16_rdhwr): Delete.
7905         (mips_finish_stub): New function.
7906         (mips_code_end): Use it to handle rdhwr stubs.
7908 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
7910         PR target/60017
7911         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
7912         when calculating size of integer atomic types.
7914 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
7916         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
7918 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
7920         PR tree-optimization/60003
7921         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
7922         * profile.c (branch_prob): Use gimple_call_builtin_p
7923         to check for BUILT_IN_SETJMP_RECEIVER.
7924         * tree-inline.c (copy_bb): Call notice_special_calls.
7926 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
7928         PR bootstrap/59985
7929         * lra-constraints.c (process_alt_operands): Update reload_sum only
7930         on the first pass.
7932 2014-01-31  Richard Henderson  <rth@redhat.com>
7934         PR middle-end/60004
7935         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
7936         until after else_eh is processed.
7938 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
7940         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
7941         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
7942         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
7943         in smmintrin.h, remove them.
7944         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
7945         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
7946         * config/i386/i386.md (ROUND_SAE): Fix value.
7947         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
7948         (const48_operand): New.
7949         * config/i386/subst.md (round), (round_expand): Use
7950         const_4_or_8_to_11_operand.
7951         (round_saeonly), (round_saeonly_expand): Use const48_operand.
7953 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
7955         * config/i386/constraints.md (Yk): Swap meaning with k.
7956         * config/i386/i386.md (movhi_internal): Change Yk to k.
7957         (movqi_internal): Ditto.
7958         (*k<logic><mode>): Ditto.
7959         (*andhi_1): Ditto.
7960         (*andqi_1): Ditto.
7961         (kandn<mode>): Ditto.
7962         (*<code>hi_1): Ditto.
7963         (*<code>qi_1): Ditto.
7964         (kxnor<mode>): Ditto.
7965         (kortestzhi): Ditto.
7966         (kortestchi): Ditto.
7967         (kunpckhi): Ditto.
7968         (*one_cmplhi2_1): Ditto.
7969         (*one_cmplqi2_1): Ditto.
7970         * config/i386/sse.md (): Change k to Yk.
7971         (avx512f_load<mode>_mask): Ditto.
7972         (avx512f_blendm<mode>): Ditto.
7973         (avx512f_store<mode>_mask): Ditto.
7974         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
7975         (avx512f_storedqu<mode>_mask): Ditto.
7976         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
7977         Ditto.
7978         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
7979         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
7980         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
7981         (avx512f_maskcmp<mode>3): Ditto.
7982         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
7983         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
7984         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
7985         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
7986         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
7987         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
7988         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
7989         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
7990         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
7991         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
7992         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
7993         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
7994         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
7995         (vec_extract_lo_<mode>_maskm): Ditto.
7996         (vec_extract_hi_<mode>_maskm): Ditto.
7997         (avx512f_vternlog<mode>_mask): Ditto.
7998         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
7999         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
8000         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
8001         (avx512f_<code>v8div16qi2_mask): Ditto.
8002         (avx512f_<code>v8div16qi2_mask_store): Ditto.
8003         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
8004         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
8005         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
8006         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
8007         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
8008         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8009         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
8010         (*avx512pf_scatterpf<mode>df_mask): Ditto.
8011         (avx512cd_maskb_vec_dupv8di): Ditto.
8012         (avx512cd_maskw_vec_dupv16si): Ditto.
8013         (avx512f_vpermi2var<mode>3_maskz): Ditto.
8014         (avx512f_vpermi2var<mode>3_mask): Ditto.
8015         (avx512f_vpermi2var<mode>3_mask): Ditto.
8016         (avx512f_vpermt2var<mode>3_maskz): Ditto.
8017         (*avx512f_gathersi<mode>): Ditto.
8018         (*avx512f_gathersi<mode>_2): Ditto.
8019         (*avx512f_gatherdi<mode>): Ditto.
8020         (*avx512f_gatherdi<mode>_2): Ditto.
8021         (*avx512f_scattersi<mode>): Ditto.
8022         (*avx512f_scatterdi<mode>): Ditto.
8023         (avx512f_compress<mode>_mask): Ditto.
8024         (avx512f_compressstore<mode>_mask): Ditto.
8025         (avx512f_expand<mode>_mask): Ditto.
8026         * config/i386/subst.md (mask): Change k to Yk.
8027         (mask_scalar_merge): Ditto.
8028         (sd): Ditto.
8030 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
8032         * doc/extend.texi (Vector Extensions): Document ?: in C++.
8034 2014-01-31  Richard Biener  <rguenther@suse.de>
8036         PR middle-end/59990
8037         * builtins.c (fold_builtin_memory_op): Make sure to not
8038         use a floating-point mode or a boolean or enumeral type for
8039         the copy operation.
8041 2014-01-30  DJ Delorie  <dj@redhat.com>
8043         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
8044         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
8045         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
8046         whenever main() has an epilogue.
8048 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8050         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
8051         unused variable "field".
8052         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
8053         (vsx_mergeh_<mode>): Likewise.
8054         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
8055         (altivec_vmrghh): Likewise.
8056         (altivec_vmrghw): Likewise.
8057         (altivec_vmrglb): Likewise.
8058         (altivec_vmrglh): Likewise.
8059         (altivec_vmrglw): Likewise.
8060         (altivec_vspltb): Add missing uses.
8061         (altivec_vsplth): Likewise.
8062         (altivec_vspltw): Likewise.
8063         (altivec_vspltsf): Likewise.
8065 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8067         PR target/59923
8068         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
8069         frame related instructions.
8071 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
8073         PR rtl-optimization/59959
8074         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
8075         any reload of register whose subreg is invalid.
8077 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
8079         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
8080         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
8081         Add missing return type - void.
8083 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8085         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
8086         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
8087         remove element index adjustment for endian (now handled in vsx.md
8088         and altivec.md).
8089         (altivec_expand_vec_perm_const): Use
8090         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
8091         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
8092         (vsx_xxspltw_<mode>): Adjust element index for little endian.
8093         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
8094         define_expand and a new define_insn *altivec_vspltb_internal;
8095         adjust for -maltivec=be on a little endian target.
8096         (altivec_vspltb_direct): New.
8097         (altivec_vsplth): Divide into a define_expand and a new
8098         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
8099         little endian target.
8100         (altivec_vsplth_direct): New.
8101         (altivec_vspltw): Divide into a define_expand and a new
8102         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
8103         little endian target.
8104         (altivec_vspltw_direct): New.
8105         (altivec_vspltsf): Divide into a define_expand and a new
8106         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
8107         a little endian target.
8109 2014-01-30  Richard Biener  <rguenther@suse.de>
8111         PR tree-optimization/59993
8112         * tree-ssa-forwprop.c (associate_pointerplus): Check we
8113         can propagate form the earlier stmt and avoid the transform
8114         when the intermediate result is needed.
8116 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
8118         * README.Portability: Fix typo.
8120 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
8122         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
8123         comparison_operator with ordered_comparison_operator.
8125 2014-01-30  Nick Clifton  <nickc@redhat.com>
8127         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
8128         Rename to mn10300_store_multiple_regs.
8129         * config/mn10300/mn10300.c: Likewise.
8130         * config/mn10300/mn10300.md (store_movm): Fix typo: call
8131         store_multiple_regs.
8132         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
8133         Call mn10300_store_multiple_regs.
8135 2014-01-30  Nick Clifton  <nickc@redhat.com>
8136             DJ Delorie  <dj@redhat.com>
8138         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
8139         %fp 2 to keep registers after it properly word-aligned.
8140         (rl78_alloc_physical_registers_umul): Handle the case where both
8141         input operands are the same.
8143 2014-01-30  Richard Biener  <rguenther@suse.de>
8145         PR tree-optimization/59903
8146         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
8147         check properly.
8149 2014-01-30  Jason Merrill  <jason@redhat.com>
8151         PR c++/59633
8152         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
8154         PR c++/59645
8155         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
8157 2014-01-30  Richard Biener  <rguenther@suse.de>
8159         PR tree-optimization/59951
8160         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
8162 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
8164         PR target/59784
8165         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
8166         SFmode to DFmode case.
8168 2014-01-29  DJ Delorie  <dj@redhat.com>
8170         * config/msp430/msp430.opt (-minrt): New.
8171         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
8172         if -minrt given.
8173         (ENDFILE_SPEC): Likewise.
8175 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
8177         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
8178         (estimate_function_body_sizes): Use it.
8180 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
8182         PR c++/58561
8183         * dwarf2out.c (is_cxx_auto): New.
8184         (is_base_type): Use it.
8185         (gen_type_die_with_usage): Likewise.
8187 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8189         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
8190         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
8191         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
8192         -maltivec=be with LE targets.
8193         (vsx_mergeh_<mode>): Likewise.
8194         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
8195         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
8196         (altivec_vmrghb): Replace with define_expand and new
8197         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
8198         (altivec_vmrghb_direct): New define_insn.
8199         (altivec_vmrghh): Replace with define_expand and new
8200         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
8201         (altivec_vmrghh_direct): New define_insn.
8202         (altivec_vmrghw): Replace with define_expand and new
8203         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
8204         (altivec_vmrghw_direct): New define_insn.
8205         (*altivec_vmrghsf): Adjust for endianness.
8206         (altivec_vmrglb): Replace with define_expand and new
8207         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
8208         (altivec_vmrglb_direct): New define_insn.
8209         (altivec_vmrglh): Replace with define_expand and new
8210         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
8211         (altivec_vmrglh_direct): New define_insn.
8212         (altivec_vmrglw): Replace with define_expand and new
8213         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
8214         (altivec_vmrglw_direct): New define_insn.
8215         (*altivec_vmrglsf): Adjust for endianness.
8216         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8217         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8218         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
8219         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
8220         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8221         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8222         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
8223         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
8225 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8227         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
8228         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
8229         whitespace.
8231 2014-01-29  Richard Biener  <rguenther@suse.de>
8233         PR tree-optimization/58742
8234         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
8235         associate_pointerplus_align.
8236         (associate_pointerplus_diff): New function.
8237         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
8238         and associate_pointerplus_diff.
8240 2014-01-29  Richard Biener  <rguenther@suse.de>
8242         * lto-streamer.h (LTO_major_version): Bump to 3.
8243         (LTO_minor_version): Reset to 0.
8245 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
8247         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
8248         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
8249         (arm_file_start): Generate correct asm header for armv7ve.
8250         * config/arm/bpabi.h: Add multilib support for armv7ve.
8251         * config/arm/driver-arm.c: Change the architectures of cortex-a7
8252         and cortex-a15 to armv7ve.
8253         * config/arm/t-aprofile: Add multilib support for armv7ve.
8254         * doc/invoke.texi: Document -march=armv7ve.
8256 2014-01-29  Richard Biener  <rguenther@suse.de>
8258         PR tree-optimization/58742
8259         * tree-ssa-forwprop.c (associate_plusminus): Return true
8260         if we changed sth, defer EH cleanup to ...
8261         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
8262         (simplify_mult): New function.
8264 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
8266         PR middle-end/59917
8267         PR tree-optimization/59920
8268         * tree.c (build_common_builtin_nodes): Remove
8269         __builtin_setjmp_dispatcher initialization.
8270         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
8271         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
8272         instead of gsi_after_labels + manually skipping debug stmts.
8273         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
8274         ignore bbs with IFN_ABNORMAL_DISPATCHER.
8275         * tree-inline.c (copy_edges_for_bb): Remove
8276         can_make_abnormal_goto argument, instead add abnormal_goto_dest
8277         argument.  Ignore computed_goto_p stmts.  Don't call
8278         make_abnormal_goto_edges.  If a call might need abnormal edges
8279         for non-local gotos, see if it already has an edge to
8280         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
8281         with true argument, don't do anything then, otherwise add
8282         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
8283         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
8284         caller.
8285         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
8286         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
8287         (lower_stmt): Don't set data->calls_builtin_setjmp.
8288         (lower_builtin_setjmp): Adjust comment.
8289         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
8290         * tree-cfg.c (found_computed_goto): Remove.
8291         (factor_computed_gotos): Remove.
8292         (make_goto_expr_edges): Return bool, true for computed gotos.
8293         Don't call make_abnormal_goto_edges.
8294         (build_gimple_cfg): Don't set found_computed_goto, don't call
8295         factor_computed_gotos.
8296         (computed_goto_p): No longer static.
8297         (make_blocks): Don't set found_computed_goto.
8298         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
8299         (make_edges): If make_goto_expr_edges returns true, push bb
8300         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
8301         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
8302         vector.  Record mapping between bbs and OpenMP regions if there
8303         are any, adjust make_gimple_omp_edges caller.  Call
8304         handle_abnormal_edges.
8305         (make_abnormal_goto_edges): Remove.
8306         * tree-cfg.h (make_abnormal_goto_edges): Remove.
8307         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
8308         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
8309         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
8310         * internal-fn.def (ABNORMAL_DISPATCHER): New.
8311         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
8312         filling *region also set *region_idx to (*region)->entry->index.
8314         PR other/58712
8315         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
8316         For REGs set ORIGINAL_REGNO.
8318 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
8320         * doc/md.texi: Mention that a target shouldn't implement
8321         vec_widen_(s|u)mul_even/odd pair if it is less efficient
8322         than hi/lo pair.
8324 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
8326         PR tree-optimization/59594
8327         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
8328         a copy of the datarefs vector rather than the vector itself.
8330 2014-01-28  Jason Merrill  <jason@redhat.com>
8332         PR c++/53756
8333         * dwarf2out.c (auto_die): New static.
8334         (gen_type_die_with_usage): Handle C++1y 'auto'.
8335         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
8336         on definition.
8338 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
8340         PR target/59672
8341         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
8342         (SPEC_X32): Likewise.
8343         (SPEC_64): Likewise.
8344         * config/i386/i386.c (ix86_option_override_internal): Turn off
8345         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
8346         for TARGET_16BIT.
8347         (x86_file_start): Output .code16gcc for TARGET_16BIT.
8348         * config/i386/i386.h (TARGET_16BIT): New macro.
8349         (TARGET_16BIT_P): Likewise.
8350         * config/i386/i386.opt: Add m16.
8351         * doc/invoke.texi: Document -m16.
8353 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
8355         PR preprocessor/59935
8356         * input.c (location_get_source_line): Bail out on when line number
8357         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
8359 2014-01-28  Richard Biener  <rguenther@suse.de>
8361         PR tree-optimization/58742
8362         * tree-ssa-forwprop.c (associate_plusminus): Handle
8363         pointer subtraction of the form (T)(P + A) - (T)P.
8365 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8367         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
8368         at const_int_cost.
8370 2014-01-28  Richard Biener  <rguenther@suse.de>
8372         Revert
8373         2014-01-28  Richard Biener  <rguenther@suse.de>
8375         PR rtl-optimization/45364
8376         PR rtl-optimization/59890
8377         * var-tracking.c (local_get_addr_clear_given_value): Handle
8378         already cleared slot.
8379         (val_reset): Handle not allocated local_get_addr_cache.
8380         (vt_find_locations): Use post-order on the inverted CFG.
8382 2014-01-28  Richard Biener  <rguenther@suse.de>
8384         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
8386 2014-01-28  Richard Biener  <rguenther@suse.de>
8388         PR rtl-optimization/45364
8389         PR rtl-optimization/59890
8390         * var-tracking.c (local_get_addr_clear_given_value): Handle
8391         already cleared slot.
8392         (val_reset): Handle not allocated local_get_addr_cache.
8393         (vt_find_locations): Use post-order on the inverted CFG.
8395 2014-01-28  Alan Modra  <amodra@gmail.com>
8397         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
8398         * configure.ac <recursive call for build != host>: Define
8399         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
8400         and LD_FOR_BUILD too.
8401         * configure: Regenerate.
8403 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
8405         * config/i386/i386.c (get_builtin_code_for_version): Separate
8406         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
8407         Broadwell from Haswell.
8409 2014-01-27  Steve Ellcey  <sellcey@mips.com>
8411         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
8412         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
8413         * config/mips/mips.c (mips_option_override): Change setting
8414         of TARGET_DSP.
8415         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
8416         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
8417         Change from Mask to Var.
8419 2014-01-27  Jeff Law  <law@redhat.com>
8421         * ipa-inline.c (inline_small_functions): Fix typo.
8423 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
8425         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
8426         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
8427         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
8428         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
8429         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
8430         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
8431         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
8432         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
8433         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
8434         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
8435         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
8436         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
8437         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
8438         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
8439         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
8440         (_mm512_storeu_epi64): Ditto.
8441         (_mm512_cmpge_epi32_mask): Ditto.
8442         (_mm512_cmpge_epu32_mask): Ditto.
8443         (_mm512_cmpge_epi64_mask): Ditto.
8444         (_mm512_cmpge_epu64_mask): Ditto.
8445         (_mm512_cmple_epi32_mask): Ditto.
8446         (_mm512_cmple_epu32_mask): Ditto.
8447         (_mm512_cmple_epi64_mask): Ditto.
8448         (_mm512_cmple_epu64_mask): Ditto.
8449         (_mm512_cmplt_epi32_mask): Ditto.
8450         (_mm512_cmplt_epu32_mask): Ditto.
8451         (_mm512_cmplt_epi64_mask): Ditto.
8452         (_mm512_cmplt_epu64_mask): Ditto.
8453         (_mm512_cmpneq_epi32_mask): Ditto.
8454         (_mm512_cmpneq_epu32_mask): Ditto.
8455         (_mm512_cmpneq_epi64_mask): Ditto.
8456         (_mm512_cmpneq_epu64_mask): Ditto.
8457         (_mm512_expand_pd): Ditto.
8458         (_mm512_expand_ps): Ditto.
8459         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
8460         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
8461         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
8462         * config/i386/i386.c (ix86_builtins): Add
8463         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
8464         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
8465         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
8466         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
8467         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
8468         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
8469         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
8470         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
8471         IX86_BUILTIN_PMOVUSQW512_MEM.
8472         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
8473         __builtin_ia32_pmovsqd512mem_mask,
8474         __builtin_ia32_pmovqd512mem_mask,
8475         __builtin_ia32_pmovusqw512mem_mask,
8476         __builtin_ia32_pmovsqw512mem_mask,
8477         __builtin_ia32_pmovqw512mem_mask,
8478         __builtin_ia32_pmovusdw512mem_mask,
8479         __builtin_ia32_pmovsdw512mem_mask,
8480         __builtin_ia32_pmovdw512mem_mask,
8481         __builtin_ia32_pmovqb512mem_mask,
8482         __builtin_ia32_pmovusqb512mem_mask,
8483         __builtin_ia32_pmovsqb512mem_mask,
8484         __builtin_ia32_pmovusdb512mem_mask,
8485         __builtin_ia32_pmovsdb512mem_mask,
8486         __builtin_ia32_pmovdb512mem_mask.
8487         (bdesc_args): Add __builtin_ia32_expanddf512,
8488         __builtin_ia32_expandsf512.
8489         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
8490         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
8491         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
8492         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
8493         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
8494         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
8495         (avx512f_<code>v8div16qi2_mask_store): This.
8496         (avx512f_expand<mode>): New.
8498 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
8500         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
8501         New.
8502         (_mm512_mask_prefetch_i64gather_pd): Ditto.
8503         (_mm512_prefetch_i32scatter_pd): Ditto.
8504         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
8505         (_mm512_prefetch_i64scatter_pd): Ditto.
8506         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
8507         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
8508         (_mm512_mask_prefetch_i64gather_ps): Ditto.
8509         (_mm512_prefetch_i32scatter_ps): Ditto.
8510         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
8511         (_mm512_prefetch_i64scatter_ps): Ditto.
8512         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
8513         * config/i386/i386-builtin-types.def: Define
8514         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
8515         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
8516         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
8517         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
8518         IX86_BUILTIN_SCATTERPFQPD.
8519         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
8520         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
8521         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
8522         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
8523         __builtin_ia32_scatterpfqps.
8524         (ix86_expand_builtin): Expand new built-ins.
8525         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
8526         fix memory access data type.
8527         (*avx512pf_gatherpf<mode>_mask): Ditto.
8528         (*avx512pf_gatherpf<mode>): Ditto.
8529         (avx512pf_scatterpf<mode>): Ditto.
8530         (*avx512pf_scatterpf<mode>_mask): Ditto.
8531         (*avx512pf_scatterpf<mode>): Ditto.
8532         (GATHER_SCATTER_SF_MEM_MODE): New.
8533         (avx512pf_gatherpf<mode>df): Ditto.
8534         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8535         (*avx512pf_scatterpf<mode>df): Ditto.
8537 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
8539         PR bootstrap/59934
8540         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
8541         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
8542         reached.
8544 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
8546         * common/config/arm/arm-common.c
8547         (arm_rewrite_mcpu): Handle multiple names.
8548         * config/arm/arm.h
8549         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
8551 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
8553         * gimple-builder.h (create_gimple_tmp): Delete.
8555 2014-01-27  Christian Bruel  <christian.bruel@st.com>
8557         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
8558         words comparisons.
8560 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
8562         * config/pa/pa.md (call): Generate indirect long calls to non-local
8563         functions when outputing 32-bit code.
8564         (call_value): Likewise except for special call to buggy powf function.
8566         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
8567         portable runtime and PIC indirect calls.
8568         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
8569         and PIC call sequences.  Use ldo instead of blr to set return register
8570         in PIC call sequence.
8572 2014-01-25  Walter Lee  <walt@tilera.com>
8574         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
8575         avoid clobbering a live register.
8577 2014-01-25  Walter Lee  <walt@tilera.com>
8579         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
8580         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
8581         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
8582         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
8584 2014-01-25  Walter Lee  <walt@tilera.com>
8586         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
8587         arguments on even registers.
8588         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
8589         STACK_BOUNDARY.
8590         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
8591         (BIGGEST_ALIGNMENT): Ditto.
8592         (BIGGEST_FIELD_ALIGNMENT): Ditto.
8594 2014-01-25  Walter Lee  <walt@tilera.com>
8596         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
8597         insns before bundling.
8598         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
8600 2014-01-25  Walter Lee  <walt@tilera.com>
8602         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
8603         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
8604         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
8606 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
8608         * config/mips/constraints.md (kl): Delete.
8609         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
8610         define expands, using...
8611         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
8612         instructions for MIPS16.
8613         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
8614         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
8616 2014-01-25  Walter Lee  <walt@tilera.com>
8618         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
8619         (clzdi2): Ditto.
8620         (ffsdi2): Ditto.
8622 2014-01-25  Walter Lee  <walt@tilera.com>
8624         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
8625         (TARGET_EXPAND_TO_RTL_HOOK): Define.
8627 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
8629         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
8630         Handle XOR.
8632 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
8634         * print-rtl.c (in_call_function_usage): New var.
8635         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
8636         EXPR_LIST mode as mode and not as reg note name.
8638         PR middle-end/59561
8639         * cfgloopmanip.c (copy_loop_info): If
8640         loop->warned_aggressive_loop_optimizations, make sure
8641         the flag is set in target loop too.
8643 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8645         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
8646         flag_cilkplus.
8647         * builtins.def: Likewise.
8648         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
8649         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
8650         * ira.c (ira_setup_eliminable_regset): Likewise.
8651         * omp-low.c (gate_expand_omp): Likewise.
8652         (execute_lower_omp): Likewise.
8653         (diagnose_sb_0): Likewise.
8654         (gate_diagnose_omp_blocks): Likewise.
8655         (simd_clone_clauses_extract): Likewise.
8656         (gate): Likewise.
8658 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8660         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
8661         correction for little endian...
8662         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
8663         here.
8665 2014-01-24  Jeff Law  <law@redhat.com>
8667         PR tree-optimization/59919
8668         * tree-vrp.c (find_assert_locations_1): Do not register asserts
8669         for non-returning calls.
8671 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
8673         * common/config/aarch64/aarch64-common.c
8674         (aarch64_rewrite_mcpu): Handle multiple names.
8675         * config/aarch64/aarch64.h
8676         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
8678 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
8680         * input.c (add_file_to_cache_tab): Handle the case where fopen
8681         returns NULL.
8683 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
8685         PR target/59929
8686         * config/i386/i386.md (pushsf splitter): Get stack adjustment
8687         from push operand if code of push isn't PRE_DEC.
8689 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
8691         PR target/59909
8692         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
8693         -mquad-memory-atomic.  Update -mquad-memory documentation to say
8694         it is only used for non-atomic loads/stores.
8696         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
8697         -mquad-memory or -mquad-memory-atomic switches.
8699         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
8700         -mquad-memory-atomic to ISA 2.07 support.
8702         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
8703         to separate support of normal quad word memory operations (ldq, stq)
8704         from the atomic quad word memory operations.
8706         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
8707         support to separate non-atomic quad word operations from atomic
8708         quad word operations.  Disable non-atomic quad word operations in
8709         little endian mode so that we don't have to swap words after the
8710         load and before the store.
8711         (quad_load_store_p): Add comment about atomic quad word support.
8712         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
8713         options printed with -mdebug=reg.
8715         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
8716         -mquad-memory-atomic as the test for whether we have quad word
8717         atomic instructions.
8718         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
8719         or -mp8-vector are used, allow byte/half-word atomic operations.
8721         * config/rs6000/sync.md (load_lockedti): Insure that the address
8722         is a proper indexed or indirect address for the lqarx instruction.
8723         On little endian systems, swap the hi/lo registers after the lqarx
8724         instruction.
8725         (load_lockedpti): Use indexed_or_indirect_operand predicate to
8726         insure the address is valid for the lqarx instruction.
8727         (store_conditionalti): Insure that the address is a proper indexed
8728         or indirect address for the stqcrx. instruction.  On little endian
8729         systems, swap the hi/lo registers before doing the stqcrx.
8730         instruction.
8731         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
8732         insure the address is valid for the stqcrx. instruction.
8734         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8735         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
8736         type of quad memory support is available.
8738 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
8740         PR regression/59915
8741         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
8742         there is a danger of looping.
8744 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
8746         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8747         force flag_ira_loop_pressure if set via command line.
8749 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
8751         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
8752         (ashr_simd): New builtin handling DI mode.
8753         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
8754         (aarch64_sshr_simddi): New match pattern.
8755         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
8756         (vshrd_n_s64): Likewise.
8757         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
8759 2014-01-23  Nick Clifton  <nickc@redhat.com>
8761         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
8762         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
8763         favour of mcu specific scripts.
8764         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
8765         430x multilibs.
8767 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
8768             Alex Velenko  <Alex.Velenko@arm.com>
8770         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
8771         (vaddv_s16): Likewise.
8772         (vaddv_s32): Likewise.
8773         (vaddv_u8): Likewise.
8774         (vaddv_u16): Likewise.
8775         (vaddv_u32): Likewise.
8776         (vaddvq_s8): Likewise.
8777         (vaddvq_s16): Likewise.
8778         (vaddvq_s32): Likewise.
8779         (vaddvq_s64): Likewise.
8780         (vaddvq_u8): Likewise.
8781         (vaddvq_u16): Likewise.
8782         (vaddvq_u32): Likewise.
8783         (vaddvq_u64): Likewise.
8784         (vaddv_f32): Likewise.
8785         (vaddvq_f32): Likewise.
8786         (vaddvq_f64): Likewise.
8787         (vmaxv_f32): Likewise.
8788         (vmaxv_s8): Likewise.
8789         (vmaxv_s16): Likewise.
8790         (vmaxv_s32): Likewise.
8791         (vmaxv_u8): Likewise.
8792         (vmaxv_u16): Likewise.
8793         (vmaxv_u32): Likewise.
8794         (vmaxvq_f32): Likewise.
8795         (vmaxvq_f64): Likewise.
8796         (vmaxvq_s8): Likewise.
8797         (vmaxvq_s16): Likewise.
8798         (vmaxvq_s32): Likewise.
8799         (vmaxvq_u8): Likewise.
8800         (vmaxvq_u16): Likewise.
8801         (vmaxvq_u32): Likewise.
8802         (vmaxnmv_f32): Likewise.
8803         (vmaxnmvq_f32): Likewise.
8804         (vmaxnmvq_f64): Likewise.
8805         (vminv_f32): Likewise.
8806         (vminv_s8): Likewise.
8807         (vminv_s16): Likewise.
8808         (vminv_s32): Likewise.
8809         (vminv_u8): Likewise.
8810         (vminv_u16): Likewise.
8811         (vminv_u32): Likewise.
8812         (vminvq_f32): Likewise.
8813         (vminvq_f64): Likewise.
8814         (vminvq_s8): Likewise.
8815         (vminvq_s16): Likewise.
8816         (vminvq_s32): Likewise.
8817         (vminvq_u8): Likewise.
8818         (vminvq_u16): Likewise.
8819         (vminvq_u32): Likewise.
8820         (vminnmv_f32): Likewise.
8821         (vminnmvq_f32): Likewise.
8822         (vminnmvq_f64): Likewise.
8824 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
8826         * config/aarch64/aarch64-simd.md
8827         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
8828         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
8829         (*aarch64_mul3_elt<mode>): Likewise.
8830         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
8831         (*aarch64_mul3_elt_to_64v2df): Likewise.
8832         (*aarch64_mla_elt<mode>): Likewise.
8833         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
8834         (*aarch64_mls_elt<mode>): Likewise.
8835         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
8836         (*aarch64_fma4_elt<mode>): Likewise.
8837         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
8838         (*aarch64_fma4_elt_to_64v2df): Likewise.
8839         (*aarch64_fnma4_elt<mode>): Likewise.
8840         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8841         (*aarch64_fnma4_elt_to_64v2df): Likewise.
8842         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
8843         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
8844         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
8845         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
8846         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8847         (aarch64_sqdmull_lane<mode>_internal): Likewise.
8848         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8850 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
8852         * config/aarch64/aarch64-simd.md
8853         (aarch64_be_checked_get_lane<mode>): New define_expand.
8854         * config/aarch64/aarch64-simd-builtins.def
8855         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
8856         New builtin definition.
8857         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
8858         Use new safe be builtin.
8860 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
8862         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
8863         New define_insn.
8864         (aarch64_be_st1<mode>): Likewise.
8865         (aarch_ld1<VALL:mode>): Define_expand modified.
8866         (aarch_st1<VALL:mode>): Likewise.
8867         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
8868         (UNSPEC_ST1): Likewise.
8870 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
8872         * config/microblaze/microblaze.md: Add trap insn and attribute
8874 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
8876         PR preprocessor/58580
8877         * input.h (location_get_source_line): Take an additional line_size
8878         parameter.
8879         (void diagnostics_file_cache_fini): Declare new function.
8880         * input.c (struct fcache): New type.
8881         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
8882         New static constants.
8883         (diagnostic_file_cache_init, total_lines_num)
8884         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
8885         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
8886         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
8887         (get_next_line, read_next_line, goto_next_line, read_line_num):
8888         New static function definitions.
8889         (diagnostic_file_cache_fini): New function.
8890         (location_get_source_line): Take an additional output line_len
8891         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
8892         read_line_num.
8893         * diagnostic.c (diagnostic_finish): Call
8894         diagnostic_file_cache_fini.
8895         (adjust_line): Take an additional input parameter for the length
8896         of the line, rather than calculating it with strlen.
8897         (diagnostic_show_locus): Adjust the use of
8898         location_get_source_line and adjust_line with respect to their new
8899         signature.  While displaying a line now, do not stop at the first
8900         null byte.  Rather, display the zero byte as a space and keep
8901         going until we reach the size of the line.
8902         * Makefile.in: Add vec.o to OBJS-libcommon
8904 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8905             Ilya Tocar     <ilya.tocar@intel.com>
8907         * config/i386/avx512fintrin.h (_mm512_kmov): New.
8908         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
8909         (__builtin_ia32_kmov16): Ditto.
8910         * config/i386/i386.md (UNSPEC_KMOV): New.
8911         (kmovw): Ditto.
8913 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8915         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
8916         (_mm512_storeu_si512): Ditto.
8918 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
8920         PR target/52125
8921         * rtl.h (get_referenced_operands): Declare.
8922         * recog.c (get_referenced_operands): New function.
8923         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
8924         operands have been referenced when recording LO_SUM references.
8926 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
8928         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
8930 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
8932         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
8933         Enable for generic and recent AMD targets.
8935 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
8937         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
8938         ARG_SIZE note when adjustment was eliminated.
8940 2014-01-22  Jeff Law  <law@redhat.com>
8942         PR tree-optimization/59597
8943         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
8944         in file.  Accept new argument REGISTERING and use it to modify
8945         dump output appropriately.
8946         (register_jump_thread): Corresponding changes.
8947         (mark_threaded_blocks): Reinstate code to cancel unprofitable
8948         thread paths involving joiner blocks.  Add code to dump cancelled
8949         jump threading paths.
8951 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
8953         PR rtl-optimization/59477
8954         * lra-constraints.c (inherit_in_ebb): Process call for living hard
8955         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
8957 2014-01-22  Tom Tromey  <tromey@redhat.com>
8959         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
8960         PARAMS.
8961         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
8963 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
8965         PR rtl-optimization/59896
8966         * lra-constraints.c (process_alt_operands): Check unused note for
8967         matched operands of insn with no output reloads.
8969 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
8971         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
8972         (mips_move_from_gpr_cost): Likewise.
8974 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
8976         PR rtl-optimization/59858
8977         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
8978         ira_class_hard_regs_num.
8979         (process_alt_operands): Increase reject for dying matched operand.
8981 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
8983         PR target/59003
8984         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
8985         smaller than size, perform several stores or loads and stores
8986         at dst + count - size to store or copy all of size bytes, rather
8987         than just last modesize bytes.
8989 2014-01-20  DJ Delorie  <dj@redhat.com>
8991         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
8992         that CLOBBERs are REGs before propogating their values.
8994 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
8996         PR middle-end/59789
8997         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
8998         (cgraph_inline_failed_type): New function.
8999         * cgraph.h (DEFCIFCODE): Add type.
9000         (cgraph_inline_failed_type_t): New enum.
9001         (cgraph_inline_failed_type): New prototype.
9002         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
9003         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
9004         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
9005         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
9006         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
9007         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
9008         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
9009         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
9010         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
9011         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
9012         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
9013         OPTIMIZATION_MISMATCH.
9014         * tree-inline.c (expand_call_inline): Emit errors during
9015         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
9017 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9019         PR target/59685
9020         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
9021         mode attribute in insn output.
9023 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
9025         * output.h (output_constant): Delete.
9026         * varasm.c (output_constant): Make private.
9028 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
9030         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
9032 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9034         PR middle-end/59860
9035         * tree.h (fold_builtin_strcat): New prototype.
9036         * builtins.c (fold_builtin_strcat): No longer static.  Add len
9037         argument, if non-NULL, don't call c_strlen.  Optimize
9038         directly into __builtin_memcpy instead of __builtin_strcpy.
9039         (fold_builtin_2): Adjust fold_builtin_strcat caller.
9040         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
9042 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
9044         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9045         for SImode_address_operand operands, having only a REG argument.
9047 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9049         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
9050         loader name using mbig-endian.
9051         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
9053 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9055         * doc/invoke.texi (-march): Clarify documentation for AArch64.
9056         (-mtune): Likewise.
9057         (-mcpu): Likewise.
9059 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
9061         * config/aarch64/aarch64-protos.h
9062         (aarch64_cannot_change_mode_class_ptr): Declare.
9063         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
9064         aarch64_cannot_change_mode_class_ptr): New.
9065         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
9066         backend hook aarch64_cannot_change_mode_class.
9068 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
9070         * common/config/aarch64/aarch64-common.c
9071         (aarch64_handle_option): Don't handle any option order logic here.
9072         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
9073         selected_cpu, warn on architecture version mismatch.
9074         (aarch64_override_options): Fix parsing order for option strings.
9076 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9077             Iain Sandoe  <iain@codesourcery.com>
9079         PR bootstrap/59496
9080         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
9081         warning.  Amend comment to reflect current functionality.
9083 2014-01-20  Richard Biener  <rguenther@suse.de>
9085         PR middle-end/59860
9086         * builtins.c (fold_builtin_strcat): Remove case better handled
9087         by tree-ssa-strlen.c.
9089 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
9091         * config/aarch64/aarch64.opt
9092         (mcpu, march, mtune): Make case-insensitive.
9094 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
9096         PR target/59880
9097         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
9098         if operands[1] is a REG or ZERO_EXTEND of a REG.
9100 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
9102         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
9104 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
9106         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
9107         long non-pic millicode calls.
9109 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
9111         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
9113 2014-01-19  Kito Cheng  <kito@0xlab.org>
9115         * builtins.c (expand_movstr): Check movstr expand done or fail.
9117 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9118             H.J. Lu  <hongjiu.lu@intel.com>
9120         PR target/59379
9121         * config/i386/i386.md (*lea<mode>): Zero-extend return register
9122         to DImode for zero-extended addresses.
9124 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
9126         PR rtl-optimization/57763
9127         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
9128         on the new indirect jump_insn and increment LABEL_NUSES (label).
9130 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
9132         PR bootstrap/59580
9133         PR bootstrap/59583
9134         * config.gcc (x86_archs): New variable.
9135         (x86_64_archs): Likewise.
9136         (x86_cpus): Likewise.
9137         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
9138         --with-arch/--with-cpu= options.
9139         Support --with-arch=/--with-cpu={nehalem,westmere,
9140         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
9142 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9144         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
9145         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
9147 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
9149         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
9151 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
9153         PR target/58944
9154         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
9155         clear cpp_get_options (parse_in)->warn_unused_macros for
9156         ix86_target_macros_internal with cpp_define.
9158 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
9160         * jump.c (delete_related_insns): Keep (use (insn))s.
9161         * reorg.c (redundant_insn): Check for barriers too.
9163 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9165         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
9167 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
9169         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
9170         call to $$dyncall when TARGET_LONG_CALLS is true.
9172 2014-01-17  Jeff Law  <law@redhat.com>
9174         * ree.c (combine_set_extension): Temporarily disable test for
9175         changing number of hard registers.
9177 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
9179         PR middle-end/58125
9180         * ipa-inline-analysis.c (inline_free_summary):
9181         Do not free summary of aliases.
9183 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
9185         PR middle-end/59706
9186         * gimplify.c (gimplify_expr): Use create_tmp_var
9187         instead of create_tmp_var_raw.  If cond doesn't have
9188         integral type, don't add the IFN_ANNOTATE builtin at all.
9190 2014-01-17  Martin Jambor  <mjambor@suse.cz>
9192         PR ipa/59736
9193         * ipa-cp.c (prev_edge_clone): New variable.
9194         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
9195         Also resize prev_edge_clone vector.
9196         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
9197         (ipcp_edge_removal_hook): New function.
9198         (ipcp_driver): Register ipcp_edge_removal_hook.
9200 2014-01-17  Andrew Pinski  <apinski@cavium.com>
9201             Steve Ellcey  <sellcey@mips.com>
9203         PR target/59462
9204         * config/mips/mips.c (mips_print_operand): Check operand mode instead
9205         of operator mode.
9207 2014-01-17  Jeff Law  <law@redhat.com>
9209         PR middle-end/57904
9210         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
9211         so that pass_ccp runs first.
9213 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9215         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
9216         (ix86_adjust_cost): Use !TARGET_XXX.
9217         (do_reorder_for_imul): Likewise.
9218         (swap_top_of_ready_list): Likewise.
9219         (ix86_sched_reorder): Likewise.
9221 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
9223         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9224         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
9225         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
9226         (intel_memset): New.  Duplicate slm_memset.
9227         (intel_cost): New.  Duplicate slm_cost.
9228         (m_INTEL): New macro.
9229         (processor_target_table): Add "intel".
9230         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
9231         with PROCESSOR_INTEL for "intel".
9232         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
9233         PROCESSOR_SILVERMONT.
9234         (ix86_issue_rate): Likewise.
9235         (ix86_adjust_cost): Likewise.
9236         (ia32_multipass_dfa_lookahead): Likewise.
9237         (swap_top_of_ready_list): Likewise.
9238         (ix86_sched_reorder): Likewise.
9239         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
9240         instead of TARGET_OPT_AGU.
9241         * config/i386/i386.h (TARGET_INTEL): New.
9242         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
9243         (processor_type): Add PROCESSOR_INTEL.
9244         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
9245         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
9247 2014-01-17  Marek Polacek  <polacek@redhat.com>
9249         PR c/58346
9250         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
9251         size is zero.
9253 2014-01-17  Richard Biener  <rguenther@suse.de>
9255         PR tree-optimization/46590
9256         * opts.c (default_options_table): Add entries for
9257         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
9258         all enabled at -O1 but not for -Og.
9259         * common.opt (fbranch-count-reg): Remove Init(1).
9260         (fmove-loop-invariants): Likewise.
9261         (ftree-pta): Likewise.
9263 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
9265         * config/i386/i386.c (ix86_data_alignment): For compatibility with
9266         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
9267         decls to at least the GCC 4.8 used alignments.
9269         PR fortran/59440
9270         * tree-nested.c (convert_nonlocal_reference_stmt,
9271         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
9272         of GIMPLE_BIND stmts, adjust associated decls.
9274 2014-01-17  Richard Biener  <rguenther@suse.de>
9276         PR tree-optimization/46590
9277         * vec.h (vec<>::bseach): New member function implementing
9278         binary search according to C89 bsearch.
9279         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
9280         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
9281         bitmap pointer again.  Make accesses_in_loop a flat array.
9282         (mem_ref_obstack): New global.
9283         (outermost_indep_loop): Adjust for mem_ref->stored changes.
9284         (mark_ref_stored): Likewise.
9285         (ref_indep_loop_p_2): Likewise.
9286         (set_ref_stored_in_loop): New helper function.
9287         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
9288         (memref_free): Adjust.
9289         (record_mem_ref_loc): Simplify.
9290         (gather_mem_refs_stmt): Adjust.
9291         (sort_locs_in_loop_postorder_cmp): New function.
9292         (analyze_memory_references): Sort accesses_in_loop after
9293         loop postorder number.
9294         (find_ref_loc_in_loop_cmp): New function.
9295         (for_all_locs_in_loop): Find relevant cluster of locs in
9296         accesses_in_loop and iterate without recursion.
9297         (execute_sm): Avoid uninit warning.
9298         (struct ref_always_accessed): Simplify.
9299         (ref_always_accessed::operator ()): Likewise.
9300         (ref_always_accessed_p): Likewise.
9301         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
9302         loop postorder numbers here.
9303         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
9304         numbers.
9306 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
9308         PR c++/57945
9309         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
9310         on decls for which assemble_alias has been called.
9312 2014-01-17  Nick Clifton  <nickc@redhat.com>
9314         * config/msp430/msp430.opt: (mcpu): New option.
9315         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
9316         (msp430_option_override): Parse target_cpu.  If the MCU name
9317         matches a generic string, clear target_mcu.
9318         (msp430_attr): Allow numeric interrupt values up to 63.
9319         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
9320         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
9321         option.
9322         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
9323         Add mcpu matches.
9324         * config/msp430/msp430.md (popm): Use %J rather than %I.
9325         (addsi3): Use msp430_nonimmediate_operand for operand 2.
9326         (addhi_cy_i): Use immediate_operand for operand 2.
9327         * doc/invoke.texi: Document -mcpu option.
9329 2014-01-17  Richard Biener  <rguenther@suse.de>
9331         PR rtl-optimization/38518
9332         * df.h (df_analyze_loop): Declare.
9333         * df-core.c: Include cfgloop.h.
9334         (df_analyze_1): Split out main part of df_analyze.
9335         (df_analyze): Adjust.
9336         (loop_inverted_post_order_compute): New function.
9337         (loop_post_order_compute): Likewise.
9338         (df_analyze_loop): New function avoiding whole-function
9339         postorder computes.
9340         * loop-invariant.c (find_defs): Use df_analyze_loop.
9341         (find_invariants): Adjust.
9342         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
9344 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
9346         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
9347         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
9349 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
9351         * ipa-ref.c (ipa_remove_stmt_references): Fix references
9352         traversal when removing references.
9354 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
9356         PR ipa/59775
9357         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
9359 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
9361         PR middle-end/56791
9362         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
9363         pushing a reload for an autoinc when we had previously reloaded an
9364         inner part of the address.
9366 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
9368         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
9369         field.
9370         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
9371         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
9372         when not giving up or versioning for alias only because of
9373         loop->safelen.
9374         (vect_analyze_data_ref_dependences): Set to true.
9375         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
9376         is a GIMPLE_PHI.
9377         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
9378         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
9379         to the condition.
9381         PR middle-end/58344
9382         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
9384         PR target/59839
9385         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
9386         operand 0 predicate for gathers, use a new pseudo as subtarget.
9388 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
9390         PR middle-end/59609
9391         * lra-constraints.c (process_alt_operands): Add printing debug info.
9392         Check absence of input/output reloads for matched operands too.
9394 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
9396         PR rtl-optimization/59835
9397         * ira.c (ira_init_register_move_cost): Increase cost for
9398         impossible modes.
9400 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
9402         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
9404 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
9406         PR target/59780
9407         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
9408         non-register objects.  Use gen_(high/low)part more consistently.
9409         Fix assertions.
9411 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9413         PR target/59844
9414         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
9415         endian support, remove tests for WORDS_BIG_ENDIAN.
9416         (p8_mfvsrd_3_<mode>): Likewise.
9417         (reload_gpr_from_vsx<mode>): Likewise.
9418         (reload_gpr_from_vsxsf): Likewise.
9419         (p8_mfvsrd_4_disf): Likewise.
9421 2014-01-16  Richard Biener  <rguenther@suse.de>
9423         PR rtl-optimization/46590
9424         * lcm.c (compute_antinout_edge): Use postorder iteration.
9425         (compute_laterin): Use inverted postorder iteration.
9427 2014-01-16  Nick Clifton  <nickc@redhat.com>
9429         PR middle-end/28865
9430         * varasm.c (output_constant): Return the number of bytes actually
9431         emitted.
9432         (output_constructor_array_range): Update the field size with the
9433         number of bytes emitted by output_constant.
9434         (output_constructor_regular_field): Likewise.  Also do not
9435         complain if the total number of bytes emitted is now greater
9436         than the expected fieldpos.
9437         * output.h (output_constant): Update prototype and descriptive comment.
9439 2014-01-16  Marek Polacek  <polacek@redhat.com>
9441         PR middle-end/59827
9442         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
9443         it is error_mark_node.
9445 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
9447         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
9448         VALID_AVX256_REG_OR_OI_MODE.
9450 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
9452         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
9453         current procedure should be profiled.
9455 2014-01-15  Andrew Pinski  <apinski@cavium.com>
9457         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
9458         of moving from/to the STACK_REG register class.
9460 2014-01-15  Richard Henderson  <rth@redhat.com>
9462         PR debug/54694
9463         * reginfo.c (global_regs_decl): Globalize.
9464         * rtl.h (global_regs_decl): Declare.
9465         * ira.c (do_reload): Diagnose frame_pointer_needed and it
9466         reserved via global_regs.
9468 2014-01-15  Teresa Johnson  <tejohnson@google.com>
9470         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
9472 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
9474         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
9475         and vmulosh rather than call gen_vec_widen_smult_*.
9476         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
9477         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
9478         (vec_widen_smult_even_v16qi): Likewise.
9479         (vec_widen_umult_even_v8hi): Likewise.
9480         (vec_widen_smult_even_v8hi): Likewise.
9481         (vec_widen_umult_odd_v16qi): Likewise.
9482         (vec_widen_smult_odd_v16qi): Likewise.
9483         (vec_widen_umult_odd_v8hi): Likewise.
9484         (vec_widen_smult_odd_v8hi): Likewise.
9485         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
9486         vmuloub rather than call gen_vec_widen_umult_*.
9487         (vec_widen_umult_lo_v16qi): Likewise.
9488         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
9489         vmulosb rather than call gen_vec_widen_smult_*.
9490         (vec_widen_smult_lo_v16qi): Likewise.
9491         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
9492         rather than call gen_vec_widen_umult_*.
9493         (vec_widen_umult_lo_v8hi): Likewise.
9494         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
9495         rather than call gen_vec_widen_smult_*.
9496         (vec_widen_smult_lo_v8hi): Likewise.
9498 2014-01-15  Jeff Law  <law@redhat.com>
9500         PR tree-optimization/59747
9501         * ree.c (find_and_remove_re): Properly handle case where a second
9502         eliminated extension requires widening a copy created for elimination
9503         of a prior extension.
9504         (combine_set_extension): Ensure that the number of hard regs needed
9505         for a destination register does not change when we widen it.
9507 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9509         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
9510         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
9511         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
9512         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
9513         (avr-*-rtems*): Likewise.
9514         (bfin*-rtems*): Likewise.
9515         (moxie-*-rtems*): Likewise.
9516         (h8300-*-rtems*): Likewise.
9517         (i[34567]86-*-rtems*): Likewise.
9518         (lm32-*-rtems*): Likewise.
9519         (m32r-*-rtems*): Likewise.
9520         (m68k-*-rtems*): Likewise.
9521         (microblaze*-*-rtems*): Likewise.
9522         (mips*-*-rtems*): Likewise.
9523         (powerpc-*-rtems*): Likewise.
9524         (sh-*-rtems*): Likewise.
9525         (sparc-*-rtems*): Likewise.
9526         (sparc64-*-rtems*): Likewise.
9527         (v850-*-rtems*): Likewise.
9528         (m32c-*-rtems*): Likewise.
9530 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
9532         PR rtl-optimization/59511
9533         * ira.c (ira_init_register_move_cost): Use memory costs for some
9534         cases of register move cost calculations.
9535         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
9536         instead of BB frequency.
9537         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
9538         * lra-assigns.c (find_hard_regno_for): Ditto.
9540 2014-01-15  Richard Biener  <rguenther@suse.de>
9542         PR tree-optimization/59822
9543         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
9544         (vectorizable_load): Use it to hoist defs of uses of invariant
9545         loads out of the loop.
9547 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
9548             Kugan Vivekanandarajah  <kuganv@linaro.org>
9550         PR target/59695
9551         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
9552         truncation.
9554 2014-01-15  Richard Biener  <rguenther@suse.de>
9556         PR rtl-optimization/59802
9557         * lcm.c (compute_available): Use inverted postorder to seed
9558         the initial worklist.
9560 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9562         PR target/59803
9563         * config/s390/s390.c (s390_preferred_reload_class): Don't return
9564         ADDR_REGS for invalid symrefs in non-PIC code.
9566 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
9568         PR other/58712
9569         * builtins.c (determine_block_size): Initialize *probable_max_size
9570         even if len_rtx is CONST_INT.
9572 2014-01-14  Andrew Pinski  <apinski@cavium.com>
9574         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
9575         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
9576         (cortexa53_tunings): Likewise.
9577         (aarch64_sched_issue_rate): New function.
9578         (TARGET_SCHED_ISSUE_RATE): Define.
9580 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
9582         * ira-costs.c (find_costs_and_classes): Add missed
9583         ira_init_register_move_cost_if_necessary.
9585 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
9587         PR target/59787
9588         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
9590 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
9592         PR target/59794
9593         * config/i386/i386.c (type_natural_mode): Add a bool parameter
9594         to indicate if type is used for function return value.  Warn ABI
9595         change if the vector mode isn't available for function return value.
9596         (ix86_function_arg_advance): Pass false to type_natural_mode.
9597         (ix86_function_arg): Likewise.
9598         (ix86_gimplify_va_arg): Likewise.
9599         (function_arg_32): Don't warn ABI change.
9600         (ix86_function_value): Pass true to type_natural_mode.
9601         (ix86_return_in_memory): Likewise.
9602         (ix86_struct_value_rtx): Removed.
9603         (TARGET_STRUCT_VALUE_RTX): Likewise.
9605 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9607         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
9608         converting a conditional jump into a conditional return.
9610 2014-01-14  Richard Biener  <rguenther@suse.de>
9612         PR tree-optimization/58921
9613         PR tree-optimization/59006
9614         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
9615         hoisting invariant stmts.
9616         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
9617         invariant loads on the preheader edge if possible.
9619 2014-01-14  Joey Ye  <joey.ye@arm.com>
9621         * doc/plugin.texi (Building GCC plugins): Update to C++.
9623 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
9625         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
9626         (_mm_rcp28_round_ss): Ditto.
9627         (_mm_rsqrt28_round_sd): Ditto.
9628         (_mm_rsqrt28_round_ss): Ditto.
9629         (_mm_rcp28_sd): Ditto.
9630         (_mm_rcp28_ss): Ditto.
9631         (_mm_rsqrt28_sd): Ditto.
9632         (_mm_rsqrt28_ss): Ditto.
9633         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
9634         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
9635         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
9636         (IX86_BUILTIN_RCP28SD): Ditto.
9637         (IX86_BUILTIN_RCP28SS): Ditto.
9638         (IX86_BUILTIN_RSQRT28SD): Ditto.
9639         (IX86_BUILTIN_RSQRT28SS): Ditto.
9640         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
9641         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
9642         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
9643         (ix86_expand_special_args_builtin): Expand new FTYPE.
9644         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
9645         (srcp14<mode>): Make insn unary.
9646         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
9647         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
9648         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
9649         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
9650         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
9651         Fix rounding: make it SAE only.
9652         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
9653         Ditto.
9654         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
9655         Ditto.
9656         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
9657         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
9658         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
9659         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
9660         (round_saeonly_mask_scalar_operand4): Ditto.
9661         (round_saeonly_mask_scalar_op3): Ditto.
9662         (round_saeonly_mask_scalar_op4): Ditto.
9664 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9666         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9667         Implement -maltivec=be for vec_insert and vec_extract.
9669 2014-01-10  DJ Delorie  <dj@redhat.com>
9671         * config/msp430/msp430.md (call_internal): Don't allow memory
9672         references with SP as the base register.
9673         (call_value_internal): Likewise.
9674         * config/msp430/constraints.md (Yc): New.  For memory references
9675         that don't use SP as a base register.
9677         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
9678         "an integer without a # prefix"
9679         * config/msp430/msp430.md (epilogue_helper): Use it.
9681 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
9683         PR target/59617
9684         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
9685         AVX512F gather builtins.
9686         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
9687         on gather decls with INTEGER_TYPE masktype.
9688         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
9689         directly into the builtin rather than hoisting it before loop.
9691         PR tree-optimization/59387
9692         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
9693         (scev_const_prop): If folded_casts and type has undefined overflow,
9694         use force_gimple_operand instead of force_gimple_operand_gsi and
9695         for each added stmt if it is assign with
9696         arith_code_with_undefined_signed_overflow, call
9697         rewrite_to_defined_overflow.
9698         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
9699         gimple-fold.h instead.
9700         (arith_code_with_undefined_signed_overflow,
9701         rewrite_to_defined_overflow): Moved to ...
9702         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
9703         rewrite_to_defined_overflow): ... here.  No longer static.
9704         Include gimplify-me.h.
9705         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
9706         rewrite_to_defined_overflow): New prototypes.
9708 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9710         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
9712 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
9714         * builtins.c (get_object_alignment_2): Minor tweak.
9715         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
9717 2014-01-13  Christian Bruel  <christian.bruel@st.com>
9719         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
9720         optimized non constant lengths.
9722 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
9724         PR libgomp/59194
9725         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
9726         load as __atomic_load_N if possible.
9728 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
9730         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
9731         target parameter.
9732         (rs6000_expand_builtin): Adjust call.
9734 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
9736         PR target/58115
9737         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
9738         * config/rs6000/rs6000.c: Include target-globals.h.
9739         (rs6000_set_current_function): Instead of doing target_reinit
9740         unconditionally, use save_target_globals_default_opts and
9741         restore_target_globals.
9743         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
9744         FPSCR.
9745         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
9746         (rs6000_expand_builtin): Handle mffs and mtfsf.
9747         (rs6000_init_builtins): Define mffs and mtfsf.
9748         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
9749         (rs6000_mffs): New pattern.
9750         (rs6000_mtfsf): New pattern.
9752 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
9754         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
9755         Start narrowing with START.  Apply candidate-use pair
9756         and check overall cost in narrowing.
9757         (iv_ca_prune): Pass new argument.
9759 2014-01-10  Jeff Law  <law@redhat.com>
9761         PR middle-end/59743
9762         * ree.c (combine_reaching_defs): Ensure the defining statement
9763         occurs before the extension when optimizing extensions with
9764         different source and destination hard registers.
9766 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
9768         PR ipa/58585
9769         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
9770         vtables into the type inheritance graph.
9772 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9774         PR rtl-optimization/59754
9775         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
9776         modes in the REGNO != REGNO case.
9778 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9780         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
9782 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9784         PR tree-optimization/59745
9785         * tree-predcom.c (tree_predictive_commoning_loop): Call
9786         free_affine_expand_cache if giving up because components is NULL.
9788         * target-globals.c (save_target_globals): Allocate < 4KB structs using
9789         GC in payload of target_globals struct instead of allocating them on
9790         the heap and the larger structs separately using GC.
9791         * target-globals.h (struct target_globals): Make regs, hard_regs,
9792         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
9793         of GTY((skip)) and change type to void *.
9794         (reset_target_globals): Cast loads from those fields to corresponding
9795         types.
9797 2014-01-10  Steve Ellcey  <sellcey@mips.com>
9799         PR plugins/59335
9800         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
9801         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
9802         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
9804 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
9806         PR target/59744
9807         * aarch64-modes.def (CC_Zmode): New flags mode.
9808         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
9809         represents an equality.
9810         (aarch64_get_condition_code): Handle CC_Zmode.
9811         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
9813 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9815         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
9816         extraction in good case.
9818 2014-01-10  Richard Biener  <rguenther@suse.de>
9820         PR tree-optimization/59374
9821         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
9822         checking after SLP discovery.  Mark stmts not participating
9823         in any SLP instance properly.
9825 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9827         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
9828         when handling a SET rtx.
9830 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9832         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
9833         (cortex-a57): Likewise.
9834         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
9836 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9838         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
9839         non-iwmmxt builtins.
9841 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
9843         PR ipa/58252
9844         PR ipa/59226
9845         * ipa-devirt.c record_target_from_binfo): Take as argument
9846         stack of binfos and lookup matching one for virtual inheritance.
9847         (possible_polymorphic_call_targets_1): Update.
9849 2014-01-10  Huacai Chen  <chenhc@lemote.com>
9851         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
9852         kernel strings for Loongson-2E/2F/3A.
9854 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
9856         PR middle-end/59670
9857         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
9858         is_gimple_call before calling gimple_call_internal_p.
9860 2014-01-09  Steve Ellcey  <sellcey@mips.com>
9862         * Makefile.in (TREE_FLOW_H): Remove.
9863         (TREE_SSA_H): Add file names from tree-flow.h.
9864         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
9865         * tree.h: Remove tree-flow.h reference.
9866         * hash-table.h: Remove tree-flow.h reference.
9867         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
9868         reference with tree-ssa-loop.h.
9870 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9872         * doc/invoke.texi: Add -maltivec={be,le} options, and document
9873         default element-order behavior for -maltivec.
9874         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
9875         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
9876         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
9877         when targeting big endian, at least for now.
9878         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
9880 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
9882         PR middle-end/47735
9883         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
9884         var satisfies use_register_for_decl, just take into account type
9885         alignment, rather than decl alignment.
9887         PR tree-optimization/59622
9888         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
9889         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
9890         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
9891         Don't devirtualize for inplace at all.  For targets.length () == 1,
9892         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
9894 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
9896         * config/i386/i386.md (cpu): Remove the unused btver1.
9898 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
9900         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
9902 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
9904         PR target/58115
9905         * tree-core.h (struct target_globals): New forward declaration.
9906         (struct tree_target_option): Add globals field.
9907         * tree.h (TREE_TARGET_GLOBALS): Define.
9908         (prepare_target_option_nodes_for_pch): New prototype.
9909         * target-globals.h (struct target_globals): Define even if
9910         !SWITCHABLE_TARGET.
9911         * tree.c (prepare_target_option_node_for_pch,
9912         prepare_target_option_nodes_for_pch): New functions.
9913         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
9914         * config/i386/i386.c: Include target-globals.h.
9915         (ix86_set_current_function): Instead of doing target_reinit
9916         unconditionally, use save_target_globals_default_opts and
9917         restore_target_globals.
9919 2014-01-09  Richard Biener  <rguenther@suse.de>
9921         PR tree-optimization/59715
9922         * tree-cfg.h (split_critical_edges): Declare.
9923         * tree-cfg.c (split_critical_edges): Export.
9924         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
9926 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
9928         * cfgexpand.c (expand_stack_vars): Optionally disable
9929         asan stack protection.
9930         (expand_used_vars): Likewise.
9931         (partition_stack_vars): Likewise.
9932         * asan.c (asan_emit_stack_protection): Optionally disable
9933         after return stack usage.
9934         (instrument_derefs): Optionally disable memory access instrumentation.
9935         (instrument_builtin_call): Likewise.
9936         (instrument_strlen_call): Likewise.
9937         (asan_protect_global): Optionally disable global variables protection.
9938         * doc/invoke.texi: Added doc for new options.
9939         * params.def: Added new options.
9940         * params.h: Likewise.
9942 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
9944         PR rtl-optimization/59724
9945         * ifcvt.c (cond_exec_process_if_block): Don't call
9946         flow_find_head_matching_sequence with 0 longest_match.
9947         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
9948         non-active insns if !stop_after.
9949         (try_head_merge_bb): Revert 2014-01-07 changes.
9951 2014-01-08  Jeff Law  <law@redhat.com>
9953         * ree.c (get_sub_rtx): New function, extracted from...
9954         (merge_def_and_ext): Here.
9955         (combine_reaching_defs): Use get_sub_rtx.
9957 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
9959         * cgraph.h (varpool_variable_node): Do not choke on null node.
9961 2014-01-08  Catherine Moore  <clm@codesourcery.com>
9963         * config/mips/mips.md (simple_return): Attempt to use JRC
9964         for microMIPS.
9965         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
9967 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
9969         PR rtl-optimization/59137
9970         * reorg.c (steal_delay_list_from_target): Call update_block for
9971         elided insns.
9972         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
9974 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9976         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
9977         two duplicate entries.
9979 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
9981         Revert:
9982         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
9984         * config/mips/mips.c (mips_truncated_op_cost): New function.
9985         (mips_rtx_costs): Adjust test for BADDU.
9986         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
9988         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
9990         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
9991         (*baddu_si): ...this new pattern.
9993 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
9995         PR ipa/59722
9996         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
9998 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
10000         PR middle-end/57748
10001         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
10002         inner_reference_p.
10003         (expand_expr, expand_normal): Adjust.
10004         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
10005         inner_reference_p. Use inner_reference_p to expand inner references.
10006         (store_expr): Adjust.
10007         * cfgexpand.c (expand_call_stmt): Adjust.
10009 2014-01-08  Rong Xu  <xur@google.com>
10011         * gcov-io.c (gcov_var): Move from gcov-io.h.
10012         (gcov_position): Ditto.
10013         (gcov_is_error): Ditto.
10014         (gcov_rewrite): Ditto.
10015         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
10016         only part to libgcc/libgcov.h.
10018 2014-01-08  Marek Polacek  <polacek@redhat.com>
10020         PR middle-end/59669
10021         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
10023 2014-01-08  Marek Polacek  <polacek@redhat.com>
10025         PR sanitizer/59667
10026         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
10028 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
10030         PR rtl-optimization/59649
10031         * stor-layout.c (get_mode_bounds): For BImode return
10032         0 and STORE_FLAG_VALUE.
10034 2014-01-08  Richard Biener  <rguenther@suse.de>
10036         PR middle-end/59630
10037         * gimple.h (is_gimple_builtin_call): Remove.
10038         (gimple_builtin_call_types_compatible_p): New.
10039         (gimple_call_builtin_p): New overload.
10040         * gimple.c (is_gimple_builtin_call): Remove.
10041         (validate_call): Rename to ...
10042         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
10043         check return types.
10044         (validate_type): New static function.
10045         (gimple_call_builtin_p): New overload and adjust.
10046         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
10047         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
10048         (gimple_fold_stmt_to_constant_1): Likewise.
10049         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
10051 2014-01-08  Richard Biener  <rguenther@suse.de>
10053         PR middle-end/59471
10054         * gimplify.c (gimplify_expr): Gimplify register-register type
10055         VIEW_CONVERT_EXPRs to separate stmts.
10057 2014-01-07  Jeff Law  <law@redhat.com>
10059         PR middle-end/53623
10060         * ree.c (combine_set_extension): Handle case where source
10061         and destination registers in an extension insn are different.
10062         (combine_reaching_defs): Allow source and destination registers
10063         in extension to be different under limited circumstances.
10064         (add_removable_extension): Remove restriction that the
10065         source and destination registers in the extension are the same.
10066         (find_and_remove_re): Emit a copy from the extension's
10067         destination to its source after the defining insn if
10068         the source and destination registers are different.
10070         PR middle-end/59285
10071         * ifcvt.c (merge_if_block): If we are merging a block with more than
10072         one successor with a block with no successors, remove any BARRIER
10073         after the second block.
10075 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
10077         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
10079 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
10081         PR target/59652
10082         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
10083         for 14-bit register offsets when INT14_OK_STRICT is false.
10085 2014-01-07  Roland Stigge  <stigge@antcom.de>
10086             Michael Meissner  <meissner@linux.vnet.ibm.com>
10088         PR 57386/target
10089         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
10090         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
10092 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
10094         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
10095         -mcpu.
10097 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
10099         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
10100         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
10101         rtx is const0_rtx or not.
10103 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
10105         PR target/58115
10106         * target-globals.c (save_target_globals): Remove this_fn_optab
10107         handling.
10108         * toplev.c: Include optabs.h.
10109         (target_reinit): Temporarily restore the global options if another
10110         set of options are in force.
10112 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
10114         PR rtl-optimization/58668
10115         * cfgcleanup.c (flow_find_cross_jump): Don't count
10116         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
10117         to determine what is counted.
10118         (flow_find_head_matching_sequence): Use active_insn_p to determine
10119         what is counted.
10120         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
10121         counting change.
10122         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
10123         determine what is counted.
10125         PR tree-optimization/59643
10126         * tree-predcom.c (split_data_refs_to_components): If one dr is
10127         read and one write, determine_offset fails and the write isn't
10128         in the bad component, just put the read into the bad component.
10130 2014-01-07  Mike Stump  <mikestump@comcast.net>
10131             Jakub Jelinek  <jakub@redhat.com>
10133         PR pch/59436
10134         * tree-core.h (struct tree_optimization_option): Change optabs
10135         type from unsigned char * to void *.
10136         * optabs.c (init_tree_optimization_optabs): Adjust
10137         TREE_OPTIMIZATION_OPTABS initialization.
10139 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
10141         PR target/59644
10142         * config/i386/i386.h (struct machine_function): Add
10143         no_drap_save_restore field.
10144         * config/i386/i386.c (ix86_save_reg): Use
10145         !cfun->machine->no_drap_save_restore instead of
10146         crtl->stack_realign_needed.
10147         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
10148         this function clears frame_pointer_needed.  Set
10149         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
10150         and DRAP reg is needed.
10152 2014-01-06  Marek Polacek  <polacek@redhat.com>
10154         PR c/57773
10155         * doc/implement-c.texi: Mention that other integer types are
10156         permitted as bit-field types in strictly conforming mode.
10158 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
10160         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
10161         is newly allocated.
10163 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
10165         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
10167 2014-01-06  Martin Jambor  <mjambor@suse.cz>
10169         PR ipa/59008
10170         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
10171         to int.
10172         * ipa-prop.c (ipa_print_node_params): Fix indentation.
10174 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
10176         PR debug/59350
10177         PR debug/59510
10178         * var-tracking.c (add_stores): Preserve the value of the source even if
10179         we don't record the store.
10181 2014-01-06  Terry Guo  <terry.guo@arm.com>
10183         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
10185 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
10187         PR bootstrap/59541
10188         * config/darwin.c (darwin_function_section): Adjust return values to
10189         correspond to optimisation changes made in r206070.
10191 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
10193         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
10194         from prefetch_block tune setting.
10195         (nocona_cost): Correct size of prefetch block to 64.
10197 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
10199         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
10200         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
10201         used to save the static chain register in the computation of the offset
10202         from which the FP registers need to be restored.
10204 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
10206         PR tree-optimization/59519
10207         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
10208         ICE if get_current_def (current_new_name) is already non-NULL, as long
10209         as it is a phi result of some other phi in *new_exit_bb that has
10210         the same argument.
10212         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
10213         or vmovdqu* for misaligned_operand.
10214         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
10215         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
10216         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
10217         aligned_mem for AVX512F masked aligned load and store builtins and for
10218         non-temporal moves.
10220 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
10222         PR tree-optimization/59651
10223         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
10224         Address range for negative step should be added by TYPE_SIZE_UNIT.
10226 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
10228         * config/m68k/m68k.c (handle_move_double): Handle pushes with
10229         overlapping registers also for registers other than the stack pointer.
10231 2014-01-03  Marek Polacek  <polacek@redhat.com>
10233         PR other/59661
10234         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
10235         __builtin_FILE.
10237 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
10239         PR target/59625
10240         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
10241         asm goto as jump.
10243         * config/i386/i386.md (MODE_SIZE): New mode attribute.
10244         (push splitter): Use <P:MODE_SIZE> instead of
10245         GET_MODE_SIZE (<P:MODE>mode).
10246         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
10247         (mov -1, reg peephole2): Likewise.
10248         * config/i386/sse.md (*mov<mode>_internal,
10249         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
10250         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
10251         *<code><mode>3, *andnot<mode>3<mask_name>,
10252         <mask_codefor><code><mode>3<mask_name>): Likewise.
10253         * config/i386/subst.md (mask_mode512bit_condition,
10254         sd_mask_mode512bit_condition): Likewise.
10256 2014-01-02  Xinliang David Li  <davidxl@google.com>
10258         PR tree-optimization/59303
10259         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
10260         (dump_predicates): Better output format.
10261         (pred_equal_p): New function.
10262         (is_neq_relop_p): Ditto.
10263         (is_neq_zero_form_p): Ditto.
10264         (pred_expr_equal_p): Ditto.
10265         (pred_neg_p): Ditto.
10266         (simplify_pred): Ditto.
10267         (simplify_preds_2): Ditto.
10268         (simplify_preds_3): Ditto.
10269         (simplify_preds_4): Ditto.
10270         (simplify_preds): Ditto.
10271         (push_pred): Ditto.
10272         (push_to_worklist): Ditto.
10273         (get_pred_info_from_cmp): Ditto.
10274         (is_degenerated_phi): Ditto.
10275         (normalize_one_pred_1): Ditto.
10276         (normalize_one_pred): Ditto.
10277         (normalize_one_pred_chain): Ditto.
10278         (normalize_preds): Ditto.
10279         (normalize_cond_1): Remove function.
10280         (normalize_cond): Ditto.
10281         (is_gcond_subset_of): Ditto.
10282         (is_subset_of_any): Ditto.
10283         (is_or_set_subset_of): Ditto.
10284         (is_and_set_subset_of): Ditto.
10285         (is_norm_cond_subset_of): Ditto.
10286         (pred_chain_length_cmp): Ditto.
10287         (convert_control_dep_chain_into_preds): Type change.
10288         (find_predicates): Ditto.
10289         (find_def_preds): Ditto.
10290         (destroy_predicates_vecs): Ditto.
10291         (find_matching_predicates_in_rest_chains): Ditto.
10292         (use_pred_not_overlap_with_undef_path_pred): Ditto.
10293         (is_pred_expr_subset): Ditto.
10294         (is_pred_chain_subset_of): Ditto.
10295         (is_included_in): Ditto.
10296         (is_superset_of): Ditto.
10298 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
10300         Update copyright years.
10302 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
10304         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
10305         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
10306         config/arc/arc.md, config/arc/arc.opt,
10307         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
10308         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
10309         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
10310         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
10311         config/linux-protos.h, config/linux.c, config/winnt-c.c,
10312         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
10313         vtable-verify.c, vtable-verify.h: Use the standard form for the
10314         copyright notice.
10316 2014-01-02  Tobias Burnus  <burnus@net-b.de>
10318         * gcc.c (process_command): Update copyright notice dates.
10319         * gcov-dump.c: Ditto.
10320         * gcov.c: Ditto.
10321         * doc/cpp.texi: Bump @copying's copyright year.
10322         * doc/cppinternals.texi: Ditto.
10323         * doc/gcc.texi: Ditto.
10324         * doc/gccint.texi: Ditto.
10325         * doc/gcov.texi: Ditto.
10326         * doc/install.texi: Ditto.
10327         * doc/invoke.texi: Ditto.
10329 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
10331         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
10333 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
10335         * config/i386/sse.md (*mov<mode>_internal): Guard
10336         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
10338         PR rtl-optimization/59647
10339         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
10340         new_rtx into UNSIGNED_FLOAT rtxes.
10342 Copyright (C) 2014 Free Software Foundation, Inc.
10344 Copying and distribution of this file, with or without modification,
10345 are permitted in any medium without royalty provided the copyright
10346 notice and this notice are preserved.