2013-09-06 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blob44ddc1a61f7905d684e7ce1072e23156cea9b6c6
1 2013-09-06  Richard Biener  <rguenther@suse.de>
3         * basic-block.h (class control_dependences): New.
4         * tree-ssa-dce.c (control_dependence_map): Remove.
5         (cd): New global.
6         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
7         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
8         find_pdom, find_control_dependence, find_all_control_dependences):
9         Move to cfganal.c.
10         (mark_control_dependent_edges_necessary, find_obviously_necessary_stmts,
11         propagate_necessity, tree_dce_init, tree_dce_done,
12         perform_tree_ssa_dce): Adjust.
13         * cfganal.c (set_control_dependence_map_bit,
14         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
15         find_all_control_dependences): Move from tree-ssa-dce.c and
16         implement as methods of control_dependences class.
17         (control_dependences::control_dependences): New.
18         (control_dependences::~control_dependences): Likewise.
19         (control_dependences::get_edges_dependent_on): Likewise.
20         (control_dependences::get_edge): Likewise.
22 2013-09-04  Jan Hubicka  <jh@suse.cz>
24         * tree.c (types_same_for_odr): Drop overactive check.
25         * ipa-devirt.c (hash_type_name): Likewise.
27 2013-09-04  Jan Hubicka  <jh@suse.cz>
29         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
30         (analyze_functions): ... here.
32 2013-09-04  Jan Hubicka  <jh@suse.cz>
34         PR middle-end/58201
35         * cgraphunit.c (analyze_functions): Clear AUX fields
36         after processing; initialize assembler name has.
38 2013-09-05  Jeff Law  <law@redhat.com>
40         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
41         from thread_around_empty_block.  Record threading path into PATH.
42         Recurse if threading through the initial block is successful.
43         (thread_across_edge): Corresponding changes to slightly simplify.
45 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
47         * config/aarch64/aarch64.md
48         (type): Remove frecpe, frecps, frecpx.
49         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
50         fix to be a TARGET_SIMD instruction.
51         (aarch64_frecps): Remove.
52         * config/aarch64/aarch64-simd.md
53         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
54         (aarch64_frecps<mode>): Handle all float/vector of float modes.
56 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
57             Sofiane Naci  <sofiane.naci@arm.com>
59         * config/arm/types.md (define_attr "type"):
60         Expand "arlo_imm"
61         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
62         Expand "arlo_reg"
63         into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm", "alu_ext",
64         "alu_reg", "alus_ext", "alus_reg", "bfm", "csel", "logic_reg",
65         "logics_reg", "rev".
66         Expand "arlo_shift"
67         into "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
68         "logics_shift_imm".
69         Expand "arlo_shift_reg"
70         into "alu_shift_reg", "alus_shift_reg", "logic_shift_reg",
71         "logics_shift_reg".
72         Expand "clz" into "clz, "rbit".
73         Rename "shift" to "shift_imm".
74         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
75         changes.
76         Update for attribute changes all occurrences of arlo_* and
77         shift* types.
78         * config/arm/arm-fixed.md: Update for attribute changes
79         all occurrences of arlo_* types.
80         * config/arm/thumb2.md: Update for attribute changes all occurrences
81         of arlo_* types.
82         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
83         (cortexa7_older_only): Likewise.
84         (cortexa7_younger):  Likewise.
85         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
86         (1020alu_shift_op): Likewise.
87         (1020alu_shift_reg_op): Likewise.
88         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
89         (alu_shift_op): Likewise.
90         (alu_shift_reg_op): Likewise.
91         * config/arm/arm1136jfs.md (11_alu_op): Update for
92         attribute changes.
93         (11_alu_shift_op): Likewise.
94         (11_alu_shift_reg_op): Likewise.
95         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
96         (9_alu_shift_reg_op): Likewise.
97         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
98         attribute changes.
99         (cortex_a15_alu_shift): Likewise.
100         (cortex_a15_alu_shift_reg): Likewise.
101         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
102         attribute changes.
103         (cortex_a5_alu_shift): Likewise.
104         * config/arm/cortex-a53.md
105         (cortex_a53_alu): Update for attribute changes.
106         (cortex_a53_alu_shift): Likewise.
107         * config/arm/cortex-a7.md
108         (cortex_a7_alu_imm): Update for attribute changes.
109         (cortex_a7_alu_reg): Likewise.
110         (cortex_a7_alu_shift): Likewise.
111         * config/arm/cortex-a8.md
112         (cortex_a8_alu): Update for attribute changes.
113         (cortex_a8_alu_shift): Likewise.
114         (cortex_a8_alu_shift_reg): Likewise.
115         * config/arm/cortex-a9.md
116         (cortex_a9_dp): Update for attribute changes.
117         (cortex_a9_dp_shift): Likewise.
118         * config/arm/cortex-m4.md
119         (cortex_m4_alu): Update for attribute changes.
120         * config/arm/cortex-r4.md
121         (cortex_r4_alu): Update for attribute changes.
122         (cortex_r4_mov): Likewise.
123         (cortex_r4_alu_shift_reg): Likewise.
124         * config/arm/fa526.md
125         (526_alu_op): Update for attribute changes.
126         (526_alu_shift_op): Likewise.
127         * config/arm/fa606te.md
128         (606te_alu_op): Update for attribute changes.
129         * config/arm/fa626te.md
130         (626te_alu_op): Update for attribute changes.
131         (626te_alu_shift_op): Likewise.
132         * config/arm/fa726te.md
133         (726te_alu_op): Update for attribute changes.
134         (726te_alu_shift_op): Likewise.
135         (726te_alu_shift_reg_op): Likewise.
136         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
137         (mp626_alu_shift_op): Likewise.
138         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
139         (pj4_alu_conds): Likewise.
140         (pj4_shift): Likewise.
141         (pj4_shift_conds): Likewise.
142         (pj4_alu_shift): Likewise.
143         (pj4_alu_shift_conds): Likewise.
144         * config/aarch64/aarch64.md: Update for attribute change
145         all occurrences of arlo_* and shift* types.
147 2013-09-05  Mike Stump  <mikestump@comcast.net>
149         * tree.h: Move documentation for tree_function_decl to tree-core.h
150         with the declaration.
152 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
154         PR target/58139
155         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
156         looking for widest mode.
158 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
160         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
162 2013-09-05  Richard Biener  <rguenther@suse.de>
164         PR tree-optimization/58137
165         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
166         Do not create vectors of pointers.
167         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
168         types for the components of the vector initializer.
169         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
170         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
172 2013-09-05  Martin Jambor  <mjambor@suse.cz>
174         * ipa-prop.c (remove_described_reference): Accept missing references,
175         return false if that hppens, otherwise return true.
176         (cgraph_node_for_jfunc): New function.
177         (try_decrement_rdesc_refcount): Likewise.
178         (try_make_edge_direct_simple_call): Use them.
179         (ipa_edge_removal_hook): Remove references from rdescs.
180         (ipa_edge_duplication_hook): Clone rdescs and their references
181         when the new edge has the same caller as the old one.
182         * cgraph.c (cgraph_resolve_speculation): Remove speculative
183         reference before removing any edges.
185 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
187         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
188         initial store.
189         * thumb2.md (thumb2_storewb_parisi): New pattern.
191 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
193         * config/aarch64/aarch64-option-extensions.def: Add
194         AARCH64_OPT_EXTENSION of 'crc'.
195         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
196         (AARCH64_ISA_CRC): Ditto.
197         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
198         description of the CRC extension.
200 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
202         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
203         * config/rs6000/linux.h: Ditto.
204         * alpha/linux.h: Ditto.
205         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
206         no_c99_libc_has_function.
207         * config/c6x/uclinux-elf.h: Ditto.
208         * config/lm32/uclinux-elf.h: Ditto.
209         * config/m68k/uclinux.h: Ditto.
210         * config/moxie/uclinux.h: Ditto.
211         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
212         (crisv32-*-linux*, cris-*-linux*): Ditto.
213         * config/bfin/bfin.c: Include "tm_p.h".
215 2013-09-05  Richard Biener  <rguenther@suse.de>
217         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
218         check for a definition without a basic-block.
220 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
221             Sofiane Naci <sofiane.naci@arm.com>
223         * config/aarch64/aarch64.md
224         (*movti_aarch64): Rename r_2_f and f_2_r.
225         (*movsf_aarch64): Likewise.
226         (*movdf_aarch64): Likewise.
227         (*movtf_aarch64): Likewise.
228         (aarch64_movdi_<mode>low): Likewise.
229         (aarch64_movdi_<mode>high): Likewise.
230         (aarch64_mov<mode>high_di): Likewise.
231         (aarch64_mov<mode>low_di): Likewise.
232         (aarch64_movtilow_tilow): Likewise.
233         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
234         values to config/arm/types.md
235         (attribute "conds"): Update for attribute change.
236         (anddi3_insn): Likewise.
237         (iordi3_insn): Likewise.
238         (xordi3_insn): Likewise.
239         (one_cmpldi2): Likewise.
240         * config/arm/types.md (type): Add Neon types.
241         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
242         use "type" attribute.
243         (movmisalign<mode>_neon_store): Likewise.
244         (movmisalign<mode>_neon_load): Likewise.
245         (vec_set<mode>_internal): Likewise.
246         (vec_setv2di_internal): Likewise.
247         (vec_extract<mode>): Likewise.
248         (vec_extractv2di): Likewise.
249         (add<mode>3_neon): Likewise.
250         (adddi3_neon): Likewise.
251         (sub<mode>3_neon): Likewise.
252         (subdi3_neon): Likewise.
253         (mul<mode>3_neon): Likewise.
254         (mul<mode>3add<mode>_neon): Likewise.
255         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
256         (fma<VCVTF:mode>4)): Likewise.
257         (fma<VCVTF:mode>4_intrinsic): Likewise.
258         (fmsub<VCVTF:mode>4)): Likewise.
259         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
260         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
261         (ior<mode>3): Likewise.
262         (and<mode>3): Likewise.
263         (anddi3_neon): Likewise.
264         (orn<mode>3_neon): Likewise.
265         (orndi3_neon): Likewise.
266         (bic<mode>3_neon): Likewise.
267         (bicdi3_neon): Likewise.
268         (xor<mode>3): Likewise.
269         (one_cmpl<mode>2): Likewise.
270         (abs<mode>2): Likewise.
271         (neg<mode>2): Likewise.
272         (umin<mode>3_neon): Likewise.
273         (umax<mode>3_neon): Likewise.
274         (smin<mode>3_neon): Likewise.
275         (smax<mode>3_neon): Likewise.
276         (vashl<mode>3): Likewise.
277         (vashr<mode>3_imm): Likewise.
278         (vlshr<mode>3_imm): Likewise.
279         (ashl<mode>3_signed): Likewise.
280         (ashl<mode>3_unsigned): Likewise.
281         (neon_load_count): Likewise.
282         (ashldi3_neon_noclobber): Likewise.
283         (signed_shift_di3_neon): Likewise.
284         (unsigned_shift_di3_neon): Likewise.
285         (ashrdi3_neon_imm_noclobber): Likewise.
286         (lshrdi3_neon_imm_noclobber): Likewise.
287         (widen_ssum<mode>3): Likewise.
288         (widen_usum<mode>3): Likewise.
289         (quad_halves_<code>v4si): Likewise.
290         (quad_halves_<code>v4sf): Likewise.
291         (quad_halves_<code>v8hi): Likewise.
292         (quad_halves_<code>v16qi): Likewise.
293         (reduc_splus_v2di): Likewise.
294         (neon_vpadd_internal<mode>): Likewise.
295         (neon_vpsmin<mode>): Likewise.
296         (neon_vpsmax<mode>): Likewise.
297         (neon_vpumin<mode>): Likewise.
298         (neon_vpumax<mode>): Likewise.
299         (ss_add<mode>_neon): Likewise.
300         (us_add<mode>_neon): Likewise.
301         (ss_sub<mode>_neon): Likewise.
302         (us_sub<mode>_neon): Likewise.
303         (neon_vadd<mode>_unspec): Likewise.
304         (neon_vaddl<mode>): Likewise.
305         (neon_vaddw<mode>): Likewise.
306         (neon_vhadd<mode>): Likewise.
307         (neon_vqadd<mode>): Likewise.
308         (neon_vaddhn<mode>): Likewise.
309         (neon_vmul<mode>): Likewise.
310         (neon_vmla<mode>): Likewise.
311         (neon_vmlal<mode>): Likewise.
312         (neon_vmls<mode>): Likewise.
313         (neon_vmlsl<mode>): Likewise.
314         (neon_vqdmulh<mode>): Likewise.
315         (neon_vqdmlal<mode>): Likewise.
316         (neon_vqdmlsl<mode>): Likewise.
317         (neon_vmull<mode>): Likewise.
318         (neon_vqdmull<mode>): Likewise.
319         (neon_vsub<mode>_unspec): Likewise.
320         (neon_vsubl<mode>): Likewise.
321         (neon_vsubw<mode>): Likewise.
322         (neon_vqsub<mode>): Likewise.
323         (neon_vhsub<mode>): Likewise.
324         (neon_vsubhn<mode>): Likewise.
325         (neon_vceq<mode>): Likewise.
326         (neon_vcge<mode>): Likewise.
327         (neon_vcgeu<mode>): Likewise.
328         (neon_vcgt<mode>): Likewise.
329         (neon_vcgtu<mode>): Likewise.
330         (neon_vcle<mode>): Likewise.
331         (neon_vclt<mode>): Likewise.
332         (neon_vcage<mode>): Likewise.
333         (neon_vcagt<mode>): Likewise.
334         (neon_vtst<mode>): Likewise.
335         (neon_vabd<mode>): Likewise.
336         (neon_vabdl<mode>): Likewise.
337         (neon_vaba<mode>): Likewise.
338         (neon_vabal<mode>): Likewise.
339         (neon_vmax<mode>): Likewise.
340         (neon_vmin<mode>): Likewise.
341         (neon_vpaddl<mode>): Likewise.
342         (neon_vpadal<mode>): Likewise.
343         (neon_vpmax<mode>): Likewise.
344         (neon_vpmin<mode>): Likewise.
345         (neon_vrecps<mode>): Likewise.
346         (neon_vrsqrts<mode>): Likewise.
347         (neon_vqabs<mode>): Likewise.
348         (neon_vqneg<mode>): Likewise.
349         (neon_vcls<mode>): Likewise.
350         (clz<mode>2): Likewise.
351         (popcount<mode>2): Likewise.
352         (neon_vrecpe): Likewise.
353         (neon_vrsqrte): Likewise.
354         (neon_vget_lane<mode>_sext_internal): Likewise.
355         (neon_vget_lane<mode>_zext_internal): Likewise.
356         (neon_vdup_n<mode>): Likewise.
357         (neon_vdup_nv2di): Likewise.
358         (neon_vdpu_lane<mode>_internal): Likewise.
359         (neon_vswp<mode>): Likewise.
360         (float<mode><V_cvtto>2): Likewise.
361         (floatuns<mode><V_cvtto>2): Likewise.
362         (fix_trunc<mode><V_cvtto>)2): Likewise
363         (fixuns_trunc<mode><V_cvtto)2): Likewise.
364         (neon_vcvt<mode>): Likewise.
365         (neon_vcvtv4sfv4hf): Likewise.
366         (neon_vcvtv4hfv4sf): Likewise.
367         (neon_vcvt_n<mode>): Likewise.
368         (neon_vmovn<mode>): Likewise.
369         (neon_vqmovn<mode>): Likewise.
370         (neon_vqmovun<mode>): Likewise.
371         (neon_vmovl<mode>): Likewise.
372         (neon_vmul_lane<mode>): Likewise.
373         (neon_vmull_lane<mode>): Likewise.
374         (neon_vqdmull_lane<mode>): Likewise.
375         (neon_vqdmulh_lane<mode>): Likewise.
376         (neon_vmla_lane<mode>): Likewise.
377         (neon_vmlal_lane<mode>): Likewise.
378         (neon_vqdmlal_lane<mode>): Likewise.
379         (neon_vmls_lane<mode>): Likewise.
380         (neon_vmlsl_lane<mode>): Likewise.
381         (neon_vqdmlsl_lane<mode>): Likewise.
382         (neon_vext<mode>): Likewise.
383         (neon_vrev64<mode>): Likewise.
384         (neon_vrev32<mode>): Likewise.
385         (neon_vrev16<mode>): Likewise.
386         (neon_vbsl<mode>_internal): Likewise.
387         (neon_vshl<mode>): Likewise.
388         (neon_vqshl<mode>): Likewise.
389         (neon_vshr_n<mode>): Likewise.
390         (neon_vshrn_n<mode>): Likewise.
391         (neon_vqshrn_n<mode>): Likewise.
392         (neon_vqshrun_n<mode>): Likewise.
393         (neon_vshl_n<mode>): Likewise.
394         (neon_vqshl_n<mode>): Likewise.
395         (neon_vqshlu_n<mode>): Likewise.
396         (neon_vshll_n<mode>): Likewise.
397         (neon_vsra_n<mode>): Likewise.
398         (neon_vsri_n<mode>): Likewise.
399         (neon_vsli_n<mode>): Likewise.
400         (neon_vtbl1v8qi): Likewise.
401         (neon_vtbl2v8qi): Likewise.
402         (neon_vtbl3v8qi): Likewise.
403         (neon_vtbl4v8qi): Likewise.
404         (neon_vtbx1v8qi): Likewise.
405         (neon_vtbx2v8qi): Likewise.
406         (neon_vtbx3v8qi): Likewise.
407         (neon_vtbx4v8qi): Likewise.
408         (neon_vtrn<mode>_internal): Likewise.
409         (neon_vzip<mode>_internal): Likewise.
410         (neon_vuzp<mode>_internal): Likewise.
411         (neon_vld1<mode>): Likewise.
412         (neon_vld1_lane<mode>): Likewise.
413         (neon_vld1_dup<mode>): Likewise.
414         (neon_vld1_dupv2di): Likewise.
415         (neon_vst1<mode>): Likewise.
416         (neon_vst1_lane<mode>): Likewise.
417         (neon_vld2<mode>): Likewise.
418         (neon_vld2_lane<mode>): Likewise.
419         (neon_vld2_dup<mode>): Likewise.
420         (neon_vst2<mode>): Likewise.
421         (neon_vst2_lane<mode>): Likewise.
422         (neon_vld3<mode>): Likewise.
423         (neon_vld3qa<mode>): Likewise.
424         (neon_vld3qb<mode>): Likewise.
425         (neon_vld3_lane<mode>): Likewise.
426         (neon_vld3_dup<mode>): Likewise.
427         (neon_vst3<mode>): Likewise.
428         (neon_vst3qa<mode>): Likewise.
429         (neon_vst3qb<mode>): Likewise.
430         (neon_vst3_lane<mode>): Likewise.
431         (neon_vld4<mode>): Likewise.
432         (neon_vld4qa<mode>): Likewise.
433         (neon_vld4qb<mode>): Likewise.
434         (neon_vld4_lane<mode>): Likewise.
435         (neon_vld4_dup<mode>): Likewise.
436         (neon_vst4<mode>): Likewise.
437         (neon_vst4qa<mode>): Likewise.
438         (neon_vst4qb<mode>): Likewise.
439         (neon_vst4_lane<mode>): Likewise.
440         (neon_vec_unpack<US>_lo_<mode>): Likewise.
441         (neon_vec_unpack<US>_hi_<mode>): Likewise.
442         (neon_vec_<US>mult_lo_<mode>): Likewise.
443         (neon_vec_<US>mult_hi_<mode>): Likewise.
444         (neon_vec_<US>shiftl_<mode>): Likewise.
445         (neon_unpack<US>_<mode>): Likewise.
446         (neon_vec_<US>mult_<mode>): Likewise.
447         (vec_pack_trunc_<mode>): Likewise.
448         (neon_vec_pack_trunk_<mode>): Likewise.
449         (neon_vabd<mode>_2): Likewise.
450         (neon_vabd<mode>_3): Likewise.
451         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
452         (thumb2_movsi_vfp): Likewise.
453         (movdi_vfp): Likewise.
454         (movdi_vfp_cortexa8): Likewise.
455         (movhf_vfp_neon): Likewise.
456         (movhf_vfp): Likewiwse.
457         (movsf_vfp): Likewiwse.
458         (thumb2_movsf_vfp): Likewiwse.
459         (movdf_vfp): Likewise.
460         (thumb2_movdf_vfp): Likewise.
461         (movsfcc_vfp): Likewise.
462         (thumb2_movsfcc_vfp): Likewise.
463         (movdfcc_vfp): Likewise.
464         (thumb2_movdfcc_vfp): Likewise.
465         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
466         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
467         (v10_v2c): Likewise.
468         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
469         attribute change.
470         (cortex_a15_neon_int_2): Likewise.
471         (cortex_a15_neon_int_3): Likewise.
472         (cortex_a15_neon_int_4): Likewise.
473         (cortex_a15_neon_int_5): Likewise.
474         (cortex_a15_neon_vqneg_vqabs): Likewise.
475         (cortex_a15_neon_vmov): Likewise.
476         (cortex_a15_neon_vaba): Likewise.
477         (cortex_a15_neon_vaba_qqq): Likewise.
478         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
479         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
480         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
481         scalar_64_32_long_scalar): Likewise.
482         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
483         (cortex_a15_neon_mla_qqq_8_16): Likewise.
484         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
485         lotype_qdd_64_32_long): Likewise.
486         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
487         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
488         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
489         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
490         (cortex_a15_neon_shift_1): Likewise.
491         (cortex_a15_neon_shift_2): Likewise.
492         (cortex_a15_neon_shift_3): Likewise.
493         (cortex_a15_neon_vshl_ddd): Likewise.
494         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
495         (cortex_a15_neon_vsra_vrsra): Likewise.
496         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
497         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
498         (cortex_a15_neon_fp_vmul_ddd): Likewise.
499         (cortex_a15_neon_fp_vmul_qqd): Likewise.
500         (cortex_a15_neon_fp_vmla_ddd): Likewise.
501         (cortex_a15_neon_fp_vmla_qqq): Likewise.
502         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
503         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
504         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
505         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
506         (cortex_a15_neon_bp_simple): Likewise.
507         (cortex_a15_neon_bp_2cycle): Likewise.
508         (cortex_a15_neon_bp_3cycle): Likewise.
509         (cortex_a15_neon_vld1_1_2_regs): Likewise.
510         (cortex_a15_neon_vld1_3_4_regs): Likewise.
511         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
512         (cortex_a15_neon_vld2_4_regs): Likewise.
513         (cortex_a15_neon_vld3_vld4): Likewise.
514         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
515         (cortex_a15_neon_vst1_3_4_regs): Likewise.
516         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
517         (cortex_a15_neon_vst3_vst4): Likewise.
518         (cortex_a15_neon_vld1_vld2_lane): Likewise.
519         (cortex_a15_neon_vld3_vld4_lane" 10
520         (cortex_a15_neon_vst1_vst2_lane): Likewise.
521         (cortex_a15_neon_vst3_vst4_lane): Likewise.
522         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
523         (cortex_a15_neon_ldm_2): Likewise.0
524         (cortex_a15_neon_stm_2): Likewise.
525         (cortex_a15_neon_mcr): Likewise.
526         (cortex_a15_neon_mcr_2_mcrr): Likewise.
527         (cortex_a15_neon_mrc): Likewise.
528         (cortex_a15_neon_mrrc): Likewise.
529         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
530         change.
531         (cortex_a15_alu_shift): Likewise.
532         (cortex_a15_alu_shift_reg): Likewise.
533         (cortex_a15_mult32): Likewise.
534         (cortex_a15_mult64): Likewise.
535         (cortex_a15_block): Likewise.
536         (cortex_a15_branch): Likewise.
537         (cortex_a15_load1): Likewise.
538         (cortex_a15_load3): Likewise.
539         (cortex_a15_store1): Likewise.
540         (cortex_a15_store3): Likewise.
541         (cortex_a15_call): Likewise.
542         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute
543         change.
544         (cortex_a5_f2r): Likewise.
545         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
546         change.
547         (cortex_a53_f2r): Likewise.
548         * config/arm/cortex-a7.md
549         (cortex_a7_branch): Update for attribute change.
550         (cortex_a7_call): Likewise.
551         (cortex_a7_alu_imm): Likewise.
552         (cortex_a7_alu_reg): Likewise.
553         (cortex_a7_alu_shift): Likewise.
554         (cortex_a7_mul): Likewise.
555         (cortex_a7_load1): Likewise.
556         (cortex_a7_store1): Likewise.
557         (cortex_a7_load2): Likewise.
558         (cortex_a7_store2): Likewise.
559         (cortex_a7_load3): Likewise.
560         (cortex_a7_store3): Likewise.
561         (cortex_a7_load4): Likewise.
562         (cortex_a7_store4): Likewise.
563         (cortex_a7_fpalu): Likewise.
564         (cortex_a7_fconst): Likewise.
565         (cortex_a7_fpmuls): Likewise.
566         (cortex_a7_neon_mul): Likewise.
567         (cortex_a7_fpmacs): Likewise.
568         (cortex_a7_neon_mla: Likewise.
569         (cortex_a7_fpmuld: Likewise.
570         (cortex_a7_fpmacd: Likewise.
571         (cortex_a7_fpfmad: Likewise.
572         (cortex_a7_fdivs: Likewise.
573         (cortex_a7_fdivd: Likewise.
574         (cortex_a7_r2f: Likewise.
575         (cortex_a7_f2r: Likewise.
576         (cortex_a7_f_flags: Likewise.
577         (cortex_a7_f_loads: Likewise.
578         (cortex_a7_f_loadd: Likewise.
579         (cortex_a7_f_stores: Likewise.
580         (cortex_a7_f_stored: Likewise.
581         (cortex_a7_neon): Likewise.
582         * config/arm/cortex-a8-neon.md
583         (cortex_a8_neon_mrc): Update for attribute change.
584         (cortex_a8_neon_mrrc): Likewise.
585         (cortex_a8_neon_int_1): Likewise.
586         (cortex_a8_neon_int_2): Likewise.
587         (cortex_a8_neon_int_3): Likewise.
588         (cortex_a8_neon_int_4): Likewise.
589         (cortex_a8_neon_int_5): Likewise.
590         (cortex_a8_neon_vqneg_vqabs): Likewise.
591         (cortex_a8_neon_vmov): Likewise.
592         (cortex_a8_neon_vaba): Likewise.
593         (cortex_a8_neon_vaba_qqq): Likewise.
594         (cortex_a8_neon_vsma): Likewise.
595         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
596         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
597         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
598         Likewise.
599         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
600         (cortex_a8_neon_mla_qqq_8_16): Likewise.
601         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
602         long_scalar_qdd_64_32_long): Likewise.
603         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
604         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
605         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
606         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
607         (cortex_a8_neon_shift_1): Likewise.
608         (cortex_a8_neon_shift_2): Likewise.
609         (cortex_a8_neon_shift_3): Likewise.
610         (cortex_a8_neon_vshl_ddd): Likewise.
611         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
612         (cortex_a8_neon_vsra_vrsra): Likewise.
613         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
614         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
615         (cortex_a8_neon_fp_vsum): Likewise.
616         (cortex_a8_neon_fp_vmul_ddd): Likewise.
617         (cortex_a8_neon_fp_vmul_qqd): Likewise.
618         (cortex_a8_neon_fp_vmla_ddd): Likewise.
619         (cortex_a8_neon_fp_vmla_qqq): Likewise.
620         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
621         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
622         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
623         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
624         (cortex_a8_neon_bp_simple): Likewise.
625         (cortex_a8_neon_bp_2cycle): Likewise.
626         (cortex_a8_neon_bp_3cycle): Likewise.
627         (cortex_a8_neon_ldr): Likewise.
628         (cortex_a8_neon_str): Likewise.
629         (cortex_a8_neon_vld1_1_2_regs): Likewise.
630         (cortex_a8_neon_vld1_3_4_regs): Likewise.
631         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
632         (cortex_a8_neon_vld2_4_regs): Likewise.
633         (cortex_a8_neon_vld3_vld4): Likewise.
634         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
635         (cortex_a8_neon_vst1_3_4_regs): Likewise.
636         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
637         (cortex_a8_neon_vst3_vst4): Likewise.
638         (cortex_a8_neon_vld1_vld2_lane): Likewise.
639         (cortex_a8_neon_vld3_vld4_lane): Likewise.
640         (cortex_a8_neon_vst1_vst2_lane): Likewise.
641         (cortex_a8_neon_vst3_vst4_lane): Likewise.
642         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
643         (cortex_a8_neon_mcr): Likewise.
644         (cortex_a8_neon_mcr_2_mcrr): Likewise.
645         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
646         change.
647         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
648         change.
649         (ca9_neon_mrrc): Likewise.
650         (cortex_a9_neon_int_1): Likewise.
651         (cortex_a9_neon_int_2): Likewise.
652         (cortex_a9_neon_int_3): Likewise.
653         (cortex_a9_neon_int_4): Likewise.
654         (cortex_a9_neon_int_5): Likewise.
655         (cortex_a9_neon_vqneg_vqabs): Likewise.
656         (cortex_a9_neon_vmov): Likewise.
657         (cortex_a9_neon_vaba): Likewise.
658         (cortex_a9_neon_vaba_qqq): Likewise.
659         (cortex_a9_neon_vsma): Likewise.
660         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
661         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
662         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
663         Likewise.
664         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
665         (cortex_a9_neon_mla_qqq_8_16): Likewise.
666         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
667         long_scalar_qdd_64_32_long): Likewise.
668         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
669         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
670         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
671         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
672         (cortex_a9_neon_shift_1): Likewise.
673         (cortex_a9_neon_shift_2): Likewise.
674         (cortex_a9_neon_shift_3): Likewise.
675         (cortex_a9_neon_vshl_ddd): Likewise.
676         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
677         (cortex_a9_neon_vsra_vrsra): Likewise.
678         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
679         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
680         (cortex_a9_neon_fp_vsum): Likewise.
681         (cortex_a9_neon_fp_vmul_ddd): Likewise.
682         (cortex_a9_neon_fp_vmul_qqd): Likewise.
683         (cortex_a9_neon_fp_vmla_ddd): Likewise.
684         (cortex_a9_neon_fp_vmla_qqq): Likewise.
685         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
686         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
687         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
688         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
689         (cortex_a9_neon_bp_simple): Likewise.
690         (cortex_a9_neon_bp_2cycle): Likewise.
691         (cortex_a9_neon_bp_3cycle): Likewise.
692         (cortex_a9_neon_ldr): Likewise.
693         (cortex_a9_neon_str): Likewise.
694         (cortex_a9_neon_vld1_1_2_regs): Likewise.
695         (cortex_a9_neon_vld1_3_4_regs): Likewise.
696         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
697         (cortex_a9_neon_vld2_4_regs): Likewise.
698         (cortex_a9_neon_vld3_vld4): Likewise.
699         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
700         (cortex_a9_neon_vst1_3_4_regs): Likewise.
701         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
702         (cortex_a9_neon_vst3_vst4): Likewise.
703         (cortex_a9_neon_vld1_vld2_lane): Likewise.
704         (cortex_a9_neon_vld3_vld4_lane): Likewise.
705         (cortex_a9_neon_vst1_vst2_lane): Likewise.
706         (cortex_a9_neon_vst3_vst4_lane): Likewise.
707         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
708         (cortex_a9_neon_mcr): Likewise.
709         (cortex_a9_neon_mcr_2_mcrr): Likewise.
710         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
711         (cortex_a9_fps): Likewise.
712         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
713         change.
714         (cortex_m4_fmuls): Likewise.
715         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
716         change.
717         (cortex_r4_mrc): Likewise.
718         * config/arm/iterators.md: Update comment referring to neon_type.
719         * config/arm/iwmmxt.md
720         (iwmmxt_arm_movdi): Update for attribute change.
721         (iwmmxt_movsi_insn): Likewise.
722         * config/arm/marvell-pj4.md
723         (pj4_vfp_to_core): Update for attribute change.
724         (pj4_core_to_vfp): Likewise.
725         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
726         attribute change.
727         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
728         (vfp_fstore): Likewise.
729         * doc/md.texi: Change references to neon_type to refer to type.
731 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
733         * tree.h (DECL_BUILT_IN): Fix typo in comment.
735 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
737         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
738         lglobl if not weak.
740 2013-09-04  Easwaran Raman  <eraman@google.com>
742         PR middle-end/57370
743         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
744         (build_and_add_sum): Use it.
745         (appears_later_in_bb): Simplify code.
747 2013-09-04  Teresa Johnson  <tejohnson@google.com>
749         * dumpfile.c (dump_finish): Don't close stderr/stdout.
751 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
753         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
755 2013-09-04  Jan Hubicka  <jh@suse.cz>
757         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
758         * ipa-devirt.c: Include diganostic.h
759         (odr_type_d): Add types and types_set.
760         (hash_type_name): Work for types with vtables during LTO.
761         (odr_hasher::remove): Fix comment; destroy types_set.
762         (add_type_duplicate): New function,
763         (get_odr_type): Use it.
764         (dump_type_inheritance_graph): Dump type duplicates.
765         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
766         graph.
767         * tree.c (types_same_for_odr): Give exact answers on types with
768         virtual tables.
770 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
772         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
773         explaining their differences.
775 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
777         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
779 2013-09-03  Jeff Law  <law@redhat.com>
781         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
782         when not threading through a joiner block.  Pass joiner/no joiner
783         state to register_jump_thread.
784         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
785         state from argument rather than implying on path length.
786         Dump the entire jump thread path into debugging dump.
787         * tree-flow.h (register_jump_thread): Update prototype.
789 2013-08-29  Xinliang David Li  <davidxl@google.com>
791         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
792         Remove a trivial gcc_assert.
794 2013-08-29  Xinliang David Li  <davidxl@google.com>
796         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
797         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
798         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
799         Delay base decl alignment adjustment.
800         * tree-vectorizer.c (vect_destroy_datarefs): New function.
801         * tree-vectorizer.h: New data structure.
802         (set_dr_misalignment): New function.
803         (dr_misalignment): Ditto.
804         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
805         (vectorizable_load): Ditto.
806         (ensure_base_align): New function.
807         (vectorize_loops): Add dbg_cnt support.
808         (execute_vect_slp): Ditto.
809         * dbgcnt.def: New debug counter.
810         * Makefile: New dependency.
812 2013-09-03  Meador Inge  <meadori@codesourcery.com>
814         Revert:
816         2013-08-30  Meador Inge  <meadori@codesourcery.com>
818         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
820 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
822         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
823         function descriptor.
825 2013-09-03  Richard Biener  <rguenther@suse.de>
827         * tree-affine.c (add_elt_to_tree): Fix association issue,
828         avoid useless converts and make sure to always return a
829         properly typed result.
831 2013-09-03  Richard Biener  <rguenther@suse.de>
833         PR middle-end/57656
834         * fold-const.c (negate_expr_p): Fix division case.
835         (negate_expr): Likewise.
837 2013-09-03  Richard Biener  <rguenther@suse.de>
839         PR lto/58285
840         * tree-streamer-out.c: Include tm.h.
841         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
843 2013-09-03  Jan Hubicka  <jh@suse.cz>
845         * tree-profile.c (tree_profiling): Cleanup CFG when done.
847 2013-09-03  Alan Modra  <amodra@gmail.com>
849         * config.gcc (powerpc*-*-linux*): Add support for little-endian
850         multilibs to big-endian target and vice versa.
851         * config/rs6000/t-linux64: Use := assignment on all vars.
852         (MULTILIB_EXTRA_OPTS): Remove fPIC.
853         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
854         * config/rs6000/t-linux64le: New file.
855         * config/rs6000/t-linux64bele: New file.
856         * config/rs6000/t-linux64lebe: New file.
858 2013-09-02  Jan Hubicka  <jh@suse.cz>
860         * ipa-inline-transform.c (inline_transform): Do not
861         optimize_inline_calls when not optimizing.
863 2013-09-02  Jan Hubicka  <jh@suse.cz>
865         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
866         duplicated nodes for assembler names.
867         * symtab.c (symtab_unregister_node): Do not attempt to unlink
868         hard registers from assembler name hash.
870 2013-09-02  Jan Hubicka  <jh@suse.cz>
872         * ipa-split.c (execute_split_functions): Split externally visible
873         functions called once.
875 2013-09-02  Martin Jambor  <mjambor@suse.cz>
877         PR ipa/58106
878         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
879         linked list.  When finding the correct duplicate, also consider also
880         the caller in additon to its inlined_to node.
882 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
884         * config/aarch64/aarch64-simd-builtins.def
885         (dup_lane_scalar): Remove.
886         * config/aarch64/aarch64-simd.md
887         (aarch64_simd_dup): Add 'w->w' alternative.
888         (aarch64_dup_lane<mode>): Allow for VALL.
889         (aarch64_dup_lane_scalar<mode>): Remove.
890         (aarch64_dup_lane_<vswap_width_name><mode>): New.
891         (aarch64_get_lane_signed<mode>): Add w->w altenative.
892         (aarch64_get_lane_unsigned<mode>): Likewise.
893         (aarch64_get_lane<mode>): Likewise.
894         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
895         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
896         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
897         (VCON): Change container of V2SF.
898         (vswap_width_name): Likewise.
899         * config/aarch64/arm_neon.h
900         (__aarch64_vdup_lane_any): New.
901         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
902         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
903         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
905 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
907         PR middle-end/56382
908         * expr.c (emit_move_complex): Do not move complex FP values as parts if
909         the source or the destination is a single hard register.
911 2013-09-02  Richard Biener  <rguenther@suse.de>
913         PR middle-end/57511
914         * tree-scalar-evolution.c (instantiate_scev_name): Allow
915         non-linear SCEVs.
917 2013-09-02  Richard Biener  <rguenther@suse.de>
919         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
920         arithmetic to sizetype.
922 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
924         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
925         Find auto-increment use both before and after candidate.
927 2013-09-02  Marek Polacek  <polacek@redhat.com>
929         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
931 2013-09-01  Jan Hubicka  <jh@suse.cz>
933         * Makefile.in: Add ipa-profile.o
934         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
935         * cgraph.c (struct cgraph_propagate_frequency_data,
936         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
937         ipa-profile.c; replace cgraph_ by ipa_ prefix.
938         * cgraph.h (cgraph_propagate_frequency): Remove.
939         * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
940         (inline_update_callee_summaries): Update.
941         * ipa-profile.c: New file.
942         * ipa-utils.h (ipa_propagate_frequency): Declare.
943         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
944         data-streamer.h, value-prof.h
945         (symtab_remove_unreachable_nodes): Update profile.
946         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
947         account_time_size, cmp_counts, dump_histogram,
948         ipa_profile_generate_summary, ipa_profile_write_summary,
949         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
950         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
951         Move to ipa-profile.c
953 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
955         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
957 2013-09-01  Jan Hubicka  <jh@suse.cz>
959         * common.opt (fdevirtualize-speculatively): New function.
960         * invoke.texi (fdevirtualize-speculatively): Document.
961         * ipa-devirt.c: Include ipa-inline.h
962         (likely_target_p): New function.
963         (ipa_devirt): New function.
964         (gate_ipa_devirt): New function.
965         (pass_data_ipa_devirt): New static var.
966         (pass_ipa_devirt): Likewise.
967         (make_pass_ipa_devirt): New function.
968         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
969         (common_handle_option): Disable devirtualization when
970         value range profiling is available.
971         * passes.def (pass_ipa_devirt): Add.
972         * timever.def (TV_IPA_DEVIRT): New timevar.
973         * tree-pass.h (make_pass_ipa_devirt): 
975 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
977         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
978         include sanitize(undefined).
980 2013-08-31  Diego Novillo  <dnovillo@google.com>
982         * Makefile.in (TREE_CORE_H): Define.
983         (TREE_H): Use.
984         (GTFILES): Add tree-core.h.
985         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
986         size the array.
987         * tree-core.h: New file.
988         Move all data structures, enum, typedefs, global
989         declarations and constants from ...
990         * tree.h: ... here.
992 2013-08-31  Jan Hubicka  <jh@suse.cz>
994         * bulitins.c (expand_builtin): Do not early exit for gcov
995         instrumented functions.
997 2013-08-31  Marek Polacek  <polacek@redhat.com>
999         * ubsan.c: Include tm_p.h.
1001 2013-08-31  Jan Hubicka  <jh@suse.cz>
1003         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
1004         warning.
1006         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
1007         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
1008         * tree-cfg.c (verify_gimple_label): ... here.
1009         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
1010         (ipa_merge_profiles): New function.
1011         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
1012         (lto_input_function_body): Likewise.
1013         * ipa-utils.h (ipa_merge_profiles): Declare.
1014         * lto-streamer.h (lto_input_function_body): Update prototype.
1015         (emit_label_in_global_context_p): Remove.
1016         * lto-symtab.c: Include ipa-utils.h
1017         (lto_cgraph_replace_node): Use ipa_merge_profiles.
1019 2013-08-31  Jan Hubicka  <jh@suse.cz>
1021         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
1023 2013-08-31  Jan Hubicka  <jh@suse.cz>
1025         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
1027 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
1029         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
1030         "cmp" RTX before signed_comparison_operator check to account
1031         for "code" changes.
1033 2013-08-30  Jan Hubicka  <jh@suse.cz>
1035         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
1036         (detect_type_change_1): Rename to ...
1037         (detect_type_change): ... this one; early return on non-polymorphic
1038         types.
1039         (detect_type_change_ssa): Add comp_type parameter; update
1040         use of detect_type_change.
1041         (compute_complex_assign_jump_func): Add param_type parameter;
1042         update use of detect_type_change_ssa.
1043         (compute_complex_ancestor_jump_func): Likewise.
1044         (ipa_get_callee_param_type): New function.
1045         (ipa_compute_jump_functions_for_edge): Compute parameter type;
1046         update calls to the jump function computation functions.
1048 2013-08-30  Teresa Johnson  <tejohnson@google.com>
1049             Steven Bosscher  <steven@gcc.gnu.org>
1051         * cfgrtl.c (fixup_new_cold_bb): New routine.
1052         (commit_edge_insertions): Invoke fixup_partitions.
1053         (find_partition_fixes): New routine.
1054         (fixup_partitions): Ditto.
1055         (verify_hot_cold_block_grouping): Update comments.
1056         (rtl_verify_edges): Invoke find_partition_fixes.
1057         (rtl_verify_bb_pointers): Update comments.
1058         (rtl_verify_bb_layout): Ditto.
1059         * basic-block.h (probably_never_executed_edge_p): Declare.
1060         (fixup_partitions): Ditto.
1061         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
1062         * bb-reorder.c (sanitize_hot_paths): New function.
1063         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
1064         sanitize_hot_paths.
1065         * predict.c (probably_never_executed_edge_p): New routine.
1066         * cfg.c (check_bb_profile): Add partition insanity warnings.
1068 2013-08-30  Meador Inge  <meadori@codesourcery.com>
1070         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
1072 2013-08-30  Marek Polacek  <polacek@redhat.com>
1074         * Makefile.in (ubsan.o): Add.
1075         (c-family/c-ubsan.o): Add.
1076         (builtins.o): Add ubsan.h dependency.
1077         * ubsan.h: New file.
1078         * ubsan.c: New file.
1079         * common.opt: Add -fsanitize=undefined option.
1080         (flag_sanitize): Add variable.
1081         (fsanitize=): Add option.  Add Driver.
1082         (fsanitize=thread): Remove option.
1083         (fsanitize=address): Likewise.
1084         (static-libubsan): New option.
1085         * doc/invoke.texi: Document the new flag and -static-libubsan.
1086         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
1087         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
1088         * builtin-attrs.def (ATTR_COLD): Define.
1089         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
1090         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
1091         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
1092         * flag-types.h (sanitize_code): New enum.
1093         * opts.c (common_handle_option): Parse command line arguments
1094         of -fsanitize=.  Add -fsanitize=unreachable option.
1095         * varasm.c (get_variable_section): Adjust.
1096         (assemble_noswitch_variable): Likewise.
1097         (assemble_variable): Likewise.
1098         (output_constant_def_contents): Likewise.
1099         (categorize_decl_for_section): Likewise.
1100         (place_block_symbol): Likewise.
1101         (output_object_block): Likewise.
1102         * builtins.def: Likewise.
1103         * toplev.c (compile_file): Likewise.
1104         (process_options): Likewise.
1105         * cppbuiltin.c: Likewise.
1106         * tsan.c (tsan_pass): Likewise.
1107         (tsan_gate): Likewise.
1108         (tsan_gate_O0): Likewise.
1109         * cfgexpand.c (partition_stack_vars): Likewise.
1110         (expand_stack_vars): Likewise.
1111         (defer_stack_allocation): Likewise.
1112         (expand_used_vars): Likewise.
1113         * cfgcleanup.c (old_insns_match_p): Likewise.
1114         * asan.c (asan_finish_file): Likewise.
1115         (asan_instrument): Likewise.
1116         (gate_asan): Likewise.
1117         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
1118         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
1119         (asan_global_struct): Use pointer_sized_int_node instead
1120         calling build_nonstandard_integer_type.
1121         (initialize_sanitizer_builtins): Likewise.
1122         (asan_finish_file): Likewise.
1123         * gcc.c: Document %{%:function(args):X}.
1124         (static_spec_functions): Add sanitize.
1125         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
1126         store funcval != NULL there.
1127         (do_spec_1): Adjust handle_spec_function caller.
1128         (handle_braces): Allow %:function(args) as condition.
1129         (sanitize_spec_function): New function.
1130         (ADD_STATIC_LIBUBSAN_LIBS): Define.
1131         (LIBUBSAN_SPEC): Likewise.
1132         (LIBUBSAN_EARLY_SPEC): Likewise.
1133         (SANITIZER_SPEC): Handle libubsan.
1134         (SANITIZER_EARLY_SPEC): Likewise.
1135         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
1136         instead of fsanitize=address.
1137         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
1138         instead of fsanitize=address*.
1139         * builtins.c: Include ubsan.h.
1140         (fold_builtin_0): Instrument __builtin_unreachable.
1141         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
1142         instead of flag_asan.
1143         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
1144         (pointer_sized_int_node): Define.
1145         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
1147 2013-08-30  Mike Stump  <mikestump@comcast.net>
1149         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
1150         with RE patterns.
1152 2013-08-29  Jan Hubicka  <jh@suse.cz>
1154         * cgraph.c (cgraph_function_body_availability): Handle weakref
1155         correctly.
1156         * passes.def: Remove pass_fixup_cfg.
1157         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
1158         track when we need to remove functions.
1159         (gate_ipa_inline): Execute inlining always; add comment why.
1160         (pass_data_ipa_inline): Remove TODO_remove_functions.
1161         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
1162         do not produce summaries.
1163         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
1164         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
1165         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
1166         constant pool and vtable.
1168 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
1170         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
1171         New arm_neon.h's internal macros to specify 64-bit constants.
1172         Avoid using stdint.h's macros.
1174 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
1176         * recog.c (verify_changes): Verify that changes[i].old is non-zero
1177         before applying REG_P.
1179 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
1181         PR tree-optimization/58277
1182         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
1183         after seeing too many stmts with vdef in between dombb and current
1184         bb, invalidate everything.
1186 2013-08-30  Richard Biener  <rguenther@suse.de>
1188         * fold-const.c (fold_single_bit_test): Fix overflow test.
1190 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
1192         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
1193         and which can live in a register, always retrieve the value on entry.
1194         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
1195         passed by invisible reference specially.
1196         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
1197         (vt_add_function_parameter): Correctly deal with a parameter passed by
1198         invisible reference.
1200 2013-08-30  Jan Hubicka  <jh@suse.cz>
1202         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
1204 2013-08-30  Richard Biener  <rguenther@suse.de>
1206         PR tree-optimization/58228
1207         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
1208         allow invariant loads in nested loop vectorization.
1210 2013-08-30  Richard Biener  <rguenther@suse.de>
1212         PR tree-optimization/58223
1213         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
1214         (has_anti_or_output_dependence): ... this and adjust to also
1215         look for output dependences.
1216         (mark_nodes_having_upstream_mem_writes): Adjust.
1217         (rdg_flag_uses): Likewise.
1219 2013-08-30  Richard Biener  <rguenther@suse.de>
1221         PR tree-optimization/58010
1222         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
1223         assert that we have a loop-closed PHI.
1225 2013-08-29  Jan Hubicka  <jh@suse.cz>
1227         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
1228         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
1229         * lto-section-in.c (lto_free_function_in_decl_state): New function.
1230         (lto_free_function_in_decl_state_for_node): New function.
1232 2013-08-29  Xinliang David Li  <davidxl@google.com>
1234         * loop-unroll.c (report_unroll_peel): Minor message change.
1235         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
1236         Emit alignment peeling message with default -fopt-info.
1237         (vect_loop_versioning): Emit loop version info message.
1238         * tree-vectorizer.c (vectorize_loops): Minor message change.
1239         (execute_vect_slp): Ditto.
1241 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
1243         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
1244         of the clone from the DECL_NAME of the original function.
1246 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
1248         * passes.c (register_pass): Add overload.
1249         * tree-pass.h (register_pass): Forward declare it.  Add comment.
1251 2013-08-29  Jan Hubicka  <jh@suse.cz>
1253         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
1254         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
1255         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
1256         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
1257         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
1258         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
1259         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
1260         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
1262 2013-08-29  Teresa Johnson  <tejohnson@google.com>
1264         * dumpfile.c (dump_loc): Output column number.
1265         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
1266         * doc/invoke.texi: Document optall -fopt-info flag.
1267         * profile.c (read_profile_edge_counts): Use new dump framework.
1268         (compute_branch_probabilities): Ditto.
1269         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
1270         when pass not in any opt group.
1271         * pass_manager.h (pass_manager::get_pass_profile): New method.
1272         * value-prof.c (check_counter): Use new dump framework.
1273         (check_ic_target): Ditto.
1274         * coverage.c (get_coverage_counts): Ditto.
1275         (coverage_init): Setup new dump framework.
1277 2013-08-29  Richard Biener  <rguenther@suse.de>
1279         PR tree-optimization/58246
1280         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
1281         handle the dominance check inside a basic-block.
1283 2013-08-29  Richard Biener  <rguenther@suse.de>
1285         PR middle-end/57287
1286         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
1287         of default defs that appear in abnormal PHI nodes.
1289 2013-08-29  Richard Biener  <rguenther@suse.de>
1291         PR tree-optimization/57685
1292         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
1293         single-use operands to avoid exponential complexity.
1295 2013-08-28  Dehao Chen  <dehao@google.com>
1297         * ipa-inline.c (edge_badness): Fix integer underflow.
1299 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
1301         * gtm-builtins.def (_ITM_free): Declare leaf.
1303 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
1305         PR target/58067
1306         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
1307         (*tls_local_dynamic_base_64_largepic): Likewise.
1308         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
1309         Remove predicate from call operand.
1310         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
1311         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
1313 2013-08-28  Jeff Law  <law@redhat.com>
1315         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
1316         checks for the number of predecessors and successors allowed.
1317         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
1318         which require copying a joiner block if there is a request which
1319         is a subpath that requires no joiner block copying.
1321 2013-08-28  Jan Hubicka  <jh@suse.cz>
1323         * lto-streamer-out.c (DFS_write_tree_body): Drop
1324         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
1325         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
1326         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
1327         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
1328         Do not read DECL_ERROR_ISSUED.
1329         (unpack_ts_decl_with_vis_value_fields): Do not read
1330         DECL_DEFER_OUTPUT.
1331         (lto_input_ts_binfo_tree_pointers): Do not read
1332         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
1333         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
1334         write DECL_ERROR_ISSUED..
1335         (pack_ts_decl_with_vis_value_fields): Do not write
1336         DECL_DEFER_OUTPUT.
1337         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
1338         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
1339         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
1340         * tree.h (tree_decl_common): Update comment.
1341         (DECL_ERROR_ISSUED): Remove.
1343 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
1345         PR middle-end/58257
1346         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
1348 2013-08-28  Jan Hubicka  <jh@suse.cz>
1350         * builtins.def (free): Declare leaf.
1352 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
1354         * gdbhooks.py: New.
1355         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
1356         * configure: Regenerate.
1358 2013-08-27  Martin Jambor  <mjambor@suse.cz>
1360         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
1361         (ipa_ancestor_jf_data): Likewise.
1362         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
1363         (ipa_get_jf_pass_through_type_preserved): New function.
1364         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
1365         (ipa_get_jf_ancestor_type_preserved): New function.
1366         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
1367         (ipa_get_jf_ancestor_result): Likewise.
1368         (propagate_vals_accross_pass_through): Use
1369         ipa_get_jf_pass_through_result to do all the value mappings.
1370         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
1371         type_preserved flag.
1372         (ipa_set_jf_cst_copy): New function.
1373         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
1374         (ipa_set_jf_arith_pass_through): Likewise.
1375         (ipa_set_ancestor_jf): Likewise.
1376         (compute_complex_assign_jump_func): Set type_preserved instead of
1377         punting.
1378         (ipa_compute_jump_functions_for_edge): Likewise.
1379         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
1380         (update_jump_functions_after_inlining): Update type_preserved.
1381         Explicitely create jump functions when combining one with pass_through.
1382         (ipa_write_jump_function): Stream the type_preserved flags.
1383         (ipa_read_jump_function): Likewise.
1385 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
1386             Aldy Hernandez  <aldyh@redhat.com>
1388         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
1389         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
1390         * function.h (struct function): Add has_force_vect_loops and
1391         has_simduid_loops.
1392         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
1393         * gimple.c (gimple_build_omp_critical): Add KIND argument and
1394         handle it.
1395         * gimple.def: Update CLAUSES comments.
1396         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
1397         (gimple_build_omp_for): Add argument to prototype.
1398         (gimple_omp_for_kind): New.
1399         (gimple_omp_for_set_kind): New.
1400         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
1401         GOVD_DATA_SHARE_CLASS.
1402         (enum omp_region_type): Add ORT_SIMD.
1403         (gimple_add_tmp_var): Handle ORT_SIMD.
1404         (gimplify_var_or_parm_decl): Same.
1405         (is_gimple_stmt): Same.
1406         (omp_firstprivatize_variable): Same.
1407         (omp_add_variable): Only use splay_tree_insert if lookup failed.
1408         (omp_notice_variable): Handle ORT_SIMD.
1409         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
1410         (omp_check_private): Handle ORT_SIMD.
1411         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
1412         OMP_CLAUSE_SAFELEN.
1413         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
1414         Handle OMP_CLAUSE_LASTPRIVATE.
1415         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
1416         OMP_CLAUSE_SAFELEN.
1417         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
1418         (gimplify_expr): Handle OMP_SIMD.
1419         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
1420         (expand_GOMP_SIMD_VF): New.
1421         (expand_GOMP_SIMD_LAST_LANE): New.
1422         * internal-fn.def (GOMP_SIMD_LANE): New.
1423         (GOMP_SIMD_VF): New.
1424         (GOMP_SIMD_LAST_LANE): New.
1425         * omp-low.c: Include target.h.
1426         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
1427         OMP_CLAUSE_SAFELEN.
1428         (check_omp_nesting_restrictions): Same.
1429         (omp_max_vf): New.
1430         (lower_rec_simd_input_clauses): New.
1431         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
1432         OMP_CLAUSE_LINEAR.
1433         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
1434         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
1435         (expand_omp_build_assign): New.
1436         (expand_omp_for_init_counts): New.
1437         (expand_omp_for_init_vars): New.
1438         (extract_omp_for_update_vars): New.
1439         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
1440         and rewrite accordingly.
1441         (expand_omp_simd): New.
1442         (expand_omp_for): Use expand_omp_simd.
1443         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
1444         (lower_omp_for): Do not lower the body.
1445         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
1446         in their own loops.
1447         * tree-flow.h (find_omp_clause): Remove prototype.
1448         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
1449         forcing vectorization of the loop, or if flag_tree_vectorize.
1450         (gate_tree_if_conversion): Similarly.
1451         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
1452         gimple_build_omp_for.
1453         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
1454         * tree-parloops (create_parallel_loop): Pass kind argument to
1455         gimple_build_omp_for.
1456         * tree-pretty-print.c (dump_omp_clause): Add cases for
1457         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
1458         OMP_CLAUSE__SIMDUID_.
1459         (dump_generic_node): Handle OMP_SIMD.
1460         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
1461         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
1462         unroll OMP_SIMD loops here.
1463         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
1464         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
1465         loop->safelen.
1466         (vect_analyze_data_refs): Handle simd loops.
1467         * tree-vect-loop.c (vectorizable_live_operation): Handle
1468         IFN_GOMP_SIMD*.
1469         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
1470         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
1471         (vectorizable_load): Same.
1472         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
1473         (struct simduid_to_vf): New.
1474         (simduid_to_vf::hash): New.
1475         (simduid_to-vf::equal): New.
1476         (struct simd_array_to_simduid): New.
1477         (simd_array_to_simduid::hash): New.
1478         (simd_array_to_simduid::equal): New.
1479         (adjust_simduid_builtins): New.
1480         (struct note_simd_array_uses_struct): New.
1481         (note_simd_array_uses_cb): New.
1482         (note_simd_array_uses): New.
1483         (vectorize_loops): Handle simd hints and adjust simd builtins
1484         accordingly.
1485         * tree-vectorizer.h (struct _stmt_vec_info): Add
1486         simd_lane_access_p field.
1487         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
1488         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
1489         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
1490         (omp_clause_code_name): Same.
1491         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
1492         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
1493         * tree.def (OMP_SIMD): New entry.
1494         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
1495         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
1496         (OMP_CLAUSE_DECL): Adjust range for new clauses.
1497         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
1498         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
1499         (OMP_CLAUSE_LINEAR_STEP): New.
1500         (OMP_CLAUSE_SAFELEN_EXPR): New.
1501         (OMP_CLAUSE__SIMDUID__DECL): New.
1502         (find_omp_clause): New prototype.
1504 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
1506         * config/i386/driver-i386.c (host_detect_local_cpu): Update
1507         Haswell processor detection.
1509 2013-08-27  Christian Widmer  <shadow@umbrox.de>
1511         PR target/57927
1512         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
1513         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
1514         AVX2 capable processors.
1516 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
1518         * config/aarch64/arm_neon.h: Replace all inline asm implementations
1519         of vget_low_* with implementations in terms of other intrinsics.
1521 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
1523         PR middle-end/57219
1524         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
1525         values to -1, 0 and 1.
1527 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
1529         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
1530         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
1531         (<optab><mode>3_insn): Remove.
1532         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
1533         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
1534         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
1535         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
1536         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
1537         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
1538         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
1539         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
1540         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
1541         (ror<mode>3_insn): Likewise.
1542         * config/aarch64/predicates.md (aarch64_simd_register): New.
1544 2013-08-27  Richard Biener  <rguenther@suse.de>
1546         PR tree-optimization/57521
1547         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
1548         one edge is non-critical.
1549         (find_phi_replacement_condition): Make sure to use a non-critical
1550         edge.  Cleanup and remove old bug workarounds.
1551         (bb_postdominates_preds): Remove.
1552         (if_convertible_loop_p_1): Do not compute post-dominators.
1553         (combine_blocks): Do not free post-dominators.
1554         (main_tree_if_conversion): Likewise.
1555         (pass_data_if_conversion): Add TODO_verify_ssa.
1557 2013-08-27  DJ Delorie  <dj@redhat.com>
1559         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
1561 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
1563         * function.c (assign_parm_find_data_types): Set passed_mode and
1564         nominal_mode to the TYPE_MODE of nominal_type for the built
1565         pointer type in case of the struct-pass-by-reference.
1567 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
1569         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
1570         (UINT16_TYPE): Change default to "unsigned int".
1572         * config/avr/avr.opt (mfract-convert-truncate): New option.
1573         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
1574         is set, round negative fractional integers according to n1169
1575         when converting to integer types.
1577 2013-08-26  Jan Hubicka  <jh@suse.cz>
1579         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
1580         methods can not be called indirectly when their address is not taken.
1582 2013-08-26  Jan Hubicka  <jh@suse.cz>
1584         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
1585         ctor_for_folding.
1587 2013-08-26  Jan Hubicka  <jh@suse.cz>
1589         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
1590         can be unshared.
1592 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
1594         * reload.c (find_valid_class): Allow classes that do not include
1595         FIRST_PSEUDO_REGISTER - 1.
1597 2013-08-26  Jan Hubicka  <jh@suse.cz>
1599         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
1600         fix edge count/frequency when speculation failed; fix type check
1601         for the direct call.
1603 2013-08-26  Jan Hubicka  <jh@suse.cz>
1605         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
1607 2013-08-26  Jan Hubicka  <jh@suse.cz>
1609         * ipa-inline-transform.c (inline_transform): Be ready for basic block
1610         to be changed by edge redirection.
1612 2013-08-26  Jan Hubicka  <jh@suse.cz>
1614         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
1615         formating; add sanity check.
1616         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
1617         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
1618         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
1619         (resolve_noninline_speculation): Update callee keys, too.
1621 2013-08-26  Jan Hubicka  <jh@suse.cz>
1623         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
1624         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
1626 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
1628         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
1629         into proper place.
1631 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
1633         * config/i386/i386.c (ix86_debug_options): Remove prototype.
1634         (x86_64_elf_select_section): Ditto.
1635         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
1636         arguments.
1637         (ix86_pass_by_reference): Ditto.
1638         (output_set_got): Ditto.
1639         (ix86_unary_operator_ok): Ditto.
1640         (ix86_expand_builtin): Ditto.
1642 2013-08-23  Jan Hubicka  <jh@suse.cz>
1644         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
1646 2013-08-23  Jan Hubicka  <jh@suse.cz>
1648         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
1649         (tree_decl_with_vis): Add FINAL field.
1651 2013-08-23  Jeff Law  <law@redhat.com>
1653         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
1654         the debugging dump when the expression is fully redundant.
1656 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1658         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
1659         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
1660         * pretty-print.c (pp_formatted_text_data): Likewise.
1661         (pp_write_text_to_stream): Likewise.
1662         (pp_write_text_as_dot_label_to_stream): Likewise.
1663         (pp_append_r): Likewise.
1664         (pp_format): Likewise.
1665         (pp_flush): Likewise.
1666         (pp_clear_output_area): Likewise.
1667         (pp_append_text): Likewise.
1668         (pp_formatted_text): Likewise.
1669         (pp_remaining_character_count_for_line): Likewise.
1670         (pp_newline): Likewise.
1671         (pp_character): Likewise.
1672         (output_buffer::~output_buffer): Define.
1673         (pretty_printer::~pretty_printer): Destruct output buffer.
1674         * pretty-print.h (output_buffer::~output_buffer): Declare.
1675         (pretty_printer::~pretty_printer): Declare virtual.
1677 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
1679         PR other/57324
1680         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
1681         HOST_WIDE_INT_M1U): New macros.
1682         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
1683         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
1684         unsigned -1 for lshift.
1685         * cse.c (cse_insn): Likewise.
1686         * double-int.c (rshift_double, lshift_double): Likewise.
1687         * builtins.c (fold_builtin_bitop): Likewise.
1688         * combine.c (force_to_mode): Likewise.
1689         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
1690         * simplify-rtx.c (simplify_const_unary_operation,
1691         simplify_const_binary_operation): Likewise.
1692         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
1693         check_va_list_escapes): Likewise.
1694         * rtlanal.c (nonzero_bits1): Likewise.
1695         * expmed.c (expand_smod_pow2): Likewise.
1696         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
1698 2013-08-23  Jan Hubicka  <jh@suse.cz>
1700         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
1701         as having address taken.
1703 2013-08-23  Jan Hubicka  <jh@suse.cz>
1705         * ipa-utils.h (method_class_type): Declare.
1706         * ipa-devirt.c (method_class_type): Export.
1708         * cgraphunit.c (analyze_functions): Do basic devirtualization;
1709         do not walk base classes of anonymous types.
1711 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
1713         PR rtl-optimization/58220
1714         PR regression/58221
1715         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
1716         handle SEQUENCE insns properly.
1718 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1720         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
1721         definition.
1722         (pp_newline_and_indent): Likewise.
1723         (pp_separate_with): Likewise.
1724         * pretty-print.c (pp_newline_and_flush): Define.
1725         (pp_newline_and_indent): Likewise.
1726         (pp_separate_with): Likewise.
1728 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
1730         PR target/58218
1731         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
1732         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
1734 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
1736         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
1737         * gcc/config/i386/i386.md (*movti_internal): Use
1738         predicate to determine if EVEX is needed.
1739         (*movsi_internal): Ditto.
1740         (*movdf_internal): Ditto.
1741         (*movsf_internal): Ditto.
1742         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
1744 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
1746         PR tree-optimization/58209
1747         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
1748         (find_tail_calls): Give up for pointer result types if m is non-NULL.
1749         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
1750         emit POINTER_PLUS_EXPR.
1751         (create_tailcall_accumulator): For pointer result type accumulate in
1752         sizetype type.
1754 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
1756         * configure.ac: Add backslashes missing from the last change.
1757         * configure: Regenerate.
1759 2013-08-22  Jan Hubicka  <jh@susue.cz>
1761         * ipa.c (function_and_variable_visibility): First remember function
1762         was global and then make it local.
1764 2013-08-22  Julian Brown  <julian@codesourcery.com>
1766         * configure.ac: Add aarch64 to list of arches which use "nop" in
1767         debug_line test.
1768         * configure: Regenerate.
1770 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1772         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
1773         gnu_libc_has_function.
1774         * config/s390/tpf.h: Likewise.
1776 2013-08-22  Jan Hubicka  <jh@susue.cz>
1778         * timevar.c (validate_phases): Add cast.
1780 2013-08-22  Jan Hubicka  <jh@susue.cz>
1782         * timevar.c (validate_phases): Use size_t for memory.
1783         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
1785 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1787         * pretty-print.h (output_buffer::output_buffer): Declare.
1788         (pretty_printer::pretty_printer): Likewise.
1789         (pp_construct): Remove.
1790         * pretty-print.c (output_buffer::output_buffer): Define.
1791         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
1792         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
1793         (print_gimple_expr): Likewise.
1794         (print_gimple_seq): Likewise.
1795         (gimple_dump_bb): Likewise.
1796         * sched-vis.c (dump_value_slim): Likewise.
1797         (dump_insn_slim): Likewise.
1798         (dump_rtl_slim): Likewise.
1799         (str_pattern_slim): Likewise.
1800         * tree-mudflap.c (mf_varname_tree): Likewise.
1801         * graph.c (print_graph_cfg): Likewise.
1802         (start_graph_dump): Likewise.
1803         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
1804         placement-new.
1805         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
1806         pretty printer initialization.
1807         * coretypes.h (diagnostic_context): Remove superflous type alias
1808         declaration.
1809         (pretty_printer): Likewise.  Declare directly as a class.
1810         (pretty_print_info): Remove declaration as class.
1811         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
1812         and pp_clear_output_area.
1813         (asan_add_global): Likewise.
1815 2013-08-22  Jan Hubicka  <jh@suse.cz>
1817         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
1818         * ipa-utils.h (update_type_inheritance_graph): Declare.
1819         (possible_polymorphic_call_target_p): Declare.
1820         (possible_polymorphic_call_target_p): New.
1821         * ipa-devirt.c: Update toplevel comments.
1822         (cached_polymorphic_call_targets): Move up.
1823         (odr_type_d): Move ID down.
1824         (polymorphic_type_binfo_p): Update comment.
1825         (odr_hasher::remove): Likewise;
1826         (get_odr_type): Set anonymous_namespace.
1827         (dump_odr_type): Dump it.
1828         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
1829         (maybe_record_node): Record node in cached_polymorphic_call_targets.
1830         (record_binfo): Add comment.
1831         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
1832         built.
1833         (devirt_node_removal_hook): Do not iCE when cache is freed.
1834         (possible_polymorphic_call_target_p): New predicate.
1835         (update_type_inheritance_graph): New function.
1837 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1838             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1839             Sergey Lega  <sergey.s.lega@intel.com>
1840             Anna Tikhonova  <anna.tikhonova@intel.com>
1841             Ilya Tocar  <ilya.tocar@intel.com>
1842             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1843             Ilya Verbin  <ilya.verbin@intel.com>
1844             Kirill Yukhin  <kirill.yukhin@intel.com>
1845             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1847         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
1848         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
1849         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
1850         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
1851         (OPTION_MASK_ISA_AVX2_UNSET): Update.
1852         (OPTION_MASK_ISA_AVX512F_UNSET): New.
1853         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
1854         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
1855         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
1856         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
1857         OPT_mavx512pf, OPT_mavx512er cases.
1858         * config/i386/constraints.md (v): New constraint.
1859         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
1860         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
1861         (bit_AVX512CD): New.
1862         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1863         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
1864         * config/i386/i386-c.c (ix86_target_macros_internal):
1865         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
1866         __AVX512PF__.
1867         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
1868         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
1869         * config/i386/i386.c (regclass_map, dbx_register_map)
1870         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
1871         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
1872         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
1873         -mavx512pf options.
1874         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
1875         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
1876         -mavx512pf options.  Fix formatting.
1877         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
1878         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
1879         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
1880         -mavx512cd, -mavx512pf options.
1881         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
1882         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
1883         (ix86_preferred_output_reload_class): Replace SSE_REGS with
1884         ALL_SSE_REGS.
1885         (ix86_hard_regno_mode_ok): Support 512-bit registers.
1886         (ix86_set_reg_reg_cost): Ditto.
1887         (x86_order_regs_for_local_alloc): Ditto.
1888         (MAX_VECT_LEN): Extend to 64-byte.
1889         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
1890         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
1891         (TARGET_AVX512ER, TARGET_AVX512CD): New.
1892         (BIGGEST_ALIGNMENT): Extend to 512-bits.
1893         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
1894         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
1895         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
1896         (SSE_REG_MODE_P): Support new modes.
1897         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
1898         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
1899         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
1900         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
1901         (REG_CLASS_CONTENTS): Add new registers.
1902         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
1903         (EXT_REX_SSE_REGNO_P): New.
1904         (HI_REGISTER_NAMES): Add new registers.
1905         * config/i386/i386.md: Define constants for new registers.
1906         (mode): Add new 512-bit modes.
1907         (prefix): Support evex prefix.
1908         (isa): Support avx512f, noavx512f, fma_avx512f.
1909         (ssemodesuffix): Add new 512-bit modes.
1910         (movxi): New.
1911         (*movxi_internal_avx512f): Ditto.
1912         (*movdi_internal): Replace constraint "x" with the new constraint "v".
1913         Support MODE_XI.
1914         (*movsi_internal): Likewise.
1915         (*movdf_internal): Likewise.
1916         (*movsf_internal): Likewise.
1917         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
1918         (<code><mode>3): Likewise.
1919         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
1920         New.
1921         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
1922         with the new constraint "v".
1923         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
1924         modes.
1925         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
1926         with the new constraint "v".
1927         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
1928         (<sse2>_storedqu<avxsizesuffix>): Likewise.
1929         (*<plusminus_insn><mode>3): Likewise.
1930         (<sse>_vm<plusminus_insn><mode>3): Likewise.
1931         (*mul<mode>3): Likewise.
1932         (<sse>_vmmul<mode>3): Likewise.
1933         (<sse>_div<mode>3): Likewise.
1934         (<sse>_vmdiv<mode>3): Likewise.
1935         (<sse>_sqrt<mode>2): Likewise.
1936         (<sse>_vmsqrt<mode>2): Likewise.
1937         (*<code><mode>3_finite): Likewise.
1938         (*<code><mode>3) <smaxmin>: Likewise.
1939         (<sse>_vm<code><mode>3): Likewise.
1940         (*<code><mode>3) <any_logic>: Likewise.
1941         (*fma_fmadd_<mode>): Likewise.
1942         (*fma_fmsub_<mode>): Likewise.
1943         (*fma_fnmadd_<mode>): Likewise.
1944         (*fma_fnmsub_<mode>): Likewise.
1945         (*fma_fmaddsub_<mode>): Likewise.
1946         (*fma_fmsubadd_<mode>): Likewise.
1947         (*fmai_fmadd_<mode>): Likewise.
1948         (*fmai_fmsub_<mode>): Likewise.
1949         (*fmai_fnmadd_<mode>): Likewise.
1950         (*fmai_fnmsub_<mode>): Likewise.
1951         (sse_cvtsi2ss): Likewise.
1952         (sse_cvtsi2ssq): Likewise.
1953         (sse_cvtss2si): Likewise.
1954         (sse_cvtss2si_2): Likewise.
1955         (sse_cvtss2siq): Likewise.
1956         (sse_cvtss2siq_2): Likewise.
1957         (sse_cvttss2si): Likewise.
1958         (sse_cvtss2siq_2): Likewise.
1959         (float<sseintvecmodelower><mode>2): Likewise.
1960         (sse2_cvtsd2si_2): Likewise.
1961         (sse2_cvtsd2siq_2): Likewise.
1962         (*<plusminus_insn><mode>3): Likewise.
1963         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1964         (*<sse4_1_avx2>_mul<mode>3): Likewise.
1965         (ashr<mode>3): Likewise.
1966         (<shift_insn><mode>3): Likewise.
1967         (avx2_<code><mode>3): Likewise.
1968         (*avx2_<code><mode>3): Likewise.
1969         (*andnot<mode>3): Likewise.
1970         (*<code><mode>3) <any_logic>: Likewise.
1971         (abs<mode>2): Likewise.
1972         (avx2_permvar<mode>): Likewise.
1973         (avx2_perm<mode>_1): Likewise.
1974         (*avx_vpermilp<mode>): Likewise.
1975         (avx_vpermilvar<mode>3): Likewise.
1976         (avx2_ashrv<mode>): Likewise.
1977         (avx2_<shift_insn>v<mode>): Likewise.
1978         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1979         -mavx512cd.
1980         * doc/rtl.texi: Document XImode.
1982 2013-08-21  Jeff Law  <law@redhat.com>
1984         * tree-flow.h (register_jump_thread): Pass vector of edges
1985         instead of each important edge.
1986         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1987         thread path into a vector and pass that to register_jump_thread.
1988         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1989         passed in edge vector to the current 3-edge form.
1991         Revert:
1992         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
1994         * dce.c (fini_dce): Call df_analyze again just in case
1995         delete_unmarked_insns removed anything.
1997 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
1999         * reload.h (struct reg_equivs): Rename to ..
2000         (struct reg_equivs_s): .. this.
2002 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
2004         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
2006 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2008         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
2010 2013-08-21  Jeff Law  <law@redhat.com>
2012         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
2013         simplify assignments too.  If the RHS collapses to a singleton
2014         range, then return the value for the range.
2016 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
2018         * config/i386/sse.md (V16): Rename to...
2019         (VMOVE): this.
2020         (mov<mode>): Update iterator name.
2021         (*mov<mode>_internal): Ditto.
2022         (push<mode>1): Ditto.
2023         (movmisalign<mode>): Ditto.
2025 2013-08-20  Jan Hubicka  <jh@suse.cz>
2027         PR bootstrap/58186
2028         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
2029         entry for direct edges.
2030         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
2032 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
2034         Revert my last two changes, r201865 and r201864:
2036         Revert r201865:
2037         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
2039         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
2040         instances can own GC refs.
2042         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
2043         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
2044         (gcc::context::gt_pch_nx): Likewise.
2045         (gcc::context::gt_pch_nx):  Likewise.
2046         * ggc.h (gt_ggc_mx <T>): New.
2047         (gt_pch_nx_with_op <T>): New.
2048         (gt_pch_nx <T>): New.
2049         * passes.c (opt_pass::gt_ggc_mx): New.
2050         (opt_pass::gt_pch_nx): New.
2051         (opt_pass::gt_pch_nx_with_op): New.
2052         (pass_manager::gt_ggc_mx): New.
2053         (pass_manager::gt_pch_nx): New.
2054         (pass_manager::gt_pch_nx_with_op): New.
2055         (pass_manager::operator new): Use
2056         ggc_internal_cleared_alloc_stat rather than xcalloc.
2057         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
2058         (pass_manager::gt_ggc_mx): New.
2059         (pass_manager::gt_pch_nx): New.
2060         (pass_manager::gt_pch_nx_with_op): New.
2061         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
2062         (opt_pass::operator new): New.
2063         (opt_pass::gt_ggc_mx): New.
2064         (opt_pass::gt_pch_nx): New.
2065         (opt_pass::gt_pch_nx_with_op): New.
2067         Revert r201864:
2068         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
2070         * Makefile.in (GTFILES): Add context.h.
2071         * context.c (gcc::context::operator new): New.
2072         (gcc::context::gt_ggc_mx): New.
2073         (gcc::context::gt_pch_nx): New.
2074         (gcc::context::gt_pch_nx): New.
2075         * context.h (gcc::context): Add GTY((user)) marking.
2076         (gcc::context::operator new): New.
2077         (gcc::context::gt_ggc_mx): New.
2078         (gcc::context::gt_pch_nx): New.
2079         (gcc::context::gt_pch_nx): New.
2080         (g): Add GTY marking.
2081         (gt_ggc_mx (gcc::context *)): New.
2082         (gt_pch_nx (gcc::context *)): New.
2083         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
2084         void *cookie)): New.
2085         * gengtype.c (open_base_files) <ifiles>: Add context.h.
2087 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
2089         * dce.c (fini_dce): Call df_analyze again just in case
2090         delete_unmarked_insns removed anything.
2092 2013-08-20  Teresa Johnson  <tejohnson@google.com>
2094         PR rtl-optimizations/57451
2095         * final.c (reemit_insn_block_notes): Prevent lexical blocks
2096         from crossing split section boundaries.
2098 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
2100         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
2101         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
2102         with MULTLIB_DEFAULTS.
2104 2013-08-20  Nick Clifton  <nickc@redhat.com>
2106         * target.def (narrow_volatile_bitfield): Note that the default
2107         value is false, not !TARGET_STRICT_ALIGN.
2108         * doc/tm.texi: Regenerate.
2110 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
2112         Fix LIB_SPEC for systems without libpthread.
2114         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
2115         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
2116         for Android.
2117         * config/i386/linux-common.h: Likewise.
2118         * config/mips/linux-common.h: Likewise.
2120 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
2122         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
2123         checks.
2125 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
2127         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
2128         instances can own GC refs.
2130         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
2131         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
2132         (gcc::context::gt_pch_nx): Likewise.
2133         (gcc::context::gt_pch_nx):  Likewise.
2134         * ggc.h (gt_ggc_mx <T>): New.
2135         (gt_pch_nx_with_op <T>): New.
2136         (gt_pch_nx <T>): New.
2137         * passes.c (opt_pass::gt_ggc_mx): New.
2138         (opt_pass::gt_pch_nx): New.
2139         (opt_pass::gt_pch_nx_with_op): New.
2140         (pass_manager::gt_ggc_mx): New.
2141         (pass_manager::gt_pch_nx): New.
2142         (pass_manager::gt_pch_nx_with_op): New.
2143         (pass_manager::operator new): Use
2144         ggc_internal_cleared_alloc_stat rather than xcalloc.
2145         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
2146         (pass_manager::gt_ggc_mx): New.
2147         (pass_manager::gt_pch_nx): New.
2148         (pass_manager::gt_pch_nx_with_op): New.
2149         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
2150         (opt_pass::operator new): New.
2151         (opt_pass::gt_ggc_mx): New.
2152         (opt_pass::gt_pch_nx): New.
2153         (opt_pass::gt_pch_nx_with_op): New.
2155 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
2157         * Makefile.in (GTFILES): Add context.h.
2158         * context.c (gcc::context::operator new): New.
2159         (gcc::context::gt_ggc_mx): New.
2160         (gcc::context::gt_pch_nx): New.
2161         (gcc::context::gt_pch_nx): New.
2162         * context.h (gcc::context): Add GTY((user)) marking.
2163         (gcc::context::operator new): New.
2164         (gcc::context::gt_ggc_mx): New.
2165         (gcc::context::gt_pch_nx): New.
2166         (gcc::context::gt_pch_nx): New.
2167         (g): Add GTY marking.
2168         (gt_ggc_mx (gcc::context *)): New.
2169         (gt_pch_nx (gcc::context *)): New.
2170         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
2171         void *cookie)): New.
2172         * gengtype.c (open_base_files) <ifiles>: Add context.h.
2174 2013-08-20  Alan Modra  <amodra@gmail.com>
2176         PR target/57865
2177         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
2178         (rs6000_emit_epilogue): Likewise.
2180 2013-08-19  Dehao Chen  <dehao@google.com>
2182         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
2184 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
2185             Jakub Jelinek  <jakub@redhat.com>
2187         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
2188         (BUILT_IN_FABSD64): Likewise.
2189         (BUILT_IN_FABSD128): Likewise.
2190         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
2191         (fold_builtin_1): Likewise.
2192         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
2193         destination and source operands.
2194         (*abstd2_fpr): Likewise.
2195         (*nabstd2_fpr): Likewise.
2197 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
2199         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
2200         JUMP_P and INSN_P.
2202 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
2204         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
2205         incomplete.
2207 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2209         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
2210         * builtins.c (default_libc_has_function): New.
2211         (gnu_libc_has_function): Ditto.
2212         (no_c99_libc_has_function): Ditto.
2213         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
2214         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
2215         (fold_builtin_sincos): Likewise.
2216         (fold_builtin_cexp): Likewise.
2217         * builtins.def (DEF_C94_BUILTIN): Likewise.
2218         (DEF_C99_BUILTIN): Likewise.
2219         (DEF_C99_C90RES_BUILTIN): Likewise.
2220         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
2221         definitions to using this define.
2222         * config/darwin-protos.h (darwin_libc_has_function): New.
2223         * config/darwin.c (darwin_libc_has_function): Ditto.
2224         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
2225         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
2226         * config/darwin.h: Ditto.
2227         * config/elfos.h: Ditto.
2228         * config/freebsd.h: Ditto.
2229         * config/i386/cygming.h: Ditto.
2230         * config/i386/djgpp.h: Ditto.
2231         * config/i386/i386-interix.h: Ditto.
2232         * config/microblaze/microblaze.h: Ditto.
2233         * config/mmix/mmix.h: Ditto.
2234         * config/gnu-user.h: Ditto.
2235         * config/ia64/hpux.h: Ditto.
2236         * config/pa/pa-hpux.h: Ditto.
2237         * config/pdp11/pdp11.h: Ditto.
2238         * config/picochip/picochip.h: Ditto.
2239         * config/linux.h: Ditto.
2240         * config/netbsd.h: Ditto.
2241         * config/openbsd.h: Ditto.
2242         * config/rs6000/aix43.h: Ditto.
2243         * config/rs6000/aix51.h: Ditto.
2244         * config/rs6000/aix52.h: Ditto.
2245         * config/rs6000/aix53.h: Ditto.
2246         * config/rs6000/aix61.h: Ditto.
2247         * config/rs6000/darwin.h: Ditto.
2248         * config/rs6000/linux.h: Ditto.
2249         * config/rs6000/linux64.h: Ditto.
2250         * config/s390/tpf.h: Ditto.
2251         * config/sol2-10.h: Ditto.
2252         * config/sol2.h: Ditto.
2253         * config/vms/vms.h: Ditto.
2254         * config/vxworks.h: Ditto.
2255         * config/linux-android.c (linux_android_libc_has_function):
2256         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
2257         * config/linux-protos.h (linux_android_libc_has_function):
2258         New declaration.
2259         * config/i386/i386.c (ix86_libc_has_function): New.
2260         * config/i386/i386-protos.h
2261         (ix86_libc_has_function): New declaration.
2262         * config/i386/i386.md
2263         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
2264         ("isinf<mode>2): Likewise.
2265         * convert.c (convert_to_integer): Using new target hook
2266         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
2267         TARGET_C99_FUNCTIONS.
2268         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
2269         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
2270         * coretypes.h (function_class): New enum for different
2271         classes of functions.
2272         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
2273         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
2274         (TARGET_HAS_SINCOS): Likewise.
2275         (TARGET_LIBC_HAS_FUNCTION): New.
2276         * doc/tm.texi: Regenerated.
2277         * targhooks.h (default_libc_has_function): New declaration.
2278         (no_c99_libc_has_function): Ditto.
2279         (gnu_libc_has_function): Ditto.
2280         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
2281         and TARGET_HAS_SINCOS.
2283 2013-08-18  Jan Hubicka  <jh@suse.cz>
2285         * Makeifle-in (ipa-devirt.o): New.
2286         (GTFILES): Add ipa-utils.h and ipa-devirt.c
2287         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
2288         (analyze_functions): Look into possible targets of polymorphic call.
2289         * dumpfile.c (dump_files): Add type-inheritance dump.
2290         * dumpfile.h (TDI_inheritance): New.
2291         * ipa-devirt.c: New file.
2292         * ipa-utils.h (odr_type_d): Forward declare.
2293         (odr_type): New type.
2294         (build_type_inheritance_graph): Declare.
2295         (possible_polymorphic_call_targets): Declare and introduce inline
2296         variant when only edge is pased.
2297         (dump_possible_polymorphic_call_targets): Likewise.
2298         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
2299         * tree.c (type_in_anonymous_namespace_p): Break out from ...
2300         (types_same_for_odr): ... here.
2301         * tree.h (type_in_anonymous_namespace_p): Declare.
2303 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
2305         PR tree-optimization/58006
2306         * tree-parloops.c (take_address_of): Don't ICE if get_name
2307         returns NULL.
2308         (eliminate_local_variables_stmt): Remove clobber stmts.
2310 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
2312         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
2313         error message is issued for an alias to undefined symbol.
2315 2013-08-18  Jan Hubicka  <jh@suse.cz>
2317         * cgraph.c (cgraph_create_indirect_edge): Discover
2318         polymorphic calls and record basic info into indirect_info.
2319         * gimple-fold.c (gimple_fold_call): When doing BINFO based
2320         devirtualization, ignore objc function calls.
2321         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
2322         call with no parm index info.
2323         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
2324         * tree.c (virtual_method_call_p): New function.
2325         * tree.h (virtual_method_call_p): Declare.
2327 2013-08-16  Jan Hubicka  <jh@suse.cz>
2329         PR middle-end/58179
2330         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
2332 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
2334         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
2335         attribute.
2337 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
2339         * gengtype.c (type_for_name): Add special-case support for
2340         locating types within the "gcc::" namespace.
2341         (open_base_files): Emit a "using namespace gcc" directive.
2343 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
2345         PR target/58160
2346         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
2347         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
2349         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
2350         array instead of each individual operand as a separate argument.
2351         (emit_fusion_gpr_load): Likewise.
2352         (expand_fusion_gpr_load): Add new function declaration.
2354         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
2355         signature to have the operands passed as an array, instead of as
2356         separate arguments.  Allow ZERO_EXTEND to be in the memory
2357         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
2358         depend on the register live/dead flags when peepholes are run.
2359         (expand_fusion_gpr_load): New function to be called from the
2360         peephole2 pass, to change the register that addis sets to be the
2361         target register.
2362         (emit_fusion_gpr_load): Change the calling signature to have the
2363         operands passed as an array, instead of as separate arguments.
2364         Allow ZERO_EXTEND to be in the memory address, and also
2365         SIGN_EXTEND if -mpower8-fusion-sign.
2367         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
2368         unspec enumeration.
2369         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
2370         adjust the register addis loads up in the peephole2 pass.  Do not
2371         depend on the register live/dead state when the peephole pass is done.
2373 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
2375         * gengtype.c (create_user_defined_type): Ensure that the kind
2376         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
2377         declaration is seen before the GTY((user)) marking.
2379 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2381         PR target/58105
2382         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
2384 2013-08-16  Jan Hubicka  <jh@suse.cz>
2386         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
2387         arugment expected_type.
2388         (gimple_fold_call): Use it.
2389         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
2390         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
2391         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
2392         (try_make_edge_direct_virtual_call): Likewise.
2393         * tree.c (obj_type_ref_class): New.
2394         * tree.h (obj_type_ref_class): Use it.
2396 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2398         * sched-vis.c (rtl_slim_pp_initialized): Remove.
2399         (rtl_slim_pp): Likewise.
2400         (init_rtl_slim_pretty_print): Likewise.
2401         (dump_value_slim):  Don't call it.  Use local pretty printer.
2402         (dump_insn_slim): Likewise.
2403         (dump_rtl_slim): Likewise.
2404         (str_pattern_slim): Likewise.
2405         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
2406         Simplify.
2408 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
2410         PR tree-optimization/58164
2411         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
2412         walk gimple_goto_dest of GIMPLE_GOTO.
2414         PR tree-optimization/58165
2415         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
2416         bi_call must be the last stmt in a bb, don't split_block, instead
2417         use fallthru edge from it and give up if there is none.
2418         Release conds vector when returning early.
2420 2013-08-14  Xinliang David Li  <davidxl@google.com>
2422         * config/i386/i386.c (ix86_option_override_internal):
2423         Remove unused variable and field.
2425 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2427         PR target/57949
2428         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
2429         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
2430         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
2431         and Linux, correct BLKmode alignment when 128-bit alignment is
2432         required and compatibility flag is not set.
2433         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
2434         for zero-size arguments when compatibility flag is not set.
2436 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
2438         PR tree-optimization/58145
2439         * tree-sra.c (build_ref_for_offset): If prev_base has
2440         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
2442 2013-08-14  Xinliang David Li  <davidxl@google.com>
2444         * config/i386/i386.c (ix86_option_override_internal):
2445         Fix uninitialized variable error.
2447 2013-08-14  Xinliang David Li  <davidxl@google.com>
2449         * config/i386/i386.opt: Define two new options.
2450         * config/i386/x86-tune.def: Add arch selector field in macros.
2451         * config/i386/i386.h: Adjust macro definition.
2452         * config/i386/i386.c (ix86_option_override_internal):
2453         Refactor the code.
2454         (parse_mtune_ctrl_str): New function.
2455         (set_ix86_tune_features): New function.
2456         (ix86_function_specific_restore): Call the new helper function.
2458 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
2460         PR rtl-optimization/57662
2461         * sel-sched.c (code_motion_process_successors): When the current insn
2462         is removed after the recursive traversal, break from the loop.
2463         Add comments and debug printouts.
2465 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
2466             Alexandre Oliva  <aoliva@redhat.com>
2468         PR target/58067
2469         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
2470         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
2471         there also UNSPEC_PLTOFF.
2473 2013-08-14  Marek Polacek  <polacek@redhat.com>
2475         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
2476         NUM_CONDITIONS bit positions.
2478 2013-08-13  Cary Coutant  <ccoutant@google.com>
2480         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
2481         (attr_checksum): Hash vector contents instead of pointer.
2482         (attr_checksum_ordered): Likewise.
2484 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
2486         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
2487         when Pmode != word_mode.  Add length_address attribute.
2488         (sse3_monitor_<mode>): Merge from sse3_monitor and
2489         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
2490         Pmode != word_mode.  Update insn length attribute.
2491         * config/i386/i386.c (ix86_option_override_internal): Update
2492         ix86_gen_monitor selection for merged sse3_monitor insn.
2494 2013-08-13  Julian Brown  <julian@codesourcery.com>
2496         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
2497         perform invalid legitimization on greater-than-word-size modes for
2498         TARGET_E500_DOUBLE.
2500 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
2502         * ira.c (setup_class_translate_array): Use aclass instead of cl
2503         for classes not fully covered by allocno classes.
2505 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
2507         PR tree-optimization/57661
2508         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
2509         * tree-inline.c (tree_function_versioning): Initialize it.
2510         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
2511         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
2512         that is not being copied.
2514         PR sanitizer/56417
2515         * asan.c (instrument_strlen_call): Fix typo in comment.
2516         Use char * type even for the lhs of POINTER_PLUS_EXPR.
2518 2013-08-13  Steve Ellcey  <sellcey@mips.com>
2520         * config/mips/mips.md (prefetch): Use lw instead of ld on
2521         loongson in 32bit mode.
2523 2013-08-13  Nick Clifton  <nickc@redhat.com>
2525         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
2527 2013-08-13  Jan Hubicka  <jh@suse.cz>
2529         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
2530         introduced edge; fix typo in sanity check.
2531         (cgraph_resolve_speculation): Export; improve diagnostic.
2532         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
2533         speculation at type mismatch.
2534         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
2535         (cgraph_resolve_speculation): Declare.
2536         (symtab_can_be_discarded): New function.
2537         * value-prof.c (gimple_ic_transform): Remove actual transform code.
2538         * ipa-inline-transform.c (speculation_removed): New global var.
2539         (clone_inlined_nodes): See if speculation can be removed.
2540         (inline_call): If speculations was removed, we growths may not match.
2541         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
2542         (speculation_useful_p): New function.
2543         (resolve_noninline_speculation): New function.
2544         (inline_small_functions): Resolve useless speculations.
2545         * ipa-inline.h (speculation_useful_p): Declare
2546         * ipa.c (can_replace_by_local_alias): Simplify.
2547         (ipa_profile): Produce speculative calls in non-lto, too;
2548         add simple cost model; produce local aliases.
2550 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
2552         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
2553         PASS_MANAGER_H.
2555 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
2557         * config/i386/i386.c (ix86_function_versions): Use error + inform.
2559 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
2561         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
2562         iterator instead of X87MODEF.
2564 2013-08-12  Perez Read  <netfirewall@gmail.com>
2566         PR target/58132
2567         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
2568         operand 0 for intel asm alternative.
2569         (*movabs<mode>_2): Ditto for operand 1.
2571 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
2573         * config/aarch64/arm_none.h
2574         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
2576 2013-08-12  Nick Clifton  <nickc@redhat.com>
2578         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
2580 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
2582         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
2583         expand for QI/HImode operand to produce more effictive code for
2584         unsigned char(short) --> float(double) conversion.
2586 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
2588         * doc/invoke.texi: Mention that -ftls-model does not force the final
2589         model.
2591 2013-08-12  Marek Polacek  <polacek@redhat.com>
2592             Marc Glisse  <marc.glisse@inria.fr>
2594         PR tree-optimization/57980
2595         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
2596         when creating -1 constant.
2598 2013-08-10  Jan Hubicka  <jh@suse.cz>
2600         Workaround binutils PR14342.
2601         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
2602         (gimple_init_edge_profiler): Likewise.
2603         (gimple_gen_ic_func_profiler): Likewise.
2605 2013-08-09  Jan Hubicka  <jh@suse.cz>
2607         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
2609 2013-08-09  Xinliang David Li  <davidxl@google.com>
2611         * config/i386/stringop.def: New file.
2612         * config/i386/stringop.opt: New file.
2613         * config/i386/i386-opts.h: Include stringopt.def.
2614         * config/i386/i386.opt: Include stringopt.opt.
2615         * config/i386/i386.c (ix86_option_override_internal):
2616         Override default size based stringop inline strategies with options.
2617         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
2618         New function.
2620 2013-08-09  Jan Hubicka  <jh@suse.cz>
2622         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
2624 2013-08-09  Jan Hubicka  <jh@suse.cz>
2626         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
2627         CGRAPH_FREQ_MAX.
2628         (dump_cgraph_node): Dump profile-id.
2629         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
2630         and common_target_probability.
2631         * lto-cgraph.c (lto_output_edge): Stream common targets.
2632         (lto_output_node): Stream profile ids.
2633         (input_node): Stream profile ids.
2634         (input_edge): Stream common targets.
2635         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
2636         * ipa.c: Include value-prof.h
2637         (ipa_profile_generate_summary): Turn indirect call statement histograms
2638         into common targets.
2639         (ipa_profile): Turn common targets into speculative edges.
2641 2013-08-09  Jan Hubicka  <jh@suse.cz>
2643         * cgraph.h (cgraph_node): Add profile_id.
2644         * value-prof.c (cgraph_node_map): Turn into pointer_map.
2645         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
2646         (del_node_map): Update.
2647         (find_func_by_funcdef_no): Replace by ...
2648         (find_func_by_profile_id): ... this one.
2649         (gimple_ic_transform): Do not remove useful histograms when
2650         speculation is not done; dump info when indirect call removal
2651         can happen at LTO.
2652         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
2653         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
2654         (__gcov_indirect_call_profiler_v2): .. this one.
2655         * profile.h (init_node_map): Update.
2656         * coverage.c (coverage_compute_profile_id): New function.
2657         * coverage.h (coverage_compute_profile_id): Declare.
2658         * tree-profile.c (init_ic_make_global_vars): Make
2659         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
2660         (gimple_init_edge_profiler): Update prototype of
2661         __gcov_indirect_call_profiler.
2662         (gimple_gen_ic_func_profiler): Simplify.
2663         (tree_profiling): Use init_node_map
2665 2013-08-09  Jan Hubicka  <jh@suse.cz>
2667         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
2668         non-speculative refs.
2669         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
2670         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
2671         (cgraph_set_call_stmt): Likewise.
2672         (cgraph_create_edge_1): Fix release checking compilatoin;
2673         clear lto_stmt_uid.
2674         (cgraph_free_edge): Free indirect info.
2675         (cgraph_turn_edge_to_speculative): New function.
2676         (cgraph_speculative_call_info): New function.
2677         (cgraph_make_edge_direct): Return direct edge; handle speculation.
2678         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
2679         (dump_cgraph_node): Dump speculation.
2680         (verify_edge_count_and_frequency): Accept speculative edges.
2681         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
2682         (verify_cgraph_node): Handle speculation.
2683         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
2684         (cgraph_set_call_stmt): Update prototype.
2685         (cgraph_make_edge_direct): Update prototype.
2686         (cgraph_speculative_call_info): Declare.
2687         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
2688         to change; update call of ipa_find_references.
2689         * ipa-ref.c (ipa_record_reference): Fix return value; clear
2690         lto_stmt_uid and speculative flags.
2691         (ipa_dump_references): Dump speculation.
2692         (ipa_clone_references): Clone speculative flag.
2693         (ipa_clone_referring): Likewise.
2694         (ipa_clone_ref): New function.
2695         (ipa_find_reference): Look into lto_stmt_uids
2696         (ipa_clear_stmts_in_references): Do not clear speculative calls.
2697         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
2698         (ipa_find_reference): Update declaration.
2699         (ipa_clone_ref): Declare.
2700         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
2701         stream speculative flag.
2702         (lto_output_ref): Stream statements uids and speculation.
2703         (input_ref): Likewise.
2704         (input_edge): Stream speuclation.
2705         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
2706         (cgraph_set_call_stmt_including_clones): Handle speculation.
2707         * ipa-inline.c (heap_edge_removal_hook): New function.
2708         (inline_small_functions): Register it.
2709         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
2710         also initialize refs.
2711         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
2712         edge to change.
2713         (try_make_edge_direct_simple_call): Likewise.
2714         (try_make_edge_direct_simple_call): Likewise.
2715         (update_indirect_edges_after_inlining): Likewise.
2716         (remove_described_reference): Look proper lto_stmt_uid.
2717         (propagate_controlled_uses): Likewise.
2718         (propagate_controlled_uses): Liekwise.
2719         * tree-inline.c (copy_bb): Copy speculative edges.
2720         (redirect_all_calls): New function.
2721         (copy_cfg_body): Do redirection after loop info is updated.
2722         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
2724 2013-08-09  Jan Hubicka  <jh@suse.cz>
2726         * lto-streamer-out.c (output_function): Renumber PHIs.
2727         * lto-streamer-in.c (input_function): Likewise.
2729 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
2731         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
2732         (get_lane_unsigned): Likewise.
2733         (dup_lane_scalar): Likewise.
2734         (get_lane): enable for VALL.
2735         * config/aarch64/aarch64-simd.md
2736         (aarch64_dup_lane_scalar<mode>): Remove.
2737         (aarch64_get_lane_signed<mode>): Likewise.
2738         (aarch64_get_lane_unsigned<mode>): Likewise.
2739         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
2740         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
2741         (aarch64_get_lane<mode>): Enable for all vector modes.
2742         (aarch64_get_lanedi): Remove misleading constraints.
2743         * config/aarch64/arm_neon.h
2744         (__aarch64_vget_lane_any): Define.
2745         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
2746         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
2747         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
2748         * config/aarch64/iterators.md (VDQQH): New.
2749         (VDQQHS): Likewise.
2750         (vwcore): Likewise.
2752 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
2754         * configure.ac: Add GAS check for LEON instructions on SPARC.
2755         * configure: Regenerate.
2756         * config.in: Likewise.
2757         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
2758         sparc*-*-* block.
2759         * config/sparc/sparc.opt (LEON, LEON3): New masks.
2760         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
2761         for LEON or LEON3.
2762         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
2763         (AS_LEON_FLAG): New macro.
2764         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
2765         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
2766         Deal with LEON and LEON3 for the memory model.
2767         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
2768         (atomic_compare_and_swap<mode>_1): Likewise.
2769         (*atomic_compare_and_swap<mode>_1): Likewise.
2771 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2773         * config/arm/neon.md (vcond): Fix floating-point vector
2774         comparisons against 0.
2776 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
2778         * lra-constraints.c (emit_spill_move): Remove assert.
2779         (process_alt_operands): Add more debugging
2780         output.  Increase reject for spilling into memory.  Decrease
2781         reject for reloading scratch.
2782         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
2784 2013-08-08  Steve Ellcey  <sellcey@mips.com>
2786         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
2787         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
2788         micromips incompatible.  Add nan2008.
2789         (MULTILIB_DIRNAMES): Add nan2008.
2790         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
2791         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
2792         and micromips incompatible.  Add nan2008.
2793         (MULTILIB_DIRNAMES): Add nan2008.
2794         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
2796 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
2798         PR rtl-optimization/58079
2799         * combine.c (combine_simplify_rtx): Avoid using SUBST if
2800         simplify_comparison has widened a comparison with an integer.
2802 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2804         * config/arm/neon.md (movmisalign<mode>): Disable when we
2805         don't allow unaligned accesses.
2806         (*movmisalign<mode>_neon_store): Likewise.
2807         (*movmisalign<mode>_neon_load): Likewise.
2808         (*movmisalign<mode>_neon_store): Likewise.
2809         (*movmisalign<mode>_neon_load): Likewise.
2811 2013-08-08  Jan Hubicka  <jh@suse.cz>
2813         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
2814         (make_pass_rebuild_cgraph_edges): Also clear references.
2815         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
2816         * ipa-inline-transform.c (inline_transform): Remove all references
2817         after inlining.
2818         * cgraphunit.c (expand_function): Remove all references after
2819         expansion.
2820         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
2821         (ipa_find_reference): Rewrite to iterator.
2822         (remove_stmt_references): Likewise.
2823         (ipa_clear_stmts_in_references): New function.
2824         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
2825         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
2826         clear references.
2827         * ipa-split.c (split_function): Remove references in split function.
2829 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
2831         PR target/57431
2832         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
2833         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
2835 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
2837         PR target/56979
2838         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
2839         suggested mode for the assignment isn't compatible with the
2840         registers required.
2842 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2844         PR target/58065
2845         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
2847 2013-08-07  Xinliang David Li  <davidxl@google.com>
2849         * config/i386/i386.opt: New option -mtune-ctrl=.
2850         * config/i386/x86-tune.def: New file.
2851         * config/i386/i386.h: include x86-tune.def.
2852         * config/i386/i386.c (ix86_option_override_internal):
2853         Parsing -mtune-ctrl= option and set tune features.
2855 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
2857         PR other/12081
2858         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
2859         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
2860         to gen_2arg_fn_t.
2862 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2864         * rtl.h (update_alignments): Declare.
2865         * final.c (grow_label_align): New function extracted from...
2866         (shorten_branches): ...here.  Call it.
2867         (update_alignments): New function.
2868         * reorg.c (sibling_labels): New variable.
2869         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
2870         the new label along with it onto the sibling_labels vector.
2871         (fill_simple_delay_slots): Adjust call to get_label_before.
2872         (fill_slots_from_thread): Likewise.
2873         (relax_delay_slots): Likewise.
2874         (make_return_insns): Likewise.
2875         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
2877 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
2879         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
2880         document its semantics.
2881         (diagnostic_report_diagnostic): Adjust accordingly.
2883 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2885         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
2886         (sparc_option_override): ...and port to new C++ pass API.
2887         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
2889 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
2891         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
2893 2013-08-06  Caroline Tice  <cmtice@google.com>
2895         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
2896         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
2897         * tree-pass.h: Add pass_vtable_verify.
2898         * varasm.c (assemble_variable): Add code to properly set the comdat
2899         section and name for the .vtable_map_vars section.
2900         (assemble_vtyv_preinit_initializer): New function.
2901         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
2902         LINK_ONCE flag.
2903         * output.h: Add function decl for assemble_vtv_preinit_initializer.
2904         * vtable-verify.c: New file.
2905         * vtable-verify.h: New file.
2906         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
2907         initialiation levels.
2908         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
2909         * passes.def: Insert pass_vtable_verify.
2910         * aclocal.m4: Reorder includes.
2911         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
2912         -fvtv-counts options.
2913         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
2914         as appropriate, if -fvtable-verify=... is used.
2915         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
2916         -fvtable-verify=... is used.
2917         * Makefile.in (OBJS):  Add vtable-verify.o to list.
2918         (vtable-verify.o): Add new build rule.
2919         (GTFILES): Add vtable-verify.c to list.
2920         * common.opt (fvtable-verify=): New flag.
2921         (vtv_priority): Values for fvtable-verify= flag.
2922         (fvtv-counts): New flag.
2923         (fvtv-debug): New flag.
2924         * tree.h (save_vtable_map_decl): New extern function decl.
2926 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2928         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
2929         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
2930         (pass_data_rl78_devirt): ...new pass_data instance and...
2931         (make_pass_rl78_devirt): ...new function.
2932         (rl78_asm_file_start): Port pass registration to new C++ API.
2934 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
2936         * coretypes.h (rtl_opt_pass): Add.
2937         (gcc::context): Add.
2938         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
2939         (epiphany_init): Port to new C++ pass API.
2940         (epiphany_optimize_mode_switching): Likewise.
2941         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2942         (pass_manager::get_pass_mode_switching): New.
2943         (pass_manager::get_pass_peephole2): New.
2944         * mode-switching.c (pass_mode_switching): Add clone method.
2945         * recog.c (pass_peephole2): Add clone method.
2946         (pass_split_all_insns): Add clone method.
2948 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
2950         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2951         (mips_option_override): ...here, porting to new C++ API for passes.
2953 2013-08-06  Jan Hubicka  <jh@suse.cz>
2955         * cgraph.c (cgraph_get_body): New function based on lto.c
2956         implementation.
2957         * cgraph.h (cgraph_get_body): Declare.
2958         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2959         LTO paths.
2960         * cgraphunit.c (expand_function): Get body prior expanding.
2961         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2962         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2963         really need.
2964         * passes.c (do_per_function_toporder): Get body.
2965         * tree-inline.c (expand_call_inline): Get body prior inlining it.
2966         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2968 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2970         PR fortran/57987
2971         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2972         is not re-finalized.  Rename second parameter to no_collect.
2974 2013-08-06  Martin Jambor  <mjambor@suse.cz>
2976         PR middle-end/58041
2977         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2978         MEM_REF has proper alignment information.
2980 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
2982         PR other/12081
2983         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2984         class insn_gen_fn.
2985         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2986         rtx (*) (rtx, ...) with insn_gen_fn.
2987         * genoutput.c (output_insn_data): Cast gen_? function pointers to
2988         insn_gen_fn::stored_funcptr.  Add initializer braces.
2990 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2992         Rewrite how instances of passes are cloned to remove assumptions
2993         about their sizes (thus allowing pass subclasses to have
2994         additional data fields, albeit non-GC-managed ones at this point).
2996         * passes.c (make_pass_instance): Now that passes have clone
2997         methods, rewrite this function to eliminate XNEW and memcpy
2998         calls that used hardcoded sizes.  Since this function no longer
2999         creates pass instances, rename it to...
3000         (add_pass_instance): ...this.  Document the old way that passes were
3001         numbered and flagged, and rework this function to continue using it.
3002         (next_pass_1): Add an initial_pass argument for use by
3003         add_pass_instance.
3004         (position_pass): When adding multiple instances of a pass, use
3005         the pass's clone method, rather than relying on the XNEW/memcpy
3006         within the former make_pass_instance (now add_pass_instance).
3007         (pass_manager::pass_manager): When invoking next_pass_1, also supply
3008         the initial instance of the current pass within the pass manager.
3010 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3012         This is the automated part of the conversion of passes from C
3013         structs to C++ classes.
3015         Patch autogenerated by refactor_passes.py from
3016         https://github.com/davidmalcolm/gcc-refactoring-scripts
3017         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
3019         * asan.c (pass_asan): Convert from a global struct to a subclass of
3020         gimple_opt_pass along with...
3021         (pass_data_asan): ...new pass_data instance and...
3022         (make_pass_asan): ...new function.
3023         (pass_asan_O0): Convert from a global struct to a subclass of
3024         gimple_opt_pass along with...
3025         (pass_data_asan_O0): ...new pass_data instance and...
3026         (make_pass_asan_O0): ...new function.
3027         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
3028         subclass of rtl_opt_pass along with...
3029         (pass_data_inc_dec): ...new pass_data instance and...
3030         (make_pass_inc_dec): ...new function.
3031         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
3032         a subclass of rtl_opt_pass along with...
3033         (pass_data_reorder_blocks): ...new pass_data instance and...
3034         (make_pass_reorder_blocks): ...new function.
3035         (pass_duplicate_computed_gotos): Convert from a global struct to a
3036         subclass of rtl_opt_pass along with...
3037         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
3038         (make_pass_duplicate_computed_gotos): ...new function.
3039         (pass_partition_blocks): Convert from a global struct to a subclass of
3040         rtl_opt_pass along with...
3041         (pass_data_partition_blocks): ...new pass_data instance and...
3042         (make_pass_partition_blocks): ...new function.
3043         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
3044         struct to a subclass of rtl_opt_pass along with...
3045         (pass_data_branch_target_load_optimize1): ...new pass_data instance
3046         and...
3047         (make_pass_branch_target_load_optimize1): ...new function.
3048         (pass_branch_target_load_optimize2): Convert from a global struct to a
3049         subclass of rtl_opt_pass along with...
3050         (pass_data_branch_target_load_optimize2): ...new pass_data instance
3051         and...
3052         (make_pass_branch_target_load_optimize2): ...new function.
3053         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
3054         of rtl_opt_pass along with...
3055         (pass_data_jump): ...new pass_data instance and...
3056         (make_pass_jump): ...new function.
3057         (pass_jump2): Convert from a global struct to a subclass of
3058         rtl_opt_pass along with...
3059         (pass_data_jump2): ...new pass_data instance and...
3060         (make_pass_jump2): ...new function.
3061         * cfgexpand.c (pass_expand): Convert from a global struct to a
3062         subclass of rtl_opt_pass along with...
3063         (pass_data_expand): ...new pass_data instance and...
3064         (make_pass_expand): ...new function.
3065         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
3066         of rtl_opt_pass along with...
3067         (pass_data_free_cfg): ...new pass_data instance and...
3068         (make_pass_free_cfg): ...new function.
3069         (pass_into_cfg_layout_mode): Convert from a global struct to a
3070         subclass of rtl_opt_pass along with...
3071         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
3072         (make_pass_into_cfg_layout_mode): ...new function.
3073         (pass_outof_cfg_layout_mode): Convert from a global struct to a
3074         subclass of rtl_opt_pass along with...
3075         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
3076         (make_pass_outof_cfg_layout_mode): ...new function.
3077         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
3078         struct to a subclass of gimple_opt_pass along with...
3079         (pass_data_build_cgraph_edges): ...new pass_data instance and...
3080         (make_pass_build_cgraph_edges): ...new function.
3081         (pass_rebuild_cgraph_edges): Convert from a global struct to a
3082         subclass of gimple_opt_pass along with...
3083         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
3084         (make_pass_rebuild_cgraph_edges): ...new function.
3085         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
3086         subclass of gimple_opt_pass along with...
3087         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
3088         and...
3089         (make_pass_remove_cgraph_callee_edges): ...new function.
3090         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
3091         struct to a subclass of rtl_opt_pass along with...
3092         (pass_data_stack_adjustments): ...new pass_data instance and...
3093         (make_pass_stack_adjustments): ...new function.
3094         * combine.c (pass_combine): Convert from a global struct to a subclass
3095         of rtl_opt_pass along with...
3096         (pass_data_combine): ...new pass_data instance and...
3097         (make_pass_combine): ...new function.
3098         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
3099         global struct to a subclass of rtl_opt_pass along with...
3100         (pass_data_compare_elim_after_reload): ...new pass_data instance
3101         and...
3102         (make_pass_compare_elim_after_reload): ...new function.
3103         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
3104         of rtl_opt_pass along with...
3105         (pass_data_rtl_cprop): ...new pass_data instance and...
3106         (make_pass_rtl_cprop): ...new function.
3107         * cse.c (pass_cse): Convert from a global struct to a subclass of
3108         rtl_opt_pass along with...
3109         (pass_data_cse): ...new pass_data instance and...
3110         (make_pass_cse): ...new function.
3111         (pass_cse2): Convert from a global struct to a subclass of
3112         rtl_opt_pass along with...
3113         (pass_data_cse2): ...new pass_data instance and...
3114         (make_pass_cse2): ...new function.
3115         (pass_cse_after_global_opts): Convert from a global struct to a
3116         subclass of rtl_opt_pass along with...
3117         (pass_data_cse_after_global_opts): ...new pass_data instance and...
3118         (make_pass_cse_after_global_opts): ...new function.
3119         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
3120         of rtl_opt_pass along with...
3121         (pass_data_ud_rtl_dce): ...new pass_data instance and...
3122         (make_pass_ud_rtl_dce): ...new function.
3123         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
3124         rtl_opt_pass along with...
3125         (pass_data_fast_rtl_dce): ...new pass_data instance and...
3126         (make_pass_fast_rtl_dce): ...new function.
3127         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
3128         a subclass of rtl_opt_pass along with...
3129         (pass_data_df_initialize_opt): ...new pass_data instance and...
3130         (make_pass_df_initialize_opt): ...new function.
3131         (pass_df_initialize_no_opt): Convert from a global struct to a
3132         subclass of rtl_opt_pass along with...
3133         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
3134         (make_pass_df_initialize_no_opt): ...new function.
3135         (pass_df_finish): Convert from a global struct to a subclass of
3136         rtl_opt_pass along with...
3137         (pass_data_df_finish): ...new pass_data instance and...
3138         (make_pass_df_finish): ...new function.
3139         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
3140         rtl_opt_pass along with...
3141         (pass_data_rtl_dse1): ...new pass_data instance and...
3142         (make_pass_rtl_dse1): ...new function.
3143         (pass_rtl_dse2): Convert from a global struct to a subclass of
3144         rtl_opt_pass along with...
3145         (pass_data_rtl_dse2): ...new pass_data instance and...
3146         (make_pass_rtl_dse2): ...new function.
3147         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
3148         subclass of rtl_opt_pass along with...
3149         (pass_data_dwarf2_frame): ...new pass_data instance and...
3150         (make_pass_dwarf2_frame): ...new function.
3151         * except.c (pass_set_nothrow_function_flags): Convert from a global
3152         struct to a subclass of rtl_opt_pass along with...
3153         (pass_data_set_nothrow_function_flags): ...new pass_data instance
3154         and...
3155         (make_pass_set_nothrow_function_flags): ...new function.
3156         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
3157         subclass of rtl_opt_pass along with...
3158         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
3159         and...
3160         (make_pass_convert_to_eh_region_ranges): ...new function.
3161         * final.c (pass_compute_alignments): Convert from a global struct to a
3162         subclass of rtl_opt_pass along with...
3163         (pass_data_compute_alignments): ...new pass_data instance and...
3164         (make_pass_compute_alignments): ...new function.
3165         (pass_final): Convert from a global struct to a subclass of
3166         rtl_opt_pass along with...
3167         (pass_data_final): ...new pass_data instance and...
3168         (make_pass_final): ...new function.
3169         (pass_shorten_branches): Convert from a global struct to a subclass of
3170         rtl_opt_pass along with...
3171         (pass_data_shorten_branches): ...new pass_data instance and...
3172         (make_pass_shorten_branches): ...new function.
3173         (pass_clean_state): Convert from a global struct to a subclass of
3174         rtl_opt_pass along with...
3175         (pass_data_clean_state): ...new pass_data instance and...
3176         (make_pass_clean_state): ...new function.
3177         * function.c (pass_instantiate_virtual_regs): Convert from a global
3178         struct to a subclass of rtl_opt_pass along with...
3179         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
3180         (make_pass_instantiate_virtual_regs): ...new function.
3181         (pass_leaf_regs): Convert from a global struct to a subclass of
3182         rtl_opt_pass along with...
3183         (pass_data_leaf_regs): ...new pass_data instance and...
3184         (make_pass_leaf_regs): ...new function.
3185         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
3186         subclass of rtl_opt_pass along with...
3187         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
3188         and...
3189         (make_pass_thread_prologue_and_epilogue): ...new function.
3190         (pass_match_asm_constraints): Convert from a global struct to a
3191         subclass of rtl_opt_pass along with...
3192         (pass_data_match_asm_constraints): ...new pass_data instance and...
3193         (make_pass_match_asm_constraints): ...new function.
3194         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
3195         subclass of rtl_opt_pass along with...
3196         (pass_data_rtl_fwprop): ...new pass_data instance and...
3197         (make_pass_rtl_fwprop): ...new function.
3198         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
3199         rtl_opt_pass along with...
3200         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
3201         (make_pass_rtl_fwprop_addr): ...new function.
3202         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
3203         rtl_opt_pass along with...
3204         (pass_data_rtl_pre): ...new pass_data instance and...
3205         (make_pass_rtl_pre): ...new function.
3206         (pass_rtl_hoist): Convert from a global struct to a subclass of
3207         rtl_opt_pass along with...
3208         (pass_data_rtl_hoist): ...new pass_data instance and...
3209         (make_pass_rtl_hoist): ...new function.
3210         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
3211         subclass of gimple_opt_pass along with...
3212         (pass_data_lower_cf): ...new pass_data instance and...
3213         (make_pass_lower_cf): ...new function.
3214         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
3215         from a global struct to a subclass of gimple_opt_pass along with...
3216         (pass_data_strength_reduction): ...new pass_data instance and...
3217         (make_pass_strength_reduction): ...new function.
3218         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
3219         of rtl_opt_pass along with...
3220         (pass_data_rtl_ifcvt): ...new pass_data instance and...
3221         (make_pass_rtl_ifcvt): ...new function.
3222         (pass_if_after_combine): Convert from a global struct to a subclass of
3223         rtl_opt_pass along with...
3224         (pass_data_if_after_combine): ...new pass_data instance and...
3225         (make_pass_if_after_combine): ...new function.
3226         (pass_if_after_reload): Convert from a global struct to a subclass of
3227         rtl_opt_pass along with...
3228         (pass_data_if_after_reload): ...new pass_data instance and...
3229         (make_pass_if_after_reload): ...new function.
3230         * init-regs.c (pass_initialize_regs): Convert from a global struct to
3231         a subclass of rtl_opt_pass along with...
3232         (pass_data_initialize_regs): ...new pass_data instance and...
3233         (make_pass_initialize_regs): ...new function.
3234         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
3235         of ipa_opt_pass_d along with...
3236         (pass_data_ipa_cp): ...new pass_data instance and...
3237         (make_pass_ipa_cp): ...new function.
3238         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
3239         global struct to a subclass of gimple_opt_pass along with...
3240         (pass_data_inline_parameters): ...new pass_data instance and...
3241         (make_pass_inline_parameters): ...new function.
3242         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
3243         subclass of gimple_opt_pass along with...
3244         (pass_data_early_inline): ...new pass_data instance and...
3245         (make_pass_early_inline): ...new function.
3246         (pass_ipa_inline): Convert from a global struct to a subclass of
3247         ipa_opt_pass_d along with...
3248         (pass_data_ipa_inline): ...new pass_data instance and...
3249         (make_pass_ipa_inline): ...new function.
3250         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
3251         struct to a subclass of gimple_opt_pass along with...
3252         (pass_data_local_pure_const): ...new pass_data instance and...
3253         (make_pass_local_pure_const): ...new function.
3254         (pass_ipa_pure_const): Convert from a global struct to a subclass of
3255         ipa_opt_pass_d along with...
3256         (pass_data_ipa_pure_const): ...new pass_data instance and...
3257         (make_pass_ipa_pure_const): ...new function.
3258         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
3259         to a subclass of ipa_opt_pass_d along with...
3260         (pass_data_ipa_reference): ...new pass_data instance and...
3261         (make_pass_ipa_reference): ...new function.
3262         * ipa-split.c (pass_split_functions): Convert from a global struct to
3263         a subclass of gimple_opt_pass along with...
3264         (pass_data_split_functions): ...new pass_data instance and...
3265         (make_pass_split_functions): ...new function.
3266         (pass_feedback_split_functions): Convert from a global struct to a
3267         subclass of gimple_opt_pass along with...
3268         (pass_data_feedback_split_functions): ...new pass_data instance and...
3269         (make_pass_feedback_split_functions): ...new function.
3270         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
3271         global struct to a subclass of simple_ipa_opt_pass along with...
3272         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
3273         instance and...
3274         (make_pass_ipa_function_and_variable_visibility): ...new function.
3275         (pass_ipa_free_inline_summary): Convert from a global struct to a
3276         subclass of simple_ipa_opt_pass along with...
3277         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
3278         (make_pass_ipa_free_inline_summary): ...new function.
3279         (pass_ipa_whole_program_visibility): Convert from a global struct to a
3280         subclass of ipa_opt_pass_d along with...
3281         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
3282         and...
3283         (make_pass_ipa_whole_program_visibility): ...new function.
3284         (pass_ipa_profile): Convert from a global struct to a subclass of
3285         ipa_opt_pass_d along with...
3286         (pass_data_ipa_profile): ...new pass_data instance and...
3287         (make_pass_ipa_profile): ...new function.
3288         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
3289         ipa_opt_pass_d along with...
3290         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
3291         (make_pass_ipa_cdtor_merge): ...new function.
3292         * ira.c (pass_ira): Convert from a global struct to a subclass of
3293         rtl_opt_pass along with...
3294         (pass_data_ira): ...new pass_data instance and...
3295         (make_pass_ira): ...new function.
3296         (pass_reload): Convert from a global struct to a subclass of
3297         rtl_opt_pass along with...
3298         (pass_data_reload): ...new pass_data instance and...
3299         (make_pass_reload): ...new function.
3300         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
3301         subclass of rtl_opt_pass along with...
3302         (pass_data_cleanup_barriers): ...new pass_data instance and...
3303         (make_pass_cleanup_barriers): ...new function.
3304         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
3305         of rtl_opt_pass along with...
3306         (pass_data_loop2): ...new pass_data instance and...
3307         (make_pass_loop2): ...new function.
3308         (pass_rtl_loop_init): Convert from a global struct to a subclass of
3309         rtl_opt_pass along with...
3310         (pass_data_rtl_loop_init): ...new pass_data instance and...
3311         (make_pass_rtl_loop_init): ...new function.
3312         (pass_rtl_loop_done): Convert from a global struct to a subclass of
3313         rtl_opt_pass along with...
3314         (pass_data_rtl_loop_done): ...new pass_data instance and...
3315         (make_pass_rtl_loop_done): ...new function.
3316         (pass_rtl_move_loop_invariants): Convert from a global struct to a
3317         subclass of rtl_opt_pass along with...
3318         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
3319         (make_pass_rtl_move_loop_invariants): ...new function.
3320         (pass_rtl_unswitch): Convert from a global struct to a subclass of
3321         rtl_opt_pass along with...
3322         (pass_data_rtl_unswitch): ...new pass_data instance and...
3323         (make_pass_rtl_unswitch): ...new function.
3324         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
3325         subclass of rtl_opt_pass along with...
3326         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
3327         and...
3328         (make_pass_rtl_unroll_and_peel_loops): ...new function.
3329         (pass_rtl_doloop): Convert from a global struct to a subclass of
3330         rtl_opt_pass along with...
3331         (pass_data_rtl_doloop): ...new pass_data instance and...
3332         (make_pass_rtl_doloop): ...new function.
3333         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
3334         a subclass of rtl_opt_pass along with...
3335         (pass_data_lower_subreg): ...new pass_data instance and...
3336         (make_pass_lower_subreg): ...new function.
3337         (pass_lower_subreg2): Convert from a global struct to a subclass of
3338         rtl_opt_pass along with...
3339         (pass_data_lower_subreg2): ...new pass_data instance and...
3340         (make_pass_lower_subreg2): ...new function.
3341         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
3342         struct to a subclass of ipa_opt_pass_d along with...
3343         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
3344         (make_pass_ipa_lto_gimple_out): ...new function.
3345         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
3346         of ipa_opt_pass_d along with...
3347         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
3348         (make_pass_ipa_lto_finish_out): ...new function.
3349         * mode-switching.c (pass_mode_switching): Convert from a global struct
3350         to a subclass of rtl_opt_pass along with...
3351         (pass_data_mode_switching): ...new pass_data instance and...
3352         (make_pass_mode_switching): ...new function.
3353         * modulo-sched.c (pass_sms): Convert from a global struct to a
3354         subclass of rtl_opt_pass along with...
3355         (pass_data_sms): ...new pass_data instance and...
3356         (make_pass_sms): ...new function.
3357         * omp-low.c (pass_expand_omp): Convert from a global struct to a
3358         subclass of gimple_opt_pass along with...
3359         (pass_data_expand_omp): ...new pass_data instance and...
3360         (make_pass_expand_omp): ...new function.
3361         (pass_lower_omp): Convert from a global struct to a subclass of
3362         gimple_opt_pass along with...
3363         (pass_data_lower_omp): ...new pass_data instance and...
3364         (make_pass_lower_omp): ...new function.
3365         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
3366         of gimple_opt_pass along with...
3367         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
3368         (make_pass_diagnose_omp_blocks): ...new function.
3369         * passes.c (pass_early_local_passes): Convert from a global struct to
3370         a subclass of simple_ipa_opt_pass along with...
3371         (pass_data_early_local_passes): ...new pass_data instance and...
3372         (make_pass_early_local_passes): ...new function.
3373         (pass_all_early_optimizations): Convert from a global struct to a
3374         subclass of gimple_opt_pass along with...
3375         (pass_data_all_early_optimizations): ...new pass_data instance and...
3376         (make_pass_all_early_optimizations): ...new function.
3377         (pass_all_optimizations): Convert from a global struct to a subclass
3378         of gimple_opt_pass along with...
3379         (pass_data_all_optimizations): ...new pass_data instance and...
3380         (make_pass_all_optimizations): ...new function.
3381         (pass_all_optimizations_g): Convert from a global struct to a subclass
3382         of gimple_opt_pass along with...
3383         (pass_data_all_optimizations_g): ...new pass_data instance and...
3384         (make_pass_all_optimizations_g): ...new function.
3385         (pass_rest_of_compilation): Convert from a global struct to a subclass
3386         of rtl_opt_pass along with...
3387         (pass_data_rest_of_compilation): ...new pass_data instance and...
3388         (make_pass_rest_of_compilation): ...new function.
3389         (pass_postreload): Convert from a global struct to a subclass of
3390         rtl_opt_pass along with...
3391         (pass_data_postreload): ...new pass_data instance and...
3392         (make_pass_postreload): ...new function.
3393         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
3394         subclass of rtl_opt_pass along with...
3395         (pass_data_gcse2): ...new pass_data instance and...
3396         (make_pass_gcse2): ...new function.
3397         * postreload.c (pass_postreload_cse): Convert from a global struct to
3398         a subclass of rtl_opt_pass along with...
3399         (pass_data_postreload_cse): ...new pass_data instance and...
3400         (make_pass_postreload_cse): ...new function.
3401         * predict.c (pass_profile): Convert from a global struct to a subclass
3402         of gimple_opt_pass along with...
3403         (pass_data_profile): ...new pass_data instance and...
3404         (make_pass_profile): ...new function.
3405         (pass_strip_predict_hints): Convert from a global struct to a subclass
3406         of gimple_opt_pass along with...
3407         (pass_data_strip_predict_hints): ...new pass_data instance and...
3408         (make_pass_strip_predict_hints): ...new function.
3409         * recog.c (pass_peephole2): Convert from a global struct to a subclass
3410         of rtl_opt_pass along with...
3411         (pass_data_peephole2): ...new pass_data instance and...
3412         (make_pass_peephole2): ...new function.
3413         (pass_split_all_insns): Convert from a global struct to a subclass of
3414         rtl_opt_pass along with...
3415         (pass_data_split_all_insns): ...new pass_data instance and...
3416         (make_pass_split_all_insns): ...new function.
3417         (pass_split_after_reload): Convert from a global struct to a subclass
3418         of rtl_opt_pass along with...
3419         (pass_data_split_after_reload): ...new pass_data instance and...
3420         (make_pass_split_after_reload): ...new function.
3421         (pass_split_before_regstack): Convert from a global struct to a
3422         subclass of rtl_opt_pass along with...
3423         (pass_data_split_before_regstack): ...new pass_data instance and...
3424         (make_pass_split_before_regstack): ...new function.
3425         (pass_split_before_sched2): Convert from a global struct to a subclass
3426         of rtl_opt_pass along with...
3427         (pass_data_split_before_sched2): ...new pass_data instance and...
3428         (make_pass_split_before_sched2): ...new function.
3429         (pass_split_for_shorten_branches): Convert from a global struct to a
3430         subclass of rtl_opt_pass along with...
3431         (pass_data_split_for_shorten_branches): ...new pass_data instance
3432         and...
3433         (make_pass_split_for_shorten_branches): ...new function.
3434         * ree.c (pass_ree): Convert from a global struct to a subclass of
3435         rtl_opt_pass along with...
3436         (pass_data_ree): ...new pass_data instance and...
3437         (make_pass_ree): ...new function.
3438         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
3439         subclass of rtl_opt_pass along with...
3440         (pass_data_stack_regs): ...new pass_data instance and...
3441         (make_pass_stack_regs): ...new function.
3442         (pass_stack_regs_run): Convert from a global struct to a subclass of
3443         rtl_opt_pass along with...
3444         (pass_data_stack_regs_run): ...new pass_data instance and...
3445         (make_pass_stack_regs_run): ...new function.
3446         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
3447         subclass of rtl_opt_pass along with...
3448         (pass_data_cprop_hardreg): ...new pass_data instance and...
3449         (make_pass_cprop_hardreg): ...new function.
3450         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
3451         subclass of rtl_opt_pass along with...
3452         (pass_data_reginfo_init): ...new pass_data instance and...
3453         (make_pass_reginfo_init): ...new function.
3454         * regmove.c (pass_regmove): Convert from a global struct to a subclass
3455         of rtl_opt_pass along with...
3456         (pass_data_regmove): ...new pass_data instance and...
3457         (make_pass_regmove): ...new function.
3458         * regrename.c (pass_regrename): Convert from a global struct to a
3459         subclass of rtl_opt_pass along with...
3460         (pass_data_regrename): ...new pass_data instance and...
3461         (make_pass_regrename): ...new function.
3462         * reorg.c (pass_delay_slots): Convert from a global struct to a
3463         subclass of rtl_opt_pass along with...
3464         (pass_data_delay_slots): ...new pass_data instance and...
3465         (make_pass_delay_slots): ...new function.
3466         (pass_machine_reorg): Convert from a global struct to a subclass of
3467         rtl_opt_pass along with...
3468         (pass_data_machine_reorg): ...new pass_data instance and...
3469         (make_pass_machine_reorg): ...new function.
3470         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
3471         of rtl_opt_pass along with...
3472         (pass_data_sched): ...new pass_data instance and...
3473         (make_pass_sched): ...new function.
3474         (pass_sched2): Convert from a global struct to a subclass of
3475         rtl_opt_pass along with...
3476         (pass_data_sched2): ...new pass_data instance and...
3477         (make_pass_sched2): ...new function.
3478         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
3479         to a subclass of rtl_opt_pass along with...
3480         (pass_data_stack_ptr_mod): ...new pass_data instance and...
3481         (make_pass_stack_ptr_mod): ...new function.
3482         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
3483         to a subclass of rtl_opt_pass along with...
3484         (pass_data_rtl_store_motion): ...new pass_data instance and...
3485         (make_pass_rtl_store_motion): ...new function.
3486         * tracer.c (pass_tracer): Convert from a global struct to a subclass
3487         of gimple_opt_pass along with...
3488         (pass_data_tracer): ...new pass_data instance and...
3489         (make_pass_tracer): ...new function.
3490         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
3491         to a subclass of gimple_opt_pass along with...
3492         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
3493         (make_pass_diagnose_tm_blocks): ...new function.
3494         (pass_lower_tm): Convert from a global struct to a subclass of
3495         gimple_opt_pass along with...
3496         (pass_data_lower_tm): ...new pass_data instance and...
3497         (make_pass_lower_tm): ...new function.
3498         (pass_tm_init): Convert from a global struct to a subclass of
3499         gimple_opt_pass along with...
3500         (pass_data_tm_init): ...new pass_data instance and...
3501         (make_pass_tm_init): ...new function.
3502         (pass_tm_mark): Convert from a global struct to a subclass of
3503         gimple_opt_pass along with...
3504         (pass_data_tm_mark): ...new pass_data instance and...
3505         (make_pass_tm_mark): ...new function.
3506         (pass_tm_edges): Convert from a global struct to a subclass of
3507         gimple_opt_pass along with...
3508         (pass_data_tm_edges): ...new pass_data instance and...
3509         (make_pass_tm_edges): ...new function.
3510         (pass_tm_memopt): Convert from a global struct to a subclass of
3511         gimple_opt_pass along with...
3512         (pass_data_tm_memopt): ...new pass_data instance and...
3513         (make_pass_tm_memopt): ...new function.
3514         (pass_ipa_tm): Convert from a global struct to a subclass of
3515         simple_ipa_opt_pass along with...
3516         (pass_data_ipa_tm): ...new pass_data instance and...
3517         (make_pass_ipa_tm): ...new function.
3518         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
3519         subclass of gimple_opt_pass along with...
3520         (pass_data_call_cdce): ...new pass_data instance and...
3521         (make_pass_call_cdce): ...new function.
3522         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
3523         subclass of gimple_opt_pass along with...
3524         (pass_data_build_cfg): ...new pass_data instance and...
3525         (make_pass_build_cfg): ...new function.
3526         (pass_split_crit_edges): Convert from a global struct to a subclass of
3527         gimple_opt_pass along with...
3528         (pass_data_split_crit_edges): ...new pass_data instance and...
3529         (make_pass_split_crit_edges): ...new function.
3530         (pass_warn_function_return): Convert from a global struct to a
3531         subclass of gimple_opt_pass along with...
3532         (pass_data_warn_function_return): ...new pass_data instance and...
3533         (make_pass_warn_function_return): ...new function.
3534         (pass_warn_function_noreturn): Convert from a global struct to a
3535         subclass of gimple_opt_pass along with...
3536         (pass_data_warn_function_noreturn): ...new pass_data instance and...
3537         (make_pass_warn_function_noreturn): ...new function.
3538         (pass_warn_unused_result): Convert from a global struct to a subclass
3539         of gimple_opt_pass along with...
3540         (pass_data_warn_unused_result): ...new pass_data instance and...
3541         (make_pass_warn_unused_result): ...new function.
3542         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
3543         a subclass of gimple_opt_pass along with...
3544         (pass_data_merge_phi): ...new pass_data instance and...
3545         (make_pass_merge_phi): ...new function.
3546         * tree-complex.c (pass_lower_complex): Convert from a global struct to
3547         a subclass of gimple_opt_pass along with...
3548         (pass_data_lower_complex): ...new pass_data instance and...
3549         (make_pass_lower_complex): ...new function.
3550         (pass_lower_complex_O0): Convert from a global struct to a subclass of
3551         gimple_opt_pass along with...
3552         (pass_data_lower_complex_O0): ...new pass_data instance and...
3553         (make_pass_lower_complex_O0): ...new function.
3554         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
3555         subclass of gimple_opt_pass along with...
3556         (pass_data_lower_eh): ...new pass_data instance and...
3557         (make_pass_lower_eh): ...new function.
3558         (pass_refactor_eh): Convert from a global struct to a subclass of
3559         gimple_opt_pass along with...
3560         (pass_data_refactor_eh): ...new pass_data instance and...
3561         (make_pass_refactor_eh): ...new function.
3562         (pass_lower_resx): Convert from a global struct to a subclass of
3563         gimple_opt_pass along with...
3564         (pass_data_lower_resx): ...new pass_data instance and...
3565         (make_pass_lower_resx): ...new function.
3566         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
3567         of gimple_opt_pass along with...
3568         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
3569         (make_pass_lower_eh_dispatch): ...new function.
3570         (pass_cleanup_eh): Convert from a global struct to a subclass of
3571         gimple_opt_pass along with...
3572         (pass_data_cleanup_eh): ...new pass_data instance and...
3573         (make_pass_cleanup_eh): ...new function.
3574         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
3575         to a subclass of simple_ipa_opt_pass along with...
3576         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
3577         (make_pass_ipa_lower_emutls): ...new function.
3578         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
3579         a subclass of gimple_opt_pass along with...
3580         (pass_data_if_conversion): ...new pass_data instance and...
3581         (make_pass_if_conversion): ...new function.
3582         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
3583         subclass of gimple_opt_pass along with...
3584         (pass_data_build_ssa): ...new pass_data instance and...
3585         (make_pass_build_ssa): ...new function.
3586         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
3587         global struct to a subclass of gimple_opt_pass along with...
3588         (pass_data_loop_distribution): ...new pass_data instance and...
3589         (make_pass_loop_distribution): ...new function.
3590         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
3591         subclass of gimple_opt_pass along with...
3592         (pass_data_mudflap_1): ...new pass_data instance and...
3593         (make_pass_mudflap_1): ...new function.
3594         (pass_mudflap_2): Convert from a global struct to a subclass of
3595         gimple_opt_pass along with...
3596         (pass_data_mudflap_2): ...new pass_data instance and...
3597         (make_pass_mudflap_2): ...new function.
3598         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
3599         subclass of gimple_opt_pass along with...
3600         (pass_data_mudflap_1): ...new pass_data instance and...
3601         (make_pass_mudflap_1): ...new function.
3602         (pass_mudflap_2): Convert from a global struct to a subclass of
3603         gimple_opt_pass along with...
3604         (pass_data_mudflap_2): ...new pass_data instance and...
3605         (make_pass_mudflap_2): ...new function.
3606         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
3607         gimple_opt_pass along with...
3608         (pass_data_nrv): ...new pass_data instance and...
3609         (make_pass_nrv): ...new function.
3610         (pass_return_slot): Convert from a global struct to a subclass of
3611         gimple_opt_pass along with...
3612         (pass_data_return_slot): ...new pass_data instance and...
3613         (make_pass_return_slot): ...new function.
3614         * tree-object-size.c (pass_object_sizes): Convert from a global struct
3615         to a subclass of gimple_opt_pass along with...
3616         (pass_data_object_sizes): ...new pass_data instance and...
3617         (make_pass_object_sizes): ...new function.
3618         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
3619         global struct to a subclass of gimple_opt_pass along with...
3620         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
3621         and...
3622         (make_pass_cleanup_cfg_post_optimizing): ...new function.
3623         (pass_fixup_cfg): Convert from a global struct to a subclass of
3624         gimple_opt_pass along with...
3625         (pass_data_fixup_cfg): ...new pass_data instance and...
3626         (make_pass_fixup_cfg): ...new function.
3627         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
3628         (make_pass_mudflap_1): ...new function.
3629         (pass_mudflap_2): Replace declaration with that of...
3630         (make_pass_mudflap_2): ...new function.
3631         (pass_asan): Replace declaration with that of...
3632         (make_pass_asan): ...new function.
3633         (pass_asan_O0): Replace declaration with that of...
3634         (make_pass_asan_O0): ...new function.
3635         (pass_tsan): Replace declaration with that of...
3636         (make_pass_tsan): ...new function.
3637         (pass_tsan_O0): Replace declaration with that of...
3638         (make_pass_tsan_O0): ...new function.
3639         (pass_lower_cf): Replace declaration with that of...
3640         (make_pass_lower_cf): ...new function.
3641         (pass_refactor_eh): Replace declaration with that of...
3642         (make_pass_refactor_eh): ...new function.
3643         (pass_lower_eh): Replace declaration with that of...
3644         (make_pass_lower_eh): ...new function.
3645         (pass_lower_eh_dispatch): Replace declaration with that of...
3646         (make_pass_lower_eh_dispatch): ...new function.
3647         (pass_lower_resx): Replace declaration with that of...
3648         (make_pass_lower_resx): ...new function.
3649         (pass_build_cfg): Replace declaration with that of...
3650         (make_pass_build_cfg): ...new function.
3651         (pass_early_tree_profile): Replace declaration with that of...
3652         (make_pass_early_tree_profile): ...new function.
3653         (pass_cleanup_eh): Replace declaration with that of...
3654         (make_pass_cleanup_eh): ...new function.
3655         (pass_sra): Replace declaration with that of...
3656         (make_pass_sra): ...new function.
3657         (pass_sra_early): Replace declaration with that of...
3658         (make_pass_sra_early): ...new function.
3659         (pass_early_ipa_sra): Replace declaration with that of...
3660         (make_pass_early_ipa_sra): ...new function.
3661         (pass_tail_recursion): Replace declaration with that of...
3662         (make_pass_tail_recursion): ...new function.
3663         (pass_tail_calls): Replace declaration with that of...
3664         (make_pass_tail_calls): ...new function.
3665         (pass_tree_loop): Replace declaration with that of...
3666         (make_pass_tree_loop): ...new function.
3667         (pass_tree_loop_init): Replace declaration with that of...
3668         (make_pass_tree_loop_init): ...new function.
3669         (pass_lim): Replace declaration with that of...
3670         (make_pass_lim): ...new function.
3671         (pass_tree_unswitch): Replace declaration with that of...
3672         (make_pass_tree_unswitch): ...new function.
3673         (pass_predcom): Replace declaration with that of...
3674         (make_pass_predcom): ...new function.
3675         (pass_iv_canon): Replace declaration with that of...
3676         (make_pass_iv_canon): ...new function.
3677         (pass_scev_cprop): Replace declaration with that of...
3678         (make_pass_scev_cprop): ...new function.
3679         (pass_empty_loop): Replace declaration with that of...
3680         (make_pass_empty_loop): ...new function.
3681         (pass_record_bounds): Replace declaration with that of...
3682         (make_pass_record_bounds): ...new function.
3683         (pass_graphite): Replace declaration with that of...
3684         (make_pass_graphite): ...new function.
3685         (pass_graphite_transforms): Replace declaration with that of...
3686         (make_pass_graphite_transforms): ...new function.
3687         (pass_if_conversion): Replace declaration with that of...
3688         (make_pass_if_conversion): ...new function.
3689         (pass_loop_distribution): Replace declaration with that of...
3690         (make_pass_loop_distribution): ...new function.
3691         (pass_vectorize): Replace declaration with that of...
3692         (make_pass_vectorize): ...new function.
3693         (pass_slp_vectorize): Replace declaration with that of...
3694         (make_pass_slp_vectorize): ...new function.
3695         (pass_complete_unroll): Replace declaration with that of...
3696         (make_pass_complete_unroll): ...new function.
3697         (pass_complete_unrolli): Replace declaration with that of...
3698         (make_pass_complete_unrolli): ...new function.
3699         (pass_parallelize_loops): Replace declaration with that of...
3700         (make_pass_parallelize_loops): ...new function.
3701         (pass_loop_prefetch): Replace declaration with that of...
3702         (make_pass_loop_prefetch): ...new function.
3703         (pass_iv_optimize): Replace declaration with that of...
3704         (make_pass_iv_optimize): ...new function.
3705         (pass_tree_loop_done): Replace declaration with that of...
3706         (make_pass_tree_loop_done): ...new function.
3707         (pass_ch): Replace declaration with that of...
3708         (make_pass_ch): ...new function.
3709         (pass_ccp): Replace declaration with that of...
3710         (make_pass_ccp): ...new function.
3711         (pass_phi_only_cprop): Replace declaration with that of...
3712         (make_pass_phi_only_cprop): ...new function.
3713         (pass_build_ssa): Replace declaration with that of...
3714         (make_pass_build_ssa): ...new function.
3715         (pass_build_alias): Replace declaration with that of...
3716         (make_pass_build_alias): ...new function.
3717         (pass_build_ealias): Replace declaration with that of...
3718         (make_pass_build_ealias): ...new function.
3719         (pass_dominator): Replace declaration with that of...
3720         (make_pass_dominator): ...new function.
3721         (pass_dce): Replace declaration with that of...
3722         (make_pass_dce): ...new function.
3723         (pass_dce_loop): Replace declaration with that of...
3724         (make_pass_dce_loop): ...new function.
3725         (pass_cd_dce): Replace declaration with that of...
3726         (make_pass_cd_dce): ...new function.
3727         (pass_call_cdce): Replace declaration with that of...
3728         (make_pass_call_cdce): ...new function.
3729         (pass_merge_phi): Replace declaration with that of...
3730         (make_pass_merge_phi): ...new function.
3731         (pass_split_crit_edges): Replace declaration with that of...
3732         (make_pass_split_crit_edges): ...new function.
3733         (pass_pre): Replace declaration with that of...
3734         (make_pass_pre): ...new function.
3735         (pass_profile): Replace declaration with that of...
3736         (make_pass_profile): ...new function.
3737         (pass_strip_predict_hints): Replace declaration with that of...
3738         (make_pass_strip_predict_hints): ...new function.
3739         (pass_lower_complex_O0): Replace declaration with that of...
3740         (make_pass_lower_complex_O0): ...new function.
3741         (pass_lower_complex): Replace declaration with that of...
3742         (make_pass_lower_complex): ...new function.
3743         (pass_lower_vector): Replace declaration with that of...
3744         (make_pass_lower_vector): ...new function.
3745         (pass_lower_vector_ssa): Replace declaration with that of...
3746         (make_pass_lower_vector_ssa): ...new function.
3747         (pass_lower_omp): Replace declaration with that of...
3748         (make_pass_lower_omp): ...new function.
3749         (pass_diagnose_omp_blocks): Replace declaration with that of...
3750         (make_pass_diagnose_omp_blocks): ...new function.
3751         (pass_expand_omp): Replace declaration with that of...
3752         (make_pass_expand_omp): ...new function.
3753         (pass_expand_omp_ssa): Replace declaration with that of...
3754         (make_pass_expand_omp_ssa): ...new function.
3755         (pass_object_sizes): Replace declaration with that of...
3756         (make_pass_object_sizes): ...new function.
3757         (pass_strlen): Replace declaration with that of...
3758         (make_pass_strlen): ...new function.
3759         (pass_fold_builtins): Replace declaration with that of...
3760         (make_pass_fold_builtins): ...new function.
3761         (pass_stdarg): Replace declaration with that of...
3762         (make_pass_stdarg): ...new function.
3763         (pass_early_warn_uninitialized): Replace declaration with that of...
3764         (make_pass_early_warn_uninitialized): ...new function.
3765         (pass_late_warn_uninitialized): Replace declaration with that of...
3766         (make_pass_late_warn_uninitialized): ...new function.
3767         (pass_cse_reciprocals): Replace declaration with that of...
3768         (make_pass_cse_reciprocals): ...new function.
3769         (pass_cse_sincos): Replace declaration with that of...
3770         (make_pass_cse_sincos): ...new function.
3771         (pass_optimize_bswap): Replace declaration with that of...
3772         (make_pass_optimize_bswap): ...new function.
3773         (pass_optimize_widening_mul): Replace declaration with that of...
3774         (make_pass_optimize_widening_mul): ...new function.
3775         (pass_warn_function_return): Replace declaration with that of...
3776         (make_pass_warn_function_return): ...new function.
3777         (pass_warn_function_noreturn): Replace declaration with that of...
3778         (make_pass_warn_function_noreturn): ...new function.
3779         (pass_cselim): Replace declaration with that of...
3780         (make_pass_cselim): ...new function.
3781         (pass_phiopt): Replace declaration with that of...
3782         (make_pass_phiopt): ...new function.
3783         (pass_forwprop): Replace declaration with that of...
3784         (make_pass_forwprop): ...new function.
3785         (pass_phiprop): Replace declaration with that of...
3786         (make_pass_phiprop): ...new function.
3787         (pass_tree_ifcombine): Replace declaration with that of...
3788         (make_pass_tree_ifcombine): ...new function.
3789         (pass_dse): Replace declaration with that of...
3790         (make_pass_dse): ...new function.
3791         (pass_nrv): Replace declaration with that of...
3792         (make_pass_nrv): ...new function.
3793         (pass_rename_ssa_copies): Replace declaration with that of...
3794         (make_pass_rename_ssa_copies): ...new function.
3795         (pass_sink_code): Replace declaration with that of...
3796         (make_pass_sink_code): ...new function.
3797         (pass_fre): Replace declaration with that of...
3798         (make_pass_fre): ...new function.
3799         (pass_check_data_deps): Replace declaration with that of...
3800         (make_pass_check_data_deps): ...new function.
3801         (pass_copy_prop): Replace declaration with that of...
3802         (make_pass_copy_prop): ...new function.
3803         (pass_vrp): Replace declaration with that of...
3804         (make_pass_vrp): ...new function.
3805         (pass_uncprop): Replace declaration with that of...
3806         (make_pass_uncprop): ...new function.
3807         (pass_return_slot): Replace declaration with that of...
3808         (make_pass_return_slot): ...new function.
3809         (pass_reassoc): Replace declaration with that of...
3810         (make_pass_reassoc): ...new function.
3811         (pass_rebuild_cgraph_edges): Replace declaration with that of...
3812         (make_pass_rebuild_cgraph_edges): ...new function.
3813         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
3814         (make_pass_remove_cgraph_callee_edges): ...new function.
3815         (pass_build_cgraph_edges): Replace declaration with that of...
3816         (make_pass_build_cgraph_edges): ...new function.
3817         (pass_local_pure_const): Replace declaration with that of...
3818         (make_pass_local_pure_const): ...new function.
3819         (pass_tracer): Replace declaration with that of...
3820         (make_pass_tracer): ...new function.
3821         (pass_warn_unused_result): Replace declaration with that of...
3822         (make_pass_warn_unused_result): ...new function.
3823         (pass_diagnose_tm_blocks): Replace declaration with that of...
3824         (make_pass_diagnose_tm_blocks): ...new function.
3825         (pass_lower_tm): Replace declaration with that of...
3826         (make_pass_lower_tm): ...new function.
3827         (pass_tm_init): Replace declaration with that of...
3828         (make_pass_tm_init): ...new function.
3829         (pass_tm_mark): Replace declaration with that of...
3830         (make_pass_tm_mark): ...new function.
3831         (pass_tm_memopt): Replace declaration with that of...
3832         (make_pass_tm_memopt): ...new function.
3833         (pass_tm_edges): Replace declaration with that of...
3834         (make_pass_tm_edges): ...new function.
3835         (pass_split_functions): Replace declaration with that of...
3836         (make_pass_split_functions): ...new function.
3837         (pass_feedback_split_functions): Replace declaration with that of...
3838         (make_pass_feedback_split_functions): ...new function.
3839         (pass_strength_reduction): Replace declaration with that of...
3840         (make_pass_strength_reduction): ...new function.
3841         (pass_ipa_lower_emutls): Replace declaration with that of...
3842         (make_pass_ipa_lower_emutls): ...new function.
3843         (pass_ipa_function_and_variable_visibility): Replace declaration with
3844         that of...
3845         (make_pass_ipa_function_and_variable_visibility): ...new function.
3846         (pass_ipa_tree_profile): Replace declaration with that of...
3847         (make_pass_ipa_tree_profile): ...new function.
3848         (pass_early_local_passes): Replace declaration with that of...
3849         (make_pass_early_local_passes): ...new function.
3850         (pass_ipa_whole_program_visibility): Replace declaration with that
3851         of...
3852         (make_pass_ipa_whole_program_visibility): ...new function.
3853         (pass_ipa_lto_gimple_out): Replace declaration with that of...
3854         (make_pass_ipa_lto_gimple_out): ...new function.
3855         (pass_ipa_increase_alignment): Replace declaration with that of...
3856         (make_pass_ipa_increase_alignment): ...new function.
3857         (pass_ipa_inline): Replace declaration with that of...
3858         (make_pass_ipa_inline): ...new function.
3859         (pass_ipa_free_lang_data): Replace declaration with that of...
3860         (make_pass_ipa_free_lang_data): ...new function.
3861         (pass_ipa_free_inline_summary): Replace declaration with that of...
3862         (make_pass_ipa_free_inline_summary): ...new function.
3863         (pass_ipa_cp): Replace declaration with that of...
3864         (make_pass_ipa_cp): ...new function.
3865         (pass_ipa_reference): Replace declaration with that of...
3866         (make_pass_ipa_reference): ...new function.
3867         (pass_ipa_pure_const): Replace declaration with that of...
3868         (make_pass_ipa_pure_const): ...new function.
3869         (pass_ipa_pta): Replace declaration with that of...
3870         (make_pass_ipa_pta): ...new function.
3871         (pass_ipa_lto_finish_out): Replace declaration with that of...
3872         (make_pass_ipa_lto_finish_out): ...new function.
3873         (pass_ipa_tm): Replace declaration with that of...
3874         (make_pass_ipa_tm): ...new function.
3875         (pass_ipa_profile): Replace declaration with that of...
3876         (make_pass_ipa_profile): ...new function.
3877         (pass_ipa_cdtor_merge): Replace declaration with that of...
3878         (make_pass_ipa_cdtor_merge): ...new function.
3879         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
3880         of...
3881         (make_pass_cleanup_cfg_post_optimizing): ...new function.
3882         (pass_init_datastructures): Replace declaration with that of...
3883         (make_pass_init_datastructures): ...new function.
3884         (pass_fixup_cfg): Replace declaration with that of...
3885         (make_pass_fixup_cfg): ...new function.
3886         (pass_expand): Replace declaration with that of...
3887         (make_pass_expand): ...new function.
3888         (pass_instantiate_virtual_regs): Replace declaration with that of...
3889         (make_pass_instantiate_virtual_regs): ...new function.
3890         (pass_rtl_fwprop): Replace declaration with that of...
3891         (make_pass_rtl_fwprop): ...new function.
3892         (pass_rtl_fwprop_addr): Replace declaration with that of...
3893         (make_pass_rtl_fwprop_addr): ...new function.
3894         (pass_jump): Replace declaration with that of...
3895         (make_pass_jump): ...new function.
3896         (pass_jump2): Replace declaration with that of...
3897         (make_pass_jump2): ...new function.
3898         (pass_lower_subreg): Replace declaration with that of...
3899         (make_pass_lower_subreg): ...new function.
3900         (pass_cse): Replace declaration with that of...
3901         (make_pass_cse): ...new function.
3902         (pass_fast_rtl_dce): Replace declaration with that of...
3903         (make_pass_fast_rtl_dce): ...new function.
3904         (pass_ud_rtl_dce): Replace declaration with that of...
3905         (make_pass_ud_rtl_dce): ...new function.
3906         (pass_rtl_dce): Replace declaration with that of...
3907         (make_pass_rtl_dce): ...new function.
3908         (pass_rtl_dse1): Replace declaration with that of...
3909         (make_pass_rtl_dse1): ...new function.
3910         (pass_rtl_dse2): Replace declaration with that of...
3911         (make_pass_rtl_dse2): ...new function.
3912         (pass_rtl_dse3): Replace declaration with that of...
3913         (make_pass_rtl_dse3): ...new function.
3914         (pass_rtl_cprop): Replace declaration with that of...
3915         (make_pass_rtl_cprop): ...new function.
3916         (pass_rtl_pre): Replace declaration with that of...
3917         (make_pass_rtl_pre): ...new function.
3918         (pass_rtl_hoist): Replace declaration with that of...
3919         (make_pass_rtl_hoist): ...new function.
3920         (pass_rtl_store_motion): Replace declaration with that of...
3921         (make_pass_rtl_store_motion): ...new function.
3922         (pass_cse_after_global_opts): Replace declaration with that of...
3923         (make_pass_cse_after_global_opts): ...new function.
3924         (pass_rtl_ifcvt): Replace declaration with that of...
3925         (make_pass_rtl_ifcvt): ...new function.
3926         (pass_into_cfg_layout_mode): Replace declaration with that of...
3927         (make_pass_into_cfg_layout_mode): ...new function.
3928         (pass_outof_cfg_layout_mode): Replace declaration with that of...
3929         (make_pass_outof_cfg_layout_mode): ...new function.
3930         (pass_loop2): Replace declaration with that of...
3931         (make_pass_loop2): ...new function.
3932         (pass_rtl_loop_init): Replace declaration with that of...
3933         (make_pass_rtl_loop_init): ...new function.
3934         (pass_rtl_move_loop_invariants): Replace declaration with that of...
3935         (make_pass_rtl_move_loop_invariants): ...new function.
3936         (pass_rtl_unswitch): Replace declaration with that of...
3937         (make_pass_rtl_unswitch): ...new function.
3938         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
3939         (make_pass_rtl_unroll_and_peel_loops): ...new function.
3940         (pass_rtl_doloop): Replace declaration with that of...
3941         (make_pass_rtl_doloop): ...new function.
3942         (pass_rtl_loop_done): Replace declaration with that of...
3943         (make_pass_rtl_loop_done): ...new function.
3944         (pass_web): Replace declaration with that of...
3945         (make_pass_web): ...new function.
3946         (pass_cse2): Replace declaration with that of...
3947         (make_pass_cse2): ...new function.
3948         (pass_df_initialize_opt): Replace declaration with that of...
3949         (make_pass_df_initialize_opt): ...new function.
3950         (pass_df_initialize_no_opt): Replace declaration with that of...
3951         (make_pass_df_initialize_no_opt): ...new function.
3952         (pass_reginfo_init): Replace declaration with that of...
3953         (make_pass_reginfo_init): ...new function.
3954         (pass_inc_dec): Replace declaration with that of...
3955         (make_pass_inc_dec): ...new function.
3956         (pass_stack_ptr_mod): Replace declaration with that of...
3957         (make_pass_stack_ptr_mod): ...new function.
3958         (pass_initialize_regs): Replace declaration with that of...
3959         (make_pass_initialize_regs): ...new function.
3960         (pass_combine): Replace declaration with that of...
3961         (make_pass_combine): ...new function.
3962         (pass_if_after_combine): Replace declaration with that of...
3963         (make_pass_if_after_combine): ...new function.
3964         (pass_ree): Replace declaration with that of...
3965         (make_pass_ree): ...new function.
3966         (pass_partition_blocks): Replace declaration with that of...
3967         (make_pass_partition_blocks): ...new function.
3968         (pass_match_asm_constraints): Replace declaration with that of...
3969         (make_pass_match_asm_constraints): ...new function.
3970         (pass_regmove): Replace declaration with that of...
3971         (make_pass_regmove): ...new function.
3972         (pass_split_all_insns): Replace declaration with that of...
3973         (make_pass_split_all_insns): ...new function.
3974         (pass_fast_rtl_byte_dce): Replace declaration with that of...
3975         (make_pass_fast_rtl_byte_dce): ...new function.
3976         (pass_lower_subreg2): Replace declaration with that of...
3977         (make_pass_lower_subreg2): ...new function.
3978         (pass_mode_switching): Replace declaration with that of...
3979         (make_pass_mode_switching): ...new function.
3980         (pass_sms): Replace declaration with that of...
3981         (make_pass_sms): ...new function.
3982         (pass_sched): Replace declaration with that of...
3983         (make_pass_sched): ...new function.
3984         (pass_ira): Replace declaration with that of...
3985         (make_pass_ira): ...new function.
3986         (pass_reload): Replace declaration with that of...
3987         (make_pass_reload): ...new function.
3988         (pass_clean_state): Replace declaration with that of...
3989         (make_pass_clean_state): ...new function.
3990         (pass_branch_prob): Replace declaration with that of...
3991         (make_pass_branch_prob): ...new function.
3992         (pass_value_profile_transformations): Replace declaration with that
3993         of...
3994         (make_pass_value_profile_transformations): ...new function.
3995         (pass_postreload_cse): Replace declaration with that of...
3996         (make_pass_postreload_cse): ...new function.
3997         (pass_gcse2): Replace declaration with that of...
3998         (make_pass_gcse2): ...new function.
3999         (pass_split_after_reload): Replace declaration with that of...
4000         (make_pass_split_after_reload): ...new function.
4001         (pass_branch_target_load_optimize1): Replace declaration with that
4002         of...
4003         (make_pass_branch_target_load_optimize1): ...new function.
4004         (pass_thread_prologue_and_epilogue): Replace declaration with that
4005         of...
4006         (make_pass_thread_prologue_and_epilogue): ...new function.
4007         (pass_stack_adjustments): Replace declaration with that of...
4008         (make_pass_stack_adjustments): ...new function.
4009         (pass_peephole2): Replace declaration with that of...
4010         (make_pass_peephole2): ...new function.
4011         (pass_if_after_reload): Replace declaration with that of...
4012         (make_pass_if_after_reload): ...new function.
4013         (pass_regrename): Replace declaration with that of...
4014         (make_pass_regrename): ...new function.
4015         (pass_cprop_hardreg): Replace declaration with that of...
4016         (make_pass_cprop_hardreg): ...new function.
4017         (pass_reorder_blocks): Replace declaration with that of...
4018         (make_pass_reorder_blocks): ...new function.
4019         (pass_branch_target_load_optimize2): Replace declaration with that
4020         of...
4021         (make_pass_branch_target_load_optimize2): ...new function.
4022         (pass_leaf_regs): Replace declaration with that of...
4023         (make_pass_leaf_regs): ...new function.
4024         (pass_split_before_sched2): Replace declaration with that of...
4025         (make_pass_split_before_sched2): ...new function.
4026         (pass_compare_elim_after_reload): Replace declaration with that of...
4027         (make_pass_compare_elim_after_reload): ...new function.
4028         (pass_sched2): Replace declaration with that of...
4029         (make_pass_sched2): ...new function.
4030         (pass_stack_regs): Replace declaration with that of...
4031         (make_pass_stack_regs): ...new function.
4032         (pass_stack_regs_run): Replace declaration with that of...
4033         (make_pass_stack_regs_run): ...new function.
4034         (pass_df_finish): Replace declaration with that of...
4035         (make_pass_df_finish): ...new function.
4036         (pass_compute_alignments): Replace declaration with that of...
4037         (make_pass_compute_alignments): ...new function.
4038         (pass_duplicate_computed_gotos): Replace declaration with that of...
4039         (make_pass_duplicate_computed_gotos): ...new function.
4040         (pass_variable_tracking): Replace declaration with that of...
4041         (make_pass_variable_tracking): ...new function.
4042         (pass_free_cfg): Replace declaration with that of...
4043         (make_pass_free_cfg): ...new function.
4044         (pass_machine_reorg): Replace declaration with that of...
4045         (make_pass_machine_reorg): ...new function.
4046         (pass_cleanup_barriers): Replace declaration with that of...
4047         (make_pass_cleanup_barriers): ...new function.
4048         (pass_delay_slots): Replace declaration with that of...
4049         (make_pass_delay_slots): ...new function.
4050         (pass_split_for_shorten_branches): Replace declaration with that of...
4051         (make_pass_split_for_shorten_branches): ...new function.
4052         (pass_split_before_regstack): Replace declaration with that of...
4053         (make_pass_split_before_regstack): ...new function.
4054         (pass_convert_to_eh_region_ranges): Replace declaration with that
4055         of...
4056         (make_pass_convert_to_eh_region_ranges): ...new function.
4057         (pass_shorten_branches): Replace declaration with that of...
4058         (make_pass_shorten_branches): ...new function.
4059         (pass_set_nothrow_function_flags): Replace declaration with that of...
4060         (make_pass_set_nothrow_function_flags): ...new function.
4061         (pass_dwarf2_frame): Replace declaration with that of...
4062         (make_pass_dwarf2_frame): ...new function.
4063         (pass_final): Replace declaration with that of...
4064         (make_pass_final): ...new function.
4065         (pass_rtl_seqabstr): Replace declaration with that of...
4066         (make_pass_rtl_seqabstr): ...new function.
4067         (pass_release_ssa_names): Replace declaration with that of...
4068         (make_pass_release_ssa_names): ...new function.
4069         (pass_early_inline): Replace declaration with that of...
4070         (make_pass_early_inline): ...new function.
4071         (pass_inline_parameters): Replace declaration with that of...
4072         (make_pass_inline_parameters): ...new function.
4073         (pass_update_address_taken): Replace declaration with that of...
4074         (make_pass_update_address_taken): ...new function.
4075         (pass_convert_switch): Replace declaration with that of...
4076         (make_pass_convert_switch): ...new function.
4077         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
4078         to a subclass of simple_ipa_opt_pass along with...
4079         (pass_data_ipa_tree_profile): ...new pass_data instance and...
4080         (make_pass_ipa_tree_profile): ...new function.
4081         * tree-sra.c (pass_sra_early): Convert from a global struct to a
4082         subclass of gimple_opt_pass along with...
4083         (pass_data_sra_early): ...new pass_data instance and...
4084         (make_pass_sra_early): ...new function.
4085         (pass_sra): Convert from a global struct to a subclass of
4086         gimple_opt_pass along with...
4087         (pass_data_sra): ...new pass_data instance and...
4088         (make_pass_sra): ...new function.
4089         (pass_early_ipa_sra): Convert from a global struct to a subclass of
4090         gimple_opt_pass along with...
4091         (pass_data_early_ipa_sra): ...new pass_data instance and...
4092         (make_pass_early_ipa_sra): ...new function.
4093         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
4094         subclass of gimple_opt_pass along with...
4095         (pass_data_ccp): ...new pass_data instance and...
4096         (make_pass_ccp): ...new function.
4097         (pass_fold_builtins): Convert from a global struct to a subclass of
4098         gimple_opt_pass along with...
4099         (pass_data_fold_builtins): ...new pass_data instance and...
4100         (make_pass_fold_builtins): ...new function.
4101         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
4102         subclass of gimple_opt_pass along with...
4103         (pass_data_copy_prop): ...new pass_data instance and...
4104         (make_pass_copy_prop): ...new function.
4105         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
4106         global struct to a subclass of gimple_opt_pass along with...
4107         (pass_data_rename_ssa_copies): ...new pass_data instance and...
4108         (make_pass_rename_ssa_copies): ...new function.
4109         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
4110         subclass of gimple_opt_pass along with...
4111         (pass_data_dce): ...new pass_data instance and...
4112         (make_pass_dce): ...new function.
4113         (pass_dce_loop): Convert from a global struct to a subclass of
4114         gimple_opt_pass along with...
4115         (pass_data_dce_loop): ...new pass_data instance and...
4116         (make_pass_dce_loop): ...new function.
4117         (pass_cd_dce): Convert from a global struct to a subclass of
4118         gimple_opt_pass along with...
4119         (pass_data_cd_dce): ...new pass_data instance and...
4120         (make_pass_cd_dce): ...new function.
4121         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
4122         subclass of gimple_opt_pass along with...
4123         (pass_data_dominator): ...new pass_data instance and...
4124         (make_pass_dominator): ...new function.
4125         (pass_phi_only_cprop): Convert from a global struct to a subclass of
4126         gimple_opt_pass along with...
4127         (pass_data_phi_only_cprop): ...new pass_data instance and...
4128         (make_pass_phi_only_cprop): ...new function.
4129         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
4130         subclass of gimple_opt_pass along with...
4131         (pass_data_dse): ...new pass_data instance and...
4132         (make_pass_dse): ...new function.
4133         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
4134         a subclass of gimple_opt_pass along with...
4135         (pass_data_forwprop): ...new pass_data instance and...
4136         (make_pass_forwprop): ...new function.
4137         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
4138         struct to a subclass of gimple_opt_pass along with...
4139         (pass_data_tree_ifcombine): ...new pass_data instance and...
4140         (make_pass_tree_ifcombine): ...new function.
4141         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
4142         subclass of gimple_opt_pass along with...
4143         (pass_data_ch): ...new pass_data instance and...
4144         (make_pass_ch): ...new function.
4145         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
4146         subclass of gimple_opt_pass along with...
4147         (pass_data_tree_loop): ...new pass_data instance and...
4148         (make_pass_tree_loop): ...new function.
4149         (pass_tree_loop_init): Convert from a global struct to a subclass of
4150         gimple_opt_pass along with...
4151         (pass_data_tree_loop_init): ...new pass_data instance and...
4152         (make_pass_tree_loop_init): ...new function.
4153         (pass_lim): Convert from a global struct to a subclass of
4154         gimple_opt_pass along with...
4155         (pass_data_lim): ...new pass_data instance and...
4156         (make_pass_lim): ...new function.
4157         (pass_tree_unswitch): Convert from a global struct to a subclass of
4158         gimple_opt_pass along with...
4159         (pass_data_tree_unswitch): ...new pass_data instance and...
4160         (make_pass_tree_unswitch): ...new function.
4161         (pass_predcom): Convert from a global struct to a subclass of
4162         gimple_opt_pass along with...
4163         (pass_data_predcom): ...new pass_data instance and...
4164         (make_pass_predcom): ...new function.
4165         (pass_vectorize): Convert from a global struct to a subclass of
4166         gimple_opt_pass along with...
4167         (pass_data_vectorize): ...new pass_data instance and...
4168         (make_pass_vectorize): ...new function.
4169         (pass_graphite): Convert from a global struct to a subclass of
4170         gimple_opt_pass along with...
4171         (pass_data_graphite): ...new pass_data instance and...
4172         (make_pass_graphite): ...new function.
4173         (pass_graphite_transforms): Convert from a global struct to a subclass
4174         of gimple_opt_pass along with...
4175         (pass_data_graphite_transforms): ...new pass_data instance and...
4176         (make_pass_graphite_transforms): ...new function.
4177         (pass_check_data_deps): Convert from a global struct to a subclass of
4178         gimple_opt_pass along with...
4179         (pass_data_check_data_deps): ...new pass_data instance and...
4180         (make_pass_check_data_deps): ...new function.
4181         (pass_iv_canon): Convert from a global struct to a subclass of
4182         gimple_opt_pass along with...
4183         (pass_data_iv_canon): ...new pass_data instance and...
4184         (make_pass_iv_canon): ...new function.
4185         (pass_scev_cprop): Convert from a global struct to a subclass of
4186         gimple_opt_pass along with...
4187         (pass_data_scev_cprop): ...new pass_data instance and...
4188         (make_pass_scev_cprop): ...new function.
4189         (pass_record_bounds): Convert from a global struct to a subclass of
4190         gimple_opt_pass along with...
4191         (pass_data_record_bounds): ...new pass_data instance and...
4192         (make_pass_record_bounds): ...new function.
4193         (pass_complete_unroll): Convert from a global struct to a subclass of
4194         gimple_opt_pass along with...
4195         (pass_data_complete_unroll): ...new pass_data instance and...
4196         (make_pass_complete_unroll): ...new function.
4197         (pass_complete_unrolli): Convert from a global struct to a subclass of
4198         gimple_opt_pass along with...
4199         (pass_data_complete_unrolli): ...new pass_data instance and...
4200         (make_pass_complete_unrolli): ...new function.
4201         (pass_parallelize_loops): Convert from a global struct to a subclass
4202         of gimple_opt_pass along with...
4203         (pass_data_parallelize_loops): ...new pass_data instance and...
4204         (make_pass_parallelize_loops): ...new function.
4205         (pass_loop_prefetch): Convert from a global struct to a subclass of
4206         gimple_opt_pass along with...
4207         (pass_data_loop_prefetch): ...new pass_data instance and...
4208         (make_pass_loop_prefetch): ...new function.
4209         (pass_iv_optimize): Convert from a global struct to a subclass of
4210         gimple_opt_pass along with...
4211         (pass_data_iv_optimize): ...new pass_data instance and...
4212         (make_pass_iv_optimize): ...new function.
4213         (pass_tree_loop_done): Convert from a global struct to a subclass of
4214         gimple_opt_pass along with...
4215         (pass_data_tree_loop_done): ...new pass_data instance and...
4216         (make_pass_tree_loop_done): ...new function.
4217         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
4218         struct to a subclass of gimple_opt_pass along with...
4219         (pass_data_cse_reciprocals): ...new pass_data instance and...
4220         (make_pass_cse_reciprocals): ...new function.
4221         (pass_cse_sincos): Convert from a global struct to a subclass of
4222         gimple_opt_pass along with...
4223         (pass_data_cse_sincos): ...new pass_data instance and...
4224         (make_pass_cse_sincos): ...new function.
4225         (pass_optimize_bswap): Convert from a global struct to a subclass of
4226         gimple_opt_pass along with...
4227         (pass_data_optimize_bswap): ...new pass_data instance and...
4228         (make_pass_optimize_bswap): ...new function.
4229         (pass_optimize_widening_mul): Convert from a global struct to a
4230         subclass of gimple_opt_pass along with...
4231         (pass_data_optimize_widening_mul): ...new pass_data instance and...
4232         (make_pass_optimize_widening_mul): ...new function.
4233         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
4234         subclass of gimple_opt_pass along with...
4235         (pass_data_phiopt): ...new pass_data instance and...
4236         (make_pass_phiopt): ...new function.
4237         (pass_cselim): Convert from a global struct to a subclass of
4238         gimple_opt_pass along with...
4239         (pass_data_cselim): ...new pass_data instance and...
4240         (make_pass_cselim): ...new function.
4241         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
4242         subclass of gimple_opt_pass along with...
4243         (pass_data_phiprop): ...new pass_data instance and...
4244         (make_pass_phiprop): ...new function.
4245         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
4246         subclass of gimple_opt_pass along with...
4247         (pass_data_pre): ...new pass_data instance and...
4248         (make_pass_pre): ...new function.
4249         (pass_fre): Convert from a global struct to a subclass of
4250         gimple_opt_pass along with...
4251         (pass_data_fre): ...new pass_data instance and...
4252         (make_pass_fre): ...new function.
4253         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
4254         subclass of gimple_opt_pass along with...
4255         (pass_data_reassoc): ...new pass_data instance and...
4256         (make_pass_reassoc): ...new function.
4257         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
4258         subclass of gimple_opt_pass along with...
4259         (pass_data_sink_code): ...new pass_data instance and...
4260         (make_pass_sink_code): ...new function.
4261         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
4262         subclass of gimple_opt_pass along with...
4263         (pass_data_strlen): ...new pass_data instance and...
4264         (make_pass_strlen): ...new function.
4265         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
4266         struct to a subclass of gimple_opt_pass along with...
4267         (pass_data_build_alias): ...new pass_data instance and...
4268         (make_pass_build_alias): ...new function.
4269         (pass_build_ealias): Convert from a global struct to a subclass of
4270         gimple_opt_pass along with...
4271         (pass_data_build_ealias): ...new pass_data instance and...
4272         (make_pass_build_ealias): ...new function.
4273         (pass_ipa_pta): Convert from a global struct to a subclass of
4274         simple_ipa_opt_pass along with...
4275         (pass_data_ipa_pta): ...new pass_data instance and...
4276         (make_pass_ipa_pta): ...new function.
4277         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
4278         subclass of gimple_opt_pass along with...
4279         (pass_data_uncprop): ...new pass_data instance and...
4280         (make_pass_uncprop): ...new function.
4281         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
4282         global struct to a subclass of gimple_opt_pass along with...
4283         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
4284         (make_pass_late_warn_uninitialized): ...new function.
4285         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
4286         to a subclass of gimple_opt_pass along with...
4287         (pass_data_init_datastructures): ...new pass_data instance and...
4288         (make_pass_init_datastructures): ...new function.
4289         (pass_early_warn_uninitialized): Convert from a global struct to a
4290         subclass of gimple_opt_pass along with...
4291         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
4292         (make_pass_early_warn_uninitialized): ...new function.
4293         (pass_update_address_taken): Convert from a global struct to a
4294         subclass of gimple_opt_pass along with...
4295         (pass_data_update_address_taken): ...new pass_data instance and...
4296         (make_pass_update_address_taken): ...new function.
4297         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
4298         struct to a subclass of gimple_opt_pass along with...
4299         (pass_data_release_ssa_names): ...new pass_data instance and...
4300         (make_pass_release_ssa_names): ...new function.
4301         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
4302         subclass of gimple_opt_pass along with...
4303         (pass_data_stdarg): ...new pass_data instance and...
4304         (make_pass_stdarg): ...new function.
4305         * tree-switch-conversion.c (pass_convert_switch): Convert from a
4306         global struct to a subclass of gimple_opt_pass along with...
4307         (pass_data_convert_switch): ...new pass_data instance and...
4308         (make_pass_convert_switch): ...new function.
4309         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
4310         to a subclass of gimple_opt_pass along with...
4311         (pass_data_tail_recursion): ...new pass_data instance and...
4312         (make_pass_tail_recursion): ...new function.
4313         (pass_tail_calls): Convert from a global struct to a subclass of
4314         gimple_opt_pass along with...
4315         (pass_data_tail_calls): ...new pass_data instance and...
4316         (make_pass_tail_calls): ...new function.
4317         * tree-vect-generic.c (pass_lower_vector): Convert from a global
4318         struct to a subclass of gimple_opt_pass along with...
4319         (pass_data_lower_vector): ...new pass_data instance and...
4320         (make_pass_lower_vector): ...new function.
4321         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
4322         gimple_opt_pass along with...
4323         (pass_data_lower_vector_ssa): ...new pass_data instance and...
4324         (make_pass_lower_vector_ssa): ...new function.
4325         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
4326         to a subclass of gimple_opt_pass along with...
4327         (pass_data_slp_vectorize): ...new pass_data instance and...
4328         (make_pass_slp_vectorize): ...new function.
4329         (pass_ipa_increase_alignment): Convert from a global struct to a
4330         subclass of simple_ipa_opt_pass along with...
4331         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
4332         (make_pass_ipa_increase_alignment): ...new function.
4333         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
4334         gimple_opt_pass along with...
4335         (pass_data_vrp): ...new pass_data instance and...
4336         (make_pass_vrp): ...new function.
4337         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
4338         subclass of simple_ipa_opt_pass along with...
4339         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
4340         (make_pass_ipa_free_lang_data): ...new function.
4341         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
4342         gimple_opt_pass along with...
4343         (pass_data_tsan): ...new pass_data instance and...
4344         (make_pass_tsan): ...new function.
4345         (pass_tsan_O0): Convert from a global struct to a subclass of
4346         gimple_opt_pass along with...
4347         (pass_data_tsan_O0): ...new pass_data instance and...
4348         (make_pass_tsan_O0): ...new function.
4349         * var-tracking.c (pass_variable_tracking): Convert from a global
4350         struct to a subclass of rtl_opt_pass along with...
4351         (pass_data_variable_tracking): ...new pass_data instance and...
4352         (make_pass_variable_tracking): ...new function.
4353         * web.c (pass_web): Convert from a global struct to a subclass of
4354         rtl_opt_pass along with...
4355         (pass_data_web): ...new pass_data instance and...
4356         (make_pass_web): ...new function.
4357         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
4358         declaration with that of...
4359         (make_pass_mode_switch_use): ...new function.
4360         (pass_resolve_sw_modes): Replace declaration with that of...
4361         (make_pass_resolve_sw_modes): ...new function.
4362         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
4363         from a global struct to a subclass of rtl_opt_pass along with...
4364         (pass_data_mode_switch_use): ...new pass_data instance and...
4365         (make_pass_mode_switch_use): ...new function.
4366         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
4367         from a global struct to a subclass of rtl_opt_pass along with...
4368         (pass_data_resolve_sw_modes): ...new pass_data instance and...
4369         (make_pass_resolve_sw_modes): ...new function.
4370         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
4371         struct to a subclass of rtl_opt_pass along with...
4372         (pass_data_insert_vzeroupper): ...new pass_data instance and...
4373         (make_pass_insert_vzeroupper): ...new function.
4374         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
4375         global struct to a subclass of rtl_opt_pass along with...
4376         (pass_data_work_around_errata): ...new pass_data instance and...
4377         (make_pass_work_around_errata): ...new function.
4378         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
4379         struct to a subclass of rtl_opt_pass along with...
4380         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
4381         (make_pass_mips_machine_reorg2): ...new function.
4383 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4385         * passes.c (pass_manager::operator new): New.
4387 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4389         Handwritten part of conversion of passes to C++ classes.
4391         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
4392         (toplev.o): Add dep on PASS_MANAGER_H.
4393         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
4394         of early local pases to reflect this moving from a global to a
4395         member of gcc::pass_manager.
4396         (cgraph_add_new_function): Likewise.
4397         * lto-cgraph.c (lto_output_node): Update for conversion of
4398         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
4399         * passes.c (opt_pass::clone): New.
4400         (opt_pass::gate): New.
4401         (opt_pass::execute): New.
4402         (opt_pass::opt_pass): New.
4403         (pass_manager::execute_early_local_passes): New.
4404         (pass_manager::execute_pass_mode_switching): new.
4405         (finish_optimization_passes): Convert to...
4406         (pass_manager::finish_optimization_passes): ...this.
4407         (finish_optimization_passes): Update for conversion of passes to
4408         C++ classes.
4409         (register_dump_files_1): Use has_gate since we cannot portably
4410         check a vtable entry against NULL.
4411         (dump_one_pass): Likewise.
4412         (ipa_write_summaries_2): Likewise.
4413         (ipa_write_optimization_summaries_1): Likewise.
4414         (ipa_read_summaries_1): Likewise.
4415         (ipa_read_optimization_summaries_1): Likewise.
4416         (execute_ipa_stmt_fixups): Likewise.
4417         (pass_manager::pass_manager): Rewrite pass-creation, invoking
4418         pass-creation functions rather than wiring up globals, and
4419         storing the results in fields of pass_manager generated using
4420         pass-instances.def.
4421         (pass_manager::dump_profile_report): Update for conversion of
4422         passes to C++ classes.
4423         (pass_manager::execute_ipa_summary_passes): Likewise.
4424         (execute_one_ipa_transform_pass): Likewise.
4425         (execute_one_pass): Use has_gate and has_execute since we cannot
4426         portably check a vtable entry against NULL.
4427         * pass_manager.h (pass_manager::finish_optimization_passes): New.
4428         (pass_manager): Use pass-instances.def to add fields for the
4429         various pass instances.
4430         * toplev.c (finalize): Update for move of
4431         finish_optimization_passes to a method of gcc::pass_manager.
4432         * toplev.h (finish_optimization_passes): Move to method of class
4433         pass_manager.
4434         * tree-pass.h (struct pass_data): New.
4435         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
4436         (opt_pass::gate): Convert to virtual function.
4437         (opt_pass::~opt_pass): New.
4438         (opt_pass::clone): New.
4439         (opt_pass::execute): Convert to virtual function.
4440         (opt_pass::opt_pass): New.
4441         (opt_pass::ctxt_): new.
4442         (gimple_opt_pass): Convert to subclass of opt_pass.
4443         (gimple_opt_pass::gimple_opt_pass): New.
4444         (rtl_opt_pass): Convert to subclass of opt_pass.
4445         (rtl_opt_pass::rtl_opt_pass): New.
4446         (ipa_opt_pass_d): Convert to subclass of opt_pass.
4447         (ipa_opt_pass_d::ipa_opt_pass_d): New.
4448         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
4449         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
4450         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
4451         invocation of pass_mode_switching to reflect this moving from a
4452         global to a member of gcc::pass_manager.
4453         (ix86_option_override): Rework how pass_insert_vzeroupper is
4454         added to the pass_manager to reflect autogenerated changes.
4455         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
4457 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
4459         PR rtl-optimization/57708
4460         * recog.c (peep2_find_free_register): Validate all regs in a
4461         multi-reg mode.
4463 2013-08-05  Jan Hubicka  <jh@suse.cz>
4465         PR lto/57602
4466         * cgraph.c (verify_cgraph_node): Accept local flags from other
4467         partitions.
4468         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
4469         (function_and_variable_visibility): Likewise.
4470         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
4472 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4474         * graph.c (init_graph_slim_pretty_print): Remove.
4475         (print_graph_cfg): Do not call it.  Use local pretty printer.
4476         (start_graph_dump): Likewise.
4478 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4480         * gimple-pretty-print.c (buffer): Remove.
4481         (initialized): Likewise.
4482         (maybe_init_pretty_print): Likewise.
4483         (print_gimple_stmt): Do not call it.  Use non-static local
4484         pretty_printer variable.
4485         (print_gimple_expr): Likewise.
4486         (print_gimple_seq): Likewise.
4487         (gimple_dump_bb): Likewise.
4489 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4491         * asan.c (asan_pp): Remove.
4492         (asan_pp_initialized): Likewise.
4493         (asan_pp_initialize): Likewise.
4494         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
4495         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
4496         (asan_add_global): Likewise.
4498 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4500         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
4501         * diagnostic.h (diagnostic_flush_buffer): Adjust.
4502         * pretty-print.c (pp_formatted_text_data): Likewise.
4503         (pp_indent): Rename from pp_base_indent.
4504         (pp_format): Rename from pp_base_format.
4505         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
4506         (pp_format_verbatim): Rename from pp_base_format_verbatim.
4507         (pp_flush): Rename from pp_base_flush.
4508         (pp_set_line_maximum_length): Rename from
4509         pp_base_set_line_maximum_length.
4510         (pp_clear_output_area): Rename from pp_base_clear_output_area.
4511         (pp_set_prefix): Rename from pp_base_set_prefix.
4512         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
4513         (pp_emit_prefix): Rename from pp_base_emit_prefix.
4514         (pp_append_text): Rename from pp_base_append_text.
4515         (pp_formatted_text): Rename from pp_base_formatted_text.
4516         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
4517         (pp_remaining_character_count_for_line): Rename from
4518         pp_base_remaining_character_count_for_line.
4519         (pp_newline): Rename from pp_base_newline.
4520         (pp_character): Rename from pp_base_character.
4521         (pp_string): Rename from pp_base_string.
4522         (pp_maybe_space): Rename from pp_base_maybe_space.
4523         * asan.c (asan_pp_string): Adjust.
4524         (asan_emit_stack_protection): Likewise.
4525         (asan_add_global): Likewise.
4526         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
4527         * tree-mudflap.c (mf_varname_tree): Likewise.
4528         * tree-pretty-print.c (pp_tree_identifier): Rename from
4529         pp_base_tree_identifier.
4530         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
4531         Declare as function.
4533 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4535         * pretty-print.h (pp_bar_bar): New.
4536         (pp_ampersand_ampersand): Likewise.
4537         (pp_less_equal): Likewise.
4538         (pp_greater_equal): Likewise.
4539         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
4540         printer functions instead of pp_string or operators and punctuators.
4541         (dump_gimple_call): Likewise.
4542         (dump_gimple_omp_for): Likewise.
4543         (dump_gimple_transaction): Likewise.
4544         (dump_gimple_phi): Likewise.
4545         (pp_gimple_stmt_1): Likewise.
4546         * sched-vis.c (print_insn): Likewise.
4547         * tree-mudflap.c (mf_varname_tree): Likewise.
4548         * tree-pretty-print.c (dump_block_node): Likewise.
4549         (dump_generic_node): Likewise.
4551 2013-08-02  Jan Hubicka  <jh@suse.cz>
4553         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
4554         boundaries.
4555         * lto-streamer-out.c (tree_is_indexable): Results decls and
4556         parm decls are not indexable.
4557         (DFS_write_tree_body): Do not follow args and results.
4558         (hash_tree): Likewise.
4559         (output_functions): Rearrange so struct function is needed
4560         only when real body is output; be able to also ouptut abstract
4561         functions; output DECL_ARGUMENTS and DECL_RESULT.
4562         (lto_output): When not in WPA, ale store abstract functions.
4563         (write_symbol): Do not care about RESULT_DECL.
4564         (output_symbol_p): Handle correctly sbtract decls.
4565         * lto-streamer-in.c (input_function): Rearrange so struct
4566         function can be NULL at entry; allow streaming of
4567         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
4568         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
4569         sanity check during LTO.
4570         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
4571         RESULT_DECl and DECL_ARGUMENTS.
4572         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
4573         Likewise.
4575 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4577         * pretty-print.h (pp_underscore): New.
4578         (pp_comma): Tidy.
4579         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
4580         printer functions instead of pp_character.
4581         (dump_binary_rhs): Likewise.
4582         (dump_ternary_rhs): Likewise.
4583         (dump_gimple_call_args): Likewise.
4584         (pp_points_to_solution): Likewise.
4585         (dump_gimple_call): Likewise.
4586         (dump_gimple_switch): Likewise.
4587         (dump_gimple_cond): Likewise.
4588         (dump_gimple_bind): Likewise.
4589         (dump_gimple_try): Likewise.
4590         (dump_gimple_omp_for): Likewise.
4591         (dump_gimple_omp_continue): Likewise.
4592         (dump_gimple_omp_single): Likewise.
4593         (dump_gimple_omp_sections): Likewise.
4594         (dump_gimple_omp_block): Likewise.
4595         (dump_gimple_omp_critical): Likewise.
4596         (dump_gimple_transaction): Likewise.
4597         (dump_gimple_asm): Likewise.
4598         (dump_gimple_phi): Likewise.
4599         (dump_gimple_omp_parallel): Likewise.
4600         (dump_gimple_omp_task): Likewise.
4601         (dump_gimple_omp_atomic_load): Likewise.
4602         (dump_gimple_omp_atomic_store): Likewise.
4603         (dump_gimple_mem_ops): Likewise.
4604         (pp_gimple_stmt_1): Likewise.
4605         (pp_cfg_jump): Likewise.
4606         (dump_implicit_edges): Likewise.
4607         (gimple_dump_bb_for_graph): Likewise.
4608         * graph.c (draw_cfg_node): Likewise.
4609         * langhooks.c (lhd_print_error_function): Likewise.
4610         * sched-vis.c (print_exp): Likewise.
4611         (print_value): Likewise.
4612         (print_pattern): Likewise.
4613         (print_insn): Likewise.
4614         (rtl_dump_bb_for_graph): Likewise.
4615         * tree-pretty-print.c (dump_function_declaration): Likewise.
4616         (dump_array_domain): Likewise.
4617         (dump_omp_clause): Likewise.
4618         (dump_location): Likewise.
4619         (dump_generic_node): Likewise.
4620         (print_struct_decl): Likewise.
4621         * diagnostic.c (diagnostic_show_locus): Use pp_space.
4623 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4625         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
4626         candidate table when replacing a candidate statement.
4627         (replace_rhs_if_not_dup): Likewise.
4628         (replace_one_candidate): Likewise.
4630 2013-08-02  Jan Hubicka  <jh@suse.cz>
4631             Martin Liska  <marxin.liska@gmail.com>
4633         * cgraphunit.c (add_new_function): Fix logic when adding from
4634         late IPA pass.
4635         (assemble_thunk): Rename to ...
4636         (expand_thunk); .. this one; export; get it working with
4637         general functions; make produced gimple valid.
4638         * cgraph.h (expand_thunk): Declare.
4640 2013-08-02  Jan Hubicka  <jh@suse.cz>
4642         * ipa-cp.c (gather_context_independent_values): Use
4643         ipa_get_param_move_cost.
4644         (get_replacement_map): Remove PARAM; move parameter folding
4645         into tree-inline.c
4646         (create_specialized_node): Update.
4647         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
4648         assert that we have gimple body; update move_cost.
4649         (count_formal_params): Assert that we have gimple body.
4650         (ipa_dump_param): New function.
4651         (ipa_alloc_node_params): Break out from ...
4652         (ipa_initialize_node_params): ... here.
4653         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
4654         (ipa_write_node_info): Stream move costs.
4655         (ipa_read_node_info): Read move costs.
4656         (ipa_update_after_lto_read): Do not recompute node params.
4657         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
4658         (ipa_get_param): Check we are not in WPA.
4659         (ipa_get_param_move_cost): New.
4660         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
4661         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
4662         parm numbers to be present.
4664 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
4666         PR rtl-optimization/58048
4667         * lra-constraints.c (process_alt_operands): Don't check asm
4668         operand on register.
4670 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
4672         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
4673         the implied StoreLoad barrier for atomic operations if before.
4675 2013-08-02  Jan Hubicka  <jh@suse.cz>
4676             Martin Liska  <marxin.liska@gmail.com>
4678         * cgraph.c (cgraph_function_body_availability): Do not check
4679         cgraph flags.
4680         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
4681         symtab_node_availability): Declare.
4682         * ipa.c (can_replace_by_local_alias): New.
4683         (function_and_variable_visibility): Use it.
4684         * symtab.c (symtab_for_node_and_aliases,
4685         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
4687 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
4689         PR rtl-optimization/57963
4690         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
4691         (lra_constraints): Use them.
4693 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
4695         * config/arm/types.md (define_attr "type"): Add "load_acq"
4696         and "store_rel".
4697         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
4698         changes.
4699         (cortex_a53_store1): Likewise.
4701 2013-08-01  Jan Hubicka  <jh@suse.cz>
4703         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
4704         partitions are not needed.
4706 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
4708         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
4709         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
4710         MAYBE_NON_Q_CLASS_P where appropriate.
4712 2013-08-01  Jan Hubicka  <jh@suse.cz>
4714         * cgraph.h (release_function_body): Declare.
4715         * tree.c (free_lang_data_in_decl): Free, parameters and return values
4716         of unused delcarations.
4718 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4720         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
4721         RTL form when subtracting a constant.
4723 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4725         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
4726         Generate canonical plus rtx with negated immediate instead of minus
4727         where appropriate.
4728         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
4730 2013-08-01  Jan Hubicka  <jh@suse.cz>
4732         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
4733         (cgraph_release_function_body): Likewise.
4734         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
4735         * cgraph.h (cgrpah_node): Rename abstract_and_needed
4736         to used_as_abstract_origin.
4737         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
4738         symbols used as abstract origins.
4739         * cgraphunit.c (analyze_functions): Update.
4740         * ipa.c (symtab_remove_unreachable_nodes): Recompute
4741         used_as_abstract_origin.
4742         * tree-inline.c (tree_function_versioning): Update
4743         used_as_abstract_origin; be ready for DECL_RESULT and
4744         DECL_ARGUMENTS to be NULL.
4746         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
4747         for abstract functions.
4748         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
4749         real symbols.
4751 2013-08-01  Jan Hubicka  <jh@suse.cz>
4753         * profile.c (compute_value_histograms): Fix thinko.
4755 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
4757         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
4758         aarch-common-protos.h to extra_headers.
4759         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
4760         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
4761         * config/aarch64/t-aarch64 (aarch-common.o): Define.
4763 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
4765         * config/aarch64/aarch64.md (define_attr "type"): Delete.
4766         Include "../arm/types.md".  Define "type" attribute for all patterns.
4767         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
4768         attribute changes.
4770 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
4772         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
4773         to support power8 load fusion.
4774         (fusion_gpr_mem_load): Likewise.
4776         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
4778         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
4779         declarations for power8 load fusion.
4780         (emit_fusion_gpr_load): Likewise.
4782         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
4783         tuning for power8, turn on fusion mode by default.  Turn on sign
4784         extending fusion mode if normal fusion mode is on, and we are at
4785         -O2 or -O3.
4786         (fusion_gpr_load_p): New function, return true if we can fuse an
4787         addis instruction with a dependent load to a GPR.
4788         (emit_fusion_gpr_load): Emit the instructions for power8 load
4789         fusion to GPRs.
4791         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
4792         (VSX load fusion peepholes): New peepholes to fuse together an
4793         addi instruction with a VSX load instruction.
4795         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
4796         peepholes to fuse an addis instruction with a load to a GPR base
4797         register.  If we are supporting sign extending fusions, convert
4798         sign extending loads to zero extending loads and add an explicit
4799         sign extension.
4801 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
4803         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
4804         aarch-common-protos.h to extra_headers.
4805         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
4806         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
4807         (arm_early_store_addr_dep): Likewise.
4808         (arm_no_early_alu_shift_dep): Likewise.
4809         (arm_no_early_alu_shift_value_dep): Likewise.
4810         (arm_no_early_mul_dep): Likewise.
4811         (arm_no_early_store_addr_dep): Likewise.
4812         (arm_mac_accumulator_is_mul_result): Likewise.
4813         (arm_mac_accumulator_is_result): Likewise.
4814         * config/arm/aarch-common.c: ... here.  New file.
4815         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
4816         here to ...
4817         (arm_early_store_addr_dep): Likewise.
4818         (arm_no_early_alu_shift_dep): Likewise.
4819         (arm_no_early_alu_shift_value_dep): Likewise.
4820         (arm_no_early_mul_dep): Likewise.
4821         (arm_no_early_store_addr_dep): Likewise.
4822         (arm_mac_accumulator_is_mul_result): Likewise.
4823         (arm_mac_accumulator_is_result): Likewise.
4824         * config/arm/aarch-common-protos.h: ... here.  New file.
4825         * config/arm/t-arm (aarch-common.o): Define.
4827 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
4829         * config/arm/arm.md: Include new file "types.md".
4830         (define_attr "type"): Move from here to ...
4831         (define_attr "mul32"): Likewise.
4832         (define_attr "mul64"): Likewise.
4833         * config/arm/types.md: ... here.  New file.
4835 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4837         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
4838         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
4840 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4842         * gen-pass-instances.awk: Fix offset of substr().
4844 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
4846         * Makefile.in (pass-instances.def): New.
4847         (passes.o): Replace dependency on passes.def with one on
4848         pass-instances.def
4850         * gen-pass-instances.awk: New.
4852         * passes.c (pass_manager::pass_manager): Use pass-instances.def
4853         rather than passes.def, updating local definition of NEXT_PASS
4854         macro to add an extra NUM parameter (currently unused).
4856 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
4858         * Makefile.in (PASS_MANAGER_H): New.
4859         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
4860         (passes.o): Likewise.
4861         (statistics.o): Likewise.
4862         (cgraphunit.o): Likewise.
4863         (context.o): Depend on PASS_MANAGER_H.
4865         * pass_manager.h: New.
4867         * cgraphunit.c (cgraph_add_new_function): Update for moves
4868         of globals to fields of pass_manager.
4869         (analyze_function): Likewise.
4870         (expand_function): Likewise.
4871         (ipa_passes): Likewise.
4872         (compile): Likewise.
4874         * context.c (context::context): New.
4875         * context.h  (context::context): New.
4876         (context::get_passes): New.
4877         (context::passes_): New.
4879         * lto-cgraph.c (input_node): Update for moves of globals to
4880         fields of pass_manager.
4882         * passes.c (all_passes): Remove, in favor of a field of the
4883         same name within the new class pass_manager.
4884         (all_small_ipa_passes): Likewise.
4885         (all_lowering_passes): Likewise.
4886         (all_regular_ipa_passes): Likewise.
4887         (all_late_ipa_passes): Likewise.
4888         (all_lto_gen_passes): Likewise.
4889         (passes_by_id): Likewise.
4890         (passes_by_id_size): Likewise.
4891         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4892         the new class pass_manager.
4893         (set_pass_for_id): Convert to...
4894         (pass_manager::set_pass_for_id): ...method.
4895         (get_pass_for_id): Convert to...
4896         (pass_manager::get_pass_for_id): ...method.
4897         (register_one_dump_file): Move body of implementation into...
4898         (pass_manager::register_one_dump_file): ...here.
4899         (register_dump_files_1): Convert to...
4900         (pass_manager::register_dump_files_1): ...method.
4901         (register_dump_files): Convert to...
4902         (pass_manager::register_dump_files): ...method.
4903         (create_pass_tab): Update for moves of globals to fields of
4904         pass_manager.
4905         (dump_passes): Move body of implementation into...
4906         (pass_manager::dump_passes): ...here.
4907         (register_pass): Move body of implementation into...
4908         (pass_manager::register_pass): ...here.
4909         (init_optimization_passes): Convert into...
4910         (pass_manager::pass_manager): ...constructor for new
4911         pass_manager class, and initialize the pass_lists array.
4912         (check_profile_consistency): Update for moves of globals to
4913         fields of pass_manager.
4914         (dump_profile_report): Move body of implementation into...
4915         (pass_manager::dump_profile_report): ...here.
4916         (ipa_write_summaries_1): Update for moves of pass lists from
4917         being globals to fields of pass_manager.
4918         (ipa_write_optimization_summaries): Likewise.
4919         (ipa_read_summaries):  Likewise.
4920         (ipa_read_optimization_summaries): Likewise.
4921         (execute_all_ipa_stmt_fixups): Likewise.
4923         * statistics.c (statistics_fini): Update for moves of globals to
4924         fields of pass_manager.
4926         * toplev.c (general_init): Replace call to
4927         init_optimization_passes with construction of the pass_manager
4928         instance.
4930         * tree-pass.h (all_passes): Remove, in favor of a field of the
4931         same name within the new class pass_manager.
4932         (all_small_ipa_passes): Likewise.
4933         (all_lowering_passes): Likewise.
4934         (all_regular_ipa_passes): Likewise.
4935         (all_lto_gen_passes): Likewise.
4936         (all_late_ipa_passes): Likewise.
4937         (passes_by_id): Likewise.
4938         (passes_by_id_size): Likewise.
4939         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4940         the new class pass_manager.
4941         (get_pass_for_id): Remove.
4943 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4945         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4946         configs.
4948 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
4950         * arm.md (mulhi3): New expand pattern.
4952 2013-07-30  Jan Hubicka  <jh@suse.cz>
4953             Martin Liska  <marxin.liska@gmail.com>
4955         * profile.c (compute_value_histograms): Do not ICE when
4956         there is mismatch only on some counters.
4958 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4960         PR rtl-optimization/57637
4961         * function.c (move_insn_for_shrink_wrap): Also check the
4962         GEN set of the LIVE problem for the liveness analysis
4963         if it exists, otherwise give up.
4965 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4967         PR tree-optimization/57993
4968         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4969         replaced statement in the candidate table.
4970         (phi_add_costs): Return infinite cost when the hidden basis does
4971         not dominate all phis on which the candidate is dependent.
4972         (replace_one_candidate): Record replaced statement in the
4973         candidate table.
4975 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
4977         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4978         (ashlv2si3): New expander.
4979         (*ashlv2si3_i): New define_insn_and_split.
4980         * predicates.md (float_operation): Allow patterns with three
4981         basic sub-patterns.
4983         PR rtl-optimization/58021
4984         * mode-switching.c (create_pre_exit): Always split off preceding
4985         insns if we are not at the basic block head.
4987 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
4989         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4990         (UCLIBC_DYNAMIC_LINKER): New macro.
4991         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4992         `-mnan=2008'.
4993         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4994         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
4995         `-mnan=2008'.
4996         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4997         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4998         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4999         for SF and DF modes.  Use ieee_quad_format for TF mode.
5000         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
5001         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
5002         (mips_option_override): Handle `-mnan=legacy'.
5003         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
5004         `-mabs=2008' and `-mnan=2008'.
5005         (OPTION_DEFAULT_SPECS): Add "nan" default.
5006         (ASM_SPEC): Handle `-mnan='.
5007         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
5008         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
5009         comment accordingly.
5010         (neg<mode>2): Likewise.
5011         * config/mips/mips.opt (mabs, mnan): New options.
5012         * doc/install.texi (Configuration): Document `--with-nan=' option.
5013         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
5014         `-mnan=' options.
5015         (MIPS Options): Document them.
5016         * config.gcc <mips*-*-*>: Handle `--with-nan='.
5017         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
5018         * configure: Regenerate.
5019         * config.in: Regenerate.
5021 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
5023         * config/i386/i386.md (float post-reload splitters): Do not check
5024         for subregs of SSE registers.
5026 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
5027             H.J. Lu  <hongjiu.lu@intel.com>
5029         PR target/57954
5030         PR target/57988
5031         * config/i386/i386.md (post-reload splitter
5032         to avoid partial SSE reg dependency stalls): New pattern.
5034 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
5036         * config/s390/s390.md ("movcc"): Swap load and store instructions.
5038 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
5040         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
5041         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
5043 2013-07-26  Cary Coutant  <ccoutant@google.com>
5045         * dwarf2out.c (die_checksum_ordered): Don't include template
5046         instantiations in signature.
5047         (is_template_parameter): New function.
5048         (is_template_instantiation): New function.
5049         (generate_skeleton_bottom_up): Don't include template instantiations
5050         in type unit DIE.
5051         (generate_skeleton): Likewise.
5052         (break_out_comdat_types): Move recursive call to break out nested
5053         types earlier.
5054         (prune_unused_types_mark_generic_parms_dies): Call
5055         is_template_parameter.
5057 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
5059         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
5060         uses vector registers.
5061         * config/aarch64/iterators.md: Add attributes rtn and vas.
5063 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5064             Richard Earnshaw  <richard.earnshaw@arm.com>
5066         * combine.c (simplify_comparison): Re-canonicalize operands
5067         where appropriate.
5068         * config/arm/arm.md (movcond_addsi): New splitter.
5070 2013-07-25  Sterling Augustine  <saugustine@google.com>
5072         * dwarf2out.c (size_of_pubnames): Move code to...
5073         (include_pubname_in_output): ...here.  New.
5074         (want_pubnames): Rearrange.
5075         (output_pubnames): Call include_pubname_in_output.  Move assertion.
5077 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
5079         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
5081 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
5083         PR target/38836
5084         * doc/extend.texi: Remove obsolete builtins. Fix
5085         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
5087 2013-07-25  Jan Hubicka  <jh@suse.cz>
5089         * cgraph.c (release_function_body): Break out from ...
5090         (cgraph_release_function_body): ... this one; also release DECL_RESULT
5091         and DECL_ARGUMENTS.
5092         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
5093         old_tree in the map.
5094         (create_specialized_node): Update.
5095         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
5096         into index.
5097         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
5098         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
5099         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
5100         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
5101         DECL_RESULT.
5103 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5105         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
5106         addsi3_carryin_alt2_<optab>): Correct output template.
5108 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5110         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
5111         Adjust for arm_restrict_it.
5112         Remove trailing whitespace.
5114 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
5116         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
5117         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
5119         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
5121 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
5123         PR rtl-optimization/57960
5124         * lra-constraints.c (process_alt_operands): Use the right mode
5125         when checking strict_low.
5127 2013-07-25  Jan Hubicka  <jh@suse.cz>
5129         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
5130         * cgraph.c (cgraph_remove_node): Do not release function body
5131         when in cgraph streaming.
5132         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
5133         in other partitions are not considered reachable; fix handling of
5134         clones.
5136 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5138         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
5140 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5142         PR target/19599
5143         PR target/57731
5144         PR target/57837
5145         * config/arm/arm.md ("*sibcall_insn): Replace use of
5146         Ss with US. Adjust output for v5 and v4t.
5147         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
5149         * config/arm/constraints.md ("Ss"): Rename to US.
5151 2013-07-25  Terry Guo  <terry.guo@arm.com>
5153         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
5154         shift_add/shift_sub0/shift_sub1 RTXs.
5156 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
5157             Anton Blanchard  <anton@au1.ibm.com>
5159         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
5160         (altivec_vpks<VI_char>ss): Likewise.
5161         (altivec_vpks<VI_char>us): Likewise.
5162         (altivec_vpku<VI_char>us): Likewise.
5163         (altivec_vpku<VI_char>um): Likewise.
5165 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
5167         Introduce context class.
5169         * Makefile.in (CONTEXT_H): New.
5170         (OBJS): Add context.o.
5171         (toplev.o): Add CONTEXT_H to dependencies.
5172         (context.o): New.
5174         * toplev.c (general_init): Create the singleton gcc::context instance.
5176         * context.c: New.
5178         * context.h: New.
5180 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
5182         PR rtl-optimization/57968
5183         * mode-switching.c (create_pre_exit): Allow instructions that
5184         don't set a return register to need a non-exit mode.
5186 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
5187             Anton Blanchard  <anton@au1.ibm.com>
5189         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
5190         operands to vperm for little endian.
5191         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
5192         of lvsl to create the control mask for a vperm for little endian.
5194 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5195             Anton Blanchard  <anton@au1.ibm.com>
5197         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
5198         two operands for little-endian.
5200 2013-07-23  Steve Ellcey  <sellcey@mips.com>
5202         * config/mips/mips.c (mips_case_values_threshold): New.
5203         (TARGET_CASE_VALUES_THRESHOLD): Define.
5205 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5206             Anton Blanchard  <anton@au1.ibm.com>
5208         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
5209         selection of field for vector splat in little endian mode.
5211 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
5213         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
5214         expanders to rs6000.md.
5215         (ior<mode>3): Likewise.
5216         (and<mode>3): Likewise.
5217         (one_cmpl<mode>2): Likewise.
5218         (nor<mode>3): Likewise.
5219         (andc<mode>3): Likewise.
5220         (eqv<mode>3): Likewise.
5221         (nand<mode>3): Likewise.
5222         (orc<mode>3): Likewise.
5224         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
5225         declaration.
5227         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
5228         to split multi-word logical operations.
5229         (rs6000_split_logical_di): Likewise.
5230         (rs6000_split_logical): Likewise.
5232         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
5233         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
5234         and allow TImode operations in 32-bit.
5235         (vsx_and<mode>3_64bit): Likewise.
5236         (vsx_ior<mode>3_32bit): Likewise.
5237         (vsx_ior<mode>3_64bit): Likewise.
5238         (vsx_xor<mode>3_32bit): Likewise.
5239         (vsx_xor<mode>3_64bit): Likewise.
5240         (vsx_one_cmpl<mode>2_32bit): Likewise.
5241         (vsx_one_cmpl<mode>2_64bit): Likewise.
5242         (vsx_nor<mode>3_32bit): Likewise.
5243         (vsx_nor<mode>3_64bit): Likewise.
5244         (vsx_andc<mode>3_32bit): Likewise.
5245         (vsx_andc<mode>3_64bit): Likewise.
5246         (vsx_eqv<mode>3_32bit): Likewise.
5247         (vsx_eqv<mode>3_64bit): Likewise.
5248         (vsx_nand<mode>3_32bit): Likewise.
5249         (vsx_nand<mode>3_64bit): Likewise.
5250         (vsx_orc<mode>3_32bit): Likewise.
5251         (vsx_orc<mode>3_64bit): Likewise.
5253         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
5254         logical types in GPRs.
5256         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
5257         logical insns to rs6000.md, and allow TImode operations in
5258         32-bit.
5259         (altivec_ior<mode>3): Likewise.
5260         (altivec_xor<mode>3): Likewise.
5261         (altivec_one_cmpl<mode>2): Likewise.
5262         (altivec_nor<mode>3): Likewise.
5263         (altivec_andc<mode>3): Likewise.
5265         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
5266         attributes for moving the 128-bit logical operations into
5267         rs6000.md.
5268         (BOOL_REGS_OUTPUT): Likewise.
5269         (BOOL_REGS_OP1): Likewise.
5270         (BOOL_REGS_OP2): Likewise.
5271         (BOOL_REGS_UNARY): Likewise.
5272         (BOOL_REGS_AND_CR0): Likewise.
5273         (one_cmpl<mode>2): Add support for DI logical operations on
5274         32-bit, splitting the operations to 32-bit.
5275         (anddi3): Likewise.
5276         (iordi3): Likewise.
5277         (xordi3): Likewise.
5278         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
5279         changes to combine the 32/64-bit code, allow logical operations on
5280         TI mode in 32-bit, and to use similar match_operator patterns like
5281         scalar mode uses.  Combine the Altivec and VSX code for logical
5282         operations, and move it here.
5283         (ior<mode>3, 128-bit types): Likewise.
5284         (xor<mode>3, 128-bit types): Likewise.
5285         (one_cmpl<mode>3, 128-bit types): Likewise.
5286         (nor<mode>3, 128-bit types): Likewise.
5287         (andc<mode>3, 128-bit types): Likewise.
5288         (eqv<mode>3, 128-bit types): Likewise.
5289         (nand<mode>3, 128-bit types): Likewise.
5290         (orc<mode>3, 128-bit types): Likewise.
5291         (and<mode>3_internal): Likewise.
5292         (bool<mode>3_internal): Likewise.
5293         (boolc<mode>3_internal1): Likewise.
5294         (boolc<mode>3_internal2): Likewise.
5295         (boolcc<mode>3_internal1): Likewise.
5296         (boolcc<mode>3_internal2): Likewise.
5297         (eqv<mode>3_internal1): Likewise.
5298         (eqv<mode>3_internal2): Likewise.
5299         (one_cmpl1<mode>3_internal): Likewise.
5301 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
5303         * config/microblaze/microblaze.c (microblaze_expand_prologue):
5304         Rename flag_stack_usage to flag_stack_usage_info.
5306 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
5308         * config/microblaze/sync.md: New file.
5309         * config/microblaze/microblaze.md: Include sync.md
5310         * config/microblaze/microblaze.c: Add print_operand 'y'.
5311         * config/microblaze/constraints.md: Add memory_contraint
5312         'Q' which is a single register.
5314 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
5316         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
5318 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
5320         * reload.c (find_reloads): Exit loop once we find this operand
5321         cannot be reloaded somehow for this alternative.
5323         * reload.c (find_reloads): Exit loop once we find a hard register.
5325         * rtlanal.c (computed_jump_p): Exit loop once we find label
5326         reference is used.
5328         * i386.c (ix86_pad_returns): Exit loop after setting replace.
5330         * cfgloopmanip.c (remove_path): Exit loop after setting
5331         irred_invalidated.
5333         * gensupport.c (subst_dup): Avoid loop if code is not
5334         MATCH_DUP nor MATCH_OP_DUP.
5336 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
5338         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
5340 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
5342         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
5343         true for SP_REGNUM if mode == ptr_mode.
5344         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
5345         with value R0_REGNUM + 31.
5347 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
5349         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
5350         pad pointer-typed argument downward.
5352 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
5354         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
5355         and __ILP32__ when the ILP32 model is in use.
5357 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
5359         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
5360         (aarch64_load_symref_appropriately): In the case of
5361         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
5362         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
5363         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
5364         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
5365         if 'mode' doesn't equal to 'ptr_mode'.
5366         (aarch64_output_mi_thunk): Add an assertion on the alignment of
5367         'vcall_offset'; change to call aarch64_emit_move differently depending
5368         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
5369         to calculate the upper bound of 'vcall_offset'.
5370         (aarch64_cannot_force_const_mem): Change to also return true if
5371         mode != ptr_mode.
5372         (aarch64_legitimize_reload_address): In the case of large
5373         displacements, add new local variable 'xmode' and an assertion
5374         based on it; change to use 'xmode' to generate the new rtx and
5375         reload.
5376         (aarch64_asm_trampoline_template): Change to generate the template
5377         differently depending on TARGET_ILP32 or not; change to use
5378         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
5379         (aarch64_trampoline_size): Removed.
5380         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
5381         and replace immediate literals with it.  Change to use 'ptr_mode'
5382         instead of 'DImode' and call convert_memory_address if the mode
5383         of 'fnaddr' doesn't equal to 'ptr_mode'.
5384         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
5385         to output symbol.
5386         (aarch64_elf_asm_destructor): Likewise.
5387         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
5388         on TARGET_ILP32 instead of aarch64_trampoline_size.
5389         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
5390         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
5391         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
5392         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
5393         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
5394         (storewb_pair<GPI:mode>_<P:mode>): ... this.
5395         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
5396         depending on the value of 'mode'.
5397         (add_losym_<mode>): New.
5398         (ldr_got_small_<mode>): New, based on ldr_got_small.
5399         (ldr_got_small): Remove.
5400         (ldr_got_small_sidi): New.
5401         * config/aarch64/iterators.md (P): New.
5402         (PTR): Change to 'ptr_mode' in the condition.
5404 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
5406         * config.gcc (aarch64*-*-*): Support --with-abi.
5407         (aarch64*-*-elf): Support --with-multilib-list.
5408         (aarch64*-*-linux*): Likewise.
5409         (supported_defaults): Add abi to aarch64*-*-*.
5410         * configure.ac: Mention AArch64 for --with-multilib-list.
5411         * configure: Re-generated.
5412         * config/aarch64/biarchilp32.h: New file.
5413         * config/aarch64/biarchlp64.h: New file.
5414         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
5415         (ABI_SPEC): Ditto.
5416         (MULTILIB_DEFAULTS): Ditto.
5417         (DRIVER_SELF_SPECS): Ditto.
5418         (ASM_SPEC): Update to also substitute -mabi.
5419         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
5420         file whose name depends on -mabi= and -mbig-endian.
5421         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
5422         TARGET_ILP32.
5423         (POINTER_SIZE): New define.
5424         (POINTERS_EXTEND_UNSIGNED): Ditto.
5425         (enum aarch64_abi_type): New enumeration tag.
5426         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
5427         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
5428         (TARGET_ILP32): New define.
5429         * config/aarch64/aarch64.opt (mabi): New.
5430         (aarch64_abi): New.
5431         (ilp32, lp64): New values for -mabi.
5432         * config/aarch64/t-aarch64 (comma): New define.
5433         (MULTILIB_OPTIONS): Ditto.
5434         (MULTILIB_DIRNAMES): Ditto.
5435         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
5436         * doc/invoke.texi: Document -mabi for AArch64.
5438 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
5440         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
5442 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
5443             Anton Blanchard  <anton@au1.ibm.com>
5445         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
5446         endianness when selecting field to splat.
5448 2013-07-22  Eric Christopher  <echristo@gmail.com>
5450         * dwarf2out.c (die_odr_checksum): New function to use
5451         CHECKSUM_ macros and ULEB128 for DIE tag.
5452         (generate_type_signature): Use.
5454 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
5456         * config.gcc (sparc*-*-*): Accept leon3 processor.
5457         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
5458         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
5459         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
5460         * config/sparc/sparc.opt (enum processor_type): Add leon3.
5461         (mfix-ut699): Adjust comment.
5462         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
5463         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
5464         (CPP_CPU_SPEC): Likewise.
5465         (ASM_CPU_SPEC): Likewise.
5466         * config/sparc/sparc.c (leon3_cost): New constant.
5467         (sparc_option_override): Add leon3 support.
5468         (mem_ref): New function.
5469         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
5470         (sparc_do_work_around_errata): Look into the instruction in the delay
5471         slot and adjust accordingly.  Add fix for the data cache nullify issues
5472         of the UT699.  Change insertion position for the NOP.
5473         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
5474         (leon3_load): New reservation.
5475         (leon_store): Bump latency to 2.
5476         (grfpu): New automaton.
5477         (grfpu_alu): New unit.
5478         (grfpu_ds): Likewise.
5479         (leon_fp_alu): Adjust.
5480         (leon_fp_mult): Delete.
5481         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
5482         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
5483         * config/sparc/sparc.md (cpu): Add leon3.
5484         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
5485         (swapsi): Likewise.
5486         (atomic_test_and_set): Likewise.
5487         (ldstub): Likewise.
5489 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
5491         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
5492         default for R5900 targets.
5493         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
5494         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
5495         * config/mips/mips.c (mips_option_override): Report an error for
5496         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
5497         for -march=r5900 -mhard-float.
5499 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
5501         * df-problems.c (can_move_insns_across): Exit loop once we
5502         find a non-fixed, non-global register.
5504         * ipa-pure-const.c (propagate_nothrow): Exit loop after
5505         setting can_throw.
5507         * omega.c (omega_eliminate_red): Break after setting red_found.
5508         (omega_problem_has_red_equations): Similarly after setting found.
5509         (omega_query_variable): Similarly after setting coupled.
5511 2013-07-22  Marek Polacek  <polacek@redhat.com>
5513         * gimplify.c: Don't include gimple.h twice.
5515 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5517         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
5518         instead of TARGET_THUMB1.
5519         (Pz): New constraint.
5520         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
5521         encodings.
5522         (compare_negsi_si): Likewise.
5523         (compare_addsi2_op0): Likewise.
5524         (compare_addsi2_op1): Likewise.
5525         (addsi3_carryin_<optab>): Likewise.
5526         (addsi3_carryin_alt2_<optab>): Likewise.
5527         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
5528         for arm_restrict_it.
5529         (subsi3_carryin): Likewise.
5530         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
5531         (minmax_arithsi): Disable for arm_restrict_it.
5532         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
5533         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
5534         (satsi_<SAT:code>_shift): Likewise.
5535         (arm_shiftsi3): Add alternative for 16-bit encoding.
5536         (arm32_movhf): Disable for arm_restrict_it.
5537         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
5538         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
5540 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
5542         * config/arm/arm.md (attribute "insn"): Delete.
5543         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
5544         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
5545         (not_shiftsi): Update for attribute change.
5546         (not_shiftsi_compare0): Likewise.
5547         (not_shiftsi_compare0_scratch): Likewise.
5548         (arm_one_cmplsi2): Likewise.
5549         (thumb1_one_cmplsi2): Likewise.
5550         (notsi_compare0): Likewise.
5551         (notsi_compare0_scratch): Likewise.
5552         (thumb1_movdi_insn): Likewise.
5553         (arm_movsi_insn): Likewise.
5554         (movhi_insn_arch4): Likewise.
5555         (movhi_bytes): Likewise.
5556         (arm_movqi_insn): Likewise.
5557         (thumb1_movqi_insn): Likewise.
5558         (arm32_movhf): Likewise.
5559         (thumb1_movhf): Likewise.
5560         (arm_movsf_soft_insn): Likewise.
5561         (thumb1_movsf_insn): Likewise.
5562         (thumb_movdf_insn): Likewise.
5563         (movsicc_insn): Likewise.
5564         (movsfcc_soft_insn): Likewise.
5565         (and_scc): Likewise.
5566         (cond_move): Likewise.
5567         (if_move_not): Likewise.
5568         (if_not_move): Likewise.
5569         (if_shift_move): Likewise.
5570         (if_move_shift): Likewise.
5571         (if_shift_shift): Likewise.
5572         (if_not_arith): Likewise.
5573         (if_arith_not): Likewise.
5574         (cond_move_not): Likewise.
5575         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
5576         (neon_mov<mode>): Likewise.
5577         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
5578         (thumb2_movsi_vfp): Likewise.
5579         (movsf_vfp): Likewise.
5580         (thumb2_movsf_vfp): Likewise.
5581         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5582         change.
5583         (cortexa7_older_only): Likewise.
5584         (cortexa7_younger): Likewise.
5585         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
5586         (1020alu_shift_op): Likewise.
5587         (1020alu_shift_reg_op): Likewise.
5588         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
5589         (alu_shift_op): Likewise.
5590         (alu_shift_reg_op): Likewise.
5591         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
5592         (11_alu_shift_op): Likewise.
5593         (11_alu_shift_reg_op): Likewise.
5594         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
5595         (9_alu_shift_reg_op): Likewise.
5596         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
5597         change.
5598         (cortex_a15_alu_shift): Likewise.
5599         (cortex_a15_alu_shift_reg): Likewise.
5600         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
5601         (cortex_a5_alu_shift): Likewise.
5602         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
5603         change.
5604         (cortex_a53_alu_shift): Likewise.
5605         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
5606         change.
5607         (cortex_a7_alu_reg): Likewise.
5608         (cortex_a7_alu_shift): Likewise.
5609         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
5610         (cortex_a8_alu_shift): Likewise.
5611         (cortex_a8_alu_shift_reg): Likewise.
5612         (cortex_a8_mov): Likewise.
5613         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
5614         (cortex_a9_dp_shift): Likewise.
5615         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
5616         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
5617         (cortex_r4_mov): Likewise.
5618         (cortex_r4_alu_shift): Likewise.
5619         (cortex_r4_alu_shift_reg): Likewise.
5620         * config/arm/fa526.md (526_alu_op): Update for attribute change.
5621         (526_alu_shift_op): Likewise.
5622         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
5623         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
5624         (626te_alu_shift_op): Likewise.
5625         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
5626         (726te_alu_op): Likewise.
5627         (726te_alu_shift_op): Likewise.
5628         (726te_alu_shift_reg_op): Likewise.
5629         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
5630         (mp626_alu_shift_op): Likewise.
5631         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
5632         (pj4_alu_e1_conds): Likewise.
5633         (pj4_alu): Likewise.
5634         (pj4_alu_conds): Likewise.
5635         (pj4_shift): Likewise.
5636         (pj4_shift_conds): Likewise.
5637         (pj4_alu_shift): Likewise.
5638         (pj4_alu_shift_conds): Likewise.
5640 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5642         * config/arm/predicates.md (shiftable_operator_strict_it):
5643         New predicate.
5644         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
5645         Disable cond_exec version for arm_restrict_it.
5646         (thumb2_smaxsi3): Convert to generate cond_exec.
5647         (thumb2_sminsi3): Likewise.
5648         (thumb32_umaxsi3): Likewise.
5649         (thumb2_uminsi3): Likewise.
5650         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
5651         (thumb2_neg_abssi2): Likewise.
5652         (thumb2_mov_scc): Add alternative for 16-bit encoding.
5653         (thumb2_movsicc_insn): Adjust alternatives.
5654         (thumb2_mov_negscc): Disable for arm_restrict_it.
5655         (thumb2_mov_negscc_strict_it): New pattern.
5656         (thumb2_mov_notscc_strict_it): New pattern.
5657         (thumb2_mov_notscc): Disable for arm_restrict_it.
5658         (thumb2_ior_scc): Likewise.
5659         (thumb2_ior_scc_strict_it): New pattern.
5660         (thumb2_cond_move): Adjust for arm_restrict_it.
5661         (thumb2_cond_arith): Disable for arm_restrict_it.
5662         (thumb2_cond_arith_strict_it): New pattern.
5663         (thumb2_cond_sub): Adjust for arm_restrict_it.
5664         (thumb2_movcond): Likewise.
5665         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
5666         (thumb2_zero_extendhisi2_v6): Likewise.
5667         (thumb2_zero_extendqisi2_v6): Likewise.
5668         (orsi_notsi_si): Likewise.
5669         (orsi_not_shiftsi_si): Likewise.
5671 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
5673         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
5674         instruction sequence is 1 byte shorter.
5676 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
5678         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
5679         it is not needed after split.
5681 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
5683         PR target/51784
5684         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
5685         second label for nonlocal goto receivers. Don't output pic base labels
5686         unless we're producing PIC; mark that action unreachable().
5687         (ix86_save_reg): If the function contains a nonlocal label, save the
5688         PIC base reg.
5689         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
5690         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
5691         (update_pic_label_number_if_needed): New.
5692         (machopic_output_function_base_name): Adjust for nonlocal receiver
5693         case.
5694         (machopic_should_output_picbase_label): New.
5695         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
5696         (nonlocal_goto_receiver): New insn and split.
5698 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
5700         * config/aarch64/aarch64-builtins.c
5701         (aarch64_fold_builtin): Fold abs in all modes.
5702         * config/aarch64/aarch64-simd-builtins.def
5703         (abs): Enable for all modes.
5704         * config/aarch64/arm_neon.h
5705         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
5706         (vabs_f64): Add missing intrinsic.
5708 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
5710         * config/aarch64/arm_neon.h (vabs_s64): New function
5712 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
5714         PR target/57516
5715         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
5716         * config/avr/avr.md (adjust_len): Add `round'.
5717         * config/avr/avr-protos.h (avr_out_round): New prototype.
5718         (avr_out_plus): Add `out_label' argument.
5719         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
5720         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
5721         Handle the case where `insn' is just a pattern.
5722         (avr_out_bitop): Handle the case where `insn' is just a pattern.
5723         (avr_out_round): New function.
5724         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
5726 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
5728         * config/microblaze/microblaze.c (microblaze_expand_prologue):
5729         Add check for flag_stack_usage to handle -fstack-usage support
5731 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
5733         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
5734         interaction for new Power8 flags and VSX.
5736 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
5738         PR middle-end/57698
5739         * tree-inline.c (expand_call_inline): Emit errors during
5740         early_inlining only if optimization is not turned on.
5742 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5744         * passes.def: New.
5746         * passes.c (init_optimization_passes): Move the construction of
5747         the pass hierarchy into a new passes.def file.
5749         * Makefile.in (passes.o): Add dependency on passes.def.
5751 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5753         * passes.c (init_optimization_passes): Introduce macros for
5754         constructing the tree of passes (INSERT_PASSES_AFTER,
5755         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
5756         TERMINATE_PASS_LIST).
5758 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
5759             Wei Mi  <wmi@google.com>
5761         PR rtl-optimization/57878
5762         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
5763         top.
5764         (reload_pseudo_compare_func): Check nregs first for reload
5765         pseudos.
5767 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5769         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
5771 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
5773         * read-rtl.c (validate_const_int): Once an invalid character is
5774         seen, quit the loop.
5776         * gengtype.c (write_roots): Similarly once we find the "deletable"
5777         or "if_marked" option.
5779 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
5781         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
5782         "xtab" and "sat".  Move value "clz" from here to ...
5783         (attriubte "type"): ... here.
5784         (satsi_<SAT:code>): Delete "insn" attribute.
5785         (satsi_<SAT:code>_shift): Likewise.
5786         (arm_zero_extendqisi2addsi): Likewise.
5787         (arm_extendqisi2addsi): Likewise.
5788         (clzsi2): Update for attribute changes.
5789         (rbitsi2): Likewise.
5790         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
5791         attribute.
5792         (arm_usatsihi): Likewise.
5793         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
5795 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
5797         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
5798         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
5799         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
5800         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
5801         in alphabetical order.
5802         (attribute "core_cycles"): Update for attribute changes.
5803         (arm_addsi3): Likewise.
5804         (addsi3_compare0): Likewise.
5805         (addsi3_compare0_scratch): Likewise.
5806         (addsi3_compare_op1): Likewise.
5807         (addsi3_compare_op2): Likewise.
5808         (compare_addsi2_op0): Likewise.
5809         (compare_addsi2_op1): Likewise.
5810         (addsi3_carryin_shift_<optab>): Likewise.
5811         (subsi3_carryin_shift): Likewise.
5812         (rsbsi3_carryin_shift): Likewise.
5813         (arm_subsi3_insn): Likewise.
5814         (subsi3_compare0): Likewise.
5815         (subsi3_compare): Likewise.
5816         (arm_andsi3_insn): Likewise.
5817         (thumb1_andsi3_insn): Likewise.
5818         (andsi3_compare0): Likewise.
5819         (andsi3_compare0_scratch): Likewise.
5820         (zeroextractsi_compare0_scratch
5821         (andsi_not_shiftsi_si): Likewise.
5822         (iorsi3_insn): Likewise.
5823         (iorsi3_compare0): Likewise.
5824         (iorsi3_compare0_scratch): Likewise.
5825         (arm_xorsi3): Likewise.
5826         (thumb1_xorsi3_insn): Likewise.
5827         (xorsi3_compare0): Likewise.
5828         (xorsi3_compare0_scratch): Likewise.
5829         (satsi_<SAT:code>_shift): Likewise.
5830         (rrx): Likewise.
5831         (arm_shiftsi3): Likewise.
5832         (shiftsi3_compare0): Likewise.
5833         (not_shiftsi): Likewise.
5834         (not_shiftsi_compare0): Likewise.
5835         (not_shiftsi_compare0_scratch): Likewise.
5836         (arm_one_cmplsi2): Likewise.
5837         (thumb_one_complsi2): Likewise.
5838         (notsi_compare0): Likewise.
5839         (notsi_compare0_scratch): Likewise.
5840         (thumb1_zero_extendhisi2): Likewise.
5841         (arm_zero_extendhisi2): Likewise.
5842         (arm_zero_extendhisi2_v6): Likewise.
5843         (arm_zero_extendhisi2addsi): Likewise.
5844         (thumb1_zero_extendqisi2): Likewise.
5845         (thumb1_zero_extendqisi2_v6): Likewise.
5846         (arm_zero_extendqisi2): Likewise.
5847         (arm_zero_extendqisi2_v6): Likewise.
5848         (arm_zero_extendqisi2addsi): Likewise.
5849         (thumb1_extendhisi2): Likewise.
5850         (arm_extendhisi2): Likewise.
5851         (arm_extendhisi2_v6): Likewise.
5852         (arm_extendqisi): Likewise.
5853         (arm_extendqisi_v6): Likewise.
5854         (arm_extendqisi2addsi): Likewise.
5855         (thumb1_extendqisi2): Likewise.
5856         (thumb1_movdi_insn): Likewise.
5857         (arm_movsi_insn): Likewise.
5858         (movsi_compare0): Likewise.
5859         (movhi_insn_arch4): Likewise.
5860         (movhi_bytes): Likewise.
5861         (arm_movqi_insn): Likewise.
5862         (thumb1_movqi_insn): Likewise.
5863         (arm32_movhf): Likewise.
5864         (thumb1_movhf): Likewise.
5865         (arm_movsf_soft_insn): Likewise.
5866         (thumb1_movsf_insn): Likewise.
5867         (movdf_soft_insn): Likewise.
5868         (thumb_movdf_insn): Likewise.
5869         (arm_cmpsi_insn): Likewise.
5870         (cmpsi_shiftsi): Likewise.
5871         (cmpsi_shiftsi_swp): Likewise.
5872         (arm_cmpsi_negshiftsi_si): Likewise.
5873         (movsicc_insn): Likewise.
5874         (movsfcc_soft_insn): Likewise.
5875         (arith_shiftsi): Likewise.
5876         (arith_shiftsi_compare0
5877         (arith_shiftsi_compare0_scratch
5878         (sub_shiftsi): Likewise.
5879         (sub_shiftsi_compare0
5880         (sub_shiftsi_compare0_scratch
5881         (and_scc): Likewise.
5882         (cond_move): Likewise.
5883         (if_plus_move): Likewise.
5884         (if_move_plus): Likewise.
5885         (if_move_not): Likewise.
5886         (if_not_move): Likewise.
5887         (if_shift_move): Likewise.
5888         (if_move_shift): Likewise.
5889         (if_shift_shift): Likewise.
5890         (if_not_arith): Likewise.
5891         (if_arith_not): Likewise.
5892         (cond_move_not): Likewise.
5893         (thumb1_ashlsi3): Set type attribute.
5894         (thumb1_ashrsi3): Likewise.
5895         (thumb1_lshrsi3): Likewise.
5896         (thumb1_rotrsi3): Likewise.
5897         (shiftsi3_compare0_scratch): Likewise.
5898         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
5899         (neon_mov<mode>): Likewise.
5900         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
5901         attribute changes.
5902         (thumb2_movsi_insn): Likewise.
5903         (thumb2_cmpsi_neg_shiftsi): Likewise.
5904         (thumb2_extendqisi_v6): Likewise.
5905         (thumb2_zero_extendhisi2_v6): Likewise.
5906         (thumb2_zero_extendqisi2_v6): Likewise.
5907         (thumb2_shiftsi3_short): Likewise.
5908         (thumb2_addsi3_compare0_scratch): Likewise.
5909         (orsi_not_shiftsi_si): Likewise.
5910         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
5911         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
5912         changes.
5913         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
5914         (1020alu_shift_op): Likewise.
5915         (1020alu_shift_reg_op): Likewise.
5916         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
5917         (alu_shift_op): Likewise.
5918         (alu_shift_reg_op): Likewise.
5919         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
5920         (11_alu_shift_op): Likewise.
5921         (11_alu_shift_reg_op): Likewise.
5922         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
5923         (9_alu_shift_reg_op): Likewise.
5924         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
5925         changes.
5926         (cortex_a15_alu_shift): Likewise.
5927         (cortex_a15_alu_shift_reg): Likewise.
5928         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
5929         changes.
5930         (cortex_a5_alu_shift): Likewise.
5931         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
5932         changes.
5933         (cortex_a53_alu_shift): Likewise.
5934         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
5935         changes.
5936         (cortex_a7_alu_reg): Likewise.
5937         (cortex_a7_alu_shift): Likewise.
5938         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
5939         changes.
5940         (cortex_a8_alu_shift): Likewise.
5941         (cortex_a8_alu_shift_reg): Likewise.
5942         (cortex_a8_mov): Likewise.
5943         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5944         (cortex_a9_dp_shift): Likewise.
5945         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5946         changes.
5947         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5948         changes.
5949         (cortex_r4_mov): Likewise.
5950         (cortex_r4_alu_shift): Likewise.
5951         (cortex_r4_alu_shift_reg): Likewise.
5952         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5953         (526_alu_shift_op): Likewise.
5954         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5955         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5956         (626te_alu_shift_op): Likewise.
5957         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5958         (726te_alu_op): Likewise.
5959         (726te_alu_shift_op): Likewise.
5960         (726te_alu_shift_reg_op): Likewise.
5961         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5962         (mp626_alu_shift_op): Likewise.
5963         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5964         (pj4_alu_e1_conds): Likewise.
5965         (pj4_alu): Likewise.
5966         (pj4_alu_conds): Likewise.
5967         (pj4_shift): Likewise.
5968         (pj4_shift_conds): Likewise.
5969         (pj4_alu_shift): Likewise.
5970         (pj4_alu_shift_conds): Likewise.
5971         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5972         changes.
5973         (cortexa7_older_only): Likewise.
5974         (cortexa7_younger): Likewise.
5976 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
5978         * ipa-pure-const.c (generate_summary): Rename to...
5979         (pure_const_generate_summary): ... this.
5981 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
5983         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5985 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
5987         PR target/57909
5988         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5989         usage in HI mode.
5991 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5993         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5994         enabled without -march=zEC12.
5995         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5996         flags to be set.
5998 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
6000         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
6001         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
6002         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
6003         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
6004         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
6005         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
6006         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
6007         ISA_HAS_FP4.
6008         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
6009         and ISA_HAS_NMADD3_NMSUB3.
6010         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
6011         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
6012         (nmsub4<mode>, nmsub3<mode>): Likewise.
6013         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
6015 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
6017         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
6018         TARGET_MIPS5400 checking.
6020 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
6021             Peter Bergner  <bergner@vnet.ibm.com>
6023         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
6024         registers in the comment.
6025         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
6026         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
6027         rather than FIRST_PSEUDO_REGISTERS.
6029 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
6031         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
6032         enable extra ISA flags with TARGET_HTM.
6034 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
6036         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
6037         Fix comment typos.
6039 2013-07-15  Cong Hou  <congh@google.com>
6041         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
6042         in compare function for sorting.
6044 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
6046         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
6047         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
6048         * config/rs6000/rs6000.opt: Add -mhtm option.
6049         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
6050         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
6051         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
6052         __HTM__ if the HTM instructions are available.
6053         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
6054         htm_spr_reg_operand): New define_predicates.
6055         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
6056         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
6057         Include htm.md.
6058         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
6059         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
6060         HTM builtin functions.
6061         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
6062         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
6063         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
6064         (rs6000_builtin_mask_calculate): Likewise.
6065         (rs6000_option_override_internal): Likewise.
6066         (bdesc_htm): Add new HTM builtin support.
6067         (htm_spr_num): New function.
6068         (htm_spr_regno): Likewise.
6069         (rs6000_htm_spr_icode): Likewise.
6070         (htm_expand_builtin): Likewise.
6071         (htm_init_builtins): Likewise.
6072         (rs6000_expand_builtin): Add support for HTM builtin functions.
6073         (rs6000_init_builtins): Likewise.
6074         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
6075         option.
6076         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
6077         (TARGET_HTM, MASK_HTM): Define macros.
6078         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
6079         (FIXED_REGISTERS): Likewise.
6080         (CALL_USED_REGISTERS): Likewise.
6081         (CALL_REALLY_USED_REGISTERS): Likewise.
6082         (REG_ALLOC_ORDER): Likewise.
6083         (enum reg_class): Likewise.
6084         (REG_CLASS_NAMES): Likewise.
6085         (REG_CLASS_CONTENTS): Likewise.
6086         (REGISTER_NAMES): Likewise.
6087         (ADDITIONAL_REGISTER_NAMES): Likewise.
6088         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
6089         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
6090         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
6091         * config/rs6000/htm.md: New file.
6092         * config/rs6000/htmintrin.h: New file.
6093         * config/rs6000/htmxlintrin.h: New file.
6095 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6097         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
6098         Define SYMBOL_TINY_GOT, update comment.
6099         * config/aarch64/aarch64.c
6100         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
6101         (aarch64_expand_mov_immediate): Likewise.
6102         (aarch64_print_operand): Likewise.
6103         (aarch64_classify_symbol): Likewise.
6104         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
6105         (ldr_got_tiny): Define.
6107 2013-07-13  Tobias Grosser  <tobias@grosser.es>
6109         PR tree-optimization/54094
6110         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
6111           scheduling dimension for the parallelism check from the polyhedral
6112           information in the AST.
6113         * graphite-dependences.c (carries_deps): Do not assume the schedule is
6114           in 2D + 1 form.
6116 2013-07-13  Jason Merrill  <jason@redhat.com>
6118         * print-tree.c (debug_vec_tree): Use debug_raw.
6119         (debug_raw (vec<tree, va_gc> &)): New.
6120         (debug_raw (vec<tree, va_gc> *)): New.
6121         * tree.h: Declare them.
6123 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
6125         * ifcvt.c (ifcvt_after_combine): New static variable.
6126         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
6127         for size.
6128         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
6129         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
6130         rest_of_handle_if_after_reload): Pass new argument for if_convert.
6132 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
6134         * config/mips/mips.c (mips_expand_call): Remove empty statement.
6136 2013-07-12  Michael Matz  <matz@suse.de>
6138         PR middle-end/55771
6139         * convert.c (convert_to_real): Reject non-float inner types.
6141 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
6143         * config/aarch64/aarch64-protos.h
6144         (aarch64_simd_immediate_valid_for_move): Remove.
6145         * config/aarch64/aarch64.c (simd_immediate_info): New member.
6146         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
6147         cases.
6148         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
6150 2013-07-11  Steve Ellcey  <sellcey@mips.com>
6152         * config/mips/mips.c (mips_conditional_register_usage): Do not
6153         use t[0-7] registers in MIPS16 mode when optimizing for size.
6155 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
6157         * config/i386/i386.c (dispatch_function_versions): Fix array
6158         indexing of function_version_info to match actual_versions.
6160 2013-07-11  Teresa Johnson  <tejohnson@google.com>
6162         * vec.h (struct va_gc): Move release out-of-line.
6163         (va_gc::release): Call ggc_free on released vec.
6165 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
6167         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
6168         Require GOT register as additional operand in UNSPEC.
6169         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
6170         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
6171         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
6172         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
6173         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
6174         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
6175         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
6177 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
6179         PR target/57631
6180         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
6181         name seen by assembler/linker rather if available.
6183 2013-07-11  Andreas Schwab  <schwab@suse.de>
6185         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
6187 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
6189         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
6191 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
6193         * doc/tm.texi.in: Move hook documentation to ....
6194         * target.def: ... here.
6196         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
6197         text on @hook line.
6198         * doc/tm.texi: Regenerate.
6200 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
6202         PR c++/57869
6203         * doc/invoke.texi: Document -Wconditionally-supported.
6205 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
6207         PR target/57844
6208         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
6209         of my_fp.
6211 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
6213         PR target/57506
6214         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
6215         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
6216         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
6217         Remove duplicate devices.
6218         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
6219         * config/avr/t-multilib: Regenerate.
6220         * config/avr/avr-tables.opt: Regenerate.
6221         * doc/avr-mmcu.texi: Regenerate.
6223 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
6225         PR target/56987
6226         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
6228 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
6230         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
6231         the cost of MULT when optimizing for size.
6233 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6235         * config/cr16/cr16-protos.h: Don't include target.h.
6237 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
6239         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
6240         adjust register size for TDmode and TFmode for VSX registers.
6242 2013-07-08  Kai Tietz  <ktietz@redhat.com>
6244         PR target/56892
6245         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
6246         hook_bool_const_tree_true.
6248 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6250         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
6251         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
6252         * config/s390/s390.md: Define FPR*_REGNUM constants.
6253         Fix FPR2_REGNUM constant (18 -> 17).
6254         ("*trunc<BFP:mode><DFP_ALL:mode>2")
6255         ("*trunc<DFP_ALL:mode><BFP:mode>2")
6256         ("trunc<BFP:mode><DFP_ALL:mode>2")
6257         ("trunc<DFP_ALL:mode><BFP:mode>2")
6258         ("*extend<BFP:mode><DFP_ALL:mode>2")
6259         ("*extend<DFP_ALL:mode><BFP:mode>2")
6260         ("extend<BFP:mode><DFP_ALL:mode>2")
6261         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
6262         FPR4_REGNUM.
6264 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
6266         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
6268 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6270         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
6271         and cfun_fpr_bit_p to cfun_fpr_save_p.
6272         (s390_frame_area, s390_register_info, s390_frame_info)
6273         (s390_emit_prologue, s390_emit_epilogue)
6274         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
6275         register numbers.
6276         * config/s390/s390.h: Define *_REGNUM macros for floating point
6277         register numbers.
6279 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
6281         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
6283 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
6285         PR rtl-optimization/57786
6286         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
6287         and break out of the loop when it is set to false.
6289 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
6291         PR target/57819
6292         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
6293         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
6294         (const_int 63)) 0)).
6295         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
6296         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
6297         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
6299         PR rtl-optimization/57829
6300         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
6301         mask bits outside of mode are just sign-extension from mode to HWI.
6303 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
6305         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
6306         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
6307         adjust_address instead of change_address to keep info about alignment.
6308         (emit_strmov): Remove.
6309         (emit_memmov): New function.
6310         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
6311         (expand_movmem_epilogue): Likewise and return updated rtx for
6312         destination.
6313         (expand_constant_movmem_prologue): Likewise and return updated rtx for
6314         destination and source.
6315         (decide_alignment): Refactor, handle vector_loop.
6316         (ix86_expand_movmem): Likewise.
6317         (ix86_expand_setmem): Likewise.
6318         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
6320 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
6322         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
6323         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
6325 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
6327         * config/i386/sse.md (sse_movlhps): Change alternative 3
6328         of operand 2 to "m".
6330 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
6332         PR target/57807
6333         * config/i386/sse.md (iptr): New mode attribute.
6334         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
6335         (<sse>_vm<plusminus_insn><mode>3): Ditto.
6336         (<sse>_vmmul<mode>3): Ditto.
6337         (<sse>_vmdiv<mode>3): Ditto.
6338         (sse_vmrcpv4sf2): Ditto.
6339         (<sse>_vmsqrt<mode>2): Ditto.
6340         (sse_vmrsqrtv4sf2): Ditto.
6341         (<sse>_vm<code><mode>3): Ditto.
6342         (avx_vmcmp<mode>3): Ditto.
6343         (<sse>_vmmaskcmp<mode>3): Ditto.
6344         (<sse>_comi): Ditto.
6345         (<sse>_ucomi): Ditto.
6346         (*xop_vmfrcz_<mode>): Ditto.
6347         (*fmai_fmadd_<mode>): Ditto.
6348         (*fmai_fmsub_<mode>): Ditto.
6349         (*fmai_fnmadd_<mode>): Ditto.
6350         (*fmai_fnmsub_<mode>): Ditto.
6351         (*fma4i_vmfmadd_<mode>): Ditto.
6352         (*fma4i_vmfmsub_<mode>): Ditto.
6353         (*fma4i_vmfnmadd_<mode>): Ditto.
6354         (*fma4i_vmfnmsub_<mode>): Ditto.
6355         (*xop_vmfrcz_<mode>): Ditto.
6356         (sse_cvtps2pi): Ditto.
6357         (sse_cvttps2pi): Ditto.
6358         (sse_cvtss2si): Ditto.
6359         (sse_cvtss2si_2): Ditto.
6360         (sse_cvtss2siq_2): Ditto.
6361         (sse_cvttss2si): Ditto.
6362         (sse_cvttss2siq): Ditto.
6363         (sse_cvtsd2si): Ditto.
6364         (sse_cvtsd2si_2): Ditto.
6365         (sse_cvtsd2siq_2): Ditto.
6366         (sse_cvttsd2si): Ditto.
6367         (sse_cvttsd2siq): Ditto.
6368         (sse_cvtsd2ss): Ditto.
6369         (sse_cvtss2sd): Ditto.
6370         (avx2_pbroadcast<mode>): Ditto.
6371         (avx2_pbroadcast<mode>_1): Ditto.
6372         (*avx_vperm_broadcast_v4sf): Ditto.
6374         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
6375         (sse_movlhps): Ditto.
6376         (sse_storehps): Ditto.
6377         (sse_loadhps): Ditto.
6378         (sse_storelps): Ditto.
6379         (sse_loadlps): Ditto.
6380         (*vec_concatv4sf): Ditto.
6381         (*vec_interleave_highv2df): Ditto.
6382         (*vec_interleave_lowv2df): Ditto.
6383         (*vec_extractv2df_1_sse): Ditto.
6384         (*vec_extractv2df_0_sse): Ditto.
6385         (sse2_storelpd): Ditto.
6386         (sse2_loadlpd): Ditto.
6387         (sse2_movsd): Ditto.
6388         (*vec_concatv4si): Ditto.
6389         (vec_concatv2di): Ditto.
6391         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
6392         for Intel asm dialect.
6393         (mmx_punpcklwd): Ditto.
6394         (mmx_punpckldq): Ditto.
6396         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
6397         for intel assembler dialect.
6399 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
6401         PR target/29776
6402         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
6403         for BUILT_IN_C{LZ,LRSB}*.
6404         * tree.h (CASE_INT_FN): Add FN##IMAX case.
6405         * tree-vrp.c (extract_range_basic): Handle
6406         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
6407         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
6408         fall thru to code calling set_value*.
6409         * builtins.c (expand_builtin): Remove *IMAX cases.
6410         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
6411         if width is bigger than 2*HWI.
6413 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
6415         PR rtl-optimization/55342
6416         * lra-int.h (lra_subreg_reload_pseudos): New.
6417         * lra.c: Add undoing optional reloads to the block diagram.
6418         (lra_subreg_reload_pseudos): New.
6419         (lra_optional_reload_pseudos): Change comments.
6420         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
6421         lra_optional_reload_pseudos after undo transformations.
6422         * lra-assigns.c (pseudo_prefix_title): New.
6423         (lra_setup_reg_renumber): Use it.
6424         (spill_for): Ditto.  Check subreg reload pseudos too.
6425         (assign_by_spills): Consider subreg reload pseudos too.
6426         * lra-constraints.c (simplify_operand_subreg): Use
6427         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
6428         (curr_insn_transform): Recognize and do optional reloads.
6429         (undo_optional_reloads): New.
6430         (lra_undo_inheritance): Call undo_optional_reloads.
6432 2013-07-05  Thomas Quinot  <quinot@adacore.com>
6434         * tree-complex.c (expand_complex_operations_1): Fix typo.
6436 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
6438         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
6439         (tune_params): New member 'const vec_costs'.
6440         * config/aarch64/aarch64.c (generic_vector_cost): New.
6441         (generic_tunings): New member 'generic_vector_cost'.
6442         (aarch64_builtin_vectorization_cost): New.
6443         (aarch64_add_stmt_cost): New.
6444         (TARGET_VECTORIZE_ADD_STMT_COST): New.
6445         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
6447 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
6449         PR target/57777
6450         * config/i386/predicates.md (vsib_address_operand): Disallow
6451         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
6453 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
6455         PR middle-end/55030
6456         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
6457         expand_builtin_setjmp_receiver.
6458         (expand_label): Adjust, call expand_builtin_setjmp_receiver
6459         with NULL for the label parameter.
6460         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
6461         the frame-pointer.  Adjust comments.
6462         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
6463         only if LABEL is non-NULL.
6465 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
6467         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
6468         (ARM_ABI_AAPCS64): Ditto.
6469         (arm_abi): Ditto.
6470         (ARM_DEFAULT_ABI): Ditto.
6472 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
6474         * config/aarch64/aarch64-builtins.c
6475         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
6476         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
6477         (st1): Likewise.
6478         * config/aarch64/aarch64-simd.md
6479         (aarch64_ld1<VALL:mode>): New.
6480         (aarch64_st1<VALL:mode>): Likewise.
6481         * config/aarch64/arm_neon.h
6482         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
6484 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
6486         * config/i386/i386.c (gate_insert_vzeroupper): Check if
6487         target ISA is AVX.
6488         (ix86_option_override_internal):Turn on all -mavx target flags by
6489         default as they are dependent on AVX anyway.
6491 2013-07-02  Cary Coutant  <ccoutant@google.com>
6493         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
6494         deterministic hash.
6495         (loc_checksum_ordered): Likewise.
6496         (hash_loc_operands): Remove inline keyword.
6498 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
6500         PR tree-optimization/57741
6501         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
6502         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
6503         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
6504         Allow REAL_CST step_exprs if flag_associative_math.
6505         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
6507 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
6509         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
6511 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
6513         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
6515 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6517         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
6518         encoding.
6519         (iorsi3_insn): Likewise.
6520         (arm_xorsi3): Likewise.
6522 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
6524         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
6525         to ...
6526         (attribute "type"): ... here, and prefix with "wmmx_".
6527         (attribute "core_cycles"): Update for attribute changes.
6528         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
6529         (tbcstv4hi): Likewise.
6530         (tbcstv2si): Likewise.
6531         (iwmmxt_iordi3): Likewise.
6532         (iwmmxt_xordi3): Likewise.
6533         (iwmmxt_anddi3): Likewise.
6534         (iwmmxt_nanddi3): Likewise.
6535         (iwmmxt_arm_movdi): Likewise.
6536         (iwmmxt_movsi_insn): Likewise.
6537         (mov<mode>_internal): Likewise.
6538         (and<mode>3_iwmmxt): Likewise.
6539         (ior<mode>3_iwmmxt): Likewise.
6540         (xor<mode>3_iwmmxt): Likewise.
6541         (add<mode>3_iwmmxt): Likewise.
6542         (ssaddv8qi3): Likewise.
6543         (ssaddv4hi3): Likewise.
6544         (ssaddv2si3): Likewise.
6545         (usaddv8qi3): Likewise.
6546         (usaddv4hi3): Likewise.
6547         (usaddv2si3): Likewise.
6548         (sub<mode>3_iwmmxt): Likewise.
6549         (sssubv8qi3): Likewise.
6550         (sssubv4hi3): Likewise.
6551         (sssubv2si3): Likewise.
6552         (ussubv8qi3): Likewise.
6553         (ussubv4hi3): Likewise.
6554         (ussubv2si3): Likewise.
6555         (mulv4hi3_iwmmxt): Likewise.
6556         (smulv4hi3_highpart): Likewise.
6557         (umulv4hi3_highpart): Likewise.
6558         (iwmmxt_wmacs): Likewise.
6559         (iwmmxt_wmacsz): Likewise.
6560         (iwmmxt_wmacu): Likewise.
6561         (iwmmxt_wmacuz): Likewise.
6562         (iwmmxt_clrdi): Likewise.
6563         (iwmmxt_clrv8qi): Likewise.
6564         (iwmmxt_clr4hi): Likewise.
6565         (iwmmxt_clr2si): Likewise.
6566         (iwmmxt_uavgrndv8qi3): Likewise.
6567         (iwmmxt_uavgrndv4hi3): Likewise.
6568         (iwmmxt_uavgv8qi3): Likewise.
6569         (iwmmxt_uavgv4hi3): Likewise.
6570         (iwmmxt_tinsrb): Likewise.
6571         (iwmmxt_tinsrh): Likewise.
6572         (iwmmxt_tinsrw): Likewise.
6573         (iwmmxt_textrmub): Likewise.
6574         (iwmmxt_textrmsb): Likewise.
6575         (iwmmxt_textrmuh): Likewise.
6576         (iwmmxt_textrmsh): Likewise.
6577         (iwmmxt_textrmw): Likewise.
6578         (iwmxxt_wshufh): Likewise.
6579         (eqv8qi3): Likewise.
6580         (eqv4hi3): Likewise.
6581         (eqv2si3): Likewise.
6582         (gtuv8qi3): Likewise.
6583         (gtuv4hi3): Likewise.
6584         (gtuv2si3): Likewise.
6585         (gtv8qi3): Likewise.
6586         (gtv4hi3): Likewise.
6587         (gtv2si3): Likewise.
6588         (smax<mode>3_iwmmxt): Likewise.
6589         (umax<mode>3_iwmmxt): Likewise.
6590         (smin<mode>3_iwmmxt): Likewise.
6591         (umin<mode>3_iwmmxt): Likewise.
6592         (iwmmxt_wpackhss): Likewise.
6593         (iwmmxt_wpackwss): Likewise.
6594         (iwmmxt_wpackdss): Likewise.
6595         (iwmmxt_wpackhus): Likewise.
6596         (iwmmxt_wpackwus): Likewise.
6597         (iwmmxt_wpackdus): Likewise.
6598         (iwmmxt_wunpckihb): Likewise.
6599         (iwmmxt_wunpckihh): Likewise.
6600         (iwmmxt_wunpckihw): Likewise.
6601         (iwmmxt_wunpckilb): Likewise.
6602         (iwmmxt_wunpckilh): Likewise.
6603         (iwmmxt_wunpckilw): Likewise.
6604         (iwmmxt_wunpckehub): Likewise.
6605         (iwmmxt_wunpckehuh): Likewise.
6606         (iwmmxt_wunpckehuw): Likewise.
6607         (iwmmxt_wunpckehsb): Likewise.
6608         (iwmmxt_wunpckehsh): Likewise.
6609         (iwmmxt_wunpckehsw): Likewise.
6610         (iwmmxt_wunpckelub): Likewise.
6611         (iwmmxt_wunpckeluh): Likewise.
6612         (iwmmxt_wunpckeluw): Likewise.
6613         (iwmmxt_wunpckelsb): Likewise.
6614         (iwmmxt_wunpckelsh): Likewise.
6615         (iwmmxt_wunpckelsw): Likewise.
6616         (ror<mode>3): Likewise.
6617         (ashr<mode>3_iwmmxt): Likewise.
6618         (lshr<mode>3_iwmmxt): Likewise.
6619         (ashl<mode>3_iwmmxt): Likewise.
6620         (ror<mode>3_di): Likewise.
6621         (ashr<mode>3_di): Likewise.
6622         (lshr<mode>3_di): Likewise.
6623         (ashl<mode>3_di): Likewise.
6624         (iwmmxt_wmadds): Likewise.
6625         (iwmmxt_wmaddu): Likewise.
6626         (iwmmxt_tmia): Likewise.
6627         (iwmmxt_tmiaph): Likewise.
6628         (iwmmxt_tmiabb): Likewise.
6629         (iwmmxt_tmiatb): Likewise.
6630         (iwmmxt_tmiabt): Likewise.
6631         (iwmmxt_tmiatt): Likewise.
6632         (iwmmxt_tmovmskb): Likewise.
6633         (iwmmxt_tmovmskh): Likewise.
6634         (iwmmxt_tmovmskw): Likewise.
6635         (iwmmxt_waccb): Likewise.
6636         (iwmmxt_wacch): Likewise.
6637         (iwmmxt_waccw): Likewise.
6638         (iwmmxt_waligni): Likewise.
6639         (iwmmxt_walignr): Likewise.
6640         (iwmmxt_walignr0): Likewise.
6641         (iwmmxt_walignr1): Likewise.
6642         (iwmmxt_walignr2): Likewise.
6643         (iwmmxt_walignr3): Likewise.
6644         (iwmmxt_wsadb): Likewise.
6645         (iwmmxt_wsadh): Likewise.
6646         (iwmmxt_wsadbz): Likewise.
6647         (iwmmxt_wsadhz): Likewise.
6648         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
6649         (iwmmxt_wabsdiffb): Likewise.
6650         (iwmmxt_wabsdiffh): Likewise.
6651         (iwmmxt_wabsdiffw): Likewise.
6652         (iwmmxt_waddsubhx): Likewise
6653         (iwmmxt_wsubaddhx): Likewise.
6654         (addc<mode>3): Likewise.
6655         (iwmmxt_avg4): Likewise.
6656         (iwmmxt_avg4r): Likewise.
6657         (iwmmxt_wmaddsx): Likewise.
6658         (iwmmxt_wmaddux): Likewise.
6659         (iwmmxt_wmaddsn): Likewise.
6660         (iwmmxt_wmaddun): Likewise.
6661         (iwmmxt_wmulwsm): Likewise.
6662         (iwmmxt_wmulwum): Likewise.
6663         (iwmmxt_wmulsmr): Likewise.
6664         (iwmmxt_wmulumr): Likewise.
6665         (iwmmxt_wmulwsmr): Likewise.
6666         (iwmmxt_wmulwumr): Likewise.
6667         (iwmmxt_wmulwl): Likewise.
6668         (iwmmxt_wqmulm): Likewise.
6669         (iwmmxt_wqmulwm): Likewise.
6670         (iwmmxt_wqmulmr): Likewise.
6671         (iwmmxt_wqmulwmr): Likewise.
6672         (iwmmxt_waddbhusm): Likewise.
6673         (iwmmxt_waddbhusl): Likewise.
6674         (iwmmxt_wqmiabb): Likewise.
6675         (iwmmxt_wqmiabt): Likewise.
6676         (iwmmxt_wqmiatb): Likewise.
6677         (iwmmxt_wqmiatt): Likewise.
6678         (iwmmxt_wqmiabbn): Likewise.
6679         (iwmmxt_wqmiabtn): Likewise.
6680         (iwmmxt_wqmiatbn): Likewise.
6681         (iwmmxt_wqmiattn): Likewise.
6682         (iwmmxt_wmiabb): Likewise.
6683         (iwmmxt_wmiabt): Likewise.
6684         (iwmmxt_wmiatb): Likewise.
6685         (iwmmxt_wmiatt): Likewise.
6686         (iwmmxt_wmiabbn): Likewise.
6687         (iwmmxt_wmiabtn): Likewise.
6688         (iwmmxt_wmiatbn): Likewise.
6689         (iwmmxt_wmiattn): Likewise.
6690         (iwmmxt_wmiawbb): Likewise.
6691         (iwmmxt_wmiawbt): Likewise.
6692         (iwmmxt_wmiawtb): Likewise.
6693         (iwmmxt_wmiawtt): Likewise.
6694         (iwmmxt_wmiawbbn): Likewise.
6695         (iwmmxt_wmiawbtn): Likewise.
6696         (iwmmxt_wmiawtbn): Likewise.
6697         (iwmmxt_wmiawttn): Likewise.
6698         (iwmmxt_wmerge): Likewise.
6699         (iwmmxt_tandc<mode>3): Likewise.
6700         (iwmmxt_torc<mode>3): Likewise.
6701         (iwmmxt_torvsc<mode>3): Likewise.
6702         (iwmmxt_textrc<mode>3): Likewise.
6703         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
6704         (wmmxt_pack): Likewise.
6705         (wmmxt_mult_c1): Likewise.
6706         (wmmxt_mult_c2): Likewise.
6707         (wmmxt_alu_c1): Likewise.
6708         (wmmxt_alu_c2): Likewise.
6709         (wmmxt_alu_c3): Likewise.
6710         (wmmxt_transfer_c1): Likewise.
6711         (wmmxt_transfer_c2): Likewise.
6712         (wmmxt_transfer_c3): Likewise.
6713         (marvell_f_iwmmxt_wstr): Likewise.
6714         (marvell_f_iwmmxt_wldr): Likewise.
6716 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
6718         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
6720 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
6722         Revert:
6723         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
6724         * lra-constraints.c (need_for_split_p): Check call used hard regs
6725         living through calls.
6727         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
6728         call used regs for call insn.
6730 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
6732         PR target/57736
6733         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
6734         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
6736 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6738         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
6740 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
6742         * lra-constraints.c (need_for_split_p): Check call used hard regs
6743         living through calls.
6745 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
6747         PR target/57744
6748         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
6749         to tie with any other modes.  Eliminate Altivec vector mode tests,
6750         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
6751         code, to return 0 if testing MODE2 for a condition, if we've
6752         already tested MODE1 for the same condition.
6754 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6756         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
6757         layout.
6759 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6761         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
6762         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
6764 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6766         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
6767         Define.
6768         (aarch64_symbolic_constant_p): Remove.
6769         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
6770         static.  Fix line length and white space.
6771         (aarch64_symbolic_constant_p): Remove.
6772         * config/aarch64/predicates.md (aarch64_valid_symref):
6773         Use aarch64_classify_symbol_expression.
6775 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6777         * config/arm/constraints.md (Ts): New constraint.
6778         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
6779         16-bit encodings.
6780         (compare_scc): Use "Ts" constraint for operand 0.
6781         (ior_scc_scc): Likewise.
6782         (and_scc_scc): Likewise.
6783         (and_scc_scc_nodom): Likewise.
6784         (ior_scc_scc_cmp): Likewise for operand 7.
6785         (and_scc_scc_cmp): Likewise.
6786         * config/arm/thumb2.md (thumb2_movsi_insn):
6787         Add alternatives for 16-bit encodings.
6788         (thumb2_movhi_insn): Likewise.
6789         (thumb2_movsicc_insn): Likewise.
6790         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
6791         (thumb2_negscc): Use "Ts" constraint.
6792         Move mvn instruction outside cond_exec block.
6793         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
6794         for 16-bit encodings.
6796 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6798         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
6799         encoding.
6800         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
6801         (mulsi3subsi): Likewise.
6802         (mulsidi3adddi): Likewise.
6803         (mulsidi3_v6): Likewise.
6804         (umulsidi3_v6): Likewise.
6805         (umulsidi3adddi_v6): Likewise.
6806         (smulsi3_highpart_v6): Likewise.
6807         (umulsi3_highpart_v6): Likewise.
6808         (mulhisi3tb): Likewise.
6809         (mulhisi3bt): Likewise.
6810         (mulhisi3tt): Likewise.
6811         (maddhisi4): Likewise.
6812         (maddhisi4tb): Likewise.
6813         (maddhisi4tt): Likewise.
6814         (maddhidi4): Likewise.
6815         (maddhidi4tb): Likewise.
6816         (maddhidi4tt): Likewise.
6817         (zeroextractsi_compare0_scratch): Likewise.
6818         (insv_zero): Likewise.
6819         (insv_t2): Likewise.
6820         (anddi_notzesidi_di): Likewise.
6821         (anddi_notsesidi_di): Likewise.
6822         (andsi_notsi_si): Likewise.
6823         (iordi_zesidi_di): Likewise.
6824         (xordi_zesidi_di): Likewise.
6825         (andsi_iorsi3_notsi): Likewise.
6826         (smax_0): Likewise.
6827         (smax_m1): Likewise.
6828         (smin_0): Likewise.
6829         (not_shiftsi): Likewise.
6830         (unaligned_loadsi): Likewise.
6831         (unaligned_loadhis): Likewise.
6832         (unaligned_loadhiu): Likewise.
6833         (unaligned_storesi): Likewise.
6834         (unaligned_storehi): Likewise.
6835         (extv_reg): Likewise.
6836         (extzv_t2): Likewise.
6837         (divsi3): Likewise.
6838         (udivsi3): Likewise.
6839         (arm_zero_extendhisi2addsi): Likewise.
6840         (arm_zero_extendqisi2addsi): Likewise.
6841         (compareqi_eq0): Likewise.
6842         (arm_extendhisi2_v6): Likewise.
6843         (arm_extendqisi2addsi): Likewise.
6844         (arm_movt): Likewise.
6845         (thumb2_ldrd): Likewise.
6846         (thumb2_ldrd_base): Likewise.
6847         (thumb2_ldrd_base_neg): Likewise.
6848         (thumb2_strd): Likewise.
6849         (thumb2_strd_base): Likewise.
6850         (thumb2_strd_base_neg): Likewise.
6851         (arm_negsi2): Add alternative for 16-bit encoding.
6852         (arm_one_cmplsi2): Likewise.
6854 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6856         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
6857         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
6858         (movdfcc): Likewise.
6859         * config/arm/vfp.md (*thumb2_movsf_vfp):
6860         Disable predication for arm_restrict_it.
6861         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
6862         (*thumb2_movdfcc_vfp): Likewise.
6863         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
6864         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
6865         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
6866         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
6867         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
6868         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
6869         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
6870         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
6871         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
6872         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
6873         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
6874         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
6875         Disable predication for arm_restrict_it.
6877 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
6879         * config/i386/bmiintrin.h (_bextr_u32): New.
6880         (_bextr_u64): Ditto.
6882 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
6884         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
6885         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
6886         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
6887         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
6888         * config/mips/n32-elf.h: ...this new file.
6890 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
6892         PR target/57224
6893         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
6894         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
6896 2013-06-27  Catherine Moore  <clm@codesourcery.com>
6898         * config/mips/mips-tables.opt: Regenerate.
6899         * config/mips/mips-cpus.def: Add m14ke and m14kec.
6900         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
6901         * doc/invoke.texi: Add -m14kc.
6903 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6905         PR target/57623
6906         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
6907         constraints of operand 1 and 2.
6909         PR target/57623
6910         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
6911         to match RTL canonicalization.  Swap predicates and
6912         constraints of operand 1 and 2.
6914 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
6916         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
6917         Process OP_INOUT regs for splitting too.
6919 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
6921         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
6922         decl before the loop, initialize to NULL.
6923         (vectorizable_load): Initialize ptr_incr to NULL.
6925 2013-06-27  Martin Jambor  <mjambor@suse.cz>
6927         PR lto/57208
6928         * ipa-ref.h (ipa_maybe_record_reference): Declare.
6929         * ipa-ref.c (ipa_maybe_record_reference): New function.
6930         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
6931         * ipa-cp.c (create_specialized_node): Record potential references from
6932         aggvals.
6933         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
6935 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6937         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
6938         parameter 'mode' of type 'enum machine_mode mode'; change to pass
6939         'mode' to force_reg.
6940         (aarch64_add_offset): Update calls to aarch64_force_temporary.
6941         (aarch64_expand_mov_immediate): Likewise.
6943 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
6945         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6946         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6948 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6950         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6951         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6952         (struct machine_function): Add tbegin_p.
6953         (s390_canonicalize_comparison): Fold CC mode compares to
6954         conditional jump if possible.
6955         (s390_emit_jump): Return the emitted jump.
6956         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6957         Handle CCRAWmode compares.
6958         (s390_option_override): Default to -mhtm if available.
6959         (s390_reg_clobbered_rtx): Handle floating point regs as well.
6960         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6961         FPRs instead of df_regs_ever_live_p.
6962         (s390_optimize_nonescaping_tx): New function.
6963         (s390_init_frame_layout): Extend clobbered_regs array to cover
6964         FPRs as well.
6965         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6966         (s390_expand_tbegin): New function.
6967         (enum s390_builtin): New enum definition.
6968         (code_for_builtin): New array definition.
6969         (s390_init_builtins): New function.
6970         (s390_expand_builtin): New function.
6971         (TARGET_INIT_BUILTINS): Define.
6972         (TARGET_EXPAND_BUILTIN): Define.
6973         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6974         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6975         (s390_alc_comparison): Likewise.
6976         * config/s390/s390-modes.def: Add CCRAWmode.
6977         * config/s390/s390.h (processor_flags): Add PF_TX.
6978         (TARGET_CPU_HTM): Define macro.
6979         (TARGET_HTM): Define macro.
6980         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6981         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6982         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6983         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6984         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6985         (TBEGIN_MASK, TBEGINC_MASK): New constants.
6986         ("*cc_to_int"): Move up.
6987         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6988         constants other than 0.
6989         ("*ccraw_to_int"): New insn and splitter definition.
6990         ("tbegin", "tbegin_nofloat", "tbegin_retry")
6991         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6992         ("tx_assist"): New expander.
6993         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6994         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6995         * config/s390/s390.opt: Add -mhtm option.
6996         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6997         * config/s390/htmxlintrin.h: New file.
6998         * config/s390/htmintrin.h: New file.
6999         * config/s390/s390intrin.h: New file.
7000         * doc/extend.texi: Document htm builtins.
7001         * config.gcc: Add the new header files to extra_headers.
7003 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
7005         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
7006         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
7008 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
7009             Pat Haugen  <pthaugen@us.ibm.com>
7010             Peter Bergner  <bergner@vnet.ibm.com>
7012         * config/rs6000/power8.md: New.
7013         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
7014         setting for power8 entry.
7015         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
7016         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
7017         test for Power4/Power5 only.
7018         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
7019         support.
7020         (force_new_group): Adjust comment.
7021         * config/rs6000/rs6000.md: Include power8.md.
7023 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
7025         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
7026         * config/arm/arm-protos.h (arm_max_conditional_execute): New
7027         declaration.
7028         (tune_params): Update comment.
7029         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
7030         (arm_max_conditional_execute): New function.
7031         (thumb2_final_prescan_insn): Use max_insn_skipped and
7032         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
7034 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
7036         PR tree-optimization/57705
7037         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
7038         SSA_NAME step, provided that it is not defined inside the loop.
7039         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
7040         (get_initial_def_for_induction): Handle SSA_NAME IV step.
7042 2013-06-25  Martin Jambor  <mjambor@suse.cz>
7044         PR middle-end/57670
7045         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
7046         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
7047         calls in the dump.
7048         (ipa_note_param_call): Initialize member_ptr flag.
7049         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
7050         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
7051         (ipa_write_indirect_edge_info): Stream member_ptr flag.
7052         (ipa_read_indirect_edge_info): Likewise.
7054 2013-06-25  Richard Biener  <rguenther@suse.de>
7056         PR middle-end/56977
7057         * passes.c (init_optimization_passes): Move pass_fold_builtins
7058         and pass_dce earlier with -Og.
7060 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
7062         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
7063         <BIT_FIELD_REF>: Remove trailing TAB.
7064         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
7065         remove blank line.
7067 2013-06-24  Martin Jambor  <mjambor@suse.cz>
7069         PR tree-optimization/57358
7070         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
7071         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
7072         (ipa_analyze_params_uses): Generate pessimistic info when true.
7074 2013-06-24  Martin Jambor  <mjambor@suse.cz>
7076         PR tree-optimization/57539
7077         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
7078         global.inlined_to of the new node to it.  All callers changed.
7079         * ipa-inline-transform.c (clone_inlined_nodes): New variable
7080         inlining_into, pass it to cgraph_clone_node.
7081         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
7082         ipa_free_edge_args_substructures.
7083         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
7084         rdesc linked list.  Do not assert rdesc edges have inlined caller.
7085         Assert we have found an rdesc in the rdesc list.
7087 2013-06-24  Richard Biener  <rguenther@suse.de>
7089         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
7090         (pointer_set_lookup): Declare.
7091         (class pointer_map): New template class implementing a
7092         generic pointer to T map.
7093         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
7094         pointer_map<T>::contains, pointer_map<T>::insert,
7095         pointer_map<T>::traverse): New functions.
7096         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
7097         (pointer_set_lookup): New function.
7098         (pointer_set_contains): Use pointer_set_lookup.
7099         (pointer_set_insert): Likewise.
7100         (insert_aux): Remove.
7101         (struct pointer_map_t): Embed a pointer_set_t.
7102         (pointer_map_create): Adjust.
7103         (pointer_map_destroy): Likewise.
7104         (pointer_map_contains): Likewise.
7105         (pointer_map_insert): Likewise.
7106         (pointer_map_traverse): Likewise.
7107         * tree-streamer.h (struct streamer_tree_cache_d): Use a
7108         pointer_map<unsigned> instead of a pointer_map_t.
7109         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
7110         (streamer_tree_cache_lookup): Likewise.
7111         (streamer_tree_cache_create): Likewise.
7112         (streamer_tree_cache_delete): Likewise.
7113         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
7114         pointer_map<unsigned> instead of a pointer_map_t.
7115         (lto_init_tree_ref_encoder): Adjust.
7116         (lto_destroy_tree_ref_encoder): Likewise.
7117         * lto-section-out.c (lto_output_decl_index): Likewise.
7118         (lto_record_function_out_decl_state): Likewise.
7119         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
7121 2013-06-24  Richard Biener  <rguenther@suse.de>
7123         PR tree-optimization/57488
7124         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
7126 2013-06-24  Alan Modra  <amodra@gmail.com>
7128         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
7129         (gen_easy_altivec_constant): Likewise.
7130         * config/rs6000/predicates.md (easy_vector_constant_add_self,
7131         easy_vector_constant_msb): Likewise.
7133 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
7135         PR target/57688
7136         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
7137         add missing return true.
7139 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
7141         PR target/52483
7142         * config/sh/predicates.md (general_extend_operand): Invoke
7143         general_movsrc_operand for memory operands.
7144         (general_movsrc_operand): Allow reg+reg addressing, do not use
7145         general_operand for memory operands.
7147 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
7149         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
7150         when current target options does not apply.
7151         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
7152         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
7153         * config/i386/bmiintrin.h: Pass appropriate target
7154         attributes to header.
7155         * config/i386/mmintrin.h: Ditto.
7156         * config/i386/nmmintrin.h: Ditto.
7157         * config/i386/avx2intrin.h: Ditto.
7158         * config/i386/fxsrintrin.h: Ditto.
7159         * config/i386/tbmintrin.h: Ditto.
7160         * config/i386/xsaveintrin.h: Ditto.
7161         * config/i386/f16cintrin.h: Ditto.
7162         * config/i386/xtestintrin.h: Ditto.
7163         * config/i386/xsaveoptintrin.h: Ditto.
7164         * config/i386/bmi2intrin.h: Ditto.
7165         * config/i386/lzcntintrin.h: Ditto.
7166         * config/i386/smmintrin.h: Ditto.
7167         * config/i386/wmmintrin.h: Ditto.
7168         * config/i386/x86intrin.h: Remove all header include guards.
7169         * config/i386/prfchwintrin.h: Ditto.
7170         * config/i386/pmmintrin.h: Ditto.
7171         * config/i386/tmmintrin.h: Ditto.
7172         * config/i386/xmmintrin.h: Ditto.
7173         * config/i386/popcntintrin.h: Ditto.
7174         * config/i386/rdseedintrin.h: Ditto.
7175         * config/i386/ammintrin.h: Ditto.
7176         * config/i386/emmintrin.h: Ditto.
7177         * config/i386/immintrin.h: Remove all header include guards.
7178         * config/i386/fma4intrin.h: Ditto.
7179         * config/i386/lwpintrin.h: Ditto.
7180         * config/i386/xopintrin.h: Ditto.
7181         * config/i386/ia32intrin.h: Ditto.
7182         * config/i386/avxintrin.h: Ditto.
7183         * config/i386/rtmintrin.h: Ditto.
7184         * config/i386/fmaintrin.h: Ditto.
7185         * config/i386/mm3dnow.h: Ditto.
7187 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
7189         * common/config/i386/i386-common.c: Handle LZCNT.
7191 2013-06-22  Andi Kleen  <ak@linux.intel.com>
7193         * doc/extend.texi: Use __atomic_store_n instead of
7194         __atomic_store in HLE example.
7196 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
7198         * config/sh/sh.c: Remove <cstdlib> workaround.
7200 2013-06-21  Andi Kleen  <ak@linux.intel.com>
7202         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
7204 2013-06-21  Andi Kleen  <ak@linux.intel.com>
7206         * doc/extend.texi: Document that __atomic_clear and
7207           __atomic_test_and_set should only be used with bool.
7209 2013-06-20  Jan Hubicka  <jh@suse.cz>
7211         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
7212         types_same_for_odr.
7213         * tree.c (decls_same_for_odr): New function.
7214         (same_for_edr): New function.
7215         (types_same_for_odr): New function.
7216         (get_binfo_at_offset): Use it.
7217         * tree.h (types_same_for_odr): Declare.
7219 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
7220             Jason Merrill  <jason@redhat.com>
7222         * system.h: Include <cstdlib> as well as <stdlib.h>.
7224 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
7226         PR target/57655
7227         * config/i386/i386.c (construct_container): Report error if
7228         long double is used with disabled x87 float returns.
7230 2013-06-20  Jan Hubicka  <jh@suse.cz>
7232         * lto-cgraph.c (input_symtab): Do not set cgraph state.
7234 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
7236         PR rtl-optimization/57425
7237         PR rtl-optimization/57569
7238         * alias.c (write_dependence_p): Remove parameters mem_mode and
7239         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
7240         Changed all callers.
7241         (canon_anti_dependence): Get comments and semantics in sync.
7242         Add parameter mem_canonicalized.  Changed all callers.
7243         * rtl.h (canon_anti_dependence): Update prototype.
7245 2013-06-20  Richard Biener  <rguenther@suse.de>
7247         * data-streamer-in.c (streamer_read_uhwi): Optimize single
7248         byte case, inline streamer_read_uchar and defer section
7249         overrun check.
7251 2013-06-20  Richard Biener  <rguenther@suse.de>
7253         PR tree-optimization/57584
7254         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
7255         SSA names into the expanded expression that take part in
7256         abnormal coalescing.
7258 2013-06-19  Sharad Singhai  <singhai@google.com>
7260         * gcov.c (print_usage): Handle new option.
7261         (process_args): Ditto.
7262         (get_gcov_intermediate_filename): New function.
7263         (output_intermediate_file): New function.
7264         (output_gcov_file): New function
7265         (generate_results): Handle new option.
7266         (release_function): Relase demangled name.
7267         (read_graph_file): Handle demangled name.
7268         (output_lines): Ditto.
7269         * doc/gcov.texi: Document gcov intermediate format.
7271 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
7273         PR bootstrap/57604
7274         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
7275         (lra_emit_add): Use the functions.  Add comment about Y as an
7276         address segment.
7278 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
7280         PR driver/57652
7281         * collect2.c (collect_atexit): New.
7282         (collect_exit): Delete.
7283         (main): Register collect_atexit with atexit.
7284         (collect_wait): Change collect_exit to exit.
7285         (do_wait): Same.
7286         * collect2.h (collect_exit): Delete.
7287         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
7289 2013-06-19  Wei Mi  <wmi@google.com>
7291         PR rtl-optimization/57518
7292         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
7293         if regno is used in paradoxical subreg.
7294         (update_equiv_regs): Check pdx_subregs[regno] before
7295         set a reg to be equivalent with a mem.
7297 2013-06-19  Matthias Klose  <doko@ubuntu.com>
7299         PR driver/57651
7300         * file-find.h (find_a_file): Add a mode parameter.
7301         * file-find.c (find_a_file): Likewise.
7302         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
7303         with X_OK for the executables.
7304         * collect2.c (main): Call find_a_file with X_OK.
7306 2013-06-19  Steve Ellcey  <sellcey@mips.com>
7308         PR target/56942
7309         * config/mips/mips.md (casesi_internal_mips16_<mode>):
7310         Use NEXT_INSN instead of next_real_insn.
7312 2013-06-19  Jan Hubicka  <jh@suse.cz>
7314         * cgraph.h (const_value_known_p): Replace by ...
7315         (ctor_for_folding): .. this one.
7316         * cgraphunit.c (process_function_and_variable_attributes): Use it.
7317         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
7318         * expr.c (expand_expr_real_1): Likewise.
7319         (string_constant): Likewise.
7320         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
7321         * ipa.c (process_references): Likewise.
7322         (symtab_remove_unreachable_nodes): Likewise.
7323         * ipa-inline-analysis.c (param_change_prob): Likewise.
7324         * gimple-fold.c (canonicalize_constructor_val): Likewise.
7325         (get_base_constructor): Likwise.
7326         * varpool.c (varpool_remove_node): Likewise.
7327         (varpool_remove_initializer): LIkewise.
7328         (dump_varpool_node): LIkwise.
7329         (const_value_known_p): Rewrite to ...
7330         (ctor_for_folding): ... this one.
7332 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
7334         PR driver/57651
7335         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
7336         PERSONALITY in $PATH derived prefixes.
7338 2013-06-19  Jeff Law  <law@redhat.com>
7340         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
7341         in comment.
7343         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
7344         (simplify_bitwise_binary): Use it to simpify certain binary ops on
7345         booleans.
7347 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
7349         * config/arm/vfp.md: Move VFP instruction classification documentation
7350         to ...
7351         * config/arm/arm.md: ... here.  Update instruction classification
7352         documentation.
7354 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
7356         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
7357         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
7358         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
7360 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
7362         PR target/57609
7363         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
7364         with NEXT_INSN.  Use tablejump_p to check for jump table data
7365         insns.
7367 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
7369         PR c++/56544
7370         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
7371         that now in C++ the value is correct per the C++ standards.
7373 2013-06-19  Richard Biener  <rguenther@suse.de>
7375         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
7376         for global context.
7378 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7380         Revert:
7381         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7383         PR target/57609
7384         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
7385         with next_active_insn.
7387 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
7389         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
7390         functions are inlined during failures to flag an error.
7391         * tree-inline.c (expand_call_inline): Allow the error to be flagged
7392         in early inline pass.
7394 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
7396         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
7397         in comments.
7399 2013-06-18  Julian Brown  <julian@codesourcery.com>
7401         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
7402         Permit virtual register pre-reload if !strict.
7403         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
7404         change.
7405         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
7406         prototype.
7407         * config/arm/neon.md (movmisalign<mode>): Use
7408         neon_perm_struct_or_reg_operand instead of
7409         neon_struct_or_register_operand.
7410         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
7411         neon_permissive_struct_operand instead of neon_struct_operand.
7412         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
7413         neon_vector_mem_operand.
7414         * config/arm/predicates.md (neon_struct_operand): Adjust call to
7415         neon_vector_mem_operand.
7416         (neon_permissive_struct_operand): New.
7417         (neon_struct_or_register_operand): Rename to...
7418         (neon_perm_struct_or_reg_operand): This. Adjust call to
7419         neon_vector_mem_operand.
7421 2013-06-18  Richard Biener  <rguenther@suse.de>
7423         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
7424         * lto-streamer.h: Include pointer-set.h.
7425         (struct lto_decl_slot): Remove.
7426         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
7427         Remove next_index entry.
7428         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
7429         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
7430         (lto_init_tree_ref_encoder): Adjust.
7431         (lto_destroy_tree_ref_encoder): Likewise.
7432         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
7433         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
7434         (lto_output_decl_index): Adjust.
7435         (lto_new_out_decl_state): Likewise.
7436         (lto_record_function_out_decl_state): Likewise.
7437         * lto-streamer-out.c (copy_function): Likewise.
7439 2013-06-18  Richard Biener  <rguenther@suse.de>
7441         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
7442         * cgraphunit.c: Include cfgloop.h.
7443         (init_lowered_empty_function): Initialize the loop tree.
7444         (assemble_thunk): Insert new BBs into loops.
7446 2013-06-18  Richard Biener  <rguenther@suse.de>
7448         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
7449         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
7450         the map from cache entry to cache index optional.
7451         (streamer_tree_cache_replace_tree): Adjust accordingly.
7452         (streamer_tree_cache_append): Likewise.
7453         (streamer_tree_cache_delete): Likewise.
7454         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
7455         streamer cache map from cache entry to cache index.
7456         * lto-streamer-out.c (create_output_block): Adjust.
7458 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
7460         * config/arm/arm.md (attribute "insn"): Move multiplication and
7461         division attributes to...
7462         (attribute "type"): ... here.  Remove mult.
7463         (attribute "mul32"): New attribute.
7464         (attribute "mul64"): Add umaal.
7465         (*arm_mulsi3): Update attributes.
7466         (*arm_mulsi3_v6): Likewise.
7467         (*thumb_mulsi3): Likewise.
7468         (*thumb_mulsi3_v6): Likewise.
7469         (*mulsi3_compare0): Likewise.
7470         (*mulsi3_compare0_v6): Likewise.
7471         (*mulsi_compare0_scratch): Likewise.
7472         (*mulsi_compare0_scratch_v6): Likewise.
7473         (*mulsi3addsi): Likewise.
7474         (*mulsi3addsi_v6): Likewise.
7475         (*mulsi3addsi_compare0): Likewise.
7476         (*mulsi3addsi_compare0_v6): Likewise.
7477         (*mulsi3addsi_compare0_scratch): Likewise.
7478         (*mulsi3addsi_compare0_scratch_v6): Likewise.
7479         (*mulsi3subsi): Likewise.
7480         (*mulsidi3adddi): Likewise.
7481         (*mulsi3addsi_v6): Likewise.
7482         (*mulsidi3adddi_v6): Likewise.
7483         (*mulsidi3_nov6): Likewise.
7484         (*mulsidi3_v6): Likewise.
7485         (*umulsidi3_nov6): Likewise.
7486         (*umulsidi3_v6): Likewise.
7487         (*umulsidi3adddi): Likewise.
7488         (*umulsidi3adddi_v6): Likewise.
7489         (*smulsi3_highpart_nov6): Likewise.
7490         (*smulsi3_highpart_v6): Likewise.
7491         (*umulsi3_highpart_nov6): Likewise.
7492         (*umulsi3_highpart_v6): Likewise.
7493         (mulhisi3): Likewise.
7494         (*mulhisi3tb): Likewise.
7495         (*mulhisi3bt): Likewise.
7496         (*mulhisi3tt): Likewise.
7497         (maddhisi4): Likewise.
7498         (*maddhisi4tb): Likewise.
7499         (*maddhisi4tt): Likewise.
7500         (maddhidi4): Likewise.
7501         (*maddhidi4tb): Likewise.
7502         (*maddhidi4tt): Likewise.
7503         (divsi3): Likewise.
7504         (udivsi3): Likewise.
7505         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
7506         (thumb2_mulsi_short_compare0): Likewise.
7507         (thumb2_mulsi_short_compare0_scratch): Likewise.
7508         * config/arm/arm1020e.md (1020mult1): Update attribute change.
7509         (1020mult2): Likewise.
7510         (1020mult3): Likewise.
7511         (1020mult4): Likewise.
7512         (1020mult5): Likewise.
7513         (1020mult6): Likewise.
7514         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
7515         change.
7516         (cortex_a15_mult64): Likewise.
7517         (cortex_a15_sdiv): Likewise.
7518         (cortex_a15_udiv): Likewise.
7519         * config/arm/arm1026ejs.md (mult1): Update attribute change.
7520         (mult2): Likewise.
7521         (mult3): Likewise.
7522         (mult4): Likewise.
7523         (mult5): Likewise.
7524         (mult6): Likewise.
7525         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
7526         (pj4_ir_div): Likewise.
7527         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
7528         (11_mult2): Likewise.
7529         (11_mult3): Likewise.
7530         (11_mult4): Likewise.
7531         (11_mult5): Likewise.
7532         (11_mult6): Likewise.
7533         (11_mult7): Likewise.
7534         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
7535         (cortex_a8_mla): Likewise.
7536         (cortex_a8_mull): Likewise.
7537         (cortex_a8_smulwy): Likewise.
7538         (cortex_a8_smlald): Likewise.
7539         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
7540         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
7541         (cortex_r4_mul_3): Likewise.
7542         (cortex_r4_mla_4): Likewise.
7543         (cortex_r4_mla_3): Likewise.
7544         (cortex_r4_smlald): Likewise.
7545         (cortex_r4_mull): Likewise.
7546         (cortex_r4_sdiv): Likewise.
7547         (cortex_r4_udiv): Likewise.
7548         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
7549         (cortex_a7_idiv): Likewise.
7550         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
7551         (9_mult2): Likewise.
7552         (9_mult3): Likewise.
7553         (9_mult4): Likewise.
7554         (9_mult5): Likewise.
7555         (9_mult6): Likewise.
7556         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
7557         (cortex_a53_sdiv): Likewise.
7558         (cortex_a53_udiv): Likewise.
7559         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
7560         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
7561         (mp626_mult2): Likewise.
7562         (mp626_mult3): Likewise.
7563         (mp626_mult4): Likewise.
7564         * config/arm/fa526.md (526_mult1): Update attribute change.
7565         (526_mult2): Likewise.
7566         * config/arm/arm-generic.md (mult): Update attribute change.
7567         (mult_ldsched_strongarm): Likewise.
7568         (mult_ldsched): Likewise.
7569         (multi_cycle): Likewise.
7570         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
7571         * config/arm/fa606te.md (606te_mult1): Update attribute change.
7572         (606te_mult2): Likewise.
7573         (606te_mult3): Likewise.
7574         (606te_mult4): Likewise.
7575         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
7576         (cortex_a9_mac16): Likewise.
7577         (cortex_a9_multiply): Likewise.
7578         (cortex_a9_mac): Likewise.
7579         (cortex_a9_multiply_long): Likewise.
7580         * config/arm/fa626te.md (626te_mult1): Update attribute change.
7581         (626te_mult2): Likewise.
7582         (626te_mult3): Likewise.
7583         (626te_mult4): Likewise.
7585 2013-06-18  Richard Biener  <rguenther@suse.de>
7587         PR lto/57334
7588         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
7590 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7592         PR target/57609
7593         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
7594         with next_active_insn.
7596 2013-06-18  Alan Modra  <amodra@gmail.com>
7598         * config/rs6000/rs6000.h (enum data_align): New.
7599         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
7600         (DATA_ABI_ALIGNMENT): Define.
7601         (CONSTANT_ALIGNMENT): Correct comment.
7602         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
7603         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
7605 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
7607         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
7608         ATTRIBUTE_UNUSED marking.
7610 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
7612         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
7613         alternative and update.
7614         (aarch64_dup_lanedi): Delete.
7615         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
7616         * config/aarch64/aarch64-simd-builtins.def: Update.
7618 2013-06-17  Richard Biener  <rguenther@suse.de>
7620         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
7621         (lto_input_scc): Declare.
7622         (lto_input_tree_1): Likewise.
7623         (struct lto_stats_d): Add num_tree_bodies_output and
7624         num_pickle_refs_output.
7625         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
7626         (lto_read_tree_1): Split out from ...
7627         (lto_read_tree): ... this.
7628         (lto_input_scc): New function.
7629         (lto_input_tree_1): Split out from ...
7630         (lto_input_tree): ... this.  Handle LTO_tree_scc.
7631         (lto_data_in_create): Create the streamer cache without hashes.
7632         * lto-streamer-out.c (create_output_block): Create the streamer
7633         cache with hashes when not doing WPA.
7634         (lto_write_tree_1): Split out from ...
7635         (lto_write_tree): ... this.
7636         (get_symbol_initial_value): New function.
7637         (lto_output_tree_1): Split out from ...
7638         (lto_output_tree): ... this.  Write trees as series of SCCs
7639         using a DFS walk via DFS_write_tree.
7640         (struct sccs, struct scc_entry): New types.
7641         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
7642         (DFS_write_tree_body): New function.
7643         (DFS_write_tree): Likewise.
7644         (hash_tree): Likewise.
7645         (scc_entry_compare): Likewise.
7646         (hash_scc): Likewise.
7647         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
7648         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
7649         TREE_CHAIN as regular reference.
7650         (streamer_read_integer_cst): Remove.
7651         (streamer_get_pickled_tree): Adjust.
7652         * tree-streamer-out.c (streamer_write_chain): Disable streaming
7653         of DECL_EXTERNALs in BLOCK_VARS for now.
7654         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
7655         reference.
7656         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
7657         Add hash value argument and record that if hashes are recorded
7658         in the cache.
7659         (streamer_tree_cache_insert_1): Adjust.
7660         (streamer_tree_cache_insert): Likewise.
7661         (streamer_tree_cache_insert_at): Rename to ...
7662         (streamer_tree_cache_replace_tree): ... this and adjust.
7663         (streamer_tree_cache_append): Adjust.
7664         (record_common_node): Likewise.
7665         (streamer_tree_cache_create): Add argument whether to
7666         record hash values together with trees.
7667         (streamer_tree_cache_delete): Adjust.
7668         * tree-streamer.h (struct streamer_tree_cache_d): Add
7669         vector of hashes.
7670         (streamer_read_integer_cst): Remove.
7671         (streamer_tree_cache_insert): Adjust.
7672         (streamer_tree_cache_append): Likewise.
7673         (streamer_tree_cache_insert_at): Rename to ...
7674         (streamer_tree_cache_replace_tree): ... this and adjust.
7675         (streamer_tree_cache_create): Add argument whether to record hashes.
7676         (streamer_tree_cache_get): Rename to ...
7677         (streamer_tree_cache_get_tree): ... this.
7678         (streamer_tree_cache_get_hash): New function.
7679         * tree.c (cache_integer_cst): New function.
7680         * tree.h (cache_integer_cst): Declare.
7681         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
7682         * lto-symtab.c (lto_varpool_replace_node): Only release
7683         DECL_INITIAL of non-prevailing decls.
7684         * varpool.c (varpool_remove_initializer): Do not release
7685         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
7687 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
7689         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
7690         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
7691         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
7692         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
7693         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
7694         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
7695         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
7696         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
7697         instead of TARGET_64BIT.
7698         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
7699         Require ISA_HAS_<D>DIV.
7701 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
7703         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
7704         (mips*-*-linux*): Move default with_llsc setting to where other
7705         defaults are set.
7706         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
7707         with_arch block.
7708         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
7709         Likewise.  Remove default with_tune setting.  Move default float
7710         setting to its own block.  Handle with_llsc in the same block as above.
7712 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
7714         PR rtl-optimization/57425
7715         PR rtl-optimization/57569
7716         * alias.c (write_dependence_p): Add new parameters mem_mode,
7717         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
7718         Changed all callers.
7719         (canon_anti_dependence): New function.
7720         * cse.c (check_dependence): Use canon_anti_dependence.
7721         * cselib.c (cselib_invalidate_mem): Likewise.
7722         * rtl.h (canon_anti_dependence): Declare.
7724 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
7726         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
7727         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
7728         ".set mips3" for 64-bit targets.
7730 2013-06-15  Dehao Chen  <dehao@google.com>
7732         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
7733         * gimple-low.c (gimple_check_call_matching_types): Likewise.
7734         (gimple_check_call_args): Likewise.
7735         * value-prof.c (check_ic_target): Likewise.
7736         * ipa-inline.c (early_inliner): Likewise.
7737         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
7738         * cgraph.c (cgraph_create_edge_1): Likewise.
7739         (cgraph_make_edge_direct): Likewise.
7741 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
7743         PR target/57615
7744         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
7745         rs6000_output_move_128bit to handle emitting quad memory
7746         operations.  Set attribute length to 8 bytes.
7748 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
7750         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
7751         New pattern.
7752         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
7753         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
7754         (aarch64_<su>mlsl<mode>): Likewise.
7756 2013-06-14  Mike Stump  <mikestump@comcast.net>
7758         * Makefile.in (TARGET_H): Add insn-codes.h.
7760 2013-06-14  Alan Modra  <amodra@gmail.com>
7762         PR middle-end/57134
7763         PR middle-end/57586
7764         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
7765         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
7766         bitfield expansion when EXPAND_MEMORY.
7767         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
7769 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
7771         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
7772         test for clearing quad memory on 32-bit later.
7774 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
7776         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
7777         (fold_negate_expr): Likewise.
7778         (fold_real_zero_addition_p): Handle vectors.
7779         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
7781 2013-06-14  Alan Modra  <amodra@gmail.com>
7783         * varasm.c (force_const_mem): Revert 2013-06-07 change.
7785 2013-06-13  Jan Hubicka  <jh@suse.cz>
7787         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
7788         Local comdats are not externally visible.
7789         * symtab.c (dump_symtab_base): Dump externally visible.
7790         (verify_symtab_base): Verify back links in the symtab hash.
7792 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
7794         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
7795         CONVERT_EXPR as equal nodes.
7797 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
7799         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
7801 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
7803         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
7804         Generalize to complex and vector.
7805         * tree.c (build_all_ones_cst): New function.
7806         * tree.h (build_all_ones_cst): Declare it.
7808 2013-06-13  Alan Modra  <amodra@gmail.com>
7810         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
7811         * config/rs6000/rs6000.md (signbittf2): New insn.
7812         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
7813         (abstf2_internal, cmptf_internal2): Likewise.
7814         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
7816 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
7817             Pat Haugen  <pthaugen@us.ibm.com>
7818             Peter Bergner  <bergner@vnet.ibm.com>
7820         * config/rs6000/rs6000.c (emit_load_locked): Add support for
7821         power8 byte, half-word, and quad-word atomic instructions.
7822         (emit_store_conditional): Likewise.
7823         (rs6000_expand_atomic_compare_and_swap): Likewise.
7824         (rs6000_expand_atomic_op): Likewise.
7826         * config/rs6000/sync.md (larx): Add new modes for power8.
7827         (stcx): Likewise.
7828         (AINT): New mode iterator to include TImode as well as normal
7829         integer modes on power8.
7830         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
7831         that VSX registers are not considered.  Use AINT mode iterator
7832         instead of INT1 to allow inclusion of quad word atomic operations
7833         on power8.
7834         (load_locked<mode>): Likewise.
7835         (store_conditional<mode>): Likewise.
7836         (atomic_compare_and_swap<mode>): Likewise.
7837         (atomic_exchange<mode>): Likewise.
7838         (atomic_nand<mode>): Likewise.
7839         (atomic_fetch_<fetchop_name><mode>): Likewise.
7840         (atomic_nand_fetch<mode>): Likewise.
7841         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
7842         each type.
7843         (ATOMIC): On power8, add QImode, HImode modes.
7844         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
7845         modes that promote to SImode.
7846         (load_lockedti): Convert TImode arguments to PTImode, so that we
7847         get a guaranteed even/odd register pair.
7848         (load_lockedpti): Likewise.
7849         (store_conditionalti): Likewise.
7850         (store_conditionalpti): Likewise.
7852         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
7853         atomic load/store instructions.
7854         (HSI): Likewise.
7856 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
7858         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
7859         loads.
7860         (insn_count): New attribute, with most cases extracted from...
7861         (length): ...here.  Redefine most cases in terms of insn_count.
7862         (single_insn): Delete.
7863         (can_delay): Use insn_count to check for single instructions.
7864         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
7865         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
7866         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
7867         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
7868         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
7869         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
7870         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
7871         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
7872         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
7873         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
7874         rather than "length".
7875         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
7876         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
7877         Use "insn_count" rather than "length".
7878         * config/mips/mips-dsp.md
7879         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
7880         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
7881         length attributes.
7883 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
7885         PR tree-optimization/57361
7886         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
7888 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
7890         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
7891         to split.
7892         (aarch64_simd_combine<mode>): New instruction expansion.
7893         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
7894         function prototype.
7895         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
7896         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
7898 2013-06-12  Jan Hubicka  <jh@suse.cz>
7900         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
7901         decl has when in streaming stage.
7902         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
7903         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
7905 2013-06-12  Roland Stigge  <stigge@antcom.de>
7907         PR target/57578
7908         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
7910 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7912         PR tree-optimization/57537
7913         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
7914         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
7916 2013-06-12  Richard Biener  <rguenther@suse.de>
7918         * data-streamer.h (streamer_write_char_stream): CSE
7919         obs->current_pointer.
7920         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
7921         streamer_write_char_stream manually and optimize the resulting loop.
7922         (streamer_write_hwi_stream): Likewise.
7924 2013-06-12  Jan Hubicka  <jh@suse.cz>
7926         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
7927         * cgraph.h (varpool_create_empty_node): Declare.
7928         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
7929         duplicated nodes.
7930         * symtab.c (symtab_unregister_node): Be lax about missin entries
7931         in node hash.
7932         (symtab_get_node): Update comment.
7933         * varpool.c (varpool_create_empty_node): Break out from ...
7934         (varpool_node_for_decl): ... here.
7935         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
7937 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
7939         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
7940         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
7941         part.  Use straight-line flow at the end.
7942         <COMPONENT_REF>: Remove superfluous else.
7943         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7945 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
7947         PR target/56564
7948         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7949         target hook even for !TREE_PUBLIC decls.  If no resolution info
7950         is available, return false for common and external decls.
7952 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7954         * config/rl78/constraints.md (U): New constraint.
7955         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7956         valloc attribute.
7958 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
7960         PR target/57589
7961         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7962         to allow returning address to AT_PLATFORM name.
7964 2013-06-11  Jan Hubicka  <jh@suse.cz>
7966         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7967         * cgraph.h (symtab_node_base): Add weakref flag.
7968         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7969         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7970         (output_weakrefs): Use weakref flag.
7971         * fold-const.c (simple_operand_p): Handle WEAK.
7972         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7973         * ipa.c (varpool_externally_visible_p): Drop weakref.
7974         (function_and_variable_visibility): Update comment; fix weakref
7975         sanity checks; do not clear DECL_WEAK on them.
7976         * lto-cgraph.c (lto_output_node): update.
7977         (lto_output_varpool_node): Update.
7978         (input_overwrite_node): Update.
7979         (input_node): Update.
7980         (input_varpool_node): Update.
7981         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7982         (lto_symtab_merge_symbols): Add sanity check.
7983         (lto_symtab_prevailing_decl): Do not special case weakrefs.
7984         * passes.c (rest_of_decl_compilation): Set static flag, too.
7985         * symtab.c (dump_symtab_base): Dump weakref.
7986         (verify_symtab_base): Sanity check weakrefs.
7987         (symtab_make_decl_local): Remove duplicated code.
7988         (symtab_alias_ultimate_target): Simplify.
7989         * varpool.c (varpool_create_variable_alias): Set weakref flag.
7991 2013-06-11  DJ Delorie  <dj@redhat.com>
7993         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7994         (rl78_unwind_word_mode): New.
7996 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
7998         * final.c (debug_prefix_maps): Make static.
8000 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
8002         * function.c (initial_trampoline): Remove stray copy.
8004 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
8006         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
8008 2013-06-11  Martin Jambor  <mjambor@suse.cz>
8010         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
8011         within bounds at the beginning of the function.
8013 2013-06-11  Alan Modra  <amodra@gmail.com>
8015         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
8016         reporting.
8017         (get_named_section): Don't NULL !DECL_P decl.
8019 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
8021         * doc/invoke.texi (core-avx2): Document.
8022         (slm): Likewise.
8023         (atom): Updated with MOVBE.
8025 2013-06-11  Richard Biener  <rguenther@suse.de>
8027         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
8029 2013-06-11  Anton Blanchard  <anton@samba.org>
8031         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
8032         correct shift value in little-endian mode.
8034 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
8036         PR target/56564
8037         * varasm.c (get_variable_align): Move #endif to the right place.
8039 2013-06-10  Cary Coutant  <ccoutant@google.com>
8041         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
8042         for hash so that hash table traversal order is deterministic.
8044 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
8045             Pat Haugen  <pthaugen@us.ibm.com>
8046             Peter Bergner  <bergner@vnet.ibm.com>
8048         * config/rs6000/vector.md (GPR move splitter): Do not split moves
8049         of vectors in GPRS if they are direct moves or quad word load or
8050         store moves.
8052         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
8053         declaration.
8054         (direct_move_p): Likewise.
8055         (quad_load_store_p): Likewise.
8057         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
8058         classes into bins based on the physical register type.
8059         (reg_class_to_reg_type): Likewise.
8060         (IS_STD_REG_TYPE): Likewise.
8061         (IS_FP_VECT_REG_TYPE): Likewise.
8062         (reload_fpr_gpr): Arrays to determine what insn to use if we can
8063         use direct move instructions.
8064         (reload_gpr_vsx): Likewise.
8065         (reload_vsx_gpr): Likewise.
8066         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
8067         information that is a simplification of register classes.  Also
8068         precalculate direct move reload helpers.
8069         (direct_move_p): New function to return true if the operation can
8070         be done as a direct move instruciton.
8071         (quad_load_store_p): New function to return true if the operation
8072         is a quad memory operation.
8073         (rs6000_legitimize_address): If quad memory, only allow register
8074         indirect for TImode addresses.
8075         (rs6000_legitimate_address_p): Likewise.
8076         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
8077         (rs6000_reload_register_type): Likewise.
8078         (register_to_reg_type): Return register type.
8079         (rs6000_secondary_reload_simple_move): New helper function for
8080         secondary reload and secondary memory needed to identify anything
8081         that is a simple move, and does not need reloading.
8082         (rs6000_secondary_reload_direct_move): New helper function for
8083         secondary reload to identify cases that can be done with several
8084         instructions via the direct move instructions.
8085         (rs6000_secondary_reload_move): New helper function for secondary
8086         reload to identify moves between register types that can be done.
8087         (rs6000_secondary_reload): Add support for quad memory operations
8088         and for direct move.
8089         (rs6000_secondary_memory_needed): Likewise.
8090         (rs6000_debug_secondary_memory_needed): Change argument names.
8091         (rs6000_output_move_128bit): New function to return the move to
8092         use for 128-bit moves, including knowing about the various
8093         limitations of quad memory operations.
8095         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
8096         memory operations.  call rs6000_output_move_128bit for the actual
8097         instruciton(s) to generate.
8098         (vsx_movti_64bit): Likewise.
8100         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
8101         (UNSPEC_P8V_MTVSRWZ): Likewise.
8102         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
8103         (UNSPEC_P8V_MTVSRD): Likewise.
8104         (UNSPEC_P8V_XXPERMDI): Likewise.
8105         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
8106         (UNSPEC_FUSION_GPR): Likewise.
8107         (FMOVE128_GPR): New iterator for direct move.
8108         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
8109         (f32_sv): Likewise.
8110         (f32_dm): Likewise.
8111         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
8112         loads and direct move instructions.
8113         (zero_extendsidi2_lfiwzx): Likewise.
8114         (extendsidi2_lfiwax): Likewise.
8115         (extendsidi2_nocell): Likewise.
8116         (floatsi<mode>2_lfiwax): Likewise.
8117         (lfiwax): Likewise.
8118         (floatunssi<mode>2_lfiwzx): Likewise.
8119         (lfiwzx): Likewise.
8120         (fix_trunc<mode>_stfiwx): Likewise.
8121         (fixuns_trunc<mode>_stfiwx): Likewise.
8122         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
8123         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
8124         (parity<mode>2_cmpb): Set length/type attr.
8125         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
8126         for 'mr.' to fast_compare.
8127         (bpermd_<mode>): Change type attr to popcnt.
8128         (p8_fmrgow_<mode>): New insns for power8 direct move support.
8129         (p8_mtvsrwz_1): Likewise.
8130         (p8_mtvsrwz_2): Likewise.
8131         (reload_fpr_from_gpr<mode>): Likewise.
8132         (p8_mtvsrd_1): Likewise.
8133         (p8_mtvsrd_2): Likewise.
8134         (p8_xxpermdi_<mode>): Likewise.
8135         (reload_vsx_from_gpr<mode>): Likewise.
8136         (reload_vsx_from_gprsf): Likewise.
8137         (p8_mfvsrd_3_<mode>): LIkewise.
8138         (reload_gpr_from_vsx<mode>): Likewise.
8139         (reload_gpr_from_vsxsf): Likewise.
8140         (p8_mfvsrd_4_disf): Likewise.
8141         (multi-word GPR splits): Do not split direct moves or quad memory
8142         operations.
8144 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
8146         * tree-into-ssa.c (interesting_blocks): Make static.
8148 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
8150         PR target/56564
8151         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
8152         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
8153         Use DATA_ABI_ALIGNMENT for that case instead if defined.
8154         (get_variable_align): New function.
8155         (get_variable_section, emit_bss, emit_common,
8156         assemble_variable_contents, place_block_symbol): Use
8157         get_variable_align instead of DECL_ALIGN.
8158         (assemble_noswitch_variable): Add align argument, use it
8159         instead of DECL_ALIGN.
8160         (assemble_variable): Adjust caller.  Use get_variable_align
8161         instead of DECL_ALIGN.
8162         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
8163         caller.
8164         (DATA_ABI_ALIGNMENT): Define.
8165         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
8166         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
8167         opt is false, only return the psABI mandated alignment increase.
8168         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
8169         (DATA_ABI_ALIGNMENT): ... this.
8170         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
8171         (DATA_ABI_ALIGNMENT): ... this.
8172         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
8173         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
8174         (DATA_ABI_ALIGNMENT): ... this.
8175         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
8176         * doc/tm.texi: Regenerated.
8178 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
8180         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
8181         cmp_code to construct REG_EQUAL note.
8183 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
8185         PR target/57568
8186         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
8187         that operands[2] doesn't overlap with operands[0].
8189 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
8190             Jan Hubicka  <jh@suse.cz>
8192         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
8193         hack to mark symbols as used.
8195 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
8197         PR rtl-optimization/57559
8198         * lra-constraints.c (process_alt_operands): Don't discourage
8199         memory with known offset for offsetable memory constraint.
8200         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
8202 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
8204         * varasm.c (struct oc_local_state): Reorder fields.
8205         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
8206         and adjust accordingly.
8207         (output_constructor): Reorder initialization code and adjust call to
8208         output_constructor_bitfield.
8210 2013-06-07  Jan Hubicka  <jh@suse.cz>
8212         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
8214 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
8216         * tree-object-size.c (unknown): Make const.
8218 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8220         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
8221         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
8222         for last alternative in the cpu_facility attribute.
8224 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8226         PR target/56315
8227         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
8228         (xordi3): Change operand 2 constraint to arm_xordi_operand.
8229         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
8230         * config/arm/constraints.md (Dg): New constraint.
8231         * config/arm/neon.md (xordi3_neon): Remove.
8232         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
8233         * config/arm/predicates.md (arm_xordi_operand): New predicate.
8235 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8237         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
8238         Clean up alternatives.
8240 2013-06-07  Alan Modra  <amodra@gmail.com>
8242         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
8243         va_list_gpr_size.
8245 2013-06-07  Alan Modra  <amodra@gmail.com>
8247         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
8249 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8251         * config/arm/constraints.md (Df): New constraint.
8252         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
8253         Correct length attribute for last two alternatives.
8255 2013-06-07  Alan Modra  <amodra@gmail.com>
8257         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8258         override user -mfp-in-toc.
8259         (offsettable_ok_by_alignment): Consider just the current access
8260         rather than the whole object, unless BLKmode.  Handle
8261         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
8262         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
8263         for -mcmodel=medium.
8264         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
8265         override user -mfp-in-toc or -msum-in-toc.  Default to
8266         -mno-fp-in-toc for -mcmodel=medium.
8268 2013-06-06  DJ Delorie  <dj@redhat.com>
8270         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
8271         TARGET_VALID_POINTER_MODE.
8273 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
8274             Pat Haugen  <pthaugen@us.ibm.com>
8275             Peter Bergner  <bergner@vnet.ibm.com>
8277         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
8278         Document new power8 builtins.
8280         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
8281         condition code register, to allow 128-bit logical operations to be
8282         done in the VSX or GPR registers.
8283         (nor<mode>3): Use the canonical form for nor.
8284         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
8285         vclz*, and vpopcnt* vector instructions.
8286         (nand<mode>3): Likewise.
8287         (orc<mode>3): Likewise.
8288         (clz<mode>2): LIkewise.
8289         (popcount<mode>2): Likewise.
8291         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
8292         that only the GPRs are recognized.
8294         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8295         support for new power8 builtins.
8297         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
8298         builtin functions.
8299         (xscvdpspn): Likewise.
8300         (vclz): Likewise.
8301         (vclzb): Likewise.
8302         (vclzh): Likewise.
8303         (vclzw): Likewise.
8304         (vclzd): Likewise.
8305         (vpopcnt): Likewise.
8306         (vpopcntb): Likewise.
8307         (vpopcnth): Likewise.
8308         (vpopcntw): Likewise.
8309         (vpopcntd): Likewise.
8310         (vgbbd): Likewise.
8311         (vmrgew): Likewise.
8312         (vmrgow): Likewise.
8313         (eqv): Likewise.
8314         (eqv_v16qi3): Likewise.
8315         (eqv_v8hi3): Likewise.
8316         (eqv_v4si3): Likewise.
8317         (eqv_v2di3): Likewise.
8318         (eqv_v4sf3): Likewise.
8319         (eqv_v2df3): Likewise.
8320         (nand): Likewise.
8321         (nand_v16qi3): Likewise.
8322         (nand_v8hi3): Likewise.
8323         (nand_v4si3): Likewise.
8324         (nand_v2di3): Likewise.
8325         (nand_v4sf3): Likewise.
8326         (nand_v2df3): Likewise.
8327         (orc): Likewise.
8328         (orc_v16qi3): Likewise.
8329         (orc_v8hi3): Likewise.
8330         (orc_v4si3): Likewise.
8331         (orc_v2di3): Likewise.
8332         (orc_v4sf3): Likewise.
8333         (orc_v2df3): Likewise.
8335         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
8336         allow power8 quad mode in 64-bit.
8337         (rs6000_builtin_vectorized_function): Add support to vectorize
8338         ISA 2.07 count leading zeros, population count builtins.
8339         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
8340         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
8341         (builtin_function_type): Add vgbbd builtin function which takes an
8342         unsigned argument.
8343         (altivec_expand_vec_perm_const): Add support for new power8 merge
8344         instructions.
8346         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
8347         that does not include TImdoe for use with 32-bit.
8348         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
8349         instructions.
8350         (UNSPEC_VSX_CVDPSPN): Likewise.
8351         (vsx_xscvdpspn): Likewise.
8352         (vsx_xscvspdpn): Likewise.
8353         (vsx_xscvdpspn_scalar): Likewise.
8354         (vsx_xscvspdpn_directmove): Likewise.
8355         (vsx_and<mode>3): Split logical operations into 32-bit and
8356         64-bit. Add support to do logical operations on TImode as well as
8357         VSX vector types.  Allow logical operations to be done in either
8358         VSX registers or in general purpose registers in 64-bit mode.  Add
8359         splitters if GPRs were used. For AND, add clobber of CCmode to
8360         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
8361         encoding.
8362         (vsx_and<mode>3_32bit): Likewise.
8363         (vsx_and<mode>3_64bit): Likewise.
8364         (vsx_ior<mode>3): Likewise.
8365         (vsx_ior<mode>3_32bit): Likewise.
8366         (vsx_ior<mode>3_64bit): Likewise.
8367         (vsx_xor<mode>3): Likewise.
8368         (vsx_xor<mode>3_32bit): Likewise.
8369         (vsx_xor<mode>3_64bit): Likewise.
8370         (vsx_one_cmpl<mode>2): Likewise.
8371         (vsx_one_cmpl<mode>2_32bit): Likewise.
8372         (vsx_one_cmpl<mode>2_64bit): Likewise.
8373         (vsx_nor<mode>3): Likewise.
8374         (vsx_nor<mode>3_32bit): Likewise.
8375         (vsx_nor<mode>3_64bit): Likewise.
8376         (vsx_andc<mode>3): Likewise.
8377         (vsx_andc<mode>3_32bit): Likewise.
8378         (vsx_andc<mode>3_64bit): Likewise.
8379         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
8380         and xxlorc instructions.
8381         (vsx_eqv<mode>3_64bit): Likewise.
8382         (vsx_nand<mode>3_32bit): Likewise.
8383         (vsx_nand<mode>3_64bit): Likewise.
8384         (vsx_orc<mode>3_32bit): Likewise.
8385         (vsx_orc<mode>3_64bit): Likewise.
8387         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
8389         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
8390         instruction.
8391         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
8392         (p8_vmrgow): Likewise.
8393         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
8394         GPRs to be split under VSX.
8395         (p8v_clz<mode>2): Add power8 count leading zero support.
8396         (p8v_popcount<mode>2): Add power8 population count support.
8397         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
8398         support.
8400         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
8401         instruction.
8403         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
8404         builtin functions.
8405         (vec_nand): Likewise.
8406         (vec_vclz): Likewise.
8407         (vec_vclzb): Likewise.
8408         (vec_vclzd): Likewise.
8409         (vec_vclzh): Likewise.
8410         (vec_vclzw): Likewise.
8411         (vec_vgbbd): Likewise.
8412         (vec_vmrgew): Likewise.
8413         (vec_vmrgow): Likewise.
8414         (vec_vpopcnt): Likewise.
8415         (vec_vpopcntb): Likewise.
8416         (vec_vpopcntd): Likewise.
8417         (vec_vpopcnth): Likewise.
8418         (vec_vpopcntw): Likewise.
8420 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
8422         PR rtl-optimization/57468
8423         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
8424         spilled pseudos.
8426 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
8428         PR rtl-optimization/57459
8429         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
8430         type when setting live regs.
8432 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
8434         * config/s390/s390.opt (mlra): New option.
8435         * config/s390/s390.c (s390_decompose_address): Check displacement
8436         for all registers for LRA.
8437         (s390_secondary_reload): Don't used secondary reloads for LRA.
8438         (s390_lra_p): New function.
8439         (TARGET_LRA_P): Define.
8440         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
8441         of attribute cpu_facility to zarch for the last alternative.
8442         (*cmpmem_short): Ditto.
8444 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
8446         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
8447         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
8448         (arm_expand_prologue): Likewise.
8450 2013-06-06  Teresa Johnson  <tejohnson@google.com>
8452         PR c++/53743
8453         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
8454         as this is now done by redirect_edge_and_branch_force.
8455         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
8456         barriers, and fix interaction with splitting.
8457         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
8458         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
8459         reflect changes made in the routine.
8460         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
8461         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
8462         since this is called in cfglayout mode, and replace partition fixup
8463         with assert as that is now done by force_nonfallthru_and_redirect.
8464         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
8465         already be marked with region crossing note.
8466         (insert_section_boundary_note): Make non-static, gate on flag
8467         has_bb_partition, rewrite to also check for multiple partitions.
8468         (rest_of_handle_reorder_blocks): Remove call to
8469         insert_section_boundary_note, now done later during free_cfg.
8470         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
8471         * bb-reorder.h (insert_section_boundary_note): Declare.
8472         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
8473         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
8474         invoke insert_section_boundary_note.
8475         (try_redirect_by_replacing_jump): Remove unnecessary
8476         check for region crossing note.
8477         (fixup_partition_crossing): New function.
8478         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
8479         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
8480         in non-cfglayout mode.
8481         (force_nonfallthru_and_redirect): Fixup partition boundaries,
8482         remove old code that tried to do this. Emit barrier correctly
8483         when we are in cfglayout mode.
8484         (last_bb_in_partition): New function.
8485         (rtl_split_edge): Correctly fixup partition boundaries.
8486         (commit_one_edge_insertion): Remove old code that tried to
8487         fixup region crossing edge since this is now handled in
8488         split_block, and set up insertion point correctly since
8489         block may now end in a jump.
8490         (verify_hot_cold_block_grouping): Guard against checking when not in
8491         linearized RTL mode.
8492         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
8493         notes.
8494         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
8495         rtl_verify_flow_info, so not called in cfglayout mode.
8496         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
8497         (fixup_reorder_chain): Remove old code that attempted to fixup region
8498         crossing note as this is now handled in force_nonfallthru_and_redirect.
8499         (duplicate_insn_chain): Don't duplicate switch section notes.
8500         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
8501         note.
8502         * basic-block.h (emit_barrier_after_bb): Declare.
8504 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8506         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
8507         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
8508         arm_usatsihi): Adjust alternatives for arm_restrict_it.
8510 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8512         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
8513         where appropriate.
8514         * config/arm/ldmstm.md: Regenerate.
8516 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8518         * config/arm/sync.md (atomic_loaddi_1):
8519         Disable predication for arm_restrict_it.
8520         (arm_load_exclusive<mode>): Likewise.
8521         (arm_load_exclusivesi): Likewise.
8522         (arm_load_exclusivedi): Likewise.
8523         (arm_load_acquire_exclusive<mode>): Likewise.
8524         (arm_load_acquire_exclusivesi): Likewise.
8525         (arm_load_acquire_exclusivedi): Likewise.
8526         (arm_store_exclusive<mode>): Likewise.
8527         (arm_store_exclusive<mode>): Likewise.
8528         (arm_store_release_exclusivedi): Likewise.
8529         (arm_store_release_exclusive<mode>): Likewise.
8531 2013-06-06  Richard Biener  <rguenther@suse.de>
8533         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
8534         after LTO_null.
8535         (lto_tag_is_tree_code_p): Adjust.
8536         (lto_tag_is_gimple_code_p): Likewise.
8537         (lto_gimple_code_to_tag): Likewise.
8538         (lto_tag_to_gimple_code): Likewise.
8539         (lto_tree_code_to_tag): Likewise.
8540         (lto_tag_to_tree_code): Likewise.
8541         * data-streamer.h (streamer_write_hwi_in_range): Use
8542         uhwi streaming to stream the normalized range.
8543         (streamer_read_hwi_in_range): Likewise.
8545 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8547         * config/arm/arm.md (enabled_for_depr_it): New attribute.
8548         (predicable_short_it): Likewise.
8549         (predicated): Likewise.
8550         (enabled): Handle above.
8551         (define_cond_exec): Set predicated attribute to yes.
8553 2013-06-05  Mike Stump  <mikestump@comcast.net>
8555         * gdbinit.in (__FUNCTION__): Add.
8557 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
8559         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
8560         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
8562 2013-06-05  Jan Hubicka  <jh@suse.cz>
8564         * varasm.c (mark_decl_referenced): Revert the removal until targets
8565         are fixed.
8567 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
8569         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
8570         instead of mark_decl_referenced.
8572 2013-06-05  Jan Hubicka  <jh@suse.cz>
8574         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
8575         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
8576         and symtab_used_from_object_file_p.
8577         (cgraph_make_node_local_1): Clear forced_by_abi.
8578         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
8579         * cgraph.h (symtab_node_base): Add forced_by_abi.
8580         (decide_is_variable_needed): Remove.
8581         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
8582         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
8583         (decide_is_symbol_needed): ... this one; handle symbols in general;
8584         always analyze virtuals; honnor forced_by_abi.
8585         (cgraph_finalize_function): Update.
8586         (varpool_finalize_decl): Update.
8587         (symbol_defined_and_needed): Remove.
8588         (analyze_functions): Update.
8589         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8590         output_refs, input_overwrite_node): Handle forced_by_abi.
8591         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
8592         (address_taken_from_non_vtable_p): ... this one.
8593         (comdat_can_be_unshared_p_1): New function.
8594         (cgraph_comdat_can_be_unshared_p): Rename to ...
8595         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
8596         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
8597         (function_and_variable_visibility): Clear forced_by_abi as needed.
8598         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
8599         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
8600         * symtab.c (dump_symtab_base): Dump forced_by_abi.
8601         * varpool.c (decide_is_variable_needed): Remove.
8603 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8605         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
8606         (arm_option_override): Override arm_restrict_it where appropriate.
8607         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
8608         * config/arm/arm.opt (mrestrict-it): New command-line option.
8609         * doc/invoke.texi: Document -mrestrict-it.
8611 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
8613         * tsan.c (tsan_atomic_table): Make const.
8615 2013-06-05  Richard Biener  <rguenther@suse.de>
8617         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
8618         index associated with the tree we are supposed to replace.
8619         * tree-streamer-out.c (pack_ts_base_value_fields): Output
8620         TREE_ASM_WRITTEN as zero for everything but SSA names.
8622 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
8624         * tree-ssa-structalias.c (call_stmt_vars): Make static.
8626 2013-06-04  Jan Hubicka  <jh@suse.cz>
8628         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
8629         (input_node, input_varpool_node): Handle correctly external same
8630         body aliases.
8631         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
8632         nodes at ltrans stage.
8634 2013-06-04  Jan Hubicka  <jh@suse.cz>
8636         * ipa-inline.c (update_caller_keys): Fix availability test.
8637         (update_callee_keys): Likewise.
8638         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
8639         to follow ELF standard.
8641 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
8643         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
8644         (mips64r5900el-*-elf*): New configurations.
8645         * config/mips/mips-cpus.def (r5900): New processor.
8646         * config/mips/mips-tables.opt: Regenerate.
8647         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
8648         (mips_issue_rate): Handle PROCESSOR_R5900.
8649         (mips_reorg_process_insns): Force reorder mode for the R5900.
8650         * config/mips/mips.h (TARGET_MIPS5900): Define.
8651         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
8652         TARGET_MIPS5900.
8653         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
8654         TARGET_MIPS5900.
8655         * config/mips/mips.md (processor): Add r5900.
8656         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
8658 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
8660         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
8661         into function to generate MOVI instruction.
8662         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
8663         (aarch64_preferred_simd_mode): Turn into wrapper.
8664         (aarch64_output_scalar_simd_mov_immediate): New function.
8665         * config/aarch64/aarch64-protos.h: Add prototype for above.
8667 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
8669         * config/aarch64/aarch64.c (simd_immediate_info): Remove
8670         element_char member.
8671         (sizetochar): Return signed char.
8672         (aarch64_simd_valid_immediate): Remove elchar and other
8673         unnecessary variables.
8674         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
8675         Calculate element_char as required.
8676         * config/aarch64/aarch64-protos.h: Update and move prototype
8677         for aarch64_output_simd_mov_immediate.
8678         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8679         Update arguments.
8681 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
8683         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
8684         information completed by aarch64_simd_valid_immediate.
8685         (aarch64_legitimate_constant_p): Update arguments.
8686         (aarch64_simd_valid_immediate): Work with struct rather than many
8687         pointers.
8688         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
8689         (aarch64_simd_make_constant): Update arguments.
8690         (aarch64_output_simd_mov_immediate): Work with struct rather than
8691         many pointers.  Output immediate directly rather than as operand.
8692         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
8693         Update prototype.
8694         * config/aarch64/constraints.md (Dn): Update arguments.
8696 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
8698         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
8699         longer static.
8700         (aarch64_simd_immediate_valid_for_move): Remove.
8701         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
8702         (aarch64_simd_make_constant): Update call.
8703         (aarch64_output_simd_mov_immediate): Update call.
8704         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
8705         Add prototype.
8706         * config/aarch64/constraints.md (Dn): Update call.
8708 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
8710         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
8711         return type to bool for prototype.
8712         (aarch64_legitimate_constant_p): Check for true instead of not -1.
8713         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
8714         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
8716 2013-06-04  Catherine Moore  <clm@codesourcery.com>
8718         * config/mips/mips.opt (meva): New.
8719         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
8720         (ASM_SPEC): Handle -meva.
8721         * doc/invoke.texi (meva):  Document.
8723 2013-06-04  Alan Modra  <amodra@gmail.com>
8725         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
8726         constant output.
8728 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8730         * rtl.def: Add extra fourth optional field to define_cond_exec.
8731         * gensupport.c (process_one_cond_exec): Process attributes from
8732         define_cond_exec.
8733         * doc/md.texi: Document fourth field in define_cond_exec.
8735 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
8737         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
8738         out the processing order as in store_bit_field_1.
8740 2013-06-04  Jan Hubicka  <jh@suse.cz>
8742         PR middle-end/57500
8743         * cgraphunit.c (cgraph_process_same_body_aliases): Create
8744         non-VAR_DECL node if it does not exist yet.
8746 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
8748         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
8749         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
8750         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
8751         target_cpu_default setting.
8753 2013-06-03  Teresa Johnson  <tejohnson@google.com>
8755         * dumpfile.c (opt_info_switch_p): Change -fopt-info
8756         default to -fopt-info=optimized instead of all.
8757         * doc/invoke.texi: Ditto.
8758         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
8759         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
8760         (execute_vect_slp): Emit BB vectorization success under
8761         MSG_OPTIMIZED_LOCATIONS.
8762         * tree-vect-slp.c (vect_slp_transform_bb): Change
8763         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
8764         * tree-vect-loop.c (vect_transform_loop): Ditto.
8766 2013-06-03  Jason Merrill  <jason@redhat.com>
8768         PR c++/57415
8769         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8770         Use TARGET_EXPR for C++.
8772 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
8774         PR rtl-optimization/57268
8775         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
8776         if DEBUG_INSN_P (insn).
8778         Reapply
8779         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
8781         PR rtl-optimization/57268
8782         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8783         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8785 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8787         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
8788         (ix86_avoid_lea_for_addr): Likewise.
8789         (exact_dependency_1): Likewise.
8790         (ix86_adjust_cost): Likewise.
8791         (swap_top_of_ready_list): Fix formatting and !reload_completed check
8792         removed.
8793         (do_reorder_for_imul): Fix typo, formatting and
8794         !reload_completed check removed.
8795         (ix86_sched_reorder): Fix typo and formatting.
8796         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
8797         list.
8799 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
8801         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
8803 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
8805         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
8806         <STRING_CST>: Likewise.
8807         <VECTOR_CST>: Likewise.
8809 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
8810             Mikael Morin  <mikael@gcc.gnu.org>
8812         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
8813         * config.in: Regenerated.
8814         * configure: Regenerated.
8816 2013-06-01  Jan Hubicka  <jh@suse.cz>
8818         PR middle-end/57366
8819         * cgraphunit.c (compile): When weakref is not supported,
8820         set up transparent aliases before final output pass.
8821         * varasm.c (assemble_alias): Do not try to do it here.
8823 2013-06-01  Jan Hubicka  <jh@suse.cz>
8825         PR middle-end/57467
8826         * passes.c (for_per_function): Skip unanalyzed functions.
8828 2013-06-01  Jan Hubicka  <jh@suse.cz>
8830         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
8831         (lto_symtab_merge_symbols_1): ... this one.
8832         (lto_symtab_merge_cgraph_nodes): Rename to ...
8833         (lto_symtab_merge_symbols): ... this one; simplify.
8834         * cgraph.c (same_body_aliases_done): Rename to ...
8835         (cpp_implicit_aliases_done): ... this one.
8836         (cgraph_create_function_alias): Update.
8837         (cgraph_same_body_alias): Update.
8838         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
8839         (verify_edge_corresponds_to_fndecl): Simplify.
8840         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
8841         (cgraph_node): Remove same_body_alias.
8842         (varpool_node): Remove alias_of and extra_name_alias.
8843         (same_body_aliases_done): Rename to ..
8844         (cpp_implicit_aliases_done): ... this one.
8845         (symtab_alias_ultimate_target): Add default parameter.
8846         (symtab_resolve_alias): New function.
8847         (fixup_same_cpp_alias_visibility): Declare.
8848         (cgraph_function_node): Add default parameter.
8849         (cgraph_node_asm_name): Likewise.
8850         (cgraph_function_or_thunk_node): Add default parameter; do
8851         not ICE when it is NULL.
8852         (varpool_variable_node): Likewise.
8853         * tree-emutls.c (create_emultls_var): Update.
8854         (ipa_lower_emutls): Update.
8855         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
8856         (cgraph_reset_node): Reset alias info.
8857         (cgraph_finalize_function): Update.
8858         (fixup_same_cpp_alias_visibility): Move to symtab.c.
8859         (analyze_function): Simplify.
8860         (cgraph_process_same_body_aliases): Simplify.
8861         (analyze_functions): Fixup same body aliases.
8862         (handle_alias_pairs): Simplify.
8863         (assemble_thunk): Update.
8864         (assemble_thunks_and_aliases): Update.
8865         (output_weakrefs): Rewrite.
8866         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
8867         (lto_output_varpool_node): Likewise.
8868         (compute_ltrans_boundary): Remve assert.
8869         (get_alias_symbol): New functoin.
8870         (input_node): Rewrite alias handling.
8871         (input_varpool_node): Likewise.
8872         * ipa-pure-const.c (propagate_pure_const): Fix formating.
8873         * ipa.c (process_references): Handle weakrefs correctly.
8874         (symtab_remove_unreachable_nodes): Likewise.
8875         * trans-mem.c (get_cg_data): Update.
8876         (ipa_tm_create_version_alias): Update.
8877         (ipa_tm_execute): Update.
8878         * symtab.c (dump_symtab_base): Dump aliases.
8879         (verify_symtab_base): Verify aliases.
8880         (symtab_node_availability): New function.
8881         (symtab_alias_ultimate_target): Simplify.
8882         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
8883         handle all the fixup cases.
8884         (symtab_resolve_alias): New function.
8885         * passes.c (ipa_write_summaries): Handle weakrefs.
8886         * varpool.c (varpool_analyze_node): Simplify.
8887         (assemble_aliases): Update.
8888         (varpool_create_variable_alias): Simplify.
8889         (varpool_extra_name_alias): Simplify.
8890         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
8891         (lto_symtab_merge_symbols): ... this one.
8893 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
8895         Revert
8896         PR rtl-optimization/57268
8897         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8898         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8900 2013-06-01  Tobias Burnus  <burnus@net-b.de>
8902         Partially reverted:
8903         2013-05-31  Tobias Burnus  <burnus@net-b.de>
8905         PR middle-end/57073
8906         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
8907         further up.
8909 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
8911         PR rtl-optimization/57268
8912         * sched-deps.c (sched_analyze_2): Flush dependence lists if
8913         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8915 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8917         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
8918         unordered comparison operators when -fno-trapping-math is in effect
8919         on the e500.
8920         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
8921         and implement unordered comparison operators properly on the e500.
8923 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
8925         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
8926         for constant scalar integers.
8927         (simplify_relational_operation_1): Likewise.
8929 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
8931         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
8932         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
8933         Fix comment.
8935 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
8936             Igor Zamyatin  <igor.zamyatin@intel.com>
8938         Silvermont (SLM) architecture performance tuning.
8939         * config/i386/i386.h (enum ix86_tune_indices): Add
8940         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
8941         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8943         * config/i386/i386.c (initial_ix86_tune_features)
8944         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8945         (ix86_lea_outperforms): Handle Silvermont tuning.
8946         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8947         call.
8948         (ix86_use_lea_for_mov): Likewise.
8949         (ix86_avoid_lea_for_addr): Likewise.
8950         (ix86_lea_for_add_ok): Likewise.
8951         (exact_dependency_1): New function.
8952         (exact_store_load_dependency): Likewise.
8953         (ix86_adjust_cost): Handle Silvermont tuning.
8954         (do_reoder_for_imul): Likewise.
8955         (swap_top_of_ready_list): New function.
8956         (ix86_sched_reorder): Changed to handle Silvermont tuning.
8958         * config/i386/i386.md (peepholes that split memory operand in fp
8959         converts): New.
8961 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8963         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8964         Remove un-necessary braces.
8966 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8968         * config/aarch64/aarch64.c (aarch64_classify_symbol):
8969         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8971 2013-05-31  Tobias Burnus  <burnus@net-b.de>
8973         PR middle-end/57073
8974         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8976 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8978         PR target/56315
8979         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8980         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8981         * config/arm/neon.md (iordi3_neon): Remove.
8982         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8983         * config/arm/predicates.md (imm_for_neon_logic_operand):
8984         Move to earlier in the file.
8985         (neon_logic_op2): Likewise.
8986         (arm_iordi_operand_neon): New predicate.
8988 2013-05-31  Richard Biener  <rguenther@suse.de>
8990         PR tree-optimization/57478
8991         PR tree-optimization/57453
8992         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8993         are life as well.
8995 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8997         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8998         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
9000 2013-05-30  Tobias Burnus  <burnus@net-b.de>
9001             Thomas Koenig  <tkoenig@gcc.gnu.org>
9003         PR middle-end/57073
9004         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
9005         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
9007 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
9009         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
9011 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
9013         * target.def (register_usage_leveling_p): New hook.
9014         * targhooks.c (default_register_usage_leveling_p): New.
9015         * targhooks.h (default_register_usage_leveling_p): New prototype.
9016         * lra-assigns.c (register_usage_leveling_p): Use the hook.
9017         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
9018         * doc/tm.texi: Update.
9019         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
9021 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
9023         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
9024         (*insv_reg<mode>): New define_insn.
9026 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
9028         PR rtl-optimization/57439
9029         * postreload.c (move2add_valid_value_p): Check that we have
9030         a zero subreg_regno_offset when accessing the register in
9031         the requested mode.
9033 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
9034             Igor Zamyatin  <igor.zamyatin@intel.com>
9036         Silvermont (SLM) architecture pipeline model, tuning and
9037         insn selection.
9038         * config.gcc: Add slm config options and target.
9040         * config/i386/slm.md: New.
9042         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
9044         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
9045         PROCESSOR_SLM.
9046         (ix86_target_macros_internal): Likewise.
9048         * gcc/config/i386/i386.c (slm_cost): New cost.
9049         (m_SLM): New macro flag.
9050         (initial_ix86_tune_features): Set m_SLM.
9051         (x86_accumulate_outgoing_args): Likewise.
9052         (x86_arch_always_fancy_math_387): Likewise.
9053         (processor_target_table): Add slm cost.
9054         (cpu_names): Add slm cpu name.
9055         (x86_option_override_internal): Set SLM ISA.
9056         (ix86_issue_rate): New case PROCESSOR_SLM.
9057         (ia32_multipass_dfa_lookahead): Likewise.
9058         (fold_builtin_cpu): Add slm.
9060         * config/i386/i386.h (TARGET_SLM): New target macro.
9061         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
9062         (processor_type): Add PROCESSOR_SLM.
9064         * config/i386/i386.md (cpu): Add new value "slm".
9065         (slm.md): Include slm.md.
9067 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
9068             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9070         * config/arm/arm-protos.h: Add and update function protos.
9071         * config/arm/arm.c (use_simple_return_p): New added.
9072         (thumb2_expand_return): Check simple_return flag.
9073         * config/arm/arm.md: Add simple_return and conditional simple_return.
9074         * config/arm/iterators.md: Add iterator for return and simple_return.
9076 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9078         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
9079         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
9080         (arm_emit_vfp_multi_reg_pop): Likewise.
9081         (thumb2_emit_ldrd_pop): Likewise.
9082         (arm_expand_epilogue): Add misc REG_CFA notes.
9083         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
9085 2013-05-29  Lawrence Crowl  <crowl@google.com>
9087         * config/arm/t-arm: Update for below.
9089         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
9090         Change type to hash_table.  Update dependent calls and types.
9092         * config/i386/t-cygming: Update for below.
9094         * config/i386/t-interix: Update for below.
9096         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
9097         Change type to hash_table.  Update dependent calls and types.
9098         (i386_find_on_wrapper_list::wrappers): Likewise.
9100         * config/ia64/t-ia64: Update for below.
9102         * config/ia64/ia64.c (bundle_state_table):
9103         Change type to hash_table.  Update dependent calls and types.
9105         * config/mips/mips.c (mips_reorg_process_insns::htab):
9106         Change type to hash_table.  Update dependent calls and types.
9108         * config/sol2.c (solaris_comdat_htab):
9109         Change type to hash_table.  Update dependent calls and types.
9111         * config/t-sol2: Update for above.
9113 2013-05-29  Teresa Johnson  <tejohnson@google.com>
9115         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
9116         functions are not yet marked as defined.
9118 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
9119             Pat Haugen  <pthaugen@us.ibm.com>
9120             Peter Bergner  <bergner@vnet.ibm.com>
9122         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
9123         instructions.
9124         (VEC_A): Likewise.
9125         (VEC_C): Likewise.
9126         (vrotl<mode>3): Likewise.
9127         (vashl<mode>3): Likewise.
9128         (vlshr<mode>3): Likewise.
9129         (vashr<mode>3): Likewise.
9131         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9132         support for power8 V2DI builtins.
9134         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
9135         power8 V2DI builtins.
9136         (vupkhsw): Likewise.
9137         (vupklsw): Likewise.
9138         (vaddudm): Likewise.
9139         (vminsd): Likewise.
9140         (vmaxsd): Likewise.
9141         (vminud): Likewise.
9142         (vmaxud): Likewise.
9143         (vpkudum): Likewise.
9144         (vpksdss): Likewise.
9145         (vpkudus): Likewise.
9146         (vpksdus): Likewise.
9147         (vrld): Likewise.
9148         (vsld): Likewise.
9149         (vsrd): Likewise.
9150         (vsrad): Likewise.
9151         (vsubudm): Likewise.
9152         (vcmpequd): Likewise.
9153         (vcmpgtsd): Likewise.
9154         (vcmpgtud): Likewise.
9155         (vcmpequd_p): Likewise.
9156         (vcmpgtsd_p): Likewise.
9157         (vcmpgtud_p): Likewise.
9158         (vupkhsw): Likewise.
9159         (vupklsw): Likewise.
9160         (vaddudm): Likewise.
9161         (vmaxsd): Likewise.
9162         (vmaxud): Likewise.
9163         (vminsd): Likewise.
9164         (vminud): Likewise.
9165         (vpksdss): Likewise.
9166         (vpksdus): Likewise.
9167         (vpkudum): Likewise.
9168         (vpkudus): Likewise.
9169         (vrld): Likewise.
9170         (vsld): Likewise.
9171         (vsrad): Likewise.
9172         (vsrd): Likewise.
9173         (vsubudm): Likewise.
9175         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
9176         support for power8 V2DI instructions.
9178         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
9179         power8 V2DI instructions.  Combine pack and unpack insns to use an
9180         iterator for each mode.  Check whether a particular mode supports
9181         Altivec instructions instead of just checking TARGET_ALTIVEC.
9182         (UNSPEC_VPKUWUM): Likewise.
9183         (UNSPEC_VPKSHSS): Likewise.
9184         (UNSPEC_VPKSWSS): Likewise.
9185         (UNSPEC_VPKUHUS): Likewise.
9186         (UNSPEC_VPKSHUS): Likewise.
9187         (UNSPEC_VPKUWUS): Likewise.
9188         (UNSPEC_VPKSWUS): Likewise.
9189         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
9190         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
9191         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
9192         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
9193         (UNSPEC_VUPKHSB): Likewise.
9194         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
9195         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
9196         (UNSPEC_VUPKHSH): Likewise.
9197         (UNSPEC_VUPKLSB): Likewise.
9198         (UNSPEC_VUPKLSH): Likewise.
9199         (VI2): Likewise.
9200         (VI_char): Likewise.
9201         (VI_scalar): Likewise.
9202         (VI_unit): Likewise.
9203         (VP): Likewise.
9204         (VP_small): Likewise.
9205         (VP_small_lc): Likewise.
9206         (VU_char): Likewise.
9207         (add<mode>3): Likewise.
9208         (altivec_vaddcuw): Likewise.
9209         (altivec_vaddu<VI_char>s): Likewise.
9210         (altivec_vadds<VI_char>s): Likewise.
9211         (sub<mode>3): Likewise.
9212         (altivec_vsubcuw): Likewise.
9213         (altivec_vsubu<VI_char>s): Likewise.
9214         (altivec_vsubs<VI_char>s): Likewise.
9215         (altivec_vavgs<VI_char>): Likewise.
9216         (altivec_vcmpbfp): Likewise.
9217         (altivec_eq<mode>): Likewise.
9218         (altivec_gt<mode>): Likewise.
9219         (altivec_gtu<mode>): Likewise.
9220         (umax<mode>3): Likewise.
9221         (smax<mode>3): Likewise.
9222         (umin<mode>3): Likewise.
9223         (smin<mode>3): Likewise.
9224         (altivec_vpkuhum): Likewise.
9225         (altivec_vpkuwum): Likewise.
9226         (altivec_vpkshss): Likewise.
9227         (altivec_vpkswss): Likewise.
9228         (altivec_vpkuhus): Likewise.
9229         (altivec_vpkshus): Likewise.
9230         (altivec_vpkuwus): Likewise.
9231         (altivec_vpkswus): Likewise.
9232         (altivec_vpks<VI_char>ss): Likewise.
9233         (altivec_vpks<VI_char>us): Likewise.
9234         (altivec_vpku<VI_char>us): Likewise.
9235         (altivec_vpku<VI_char>um): Likewise.
9236         (altivec_vrl<VI_char>): Likewise.
9237         (altivec_vsl<VI_char>): Likewise.
9238         (altivec_vsr<VI_char>): Likewise.
9239         (altivec_vsra<VI_char>): Likewise.
9240         (altivec_vsldoi_<mode>): Likewise.
9241         (altivec_vupkhsb): Likewise.
9242         (altivec_vupkhs<VU_char>): Likewise.
9243         (altivec_vupkls<VU_char>): Likewise.
9244         (altivec_vupkhsh): Likewise.
9245         (altivec_vupklsb): Likewise.
9246         (altivec_vupklsh): Likewise.
9247         (altivec_vcmpequ<VI_char>_p): Likewise.
9248         (altivec_vcmpgts<VI_char>_p): Likewise.
9249         (altivec_vcmpgtu<VI_char>_p): Likewise.
9250         (abs<mode>2): Likewise.
9251         (vec_unpacks_hi_v16qi): Likewise.
9252         (vec_unpacks_hi_v8hi): Likewise.
9253         (vec_unpacks_lo_v16qi): Likewise.
9254         (vec_unpacks_hi_<VP_small_lc>): Likewise.
9255         (vec_unpacks_lo_v8hi): Likewise.
9256         (vec_unpacks_lo_<VP_small_lc>): Likewise.
9257         (vec_pack_trunc_v8h): Likewise.
9258         (vec_pack_trunc_v4si): Likewise.
9259         (vec_pack_trunc_<mode>): Likewise.
9261         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
9262         V2DI builtins.
9263         (vec_vmaxsd): Likewise.
9264         (vec_vmaxud): Likewise.
9265         (vec_vminsd): Likewise.
9266         (vec_vminud): Likewise.
9267         (vec_vpksdss): Likewise.
9268         (vec_vpksdus): Likewise.
9269         (vec_vpkudum): Likewise.
9270         (vec_vpkudus): Likewise.
9271         (vec_vrld): Likewise.
9272         (vec_vsld): Likewise.
9273         (vec_vsrad): Likewise.
9274         (vec_vsrd): Likewise.
9275         (vec_vsubudm): Likewise.
9276         (vec_vupkhsw): Likewise.
9277         (vec_vupklsw): Likewise.
9279 2013-05-29  Jan Hubicka  <jh@suse.cz>
9281         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
9282         flags; reorder rest of fields in more consistent way.
9283         (varpool_node): Remove analyzed, finalized and alias.
9284         (cgraph_ndoe): Likewise.
9285         (symtab_alias_ultimate_target): New function.
9286         (cgraph_function_node): Move offline.
9287         (cgraph_reset_node): Declare.
9288         (cgraph_comdat_can_be_unshared_p): Remove.
9289         (varpool_remove_initializer): Declare.
9290         (varpool_first_defined_variable, varpool_next_defined_variable
9291         cgraph_first_defined_function, cgraph_next_defined_function): Update.
9292         (cgraph_function_with_gimple_body_p): Update.
9293         (varpool_all_refs_explicit_p): Update.
9294         (symtab_alias_target): New function.
9295         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
9296         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
9297         (cgraph_function_or_thunk_node): Simplify using
9298         symtab_alias_ultimate_target.
9299         (varpool_variable_node): Likewise.
9300         * cgraph.c (cgraph_create_function_alias): Update.
9301         (cgraph_add_thunk): Update.
9302         (cgraph_remove_node): Update.
9303         (dump_cgraph_node): Do not dump removed flags.
9304         (cgraph_function_body_availability): Update.
9305         (cgraph_propagate_frequency): Update.
9306         (verify_cgraph_node): Check sanity of local flag.
9307         (cgraph_function_node): Move here from cgraph.h; revamp for
9308         cgraph_function_or_thunk_node.
9309         * lto-symtab.c (lto_varpool_replace_node): Update.
9310         (lto_symtab_resolve_can_prevail_p): Update.
9311         (lto_symtab_merge_cgraph_nodes): Update.
9312         * ipa-cp.c (determine_versionability, initialize_node_lattices,
9313         propagate_constants_accross_call, devirtualization_time_bonus,
9314         ipcp_propagate_stage): Update.
9315         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
9316         * ipa-inline-transform.c (clone_inlined_nodes,
9317         preserve_function_body_p): Update.
9318         * ipa-reference.c (propagate): Update.
9319         (write_node_summary_p): Update.
9320         * toplev.c (wrapup_global_declaration_2): Update.
9321         * cgraphunit.c (cgraph_analyze_function): Rename to ...
9322         (analyze_function) ... this one.
9323         (cgraph_process_new_functions): Update.
9324         (cgraph_reset_node): Export.
9325         (cgraph_finalize_function): Update.
9326         (cgraph_add_new_function): Update.
9327         (process_function_and_variable_attributes): Update.
9328         (varpool_finalize_decl): Update.
9329         (symbol_finalized): Remove.
9330         (symbol_finalized_and_needed): Rename to ...
9331         (symbol_defined_and_needed): ... update.
9332         (cgraph_analyze_functions): Update.
9333         (handle_alias_pairs): Update.
9334         (mark_functions_to_output): Update.
9335         (assemble_thunk): Update.
9336         (output_in_order): Update.
9337         (output_weakrefs): Update.
9338         (finalize_compilation_unit): Update.
9339         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
9340         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
9341         input_node, input_varpool_node): Update.
9342         * dbxout.c (dbxout_expand_expr): Update.
9343         * cgraphclones.c (cgraph_clone_node): Update.
9344         (cgraph_copy_node_for_versioning): Update.
9345         (cgraph_materialize_clone): Update.
9346         (cgraph_materialize_all_clones): Update.
9347         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
9348         propagate_pure_const, propagate_nothrow): Update.
9349         * lto-streamer-out.c (lto_output, write_symbol): Update.
9350         * ipa-utils.c (ipa_reverse_postorder): Update.
9351         * ipa-inline.c (can_inline_edge_p): Update.
9352         (update_caller_keys, ipa_inline): Update.
9353         * dwarf2out.c (reference_to_unused,
9354         premark_types_used_by_global_vars_helper): Update.
9355         * tree-eh.c (tree_could_trap_p): Update.
9356         * ipa-split.c (consider_split, execute_split_functions): Update.
9357         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
9358         has_addr_references_p): Update; move ahead in file for better
9359         readability.
9360         (process_references): Simplify.
9361         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
9362         bodies are removed.
9363         (cgraph_comdat_can_be_unshared_p): Make static.
9364         (cgraph_externally_visible_p): Update.
9365         (varpool_externally_visible_p): Update.
9366         (function_and_variable_visibility): Update.
9367         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
9368         ipa_tm_mark_force_output_node): Update.
9369         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
9370         estimate_edge_devirt_benefit, inline_generate_summary,
9371         inline_write_summary): Update.
9372         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
9373         * ipa-prop.c (ipa_compute_jump_functions): Update.
9374         (ipa_print_node_params, ipa_prop_read_section,
9375         ipa_update_after_lto_read, read_replacements_section): Update.
9376         * varasm.c (mark_decl_referenced): Update.
9377         (assemble_alias, dump_tm_clone_pairs): Update.
9378         * tree-inline.c (copy_bb): Update.
9379         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
9380         Update.
9381         * symtab.c (dump_symtab_base): Print new flags.
9382         (verify_symtab_base): Verify new flags.
9383         (symtab_alias_ultimate_target): New function.
9384         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
9385         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
9386         Update.
9387         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
9388         Update.
9389         * i386.c (ix86_get_function_versions_dispatcher,
9390         ix86_generate_version_dispatcher_body): Update.
9391         (fold_builtin_cpu): Use varpool_add_new_variable.
9392         * varpool.c (varpool_remove_initializer): Break out from ...
9393         (varpool_remove_node): ... this one.
9394         (dump_varpool_node, varpool_node_for_asm,
9395         cgraph_variable_initializer_availability, varpool_analyze_node,
9396         varpool_assemble_decl, varpool_remove_unreferenced_decls,
9397         varpool_finalize_named_section_flags, varpool_create_variable_alias):
9398         Update.
9400 2013-05-29  Jan Hubicka  <jh@suse.cz>
9402         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
9404 2013-05-29  Easwaran Raman  <eraman@google.com>
9406         PR tree-optimization/57442
9407         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
9408         when control exits the main loop.
9410 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
9412         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
9413         and RX600.
9414         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
9415         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
9416         * rx/t-rx: Add rx100 under multi library matches option for nofpu
9417         option.
9419 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9421         PR tree-optimization/57441
9422         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
9423         Don't limit size of incr_vec to number of candidates.
9425 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
9427         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
9428         and mips16 directories.
9429         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
9430         (MULTILIB_DIRNAMES): Ditto.
9431         (MULTILIB_EXCEPTIONS): Add new exceptions.
9432         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
9433         (MULTILIB_DIRNAMES): Ditto.
9434         (MULTILIB_EXCEPTIONS): Add new exceptions.
9436 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
9437             Marcus Shawcroft  <marcus.shawcroft@arm.com>
9439         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
9440         SYMBOL_TINY_ABSOLUTE.
9441         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
9442         SYMBOL_TINY_ABSOLUTE.
9443         (aarch64_expand_mov_immediate): Likewise.
9444         (aarch64_classify_symbol): Likewise.
9445         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
9446         Permit SYMBOL_TINY_ABSOLUTE.
9447         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
9449 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
9450             Marcus Shawcroft  <marcus.shawcroft@arm.com>
9452         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
9453         Refactor if/switch.  Replace gcc_assert with if.
9455 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9457         * config/i386/i386.c (initial_ix86_tune_features): Enable
9458         FP Reassociation for AMD bdver1 and bdver2.
9460 2013-05-29  Martin Jambor  <mjambor@suse.cz>
9462         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
9463         and IMAGPART_EXPR do not occur within other handled_components.
9465 2013-05-29  Richard Biener  <rguenther@suse.de>
9467         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
9468         access on whether the use is in the BB we currently try to
9469         vectorize.
9470         (vect_bb_vectorization_profitable_p): Pass the BB we currently
9471         vectorize to vect_bb_slp_scalar_cost.
9473 2013-05-29  Richard Biener  <rguenther@suse.de>
9475         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
9476         computing scalar cost offsetted by stmts that are kept live
9477         by scalar uses.
9478         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
9479         for computation of scalar cost.
9481 2013-05-28  Steve Ellcey  <sellcey@mips.com>
9483         * config/mips/mips-cpus.def (mips32r2): Change processor type.
9485 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
9487         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
9488         array notation built-in reduction functions.
9489         * doc/passes.texi (Passes): Added documentation about changes done
9490         for Cilk Plus.
9491         * doc/invoke.texi (C Dialect Options): Added documentation about
9492         the -fcilkplus flag.
9493         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
9494         (BUILTINS_DEF): Depend on cilkplus.def.
9495         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
9496         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
9497         * cilkplus.def: New file.
9499 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
9501         PR rtl-optimization/57439
9502         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
9504 2013-05-28  Easwaran Raman  <eraman@google.com>
9506         PR tree-optimization/57337
9507         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
9508         (find_insert_point): Correctly identify the insertion point
9509         when two statements with the same UID is compared.
9511 2013-05-28  Richard Biener  <rguenther@suse.de>
9513         PR tree-optimization/56787
9514         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
9515         from the list of data references.
9516         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
9517         clobbers.
9518         (vect_analyze_loop_operations): Likewise.
9519         (vect_transform_loop): Remove clobbers.
9521 2013-05-28  Martin Jambor  <mjambor@suse.cz>
9523         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
9524         and REALPART_EXPRs have scalar type.
9526 2013-05-28  Richard Biener  <rguenther@suse.de>
9528         PR tree-optimization/57411
9529         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
9530         virtual operands.
9531         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
9532         virtual operand propagation.
9534 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
9536         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
9537         destination register for bmasksi_vis.
9538         (vector_init_bshuffle): Likewise.
9539         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
9541 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
9543         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
9544         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
9545         mode if the instruction isn't available in the original mode.
9546         * config/sparc/sparc.opt (mfix-ut699): New option.
9547         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
9548         (divdf3): Turn into expander.
9549         (divdf3_nofix): New insn.
9550         (divdf3_fix): Likewise.
9551         (divsf3): Disable if -mfix-ut699.
9552         (sqrtdf2): Turn into expander.
9553         (sqrtdf2_nofix): New insn.
9554         (sqrtdf2_fix): Likewise.
9555         (sqrtsf2): Disable if -mfix-ut699.
9557 2013-05-27  Richard Biener  <rguenther@suse.de>
9559         PR middle-end/57412
9560         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
9561         block for the new loop.
9563 2013-05-27  Richard Biener  <rguenther@suse.de>
9565         PR tree-optimization/57343
9566         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
9567         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
9568         (number_of_iterations_cond): Do not build the folded tree.
9570 2013-05-27  Richard Biener  <rguenther@suse.de>
9572         Revert
9573         PR middle-end/57381
9574         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
9575         OEP_CONSTANT_ADDRESS_OF retained.
9577         PR tree-optimization/57417
9578         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
9579         for unchanged base.
9580         (set_ssa_val_to): Compare addresses using
9581         get_addr_base_and_unit_offset.
9583 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
9585         PR rtl-optimization/56833
9586         * postreload.c (move2add_record_mode): New function.
9587         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
9588         (move2add_use_add2_insn): Use move2add_record_sym_value.
9589         (move2add_use_add3_insn): Likewise.
9590         (reload_cse_move2add): Use move2add_valid_value_p and
9591         move2add_record_mode.  Invalidate call-clobbered and REG_INC
9592         affected regs by setting reg_mode to VOIDmode.
9593         (move2add_note_store): Don't pretend the inside of a SUBREG is
9594         the actual destination.  Invalidate single/leading registers by
9595         setting reg_mode to VOIDmode.
9596         Use move2add_record_sym_value, move2add_valid_value_p and
9597         move2add_record_mode.
9599 2013-05-27  Richard Biener  <rguenther@suse.de>
9601         PR tree-optimization/57396
9602         * tree-affine.c (double_int_constant_multiple_p): Properly
9603         return false for val == 0 and div != 0.
9605 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
9607         * config/mips/mips.h: Use #elif in preprocessor conditions.
9609 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
9611         PR target/53916
9612         * config/mips/constraints.md (kl): New constraint.
9613         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
9614         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
9615         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
9616         from LO for MIPS16.
9617         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
9619 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
9621         PR target/55777
9622         * config/mips/mips.c (mips_can_inline_p): New function.
9623         (TARGET_CAN_INLINE_P): Define.
9625 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
9627         * sched-int.h (ds_t, dw_t): Make unsigned int.
9628         Fix documentation that describes how all the ds_t bits are used.
9629         Reserve the last bit for delayed-branch scheduling.
9630         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
9631         (BITS_PER_DEP_WEAK): Fix definition and documentation.
9632         (gen_dep_weak_1): Remove prototype.
9633         * sched-deps.c (get_dep_weak_1): Make static.
9634         * target.def (speculate_insn, needs_block_p, gen_spec_check,
9635         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
9636         * doc/tm.texi: Regenerate.
9637         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
9639 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
9641         PR debug/56950
9642         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
9644 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
9645             Sandra Loosemore  <sandra@codesourcery.com>
9647         * config.gcc (powerpc-*): Allow native for with-cpu.
9649 2013-05-24  Jeff Law  <law@redhat.com>
9651         PR tree-optimization/57124
9652         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
9653         conversion feeding a condition if the range has an overflow
9654         if -fstrict-overflow.  Add warnings for when we do make the
9655         transformation.
9657 2013-05-24  Dehao Chen  <dehao@google.com>
9659         * tree-cfg.c (locus_discrim_map): Fix the typo.
9660         (locus_discrim_hasher): Likewise.
9661         (locus_discrim_hasher::hash): Likewise.
9662         (locus_discrim_hasher::equal): Likewise.
9664 2013-05-24  Martin Jambor  <mjambor@suse.cz>
9666         PR tree-optimization/57294
9667         * cgraph.h (ipa_record_stmt_references): Declare.
9668         * cgraphbuild.c (ipa_record_stmt_references): New function.
9669         (build_cgraph_edges): Use ipa_record_stmt_references.
9670         (rebuild_cgraph_edges): Likewise.
9671         (cgraph_rebuild_references): Likewise.
9672         * ipa-prop.c (ipa_modify_call_arguments): Discard references
9673         associated with the old statement and build references from the
9674         newly built statements.
9675         * ipa-ref.c (ipa_remove_stmt_references): New function.
9676         * ipa-ref.h (ipa_remove_stmt_references): Declare.
9678 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
9680         * lra-constraints.c (emit_spill_move): Use smaller mode for
9681         mem-mem moves.
9682         (check_and_process_move): Consider mem-reg moves for secondary
9683         too.
9684         (curr_insn_transform): Don't lose insns emitted before for
9685         secondary memory moves.
9686         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
9687         reg set up in the current insn.
9689 2013-05-24  Dehao Chen  <dehao@google.com>
9691         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
9692         hash function.
9693         (locus_descrim_hasher::equal): Likewise.
9694         (build_gimple_cfg): New discriminator assignment algorithm.
9695         (make_edges): Likewise.
9696         (next_discriminator_for_locus): Likewise.
9697         (same_line_p): Likewise.
9698         (assign_discriminators): Likewise.
9699         (make_cond_expr_edges): Likewise.
9700         (make_gimple_switch_edges): Likewise.
9701         (make_goto_expr_edges): Likewise.
9702         (make_gimple_asm_edges): Likewise.
9704 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
9706         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
9707         X format specifier to only display bottom 16 bits.
9708         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
9709         immediate to match for operand 2, since it will be masked.
9711 2013-05-24  Richard Biener  <rguenther@suse.de>
9713         PR tree-optimization/57287
9714         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
9715         all SSA names that occur in abnormal PHIs.
9717 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9719         PR tree-ssa/57385
9720         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
9721         that index is not negative.
9723 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
9725         PR rtl-optimization/55177
9726         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
9727         (simplify_byte_swapping_operation): New.
9728         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
9729         (simplify_relational_operation_1): Deal with BSWAP.
9731 2013-05-23  Richard Henderson  <rth@redhat.com>
9733         PR target/56742
9734         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
9735         (ix86_reorg): Call it.
9737 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
9739         PR target/57379
9740         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
9741         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
9742         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
9744 2013-05-23  Christian Bruel  <christian.bruel@st.com>
9746         PR debug/57351
9747         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
9749 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
9750             Marcus Shawcroft  <marcus.shawcroft@arm.com>
9752         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
9753         * config/aarch64/constraints.md (Usa): Remove.
9754         * doc/md.texi (AArch64 Usa): Remove.
9756 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
9757             Marcus Shawcroft  <marcus.shawcroft@arm.com>
9759         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
9760         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
9761         * config/aarch64/predicates.md (aarch64_const_address): Remove.
9762         (aarch64_mov_operand): Use aarch64_mov_operand_p.
9764 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
9766         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
9767         instruction (AdvSIMD).
9768         * config/aarch64/aarch64-builtins.c
9769         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
9770         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
9772 2013-05-23  Martin Jambor  <mjambor@suse.cz>
9774         PR middle-end/57347
9775         * tree.h (contains_bitfld_component_ref_p): Declare.
9776         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
9777         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
9778         caller.
9779         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
9780         not access a bit-field.  Assert all final offsets are byte-aligned.
9782 2013-05-23  Richard Biener  <rguenther@suse.de>
9784         PR tree-optimization/57380
9785         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
9786         least one invariant or re-used load.
9787         * passes.c (init_optimization_passes): Move pass_phiprop before
9788         pass_forwprop.
9790 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
9792         * config/aarch64/aarch64-simd.md
9793         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
9795 2013-05-23  Richard Biener  <rguenther@suse.de>
9797         PR middle-end/57381
9798         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
9799         OEP_CONSTANT_ADDRESS_OF retained.
9801 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
9803         PR middle-end/57344
9804         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
9805         don't lower unit.  Handle unit not being always BITS_PER_WORD.
9807 2013-05-23  Richard Biener  <rguenther@suse.de>
9809         PR rtl-optimization/57341
9810         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
9811         instead of true_dependence.
9813 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
9815         * bb-reorder.c (branch_threshold): Make const.
9816         (exec_threshold): Ditto.
9818 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
9819             Pat Haugen  <pthaugen@us.ibm.com>
9820             Peter Bergner  <bergner@vnet.ibm.com>
9822         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
9823         documentation for the power8 crypto builtins.
9825         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
9827         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
9828         macros for defining power8 builtin functions.
9829         (BU_P8V_AV_2): Likewise.
9830         (BU_P8V_AV_P): Likewise.
9831         (BU_P8V_VSX_1): Likewise.
9832         (BU_P8V_OVERLOAD_1): Likewise.
9833         (BU_P8V_OVERLOAD_2): Likewise.
9834         (BU_CRYPTO_1): Likewise.
9835         (BU_CRYPTO_2): Likewise.
9836         (BU_CRYPTO_3): Likewise.
9837         (BU_CRYPTO_OVERLOAD_1): Likewise.
9838         (BU_CRYPTO_OVERLOAD_2): Likewise.
9839         (XSCVSPDP): Fix typo, point to the correct instruction.
9840         (VCIPHER): Add power8 crypto builtins.
9841         (VCIPHERLAST): Likewise.
9842         (VNCIPHER): Likewise.
9843         (VNCIPHERLAST): Likewise.
9844         (VPMSUMB): Likewise.
9845         (VPMSUMH): Likewise.
9846         (VPMSUMW): Likewise.
9847         (VPERMXOR_V2DI): Likewise.
9848         (VPERMXOR_V4SI: Likewise.
9849         (VPERMXOR_V8HI: Likewise.
9850         (VPERMXOR_V16QI: Likewise.
9851         (VSHASIGMAW): Likewise.
9852         (VSHASIGMAD): Likewise.
9853         (VPMSUM): Likewise.
9854         (VPERMXOR): Likewise.
9855         (VSHASIGMA): Likewise.
9857         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
9858         __CRYPTO__ if the crypto instructions are available.
9859         (altivec_overloaded_builtins): Add support for overloaded power8
9860         builtins.
9862         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
9863         support for power8 crypto builtins.
9864         (builtin_function_type): Likewise.
9865         (altivec_init_builtins): Add support for builtins that take vector
9866         long long (V2DI) arguments.
9868         * config/rs6000/crypto.md: New file, define power8 crypto
9869         instructions.
9871 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
9872             Pat Haugen  <pthaugen@us.ibm.com>
9873             Peter Bergner  <bergner@vnet.ibm.com>
9875         * doc/invoke.texi (Option Summary): Add power8 options.
9876         (RS/6000 and PowerPC Options): Likewise.
9878         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
9879         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
9880         wm, wn, wr documentation.
9882         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
9883         registers if direct move instructions are enabled.
9884         (wn): New constraint for no registers.
9885         (wq): New constraint for quad word even GPR registers.
9886         (wr): New constraint if 64-bit instructions are enabled.
9887         (wv): New constraint if power8 vector instructions are enabled.
9888         (wQ): New constraint for quad word memory locations.
9890         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
9891         constraint for 0..15 for crypto instructions.
9892         (gpc_reg_operand): If VSX allow registers in VSX registers as well
9893         as GPR and floating point registers.
9894         (int_reg_operand): New predicate to match only GPR registers.
9895         (base_reg_operand): New predicate to match base registers.
9896         (quad_int_reg_operand): New predicate to match even GPR registers
9897         for quad memory operations.
9898         (vsx_reg_or_cint_operand): New predicate to allow vector logical
9899         operations in both GPR and VSX registers.
9900         (quad_memory_operand): New predicate for quad memory operations.
9901         (reg_or_indexed_operand): New predicate for direct move support.
9903         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
9904         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
9905         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
9906         (POWERPC_MASKS): Add power8 options.
9907         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
9908         various options.
9910         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9911         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
9913         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
9914         (-mpower8-fusion): New power8 options.
9915         (-mpower8-fusion-sign): Likewise.
9916         (-mpower8-vector): Likewise.
9917         (-mcrypto): Likewise.
9918         (-mdirect-move): Likewise.
9919         (-mquad-memory): Likewise.
9921         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
9922         power8.
9923         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
9924         registers.
9925         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
9926         (rs6000_debug_vector_unit): Add p8_vector.
9927         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
9928         definitions.  Also print fusion state.
9929         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
9930         (rs6000_builtin_mask_calculate): Add power8 builtin support.
9931         (rs6000_option_override_internal): Add support for power8.
9932         (rs6000_common_init_builtins): Add debugging for skipped builtins
9933         if -mdebug=builtin.
9934         (rs6000_adjust_cost): Add power8 support.
9935         (rs6000_issue_rate): Likewise.
9936         (insn_must_be_first_in_group): Likewise.
9937         (insn_must_be_last_in_group): Likewise.
9938         (force_new_group): Likewise.
9939         (rs6000_register_move_cost): Likewise.
9940         (rs6000_opt_masks): Likewise.
9942         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9943         power8 capable assembler, default to power7 options.
9944         (TARGET_DIRECT_MOVE): Likewise.
9945         (TARGET_CRYPTO): Likewise.
9946         (TARGET_P8_VECTOR): Likewise.
9947         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9948         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9949         (VECTOR_MEM_P8_VECTOR_P): Likewise.
9950         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9951         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9952         (TARGET_XSCVDPSPN): Likewise.
9953         (TARGET_XSCVSPDPN): Likewsie.
9954         (TARGET_SYNC_HI_QI): Likewise.
9955         (TARGET_SYNC_TI): Likewise.
9956         (MASK_CRYPTO): Likewise.
9957         (MASK_DIRECT_MOVE): Likewise.
9958         (MASK_P8_FUSION): Likewise.
9959         (MASK_P8_VECTOR): Likewise.
9960         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9961         temporary used by some of the direct move instructions to get two FP
9962         temporary registers does not force creation of a stack frame.
9963         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9964         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9965         that any VSX registers are tieable, even if they are also an
9966         Altivec vector mode.
9967         (r6000_reg_class_enum): Add wm, wr, wv constraints.
9968         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9969         (RS6000_BTM_CRYPTO): Likewise.
9970         (RS6000_BTM_COMMON): Likewise.
9972         * config/rs6000/rs6000.md (cpu attribute): Add power8.
9973         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9974         (enum rs6000_vector): Add power8 vector support.
9976 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9978         PR target/19599
9979         PR target/57340
9980         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9981         (any_sibcall_could_use_r3): this and handle indirect calls.
9982         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9984 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9986         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9988 2013-05-22  Richard Biener  <rguenther@suse.de>
9990         PR middle-end/57349
9991         * profile.c (branch_prob): Do not split blocks that are
9992         abnormally receiving from ECF_RETURNS_TWICE functions.
9994 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9996         * recog.c (offsettable_address_addr_space_p): Fix calculation of
9997         address mode.  Move pointer mode initialization to the same place.
9999 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
10001         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
10002         while it has any effect.
10004 2013-05-21  Easwaran Raman  <eraman@google.com>
10006         PR tree-optimization/57322
10007         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
10008         UID of the statement added to the BB to be 1.
10010 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
10012         PR tree-optimization/57331
10013         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
10014         of conversion from pointer type to integral type with integer.
10016 2013-05-21  Martin Jambor  <mjambor@suse.cz>
10018         PR lto/57289
10019         * ipa-prop.c (ipa_read_node_info): Process param_used and
10020         controlled_uses in the same order as when writing.
10022 2013-05-21  Magnus Granberg  <baldrick@free.fr>
10024         PR plugins/56754
10025         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
10027 2013-05-21  Richard Biener  <rguenther@suse.de>
10029         PR tree-optimization/57318
10030         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
10031         estimate stmts with side-effects as likely eliminated.
10033 2013-05-21  Richard Biener  <rguenther@suse.de>
10035         PR tree-optimization/57330
10036         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
10037         preserve the call stmts fntype.
10039 2013-05-21  Richard Biener  <rguenther@suse.de>
10041         PR tree-optimization/57303
10042         * tree-ssa-sink.c (statement_sink_location): Improve killing
10043         stmt detection and properly handle self-assignments.
10045 2013-05-21  Christian Bruel  <christian.bruel@st.com>
10047         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
10048         spanning registers. LEAF_REG_REMAP is supported only for contiguous
10049         registers. Set register size out of the PARALLEL loop.
10051 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
10053         PR target/56547
10054         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
10055         (*fmasf4, *fmasf4_media): New insns.
10057 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
10059         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
10060         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
10061         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
10062         (mips_idiv_insns): Update the comments to say that the returned
10063         instruction counts are in units of BASE_INSN_LENGTH.
10064         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
10065         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
10066         using 2 rather than 4 as the length of indirect MIPS16 and
10067         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
10068         length of a NOP.  Don't divide MIPS16 lengths by 2.
10069         (mips16_split_long_branches): Assume a branch is long if the
10070         length is greater than 4 rather than 8.
10071         * config/mips/mips.md (length): Give MIPS16 lengths directly,
10072         rather than multiplying them by 2.  Multiply instruction counts
10073         by BASE_INSN_LENGTH rather than 4.
10074         (*jump_mips16, tls_get_tp_mips16_<mode>)
10075         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
10077 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
10079         * config/mips/mips.md (extended_mips16): Remove branch case.
10080         (length): Remove duplicated extended_mips16 test.
10082 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
10084         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
10086 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
10088         * recog.h (Recog_data): Rename to...
10089         (recog_data_d): ...this.
10090         (recog_data): Update accordingly.
10091         * recog.c (recog_data): Likewise.
10092         * reload.c (save_recog_data): Likewise.
10093         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
10094         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
10096 2013-05-17  Julian Brown  <julian@codesourcery.com>
10098         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
10099         found in a REG_EQUAL note, invalidate it.
10101 2013-05-17  Easwaran Raman  <eraman@google.com>
10103         * tree-ssa-reassoc.c (find_insert_point): New function.
10104         (insert_stmt_after): Likewise.
10105         (get_def_stmt): Likewise.
10106         (ensure_ops_are_available): Likewise.
10107         (not_dominated_by): Likewise.
10108         (rewrite_expr_tree): Do not move statements beyond what is
10109         necessary. Remove call to swap_ops_for_binary_stmt...
10110         (reassociate_bb): ... and move it here.
10111         (build_and_add_sum): Assign UIDs for new statements.
10112         (linearize_expr): Likewise.
10113         (do_reassoc): Renumber gimple statement UIDs.
10115 2013-05-17  Jan Hubicka  <jh@suse.cz>
10117         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
10118         weakrefs.
10119         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
10120         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
10121         weakrefs.
10122         (output_weakrefs): Update.
10124 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
10125             Martin Jambor  <mjambor@suse.cz>
10127         PR middle-end/57276
10128         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
10129         value that corresponds to the given aggval is found in values vector.
10131 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
10133         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
10134         sse, sse2, sse3, ssse3 and sse4a flags to options.
10136 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
10138         * gengtype-state.c: (s_expr_writer): New class, to handle
10139         prettifying of output layout of s-expressions.
10140         (state_writer): New class, to write out gtype.state.
10141         (state_written_type_count): Move this variable into member data of
10142         state_writer.
10143         (s_expr_writer::s_expr_writer): New code: constructor for new class
10144         (state_writer::state_writer(): ditto
10145         (s_expr_writer::write_new_line): New function
10146         (s_expr_writer::write_any_indent): ditto
10147         (s_expr_writer::begin_s_expr): ditto
10148         (s_expr_writer::end_s_expr): ditto
10149         (write_state_fileloc): convert to method of state_writer...
10150         (state_writer:: write_state_fileloc): ...and use methods of
10151         s_expr_writer to write indentation into the gtype.state output file
10152         to visually represent the hierarchical structure of the list
10153         structures
10154         (write_state_fields): ditto, renaming to...
10155         (state_writer::write_state_fields)
10156         (write_state_a_string): ditto, renaming to...
10157         (state_writer::write_state_a_string)
10158         (write_state_string_option): ditto, renaming to...
10159         (state_writer::write_state_string_option)
10160         (write_state_type_option): ditto, renaming to...
10161         (state_writer::write_state_type_option)
10162         (write_state_nested_option): ditto, renaming to...
10163         (state_writer::write_state_nested_option)
10164         (write_state_option): ditto, renaming to...
10165         (state_writer::write_state_option)
10166         (write_state_options): ditto, renaming to...
10167         (state_writer::write_state_options)
10168         (write_state_lang_bitmap): ditto, renaming to...
10169         (state_writer::write_state_lang_bitmap)
10170         (write_state_version): ditto, renaming to...
10171         (state_writer::write_state_version)
10172         (write_state_scalar_type): ditto, renaming to...
10173         (state_writer::write_state_scalar_type)
10174         (write_state_string_type): ditto, renaming to...
10175         (state_writer::write_state_string_type)
10176         (write_state_undefined_type): ditto, renaming to...
10177         (state_writer::write_state_undefined_type)
10178         (write_state_struct_union_type): ditto, renaming to...
10179         (state_writer::write_state_struct_union_type)
10180         (write_state_struct_type): ditto, renaming to...
10181         (state_writer::write_state_struct_type)
10182         (write_state_user_struct_type): ditto, renaming to...
10183         (state_writer::write_state_user_struct_type)
10184         (write_state_lang_struct_type): ditto, renaming to...
10185         (state_writer::write_state_lang_struct_type)
10186         (write_state_param_struct_type): ditto, renaming to...
10187         (state_writer::write_state_param_struct_type)
10188         (write_state_pointer_type): ditto, renaming to...
10189         (state_writer::write_state_pointer_type)
10190         (write_state_array_type): ditto, renaming to...
10191         (state_writer::write_state_array_type)
10192         (write_state_gc_used): ditto, renaming to...
10193         (state_writer::write_state_gc_used)
10194         (write_state_common_type_content): ditto, renaming to...
10195         (state_writer::write_state_common_type_content)
10196         (write_state_type): ditto, renaming to...
10197         (state_writer::write_state_type)
10198         (write_state_pair_list): ditto, renaming to...
10199         (state_writer::write_state_pair_list)
10200         (write_state_pair): ditto, renaming to...
10201         (state_writer::write_state_pair)
10202         (write_state_typedefs): ditto, renaming to...
10203         (state_writer::write_state_typedefs)
10204         (write_state_structures): ditto, renaming to...
10205         (state_writer::write_state_structures)
10206         (write_state_param_structs): ditto, renaming to...
10207         (state_writer::write_state_param_structs)
10208         (write_state_variables): ditto, renaming to...
10209         (state_writer::write_state_variables)
10210         (write_state_srcdir): ditto, renaming to...
10211         (state_writer::write_state_srcdir)
10212         (write_state_files_list): ditto, renaming to...
10213         (state_writer::write_state_files_list)
10214         (write_state_languages): ditto, renaming to...
10215         (state_writer::write_state_languages)
10216         (write_state): create a state_writer instance and use it when
10217         writing out the state file
10219 2013-05-17  Mike Stump  <mikestump@comcast.net>
10221         PR rtl-optimization/57304
10222         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
10223         accessing DF_REF_REAL_LOC.
10225 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
10227         PR rtl-optimization/57281
10228         PR rtl-optimization/57300
10229         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
10230         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
10231         what the other splitter did if the registers are dead.
10233 2013-05-17  Richard Biener  <rguenther@suse.de>
10235         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
10236         MEM_REF offsets.
10238 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
10240         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
10241         linking.
10243 2013-05-17  Marek Polacek  <polacek@redhat.com>
10245         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
10246         length when doing non-zero store of storing '\0' to '\0'.
10248 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
10250         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
10251         vect_external_def oprnd1 with loop_vinfo, try to emit
10252         optional cast, negation and and stmts on the loop preheader
10253         edge instead of into the pattern def seq.
10255         PR tree-optimization/57051
10256         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
10257         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
10259 2013-05-16  Nick Clifton  <nickc@redhat.com>
10261         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
10262         (rl78_is_naked_func): New function.
10263         (rl78_expand_prologue): Skip prologue generation for naked functions.
10264         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
10265         * doc/extend.texi (naked): Add RL78 to the list of processors
10266         that supports this attribute.
10268 2013-05-16  Jeff Law  <law@redhat.com>
10270         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
10272 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
10274         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
10275         cache parameters using detect_caches_amd also for CYRIX,
10276         NSC and TM2 signatures.
10278 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
10279             Dzianis Kahanovich  <mahatma@eu.by>
10281         PR target/45359
10282         PR target/46396
10283         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
10284         VIA/Centaur processors and determine their cache parameters
10285         using detect_caches_amd.
10287 2013-05-16  Teresa Johnson  <tejohnson@google.com>
10289         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
10290         (rtl_verify_edges): New function.
10291         (rtl_verify_bb_insns): Ditto.
10292         (rtl_verify_bb_pointers): Ditto.
10293         (rtl_verify_bb_insn_chain): Ditto.
10294         (rtl_verify_fallthru): Ditto.
10295         (rtl_verify_bb_layout): Ditto.
10296         (rtl_verify_flow_info_1): Outline checks into new functions.
10297         (rtl_verify_flow_info): Ditto.
10299 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
10301         * cfghooks.c (copy_bbs): Add update_dominance argument.
10302         * cfghooks.h (copy_bbs): Update prototype.
10303         * tree-cfg.c (gimple_duplicate_sese_region):
10304         Add update_dominance argument.
10305         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
10306         * tree-ssa-loop-ch.c (copy_loop_headers): Update
10307         gimple_duplicate_sese_region call.
10308         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
10309         Update copy_bbs call.
10310         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
10311         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
10313 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
10315         * tree-vectorizer.h (NUM_PATTERNS): Increment.
10316         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
10317         vect_recog_rotate_pattern.
10318         (vect_recog_rotate_pattern): New function.
10320 2013-05-16  Jason Merrill  <jason@redhat.com>
10322         * Makefile.in (LLINKER): New variable.
10323         (mostlyclean): Remove link mutex.
10324         * configure.ac: Handle --enable-link-mutex.
10325         * lock-and-run.sh: New script.
10327 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
10329         PR target/19599
10330         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
10331         for NULL decl.
10333 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
10335         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
10337 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
10339         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
10340         * config/arm/arm.c (next_consecutive_mem): New function.
10341         (gen_movmem_ldrd_strd): Likewise.
10342         * config/arm/arm.md (movmemqi): Update condition and code.
10343         (unaligned_loaddi, unaligned_storedi): New patterns.
10345 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
10347         * config.gcc: Obsolete *-*-solaris2.9*.
10348         * doc/install.texi (Specific, *-*-solaris2*): Document it.
10350 2013-05-16  Richard Biener  <rguenther@suse.de>
10352         * passes.c (init_optimization_passes): Move pass_parallelize_loops
10353         earlier, after GRAPHITE transforms and IV canonicalization.
10355 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
10357         * omp-low.c (extract_omp_for_data): For collapsed loops,
10358         if at least one of the loops is known at compile time to
10359         iterate zero times, set count to 0.
10360         (expand_omp_regimplify_p): New function.
10361         (expand_omp_for_generic): For collapsed loops, if at least
10362         one of the loops isn't known to iterate at least once,
10363         add runtime check with setting count to 0.
10364         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
10365         For unsigned types if it isn't known at compile time that
10366         the loop will iterate at least once, add runtime check to bypass
10367         the whole loop if initial condition isn't true.
10369 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
10371         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
10373 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
10375         PR middle-end/57286
10376         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
10377         transformations to avoid an infinite loop.
10379 2013-05-16  Marek Polacek  <polacek@redhat.com>
10381         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
10383 2013-05-15  Leif Ekblad  <leif@rdos.net>
10385         * config/i386/i386.c (ix86_decompose_address): Use
10386         DEFAULT_TLS_SEG_REG to access TLS segment register.
10387         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
10388         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
10389         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
10391 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
10393         PR target/57260
10394         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
10395         sibling calls to functions that would normally be lazily bound,
10396         unless $gp is call-clobbered.
10398 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
10400         * config/i386/i386.c (ix86_option_override_internal): Update
10401         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
10402         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
10403         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
10404         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
10405         of TARGET_3DNOW.
10406         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
10408 2013-05-15  Andreas Schwab  <schwab@suse.de>
10410         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
10411         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
10412         third operand.
10414 2013-05-15  Teresa Johnson  <tejohnson@google.com>
10416         * loop-unroll.c (report_unroll_peel): Check decision before
10417         emitting unroll/peel message.
10419 2013-05-15  Teresa Johnson  <tejohnson@google.com>
10421         * function.h (has_bb_partition): New rtl_data flag.
10422         (bb_reorder_complete): Ditto.
10423         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
10424         instead of flag_reorder_blocks_and_partition.
10425         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
10426         with some enhancements.
10427         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
10428         * bb-reorder.c (connect_traces): Check for has_bb_partition
10429         instead of flag_reorder_blocks_and_partition.
10430         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
10431         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
10432         verify_hot_cold_block_grouping.
10433         (partition_hot_cold_basic_blocks): Set has_bb_partition.
10435 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
10437         PR target/19599
10438         * config/arm/predicates.md (call_insn_operand): New predicate.
10439         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
10440         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
10441         if insn is not a tail call.
10442         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
10443         registers.
10444         * config/arm/arm.h (enum reg_class): New caller save register class.
10445         (REG_CLASS_NAMES): Likewise.
10446         (REG_CLASS_CONTENTS): Likewise.
10447         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
10448         without decls.
10450 2013-05-15  Richard Biener  <rguenther@suse.de>
10452         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
10453         of MSG_OPTIMIZED_LOCATIONS.
10454         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
10455         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
10456         message.
10457         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
10458         of MSG_OPTIMIZED_LOCATIONS.
10459         (execute_vect_slp): Likewise.
10460         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
10461         (vect_create_cond_for_alias_checks): Likewise.
10462         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
10463         (vect_recog_widen_mult_pattern): Likewise.
10464         (vect_recog_widen_sum_pattern): Likewise.
10465         (vect_recog_over_widening_pattern): Likewise.
10466         (vect_recog_widen_shift_pattern): Likewise.
10467         (vect_recog_vector_vector_shift_pattern): Likewise.
10468         (vect_recog_divmod_pattern): Likewise.
10469         (vect_recog_mixed_size_cond_pattern): Likewise.
10470         (vect_recog_bool_pattern): Likewise.
10471         (vect_pattern_recog_1): Likewise.
10473 2013-05-15  Martin Jambor  <mjambor@suse.cz>
10475         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
10476         non-functions to builtin_unreachable.
10477         * ipa-inline-transform.c (inline_call): Do not assert estimates were
10478         correct when new direct edges were discovered.
10480 2013-05-15  Martin Jambor  <mjambor@suse.cz>
10482         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
10483         header, print symbol order instead of node uid, print more information
10484         about indirect edge targets.
10485         (ipa_make_edge_direct_to_target): Print symbol order instead of node
10486         uids.
10487         (ipa_make_edge_direct_to_target): Likewise.
10488         (remove_described_reference): Likewise.
10489         (propagate_controlled_uses): Likewise.
10490         (ipa_print_node_params): Also print symbol order.
10491         (ipcp_transform_function): Print symbol order instead of node uids.
10492         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
10493         (cgraph_get_create_real_symbol_node): Likewise.
10494         * ipa-cp.c (print_lattice): Likewise.
10495         (print_all_lattices): Likewise.
10496         (determine_versionability): Likewise.
10497         (initialize_node_lattices): Likewise.
10498         (estimate_local_effects): Likewise.
10499         (update_profiling_info): Likewise.
10500         (create_specialized_node): Likewise.
10501         (perhaps_add_new_callers): Likewise.
10502         (decide_about_value): Likewise.
10503         (decide_whether_version_node): Likewise.
10504         (identify_dead_nodes): Likewise.
10505         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
10506         (dump_inline_summary): Likewise.
10507         (estimate_node_size_and_time): Likewise.
10508         (inline_analyze_function): Likewise.
10509         * ipa-inline.c (report_inline_failed_reason): Likewise.
10510         (want_early_inline_function_p): Likewise.
10511         (edge_badness): Likewise.
10512         (update_edge_key): Likewise.
10513         (inline_small_functions): Likewise.  Add dumping of order to two other
10514         dumps.
10515         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
10516         instead of node uids.
10517         (propagate_pure_const): Likewise.
10518         (propagate_pure_const): Likewise.
10519         * ipa-utils.c (dump_cgraph_node_set): Likewise.
10520         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
10521         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
10522         of node uids.
10523         * tree-pretty-print.c (dump_function_header): Likewise.
10524         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
10525         Print symbol order instead of node uids.
10527 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10529         * config/s390/s390.c (s390_register_move_cost): Don't impose the
10530         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
10532 2013-05-15  Richard Biener  <rguenther@suse.de>
10534         PR tree-optimization/57275
10535         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
10536         return value for fail to do runtime alias checks for gather loads.
10538 2013-05-15  Jan Hubicka  <jh@suse.cz>
10540         PR lto/57038
10541         PR lto/47375
10542         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
10543         weakrefs are not external.
10544         (lto_symtab_merge_decls): Fix thinko when dealing with
10545         non-lto_symtab decls.
10546         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
10547         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
10548         * varpool.c (dump_varpool_node): Dump more flags.
10550 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
10552         * config/i386/i386.c (processor_alias_table): Add instruction
10553         FSGSBASE for AMD bdver3 architecture.
10555 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
10557         * tree.c (warn_deprecated_use): Print file:line using locus color.
10558         * diagnostic.c (diagnostic_report_current_module): Print file:line
10559         and file:line:column using locus color.
10561 2013-05-14  Mike Stump  <mikestump@comcast.net>
10563         * gdbinit.in: Add __null.
10565 2013-05-14  Mike Stump  <mikestump@comcast.net>
10567         * recog.h: Rename struct recog_data to Recog_data.
10568         * recog.c: Likewise.
10569         * reload.c (can_reload_into): Likewise.
10570         * config/picochip/picochip.c: Likewise.
10572 2013-05-14  Mike Stump  <mikestump@comcast.net>
10574         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
10576 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
10578         * resource.h (struct resources): Remove unch_memory member.
10579         (CLEAR_RESOURCE): Don't clear unch_memory.
10580         * resource.c (mark_referenced_resources): Don't set it.
10581         (mark_set_resources): Likewise.
10582         (mark_target_live_regs): Don't clear it.
10583         (init_resource_info): Likewise.
10584         * reorg.c (resource_conflicts_p): Don't compare it.
10585         (redundant_insn): Don't set it.
10587         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
10588         Remove prototypes.
10589         * emit-rtl.c (next_label): Remove unused function.
10590         (skip_consecutive_labels, link_cc0_insns): Move to ...
10591         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
10592         only place where these functions are used, and make them static.
10594 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
10596         * fold-const.c (fold_negate_expr): Handle vectors.
10597         (fold_truth_not_expr): Make it static.
10598         (fold_invert_truthvalue): New static function.
10599         (invert_truthvalue_loc): Handle vectors. Do not call
10600         fold_truth_not_expr directly.
10601         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
10602         <TRUTH_NOT_EXPR>: Do not cast to boolean.
10603         (fold_comparison): Handle vector constants.
10604         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
10605         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
10606         * tree.h (fold_truth_not_expr): Remove declaration.
10608 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
10610         * config/aarch64/aarch64-simd.md
10611         (aarch64_vcond_internal<mode>): Rename to...
10612         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
10613         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
10614         float modes. Clarify all iterator modes.
10615         (vcond<mode><mode>): Use new name for vcond expanders.
10616         (vcond<v_cmp_result><mode>): Likewise.
10617         (vcondu<mode><mode>: Likewise.
10618         * config/aarch64/iterators.md (VDQF_COND): New.
10620 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
10622         PR bootstrap/57266
10623         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
10624         variable for the shift amount. Check that we shift by non-negative
10625         amounts.
10627 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
10629         PR target/42017
10630         * config/arm/arm.h (EPILOGUE_USES): Only return true
10631         for LR_REGNUM after epilogue_completed.
10633 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
10635         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
10636         is error_mark_node.
10638 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
10640         PR target/57261
10641         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
10642         and Solaris 11+/x86 with gld.
10643         * configure: Regenerate.
10645 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
10647         * expmed.c (expand_shift_1): Canonicalize rotates by
10648         constant bitsize / 2 to bitsize - 1.
10649         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
10650         case ROTATERT>: Likewise.
10652         Revert:
10653         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
10655         * config/i386/i386.md (rotateinv): New code attr.
10656         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
10657         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
10658         roll $31, %eax, etc.
10660 2013-05-14  Richard Biener  <rguenther@suse.de>
10662         PR middle-end/57235
10663         * tree-eh.c (sink_clobbers): Give up for successors with
10664         multiple predecessors and no virtual uses.
10666 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
10668         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
10669         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
10671 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
10673         PR middle-end/57251
10674         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
10675         the case when both op0 and op1 have VOIDmode.
10677 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10679         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
10680         in multiply-accumulate mode.
10682 2013-05-13  Guozhi Wei  <carrot@google.com>
10684         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
10686 2013-05-13  Kai Tietz  <ktietz@redhat.com>
10688         PR target/56975
10689         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
10690         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
10691         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
10692         * config/i386/i386.c (ix86_option_override_internal): Likewise.
10693         (ix86_expand_prologue): Likewise.
10694         (ix86_expand_split_stack_prologue): Likewise.
10695         (legitimate_pic_address_disp_p): Likewise.
10696         (legitimize_pic_address): Likewise.
10697         (legitimize_tls_address): Likewise.
10698         (legitimize_pe_coff_symbol): Likewise.
10699         (output_pic_addr_const): Likewise.
10700         (construct_plt_address): Likewise.
10701         (ix86_expand_call): Likewise.
10702         (x86_output_mi_thunk): Likewise.
10703         (x86_function_profiler): Likewise.
10705 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
10707         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
10708         similar switch cases.
10709         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
10710         (aarch64_simd_mov_to_<mode>low): Delete.
10711         (aarch64_simd_mov_to_<mode>high): Delete.
10712         (move_lo_quad_<mode>): Add w<-r alternative.
10713         (aarch64_simd_move_hi_quad_<mode>): Likewise.
10714         (aarch64_simd_mov_from_*): Update type attribute.
10715         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
10716         statement.
10718 2013-05-13  Jan Hubicka  <jh@suse.cz>
10720         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
10721         * config/i386/i386.c (ix86_compute_frame_layout,
10722         ix86_expand_epilogue, emit_i387_cw_initialization,
10723         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
10724         ix86_local_alignment): Fix use of size/speed predicates.
10726 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
10728         PR tree-optimization/45216
10729         PR tree-optimization/57157
10730         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
10731         the (-Y) & (B - 1) variant if OP is |.
10732         * expmed.c (expand_shift_1): For rotations by const0_rtx just
10733         return shifted.  Use (-op1) & (prec - 1) as other_amount
10734         instead of prec - op1.
10736 2013-05-13  Martin Jambor  <mjambor@suse.cz>
10738         PR middle-end/42371
10739         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
10740         (ipa_constant_data): New type.
10741         (ipa_jump_func): Use ipa_constant_data to hold information about
10742         constant jump functions.
10743         (ipa_get_jf_constant): Adjust to jump function type changes.
10744         (ipa_get_jf_constant_rdesc): New function.
10745         (ipa_param_descriptor): New field controlled_uses.
10746         (ipa_get_controlled_uses): New function.
10747         (ipa_set_controlled_uses): Likewise.
10748         * ipa-ref.h (ipa_find_reference): Declare.
10749         * ipa-prop.c (ipa_cst_ref_desc): New type.
10750         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
10751         changes.
10752         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
10753         New parameter cs.  Adjust all callers.
10754         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
10755         (remove_described_reference): New function.
10756         (jfunc_rdesc_usable): Likewise.
10757         (try_make_edge_direct_simple_call): Decrement controlled use count,
10758         attempt to remove reference if it hits zero.
10759         (combine_controlled_uses_counters): New function.
10760         (propagate_controlled_uses): Likewise.
10761         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
10762         (ipa_edge_duplication_hook): Duplicate reference descriptions.
10763         (ipa_print_node_params): Print described use counter.
10764         (ipa_write_jump_function): Adjust to jump function type changes.
10765         (ipa_read_jump_function): New parameter CS, pass it to
10766         ipa_set_jf_constant.  Adjust caller.
10767         (ipa_write_node_info): Stream controlled use count
10768         (ipa_read_node_info): Likewise.
10769         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
10770         asserting.
10771         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
10772         count.  Remove cloning-added reference if it reaches zero.
10773         * ipa-ref.c (ipa_find_reference): New function.
10775 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
10777         * config/i386/i386.c (processor_target_table): Modified default
10778         alignment values for AMD BD and BT architectures.
10780 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
10782         * tree-vect-generic.c (uniform_vector_p): Move ...
10783         * tree.c (uniform_vector_p): ... here.
10784         * tree.h (uniform_vector_p): Declare it.
10785         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
10786         into a scalar.
10788 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
10790         PR tree-optimization/57230
10791         * tree-ssa-strlen.c (handle_char_store): Record length for
10792         array store from STRING_CST.
10794         PR tree-optimization/57230
10795         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
10796         check.
10798 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
10800         * config/epiphany/epiphany.c (epiphany_init): Check size of
10801         NUM_MODES_FOR_MODE_SWITCHING.
10802         (epiphany_expand_prologue):
10803         Remove CONFIG_REGNUM initial value handling code.
10804         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
10805         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
10806         (emit_set_fp_mode, epiphany_mode_after): Likewise.
10807         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
10808         Don't return 1 for FP_MODE_NONE.
10809         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
10810         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
10811         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
10812         * config/epiphany/epiphany.md (save_config): New pattern.
10814 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
10816         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
10818 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
10820         * config/i386/i386.md (memory): Handle sseishft1.
10821         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
10822         (*vec_extractv2di_1): Ditto.
10824 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
10826         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
10827         saved registers.
10829 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
10831         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
10832         Add mthumb/march=armv7-a multilib.
10833         Add mthumb/march=armv7-r multilib.
10834         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
10836 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
10838         * config/v850/t-rtems: Add more multilibs.
10840 2013-05-10  Richard Biener  <rguenther@suse.de>
10842         PR tree-optimization/57214
10843         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
10844         not propagate from SSA names that occur in abnormal PHI nodes.
10846 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
10848         * stor-layout.c (element_precision): New function.
10849         * machmode.h (element_precision): Declare it.
10850         * tree.c (build_minus_one_cst): New function.
10851         (element_precision): Likewise.
10852         * tree.h (build_minus_one_cst): Declare new function.
10853         (element_precision): Likewise.
10854         * fold-const.c (operand_equal_p): Use element_precision.
10855         (fold_binary_loc): Handle vector types.
10856         * convert.c (convert_to_integer): Use element_precision.
10857         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
10858         separately.
10860 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
10862         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
10863         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
10864         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
10865         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10866         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
10867         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
10868         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
10869         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10870         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
10871         (Uuw8): New constraints.
10872         (Usb4): Move into alphabetical order.
10873         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
10874         (sd8_operand, ub8_operand, uw8_operand): New predicates.
10875         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
10876         previously unnamed patterns.
10877         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
10878         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
10879         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
10880         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
10881         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
10882         of specific lengths.
10884 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
10886         * config/i386/i386.md (rotateinv): New code attr.
10887         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
10888         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
10889         roll $31, %eax, etc.
10891         PR tree-optimization/45216
10892         PR tree-optimization/57157
10893         * tree-ssa-forwprop.c (simplify_rotate): New function.
10894         (ssa_forward_propagate_and_combine): Call it.
10896 2013-05-10  Richard Biener  <rguenther@suse.de>
10898         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
10899         disable peeling when we version for aliasing.
10900         (vector_alignment_reachable_p): Honor explicit user alignment.
10901         (vect_supportable_dr_alignment): Likewise.
10902         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
10903         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
10904         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
10905         then peeling to arrange for the cost-model check to come first.
10907 2013-05-10  Alan Modra  <amodra@gmail.com>
10909         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
10910         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
10911         * configure: Regenerate.
10913 2013-05-10  Alan Modra  <amodra@gmail.com>
10915         PR target/55033
10916         * varasm.c (default_elf_select_section): Move !DECL_P check..
10917         (get_named_section): ..to here before calling get_section_name.
10918         Adjust assertion.
10919         (default_section_type_flags): Add DECL_P check.
10920         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
10921         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
10923 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
10925         * config/epiphany/epiphany.c (epiphany_expand_prologue):
10926         When using gen_stack_adjust_str with a register offset, add a
10927         REG_FRAME_RELATED_EXPR note.
10929 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
10931         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
10932         (*vec_extractv4si_zext_mem): Ditto.
10933         (*vec_extractv2di): Add 0->x and x->x alternatives.
10934         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
10935         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
10937 2013-05-09  Jason Merrill  <jason@redhat.com>
10939         N3639 C++1y VLA support
10940         * gimplify.c (gimplify_vla_decl): Don't touch an existing
10941         DECL_VALUE_EXPR.
10943         * tree.c (build_constructor_va): New.
10944         * tree.h: Declare it.
10946 2013-05-09  Martin Jambor  <mjambor@suse.cz>
10948         PR lto/57084
10949         * gimple-fold.c (canonicalize_constructor_val): Call
10950         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10952 2013-05-09  Jan Hubicka  <jh@suse.cz>
10953             Richard Biener  <rguenther@suse.de>
10955         PR lto/54095
10956         * symtab.c (symtab_make_decl_local): Do not add private names.
10958 2013-05-09  Jan Hubicka  <jh@suse.cz>
10960         PR lto/54095
10961         * symtab.c (insert_to_assembler_name_hash): Handle clones.
10962         (unlink_from_assembler_name_hash): Likewise.
10963         (symtab_prevail_in_asm_name_hash, symtab_register_node,
10964         symtab_unregister_node, symtab_initialize_asm_name_hash,
10965         change_decl_assembler_name): Update.
10967 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
10969         * config/aarch64/aarch64.md: New movtf split.
10970         (*movtf_aarch64): Update.
10971         (aarch64_movdi_tilow): Handle TF modes and rename to
10972         aarch64_movdi_<mode>low.
10973         (aarch64_movdi_tihigh): Handle TF modes and rename to
10974         aarch64_movdi_<mode>high
10975         (aarch64_movtihigh_di): Handle TF modes and rename to
10976         aarch64_mov<mode>high_di
10977         (aarch64_movtilow_di): Handle TF modes and rename to
10978         aarch64_mov<mode>low_di
10979         (aarch64_movtilow_tilow): Remove spurious whitespace.
10980         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10981         splits.
10982         (aarch64_print_operand): Update.
10984 2013-05-09  Alan Modra  <amodra@gmail.com>
10986         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10987         powerpc64le.
10988         * configure: Regenerate.
10990 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
10992         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10993         splitter preparation statements.
10994         * config/i386/sse.md (*vec_extract* splitters): Ditto.
10995         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10996         adjust_address_nv.
10998 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11000         * gimple-ssa-strength-reduction.c (count_candidates): Change
11001         return value to int.
11002         (analyze_candidates_and_replace): Change type of length to int.
11004 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
11006         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
11007         (*vec_extract<mode>): Use VI12_128 mode iterator.
11008         (*vec_extract<mode>_mem): Ditto.
11009         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
11010         attribute.
11012 2013-05-08  Diego Novillo  <dnovillo@google.com>
11014         PR bootstrap/54659
11016         Revert:
11017         2012-08-17  Diego Novillo  <dnovillo@google.com>
11019         PR bootstrap/54281
11020         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
11021         * config.in: Regenerate.
11022         * configure: Regenerate.
11023         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
11025 2013-05-08  Jan Hubicka  <jh@suse.cz>
11027         PR lto/54095
11028         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
11029         * cgraph.h (symtab_node_base): Add unique_name.
11030         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
11031         input_overwrite_node, input_varpool_node): Stream unique_name.
11032         * cgraphclones.c (cgraph_create_virtual_clone,
11033         cgraph_function_versioning): Set unique_name.
11034         * ipa.c (function_and_variable_visibility): Set unique_name.
11036 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11038         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
11039         (alloc_cand_and_find_basis): Restrict conditional candidate
11040         processing to CAND_MULTs.
11042 2013-05-08  Jan Hubicka  <jh@suse.cz>
11044         PR lto/54095
11045         lto-symtab.c (lto_symtab_symbol_p): New function.
11046         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
11047         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
11048         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
11049         Skip static symbols.
11051 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
11053         PR tree-optimization/57200
11054         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
11055         Only call inform if the preceding warning_at returns true.
11057 2013-05-07  Han Shen  <shenhan@google.com>
11059         * cfgexpand.c (record_or_union_type_has_array_p): New function.
11060         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
11061         * common.opt (fstack-protector-strong): New option.
11062         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
11063         * doc/invoke.texi (Optimization Options): Document
11064         "-fstack-protector-strong".
11065         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
11067 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
11069         * config/mips/mips.c (mips_machine_reorg2): Return 0.
11071 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
11073         * ira.c (update_equiv_regs): Add insn having equiv memory even if
11074         it is not lhs of the insn.
11075         (setup_reg_equiv): Remove insn having equiv memory which it is not
11076         lhs of the insn.
11077         * lra-constraints.c (process_address): Try to improve generation
11078         code for address base + disp.
11079         (lra_constraints): Make correct the code for checking insn setting
11080         up backward equivalence.  Remove insn only if it is in the init
11081         insn list.
11082         * lra-eliminations.c (update_reg_eliminate): Change return value.
11083         (lra_eliminate): Use the result.
11085 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
11087         * config/i386/sse.md (ssescalarnummask): New mode attribute.
11088         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
11089         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
11090         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
11091         register target operands.
11092         (*vec_extractv8hi_sse2): New pattern.
11093         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
11094         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
11095         (*vec_extract<mode>_mem): New insn and split pattern.
11097 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
11099         * config/arm/arm.c (arm_asan_shadow_offset): New function.
11100         (TARGET_ASAN_SHADOW_OFFSET): Define.
11101         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
11102         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
11104 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11106         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
11107         (incr_vec_index): Return -1 if increment not found.
11108         (create_add_on_incoming_edge): Assert if increment not found.
11109         (record_increment): Limit number of increments recorded.
11110         (all_phi_incrs_profitable): Return false if an increment not found.
11111         (replace_profitable_candidates): Don't process increments that were
11112         not recorded.
11113         (analyze_candidates_and_replace): Limit size of incr_vec.
11115 2013-05-07  Richard Biener  <rguenther@suse.de>
11117         * calls.c (special_function_p): setjmp-like functions are leaf.
11118         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
11119         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
11121 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
11123         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
11124         (aarch64_simd_mov<mode>): New expander.
11125         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
11126         (aarch64_simd_mov_to_<mode>high): Likewise.
11127         (aarch64_simd_mov_from_<mode>low): Likewise.
11128         (aarch64_simd_mov_from_<mode>high): Likewise.
11129         (aarch64_dup_lane<mode>): Update.
11130         (aarch64_dup_lanedi): New instruction pattern.
11131         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
11132         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
11134 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11136         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
11137         (replace_mult_candidate): Remove unnecessary argument; remove
11138         unnecessary parameter from call to introduce_cast_before_cand.
11139         (replace_unconditional_candidate): Remove unnecessary parameter
11140         from call to replace_mult_candidate.
11141         (replace_conditional_candidate): Likewise.
11142         (insert_initializers): Use make_temp_ssa_name.
11143         (introduce_cast_before_cand): Remove unnecessary argument; use
11144         make_temp_ssa_name.
11145         (replace_one_candidate): Remove unnecessary argument; remove
11146         unnecessary parameter from calls to introduce_cast_before_cand.
11147         (replace_profitable_candidates): Remove unnecessary parameters
11148         from calls to replace_one_candidate.
11150 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11152         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11153         phi def as possibly hiding a basis for a CAND_ADD whose operands
11154         have been commuted in the analysis.
11155         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
11157 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11159         * config/aarch64/aarch64.md
11160         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
11161         shift value between 0-4.
11163 2013-05-07  Richard Biener  <rguenther@suse.de>
11165         * double-int.h (rshift): New overload.
11166         * double-int.c (rshift): New function.
11167         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
11168         (create_reference_ops_from_ref): Remove.
11169         (vn_reference_insert): Use shared ops for constructing the
11170         reference and copy it.
11172 2013-05-07  Richard Biener  <rguenther@suse.de>
11174         PR middle-end/57190
11175         * tree-eh.c (sink_clobbers): Properly propagate
11176         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
11178 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
11180         PR tree-optimization/57149
11181         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
11182         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
11183         collect_phi_def_edges, execute_late_warn_uninitialized): Use
11184         uninit_undefined_value_p instead of ssa_undefined_value_p.
11186         PR debug/57184
11187         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
11188         for modifier == EXPAND_INITIALIZER.
11190 2013-05-07  Anton Blanchard  <anton@samba.org>
11192         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
11193         for powerpc64 little endian.
11194         * configure: Regenerate.
11196 2013-05-06  Graham Stott  <grahams@btinternet.com>
11198         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
11199         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
11200         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
11201         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
11203 2013-05-06  Graham Stott  <grahams@btinternet.com>
11205         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
11206         codes which allow non-lvalues.
11208 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
11210         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
11211         components are all 1s.
11212         (integer_minus_onep): New function.
11213         * tree.h (integer_minus_onep): Declare it.
11214         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
11215         integer_minus_onep instead of integer_all_onesp.
11217 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
11219         PR target/52933
11220         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
11221         variations of these patterns.
11223 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
11225         * config/i386/i386.md (isa): Add x64_sse4 member.
11226         (enabled): Handle x64_sse4.
11227         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
11228         instruction for 64bit SSE4_1 targets.  Update insn attributes.
11229         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
11230         instruction for SSE4_1 targets.  Update insn attributes.
11231         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
11232         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
11233         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
11234         const_1 selector.
11235         (*vec_extractv4si): Rename from *sse4_1_pextrd.
11236         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
11237         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
11239 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
11241         PR target/57108
11242         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
11244 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
11246         * final.c (do_assembler_dialects): Don't handle curly braces and
11247         vertical bar escaped by % as dialect delimiters.
11248         (output_asm_insn): Print curly braces and vertical bar if escaped
11249         by % and ASSEMBLER_DIALECT defined.
11250         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
11251         * doc/tm.texi: Regenerated.
11253 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
11255         * config/mips/mips.c: Include tree-pass.h.
11256         (mips_reorg): Split in pre- and post-dbr_schedule parts.
11257         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
11258         (pass_mips_machine_reorg2): New machine specific pass.
11259         (insert_pass_mips_machine_reorg2): New pass plugin definition.
11260         (mips_option_override): Register the new pass.
11261         * rtl.h (cleanup_barriers): Remove prototype.
11262         (dbr_schedule): Likewise.
11263         * jump.c (cleanup_barriers): Make static.
11264         * reorg.c (dbr_schedule): Likewise.
11266 2013-05-06  Richard Biener  <rguenther@suse.de>
11268         PR tree-optimization/57185
11269         * tree-parloops.c (add_field_for_reduction): Handle anonymous
11270         SSA names properly.
11272 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
11274         PR target/57106
11275         * config/i386/i386.c (add_parameter_dependencies): Add dependence
11276         between "first_arg" and "insn", not "last" and "insn".
11278 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11280         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
11281         (find_candidates_in_block): Re-enable slsr_process_phi.
11282         (create_phi_basis): Fix double counting of candidate adjustment.
11284 2013-05-06  Richard Biener  <rguenther@suse.de>
11286         PR middle-end/57147
11287         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
11288         the edge is also fallthru, preserve it and just clear the
11289         abnormal flag.
11290         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
11291         also complex, preserve that and just clear the fallthru flag.
11292         * tree-inline.c (update_ssa_across_abnormal_edges): Also
11293         update virtual operands.
11295 2013-05-06  Alan Modra  <amodra@gmail.com>
11297         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
11298         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
11299         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
11300         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
11301         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
11302         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
11304 2013-05-06  Alan Modra  <amodra@gmail.com>
11306         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
11307         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
11308         (DEFAULT_ASM_ENDIAN): Define.
11309         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
11310         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
11311         Update -K PIC clause from sysv4.h.
11312         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
11313         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
11315 2013-05-06  Alan Modra  <amodra@gmail.com>
11317         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
11318         twice for little-endian.
11319         (ashrdi3_no_power, ashrdi3): Support little-endian.
11321 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
11323         PR target/55303
11324         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
11325         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
11326         related expanders.
11327         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
11328         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
11329         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
11330         New predicates.
11332 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
11333             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
11335         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
11336         * config/pa/pa.opt: Make mbig-switch a no-op.
11337         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
11338         (CASE_VECTOR_MODE): Always return SImode.
11339         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
11340         for the !TARGET_BIG_SWITCH case.
11341         * config/pa/pa-linux.h: Likewise.
11342         * config/pa/pa-openbsd.h: Likewise.
11343         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
11344         * config/pa/pa.md (short_jump): Remove define_insn.
11345         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
11346         (casesi0): Remove define_insn.
11347         (type): Remove btable_branch.
11348         (pa_combine_type): Likewise.
11349         (in_nullified_branch_delay): Likewise.
11350         (in_call_delay): Likewise.
11351         (define_delay): Likewise.
11352         (define_insn_reservation "Z3"): Likewise.
11353         (define_insn_reservation "Z4"): Likewise.
11354         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
11355         (pa_adjust_insn_length): Remove adjustment for btable branches.
11356         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
11357         and mno-big-switch
11359 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
11361         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
11362         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
11363         Add m->r,x alternatives.
11364         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
11365         splitters using SWI48x mode iterator.
11366         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
11367         TARGET_64BIT.  Add m->x alternative.
11368         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
11369         Add o->x alternative.  Enable for TARGET_SSE.
11370         (sse_storeq): Remove expander.
11371         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
11372         with memory input operand.
11373         (*vec_extractv2di_1 splitter): New.
11374         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
11375         * config/i386/i386.md (ssevecmodelower): New mode attribute.
11377 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
11379         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
11380         (INT_LOWPART): Delete.
11381         (extract_MB): Adjust.
11382         (extract_ME): Adjust.
11383         (print_operand): Adjust.
11385 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
11387         * config/rs6000/predicates.md (reg_or_add_cint_operand,
11388         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
11389         (reg_or_logical_cint_operand, easy_fp_constant,
11390         logical_const_operand): Delete "CONST_DOUBLE" case.
11391         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
11392         "HOST_BITS_PER_WIDE_INT == 64" test.
11393         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
11394         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
11395         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
11396         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
11397         test.
11398         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
11399         CONST_DOUBLE DImode/VOIDmode case.
11400         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
11401         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
11402         CONST_DOUBLE VOIDmode case.
11403         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
11404         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
11405         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
11406         Delete CONST_DOUBLE case.
11407         (splitters for mov FMOVE64 const_double): Delete
11408         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
11409         "HOST_BITS_PER_WIDE_INT >= 64" test.
11410         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
11411         case.
11412         (mov DI const_double): Delete.
11414 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
11416         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
11417         on op shows all bits zero in mode of a lowpart subreg, return zero.
11419 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
11421         PR target/57150
11422         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
11423         to save TFmode registers and DImode to save TImode registers for
11424         caller save operations.
11425         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
11426         mark being partially clobbered since they only use the first
11427         double word.
11429         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
11430         and TDmode only use the upper 64-bits of each VSX register.
11432 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11434         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
11435         (find_candidates_in_block): Disable slsr_process_phi.
11437 2013-05-03  Guozhi Wei  <carrot@google.com>
11439         * coverage.c (coverage_obj_init): Move the construction of gcov
11440         constructor to ...
11441         (build_init_ctor): ... here.
11443 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11445         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
11446         (slsr_cand_d): Redefine def_phi.
11447         (stride_status, phi_adjust_status, count_phis_status): New enums.
11448         (find_phi_def): New.
11449         (find_basis_for_base_expr): New.
11450         (find_basis_for_candidate): Handle hidden bases.
11451         (alloc_cand_and_find_basis): Handle phi candidates.
11452         (slsr_process_phi): New.
11453         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
11454         (create_mul_imm_cand): Likewise.
11455         (create_add_ssa_cand): Exclude phi base candidates.
11456         (create_add_imm_cand): Likewise.
11457         (slsr_process_cast): Likewise.
11458         (slsr_process_copy): Likewise.
11459         (find_candidates_in_block): Handle phi candidates.
11460         (dump_candidate): Likewise.
11461         (unconditional_cands): Delete.
11462         (unconditional_cands_with_known_stride_p): Delete.
11463         (phi_dependent_cand_p): New.
11464         (cand_increment): Handle phi-dependent candidates.
11465         (replace_dependent): Delete.
11466         (replace_mult_candidate): New.
11467         (replace_unconditional_candidate): New.
11468         (incr_vec_index): Move to avoid forward reference.
11469         (create_add_on_incoming_edge): New.
11470         (create_phi_basis): New.
11471         (replace_dependents): Delete.
11472         (replace_conditional_candidate): New.
11473         (phi_add_costs): New.
11474         (replace_uncond_cands_and_profitable_phis): New.
11475         (record_increment): Handle phi adjustments.
11476         (record_phi_increments): New.
11477         (record_increments): Handle phi adjustments.
11478         (phi_incr_cost): New.
11479         (lowest_cost_path): Handle phis.
11480         (total_savings): Likewise.
11481         (analyze_increments): Likewise.
11482         (ncd_with_phi): New.
11483         (ncd_of_cand_and_phis): New.
11484         (nearest_common_dominator_for_cands): Handle phi increments.
11485         (all_phi_incrs_profitable): New.
11486         (replace_profitable_candidates): Handle phi-dependent candidates.
11487         (analyze_candidates_and_replace): Likewise.
11489 2013-05-03  Teresa Johnson  <tejohnson@google.com>
11491         PR bootstrap/57154
11492         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
11493         do not exceed REG_BR_PROB_BASE.
11495 2013-05-03  Jeff Law  <law@redhat.com>
11497         PR tree-optimization/57144
11498         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
11499         operand of the condition will bit into the new type when eliminating
11500         a cast feeding a condition.
11502 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
11504         PR rtl-optimization/57130
11505         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
11506         of COMPARE as in_code to the recursive call if needed.
11508 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
11510         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
11511         (enabled): Handle new members.
11512         * config/i386/sse.md (*vec_concatv2si): Merge from
11513         *vec_concatv2si_sse2 and vec_concatv2si_sse.
11514         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
11516 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
11518         PR tree-optimization/57027
11519         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
11520         for fnms opportunity, check we got the prerequisite kind
11521         of tree / gimple before using accessor functions.
11523 2013-05-03  Richard Biener  <rguenther@suse.de>
11525         * double-int.h (lshift): New overload without precision
11526         and arith argument.
11527         (operator *=, operator +=, operator -=): Move ...
11528         * double-int.c (operator *=, operator +=, operator -=): ... here
11529         and implement more efficiently.
11530         (mul_double_with_sign): Remove.
11531         (lshift_double): Adjust to take unsinged shift argument, push
11532         dispatching code to callers.
11533         (mul_double_wide_with_sign): Add early out for callers that
11534         are not interested in high parts or overflow.
11535         (lshift): New function.
11536         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
11537         dispatch code here.
11538         (lrotate, rrotate): Use logical shifts.
11539         * expr.c (get_inner_reference): Use lshift.
11540         * fixed-value.c (do_fixed_divide): Likewise.
11541         * tree-dfa.c (get_ref_base_and_extent): Likewise.
11542         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
11543         (indirect_refs_may_alias_p): Likewise.
11544         (stmt_kills_ref_p_1): Likewise.
11546 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
11548         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
11550 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
11552         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
11553         scalar form of FABD instruction.
11555 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
11557         * lra-constraints.c (process_alt_operands): Add checking alt
11558         number to choose the best alternative.
11560 2013-05-02  Richard Biener  <rguenther@suse.de>
11562         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
11563         bitmap and its handling.
11564         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
11566 2013-05-02  Richard Biener  <rguenther@suse.de>
11568         PR middle-end/57140
11569         * tree-inline.c (copy_loops): Properly handle removed loops.
11570         (copy_cfg_body): Mark destination loops for fixup if source
11571         loops needed fixup.
11573 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
11575         PR target/56732
11576         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
11577         generating simple_return for naked functions.
11579 2013-05-02  Martin Jambor  <mjambor@suse.cz>
11581         PR middle-end/56988
11582         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
11583         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
11584         flags match.
11585         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
11586         ipa_agg_replacement_value structures.
11587         (known_aggs_to_agg_replacement_list): Likewise.
11588         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
11589         (read_agg_replacement_chain): Likewise.
11590         (ipcp_transform_function): Also check that by_ref flags match.
11592 2013-05-02  Richard Biener  <rguenther@suse.de>
11594         * graphds.h (struct graph): Add obstack member.
11595         * graphds.c (new_graph): Initialize obstack and allocate
11596         vertices from it.
11597         (add_edge): Allocate edge from the obstack.
11598         (free_graph): Free the obstack instead of all edges and vertices.
11600 2013-05-02  Teresa Johnson  <tejohnson@google.com>
11602         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
11603         divides.
11604         * cfg.c (update_bb_profile_for_threading): Ditto.
11605         * tree-inline.c (copy_bb): Ditto.
11606         (copy_edges_for_bb): Ditto.
11607         (initialize_cfun): Ditto.
11608         (copy_cfg_body): Ditto.
11609         (expand_call_inline): Ditto.
11610         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
11611         (estimate_node_size_and_time): Ditto.
11612         (inline_merge_summary): Ditto.
11613         * cgraphclones.c (cgraph_clone_edge): Ditto.
11614         (cgraph_clone_node): Ditto.
11615         * sched-rgn.c (compute_dom_prob_ps): Ditto.
11616         (compute_trg_info): Ditto.
11618 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
11620         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
11621         S reg when fp attribute set.
11622         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
11624 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
11626         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
11627         New pattern.
11628         (*and_one_cmplsi3_compare0_uxtw): Likewise.
11629         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
11630         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
11632 2013-05-02  Richard Biener  <rguenther@suse.de>
11634         * tree-scalar-evolution.c (scev_info_hasher): Remove.
11635         (struct instantiate_cache_entry): New type.
11636         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
11637         (struct instantiate_cache_type): New type.
11638         (set_instantiated_value, get_instantiated_value): Remove.
11639         (get_instantiated_value_entry): New function.
11640         (instantiate_scev_name): Use the new cache and adjust.
11641         (instantiate_scev_poly): Adjust.
11642         (instantiate_scev_binary): Likewise.
11643         (instantiate_array_ref): Likewise.
11644         (instantiate_scev_convert): Likewise.
11645         (instantiate_scev_not): Likewise.
11646         (instantiate_scev_3): Likewise.
11647         (instantiate_scev_2): Likewise.
11648         (instantiate_scev_r): Likewise.
11649         (instantiate_scev): Likewise.
11650         (resolve_mixers): Likewise.
11652 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
11654         PR target/57091
11655         * lra-constraints.c (best_small_class_operands_num): Remove.
11656         (process_alt_operands): Remove small_class_operands_num.  Take
11657         small classes operands into losers and only if the operand is not
11658         matched.  Modify debugging output.
11659         (curr_insn_transform): Remove best_small_class_operands_num.
11660         Print insn name.
11662 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11664         * config/aarch64/aarch64-builtins.c
11665         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
11666         * config/aarch64/aarch64-simd-builtins.def
11667         (reduc_splus_): Add new modes.
11668         (reduc_uplus_): New.
11669         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
11670         (reduc_uplus_v4sf): Likewise.
11671         (reduc_splus_v4sf): Likewise.
11672         (aarch64_addv<mode>): Likewise.
11673         (reduc_uplus_<mode>): Likewise.
11674         (reduc_splus_<mode>): Likewise.
11675         (aarch64_addvv2di): Likewise.
11676         (reduc_uplus_v2di): Likewise.
11677         (reduc_splus_v2di): Likewise.
11678         (aarch64_addvv2si): Likewise.
11679         (reduc_uplus_v2si): Likewise.
11680         (reduc_splus_v2si): Likewise.
11681         (reduc_<sur>plus_<mode>): New.
11682         (reduc_<sur>plus_v2di): Likewise.
11683         (reduc_<sur>plus_v2si): Likewise.
11684         (reduc_<sur>plus_v4sf): Likewise.
11685         (aarch64_addpv4sf): Likewise.
11686         * config/aarch64/arm_neon.h
11687         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
11688         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
11689         add UNSPEC_SADDV, UNSPEC_UADDV.
11690         (SUADDV): New.
11691         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
11693 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11695         * config/aarch64/arm_neon.h
11696         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
11698 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11700         * config/aarch64/aarch64-builtins
11701         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
11703 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11705         * config/aarch64/aarch64-simd-builtins.def
11706         (reduc_smax_): New.
11707         (reduc_smin_): Likewise.
11708         (reduc_umax_): Likewise.
11709         (reduc_umin_): Likewise.
11710         (reduc_smax_nan_): Likewise.
11711         (reduc_smin_nan_): Likewise.
11712         (fmax): Remove.
11713         (fmin): Likewise.
11714         (smax): Update for V2SF, V4SF and V2DF modes.
11715         (smin): Likewise.
11716         (smax_nan): New.
11717         (smin_nan): Likewise.
11718         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
11719         (<su><maxmin><mode>3): ...This, refactor.
11720         (s<maxmin><mode>3): New.
11721         (<maxmin_uns><mode>3): Likewise.
11722         (reduc_<maxmin_uns>_<mode>): Refactor.
11723         (reduc_<maxmin_uns>_v4sf): Likewise.
11724         (reduc_<maxmin_uns>_v2si): Likewise.
11725         (aarch64_<fmaxmin><mode>: Remove.
11726         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
11727         new builtin names.
11728         (vmin<q>_f<32,64>): Likewise.
11729         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
11730         (FMAXMIN): New.
11731         (su): Add mappings for smax, smin, umax, umin.
11732         (maxmin): New.
11733         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
11734         (FMAXMIN): Rename as...
11735         (FMAXMIN_UNS): ...This.
11736         (maxminv): Remove.
11737         (fmaxminv): Likewise.
11738         (fmaxmin): Likewise.
11739         (maxmin_uns): New.
11740         (maxmin_uns_op): Likewise.
11742 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11744         * config/aarch64/arm_neon.h
11745         (vac<ge, gt><sd>_f<32, 64>): Rename to...
11746         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
11747         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
11749 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11751         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
11752         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
11754 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11756         * config/aarch64/aarch64-simd.md
11757         (vcond<mode>_internal): Handle special cases for constant masks.
11758         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
11759         (vcondu<mode><mode>): Likewise.
11760         (vcond<v_cmp_result><mode>): New.
11762 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11764         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
11765         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
11766         * config/aarch64/aarch64-simd-builtins.def
11767         (cmeq): Update to BUILTIN_VALLDI.
11768         (cmgt): Likewise.
11769         (cmge): Likewise.
11770         (cmle): Likewise.
11771         (cmlt): Likewise.
11772         * config/aarch64/arm_neon.h
11773         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
11774         to builtins or C as appropriate.
11776 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
11778         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
11779         (cmgeu): ...This.
11780         (cmhi): Rename to...
11781         (cmgtu): ...This.
11782         * config/aarch64/aarch64-simd.md
11783         (simd_mode): Add SF.
11784         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
11785         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
11786         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
11787         (cstore<mode>_neg): ...This.
11788         * config/aarch64/iterators.md
11789         (VALLF): new.
11790         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
11791         (COMPARISONS): New.
11792         (UCOMPARISONS): Likewise.
11793         (optab): Add missing comparisons.
11794         (n_optab): New.
11795         (cmp_1): Likewise.
11796         (cmp_2): Likewise.
11797         (CMP): Likewise.
11798         (cmp): Remove.
11799         (VCMP_S): Likewise.
11800         (VCMP_U): Likewise.
11801         (V_cmp_result): Add DF, SF modes.
11802         (v_cmp_result): Likewise.
11803         (v): Likewise.
11804         (vmtype): Likewise.
11805         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
11807 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
11809         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
11810         define_insn to define_insn_and_split.
11811         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
11812         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
11813         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
11814         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
11815         (thumb2_negscc): Likewise.
11817 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
11819         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
11821 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
11823         * config/arm/thumb2.md: Remove trailing whitespaces.
11825 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11827         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
11828         Use gen_int_mode rather than GEN_INT.
11830 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
11832         * value-prof.c (stream_in_histogram_value): Remove the strayed
11833         debug_gimple_stmt.
11835 2013-04-30  Richard Biener  <rguenther@suse.de>
11837         PR middle-end/57122
11838         * cfghooks.c (split_edge): Properly check for the loop latch edge.
11840 2013-04-30  Richard Biener  <rguenther@suse.de>
11842         PR middle-end/57107
11843         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
11845 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
11847         PR rtl-optimization/56957
11848         PR rtl-optimization/57105
11849         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
11850         variable.  Use just INSN_UID for determining whether an insn
11851         should be only disconnected from the insn stream.
11852         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
11854 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
11856         PR tree-optimization/57104
11857         * tsan.c (instrument_expr): Don't instrument accesses to
11858         DECL_HARD_REGISTER VAR_DECLs.
11860 2013-04-30  Richard Biener  <rguenther@suse.de>
11862         * function.h (loops_for_fn): New inline function.
11863         (set_loops_for_fn): Likewise.
11864         * cfgloop.h (place_new_loop): Add struct function parameter.
11865         (get_loop): Likewise.
11866         (get_loops): Likewise.
11867         (number_of_loops): Likewise.
11868         (fel_next): Adjust.
11869         (fel_init): Likewise.
11870         * cfg.c (get_loop_copy): Adjust.
11871         * cfgloop.c (flow_loops_dump): Likewise.
11872         (record_loop_exits): Likewise.
11873         (verify_loop_structure): Likewise.
11874         * cfgloopanal.c (mark_irreducible_loops): Likewise.
11875         (estimate_reg_pressure_cost): Likewise.
11876         (mark_loop_exit_edges): Likewise.
11877         * cfgloopmanip.c (place_new_loop): Likewise.
11878         (add_loop): Likewise.
11879         (duplicate_loop): Likewise.
11880         * graph.c (draw_cfg_nodes): Likewise.
11881         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
11882         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
11883         (extract_affine_chrec): Likewise.
11884         (build_scop_iteration_domain): Likewise.
11885         * graphite.c (graphite_initialize): Likewise.
11886         * ira-build.c (create_loop_tree_nodes): Likewise.
11887         (more_one_region_p): Likewise.
11888         (rebuild_regno_allocno_maps): Likewise.
11889         (mark_loops_for_removal): Likewise.
11890         (mark_all_loops_for_removal): Likewise.
11891         (remove_unnecessary_regions): Likewise.
11892         (ira_build): Likewise.
11893         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
11894         * loop-init.c (fix_loop_structure): Likewise.
11895         (gate_rtl_move_loop_invariants): Likewise.
11896         (gate_rtl_unswitch): Likewise.
11897         (gate_rtl_unroll_and_peel_loops): Likewise.
11898         (rtl_doloop): Likewise.
11899         * lto-streamer-in.c (input_cfg): Likewise.
11900         * lto-streamer-out.c (output_cfg): Likewise.
11901         * modulo-sched.c (sms_schedule): Likewise.
11902         * predict.c (tree_estimate_probability): Likewise.
11903         (tree_estimate_probability_driver): Likewise.
11904         (estimate_loops): Likewise.
11905         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
11906         (move_sese_region_to_fn): Likewise.
11907         (debug_loop_num): Likewise.
11908         * tree-chrec.c (chrec_evaluate): Likewise.
11909         (hide_evolution_in_other_loops_than_loop): Likewise.
11910         (chrec_component_in_loop_num): Likewise.
11911         (reset_evolution_in_loop): Likewise.
11912         (evolution_function_is_invariant_rec_p): Likewise.
11913         * tree-if-conv.c (main_tree_if_conversion): Likewise.
11914         * tree-inline.c (copy_loops): Likewise.
11915         (copy_cfg_body): Likewise.
11916         (tree_function_versioning): Likewise.
11917         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
11918         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
11919         Likewise.
11920         (add_to_evolution_1): Likewise.
11921         (scev_const_prop): Likewise.
11922         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
11923         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
11924         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
11925         (tree_ssa_lim_initialize): Likewise.
11926         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
11927         (verify_loop_closed_ssa): Likewise.
11928         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
11929         (tree_ssa_loop_im): Likewise.
11930         (tree_ssa_loop_unswitch): Likewise.
11931         (tree_vectorize): Likewise.
11932         (check_data_deps): Likewise.
11933         (tree_ssa_loop_ivcanon): Likewise.
11934         (tree_ssa_loop_bounds): Likewise.
11935         (tree_complete_unroll): Likewise.
11936         (tree_complete_unroll_inner): Likewise.
11937         (tree_parallelize_loops): Likewise.
11938         (tree_ssa_loop_prefetch): Likewise.
11939         (tree_ssa_loop_ivopts): Likewise.
11940         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
11941         * tree-vectorizer.c (vectorize_loops): Likewise.
11943 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
11945         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11946         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11947         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11948         with EABI_LINK_SPEC.
11950 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11952         PR target/44578
11953         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11954         alternative.
11956 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
11958         PR target/57097
11959         * lra-constraints.c (process_alt_operands): Discourage a bit more
11960         using memory for pseudos.  Print cost dump for alternatives.
11961         Modify cost values for conflicts with early clobbers.
11962         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11964 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
11966         PR target/57098
11967         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11969 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11971         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11972         from/to S register.
11973         (movdi_aarch64): Support LDR/STR from/to D register.
11975 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
11977         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11978         or higher by default.
11980 2013-04-29  Richard Biener  <rguenther@suse.de>
11982         PR middle-end/57075
11983         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11984         even if not adding abnormal edges for calls that can make
11985         abnormal gotos.
11987 2013-04-29  Richard Biener  <rguenther@suse.de>
11989         PR middle-end/57103
11990         * tree-cfg.c (move_stmt_op): Fix condition under which to update
11991         TREE_BLOCK.
11992         (move_stmt_r): Remove redundant checking.
11994 2013-04-29  Teresa Johnson  <tejohnson@google.com>
11996         PR bootstrap/57077
11997         * basic-block.h (apply_scale): New function.
11998         (apply_probability): Use apply_scale.
11999         * gimple-streamer-in.c (input_bb): Ditto.
12000         * lto-streamer-in.c (input_cfg): Ditto.
12001         * lto-cgraph.c (merge_profile_summaries): Ditto.
12002         * tree-optimize.c (execute_fixup_cfg): Ditto.
12003         * tree-inline.c (copy_bb): Update comment to use apply_scale.
12004         (copy_edges_for_bb): Ditto.
12005         (copy_cfg_body): Ditto.
12007 2013-04-29  Tom de Vries  <tom@codesourcery.com>
12009         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
12010         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
12011         (tail_merge_optimize): Handle current_loops == NULL.
12013 2013-04-26  Jeff Law  <law@redhat.com>
12015         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
12016         (simplify_cond_using_ranges): Generalize code to simplify
12017         COND_EXPRs where one argument is a constant and the other
12018         is an SSA_NAME created by an integral type conversion.
12020 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12022         * config/arm/arm.md (store_minmaxsi): Use only when
12023         optimize_insn_for_size_p.
12025 2013-04-29  Christian Bruel  <christian.bruel@st.com>
12027         PR target/57108
12028         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
12030 2013-04-29  Richard Biener  <rguenther@suse.de>
12032         PR middle-end/57089
12033         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
12034         loop tree make sure to schedule a fixup for the child as well.
12035         (expand_omp_for_generic): Properly add loops.
12036         (expand_omp_for_static_nochunk): Likewise.
12037         (expand_omp_for_static_chunk): Likewise.
12038         (expand_omp_for): For the degenerate case fixup loops.
12039         (expand_omp_sections): Fix default bb placement in loops.
12040         (expand_omp_atomic_pipeline): Properly add loops.
12042 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12044         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
12046 2013-04-29  Tom de Vries  <tom@codesourcery.com>
12048         * tree-ssa-tail-merge.c: Update header comment.
12050 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12052         * config/aarch64/arm_neon.h
12053         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
12054         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
12055         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
12056         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
12057         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
12058         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
12059         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
12060         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
12062 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12064         * config/aarch64/aarch64-simd.md
12065         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
12066         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
12067         fix_trunc, fixuns_trunc.
12068         (ftrunc<VDQF:mode>2): New.
12069         * config/aarch64/iterators.md (optab): Add fix, fixuns.
12070         (fix_trunc_optab): New.
12072 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12074         * config/aarch64/aarch64-builtins.c
12075         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
12076         iceilf, lround, iroundf.
12078 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
12080         PR target/54349
12081         * config/i386/i386.h (enum ix86_tune_indices)
12082         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
12083         New, split from X86_TUNE_INTER_UNIT_MOVES.
12084         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
12085         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
12086         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
12087         (TARGET_INTER_UNIT_MOVES): Remove.
12088         * config/i386/i386.c (initial_ix86_tune_features): Update.
12089         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
12090         (ix86_expand_convert_uns_didf_sse): Use
12091         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
12092         (ix86_expand_vector_init_one_nonzero): Ditto.
12093         (ix86_expand_vector_init_interleave): Ditto.
12094         (inline_secondary_memory_needed): Return true for moves from SSE class
12095         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
12096         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12097         * config/i386/constraints.md (Yi, Ym): Depend on
12098         TARGET_INTER_UNIT_MOVES_TO_VEC.
12099         (Yj, Yn): New constraints.
12100         * config/i386/i386.md (*movdi_internal): Change constraints of
12101         operand 1 from Yi to Yj and from Ym to Yn.
12102         (*movsi_internal): Ditto.
12103         (*movdf_internal): Ditto.
12104         (*movsf_internal): Ditto.
12105         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
12106         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
12107         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
12108         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
12109         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
12110         * config/i386/sse.md (movdi_to_sse): Ditto.
12111         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
12112         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
12113         TARGET_INTER_UNIT_MOVES.
12114         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
12115         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
12116         instead of TARGET_INTER_UNIT_MOVES.
12117         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
12118         operand 1 from Yi to Yj and from Ym to Yn.
12120 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12122         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
12123         (float_truncate_hi_): Likewise.
12124         (float_extend_lo_): Likewise.
12125         (float_truncate_lo_): Likewise.
12126         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
12127         (aarch64_float_extend_lo_v2df): Likewise.
12128         (vec_unpacks_hi_v4sf): Likewise.
12129         (aarch64_float_truncate_lo_v2sf): Likewise.
12130         (aarch64_float_truncate_hi_v4sf): Likewise.
12131         (vec_pack_trunc_v2df): Likewise.
12132         (vec_pack_trunc_df): Likewise.
12134 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12136         * config/aarch64/aarch64-builtins.c
12137         (aarch64_fold_builtin): Fold float conversions.
12138         * config/aarch64/aarch64-simd-builtins.def
12139         (floatv2si, floatv4si, floatv2di): New.
12140         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
12141         * config/aarch64/aarch64-simd.md
12142         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
12143         * config/aarch64/iterators.md (FLOATUORS): New.
12144         (optab): Add float, floatuns.
12145         (su_optab): Likewise.
12147 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12149         * config/aarch64/aarch64-builtins.c
12150         (aarch64_builtin_vectorized_function): Use new names for
12151         fcvt builtins.
12152         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
12153         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
12154         (fcvtzu): Split as...
12155         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
12156         (fcvtas): Split as...
12157         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
12158         (fcvtau): Split as...
12159         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
12160         (fcvtps): Split as...
12161         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
12162         (fcvtpu): Split as...
12163         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
12164         (fcvtms): Split as...
12165         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
12166         (fcvtmu): Split as...
12167         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
12168         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
12169         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
12170         (lfrintnusf, lfrintnudf): Likewise.
12171         * config/aarch64/aarch64-simd.md
12172         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
12173         define_insn.
12174         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
12175         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
12176         (fcvt_pattern): Likewise.
12178 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12180         * config/aarch64/aarch64-simd.md
12181         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
12182         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
12184 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12186         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
12187         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
12188         (vrnd<a,m,n,p>_f32): Implement using builtins.
12189         (vrnd<i,x><q>_f<32, 64>): New.
12191 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
12193         * config/aarch64/aarch64-builtins.c
12194         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
12195         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
12196         (frintz): Rename to...
12197         (btrunc): ...this.
12198         (frintp): Rename to...
12199         (ceil): ...this.
12200         (frintm): Rename to...
12201         (floor): ...this.
12202         (frinti): Rename to...
12203         (nearbyint): ...this.
12204         (frintx): Rename to...
12205         (rint): ...this.
12206         (frinta): Rename to...
12207         (round): ...this.
12208         * config/aarch64/aarch64-simd.md
12209         (aarch64_frint<frint_suffix><mode>): Delete.
12210         (<frint_pattern><mode>2): Convert to insn.
12211         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
12212         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
12213         (frint_pattern): Likewise.
12214         (frint_suffix): Likewise.
12216 2013-04-29  Richard Biener  <rguenther@suse.de>
12218         PR tree-optimization/57081
12219         * loop-init.c: Include tree-flow.h.
12220         (loop_optimizer_finalize): Free number of iteration estimates.
12221         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
12223 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
12225         PR tree-optimization/57083
12226         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
12227         non-singleton shift count range, zero extend low_bound for uns case.
12229         * config/i386/predicates.md (general_vector_operand): New predicate.
12230         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
12231         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
12232         if they aren't nonimmediate operands.  If their original values
12233         satisfy const_vector_equal_evenodd_p, don't shift them.
12234         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
12235         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
12236         if not nonimmediate_operand.
12237         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
12238         instead of register_operand.
12239         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
12241 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
12243         * stor-layout.c (finalize_size_functions): Allocate a structure and
12244         reset cfun before dumping the functions.
12246 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
12248         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
12250         PR target/56866
12251         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
12252         use xop_pmacsdqh if uns_p.
12253         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
12254         the immediate rotate count.
12256 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
12258         * rtl.h (struct rtx_def): Add comment for field jump.
12259         (LRA_SUBREG_P): New macro.
12260         * recog.c (register_operand): Check LRA_SUBREG_P.
12261         * lra.c (lra): Add note at the end of RTL code. Align non-empty
12262         stack frame.
12263         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
12264         (lra_final_code_change): Skip subreg change for operators.
12265         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
12266         if there are no operand changes.
12267         * lra-constraints.c (curr_insn_set): New.
12268         (match_reload): Set LRA_SUBREG_P.
12269         (emit_spill_move): Ditto.
12270         (check_and_process_move): Use curr_insn_set. Process only single
12271         set insns.  Don't initialize sec_mem_p and change_p.
12272         (simplify_operand_subreg): Use LRA_SUBREG_P.
12273         (reg_in_class_p): New function.
12274         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
12275         of #ifdef.  Add code to remove cycling.
12276         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
12277         non-null disp.  Reload inner instead of disp when base and index
12278         are null.  Try to put lo_sum into register.
12279         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
12280         (check_and_process_move): Move code for move cost check to
12281         simple_move_p.  Remove equiv_substitution.
12282         (simple_move_p): New function.
12283         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
12284         curr_insn_set.  Call check_and_process_move only for single set
12285         insns.  Use the new function.  Move call of check_and_process_move
12286         after operand equiv substitution and address process.
12288 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
12290         PR go/57045
12291         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
12292         with nonlocal goto receivers or returns twice calls, ignore
12293         unininitialized values from abnormal edges to nl goto receiver
12294         or returns twice call.
12296 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
12298         PR tree-optimization/57051
12299         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
12300         and VEC_RSHIFT_EXPR if shift count is a multiple of element
12301         bitsize.
12303 2013-04-26  Richard Biener  <rguenther@suse.de>
12305         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
12306         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
12307         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
12308         (fixup_loop_arrays_after_move): New function.
12309         (move_sese_region_to_fn): Properly outline the loop tree parts
12310         of the SESE region.
12312 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
12314         * config/i386/i386.md (type, unit): Fix long lines.
12316 2013-04-26  Richard Biener  <rguenther@suse.de>
12318         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
12319         (lto-streamer-out.o): Likewise.
12320         * cfgloop.c (init_loops_structure): Export, add struct function
12321         argument and adjust.
12322         (flow_loops_find): Adjust.
12323         * cfgloop.h (enum loop_estimation): Add EST_LAST.
12324         (init_loops_structure): Declare.
12325         * lto-streamer-in.c: Include cfgloop.h.
12326         (input_cfg): Input the loop tree.
12327         * lto-streamer-out.c: Include cfgloop.h.
12328         (output_cfg): Output the loop tree.
12329         (output_struct_function_base): Do not drop PROP_loops.
12331 2013-03-26  Richard Biener  <rguenther@suse.de>
12333         * tree-cfg.c (execute_build_cfg): Build the loop tree.
12334         (pass_build_cfg): Provide PROP_loops.
12335         (move_sese_region_to_fn): Remove loops that are outlined into fn
12336         for now.
12337         * tree-inline.c: Include cfgloop.h.
12338         (initialize_cfun): Do not drop PROP_loops.
12339         (copy_loops): New function.
12340         (copy_cfg_body): Copy loop structure.
12341         (tree_function_versioning): Initialize destination loop tree.
12342         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
12343         (pass_parallelize_loops): Do IL verification.
12344         * loop-init.c (loop_optimizer_init): Fixup loops if required.
12345         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
12346         the CFG make sure we fixup loops as well.
12347         * tree-ssa-tail-merge.c: Include cfgloop.h.
12348         (replace_block_by): When merging loop latches mark loops for fixup.
12349         * lto-streamer-out.c (output_struct_function_base): Drop
12350         PROP_loops for now.
12351         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
12352         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
12353         * ipa-split.c: Include cfgloop.h.
12354         (split_function): Add the new return block to the loop tree root.
12355         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
12356         whether we have removed the forwarder block.
12357         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
12358         * cfgloop.h (place_new_loop): Declare.
12359         * cfgloopmanip.c (place_new_loop): Export.
12360         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
12361         (tree-switch-conversion.o): Likewise.
12362         (tree-complex.o): Likewise.
12363         (tree-inline.o): Likewise.
12364         (tree-ssa-tailmerge.o): Likewise.
12365         (ipa-split.o): Likewise.
12366         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
12367         (tree-ssa-copy.o): Likewise.
12368         * tree-switch-conversion.c: Include cfgloop.h
12369         (process_switch): If we emit a bit-test cascade, schedule loops
12370         for fixup.
12371         * tree-complex.c: Include cfgloop.h.
12372         (expand_complex_div_wide): Properly add new basic-blocks to loops.
12373         * asan.c: Include cfgloop.h.
12374         (create_cond_insert_point): Properly add new basic-blocks to
12375         loops, schedule loop fixup.
12376         * cfgloop.c (verify_loop_structure): Check that looks are not
12377         marked for fixup.
12378         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
12379         to loops.
12380         (expand_omp_for_generic): Likewise.
12381         (expand_omp_sections): Likewise.
12382         (expand_omp_atomic_pipeline): Schedule loops for fixup.
12383         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
12384         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
12385         is initialized, not when loops are present.
12386         * tree-parloops.c (parallelize_loops): Remove checking here.
12387         * passes.c (init_optimization_passes): Schedule a copy-propagation
12388         pass before complete unrolling of inner loops.
12390 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
12392         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
12393         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
12394         (colorize_init): Add argument to _WIN32 version.
12395         * toplev.c: Include diagnostic-color.h.
12396         (process_options): Default to -fdiagnostics-color=auto if
12397         GCC_COLORS env var is in the environment.
12398         * common.opt (fdiagnostics-color=): Add Var and Init.
12399         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
12400         env var is in the environment, the default is auto rather than never.
12402         * diagnostic.h (file_name_as_prefix): Add context argument.
12403         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
12404         the string as locus.
12405         * langhooks.c (lhd_print_error_function): Adjust caller.
12407 2013-04-25  Lawrence Crowl  <crowl@google.com>
12409         * var-tracking.c (shared_hash_def::htab):
12410         Change type to hash_table.  Update dependent calls and types.
12412 2013-04-25  Lawrence Crowl  <crowl@google.com>
12414         * Makefile.in: Update as needed below.
12416         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
12417         Move declaration to after the type's method definitons.
12419         * attribs.c (htab_t scoped_attributes::attribute_hash):
12420         Change type to hash_table.  Update dependent calls and types.
12422         * bitmap.c (htab_t bitmap_desc_hash):
12423         Change type to hash_table.  Update dependent calls and types.
12425         * cselib.c (htab_t cselib_hash_table):
12426         Change type to hash_table.  Update dependent calls and types.
12428         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
12429         (hash_string_slot_node): Move implementation into lto-streamer.h
12430         struct string_slot_hasher.
12431         (eq_string_slot_node): Likewise.
12433         * data-streamer-out.c: Update output_block::string_hash_table
12434         dependent calls and types.
12436         * dwarf2cfi.c (htab_t trace_index):
12437         Change type to hash_table.  Update dependent calls and types.
12439         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
12440         Change type to hash_table.  Update dependent calls and types.
12441         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
12442         (htab_t optimize_external_refs::map): Likewise.
12443         (htab_t output_comp_unit::extern_map): Likewise.
12444         (htab_t output_comdat_type_unit::extern_map): Likewise.
12445         (htab_t output_macinfo::macinfo_htab): Likewise.
12446         (htab_t optimize_location_lists::htab): Likewise.
12447         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
12449         * except.c (htab_t ehspec_hash_type):
12450         Change type to hash_table.  Update dependent calls and types.
12451         (assign_filter_values::ttypes): Likewise.
12452         (assign_filter_values::ehspec): Likewise.
12453         (sjlj_assign_call_site_values::ar_hash): Likewise.
12454         (convert_to_eh_region_ranges::ar_hash): Likewise.
12456         * gcse.c (htab_t pre_ldst_table):
12457         Change type to hash_table.  Update dependent calls and types.
12459         * ggc-common.c (htab_t saving_htab):
12460         Change type to hash_table.  Update dependent calls and types.
12461         (htab_t loc_hash): Likewise.
12462         (htab_t ptr_hash): Likewise.
12463         (call_count): Rename ggc_call_count.
12464         (call_alloc): Rename ggc_call_alloc.
12465         (loc_descriptor): Rename make_loc_descriptor.
12466         (add_statistics): Rename ggc_add_statistics.
12468         * ggc-common.c (saving_htab):
12469         Change type to hash_table.  Update dependent calls and types.
12471         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
12472         (push_gimplify_context): Likewise.
12473         (pop_gimplify_context): Likewise.
12474         (struct gimple_temp_hash_elt): Added.
12475         (struct gimplify_hasher): Likewise.
12476         (struct gimplify_ctx.temp_htab):
12477         Change type to hash_table.  Update dependent calls and types.
12479         * gimple-fold.c: Include gimplify-ctx.h.
12481         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
12482         Change type to hash_table.  Update dependent calls and types.
12483         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
12484         avoid potential global name collision.
12486         * gimplify.c: Include gimplify-ctx.h.
12487         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
12488         (htab_t gimplify_ctx::temp_htab):
12489         Update dependent calls and types for new type hash_table.
12490         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
12491         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
12493         * gimplify-ctx.h: New.
12494         (struct gimple_temp_hash_elt): Move from gimplify.c.
12495         (class gimplify_hasher): New.
12496         (struct gimplify_ctx): Move from gimple.h.
12497         (htab_t gimplify_ctx::temp_htab):
12498         Change type to hash_table.  Update dependent calls and types.
12500         * graphite-clast-to-gimple.c: Include graphite-htab.h.
12501         (htab_t ivs_params::newivs_index):
12502         Change type to hash_table.  Update dependent calls and types.
12503         (htab_t ivs_params::params_index): Likewise.
12504         (htab_t print_generated_program::params_index): Likewise.
12505         (htab_t gloog::newivs_index): Likewise.
12506         (htab_t gloog::params_index): Likewise.
12508         * graphite.c: Include graphite-htab.h.
12509         4htab_t graphite_transform_loops::bb_pbb_mapping):
12510         Change type to hash_table.  Update dependent calls and types.
12512         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
12513         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
12514         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
12516         * graphite-dependences.c: Include graphite-htab.h.
12517         (loop_is_parallel_p): Change hash table type of parameter.
12519         * graphite-htab.h: New.
12520         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
12521         (extern find_pbb_via_hash): Move from graphite-poly.h.
12522         (extern loop_is_parallel_p): Move from graphite-poly.h.
12523         (extern get_loop_body_pbbs): Move from graphite-poly.h.
12525         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
12526         (extern loop_is_parallel_p): Move to graphite-htab.h.
12527         (extern get_loop_body_pbbs): Move to graphite-htab.h.
12529         * haifa-sched.c (htab_t delay_htab):
12530         Change type to hash_table.  Update dependent calls and types.
12531         (htab_t delay_htab_i2): Likewise.
12533         * ira-color.c (htab_t allocno_hard_regs_htab):
12534         Change type to hash_table.  Update dependent calls and types.
12536         * ira-costs.c (htab_t cost_classes_htab):
12537         Change type to hash_table.  Update dependent calls and types.
12539         * loop-invariant.c (htab_t merge_identical_invariants::eq):
12540         Change type to hash_table.  Update dependent calls and types.
12542         * loop-iv.c (htab_t bivs):
12543         Change type to hash_table.  Update dependent calls and types.
12545         * loop-unroll.c (htab_t opt_info::insns_to_split):
12546         Change type to hash_table.  Update dependent calls and types.
12547         (htab_t opt_info::insns_with_var_to_expand): Likewise.
12549         * lto-streamer.h (struct string_slot): Move from data-streamer.h
12550         (struct string_slot_hasher): New.
12551         (htab_t output_block::string_hash_table):
12552         Change type to hash_table.  Update dependent calls and types.
12554         * lto-streamer-in.c (freeing_string_slot_hasher): New.
12555         (htab_t file_name_hash_table):
12556         Change type to hash_table.  Update dependent calls and types.
12558         * lto-streamer-out.c: Update output_block::string_hash_table dependent
12559         calls and types.
12561         * lto-streamer.c (htab_t tree_htab):
12562         Change type to hash_table.  Update dependent calls and types.
12564         * omp-low.c: Include gimplify-ctx.h.
12566         * passes.c (htab_t name_to_pass_map):
12567         Change type to hash_table.  Update dependent calls and types.
12568         (pass_traverse): Rename to passes_pass_traverse.
12570         * plugin.c (htab_t event_tab):
12571         Change type to hash_table.  Update dependent calls and types.
12573         * postreload-gcse.c (htab_t expr_table):
12574         Change type to hash_table.  Update dependent calls and types.
12575         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
12577         * sese.c (debug_rename_map_1): Make extern.
12578         (htab_t copy_bb_and_scalar_dependences::rename_map):
12579         Change type to hash_table.  Update dependent calls and types.
12581         * sese.h (extern debug_rename_map): Move to .c file.
12583         * store-motion.c (htab_t store_motion_mems_table):
12584         Change type to hash_table.  Update dependent calls and types.
12586         * trans-mem.c (htab_t tm_new_mem_hash):
12587         Change type to hash_table.  Update dependent calls and types.
12589         * tree-browser.c (htab_t TB_up_ht):
12590         Change type to hash_table.  Update dependent calls and types.
12592         * tree-cfg.c (htab_t discriminator_per_locus):
12593         Change type to hash_table.  Update dependent calls and types.
12595         * tree-complex.c: Include tree-hasher.h
12596         (htab_t complex_variable_components):
12597         Change type to hash_table.  Update dependent calls and types.
12599         * tree-eh.c (htab_t finally_tree):
12600         Change type to hash_table.  Update dependent calls and types.
12602         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
12603         struct int_tree_hasher.
12604         (extern int_tree_map_eq): Likewise.
12605         (uid_decl_map_hash): Removed.
12606         (extern decl_tree_map_eq): Likewise.
12608         * tree-hasher.h: New.
12609         (struct int_tree_hasher): New.
12610         (typedef int_tree_htab_type): New.
12612         * tree-inline.c: Include gimplify-ctx.h.
12614         * tree-mudflap.c: Include gimplify-ctx.h.
12616         * tree-parloops.c: Include tree-hasher.h.
12617         (htab_t eliminate_local_variables_stmt::decl_address):
12618         Change type to hash_table.  Update dependent calls and types.
12619         (htab_t separate_decls_in_region::decl_copies): Likewise.
12621         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
12622         Change type to hash_table.  Update dependent calls and types.
12624         * tree-sra.c (candidates):
12625         Change type to hash_table.  Update dependent calls and types.
12627         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
12628         in tree-flow.h.
12629         (int_tree_map_hash): Likewise.
12631         * tree-ssa-dom.c (htab_t avail_exprs):
12632         Change type to hash_table.  Update dependent calls and types.
12634         * tree-ssa-live.c (var_map_base_init::tree_to_index):
12635         Change type to hash_table.  Update dependent calls and types.
12637         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
12638         Change type to hash_table.  Update dependent calls and types.
12640         * tree-ssa-phiopt.c (seen_ssa_names):
12641         Change type to hash_table.  Update dependent calls and types.
12643         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
12644         Change type to hash_table.  Update dependent calls and types.
12646         * tree-ssa-uncprop.c (equiv):
12647         Change type to hash_table.  Update dependent calls and types.
12649 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
12651         PR rtl-optimization/57003
12652         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
12653         call note_stores with kill_clobbered_value callback again after
12654         killing regs_invalidated_by_call.
12656 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
12658         * config/aarch64/aarch64-simd.md
12659         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
12660         (aarch64_simd_bsl<mode>): Likewise.
12661         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
12663 2013-04-25  Marek Polacek  <polacek@redhat.com>
12665         PR tree-optimization/57066
12666         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
12668 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
12670         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
12672 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
12674         * config/aarch64/aarch64-builtins.c
12675         (aarch64_fold_builtin): New.
12676         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
12677         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
12678         * config/aarch64/aarch64-simd-builtins.def (abs): New.
12679         * config/aarch64/arm_neon.h
12680         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
12682 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
12683             Tejas Belagod  <tejas.belagod@arm.com>
12685         * config/aarch64/aarch64-builtins.c
12686         (aarch64_gimple_fold_builtin): New.
12687         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
12688         * config/aarch64/aarch64-simd-builtins.def (addv): New.
12689         * config/aarch64/aarch64-simd.md (addpv4sf): New.
12690         (addvv4sf): Update.
12691         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
12693 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12695         * config/aarch64/aarch64.md
12696         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
12698 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12700         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
12701         (*ngcsi_uxtw): New pattern.
12703 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12704             Julian Brown  <julian@codesourcery.com>
12706         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
12707         (TB_DREG): Add T_V4HF.
12708         (v4hf_UP): New macro.
12709         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
12710         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
12711         Handle initialisation of V4HF. Adjust initialisation of reinterpret
12712         built-ins.
12713         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
12714         (arm_vector_mode_supported_p): Handle V4HF.
12715         (arm_mangle_map): Handle V4HFmode.
12716         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
12717         * config/arm/arm_neon_builtins.def: Add entries for
12718         vcvtv4hfv4sf, vcvtv4sfv4hf.
12719         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
12720         (neon_vcvtv4hfv4sf): Likewise.
12721         * config/arm/neon-gen.ml: Handle half-precision floating point
12722         features.
12723         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
12724         * config/arm/arm_neon.h: Regenerate.
12725         * config/arm/neon.ml (type elts): Add F16.
12726         (type vectype): Add T_float16x4, T_floatHF.
12727         (type vecmode): Add V4HF.
12728         (type features): Add Requires_FP_bit feature.
12729         (elt_width): Handle F16.
12730         (elt_class): Likewise.
12731         (elt_of_class_width): Likewise.
12732         (mode_of_elt): Refactor.
12733         (type_for_elt): Handle F16, fix error messages.
12734         (vectype_size): Handle T_float16x4.
12735         (vcvt_sh): New function.
12736         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
12737         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
12738         (string_of_mode): Handle V4HF.
12739         * doc/arm-neon-intrinsics.texi: Regenerate.
12741 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
12743         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
12744         format specifier in 'X' case.
12746 2013-04-25  Alan Modra  <amodra@gmail.com>
12748         PR target/57052
12749         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
12750         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
12751         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
12752         Repeat for many other rotate/shift and mask patterns using subregs.
12753         Name lshiftrt insns.
12754         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
12755         on WORDS_BIG_ENDIAN.
12757 2013-04-25  Alan Modra  <amodra@gmail.com>
12759         * config.gcc: Support little-endian powerpc-linux targets.
12760         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
12761         (LINK_OS_LINUX_SPEC): Define.
12762         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
12763         Preserve MASK_LITTLE_ENDIAN.
12764         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
12765         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
12766         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
12767         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
12768         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
12769         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
12770         Correct fp word order for little-endian.  Don't shift toc entries
12771         smaller than a word for little-endian.
12772         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
12773         (bswapdi2 splits): Correct low-part subreg for little-endian.
12774         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
12775         low/high where such is correct only for be.
12776         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
12777         little-endian for -mcall-aixdesc.
12779 2013-04-25  Alan Modra  <amodra@gmail.com>
12781         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
12782         replace_equiv_address_nv.
12784 2013-04-25  Alan Modra  <amodra@gmail.com>
12786         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
12788 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12790         Revert:
12791         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12792         * rtl.h (struct rtx_def): ...
12794 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12796         PR rtl-optimizations/57046
12797         * lra-constraints (split_reg): Set up lra_risky_transformations_p
12798         for multi-reg splits.
12800 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
12802         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
12804 2013-04-24  Sterling Augustine  <saugustine@google.com>
12806         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
12807         (comp_dir_string, debug_str_dwo_section): New.
12808         (DEBUG_STR_DWO_SECTION): Rename to ...
12809         (DEBUG_DWO_STR_SECTION): ... this.
12810         (DEBUG_NORM_STR_SECTION): Delete.
12811         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
12812         (DEBUG_STR_DWO_SECTION_FLAGS): New.
12813         (find_AT_string): Move most logic to ...
12814         (find_AT_string_in_table): ... here.  New.
12815         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
12816         add_skeleton_AT_string.  Delete logic.
12817         (output_skeleton_debug_sections): Remove call to
12818         add_top_level_skeleton_die_attrs.
12819         (add_comp_dir_attribute): Move logic to comp_dir_string.
12820         (dwarf2out_init): Initialize debug_str_dwo_section.
12821         (output_indirect_string): Call find_string_form.
12822         (output_indirect_strings): Rewrite.
12823         (prune_unused_types): Empty skeleton_debug_str_hash.
12824         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
12825         (dwarf2out_finish):  Call output_indirect_strings.
12827 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
12829         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
12831 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
12833         * rtl.h (struct rtx_def): Add comment for field jump.
12834         (LRA_SUBREG_P): New macro.
12835         * recog.c (register_operand): Check LRA_SUBREG_P.
12836         * lra.c (lra): Add note at the end of RTL code. Align non-empty
12837         stack frame.
12838         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
12839         (lra_final_code_change): Skip subreg change for operators.
12840         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
12841         if there are no operand changes.
12842         * lra-constraints.c (curr_insn_set): New.
12843         (match_reload): Set LRA_SUBREG_P.
12844         (emit_spill_move): Ditto.
12845         (check_and_process_move): Use curr_insn_set. Process only single
12846         set insns.  Don't initialize sec_mem_p and change_p.
12847         (simplify_operand_subreg): Use LRA_SUBREG_P.
12848         (reg_in_class_p): New function.
12849         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
12850         of #ifdef.  Add code to remove cycling.
12851         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
12852         non-null disp.  Reload inner instead of disp when base and index
12853         are null.  Try to put lo_sum into register.
12854         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
12855         (check_and_process_move): Move code for move cost check to
12856         simple_move_p.  Remove equiv_substitution.
12857         (simple_move_p): New function.
12858         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
12859         curr_insn_set.  Call check_and_process_move only for single set
12860         insns.  Use the new function.  Move call of check_and_process_move
12861         after operand equiv substitution and address process.
12863 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
12865         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
12866         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
12867         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
12869 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
12871         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
12873 2013-04-24  Marek Polacek  <polacek@redhat.com>
12875         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
12876         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
12877         (select_loops_exit_conditions): Likewise.
12878         (number_of_iterations_for_all_loops): Likewise.
12879         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
12880         (scev_analysis): Likewise.
12882 2013-04-02  Catherine Moore  <clm@codesourcery.com>
12883             Chao-ying Fu  <fu@mips.com>
12885         * config/mips/micromips.md (jraddiusp): New pattern.
12886         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
12887         instruction if possible.
12889 2013-04-24  Alan Modra  <amodra@gmail.com>
12891         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
12893 2013-04-24  Julian Brown  <julian@codesourcery.com>
12894             Chung-Lin Tang  <cltang@codesourcery.com>
12896         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
12897         dependency behavior in enumeration type DIE generation. Add TODO note
12898         to comments about future DW_FORM_sdata/udata re-work of related code.
12900 2013-04-23  Lawrence Crowl  <crowl@google.com>
12902         * Makefile.in: Update as needed below.
12904         * hash-table.h (class hash_table):
12905         Correct many methods with parameter types compare_type to the correct
12906         value_type.  (Correct code was unlikely to notice the change.)
12907         (hash_table::elements_with_deleted) New.
12908         (class hashtable::iterator): New.
12909         (hashtable::begin()): New.
12910         (hashtable::end()): New.
12911         (FOR_EACH_HASH_TABLE_ELEMENT): New.
12913         * statistics.c (statistics_hashes):
12914         Change type to hash_table.  Update dependent calls and types.
12916         * tree-into-ssa.c (var_infos):
12917         Change type to hash_table.  Update dependent calls and types.
12919         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
12920         Change type to hash_table.  Update dependent calls and types.
12922         * tree-ssa-loop-im.c (struct mem_ref.refs):
12923         Change type to hash_table.  Update dependent calls and types.
12925         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
12926         Change type to hash_table.  Update dependent calls and types.
12928         * tree-ssa-sccvn.c (vn_tables_s::nary):
12929         Change type to hash_table.  Update dependent calls and types.
12930         (vn_tables_s::phis): Likewise.
12931         (vn_tables_s::references): Likewise.
12933         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
12934         (vn_reference_eq): Update parameter and return types.
12936         * tree-ssa-structalias.c (pointer_equiv_class_table):
12937         Change type to hash_table.  Update dependent calls and types.
12938         (location_equiv_class_table): Likewise.
12940         * tree-vect-data-refs.c: Consequential changes for making
12941         peeling a hash_table.
12943         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12944         (destroy_loop_vec_info): Dependent hash_table update.
12946         * tree-vectorizer.h (peeling_htab):
12947         Change type to hash_table.  Update dependent calls and types.
12949 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
12951         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12952         to check the register content is equal or not.
12953         * lra-constraints.c (match_reload): Use lra_assign_reg_val
12954         to assign register content record.
12955         * lra-eliminations.c (update_reg_eliminate): Use
12956         lra_update_reg_val_offset to update register content offset.
12957         * lra-int.h (struct lra_reg): Add offset member.
12958         (lra_reg_val_equal_p): New static inline function.
12959         (lra_update_reg_val_offset): New static inline function.
12960         (lra_assign_reg_val): New static inline function.
12961         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12962         to assign register content record.
12963         (initialize_lra_reg_info_element): Initial offset to zero.
12965 2013-04-23  Catherine Moore  <clm@codesourcery.com>
12967         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12968         operands.  Record compression.
12970 2013-04-23  Xinliang David Li  <davidxl@google.com>
12972         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12974 2013-04-23  Richard Biener  <rguenther@suse.de>
12976         PR middle-end/57036
12977         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12978         parameter, only add abnormal goto edges from the copied body
12979         if the call could perform abnormal gotos.
12980         (copy_cfg_body): Adjust.
12982 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
12984         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12986 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
12988         * coretypes.h (gimple_stmt_iterator): Add struct to make
12989         compatible with C.
12991 2013-04-23  Richard Biener  <rguenther@suse.de>
12993         PR tree-optimization/57026
12994         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12995         from SSA names occuring in abnormal PHI nodes.
12997 2013-04-22  Andi Kleen  <ak@linux.intel.com>
12999         * lto/lto.c (print_lto_report_1): Fix LTO report names.
13001 2013-04-22  Andi Kleen  <ak@linux.intel.com>
13003         * lto/lto.c (print_lto_report_1): Declare early.
13004         (read_cgraph_and_symbols): Call print_lto_report_1 early.
13006 2013-04-22  Andi Kleen  <ak@linux.intel.com>
13008         * common.opt (-flto-report-wpa): Add.
13009         * doc/invoke.texi (-flto-report-wpa): Add.
13010         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
13011         (lto_main): dito.
13013 2013-04-22  Xinliang David Li  <davidxl@google.com>
13015         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
13016         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
13017         * Makefile.in: New dependency
13019             David Daney <ddaney.cavm@gmail.com>
13021         * configure.ac (gcc_cv_as_micromips_support):  Use the
13022         --fatal-warnings option.
13023         * configure: Regenerate.
13025 2013-04-22  Marek Polacek  <polacek@redhat.com>
13027         PR sanitizer/56990
13028         * tsan.c (instrument_expr): Don't instrument expression
13029         in case its size is zero.
13031 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
13033         PR target/57032
13034         Revert:
13035         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
13037         * config/alpha/alpha.c (TARGET_LRA_P): New define.
13039 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
13041         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
13042         (gimple_stmt_iterator): New typedef.
13043         * gimple.h (gimple_stmt_iterator): Rename to...
13044         (gimple_stmt_iterator_d): ... This.
13045         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
13046         trees be valid for GIMPLE and GENERIC.
13047         (TARGET_GIMPLE_FOLD_BUILTIN): New.
13048         * gimple-fold.c (gimple_fold_call): Call target hook
13049         gimple_fold_builtin.
13050         * hooks.c (hook_bool_gsiptr_false): New.
13051         * hooks.h (hook_bool_gsiptr_false): New.
13052         * target.def (fold_stmt): New.
13053         * doc/tm.texi: Regenerate.
13055 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
13057         PR target/57018
13058         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
13059         a set sp if no stack realignment.
13061 2013-04-22  Nick Clifton  <nickc@redhat.com>
13063         * config.gcc (tilegx-linux): Extend extra_objs rather than
13064         overwriting it.
13065         (tilepro-linux): Likewise.
13067 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
13069         * config/aarch64/aarch64-builtins.c
13070         (CF): Remove.
13071         (CF0, CF1, CF2, CF3, CF4, CF10): New.
13072         (VAR<1-12>): Add MAP parameter.
13073         (BUILTIN_*): Likewise.
13074         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
13075         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
13076         (aarch64_ushl_n<mode>): Likewise.
13077         (aarch64_sshr_n<mode>): Likewise.
13078         (aarch64_ushr_n<mode>): Likewise.
13079         (aarch64_<maxmin><mode>): Likewise.
13080         (aarch64_sqrt<mode>): Likewise.
13081         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
13082         (vshr<q>_n_*): Likewise.
13084 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
13086         * config/aarch64/aarch64-builtins.c
13087         (aarch64_simd_builtin_type_mode): Handle SF types.
13088         (sf_UP): Define.
13089         (BUILTIN_GPF): Define.
13090         (aarch64_init_simd_builtins): Handle SF types.
13091         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
13092         (frecps): Likewise.
13093         (frecpx): Likewise.
13094         * config/aarch64/aarch64-simd.md
13095         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
13096         (aarch64_frecpe<mode>): New.
13097         (aarch64_frecps<mode>): Likewise.
13098         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
13099         (v8type): Add frecp<esx>.
13100         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
13101         (aarch64_frecps<mode>): Likewise.
13102         * config/aarch64/iterators.md (FRECP): New.
13103         (frecp_suffix): Likewise.
13104         * config/aarch64/arm_neon.h
13105         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
13107 2013-04-22  Christian Bruel  <christian.bruel@st.com>
13109         PR target/56995
13110         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
13111         (REG_CLASS_NAMES): Idem.
13112         (REG_CLASS_CONTENTS): Idem.
13113         (REGCLASS_HAS_FP_REG): Idem.
13114         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
13115         (sh_conditional_register_usage): Idem.
13117 2013-04-21  Jeff Law  <law@redhat.com>
13119         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
13120         (ssa_forward_propagate_and_combine): Use it.
13122 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
13124         * lra.c: Update the flow chart diagram.
13126 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
13128         PR rtl-optimization/56847
13129         * lra-constraints.c (process_alt_operands): Discourage alternative
13130         with non-matche doffsettable memory constraint fro memory with
13131         known offset.
13133 2013-04-19  Richard Biener  <rguenther@suse.de>
13135         PR tree-optimization/56982
13136         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
13137         function.
13138         * gimplify.c (gimplify_call_expr): Notice special calls.
13139         (gimplify_modify_expr): Likewise.
13140         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
13141         abnormal control flow receivers.
13142         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
13143         in the same way as cfun->has_nonlocal_labels.
13144         (gimple_purge_dead_abnormal_call_edges): Likewise.
13145         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
13146         receivers start a basic-block.
13148 2013-04-19  Richard Biener  <rguenther@suse.de>
13150         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
13151         member ...
13152         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
13153         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
13154         (SLP_TREE_LOAD_PERMUTATION): Add.
13155         (vect_transform_slp_perm_load): Adjust prototype.
13156         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
13157         (vect_free_slp_instance): Likewise.
13158         (vect_create_new_slp_node): Likewise.
13159         (vect_supported_slp_permutation_p): Remove.
13160         (vect_slp_rearrange_stmts): Adjust.
13161         (vect_supported_load_permutation_p): Likewise.  Inline
13162         vect_supported_slp_permutation_p here.
13163         (vect_analyze_slp_instance): Compute load permutations per
13164         slp node instead of per instance.
13165         (vect_get_slp_defs): Adjust.
13166         (vect_transform_slp_perm_load): Likewise.
13167         (vect_schedule_slp_instance): Remove redundant code.
13168         (vect_schedule_slp): Remove hack for PR56270, add it ...
13169         * tree-vect-stmts.c (vectorizable_load): ... here, do not
13170         CSE loads for SLP.  Adjust.
13172 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
13174         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
13175         spelling in two comments.
13177 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
13179         PR target/56797
13180         * config/arm/arm.c (load_multiple_sequence): Require SP
13181         as base register for loads if SP is in the register list.
13183 2013-04-19  Martin Jambor  <mjambor@suse.cz>
13185         PR tree-optimization/56718
13186         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
13187         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
13188         and made public.  Adjusted all callers.
13189         (ipa_intraprocedural_devirtualization): New function.
13190         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
13191         (ipa_intraprocedural_devirtualization): Likewise.
13192         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
13194 2013-04-19  Richard Biener  <rguenther@suse.de>
13196         PR tree-optimization/57000
13197         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
13199 2013-04-19  Terry Guo  <terry.guo@arm.com>
13201         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
13202         Replace with ...
13203         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
13204         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
13205         (cortex_m4_fmacs): Use new reservations.
13206         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
13208 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
13210         PR rtl-optimization/56999
13211         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
13212         related code.
13213         (lra_coalesce): Remove split_origin_bitmap and related code.
13214         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
13215         ranges if necessary.
13217 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
13219         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
13220         New array.
13221         (ix86_expand_call): Remove clobbered_registers array and use
13222         x86_64_ms_sysv_extra_clobbered_registers instead.
13223         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
13224         Declare here.
13225         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
13226         predicate.
13227         * config/i386/i386.md (*call_rex64_ms_sysv): Use
13228         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
13229         (*call_value_rex64_ms_sysv): Ditto.
13231 2013-04-18  Cary Coutant  <ccoutant@google.com>
13233         * dwarf2out.c (output_pubnames): Check die_perennial_p of
13234         parent instead of die_mark.
13236 2013-04-18  Diego Novillo  <dnovillo@google.com>
13238         * gimple.c (create_gimple_tmp): New.
13239         (get_expr_type): New.
13240         (build_assign): New.
13241         (build_type_cast): New.
13242         * gimple.h (enum ssa_mode): Define.
13243         (gimple_seq_set_location): New.
13244         * asan.c (build_check_stmt): Change some gimple_build_* calls
13245         to use build_assign and build_type_cast.
13247 2013-04-18  Richard Biener  <rguenther@suse.de>
13249         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
13250         handle negative step.  Remove redundant checks.
13251         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
13252         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
13253         for negative step and grouped loads fail to vectorize.
13255 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
13257         * emit-rtl.c (reset_insn_used_flags): New function.
13258         (reset_all_used_flags): Use it.
13259         (verify_insn_sharing): New function.
13260         (verify_rtl_sharing): Fix verification for SEQUENCEs.
13262 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
13264         PR tree-optimization/56984
13265         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
13266         and (x >> M) >= N don't register any assertion if N << M is the
13267         minimum value.
13269 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
13271         * lower-subreg.c (resolve_simple_move): If called self-recursive,
13272         do not delete_insn insns that have not yet been emitted, only
13273         unlink them with remove_insn.
13274         * df-scan.c (df_insn_delete): Revert r197492.
13276 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
13278         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
13279         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
13281 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
13283         * config/arm/arm.md (movsicc_insn): Convert define_insn into
13284         define_insn_and_split.
13285         (and_scc,ior_scc,negscc): Likewise.
13286         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
13288 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
13290         * config/arm/arm.c (use_return_insn): Return 0 for targets that
13291         can benefit from using a sequence of LDRD instructions in epilogue
13292         instead of a single LDM instruction.
13294 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
13296         PR 45688
13297         * doc/extend.texi: Fix typo.
13299 2013-04-17  Richard Biener  <rguenther@suse.de>
13301         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
13302         (vect_build_slp_tree): ... here.
13303         (vect_build_slp_tree_1): Compute which stmts of the SLP group
13304         match.  Remove special-casing of mismatched complex loads.
13305         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
13306         re-try the match with swapped commutative operands.
13307         (vect_supported_load_permutation_p): Remove special-casing of
13308         mismatched complex loads.
13309         (vect_analyze_slp_instance): Adjust.
13311 2013-04-17  Richard Biener  <rguenther@suse.de>
13313         PR rtl-optimization/56921
13314         * cfgloop.h (struct loop): Add simple_loop_desc member.
13315         (struct niter_desc): Mark with GTY(()).
13316         (simple_loop_desc): Do not use aux field but simple_loop_desc.
13317         * loop-iv.c (get_simple_loop_desc): Likewise.
13318         (free_simple_loop_desc): Likewise.
13320         Revert
13321         2013-04-16  Richard Biener  <rguenther@suse.de>
13323         PR rtl-optimization/56921
13324         * loop-init.c (pass_rtl_move_loop_invariants): Add
13325         TODO_do_not_ggc_collect to todo_flags_finish.
13326         (pass_rtl_unswitch): Same.
13327         (pass_rtl_unroll_and_peel_loops): Same.
13328         (pass_rtl_doloop): Same.
13330 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
13332         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
13333         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
13334         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
13335         references.
13336         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
13337         * tree-streamer.c (record_common_node): Adjust reference in comment.
13339 2013-04-17  Terry Guo  <terry.guo@arm.com>
13341         * config/arm/cortex-m4.md: Add a new bypass.
13343 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13345         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
13346         New pattern.
13347         (*subs_<optab><mode>_multp2): New pattern.
13348         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
13349         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
13351 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13353         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
13354         (*subs_mul_imm_<mode>): New pattern.
13356 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
13358         PR target/56948
13359         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
13360         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
13361         (vsx_movti_32bit): Change j->wa to O->wa.
13363 2013-04-16  Richard Biener  <rguenther@suse.de>
13365         PR rtl-optimization/56921
13366         * loop-init.c (pass_rtl_move_loop_invariants): Add
13367         TODO_do_not_ggc_collect to todo_flags_finish.
13368         (pass_rtl_unswitch): Same.
13369         (pass_rtl_unroll_and_peel_loops): Same.
13370         (pass_rtl_doloop): Same.
13372 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
13374         * config/arm/arm.c (emit_multi_reg_push): New declaration
13375         for an existing function.
13376         (arm_emit_strd_push): New function.
13377         (arm_expand_prologue): Used here.
13378         (arm_emit_ldrd_pop): New function.
13379         (arm_expand_epilogue): Used here.
13380         (arm_get_frame_offsets): Update condition.
13381         (arm_emit_multi_reg_pop): Add a special case for load of a single
13382         register with writeback.
13384 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
13386         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
13387         description.
13389 2013-04-16  Richard Biener  <rguenther@suse.de>
13391         PR tree-optimization/56756
13392         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
13393         (first_mem_ref_loc): New.
13394         (execute_sm): Place the load temporarily before a previous
13395         access instead of in the latch edge to ensure its SSA dependencies
13396         are defined at points dominating the load.
13398 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
13400         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
13401         correct fix by moving header and footer insn to the footer of
13402         the merged basic block.  Clear BB_END of the merged-away block.
13404         PR middle-end/43631
13405         * emit-rtl.c (make_note_raw): New function.
13406         (link_insn_into_chain): New static inline function.
13407         (add_insn): Use it.
13408         (add_insn_before, add_insn_after): Factor insn chain linking code...
13409         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
13410         using link_insn_into_chain.
13411         (note_outside_basic_block_p): New helper function for emit_note_after
13412         and emit_note_before.
13413         (emit_note_after): Use nobb variant of add_insn_after if the note
13414         should not be contained in a basic block.
13415         (emit_note_before): Use nobb variant of add_insn_before if the note
13416         should not be contained in a basic block.
13417         (emit_note_copy): Use make_note_raw.
13418         (emit_note): Likewise.
13419         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
13420         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
13421         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
13422         the moved barrier the tail of the basic block it follows.
13423         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
13425 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
13427         PR tree-optimization/56962
13428         * gimple-ssa-strength-reduction.c (record_increment): Only set
13429         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
13430         either rhs1 or rhs2 is equal to c->base_expr.
13432 2013-04-15  Richard Biener  <rguenther@suse.de>
13434         PR tree-optimization/56933
13435         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
13436         member.
13437         (GROUP_READ_WRITE_DEPENDENCE): Remove.
13438         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
13439         * tree-vect-data-refs.c (vect_analyze_group_access): Move
13440         dependence check ...
13441         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
13442         ... here.
13443         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
13444         GROUP_READ_WRITE_DEPENDENCE.
13446 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13448         * emit-rtl.c (reset_all_used_flags): New function.
13449         (verify_rtl_sharing): Call reset_all_used_flags before and after
13450         performing the checks.
13452 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13454         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
13455         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
13456         * config/arm/constraints.md (De): New constraint.
13457         * config/arm/neon.md (anddi3_neon): Delete.
13458         (neon_vand<mode>): Expand to standard anddi3 pattern.
13459         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
13460         Move earlier in the file.
13461         (neon_inv_logic_op2): Likewise.
13462         (arm_anddi_operand_neon): New predicate.
13464 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
13466         * configure.ac (gcc_cv_ld_as_needed): Set
13467         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
13468         Use -z ignore, -z record on *-*-solaris2*.
13469         (HAVE_LD_AS_NEEDED): Update comment.
13470         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
13471         * configure: Regenerate.
13472         * config.in: Regenerate.
13473         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
13474         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
13475         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
13476         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
13477         equivalents.  Fix markup.
13478         * doc/tm.texi: Regenerate.
13480 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
13482         * config/i386/i386.opt: New option mstack-protector-guard=.
13483         * config/i386/i386-opts.h: Add enum stack_protector_guard.
13484         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
13485         TARGET_SSP_TLS_GUARD.
13486         * config/i386/i386.c (ix86_option_override_internal): Set
13487         ix86_stack_protector_guard.
13488         * config/i386/i386.md (stack_protect_set): Enable for
13489         TARGET_SSP_TLS_GUARD only.
13490         (stack_protect_set_<mode>): Ditto.
13491         (stack_protect_test): Ditto.
13492         (stack_protect_test_<mode>): Ditto.
13493         * doc/invoke.texi (i386 Option): Document.
13495 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
13497         PR target/56890
13498         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
13499         (S_MODES): Set H_MODE bit.
13500         (SF_MODES): Set only S_MODE and SF_MODE bits.
13501         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
13502         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
13503         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
13504         <MODE_FLOAT>: Likewise.
13506 2013-04-15  Joey Ye  <joey.ye@arm.com>
13508         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
13510 2013-04-15  Joey Ye  <joey.ye@arm.com>
13512         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
13513         for real far jump.
13514         (thumb_far_jump_used_p): Count instruction size and set
13515         far_jump_used.
13517 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
13519         * reorg.c (fill_simple_delay_slots): Reindent block of code.
13520         * resource.c (mark_target_live_regs): Reformat conditional block.
13522 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
13524         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
13525         notes, they are emitted only just before final.
13526         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
13528 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
13530         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
13531         * cfgrtl.c (delete_insn): Call it here instead.
13532         * lra-spills.c (lra_final_code_change): Use delete_insn.
13533         * haifa-sched.c (sched_remove_insn): Likewise.
13534         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
13535         returning to the nop pool.
13536         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
13537         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
13539 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
13541         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
13542         * doc/tm.texi: Regenerated.
13544 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
13546         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
13547         QImode checks.
13549 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
13551         * df-core.c (df_find_def): Compare register numbers.
13552         (df_find_use): Likewise.
13554 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
13556         PR target/56903
13557         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
13558         lra_in_progress for return.
13560 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
13562         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
13563         define_insn into define_insn_and_split and emit movsicc patterns.
13565 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
13567         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
13569 2013-04-12  Richard Biener  <rguenther@suse.de>
13571         * tree-pass.h (TODO_do_not_ggc_collect): New.
13572         * passes.c (execute_one_ipa_transform_pass): Honor
13573         TODO_do_not_ggc_collect.
13574         (execute_one_pass): Likewise.
13576         Revert
13577         2013-04-10  Richard Biener  <rguenther@suse.de>
13579         * passes.c (init_optimization_passes): Remove reload pass.
13580         * ira.c (do_reload): Merge into ...
13581         (ira): ... this.
13582         (rest_of_handle_reload): Remove.
13583         (pass_reload): Likewise.
13584         * config/i386/i386.c (ix86_option_override): Refer to ira instead
13585         of reload for vzeroupper pass placement.
13587 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
13589         PR tree-optimization/56918
13590         PR tree-optimization/56920
13591         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
13592         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
13593         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
13594         use wide_mul_with_sign method.
13596 2013-04-12  Richard Biener  <rguenther@suse.de>
13598         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
13599         not be considered a gimple constant.
13601 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
13603         * fold-const.c (const_binop): Handle vector shifts by a scalar.
13604         (fold_binary_loc): Call const_binop also for mixed vector-scalar
13605         operations.
13607 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
13608             Jakub Jelinek  <jakub@redhat.com>
13610         * opts.c: Include diagnostic-color.h.
13611         (common_handle_option): Handle OPT_fdiagnostics_color_.
13612         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
13613         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
13614         (diagnostic-color.o): New.
13615         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
13616         (diagnostic_color_rule): New enum.
13617         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
13618         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
13619         the location string.
13620         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
13621         either NULL, or color kind.
13622         * diagnostic-color.c: New file.
13623         * diagnostic-color.h: New file.
13624         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
13625         arguments.
13626         * doc/invoke.texi (-fdiagnostics-color): Document.
13627         * pretty-print.h (pp_show_color): Define.
13628         (struct pretty_print_info): Add show_color field.
13629         * diagnostic.c: Include diagnostic-color.h.
13630         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
13631         macros.  Colorize error:, warning: etc. strings and also the location
13632         string.
13633         (diagnostic_show_locus): Colorize the caret line.
13634         * pretty-print.c: Include diagnostic-color.h.
13635         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
13636         inside of %< %> quotes or quoted through q format modifier.
13638 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13640         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
13642 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13644         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
13645         code in CC_NZ mode.
13646         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
13647         pattern.
13649 2013-04-11  Marek Polacek  <polacek@redhat.com>
13651         PR tree-optimization/48184
13652         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
13654 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
13656         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
13657         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
13658         (skip_simple_arithmetic): Tidy up.
13659         * tree.h (skip_simple_constant_arithmetic): Declare.
13661 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13663         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
13665 2013-04-11  Richard Biener  <rguenther@suse.de>
13667         * tree-vect-loop.c (get_initial_def_for_induction): Properly
13668         generate vector constants.
13670 2013-04-11  Richard Biener  <rguenther@suse.de>
13672         PR tree-optimization/56878
13673         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
13674         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
13675         New function.
13676         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
13677         Prefer to align the DR with the most invariant base address.
13679 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
13681         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
13682         comment.
13684 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
13686         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
13687         floating-point vector comparisons against 0.
13689 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
13691         PR tree-optimization/56899
13692         * fold-const.c (extract_muldiv_1): Apply distributive law
13693         only if TYPE_OVERFLOW_WRAPS (ctype).
13695 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
13697         PR target/56124
13698         * ira-costs.c (scan_one_insn): Check whether the source rtx of
13699         loading has side effect.
13701 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
13703         * config/sparc/sparc.c: Include tree-pass.h.
13704         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
13705         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
13706         head of file.  Change return type.  Split off gate function.
13707         (sparc_gate_work_around_errata): New function.
13708         (pass_work_around_errata): New pass definition.
13709         (insert_pass_work_around_errata) New pass insert definition to
13710         insert pass_work_around_errata just after delayed-branch scheduling.
13711         (sparc_option_override): Insert the pass.
13712         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
13714 2013-04-10  David S. Miller  <davem@davemloft.net>
13716         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
13717         or -mcpu=hypersparc.
13719         * target.def (cstore_mode): New hook.
13720         * target.h: Include insn-codes.h
13721         * targhooks.c: Likewise.
13722         (default_cstore_mode): New function.
13723         * targhooks.h: Declare it.
13724         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
13725         * doc/tm.texi: Rebuild.
13726         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
13727         target hook, rather than inspecting the insn_data.
13728         * config/sparc/sparc.c (sparc_cstore_mode): New function.
13729         (TARGET_CSTORE_MODE): Redefine.
13730         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
13731         result patterns.
13732         * config/sparc/predicates.md (cstore_result_operand): New special
13733         predicate.
13734         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
13735         Use it for operand 0.
13736         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
13737         (*snesi_special): Likewise.
13738         (*snesi_zero): Likewise.
13739         (*seqsi_zero): Likewise.
13740         (*sltu_insn): Likewise.
13741         (*sgeu_insn): Likewise.
13742         (*seqdi_special): Make operand 0 and comparison operation be of
13743         DImode.
13744         (*snedi_special): Likewise.
13745         (*snedi_special_vis3): Likewise.
13746         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
13747         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
13748         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
13749         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
13750         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
13751         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
13752         (*sltu_extend_sp64): Likewise.
13753         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
13754         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
13755         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
13756         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
13757         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
13759 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
13761         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
13762         (aarch64_start_file): Use the new function.
13764 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
13765             Jason Merrill  <jason@redhat.com>
13767         * common.opt: Add -gdwarf.
13768         * opts.c (common_handle_option): Handle it.
13769         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
13771 2013-04-10  Richard Biener  <rguenther@suse.de>
13773         * passes.c (execute_todo): Do not call ggc_collect conditional here.
13774         (execute_one_ipa_transform_pass): But unconditionally here.
13775         (execute_one_pass): And here.
13776         (init_optimization_passes): Remove reload pass.
13777         * tree-pass.h (TODO_ggc_collect): Remove.
13778         (pass_reload): Likewise.
13779         * ira.c (do_reload): Merge into ...
13780         (ira): ... this.
13781         (rest_of_handle_reload): Remove.
13782         (pass_reload): Likewise.
13783         * config/i386/i386.c (ix86_option_override): Refer to ira instead
13784         of reload for vzeroupper pass placement.
13785         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
13786         and todo_flags_finish of all passes.
13788 2013-04-10  Richard Biener  <rguenther@suse.de>
13790         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
13791         first_const_oprnd field, rename first_def_type to first_op_type.
13792         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
13793         (vect_get_and_check_slp_defs): Always use the type of the
13794         operand.  Allow mixed vect_external_def, vect_constant_def types.
13795         (vect_get_constant_vectors): Handle mixed vect_external_def,
13796         vect_constant_def types.
13798 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
13800         PR tree-optimization/55524
13801         * tree-ssa-math-opts.c
13802         (convert_mult_to_fma): Don't use an fms construct
13803         when we don't have an fms operation, but fnma, and it looks
13804         likely that we'll be able to use the latter.
13806 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
13808         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
13809         function.
13810         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
13811         inline fail caused by overwritable functions.
13813 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
13815         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
13816         unnecessary bits in the constant power of two case.
13818 2013-04-10  Richard Biener  <rguenther@suse.de>
13820         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
13821         broken code swapping operands.
13822         (vect_build_slp_tree): Do not compute load permutations here.
13823         (vect_analyze_slp_instance): Compute load permutations here,
13824         after building the SLP tree.
13826 2013-04-09  Christian Bruel  <christian.bruel@st.com>
13828         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
13829         of next/prev_real_insn.
13831 2013-04-09  Jan Hubicka  <jh@suse.cz>
13833         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
13834         Drop aliased parameter.
13835         (function_and_variable_visibility): Do not handle alias pairs.
13836         * cgraph.c (varpool_externally_visible_p): Update prototype.
13837         * varpool.c (varpool_add_new_variable): Update.
13839 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13841         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
13843 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
13845         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
13847         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
13849 2013-04-09  Marek Polacek  <polacek@redhat.com>
13851         PR tree-optimization/48762
13852         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
13854 2013-04-09  Richard Biener  <rguenther@suse.de>
13856         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
13857         dealing with cost.
13858         (vect_build_slp_tree): Likewise.
13859         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
13860         calculating the cost of a SLP instance.
13861         (vect_analyze_slp_instance): Use it from here, after building
13862         the SLP tree.
13864 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13866         PR middle-end/56883
13867         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
13868         expand_omp_for_static_chunk): Use simple_p = true in
13869         force_gimple_operand_gsi calls when assigning to addressable decls.
13871 2013-04-09  Jeff Law  <law@redhat.com>
13873         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
13874         when the boolean was created by converting a wider object which
13875         had a boolean range.
13877 2013-04-09  Richard Biener  <rguenther@suse.de>
13879         * tree-vectorizer.h (slp_void_p): Remove.
13880         (slp_tree): Typedef before _slp_tree declaration.
13881         (struct _slp_tree): Use a vector of slp_tree as children.
13882         (vect_get_place_in_interleaving_chain): Remove.
13883         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
13884         Move ...
13885         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
13886         and make static.
13887         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
13888         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
13889         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
13890         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
13891         Use slp_node instead of slp_void_p and adjust.
13893 2013-04-09  Richard Biener  <rguenther@suse.de>
13895         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
13896         work that is not necessary.
13898 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
13900         PR tree-optimization/56854
13901         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
13902         forward into clobber stmts if it would change MEM_REF lhs into
13903         non-MEM_REF.
13905 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
13907         * tree.c (type_hash_lookup, type_hash_add): Make static.
13908         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
13910 2013-04-09  Richard Biener  <rguenther@suse.de>
13912         * tree.h (unsave_expr_now): Remove.
13913         * tree-inline.c (mark_local_for_remap_r): Remove.
13914         (unsave_expr_1): Likewise.
13915         (unsave_r): Likewise.
13916         (unsave_expr_now): Likewise.
13917         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
13918         (propagate_tree_value): Likewise.
13920 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
13922         * doc/rtl.texi (sequence): Rewrite documentation to match the
13923         current use of SEQUENCE rtl objects.
13924         * rtl.def (SEQUENCE): Likewise.
13926         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
13927         Update documentation.
13928         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
13929         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
13931         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
13933 2013-04-08  Teresa Johnson  <tejohnson@google.com>
13935         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
13936         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
13937         methods.
13938         (estimate_edge_size_and_time): Add comment to suggest using rounding
13939         methods.
13940         (estimate_node_size_and_time): Ditto.
13941         (remap_edge_change_prob): Use helper rounding divide methods.
13942         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13943         (gimple_mod_pow2_value_transform): Ditto.
13944         (gimple_mod_subtract_transform): Ditto.
13945         (gimple_ic_transform): Ditto.
13946         (gimple_stringops_transform): Ditto.
13947         * stmt.c (conditional_probability): Ditto.
13948         (emit_case_dispatch_table): Ditto.
13949         * lto-cgraph.c (merge_profile_summaries): Ditto.
13950         * tree-optimize.c (execute_fixup_cfg): Ditto.
13951         * cfgcleanup.c (try_forward_edges): Ditto.
13952         * cfgloopmanip.c (scale_loop_profile): Ditto.
13953         (loopify): Ditto.
13954         (duplicate_loop_to_header_edge): Ditto.
13955         (lv_adjust_loop_entry_edge): Ditto.
13956         * tree-vect-loop.c (vect_transform_loop): Ditto.
13957         * profile.c (compute_branch_probabilities): Ditto.
13958         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13959         * lto-streamer-in.c (input_cfg): Ditto.
13960         * gimple-streamer-in.c (input_bb): Ditto.
13961         * ipa-cp.c (update_profiling_info): Ditto.
13962         (update_specialized_profile): Ditto.
13963         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13964         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13965         rounding methods.
13966         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13967         (compute_trg_info): Ditto.
13968         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13969         (purge_dead_edges): Ditto.
13970         * loop-unswitch.c (unswitch_loop): Ditto.
13971         * cgraphclones.c (cgraph_clone_edge): Ditto.
13972         (cgraph_clone_node): Ditto.
13973         * tree-inline.c (copy_bb): Ditto.
13974         (copy_edges_for_bb): Ditto.
13975         (initialize_cfun): Ditto.
13976         (copy_cfg_body): Ditto.
13977         (expand_call_inline): Ditto.
13979 2013-04-08  Kai Tietz  <ktietz@redhat.com>
13981         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13982         TARGET_CYGWIN64 by TARGET_64BIT.
13984 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
13986         * config/epiphany/epiphany.md (GPR_1): New constant.
13987         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13988         * config/epiphany/epiphany.c (gen_compare_reg):
13989         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13990         is already in place.
13991         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13992         Don't require being called during rtl expansion; If y operlaps r0,
13993         return 0.
13994         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13995         (epiphany_expand_epilogue): Likewise.
13997         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13998         Don't use CC_FPmode for ORDERED / UNORDERED.
13999         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
14001         * config/epiphany/constraints.md (CnL): New constraint.
14002         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
14003         * config/epiphany/predicates.md (add_operand): Allow 1024.
14005         * config/epiphany/epiphany.md (logical_op): New code iterator.
14006         (op_mnc): New code attribute.
14007         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
14008         (mov_f+1, mov_f+2): New peephole2 patterns.
14010         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
14011         (cstoresi4): Also allow re-use of zero result when doing a NE
14012         comparison to a non-zero operand.
14013         Use (clobber (scratch)) for first insn if the gpr output is not needed.
14015         * config/epiphany/epiphany.md (<insn_opname>v2si3):
14016         Use gen_addsi3_i / gen_subsi3_i.
14018 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
14020         PR c++/34949
14021         PR c++/50243
14022         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
14023         contain anything but clobbers, at most one __builtin_stack_restore,
14024         optionally debug stmts and final resx, and if it has at least one
14025         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
14026         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
14027         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
14028         which isn't defaut definition, remove them.
14029         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
14030         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
14031         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
14032         with MEM_REF LHS with SSA_NAME address.
14034 2013-04-08  Jeff Law  <law@redhat.com>
14036         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
14038 2013-04-08  Richard Biener  <rguenther@suse.de>
14040         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
14041         extra newline.
14042         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
14043         determined vector type.
14044         (vect_analyze_data_refs): Likewise.
14045         (vect_get_new_vect_var): Adjust.
14046         (vect_create_destination_var): Preserve SSA name versions.
14047         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
14048         not dump anything here.
14050 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
14052         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
14053         Add member lr_slot_known.
14054         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
14055         if necessary.
14056         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
14057         Remove code that sets lr_slot_offset according to what a previous
14058         version of epiphany_emit_save_restore used to do.
14059         (epiphany_emit_save_restore): When doing an lr save or restore,
14060         set/verify lr_slot_known and lr_slot_offset.
14062 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
14064         PR target/54338
14065         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
14066         in ALL_REGS.
14068 2013-04-08  Richard Biener  <rguenther@suse.de>
14070         * alias.c (find_base_term): Fix thinko in previous change.
14072 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
14074         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
14075         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
14076         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
14077         if possible to compute val.
14078         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
14079         For QImode integers don't require anything about precision.  Use
14080         const_with_all_bytes_same to find out if the constant doesn't have
14081         repeated bytes in it.
14083 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14085         * config/s390/s390.c (s390_expand_insv): Only accept insertions
14086         within mode size.
14088 2013-04-08  Marek Polacek  <polacek@redhat.com>
14090         PR rtl-optimization/48182
14091         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
14092         value to 1.
14094 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14096         PR target/55487
14097         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
14098         nuses, make sure we have a label.
14100 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14102         PR target/56843
14103         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
14104         (rs6000_emit_swdiv_low_precision): Remove.
14105         (rs6000_emit_swdiv): Rewrite to handle between one and four
14106         iterations of Newton-Raphson generally; modify required number of
14107         iterations for some cases.
14108         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
14110 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
14112         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
14113         set-but-unused variable.
14115         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
14116         basic blocks of released function bodies garbage-collectable.
14118         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
14119         (struct rtl_opt_pass): Add TODO_df_finish.
14121         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
14123 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14125         * config/arm/constraints.md (q): New constraint.
14126         * config/arm/ldrdstrd.md: New file.
14127         * config/arm/arm.md (ldrdstrd.md) New include.
14128         (arm_movdi): Use "q" instead of "r" constraint
14129         for double-word memory access.
14130         (movdf_soft_insn): Likewise.
14131         * config/arm/vfp.md (movdi_vfp): Likewise.
14132         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
14133         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
14134         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
14135         (mem_ok_for_ldrd_strd): Likewise.
14136         (output_move_double): Update assertion.
14138 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14140         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
14142 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14144         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
14145         define_insn_and_split.
14146         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
14148 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14150         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
14151         define_insn_and_split.
14152         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
14153         (shiftsi3_compare): New pattern.
14154         (rrx): New pattern.
14155         * config/arm/unspecs.md (UNSPEC_RRX): New.
14157 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14159         * config/arm/arm.md (negdi_extendsidi): New pattern.
14160         (negdi_zero_extendsidi): Likewise.
14162 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14164         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
14165         define_insn_and_split.
14166         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
14167         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
14169 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14171         * config/arm/arm.md (arm_subdi3): Convert define_insn into
14172         define_insn_and_split.
14173         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
14174         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
14176 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14178         * config/arm/arm.md (subsi3_carryin): New pattern.
14179         (subsi3_carryin_const): Likewise.
14180         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
14181         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
14183 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14185         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
14187 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
14189         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
14190         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
14192 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14194         * config/arm/arm.c (arm_expand_builtin): Change fcode
14195         type to unsigned int.
14197 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14199         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
14201 2013-04-04  Ian Lance Taylor  <iant@google.com>
14203         * doc/standards.texi (Standards): The Go frontend supports the Go 1
14204         language standard.
14206 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
14208         PR middle-end/56729
14209         * df-scan.c (df_insn_delete): Disable failing assert.
14211 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14213         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
14214         New function prototype.
14215         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
14216         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
14217         (arm_builtin_vectorized_function): New function.
14219 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14221         * config/arm/arm_neon_builtins.def: New file.
14222         * config/arm/arm.c (neon_builtin_data): Move contents to
14223         arm_neon_builtins.def.
14224         (enum arm_builtins): Include neon builtin definitions.
14225         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
14226         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
14228 2013-04-04  Marek Polacek  <polacek@redhat.com>
14230         PR tree-optimization/48186
14231         * predict.c (maybe_hot_frequency_p): Return false if
14232         HOT_BB_FREQUENCY_FRACTION is 0.
14233         (cgraph_maybe_hot_edge_p): Likewise.
14235 2013-04-04  Richard Biener  <rguenther@suse.de>
14237         PR tree-optimization/56826
14238         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
14239         more accurately.
14241 2013-04-04  Richard Biener  <rguenther@suse.de>
14243         PR tree-optimization/56213
14244         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
14245         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
14247 2013-04-04  Richard Biener  <rguenther@suse.de>
14249         PR tree-optimization/56837
14250         * tree-loop-distribution.c (classify_partition): For non-zero
14251         values require that the value has the same precision as its
14252         mode to be useful as memset value.
14254 2013-04-03  Nick Clifton  <nickc@redhat.com>
14256         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
14257         (fmssf4): Use fmsf.s on E3V5 architectures.
14258         (fnmasf4): Use fnmaf.s on E3V5 architectures.
14259         (fnmssf4): Use fnmsf.s on E3V5 architectures.
14261 2013-04-03  Jeff Law  <law@redhat.com>
14263         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
14264         (lra-eliminations.o): Likewise.
14266 2013-04-03  Teresa Johnson  <tejohnson@google.com>
14268         * gcov-io.c (compute_working_sets): Moved most of body of old
14269         compute_working_sets here from profile.c.
14270         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
14271         (gcov_working_set_t): Moved typedef here from basic-block.h
14272         (compute_working_set): Declare.
14273         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
14274         (get_working_sets): Renamed from compute_working_set,
14275         replace most of body with call to new compute_working_sets.
14276         (get_exec_counts): Replace call to compute_working_sets
14277         to get_working_sets.
14278         * profile.h (get_working_sets): Renamed from compute_working_set.
14279         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
14280         to get_working_sets.
14281         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
14282         * gcov-dump.c (dump_working_sets): New function.
14284 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
14286         * hwint.c (sext_hwi, zext_hwi): New functions.
14287         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
14288         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
14289         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
14290         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
14291         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
14292         (sext_hwi, zext_hwi): New functions.
14294 2013-04-03  Jeff Law  <law@redhat.com>
14296         PR tree-optimization/56799
14297         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
14298         back test for widening conversion erroneously dropped in prior change.
14300 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14302         PR target/56809
14303         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
14304         instead of next_real_insn.
14306 2013-04-03  Marek Polacek  <polacek@redhat.com>
14308         PR sanitizer/55702
14309         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
14311 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14313         PR target/56809
14314         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
14315         next_real_insn.
14316         (thumb1_output_casesi): Likewise.
14317         (thumb2_output_casesi): Likewise.
14319 2013-04-03  Richard Biener  <rguenther@suse.de>
14321         PR tree-optimization/56817
14322         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
14323         Split out ...
14324         (tree_unroll_loops_completely_1): ... new function to manually
14325         walk the loop tree, properly defering outer loops of unrolled
14326         loops to later iterations.
14328 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
14330         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
14331         (vectorizable_load): Likewise.
14332         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
14333         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
14335 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
14337         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
14338         BIT_FIELD_REF.
14340 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
14342         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
14344 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
14346         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
14348 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
14350         PR tree-optimization/56790
14351         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
14352         folding.
14354 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
14356         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
14357         Handle VEC_MERGE.
14358         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
14359         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
14360         equal arguments.
14362 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
14364         PR c/19449
14365         * tree.h (force_folding_builtin_constant_p): New decl.
14366         * builtins.c (force_folding_builtin_constant_p): New variable.
14367         (fold_builtin_constant_p): Fold immediately also if
14368         force_folding_builtin_constant_p.
14370 2013-04-03  Richard Biener  <rguenther@suse.de>
14372         PR tree-optimization/56812
14373         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
14374         DRs of the same interleaving chain are independent.
14376 2013-04-02  Jason Merrill  <jason@redhat.com>
14378         * gdbinit.in (pbb): Use debug fn.
14380 2013-04-02  Lawrence Crowl  <crowl@google.com>
14382         * sese.h (struct ivtype_map_elt_s): Remove unused.
14383         (extern debug_ivtype_map): Remove unused.
14384         (extern eq_ivtype_map_elts): Remove unused.
14385         * sese.c (debug_ivtype_map): Removed unused.
14386         (debug_ivtype_map_1): Removed unused.
14387         (debug_ivtype_elt): Remove unused.
14388         (eq_ivtype_map_elts): Remove unused.
14391 2013-04-02  Kai Tietz  <ktietz@redhat.com>
14393         PR target/52790
14394         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
14395         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
14396         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
14397         function.
14398         (legitimize_pe_coff_symbol): Likewise.
14399         (is_imported_p): New helper-function.
14400         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
14401         for Windows x64 targets.
14402         (ix86_expand_prologue): Optimize for pe-coff targets.
14403         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
14404         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
14405         medium/large code-model.
14406         (legitimize_pic_address): Likewise.
14407         (legitimize_tls_address): Likewise.
14408         (ix86_expand_call): Likewise.
14409         (x86_output_mi_thunk): Likewise.
14410         (get_dllimport_decl): Add new beimport argument.
14411         (construct_plt_address): Don't assert for x64 pe-coff targets.
14412         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
14413         targets.
14414         (SYMBOL_FLAG_STUBVAR): New macro.
14415         (SYMBOL_REF_STUBVAR_P): Likewise.
14416         * config/i386/winnt.c (stub_list): New structure.
14417         (stub_head): New local variable.
14418         (i386_pe_record_stub): New function.
14419         (i386_pe_file_end): Emit refptr-stubs.
14421 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
14423         PR rtl-optimization/56745
14424         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
14425         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
14427         PR c++/34949
14428         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
14429         and both of them are MEM_REFs, just compare first argument for
14430         equality and attempt to deal even with differing offsets.
14432         PR c++/34949
14433         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
14434         of gimple_clobber_p to be MEM_REF.
14435         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
14436         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
14437         after gimplification.
14438         * asan.c (get_mem_ref_of_assignment): Don't instrument
14439         gimple_clobber_p stmts.
14440         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
14441         gimple_clobber_p stmt if they have MEM_REF lhs and
14442         are dead because of another gimple_clobber_p stmt.
14443         * tree-ssa-live.c (clear_unused_block_pointer): Treat
14444         gimple_clobber_p stmts like debug stmts.
14445         (remove_unused_locals): Remove clobbers with MEM_REF lhs
14446         that refer to unused VAR_DECLs or uninitialized values.
14447         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
14448         gimple_clobber_p stmts if they refer to removed parameters.
14449         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
14450         formatting.
14452 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
14454         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
14455         using SWI48 mode attribute.
14457 2013-04-02  Wei Mi  <wmi@google.com>
14459         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
14460         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
14461         *<rotate_insn><mode>3_mask in i386.md.
14463 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
14465         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
14467 2013-04-02  Richard Biener  <rguenther@suse.de>
14469         PR tree-optimization/56778
14470         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
14471         Runtime alias tests are not supported for gather loads.
14472         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
14473         stmts referenced from SSA operands before updating SSA form.
14475 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
14476             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14478         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
14479         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
14480         * config/arm/cortex-a53.md: New file.
14481         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
14482         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
14483         * config/arm/arm.c (arm_issue_rate): Likewise.
14484         * config/arm/arm-tune.md: Regenerate
14485         * config/arm/arm-tables.opt: Regenerate.
14486         * config/arm/arm-cores.def: Add cortex-a53.
14488 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
14490         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
14491         non-static link.
14493 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
14495         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
14496         scalar load/store operations using B/H registers.
14497         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
14499 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
14501         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
14502         scalar move.
14503         * config/aarch64/aarch64.c
14504         (aarch64_simd_scalar_immediate_valid_for_move): New.
14505         * config/aarch64/aarch64-protos.h
14506         (aarch64_simd_scalar_immediate_valid_for_move): New.
14507         * config/aarch64/constraints.md (Dh, Dq): New.
14508         * config/aarch64/iterators.md (hq): New.
14510 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
14512         * reorg.c (get_branch_condition): Deal with conditional returns.
14513         (fill_simple_delay_slots): Remove dead code dealing with jumps.
14515 2013-04-01  Wei Mi  <wmi@google.com>
14517         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
14518         Truncate operand 2 using %b asm operand modifier.
14519         (*<shift_insn><mode>3_mask): Ditto.
14520         (*<rotate_insn><mode>3_mask): Ditto.
14522 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
14524         PR middle-end/56798
14525         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
14527 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
14529         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
14530         of next_real_insn.
14531         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
14533 2013-03-30  Lawrence Crowl  <crowl@google.com>
14535         * dse.c (clear_alias_sets): Remove never set.
14536         (disqualified_clear_alias_sets): Remove never set.
14537         (clear_alias_mode_pool): Remove never set.
14538         (dse_step0): Remove condition that is never true.
14539         (canon_address): Remove condition that is never true.
14540         (dse_step7): Remove condition that is never true.
14541         (rest_of_handle_dse): Remove condition that is never true.
14542         (rest_of_handle_dse::did_global): Remove never read from above.
14543         (dse_step2_spill): Remove never called from above.
14544         (dse_step5_spill): Remove never called from above.
14546 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
14548         * doc/md.texi (Standard Names) <casesi>: Update documentation for
14549         JUMP_TABLE_DATA changes.
14550         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
14551         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
14552         (Insns) <jump_table_data>: New entry.
14553         * doc/tm.texi: Regenerate.
14555         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
14557         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
14558         for table jump at the end of a basic block using tablejump_p.
14559         * targhooks.c (default_invalid_within_doloop): Likewise.
14560         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
14561         target hook implementation that is identical to the default hook.
14562         (rs6000_invalid_within_doloop): Remove.
14564         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
14565         unused variable from tablejump_p call.
14567         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
14568         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
14569         (INSN_DELETED_P): Likewise.
14570         (emit_jump_table_data): New prototype.
14571         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
14572         after 4th as unused.
14573         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
14574         * sched-vis.c (print_insn): Likewise.
14575         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
14576         insn for compatibility with back ends that use next_active_insn to
14577         identify jump table data.
14578         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
14579         (remove_insn): Likewise.
14580         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
14581         to be emitted.
14582         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
14583         (emit_jump_table_data): New function.
14585         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
14586         basic block, a JUMP_TABLE_DATA never is.
14587         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
14588         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
14589         off from code handling real insns.
14590         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
14591         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
14592         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
14593         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
14594         is not a NONDEBUG_INSN_P.
14595         * ira-costs.c (scan_one_insn): Likewise.
14596         * jump.c (mark_all_labels): Likewise.
14597         (mark_jump_label_1): Likewise.
14598         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
14599         * lra.c (get_insn_freq): Expect all insns reaching here to be in
14600         a basic block.
14601         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
14602         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
14603         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
14604         JUMP_TABLE_DATA_P insns.
14605         (calculate_elim_costs_all_insns): Likewise.
14606         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
14607         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
14608         (delete_output_reload): Code style fixups.
14609         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
14610         insn flags on this non-insn.
14611         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
14612         as scheduling barriers, for pre-change compatibility.
14613         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
14614         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
14616         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
14617         redundant JUMP_TABLE_DATA_P test.
14618         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
14619         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
14620         (frv_for_each_packet): Likewise.
14621         * config/i386/i386.c (min_insn_size): Likewise.
14622         (ix86_avoid_jump_mispredicts): Likewise.
14623         * config/m32r/m32r.c (m32r_is_insn): Likewise.
14624         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
14625         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
14626         (mips16_insn_length): Robustify.
14627         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
14628         (mips16_split_long_branches): Likewise.
14629         * config/pa/pa.c (pa_combine_instructions): Likewise.
14630         * config/rs6000/rs6000.c (get_next_active_insn): Treat
14631         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
14632         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
14633         as contributing to pool range lengths.
14634         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
14635         Remove redundant JUMP_TABLE_DATA_P test.
14636         (sh_loop_align): Likewise.
14637         (split_branches): Likewise.
14638         (sh_insn_length_adjustment): Likewise.
14639         * config/spu/spu.c (get_branch_target): Likewise.
14641 2013-03-29  Jan Hubicka  <jh@suse.cz>
14643         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
14644         gcov streaming; stream hot bb threshold to ltrans.
14645         * predict.c (get_hot_bb_threshold): Break out from ....
14646         (maybe_hot_count_p): ... here.
14647         (set_hot_bb_threshold): New function.
14648         * lto-section-in.c (lto_section_name): Add profile.
14649         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
14650         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
14651         and data-streamer.h
14652         (histogram_entry): New structure.
14653         (histogram, histogram_pool): New global vars.
14654         (histogram_hash): New structure.
14655         (histogram_hash::hash): New method.
14656         (histogram_hash::equal): Likewise.
14657         (account_time_size): New function.
14658         (cmp_counts): New function.
14659         (dump_histogram): New function.
14660         (ipa_profile_generate_summary): New function.
14661         (ipa_profile_write_summary): New function.
14662         (ipa_profile_read_summary): New function.
14663         (ipa_profile): Decide on threshold.
14664         (pass_ipa_profile): Add ipa_profile_write_summary and
14665         ipa_profile_read_summary.
14666         * Makefile.in (ipa.o): Update dependencies.
14667         * lto-streamer.h (LTO_section_ipa_profile): New section.
14669 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14671         * tree.h (VAR_P): New.
14673 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
14675         PR lto/56777
14676         * doc/invoke.texi ([-fwhole-program]): Fix typo.
14678 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
14680         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
14681         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
14682         (control_flow_insn_p): Likewise.
14683         * cfgrtl.c (duplicate_insn_chain): Likewise.
14684         * final.c (get_attr_length_1): Likewise.
14685         (shorten_branches): Likewise.
14686         (final_scan_insn): Likewise.
14687         * function.c (instantiate_virtual_regs): Likewise.
14688         * gcse.c (insert_insn_end_basic_block): Likewise.
14689         * ira-costs.c (scan_one_insn): Likewise.
14690         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
14691         * lra.c (check_rtl): Likewise.
14692         * reload1.c (elimination_costs_in_insn): Likewise.
14693         * reorg.c (follow_jumps): Likewise.
14695         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
14696         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
14697         (thumb_far_jump_used_p): Likewise.
14698         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
14699         (workaround_speculation): Likewise.
14700         (add_sched_insns_for_speculation): Likewise.
14701         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
14702         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
14703         (frv_for_each_packet): Likewise.
14704         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
14705         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
14706         (final_emit_insn_group_barriers): Likewise.
14707         * config/m32r/m32r.c (m32r_is_insn): Likewise.
14708         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
14709         (mips16_insn_length): Likewise.
14710         * config/pa/pa.c (pa_reorg): Likewise.
14711         (pa_combine_instructions): Likewise.
14712         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
14713         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
14714         (sh_reorg): Likewise.
14715         (split_branches): Likewise.
14716         * config/spu/spu.c (get_branch_target): Likewise.
14718         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
14719         JUMP_TABLE_DATA_P.
14721 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
14723         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
14724         Fix declaration name.
14726 2013-03-28  Lawrence Crowl  <crowl@google.com>
14728         * graphds.h (struct graph.indicies): Remove unused.
14729         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
14730         (SCOP_ORIGINAL_PDDRS): Remove unused.
14731         * sese.h (extern insert_loop_close_phis): Removed unused.
14732         (extern insert_guard_phis): Removed unused.
14733         (extern ivtype_map_elt_info): Removed unused.
14734         (new_ivtype_map_elt): Removed unused.
14735         * sese.c (ivtype_map_elt_info): Removed unused.
14737 2013-03-28  Lawrence Crowl  <crowl@google.com>
14739         * Makefile.in: Add several missing include dependences.
14740         (DUMPFILE_H): New.
14741         (test-dump.o): New.  This object is not added to any executable,
14742         but is present for ad-hoc testing.
14743         * bitmap.c
14744         (debug (const bitmap_head_def &)): New.
14745         (debug (const bitmap_head_def *)): New.
14746         * bitmap.h
14747         (extern debug (const bitmap_head_def &)): New.
14748         (extern debug (const bitmap_head_def *)): New.
14749         * cfg.c
14750         (debug (edge_def &)): New.
14751         (debug (edge_def *)): New.
14752         * cfghooks.c
14753         (debug (basic_block_def &)): New.
14754         (debug (basic_block_def *)): New.
14755         * dumpfile.h
14756         (dump_node (const_tree, int, FILE *)): Correct source file.
14757         * dwarf2out.c
14758         (debug (die_struct &)): New.
14759         (debug (die_struct *)): New.
14760         * dwarf2out.h
14761         (extern debug (die_struct &)): New.
14762         (extern debug (die_struct *)): New.
14763         * gimple-pretty-print.c
14764         (debug (gimple_statement_d &)): New.
14765         (debug (gimple_statement_d *)): New.
14766         * gimple-pretty-print.h
14767         (extern debug (gimple_statement_d &)): New.
14768         (extern debug (gimple_statement_d *)): New.
14769         * ira-build.c
14770         (debug (ira_allocno_copy &)): New.
14771         (debug (ira_allocno_copy *)): New.
14772         (debug (ira_allocno &)): New.
14773         (debug (ira_allocno *)): New.
14774         * ira-int.h
14775         (extern debug (ira_allocno_copy &)): New.
14776         (extern debug (ira_allocno_copy *)): New.
14777         (extern debug (ira_allocno &)): New.
14778         (extern debug (ira_allocno *)): New.
14779         * ira-lives.c
14780         (debug (live_range &)): New.
14781         (debug (live_range *)): New.
14782         * lra-int.h
14783         (debug (lra_live_range &)): New.
14784         (debug (lra_live_range *)): New.
14785         * lra-lives.c
14786         (debug (lra_live_range &)): New.
14787         (debug (lra_live_range *)): New.
14788         * omega.c
14789         (debug (omega_pb_d &)): New.
14790         (debug (omega_pb_d *)): New.
14791         * omega.h
14792         (extern debug (omega_pb_d &)): New.
14793         (extern debug (omega_pb_d *)): New.
14794         * print-rtl.c
14795         (debug (const rtx_def &)): New.
14796         (debug (const rtx_def *)): New.
14797         * print-tree.c
14798         (debug_tree (tree): Move within file.
14799         (debug_raw (const tree_node &)): New.
14800         (debug_raw (const tree_node *)): New.
14801         (dump_tree_via_hooks (const tree_node *, int)): New.
14802         (debug (const tree_node &)): New.
14803         (debug (const tree_node *)): New.
14804         (debug_verbose (const tree_node &)): New.
14805         (debug_verbose (const tree_node *)): New.
14806         (debug_head (const tree_node &)): New.
14807         (debug_head (const tree_node *)): New.
14808         (debug_body (const tree_node &)): New.
14809         (debug_body (const tree_node *)): New.
14810         (debug_vec_tree (tree): Move and reimplement in terms of dump.
14811         (debug (vec<tree, va_gc> &)): New.
14812         (debug (vec<tree, va_gc> *)): New.
14813         * rtl.h
14814         (extern debug (const rtx_def &)): New.
14815         (extern debug (const rtx_def *)): New.
14816         * sbitmap.c
14817         (debug_raw (simple_bitmap_def &)): New.
14818         (debug_raw (simple_bitmap_def *)): New.
14819         (debug (simple_bitmap_def &)): New.
14820         (debug (simple_bitmap_def *)): New.
14821         * sbitmap.h
14822         (extern debug (simple_bitmap_def &)): New.
14823         (extern debug (simple_bitmap_def *)): New.
14824         (extern debug_raw (simple_bitmap_def &)): New.
14825         (extern debug_raw (simple_bitmap_def *)): New.
14826         * sel-sched-dump.c
14827         (debug (vinsn_def &)): New.
14828         (debug (vinsn_def *)): New.
14829         (debug_verbose (vinsn_def &)): New.
14830         (debug_verbose (vinsn_def *)): New.
14831         (debug (expr_def &)): New.
14832         (debug (expr_def *)): New.
14833         (debug_verbose (expr_def &)): New.
14834         (debug_verbose (expr_def *)): New.
14835         (debug (vec<rtx> &)): New.
14836         (debug (vec<rtx> *)): New.
14837         * sel-sched-dump.h
14838         (extern debug (vinsn_def &)): New.
14839         (extern debug (vinsn_def *)): New.
14840         (extern debug_verbose (vinsn_def &)): New.
14841         (extern debug_verbose (vinsn_def *)): New.
14842         (extern debug (expr_def &)): New.
14843         (extern debug (expr_def *)): New.
14844         (extern debug_verbose (expr_def &)): New.
14845         (extern debug_verbose (expr_def *)): New.
14846         (extern debug (vec<rtx> &)): New.
14847         (extern debug (vec<rtx> *)): New.
14848         * sel-sched-ir.h
14849         (_list_iter_cond_expr): Make inline instead of static.
14850         * sreal.c
14851         (debug (sreal &)): New.
14852         (debug (sreal *)): New.
14853         * sreal.h
14854         (extern debug (sreal &)): New.
14855         (extern debug (sreal *)): New.
14856         * tree.h
14857         (extern debug_raw (const tree_node &)): New.
14858         (extern debug_raw (const tree_node *)): New.
14859         (extern debug (const tree_node &)): New.
14860         (extern debug (const tree_node *)): New.
14861         (extern debug_verbose (const tree_node &)): New.
14862         (extern debug_verbose (const tree_node *)): New.
14863         (extern debug_head (const tree_node &)): New.
14864         (extern debug_head (const tree_node *)): New.
14865         (extern debug_body (const tree_node &)): New.
14866         (extern debug_body (const tree_node *)): New.
14867         (extern debug (vec<tree, va_gc> &)): New.
14868         (extern debug (vec<tree, va_gc> *)): New.
14869         * tree-cfg.c
14870         (debug (struct loop &)): New.
14871         (debug (struct loop *)): New.
14872         (debug_verbose (struct loop &)): New.
14873         (debug_verbose (struct loop *)): New.
14874         * tree-dump.c: Add header dependence.
14875         * tree-flow.h
14876         (extern debug (struct loop &)): New.
14877         (extern debug (struct loop *)): New.
14878         (extern debug_verbose (struct loop &)): New.
14879         (extern debug_verbose (struct loop *)): New.
14880         * tree-data-ref.c
14881         (debug (data_reference &)): New.
14882         (debug (data_reference *)): New.
14883         (debug (vec<data_reference_p> &)): New.
14884         (debug (vec<data_reference_p> *)): New.
14885         (debug (vec<ddr_p> &)): New.
14886         (debug (vec<ddr_p> *)): New.
14887         * tree-data-ref.h
14888         (extern debug (data_reference &)): New.
14889         (extern debug (data_reference *)): New.
14890         (extern debug (vec<data_reference_p> &)): New.
14891         (extern debug (vec<data_reference_p> *)): New.
14892         (extern debug (vec<ddr_p> &)): New.
14893         (extern debug (vec<ddr_p> *)): New.
14894         * tree-ssa-alias.c
14895         (debug (pt_solution &)): New.
14896         (debug (pt_solution *)): New.
14897         * tree-ssa-alias.h
14898         (extern debug (pt_solution &)): New.
14899         (extern debug (pt_solution *)): New.
14900         * tree-ssa-alias.c
14901         (debug (_var_map &)): New.
14902         (debug (_var_map *)): New.
14903         (debug (tree_live_info_d &)): New.
14904         (debug (tree_live_info_d *)): New.
14905         * tree-ssa-alias.h
14906         (extern debug (_var_map &)): New.
14907         (extern debug (_var_map *)): New.
14908         (extern debug (tree_live_info_d &)): New.
14909         (extern debug (tree_live_info_d *)): New.
14911 2013-03-28  Jan Hubicka  <jh@suse.cz>
14913         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
14915 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
14917         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
14918         record only when desired or required.
14920 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
14922         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
14923         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
14925 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14927         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
14928         (*andsi3_compare0_uxtw): New pattern.
14929         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
14930         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
14932 2013-03-28  Jan Hubicka  <jh@suse.cz>
14934         * data-streamer-in.c (streamer_read_gcov_count): New function.
14935         * gimple-streamer-out.c: Include value-prof.h.
14936         (output_gimple_stmt): Output histogram.
14937         (output_bb): Use streamer_write_gcov_count.
14938         * value-prof.c: Include data-streamer.h
14939         (dump_histogram_value): Add HIST_TYPE_MAX.
14940         (stream_out_histogram_value): New function.
14941         (stream_in_histogram_value): New function.
14942         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14943         (stream_out_histogram_value, stream_in_histogram_value): Declare.
14944         * data-streamer-out.c (streamer_write_gcov_count): New function.
14945         (streamer_write_gcov_count_stream): New function.
14946         * lto-cgraph.c (lto_output_edge): Update counter streaming.
14947         (lto_output_node): Likewise.
14948         (input_node, input_edge): Likewise.
14949         * lto-streamer-out.c (output_cfg): Update streaming.
14950         * lto-streamer-in.c (input_cfg): Likewise.
14951         * data-streamer.h (streamer_write_gcov_count,
14952         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14953         * gimple-streamer-in.c: Include value-prof.h
14954         (input_gimple_stmt): Input histograms.
14955         (input_bb): Update profile streaming.
14957 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14959         * genmodes.c (emit_max_int): New function.
14960         (emit_insn_modes_h): Added call to emit_max_function.
14961         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14962         Added doc.
14963         * machmode.def: Fixed comment.
14965 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
14967         * combine.c (try_combine): Removed useless assert.
14968         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14970 2013-03-28  Marek Polacek  <polacek@redhat.com>
14971             Richard Biener  <rguenther@suse.de>
14973         PR tree-optimization/56695
14974         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14975         build signed result of a vector comparison.
14976         * tree-cfg.c (verify_gimple_comparison): Check that a result
14977         of a vector comparison has signed type.
14979 2013-03-28  Richard Biener  <rguenther@suse.de>
14981         PR tree-optimization/37021
14982         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14983         do not restrict gaps between groups.
14984         * tree-vect-stmts.c (vectorizable_load): Properly account for
14985         a gap between groups.
14987 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
14989         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14990         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14991         is not enabled.
14993 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
14995         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14996         * doc/extend.texi (Named Address Spaces): Ditto.
14997         (Variable Attributes): Ditto.
14999 2013-03-27  Kai Tietz  <ktietz@redhat.com>
15001         * config.build: Add support for cygwin x64 target.
15002         * config.gcc: Likewise.
15003         * config.host: Likewise.
15004         * configure.ac: Likewise
15005         * configure: Regenerated.
15007 2013-03-27  Kai Tietz  <ktietz@redhat.com>
15009         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
15010         * config/i386/t-cygwin-w64: New file.
15011         * config/i386/cygwin-w64.h: New file.
15012         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
15013         and add support for x64-cygwin target.
15014         (CPP_SPEC): Likewise.
15015         (CXX_WRAP_SPEC_LIST): Undefine before define.
15016         (LIBGCJ_SONAME): Use 15 as version.
15018 2013-03-27  Richard Biener  <rguenther@suse.de>
15020         PR tree-optimization/56716
15021         * tree-ssa-structalias.c (perform_var_substitution): Adjust
15022         dumping for ref nodes.
15024 2013-03-27  Martin Jambor  <mjambor@suse.cz>
15026         PR tree-optimization/55334
15027         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
15028         restricted pointers to arrays.
15030 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
15032         * Makefile.in (.SUFFIXES): Add .cc.
15033         (.c.o): Apply same recipe for implicit rule .cc.o.
15035 2013-03-27  Richard Biener  <rguenther@suse.de>
15037         PR tree-optimization/37021
15038         * tree-vect-data-refs.c (vect_check_strided_load): Allow
15039         REALPART/IMAGPART_EXPRs around the supported refs.
15040         * tree-ssa-structalias.c (find_func_aliases): Assume that
15041         floating-point values are not used to transfer pointers.
15043 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
15045         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
15046         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
15047         * gcc/doc/tm.texi: Regenerate.
15048         * gcc/targhooks.h (default_has_ifunc_p): New.
15049         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
15050         * gcc/config/linux-protos.h: New file.
15051         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
15052         this hook for linux which disables support of indirect functions in
15053         android.
15054         * gcc/config/linux-android.c: New file.
15055         * gcc/config/t-linux-android.c: Ditto.
15056         * gcc/config.gcc: Added new object file linux-android.o.
15057         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
15058         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
15059         * gcc/varasm.c (do_assemble_alias): Likewise.
15060         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
15061         doesn't support indirect functions.
15062         * configure: Regenerate.
15064 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
15066         PR target/56102
15067         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
15068         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
15069         mult-word mode.
15071 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15073         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
15075 2013-03-27  Terry Guo  <terry.guo@arm.com>
15077         * config/arm/arm-cores.def: Added core cortex-r7.
15078         * config/arm/arm-tune.md: Regenerated.
15079         * config/arm/arm-tables.opt: Regenerated.
15080         * doc/invoke.texi: Added entry for core cortex-r7.
15082 2013-03-27  Walter Lee  <walt@tilera.com>
15084         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
15085         double-decrement of next_scratch_regno.
15087 2013-03-27  Walter Lee  <walt@tilera.com>
15089         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
15090         input operands.
15091         (insn_v1mulus): Ditto.
15092         (insn_v2muls): Ditto.
15094 2013-03-27  Walter Lee  <walt@tilera.com>
15096         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
15097         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
15099 2013-03-27  Walter Lee  <walt@tilera.com>
15101         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
15102         (*sibcall_value): Ditto.
15104 2013-03-27  Walter Lee  <walt@tilera.com>
15106         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
15107         (insn_mnz_v8qi): ... this ...
15108         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
15109         vector equivalent.
15110         (insn_v<n>mnz): Replaced by ...
15111         (insn_v1mnz): ... this ...
15112         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
15113         equivalent.
15114         (insn_mz_<mode>): Replaced by ...
15115         (insn_mz_v8qi): ... this ...
15116         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
15117         vector equivalent.
15118         (insn_v<n>mz): Replaced by ...
15119         (insn_v1mz): ... this ...
15120         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
15121         equivalent.
15123 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
15125         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
15127 2013-03-26  Roland McGrath  <mcgrathr@google.com>
15129         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
15130         than fprintf with a non-constant, non-format string.
15132 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
15134         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
15135         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
15136         operand 0 predicate.
15137         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
15138         attribute.  Use general_x64nomem_operand as operand 1 predicate.
15139         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
15140         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
15141         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
15142         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
15143         (mov<mode>_insv_1): Remove expander.  Merge insn with
15144         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
15145         Use general_x64nomem_operand as operand 1 predicate.
15146         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
15147         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
15148         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
15149         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
15150         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
15151         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
15152         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
15153         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
15154         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
15155         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
15156         (general_x64nomem_operand): Ditto.
15158 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
15160         * config/rtems.opt: Add -pthread option.
15162 2013-03-26  Richard Biener  <rguenther@suse.de>
15164         * alias.c (find_base_term): Avoid redundant and not used recursion.
15165         (base_alias_check): Get the initial base term from the caller.
15166         (true_dependence_1): Compute and pass base terms to base_alias_check.
15167         (write_dependence_p): Likewise.
15168         (may_alias_p): Likewise.
15170 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
15172         * config/aarch64/aarch64.c (aarch64_classify_address): Support
15173         PC-relative load in SI modes and above only.
15175 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
15177         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
15178         * config/arm/iwmmxt.md (WCGR0): Update.
15179         (WCGR1, WCGR2, WCGR3): Likewise.
15181 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
15183         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
15184         Use x64 and nox64 isa attributes.
15186 2013-03-26  Richard Biener  <rguenther@suse.de>
15188         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
15189         alignment computations and rely on get_object_alignment_1
15190         for the !TYPE_P case.
15191         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
15193 2013-03-26  Walter Lee  <walt@tilera.com>
15195         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
15196         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
15198 2013-03-25  Jeff Law  <law@redhat.com>
15200         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
15201         check for INTEGRAL_TYPE_P that was missing due to checking in
15202         wrong version of prior patch.
15204 2013-03-25  Walter Lee  <walt@tilera.com>
15206         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
15207         TILEGX_INSN_SHUFFLEBYTES1.
15208         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
15209         shufflebytes1.
15210         (tilegx_builtins): Ditto.
15211         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
15213 2013-03-25  Walter Lee  <walt@tilera.com>
15215         * config/tilegx/tilegx.md (floatsisf2): New pattern.
15216         (floatunssisf2): New pattern.
15217         (floatsidf2): New pattern.
15218         (floatunssidf2): New pattern.
15220 2013-03-25  Walter Lee  <walt@tilera.com>
15222         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
15223         tests for constraint J, K, N, P.
15225 2013-03-25  Walter Lee  <walt@tilera.com>
15227         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
15228         Use indirect/pcrel encoding.
15229         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
15230         Ditto.
15232 2013-03-25  Steve Ellcey  <sellcey@mips.com>
15234         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
15235         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
15236         * config/mips/mips.c (mips_option_override): Set IMADD default.
15237         * config/mips/mips.h (PTF_AVOID_IMADD): New.
15238         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
15239         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
15240         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
15241         * doc/invoke.texi (-mimadd/-mno-imadd): New.
15243 2013-03-25  Jeff Law  <law@redhat.com>
15245         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
15246         slightly to avoid creating and folding useless trees.  Simplify
15247         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
15249 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
15251         * config/i386/i386.md (*zero_extendsidi2): Merge with
15252         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
15253         * config/i386/predicates.md (x86_64_zext_operand): Rename from
15254         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
15255         targets.  Clarify comment.
15257 2013-03-25  Martin Jambor  <mjambor@suse.cz>
15259         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
15260         pass-through jump functions differently.
15261         (ipa_read_jump_function): Likewise.  Also use setter functions to set
15262         up jump functions.
15264 2013-03-25  Martin Jambor  <mjambor@suse.cz>
15266         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
15267         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
15268         process it.
15269         (ipa_get_indirect_edge_target): New function.
15270         (devirtualization_time_bonus): New parameter known_aggs, pass it to
15271         ipa_get_indirect_edge_target.  Update all callers.
15272         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
15273         ipa_get_indirect_edge_target_1 instead of calling
15274         ipa_get_indirect_edge_target.
15275         (create_specialized_node): Pass aggvlas to
15276         ipcp_discover_new_direct_edges.
15278 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15280         * config/arm/arm.md (f_sels, f_seld): New types.
15281         (*cmov<mode>): New pattern.
15282         * config/arm/predicates.md (arm_vsel_comparison_operator): New
15283         predicate.
15285 2013-03-25  Kai Tietz  <ktietz@redhat.com>
15287         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
15288         POSIX-printf for mingw-hosted  builds.
15290 2013-03-25  Richard Biener  <rguenther@suse.de>
15292         PR middle-end/56694
15293         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
15294         must-not-throw stmt location.
15296 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15298         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
15299         Emit load-acquire versions when acq is true.
15300         (arm_emit_store_exclusive): Add rel parameter.
15301         Emit store-release versions when rel is true.
15302         (arm_split_compare_and_swap): Use acquire-release instructions
15303         instead.
15304         of barriers when appropriate.
15305         (arm_split_atomic_op): Likewise.
15306         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
15307         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
15308         (VUNSPEC_SLX): Likewise.
15309         (VUNSPEC_LDA): Likewise.
15310         (VUNSPEC_STL): Likewise.
15311         * config/arm/sync.md (atomic_load<mode>): New pattern.
15312         (atomic_store<mode>): Likewise.
15313         (arm_load_acquire_exclusive<mode>): Likewise.
15314         (arm_load_acquire_exclusivesi): Likewise.
15315         (arm_load_acquire_exclusivedi): Likewise.
15316         (arm_store_release_exclusive<mode>): Likewise.
15318 2013-03-25  Catherine Moore  <clm@codesourcery.com>
15320         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
15321         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
15322         * config/mip/predicates.md (lwsp_swsp_operand,
15323         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
15324         sb16_operand, db4_operand, db7_operand, ib3_operand,
15325         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
15326         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
15327         andi16_operand): New predicates.
15328         * config/mips/mips.md (compression): New attribute.
15329         (enabled): New attribute.
15330         (length): Consider compression in computing length.
15331         (shift_compression): New code attribute.
15332         (*add<mode>3): New operands. Record compression.
15333         (sub<mode>3): Likewise.
15334         (one_cmpl<mode>2): Likewise.
15335         (*and<mode>3): Likewise.
15336         (*ior<mode>3): Likewise.
15337         (unnamed pattern for xor): Likewise.
15338         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
15339         (*<optab><mode>3): Likewise.
15340         (*mov<mode>_internal: Likewise.
15341         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
15342         (mips_unsigned_immediate_p): New.
15343         (umips_lwsp_swsp_address_p): New.
15344         (m16_based_address_p): New.
15345         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
15346         (mips_unsigned_immediate_p): New prototype.
15347         (lwsp_swsp_address_p): New prototype.
15348         (m16_based_address_p): New prototype.
15349         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
15350         (mips_signed_immediate_p): New function.
15351         (m16_based_address_p): New function.
15352         (lwsp_swsp_address_p): New function.
15353         (mips_print_operand_punctuation): Recognize short delay slot insns
15354         for microMIPS.add<mode>3"
15356 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15358         PR target/56720
15359         * config/arm/iterators.md (v_cmp_result): New mode attribute.
15360         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
15362 2013-03-25  Richard Biener  <rguenther@suse.de>
15364         PR tree-optimization/56689
15365         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
15366         any edge.
15368 2013-03-25  Richard Biener  <rguenther@suse.de>
15370         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
15371         of bitmap.
15372         (memory_references): Likewise.
15373         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
15374         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
15375         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
15376         (gather_mem_refs_in_loops): Fold into ...
15377         (analyze_memory_references): ... this.  Move initialization
15378         to tree_ssa_lim_initialize.
15379         (fill_always_executed_in): Rename to ...
15380         (fill_always_executed_in_1): ... this.
15381         (fill_always_executed_in): Move contains_call computation to
15382         this new function from ...
15383         (tree_ssa_lim_initialize): ... here.
15384         (tree_ssa_lim): Call fill_always_executed_in.
15386 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
15388         * postreload.c (reload_combine): Fix code detecting returns.
15390 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
15392         * function.c (emit_use_return_register_into_block): On cc0 targets,
15393         do not emit the sequence between cc0 setter and user.
15395 2013-03-25  Kai Tietz  <ktietz@redhat.com>
15397         * config/i386/predicates.md (local_symbolic_operand): Interpret
15398         dll-imported symbols as none-local.
15400 2013-03-25  Richard Biener  <rguenther@suse.de>
15402         * tree-ssa-loop-im.c (struct depend): Remove.
15403         (struct lim_aux_data): Make depends a vec of gimples.
15404         (free_lim_aux_data): Adjust.
15405         (add_dependency): Likewise.
15406         (set_level): Likewise.
15408 2013-03-25  Richard Biener  <rguenther@suse.de>
15410         PR middle-end/56434
15411         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
15412         the pointer returned by calls with ECF_MALLOC set.
15414 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
15416         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
15418 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
15420         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
15421         using MMXMODE mode iterator.
15422         (*move<mode>_internal): Merge with *movv2sf_internal and
15423         *movv2sf_internal_rex64 using MMXMODE mode iterator.
15425 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
15427         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
15428         (record_last_mem_set_info): Likewise.
15430         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
15431         of XNEWVEC followed by memset.
15432         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
15434 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
15436         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
15437         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
15438         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
15439         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
15440         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
15441         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
15442         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
15443         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
15444         BARRIER_P instead of GET_CODE.
15446 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
15448         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
15449         inaccuracy in the probing code.
15451         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
15452         (ctrapdi4): Likewise.
15454 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
15456         * calls.c (expand_call): Add missing guard to code handling return
15457         of non-BLKmode structures in MSB.
15458         * function.c (expand_function_end): Likewise.
15460 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
15462         * combine.c (try_combine): Adjust comment.  Do not add the set of
15463         insn #0 if the destination indirectly is set or dies in insn #2.
15464         Tidy up code to distribute a new note.
15466 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
15468         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
15469         also for alternatives 16 and 17.
15471 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
15473         * config/i386/sse.md (*mov<mode>_internal): Merge with
15474         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
15475         Emit insn template depending on type attribute.  Use
15476         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
15477         movd instead of movq mnemonic for interunit moves.  Rewrite mode
15478         attribute calculation.  Remove unit attribute calculation.
15479         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
15480         Set prefix_data16 attribute for DImode ssemov types.
15481         Use Ym instead of y for SSE-MMX conversion alternatives.
15482         Reorder operand constraints.
15484 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
15486         * df.h (df_insn_delete): Adjust prototype.
15487         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
15488         and let it decide whether mark the basic block dirty.
15489         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
15490         * df-scan.c (df_insn_info_delete): New helper function, split
15491         off from df_insn_delete.
15492         (df_scan_free_bb_info): Use it.
15493         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
15494         Likewise.
15495         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
15496         that the insn is actually an insn and it has a non-NULL basic block.
15497         Do not mark basic block dirty if only deleting a DEBUG_INSN.
15499 2013-03-22  Richard Biener  <rguenther@suse.de>
15501         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
15502         dep_ref members.
15503         (mem_ref_alloc): Do not allocate them.
15504         (refs_independent_p): Do not query or maintain a cache.
15506 2013-03-22  Richard Biener  <rguenther@suse.de>
15508         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
15509         (gather_mem_refs_in_loops): Do not compute it.
15510         (analyze_memory_references): Do not allocate it.
15511         (tree_ssa_lim_finalize): Do not free it.
15512         (for_all_locs_in_loop): Do not query all_refs_in_loop.
15514 2013-03-22  Richard Biener  <rguenther@suse.de>
15516         * is-a.h (as_a): Use gcc_checking_assert.
15518 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
15520         * config/aarch64/aarch64.c (aarch64_print_operand): New
15521         format specifier for printing a constant in hex.
15522         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
15523         format specifier for printing second operand.
15525 2013-03-22  Richard Biener  <rguenther@suse.de>
15527         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
15528         bitmaps.
15529         (gather_mem_refs_in_loops): Perform store accumulation here.
15530         (create_vop_ref_mapping_loop): Remove.
15531         (create_vop_ref_mapping): Likewise.
15532         (analyze_memory_references): Initialize refs_stored_in_loop.
15533         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
15534         (record_indep_loop): Remove.
15535         (record_dep_loop): New function.
15536         (ref_indep_loop_p_1): Adjust to only walk over references
15537         in the loop, not its subloops.
15538         (ref_indep_loop_p): Rename to ...
15539         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
15540         maintaining a more fine-grained cache.
15541         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
15542         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
15544 2013-03-22  Richard Biener  <rguenther@suse.de>
15546         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
15547         (struct mem_ref): Make accesses_in_loop a vec of a vec of
15548         aggregate mem_ref_loc.
15549         (free_mem_ref_locs): Inline into ...
15550         (memref_free): ... this and adjust.
15551         (mem_ref_alloc): Adjust.
15552         (mem_ref_locs_alloc): Remove.
15553         (record_mem_ref_loc): Adjust.
15554         (get_all_locs_in_loop): Rewrite into ...
15555         (for_all_locs_in_loop): ... this iterator.
15556         (rewrite_mem_ref_loc): New functor.
15557         (rewrite_mem_refs): Use for_all_locs_in_loop.
15558         (sm_set_flag_if_changed): New functor.
15559         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
15560         (ref_always_accessed): New functor.
15561         (ref_always_accessed_p): Use for_all_locs_in_loop.
15563 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
15565         * tree-pass.h (PROP_gimple_lvec): New.
15566         * passes.c (dump_properties): Handle PROP_gimple_lvec.
15567         (init_optimization_passes): Move pass_lower_vector.
15568         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
15569         PROP_gimple_lvec.
15570         (pass_lower_vector): Provide PROP_gimple_lvec.
15571         (pass_lower_vector_ssa): Likewise.
15572         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
15574 2013-03-21  Mark Wielaard  <mjw@redhat.com>
15576         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
15578 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
15580         * config/i386/i386.md (*movdi_internal): Disparage slightly
15581         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
15582         conversion alternatives.
15584 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
15586         PR middle-end/48087
15587         * diagnostic.def (DK_WERROR): New kind.
15588         * diagnostic.h (werrorcount): Define.
15589         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
15590         promoted to DK_ERROR, increment DK_WERROR counter instead of
15591         DK_ERROR counter.
15592         * toplev.c (toplev_main): Call print_ignored_options even if
15593         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
15594         even if just werrorcount is non-zero.
15596         PR debug/55608
15597         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
15598         on failure.
15599         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
15600         (string_cst_pool_decl): New function.
15601         (optimize_one_addr_into_implicit_ptr): New function.
15602         (resolve_addr_in_expr): Optimize DWARF location expression
15603         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
15604         which doesn't live in memory, but has DW_AT_location or
15605         DW_AT_const_value, or refers to a string literal, into
15606         DW_OP_GNU_implicit_pointer.
15607         (optimize_location_into_implicit_ptr): New function.
15608         (resolve_addr): If removing DW_AT_location of a variable because
15609         it was DW_OP_addr of address of the variable, but the variable doesn't
15610         live in memory, try to emit const value attribute for the initializer.
15612 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
15614         * tree.h (VECTOR_TYPE_P): New macro.
15615         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
15616         TYPE_MODE): Use it.
15617         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
15618         VEC_COND_EXPR cannot be lvalues.
15619         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
15621 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
15623         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
15624         Restrict the transformation to equal modes.
15626 2013-03-21  Richard Biener  <rguenther@suse.de>
15628         PR tree-optimization/39326
15629         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
15630         (MEM_ANALYZABLE): Adjust.
15631         (record_mem_ref_loc): Move bitmap ops ...
15632         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
15633         unanalyzable refs, do not record locations for it.
15634         (analyze_memory_references): Allocate ref zero as shared
15635         unanalyzable ref.
15636         (refs_independent_p): Do not test for unanalyzed mems here.
15637         (ref_indep_loop_p_1): Special-case disambiguation against
15638         the unanalyzed ref.
15639         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
15641 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
15643         * config/arm/arm-protos.h (tune_params): Add
15644         prefer_neon_for_64bits field.
15645         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
15646         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
15647         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
15648         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
15649         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
15650         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
15651         (arm_option_override): Handle -mneon-for-64bits new option.
15652         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
15653         (prefer_neon_for_64bits): Declare new variable.
15654         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
15655         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
15656         (arch_enabled): Handle new arch types. Remove support for onlya8
15657         and nota8.
15658         (one_cmpldi2): Use new arch names.
15659         (zero_extend<mode>di2, extend<mode>di2): Ditto.
15660         * config/arm/arm.opt (mneon-for-64bits): Add option.
15661         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
15662         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
15663         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
15664         of onlya8.
15665         * doc/invoke.texi (-mneon-for-64bits): Document.
15667 2013-03-21  Richard Biener  <rguenther@suse.de>
15669         PR tree-optimization/39326
15670         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
15671         (sort_bbs_in_loop_postorder_cmp): New function.
15672         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
15674 2013-03-21  Richard Biener  <rguenther@suse.de>
15676         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
15677         (vect_insert_into_interleaving_chain): Likewise.
15678         (vect_drs_dependent_in_basic_block): Inline ...
15679         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
15680         split out from ...
15681         (vect_analyze_data_ref_dependence): ... here.  Simplify.
15682         (vect_check_interleaving): Simplify.
15683         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
15684         (vect_slp_analyze_data_ref_dependences): ... this new function.
15685         (dr_group_sort_cmp): New function.
15686         (vect_analyze_data_ref_accesses): Compute data-reference groups
15687         here instead of in vect_analyze_data_ref_dependence.  Use
15688         a more efficient algorithm.
15689         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
15690         vect_slp_analyze_data_ref_dependences.  Call
15691         vect_analyze_data_ref_accesses earlier.
15692         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
15693         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
15694         (vect_slp_analyze_data_ref_dependences): New prototype.
15696 2013-03-21  Richard Biener  <rguenther@suse.de>
15698         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
15699         ref is stored in the loop.
15700         (find_refs_for_sm): Walk only over all stores.
15701         (store_motion_loop): Allocate from lim_bitmap_obstack.
15702         (store_motion): Likewise.
15704 2013-03-21  Richard Biener  <rguenther@suse.de>
15706         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
15707         Update virtual SSA form.
15709 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
15711         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
15712         * configure: Regenerate.
15713         * config.in: Regenerate.
15714         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
15715         if !HAVE_LD_EH_FRAME_CIEV3.
15717 2013-03-21  Richard Biener  <rguenther@suse.de>
15719         * tree-cfg.c (verify_expr_no_block): New function.
15720         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
15721         nor DECL_VALUE_EXPR have locations with associated blocks.
15722         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
15723         (clear_unused_block_pointer): Remove code dealing with
15724         blocks in DECL_DEBUG_EXPR locations.
15726 2013-03-21  Richard Biener  <rguenther@suse.de>
15728         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
15729         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
15730         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
15731         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
15732         instead of DECL_DEBUG_EXPR_IS_FROM.
15733         * gimplify.c (gimplify_modify_expr): Likewise.
15734         * tree-cfg.c (verify_expr_location_1): Likewise.
15735         * tree-complex.c (create_one_component_var): Likewise.
15736         * tree-sra.c (create_access_replacement): Likewise.
15737         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
15738         (clear_unused_block_pointer): Likewise.
15739         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
15740         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
15741         * var-tracking.c (var_debug_decl): Likewise.
15742         (track_expr_p): Likewise.
15743         * tree-inline.c (add_local_variables): Likewise.  Set
15744         DECL_HAS_DEBUG_EXPR_P after copying it.
15745         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
15746         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
15748 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
15750         PR bootstrap/56656
15751         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
15752         * configure: Regenerate.
15753         * config.in: Regenerate.
15754         * config/i386/i386.md (*movdf_internal): Use
15755         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
15756         movd instead of movq mnemonic for interunit moves.
15757         (*movdi_internal): Ditto.
15759 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15761         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
15762         (abd<mode>_3): New pattern.
15763         (aba<mode>_3): New pattern.
15764         (fabd<mode>_3): New pattern.
15766 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15768         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
15769         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
15770         occurrence of REGISTER_PREFIX as its empty string.
15772 2013-03-20  Jeff Law  <law@redhat.com>
15774         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
15775         addititional equivalences for equality comparisons between an SSA_NAME
15776         and a constant where the SSA_NAME was set from a widening conversion.
15778 2013-03-20  Walter Lee  <walt@tilera.com>
15780         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
15782 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15784         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
15785         depending on type attribute.
15786         (*movti_internal): Ditto.
15787         (*movtf_internal): Ditto.
15788         (*movxf_internal): Ditto.
15789         (*movdf_internal): Ditto.
15790         (*movsf_internal): Ditto.
15792 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15794         * config/i386/i386.md (*movti_internal): Set prefix attribute to
15795         maybe_vex for sselog1 and ssemov types.
15796         (*movdi_internal): Reorder operand constraints.
15797         (*movsi_internal): Ditto.  Set prefix attribute to
15798         maybe_vex for sselog1 and ssemov types.
15799         (*movtf_internal): Set prefix attribute to maybe_vex
15800         for sselog1 and ssemov types.
15801         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
15802         DImode ssemov types.  Reorder operand constraints.
15803         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
15804         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
15805         attribute for SImode ssemov types.  Reorder operand constraints.
15807 2013-03-20  Martin Jambor  <mjambor@suse.cz>
15809         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
15810         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
15812 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
15814         * config/rs6000/predicates.md (indexed_address, update_address_mem
15815         update_indexed_address_mem): New predicates.
15816         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
15817         attribute for load/store instructions.
15818         * config/rs6000/dfp.md (movsd_store): Likewise.
15819         (movsd_load): Likewise.
15820         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
15821         (unnamed HI->DI extend define_insn): Likewise.
15822         (unnamed SI->DI extend define_insn): Likewise.
15823         (unnamed QI->SI extend define_insn): Likewise.
15824         (unnamed QI->HI extend define_insn): Likewise.
15825         (unnamed HI->SI extend define_insn): Likewise.
15826         (unnamed HI->SI extend define_insn): Likewise.
15827         (extendsfdf2_fpr): Likewise.
15828         (movsi_internal1): Likewise.
15829         (movsi_internal1_single): Likewise.
15830         (movhi_internal): Likewise.
15831         (movqi_internal): Likewise.
15832         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
15833         attribute for load/store instructions.
15834         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
15835         instructions.
15836         (mov<mode>_softfloat): Likewise.
15837         (mov<mode>_hardfloat32): Likewise.
15838         (mov<mode>_hardfloat64): Likewise.
15839         (mov<mode>_softfloat64): Likewise.
15840         (movdi_internal32): Likewise.
15841         (movdi_internal64): Likewise.
15842         (probe_stack_<mode>): Likewise.
15844 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
15846         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
15847         floating point, and decimal floating point to reload iterator.
15849         * config/rs6000/constraints.md (wl constraint): New constraints to
15850         return FLOAT_REGS if certain options are used to reduce the number
15851         of separate patterns that exist in the file.
15852         (wx constraint): Likewise.
15853         (wz constraint): Likewise.
15855         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15856         -mdebug=reg, print wg, wl, wx, and wz constraints.
15857         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15858         Initialize the reload functions for 64-bit binary/decimal floating
15859         point types.
15860         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15861         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15862         create the buffer on the stack to overcome not having a 32-bit
15863         load and store.
15864         (rs6000_emit_move): Likewise.
15865         (rs6000_secondary_memory_needed_rtx): Likewise.
15866         (rs6000_alloc_sdmode_stack_slot): Likewise.
15867         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15868         via xxlxor, just like DFmode 0.0.
15870         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
15871         define as 1 if we are running on a power7 or newer.
15872         (enum r6000_reg_class_enum): Add new constraints.
15874         * config/rs6000/dfp.md (movsd): Delete, combine with binary
15875         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15876         with other moves by using conditional constraits (wg).  Use LFIWZX
15877         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
15878         (movsd splitter): Likewise.
15879         (movsd_hardfloat): Likewise.
15880         (movsd_softfloat): Likewise.
15882         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
15883         binary and decimal floating point moves.
15884         (fmove_ok): New attributes to combine binary and decimal floating
15885         point moves, and to combine power6x (mfpgpr) moves along normal
15886         floating moves.
15887         (real_value_to_target): Likewise.
15888         (f32_lr): Likewise.
15889         (f32_lm): Likewise.
15890         (f32_li): Likewise.
15891         (f32_sr): Likewise.
15892         (f32_sm): Likewise.
15893         (f32_si): Likewise.
15894         (movsf): Combine binary and decimal floating point moves.  Combine
15895         power6x (mfpgpr) moves with other moves by using conditional
15896         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
15897         (mov<mode> for SFmode/SDmode); Likewise.
15898         (SFmode/SDmode splitters): Likewise.
15899         (movsf_hardfloat): Likewise.
15900         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
15901         (movsf_softfloat): Likewise.
15902         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
15904         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
15905         wx and wz constraints.
15907         * config/rs6000/constraints.md (wg constraint): New constraint to
15908         return FLOAT_REGS if -mmfpgpr (power6x) was used.
15910         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
15911         constraint.
15913         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15914         -mdebug=reg, print wg, wl, wx, and wz constraints.
15915         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15916         Initialize the reload functions for 64-bit binary/decimal floating
15917         point types.
15918         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15919         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15920         create the buffer on the stack to overcome not having a 32-bit
15921         load and store.
15922         (rs6000_emit_move): Likewise.
15923         (rs6000_secondary_memory_needed_rtx): Likewise.
15924         (rs6000_alloc_sdmode_stack_slot): Likewise.
15925         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15926         via xxlxor, just like DFmode 0.0.
15928         * config/rs6000/dfp.md (movdd): Delete, combine with binary
15929         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
15930         with other moves by using conditional constraits (wg).  Use LFIWZX
15931         and STFIWX for loading SDmode on power7.
15932         (movdd splitters): Likewise.
15933         (movdd_hardfloat32): Likewise.
15934         (movdd_softfloat32): Likewise.
15935         (movdd_hardfloat64_mfpgpr): Likewise.
15936         (movdd_hardfloat64): Likewise.
15937         (movdd_softfloat64): Likewise.
15939         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
15940         64-bit binary and decimal floating point moves.
15941         (FMOVE64X): Likewise.
15942         (movdf): Combine 64-bit binary and decimal floating point moves.
15943         Combine power6x (mfpgpr) moves with other moves by using
15944         conditional constraits (wg).
15945         (mov<mode> for DFmode/DDmode): Likewise.
15946         (DFmode/DDmode splitters): Likewise.
15947         (movdf_hardfloat32): Likewise.
15948         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15949         (movdf_softfloat32): Likewise.
15950         (movdf_hardfloat64_mfpgpr): Likewise.
15951         (movdf_hardfloat64): Likewise.
15952         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15953         (movdf_softfloat64): Likewise.
15954         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15955         (reload_<mode>_load): Move to later in the file so they aren't in
15956         the middle of the floating point move insns.
15957         (reload_<mode>_store): Likewise.
15959         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15960         constraint.
15962         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15963         constraint if -mdebug=reg.
15964         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15965         Enable using dd reload support if needed.
15967         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15968         binary and decimal floating point moves in rs6000.md.
15969         (movtd_internal): Likewise.
15971         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15972         decimal floating point moves.
15973         (movtf): Likewise.
15974         (movtf_internal): Likewise.
15975         (mov<mode>_internal, TDmode/TFmode): Likewise.
15976         (movtf_softfloat): Likewise.
15977         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15979         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15980         movdi_internal64, using wg constraint for move direct operations.
15981         (movdi_internal64): Likewise.
15983         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15984         MODES_TIEABLE_P for selected modes.  Print the numerical value of
15985         the various virtual registers. Use GPR/FPR first/last values,
15986         instead of hard coding the register numbers.  Print which modes
15987         have reload functions registered.
15988         (rs6000_option_override_internal): If -mdebug=reg, trace the options
15989         settings before/after setting cpu, target and subtarget settings.
15990         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15991         and for secondary reload failures in rs6000_secondary_reload_inner.
15992         (rs6000_secondary_reload_fail): Likewise.
15993         (rs6000_secondary_reload_inner): Likewise.
15995         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15996         macros for first/last GPR and FPR registers.
15997         (LAST_GPR_REGNO): Likewise.
15998         (FIRST_FPR_REGNO): Likewise.
15999         (LAST_FPR_REGNO): Likewise.
16001         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
16002         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
16003         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
16004         (vcond<mode><mode>): Likewise.
16005         (vcondu<mode><mode>): Likewise.
16006         (vector_gtu<mode>): Likewise.
16007         (vector_gte<mode>): Likewise.
16008         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
16009         to prevent the compiler from converting DImode operations to TImode.
16010         (ior<mode>3): Likewise.
16011         (and<mode>3): Likewise.
16012         (one_cmpl<mode>2): Likewise.
16013         (nor<mode>3): Likewise.
16014         (andc<mode>3): Likewise.
16016         * config/rs6000/constraints.md (wt constraint): New constraint
16017         that returns VSX_REGS if TImode is allowed in VSX registers.
16019         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
16020         constant under VSX.
16022         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
16023         similar to TImode, but it is restricted to being in the GPRs.
16025         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
16026         TImode to occupy a single VSX register.
16028         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
16029         -mvsx-timode for power7/power8.
16030         (power7 cpu): Likewise.
16031         (power8 cpu): Likewise.
16033         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
16034         sure that TFmode/TDmode take up two registers if they are ever
16035         allowed in the upper VSX registers.
16036         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
16037         registers.
16038         (rs6000_init_hard_regno_mode_ok): Likewise.
16039         (rs6000_debug_reg_global): Add debugging for PTImode and wt
16040         constraint.  Print if LRA is turned on.
16041         (rs6000_option_override_internal): Give an error if -mvsx-timode
16042         and VSX is not enabled.
16043         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
16044         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
16045         to reg+offset addressing.  Use PTImode when checking offset
16046         addresses for validity.
16047         (reg_offset_addressing_ok_p): Likewise.
16048         (rs6000_legitimate_offset_address_p): Likewise.
16049         (rs6000_legitimize_address): Likewise.
16050         (rs6000_legitimize_reload_address): Likewise.
16051         (rs6000_legitimate_address_p): Likewise.
16052         (rs6000_eliminate_indexed_memrefs): Likewise.
16053         (rs6000_emit_move): Likewise.
16054         (rs6000_secondary_reload): Likewise.
16055         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
16056         reloads to fpr registers to continue to use reg+offset addressing,
16057         but 64-bit reloads to altivec registers need reg+reg addressing.
16058         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
16059         it.  Treat LO_SUM like a PLUS operation.
16060         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
16061         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
16062         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
16063         registers to share a register with a smaller sized type, since VSX
16064         puts scalars in the upper 64-bits.
16065         (print_operand): Add support for PTImode.
16066         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
16067         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
16068         registers, but don't have arithmetic support.
16069         (rs6000_memory_move_cost): Add test for VSX.
16070         (rs6000_opt_masks): Add -mvsx-timode.
16072         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
16073         for TImode.
16074         (VSs): Likewise.
16075         (VSr): Use wt constraint for TImode.
16076         (VSv): Drop TImode support.
16077         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
16078         (vsx_movti_64bit): Likewise.
16079         (vsx_movti_32bit): Likewise.
16080         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
16081         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
16082         one '?' on the appropriate output constraint.  Do not allow TImode
16083         logical operations on 32-bit systems.
16084         (vsx_ior<mode>3): Likewise.
16085         (vsx_xor<mode>3): Likewise.
16086         (vsx_one_cmpl<mode>2): Likewise.
16087         (vsx_nor<mode>3): Likewise.
16088         (vsx_andc<mode>3): Likewise.
16089         (vsx_concat_<mode>): Likewise.
16090         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
16092         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
16093         OPTION_MASK_VSX_TIMODE.
16094         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
16095         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
16097         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
16098         (TI2 iterator): New iterator for TImode, PTImode.
16099         (wd mode attribute): Add values for vector types.
16100         (movti_string): Replace TI move operations with operations for TImode
16101         and PTImode.  Add support for TImode being allowed in VSX registers.
16102         (mov<mode>_string, TImode/PTImode): Likewise.
16103         (movti_ppc64): Likewise.
16104         (mov<mode>_ppc64, TImode/PTImode): Likewise.
16105         (TI mode splitters): Likewise.
16107         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
16108         constraint.
16110 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
16112         PR tree-optimization/56355
16113         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
16114         Also handle integers with undefined overflow.
16116 2013-03-20  Catherine Moore  <clm@codesourcery.com>
16117             Maciej W. Rozycki  <macro@codesourcery.com>
16118             Tom de Vries  <tom@codesourcery.com>
16119             Nathan Sidwell  <nathan@codesourcery.com>
16120             Iain Sandoe  <iain@codesourcery.com>
16121             Nathan Froyd  <froydnj@codesourcery.com>
16122             Chao-ying Fu  <fu@mips.com>
16124         * doc/extend.texi: (micromips, nomicromips, nocompression):
16125         Document new function attributes.
16126         * doc/invoke.texi (minterlink-compressed, mmicromips,
16127         m14k, m14ke, m14kec): Document new options.
16128         (minterlink-mips16): Update documentation.
16129         * doc/md.texi (ZC, ZD): Document new constraints.
16130         * configure.ac (gcc_cv_as_micromips): Check if linker
16131         supports the .set micromips directive.
16132         * configure: Regenerate.
16133         * config.in: Regenerate.
16134         * config/mips/mips-tables.opt: Regenerate.
16135         * config/mips/micromips.md: New file.
16136         * constraints.md (ZC, ZD): New constraints.
16137         * config/mips/predicates.md (movep_src_register): New predicate.
16138         (movep_src_operand): New predicate.
16139         (non_volatile_mem_operand): New predicate.
16140         * config/mips/mips.md (multimem): New type.
16141         (length): Differentiate between 17-bit and 18-bit branch offsets.
16142         (MOVEP1, MOVEP2): New mode iterator.
16143         (mov_<load>l): Use ZC constraint.
16144         (mov_<load>r): Likewise.
16145         (mov_<store>l): Likewise.
16146         (mov_<store>r): Likewise.
16147         (*branch_equality<mode>_inverted): Add microMIPS support.
16148         (*branch_equality<mode>): Likewise.
16149         (*jump_absolute): Likewise.
16150         (indirect_jump_<mode>): Likewise.
16151         (tablejump_<mode>): Likewise.
16152         (<optab>_internal): Likewise.
16153         (sibcall_internal): Likewise.
16154         (sibcall_value_internal): Likewise.
16155         (prefetch): Use constraint ZD.
16156         * config/mips/mips.opt (minterlink-compressed): New option.
16157         (minterlink-mips16): Now an alias for minterlink-compressed.
16158         (mmicromips): New option.
16159         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16160         (compare_and_swap_12): Likewise.
16161         (sync_add<mode>): Likewise.
16162         (sync_<optab>_12): Likewise.
16163         (sync_old_<optab>_12): Likewise.
16164         (sync_new_<optab>_12): Likewise.
16165         (sync_nand_12): Likewise.
16166         (sync_old_nand_12): Likewise.
16167         (sync_new_nand_12): Likewise.
16168         (sync_sub<mode>): Likewise.
16169         (sync_old_add<mode>): Likewise.
16170         (sync_old_sub<mode>): Likewise.
16171         (sync_new_add<mode>): Likewise.
16172         (sync_new_sub<mode>): Likewise.
16173         (sync_<optab><mode>): Likewise.
16174         (sync_old_<optab><mode>): Likewise.
16175         (sync_new_<optab><mode>): Likewise.
16176         (sync_nand<mode>): Likewise.
16177         (sync_old_nand<mode>): Likewise.
16178         (sync_new_nand<mode>): Likewise.
16179         (sync_lock_test_and_set<mode>): Likewise.
16180         (test_and_set_12): Likewise.
16181         (atomic_compare_and_swap<mode>): Likewise.
16182         (atomic_exchange<mode>_llsc): Likewise.
16183         (atomic_fetch_add<mode>_llsc): Likewise.
16184         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16185         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16186         (umips_save_restore_pattern_p): Likewise.
16187         (umips_load_store_pair_p): Likewise.
16188         (umips_output_load_store_pair): Likewise.
16189         (umips_movep_target_p): Likewise.
16190         (umips_12bit_offset_address_p): Likewise.
16191         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16192         (mips_base_mips16): Rename this...
16193         (mips_base_compression_flags): ...to this. Update all uses.
16194         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16195         (mips_mips16_decl_p): Delete.
16196         (mips_nomips16_decl_p): Delete.
16197         (mips_get_compress_on_flags): New function.
16198         (mips_get_compress_off_flags): New function.
16199         (mips_get_compress_mode): New function.
16200         (mips_get_compress_on_name): New function.
16201         (mips_get_compress_off_name): New function.
16202         (mips_insert_attributes): Support multiple compression types.
16203         (mips_merge_decl_attributes): Likewise.
16204         (umips_12bit_offset_address_p): New function.
16205         (mips_start_function_definition): Emit .set micromips directive.
16206         (mips_call_may_need_jalx_p): New function.
16207         (mips_function_ok_for_sibcall): Add microMIPS support.
16208         (mips_print_operand_punctuation): Support short delay slots and
16209         compact jumps.
16210         (umips_swm_mask, umips_swm_encoding): New.
16211         (umips_build_save_restore): New function.
16212         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16213         (was_mips16_p): Remove.
16214         (old_compression_mode): New.
16215         (mips_set_compression_mode): New function.
16216         (mips_set_current_function): Add microMIPS support.
16217         (mips_option_override): Likewise.
16218         (umips_save_restore_pattern_p): New function.
16219         (umips_output_save_restore): New function.
16220         (umips_load_store_pair_p_1): New function.
16221         (umips_load_store_pair_p): New function.
16222         (umips_output_load_store_pair_1): New function.
16223         (umips_output_load_store_pair): New function.
16224         (umips_movep_target_p) New function.
16225         (mips_prepare_pch_save): Add microMIPS support.
16226         * config/mips/mips.h (TARGET_COMPRESSION): New.
16227         (TARGET_CPU_CPP_BUILTINS): Update macro
16228         to use new compression flags and to support microMIPS.
16229         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16230         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16231         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16232         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16233         (ASM_SPEC): Support mmicromips and mno-micromips.
16234         (M16STORE_REG_P): New macro.
16235         (MIPS_CALL): Support TARGET_MICROMIPS.
16236         (MICROMIPS_J): New macro.
16237         (mips_base_mips16): Rename this...
16238         (mips_base_compression_flags): ...to this.
16239         (UMIPS_12BIT_OFFSET_P): New macro.
16240         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16241         (MULTILIB_DIRNAMES): Likewise.
16242 2013-03-20  Richard Biener  <rguenther@suse.de>
16244         PR tree-optimization/56661
16245         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
16246         the result does not have to be distinct.
16248 2013-03-20  Richard Biener  <rguenther@suse.de>
16250         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
16251         remap_gimple_op_r.
16253 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16254             Steven Bosscher  <steven@gcc.gnu.org>
16256         PR rtl-optimization/56605
16257         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
16259 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
16261         PR bootstrap/56656
16262         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
16263         that require movd instead of movq.
16265 2013-03-20  Richard Biener  <rguenther@suse.de>
16267         * tree-ssa-structalias.c (struct variable_info): Add pointer
16268         to the first field of an aggregate with sub-vars.  Make
16269         this and the pointer to the next subfield its ID.
16270         (vi_next): New function.
16271         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
16272         storedanything_id, integer_id): Increment by one.
16273         (new_var_info, get_call_vi, lookup_call_clobber_vi,
16274         get_call_clobber_vi): Adjust.
16275         (solution_set_expand): Simplify and speedup.
16276         (solution_set_add): Inline into ...
16277         (set_union_with_increment): ... this.  Adjust accordingly.
16278         (do_sd_constraint): Likewise.
16279         (do_ds_constraint): Likewise.
16280         (do_complex_constraint): Simplify.
16281         (build_pred_graph): Adjust.
16282         (solve_graph): Likewise.  Simplify and speedup.
16283         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
16284         get_constraint_for_component_ref, get_constraint_for_1,
16285         first_vi_for_offset, first_or_preceding_vi_for_offset,
16286         create_function_info_for, create_variable_info_for_1,
16287         create_variable_info_for, intra_create_variable_infos): Adjust.
16288         (init_base_vars): Push NULL for ID zero.
16289         (compute_points_to_sets): Adjust.
16291 2013-03-20  Richard Biener  <rguenther@suse.de>
16293         * cfgloop.c (verify_loop_structure): Streamline and avoid
16294         ICEing on corrupt loop tree.
16295         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
16296         loop tree.
16298 2013-03-20  Richard Biener  <rguenther@suse.de>
16300         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
16301         check whether an SSA update is needed.
16303 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
16305         * config/mips/constraints.md (T): Rename to...
16306         (Yf): ...this.
16307         (U): Rename to...
16308         (Yd): ...this.
16309         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
16310         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
16312 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
16314         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
16315         (*subsi3_carryin_uxtw): Likewise.
16317 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
16319         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
16320         (*rorsi3_insn_uxtw): Likewise.
16322 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
16324         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
16325         (*extrsi5_insn_uxtw): Likewise.
16327 2013-03-19  Richard Biener  <rguenther@suse.de>
16329         PR tree-optimization/56273
16330         * passes.c (init_optimization_passes): Move second VRP after DOM.
16332 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
16334         * config/i386/i386.md (*movti_internal): Merge from
16335         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
16336         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
16337         nox64 isa attributes.
16339 2013-03-18  Richard Biener  <rguenther@suse.de>
16341         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
16342         (unite): Likewise.
16343         (merge_node_constraints): Likewise.
16344         (build_succ_graph): Likewise.
16345         (valid_graph_edge): Inline into single caller.
16346         (unify_nodes): Likewise.  Use bitmap_set_bit return value
16347         and cache varinfo.
16348         (scc_visit): Fix formatting and variable use.
16349         (do_sd_constraint): Use gcc_checking_assert.
16350         (do_ds_constraint): Likewise.
16351         (do_complex_constraint): Likewise.
16352         (condense_visit): Likewise.  Cleanup.
16353         (dump_pred_graph): New function.
16354         (perform_var_substitution): Dump the pred-graph before
16355         variable substitution.
16356         (find_equivalent_node): Use gcc_checking_assert.
16357         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
16359 2013-03-18  Richard Biener  <rguenther@suse.de>
16361         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
16362         Remove cond_expr_stmt_list argument and do not gimplify the
16363         built expression.
16364         (vect_loop_versioning): Adjust.
16365         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
16366         Cleanup to use less temporaries.
16367         (vect_create_data_ref_ptr): Cleanup.
16369 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
16371         PR tree-optimization/56635
16372         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
16373         require types_compatible_p types.
16375 2013-03-18  Nick Clifton  <nickc@redhat.com>
16377         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
16378         spurious backslash.
16380         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
16381         Add missing line to comment describing function.
16383 2013-03-18  Richard Biener  <rguenther@suse.de>
16385         PR tree-optimization/56210
16386         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
16387         Handle string / character search functions.
16388         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
16390 2013-03-18  Richard Biener  <rguenther@suse.de>
16392         PR middle-end/56483
16393         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
16394         and implement properly.
16395         * gimple.h (gimple_cond_single_var_p): Remove.
16397 2013-03-18  Richard Biener  <rguenther@suse.de>
16399         * tree-data-ref.h (find_data_references_in_loop): Declare.
16400         * tree-data-ref.c (get_references_in_stmt): Use a stack
16401         vector pre-allocated in the callers.
16402         (find_data_references_in_stmt): Adjust.
16403         (graphite_find_data_references_in_stmt): Likewise.
16404         (create_rdg_vertices): Likewise.
16405         (find_data_references_in_loop): Export.
16406         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
16407         Compute dependences here...
16408         (vect_analyze_data_refs): ...not here.  When we encounter
16409         a non-vectorizable data reference in basic-block vectorization
16410         truncate the data reference vector.  Do not bother to
16411         fixup data-dependence information for gather loads.
16412         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
16413         of data references, as reported.
16415 2013-03-18  Richard Biener  <rguenther@suse.de>
16417         PR tree-optimization/3713
16418         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
16419         has_constants and expr.
16420         (stmt_has_constants): Properly valueize SSA names when deciding
16421         whether the stmt has constants.
16423 2013-03-18  Richard Biener  <rguenther@suse.de>
16425         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
16426         whole function when there is nothing to do.
16427         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
16428         * tree-vectorizer.c (vectorize_loops): Update virtual and
16429         loop-closed SSA once.
16430         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
16432 2013-03-18  Richard Biener  <rguenther@suse.de>
16434         PR middle-end/56113
16435         * domwalk.c (bb_postorder): New global static.
16436         (cmp_bb_postorder): New function.
16437         (walk_dominator_tree): Replace scheme imposing an order for
16438         visiting dominator sons by one sorting them at the time they
16439         are pushed on the stack.
16441 2013-03-18  Richard Biener  <rguenther@suse.de>
16443         PR tree-optimization/39326
16444         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
16445         (struct mem_ref): Replace mem member with ao_ref typed member.
16446         (MEM_ANALYZABLE): Adjust.
16447         (memref_eq): Likewise.
16448         (mem_ref_alloc): Likewise.
16449         (gather_mem_refs_stmt): Likewise.
16450         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
16451         (execute_sm_if_changed_flag_set): Adjust.
16452         (execute_sm): Likewise.
16453         (ref_always_accessed_p): Likewise.
16454         (refs_independent_p): Likewise.
16455         (can_sm_ref_p): Likewise.
16457 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
16459         PR c/56566
16460         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
16461         return 1 even for !unsignedp.
16463 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
16465         * config/i386/i386.md (isa): Add x64 and nox64.
16466         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
16467         (*pushtf): Enable *roF alternative for x64 isa only.
16468         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
16469         mode attribute of integer alternatives to DImode for TARGET_64BIT.
16470         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
16471         (*movtf_internal): Merge from *movtf_internal_rex64 and
16472         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
16473         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
16474         nox64 isa attributes.
16475         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
16476         nox64 isa attributes.
16477         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
16479 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
16481         * config/alpha/alpha.c (TARGET_LRA_P): New define.
16483 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
16485         PR target/56640
16486         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
16487         class names.  Remove trailing comma after "ALL_REGS".
16489 2013-03-16  Jan Hubicka  <jh@suse.cz>
16491         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
16492         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
16493         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
16494         of cgraph_get_create_node.
16495         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16497 2013-03-16  Jason Merrill  <jason@redhat.com>
16499         PR debug/49090
16500         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
16501         with DW_AT_default_value.
16503 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
16505         * BASE-VER: Set to 4.9.0.
16507 2013-03-14  Andi Kleen  <ak@linux.intel.com>
16509         PR target/56619
16510         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
16511         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
16512         Document _x* TSX intrinsics.
16514 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
16515             David Holsgrove  <david.holsgrove@xilinx.com>
16517         * configure.ac: Add MicroBlaze TLS support detection.
16518         * configure: Regenerate.
16519         * config/microblaze/microblaze-protos.h
16520         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
16521         symbol_mentioned_p, label_mentioned_p): Add prototypes.
16522         * config/microblaze/microblaze.c (microblaze_address_type): Add
16523         ADDRESS_TLS and tls_reloc address types.
16524         (microblaze_address_info): Add tls_reloc.
16525         (TARGET_HAVE_TLS): Define.
16526         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
16527          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
16528          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
16529         load_tls_operand,  microblaze_call_tls_get_addr,
16530         microblaze_legitimize_tls_address): New functions.
16531         (microblaze_classify_unspec): Handle UNSPEC_TLS.
16532         (get_base_reg): Use microblaze_tls_symbol_p.
16533         (microblaze_classify_address): Handle TLS.
16534         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
16535         label_mentioned_p and microblaze_tls_referenced_p.
16536         (microblaze_legitimize_address): Handle TLS.
16537         (microblaze_address_insns): Handle ADDRESS_TLS.
16538         (pic_address_needs_scratch): Handle TLS.
16539         (print_operand_address): Handle TLS.
16540         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
16541         (microblaze_expand_move): Handle TLS.
16542         (microblaze_legitimate_constant_p): Check
16543         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
16544         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16545         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
16546         (PIC_OFFSET_TABLE_REGNUM): Set.
16547         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
16548         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
16549         (addsi3, movsi_internal2, movdf_internal): Update constraints
16550         * config/microblaze/predicates.md (arith_plus_operand): Define
16551         (move_operand): Redefine as move_src_operand,
16552         check microblaze_tls_referenced_p.
16554 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
16556         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
16557         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
16559 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
16561         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
16562         CC mode for AND.
16564 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
16566         PR tree-optimization/53265
16567         * common.opt (Waggressive-loop-optimizations): New option.
16568         * tree-ssa-loop-niter.c: Include tree-pass.h.
16569         (do_warn_aggressive_loop_optimizations): New function.
16570         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
16571         if number_of_latch_executions returned constant.
16572         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
16573         early.  If number_of_latch_executions returned constant, set
16574         nb_iterations_upper_bound back to it.
16575         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
16576         field.
16577         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
16578         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
16580         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
16581         (MULTILIB_OSDIRNAMES): Set.
16582         * genmultilib: If defaultosdirname doesn't start with :: , set
16583         defaultosdirname2 instead, clear it and emit two . multilib_raw
16584         entries instead of just one.
16586 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
16588         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
16589         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
16590         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
16591         (SUBTARGET_OVERRIDE_OPTIONS): New.
16593 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
16595         PR target/49880
16596         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
16597         (musermode): Convert to Var(TARGET_USERMODE).
16598         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
16599         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
16600         * config/sh/sh.c (sh_option_override): Use
16601         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
16602         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
16603         condition.
16604         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
16605         TARGET_SH4.
16606         (udivsi3_i4_single, divsi3_i4_single): Use
16607         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
16609 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
16611         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
16612         default setting.
16614 2013-03-13  Richard Biener  <rguenther@suse.de>
16616         PR tree-optimization/56608
16617         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
16618         calls when vectorizing basic-blocks.
16620 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
16622         PR plugins/45078
16623         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
16624         tm_file.
16626 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
16628         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
16630 2013-03-11  Jan Hubicka  <jh@suse.cz>
16632         PR lto/56557
16633         * lto-streamer-out.c (output_symbol_p): Skip references from
16634         constructors of external variables.
16636 2013-03-11  Jan Hubicka  <jh@suse.cz>
16638         PR middle-end/56571
16639         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
16640         from pseudos.
16641         * emit-rtl.c (verify_rtx_sharing): Likewise.
16642         (copy_insn_1): Likewise.
16643         * rtl.c (copy_rtx): Likewise.
16645 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
16647         PR target/56591
16648         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
16649         output_operand_lossage message.
16651 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
16653         PR target/56470
16654         * arm.c (shift_op): Validate RTL pattern on the fly.
16655         (arm_print_operand, case 'S'): Don't use shift_operator to validate
16656         the RTL.
16658 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16660         PR target/56347
16661         * config/pa/pa.md (call_value): Check for calls to powf and direct to
16662         new call patterns that clobber %fr12.
16663         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
16664         split and postreload patterns.
16665         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
16666         registers %fr12 and %fr12R as call used.
16668 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
16670         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
16671         (canon_address, record_store, replace_read, check_mem_read_rtx,
16672         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
16673         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
16674         rest_of_handle_dse): Likewise.
16676 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
16678         PR middle-end/56524
16679         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
16680         Add base_optabs.
16681         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
16682         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
16683         (save_optabs_if_changed): Replace with...
16684         (init_tree_optimization_optabs): ...this.
16685         * optabs.c (save_optabs_if_changed): Rename to...
16686         (init_tree_optimization_optabs): ...this.  Take the optimization node
16687         as argument.  Do nothing if the base optabs are already correct.
16688         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
16689         to recompute optabs.
16690         * function.h (function): Remove optabs field.
16691         * function.c (invoke_set_current_function_hook): Call
16692         init_tree_optimization_optabs.  Use the result to initialize
16693         this_fn_optabs.
16695 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
16697         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
16698         if GTMA_HAS_NO_INSTRUMENTATION.
16699         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
16700         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
16701         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
16702         * gimple-pretty-print.c (dump_gimple_transaction): Handle
16703         GTMA_HAS_NO_INSTRUMENTATION.
16705 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
16707         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
16708         libasan_preinit.o.
16710 2013-03-08  Marek Polacek  <polacek@redhat.com>
16711             Jakub Jelinek  <jakub@redhat.com>
16713         PR tree-optimization/56478
16714         * predict.c (is_comparison_with_loop_invariant_p): Change the
16715         type of loop_step to tree.
16716         (predict_loops): Adjust.
16717         (predict_iv_comparison): Perform the computations on double_ints.
16719 2013-03-08  Richard Biener  <rguenther@suse.de>
16721         PR tree-optimization/56570
16722         * tree-cfg.c (verify_expr_location_1): Verify locations for
16723         DECL_DEBUG_EXPR.
16724         * tree-sra.c (create_access_replacement): Strip locations
16725         from DECL_DEBUG_EXPRs.
16727 2013-03-08  Richard Biener  <rguenther@suse.de>
16729         * tree-inline.c (expand_call_inline): Do not associate
16730         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
16731         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
16733 2013-03-08  Richard Biener  <rguenther@suse.de>
16735         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
16736         or block changes with -Og.  Fix for location / block encoding
16737         changes and PHI arguments with locations.
16739 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
16741         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
16742         for all counters.
16743         (struct output_info): Likewise.
16744         (register_overhead): Remove bad gcc_assert.
16745         (bitmap_find_bit): If there is only a single bitmap element, do not
16746         count a miss as a search.
16747         (print_statistics): Update for counter type changes.
16748         (dump_bitmap_statistics): Likewise.  Print headers such that they
16749         are properly lined up with the printed counters.
16751 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
16753         PR tree-optimization/56559
16754         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
16755         check that it has only a single use.
16757 2013-03-07  Richard Biener  <rguenther@suse.de>
16759         * doc/invoke.texi (fwhole-program): Discourage use in combination
16760         with -flto.
16762 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
16764         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
16766         PR tree-optimization/56539
16767         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
16768         instead of GSI_CONTINUE_LINKING as last argument to
16769         force_gimple_operand_gsi.  Adjust function comment.
16771         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
16772         aarch64-cores.def.
16774         PR middle-end/56548
16775         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
16776         promoted mode, convert the result back to the original mode.
16778 2013-03-06  Richard Biener  <rguenther@suse.de>
16780         PR middle-end/56294
16781         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
16782         (insert_updated_phi_nodes_compare_uids): New function.
16783         (update_ssa): Sort symbols_to_rename after UID before
16784         traversing it to insert PHI nodes.
16786 2013-03-06  Richard Biener  <rguenther@suse.de>
16788         PR middle-end/50494
16789         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
16790         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
16792         Revert
16793         2013-02-13  Richard Biener  <rguenther@suse.de>
16795         PR lto/50494
16796         * varasm.c (output_constant_def_1): Get the decl representing
16797         the constant as argument.
16798         (output_constant_def): Wrap output_constant_def_1.
16799         (make_decl_rtl): Use output_constant_def_1 with the decl
16800         representing the constant.
16801         (build_constant_desc): Optionally re-use a decl already
16802         representing the constant.
16803         (tree_output_constant_def): Adjust.
16805 2013-03-06  Joey Ye  <joey.ye@arm.com>
16807         PR lto/50293
16808         * gcc.c (convert_white_space): New function.
16809         (main): Handles white space in function name.
16811 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
16813         PR target/56529
16814         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
16815         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
16816         to SH_DIV_CALL_TABLE for TARGET_SH2.
16817         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
16818         list.
16819         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
16820         call-table options.
16822 2013-03-05  Sterling Augustine  <saugustine@google.com>
16823             Cary Coutant  <ccoutant@google.com>
16825         PR debug/55364
16826         * dwarf2out.c (resolve_addr): Don't call
16827         remove_loc_list_addr_table_entries a second time for the same
16828         expression.
16830 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16832         PR debug/56510
16833         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
16834         (avoid_complex_debug_insns): New function.
16835         (expand_debug_locations): Call it.
16837         PR rtl-optimization/56484
16838         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
16839         lifetimes of hard registers on small register class machines.
16841 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16843         * config/microblaze/microblaze-protos.h: Rename
16844         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
16845         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
16846         fast_interrupt.
16847         (microblaze_fast_interrupt_function_p): New function.
16848         (microblaze_is_interrupt_handler): Rename to
16849         microblaze_is_interrupt_variant and add fast_interrupt check.
16850         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
16851         (save_restore_insns): Likewise.
16852         (compute_frame_size): Likewise.
16853         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
16854         (microblaze_globalize_label): Likewise.
16855         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
16856         * config/microblaze/microblaze.md: Use wrapper
16857         microblaze_is_interrupt_variant.
16859 2013-03-05  Kai Tietz  <ktietz@redhat.com>
16861         * sdbout.c (sdbout_one_type): Switch to current function's section
16862         supporting cold/hot.
16864 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
16866         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
16867         -mxl-reorder.
16869 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16871         PR middle-end/56461
16872         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
16873         if VALGRIND_GET_VBITS is defined, temporarily make object
16874         memory all defined, and restore previous valgrind addressability
16875         and definability afterwards.  Free this_object at the end.
16877         PR middle-end/56461
16878         * lra.c (lra): Call lra_clear_live_ranges if live_p,
16879         right before calling lra_create_live_ranges, also call it
16880         when clearing live_p.  Only call lra_clear_live_ranges
16881         at the end if live_p.
16883         PR middle-end/56461
16884         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
16886 2013-03-05  Richard Biener  <rguenther@suse.de>
16888         PR tree-optimization/56521
16889         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
16890         value-id.
16892 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
16894         PR c++/55135
16895         * except.h (remove_unreachable_eh_regions): New prototype.
16896         * except.c (remove_eh_handler_splicer): New function, split out
16897         of remove_eh_handler.
16898         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
16899         warning about running it on many EH regions one at a time.
16900         (remove_unreachable_eh_regions_worker): New function, walk the
16901         EH tree in depth-first order and remove non-marked regions.
16902         (remove_unreachable_eh_regions): New function.
16903         * tree-eh.c (mark_reachable_handlers): New function, split out
16904         from remove_unreachable_handlers.
16905         (remove_unreachable_handlers): Use mark_reachable_handlers and
16906         remove_unreachable_eh_regions.
16907         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
16908         and remove_unreachable_eh_regions.
16910 2013-03-05  Richard Biener  <rguenther@suse.de>
16912         PR middle-end/56525
16913         * loop-init.c (fix_loop_structure): Remove loops in two stages,
16914         not freeing them until the end.
16916 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16918         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
16920 2013-03-05  Richard Biener  <rguenther@suse.de>
16922         PR tree-optimization/56270
16923         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
16924         of loads after scheduling an SLP instance.
16926 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16928         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
16929         tic6x.exp.
16930         (check_gcc_parallelize): Run guality.exp as a separate job from
16931         vect.exp with unsorted.exp and $(dg_target_exps) separately from
16932         struct-layout-1.exp with stackalign.exp.
16934         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
16936         PR middle-end/56461
16937         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
16938         load_index sbitmap even if some bit in it isn't set.
16940         PR middle-end/56461
16941         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16942         (discover_iteration_bound_by_body_walk): Change queues to
16943         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
16944         spelling in comment.  Call safe_push on queues[bound_index] directly.
16945         Release queues[queue_index] in every iteration unconditionally.
16946         Release bounds vector.
16948         PR middle-end/56461
16949         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16950         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16951         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16952         inner_phis vector.
16954 2013-03-05  Richard Biener  <rguenther@suse.de>
16956         PR lto/56515
16957         * tree-inline.c (remap_blocks_to_null): New function.
16958         (expand_call_inline): When expanding a call stmt without
16959         an associated block inline remap all callee blocks to NULL.
16961 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
16963         PR rtl-optimization/56494
16964         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16965         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16966         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16968         PR middle-end/56461
16969         * sel-sched-ir.c (free_sched_pools): Release
16970         succs_info_pool.stack[succs_info_pool.max_top] vectors too
16971         if succs_info_pool.max_top isn't -1.
16973         PR bootstrap/56509
16974         * opts.c (opts_obstack, opts_concat): Moved to...
16975         * opts-common.c (opts_obstack, opts_concat): ... here.
16977 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
16979         PR middle-end/56461
16980         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16982 2013-03-04  Martin Jambor  <mjambor@suse.cz>
16984         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16985         all appropriate places.
16987 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
16989         PR tree-optimization/56424
16990         * ipa-split.c (split_function): Do not set the RSO flag if result is
16991         not by reference and its type is a register type.
16993 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
16995         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16996         (microblaze_legitimate_pic_operand): Likewise
16997         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16998         new function microblaze_legitimate_pic_operand
16999         * config/microblaze/microblaze-protos.h
17000         (microblaze_legitimate_pic_operand): Declare.
17002 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17004         * config/microblaze/predicates.md (call_insn_simple_operand):
17005         New predicate for supported rtx code types.
17006         * config/microblaze/microblaze.md (call_internal1): Use
17007         call_insn_simple_operand predicate.
17009 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
17011         PR middle-end/56461
17012         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
17013         partitions.ordered_remove.
17015         PR middle-end/56461
17016         * tree-vect-stmts.c (vectorizable_conversion): Don't call
17017         vec_oprnds0.create (1) for modifier == NONE.
17019         PR middle-end/56461
17020         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
17021         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
17022         vec_oprnds1 right before pushing anything to it for
17023         scalar_shift_arg.
17025         PR middle-end/56461
17026         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
17027         set nbbs to 0 instead of having separate code path.
17028         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
17029         instead of false as last argument if returning NULL.
17031 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
17033         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
17034         the attribute is now called "target" instead of "option".
17035         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
17036         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
17037         attribute/pragma name for TARGET_OPTION_VALID_P and
17038         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
17039         * doc/tm.texi: Regenerated.
17041 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
17043         * config/microblaze/microblaze.c:
17044         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
17045         * config/microblaze/microblaze.h: Add -mxl-reorder to
17046         DRIVER_SELF_SPECS.
17047         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
17048         instructions emitted if TARGET_REORDER.
17049         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
17050         or 0 for -m/-mno case, but initialises as 2 to detect default use case
17051         separately.
17053 2013-03-01  Xinliang David Li  <davidxl@google.com>
17055         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
17056         walk length.
17058 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
17060         PR middle-end/56461
17061         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
17062         vector even when returning true.  Fix up function comment formatting.
17064         PR middle-end/56461
17065         * ira-build.c (ira_loop_nodes_count): New variable.
17066         (create_loop_tree_nodes): Initialize it.
17067         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
17069         PR middle-end/56461
17070         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
17071         method on dr_chain and result_chain.
17072         * tree-vect-stmts.c (vectorizable_store): Only call
17073         result_chain.create if j == 0.
17075         PR middle-end/56461
17076         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
17077         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
17078         before overwriting it.
17080 2013-03-01  Tobias Burnus  <burnus@net-b.de>
17082         * doc/extended.texi (C Extensions): Change order in @menu
17083         to match @node.
17084         (Other MIPS Built-in Functions): Move last MIPS entry before
17085         "picoChip Built-in Functions".
17086         (SH Built-in Functions): Move after RX Built-in Functions.
17087         * doc/gcc.texi (Introduction): Change order in @menu
17088         to match @node.
17089         * doc/md.texi (Constraints): Ditto.
17090         * gty.texi (Type Information): Ditto.
17091         (User-provided marking routines for template types): Make
17092         subsection.
17093         * doc/invoke.texi (AArch64 Options): Move before
17094         "Adapteva Epiphany Options".
17096 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
17097             Jakub Jelinek  <jakub@redhat.com>
17099         PR sanitizer/56454
17100         * asan.c (gate_asan): Lookup no_sanitize_address instead of
17101         no_address_safety_analysis attribute.
17102         * doc/extend.texi (no_address_safety_attribute): Rename to
17103         no_sanitize_address attribute, mention no_address_safety_analysis
17104         attribute as deprecated alias.
17106 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
17108         PR middle-end/56461
17109         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
17110         type to vec<vec<tree> > *.
17111         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
17112         to be vec<tree> instead of vec<tree> *, set vec_defs
17113         to vNULL and call vec_defs.create (number_of_vects), adjust other
17114         uses of vec_defs.
17115         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
17116         vectorizable_condition): Adjust vect_get_slp_defs callers.
17118 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
17120         * config/aarch64/aarch64.c
17121         (aarch64_float_const_representable): Remove unused variable.
17123 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
17125         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
17127 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
17129         * config/aarch64/aarch64-builtins.c
17130         (aarch64_init_simd_builtins): Make static.
17132 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
17134         * config/aarch64/aarch64.c
17135         (aarch64_simd_make_constant): Make static.
17137 2013-02-28  Martin Jambor  <mjambor@suse.cz>
17139         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
17140         with no initialization to the RHS of debug statements.
17142 2013-02-28  Martin Jambor  <mjambor@suse.cz>
17144         PR tree-optimization/56294
17145         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
17146         Adjust dumping.
17147         (get_access_replacement): Do not call create_access_replacement.
17148         Assert a replacement exists.
17149         (get_repl_default_def_ssa_name): Create the replacement declaration
17150         itself.
17152 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17154         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
17155         final_end_function.
17157 2013-02-28  Marek Polacek  <polacek@redhat.com>
17159         PR rtl-optimization/56466
17160         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
17161         if we're changing a loop.
17162         (peel_loops_completely): Likewise.
17164 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
17166         PR c++/55813
17167         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
17169 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
17171         PR target/56445
17172         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
17173         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
17174         INTX_FTYPE_FX, FX_FTYPE_INTX.
17175         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
17177 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
17179         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
17180         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
17181         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
17182         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
17183         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
17184         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
17185         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
17186         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
17187         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
17188         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
17189         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
17190         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
17191         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
17192         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
17193         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
17194         (avrxmega6): Increase max flash segments from 5 to 6.
17195         * config/avr/t-multilib: Regenerate.
17196         * config/avr/avr-tables.opt: Regenerate.
17197         * doc/avr-mmcu.texi: Regenerate.
17199 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
17201         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
17202         (avr_device_to_arch): Rename to avr_device_to_ld.
17203         (avr_device_to_as): New prototype.
17204         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
17205         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
17206         * config/avr/driver-avr.c (avr_device_to_as): New.
17207         (avr_device_to_arch): Rename to avr_device_to_ld.
17209 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
17211         PR middle-end/56461
17212         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
17213         method on dr_chain and result_chain.
17215         PR middle-end/56461
17216         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
17217         pointer_set_destroy on not_executed_last_iteration.
17219         PR middle-end/56461
17220         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
17222         PR middle-end/56461
17223         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
17224         FOR_EACH_DEFINED_FUNCTION when freeing state.
17226         PR middle-end/56461
17227         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
17228         pool_free.
17229         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
17230         overwriting it.
17232         PR middle-end/56461
17233         * ipa-cp.c (decide_whether_version_node): Call vec_free on
17234         known_aggs[i].items and release known_aggs vector.
17236         PR middle-end/56461
17237         * ipa-reference.c (propagate): Free node_info even for alias nodes.
17239 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17241         * config/microblaze/microblaze.c (microblaze_emit_compare):
17242         Use xor for EQ/NE comparisions.
17243         * config/microblaze/microblaze.md (cstoresf4): Add constraints
17244         (cbranchsf4): Adjust operator to comparison_operator.
17246 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
17248         PR middle-end/56461
17249         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
17250         vector.
17251         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
17252         vec_safe_push, always update *slot.
17253         (redirect_edge_var_map_clear): Use vec_free.
17254         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
17255         (free_var_map_entry): Use vec_free.
17256         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
17257         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
17259 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
17261         PR middle-end/45472
17262         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
17263         when the may_trap_p bit of the exprs being merged differs.
17264         Reorder tests for speculativeness in the logical and operator.
17266 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
17268         * incpath.c (add_standard_paths): Use reconcat instead of concat
17269         where appropriate and avoid leaking memory.
17271         * opts.h: Include obstack.h.
17272         (opts_concat): New prototype.
17273         (opts_obstack): New declaration.
17274         * opts.c (opts_concat): New function.
17275         (opts_obstack): New variable.
17276         (init_options_struct): Call gcc_init_obstack on opts_obstack.
17277         (finish_options): Use opts_concat instead of concat
17278         and XOBNEWVEC instead of XNEWVEC.
17279         * opts-common.c (generate_canonical_option, decode_cmdline_option,
17280         generate_option): Likewise.
17281         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
17282         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
17284         PR target/56455
17285         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
17286         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
17288 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
17290         PR middle-end/56461
17291         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
17293 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
17295         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
17296         (arm_block_move_unaligned_straight): Likewise.
17297         (arm_adjust_block_mem): Likewise.
17299 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
17301         PR target/48901
17302         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
17303         temp, cond and label.
17304         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
17306         PR target/52500
17307         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
17308         * config/c6x/c6x.h (dbx_register_map): Update declaration.
17310         PR target/52501
17311         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
17312         of prologue/epilogue functions.
17314         PR target/52550
17315         * config/tilegx/tilegx.c (tilegx_expand_prologue):
17316         Remove unused variable cfa_offset.
17317         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
17319         PR target/54639
17320         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
17321         type promotion to unsigned.
17323         PR target/54640
17324         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
17325         for HOST_WIDE_INT of 32 bit / same size as int.
17326         (arm_block_move_unaligned_straight): Likewise.
17327         (arm_adjust_block_mem): Likewise.
17329         PR target/54662
17330         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
17331         ALL_CFLAGS.
17333 2013-02-26  Marek Polacek  <polacek@redhat.com>
17335         PR tree-optimization/56426
17336         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
17338 2013-02-26  Richard Biener  <rguenther@suse.de>
17340         PR target/56444
17341         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
17342         unused variable loops.
17344 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
17346         PR tree-optimization/56448
17347         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
17348         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
17349         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
17350         later operands of the references, or even first operand for
17351         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
17353         PR tree-optimization/56443
17354         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
17355         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
17356         to type_for_mode langhook.
17358 2013-02-25  Matt Turner  <mattst88@gmail.com>
17360         * doc/invoke.texi: Document r4700.
17362 2013-02-25  Richard Biener  <rguenther@suse.de>
17364         PR tree-optimization/56175
17365         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
17366         split out from ...
17367         (simplify_bitwise_binary): ... here.  Also guard the conversion
17368         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
17370 2013-02-25  Catherine Moore  <clm@codesourcery.com>
17372         Revert:
17373         2013-02-24  Catherine Moore  <clm@codesourcery.com>
17374             Maciej W. Rozycki  <macro@codesourcery.com>
17375             Tom de Vries  <tom@codesourcery.com>
17376             Nathan Sidwell  <nathan@codesourcery.com>
17377             Iain Sandoe  <iain@codesourcery.com>
17378             Nathan Froyd  <froydnj@codesourcery.com>
17379             Chao-ying Fu  <fu@mips.com>
17381         * doc/extend.texi: (micromips, nomicromips, nocompression):
17382         Document new function attributes.
17383         * doc/invoke.texi (minterlink-compressed, mmicromips,
17384         m14k, m14ke, m14kec): Document new options.
17385         (minterlink-mips16): Update documentation.
17386         * doc/md.texi (ZC, ZD): Document new constraints.
17387         * configure.ac (gcc_cv_as_micromips): Check if linker
17388         supports the .set micromips directive.
17389         * configure: Regenerate.
17390         * config.in: Regenerate.
17391         * config/mips/mips-tables.opt: Regenerate.
17392         * config/mips/micromips.md: New file.
17393         * constraints.md (ZC, AD): New constraints.
17394         * config/mips/predicates.md (movep_src_register): New predicate.
17395         (movep_src_operand): New predicate.
17396         (non_volatile_mem_operand): New predicate.
17397         * config/mips/mips.md (multimem): New type.
17398         (length): Differentiate between 17-bit and 18-bit branch offsets.
17399         (MOVEP1, MOVEP2): New mode iterator.
17400         (mov_<load>l): Use ZC constraint.
17401         (mov_<load>r): Likewise.
17402         (mov_<store>l): Likewise.
17403         (mov_<store>r): Likewise.
17404         (*branch_equality<mode>_inverted): Add microMIPS support.
17405         (*branch_equality<mode>): Likewise.
17406         (*jump_absolute): Likewise.
17407         (indirect_jump_<mode>): Likewise.
17408         (tablejump_<mode>): Likewise.
17409         (<optab>_internal): Likewise.
17410         (sibcall_internal): Likewise.
17411         (sibcall_value_internal): Likewise.
17412         (prefetch): Use constraint ZD.
17413         * config/mips/mips.opt (minterlink-compressed): New option.
17414         (minterlink-mips16): Now an alias for minterlink-compressed.
17415         (mmicromips): New option.
17416         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17417         (compare_and_swap_12): Likewise.
17418         (sync_add<mode>): Likewise.
17419         (sync_<optab>_12): Likewise.
17420         (sync_old_<optab>_12): Likewise.
17421         (sync_new_<optab>_12): Likewise.
17422         (sync_nand_12): Likewise.
17423         (sync_old_nand_12): Likewise.
17424         (sync_new_nand_12): Likewise.
17425         (sync_sub<mode>): Likewise.
17426         (sync_old_add<mode>): Likewise.
17427         (sync_old_sub<mode>): Likewise.
17428         (sync_new_add<mode>): Likewise.
17429         (sync_new_sub<mode>): Likewise.
17430         (sync_<optab><mode>): Likewise.
17431         (sync_old_<optab><mode>): Likewise.
17432         (sync_new_<optab><mode>): Likewise.
17433         (sync_nand<mode>): Likewise.
17434         (sync_old_nand<mode>): Likewise.
17435         (sync_new_nand<mode>): Likewise.
17436         (sync_lock_test_and_set<mode>): Likewise.
17437         (test_and_set_12): Likewise.
17438         (atomic_compare_and_swap<mode>): Likewise.
17439         (atomic_exchange<mode>_llsc): Likewise.
17440         (atomic_fetch_add<mode>_llsc): Likewise.
17441         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17442         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17443         (umips_save_restore_pattern_p): Likewise.
17444         (umips_load_store_pair_p): Likewise.
17445         (umips_output_load_store_pair): Likewise.
17446         (umips_movep_target_p): Likewise.
17447         (umips_12bit_offset_address_p): Likewise.
17448         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17449         (mips_base_mips16): Rename this...
17450         (mips_base_compression_flags): ...to this. Update all uses.
17451         (mips_attribute_table): Add micromips, nomicromips and nocompression.
17452         (mips_mips16_decl_p): Delete.
17453         (mips_nomips16_decl_p): Delete.
17454         (mips_get_compress_on_flags): New function.
17455         (mips_get_compress_off_flags): New function.
17456         (mips_get_compress_mode): New function.
17457         (mips_get_compress_on_name): New function.
17458         (mips_get_compress_off_name): New function.
17459         (mips_insert_attributes): Support multiple compression types.
17460         (mips_merge_decl_attributes): Likewise.
17461         (umips_12bit_offset_address_p): New function.
17462         (mips_start_function_definition): Emit .set micromips directive.
17463         (mips_call_may_need_jalx_p): New function.
17464         (mips_function_ok_for_sibcall): Add microMIPS support.
17465         (mips_print_operand_punctuation): Support short delay slots and
17466         compact jumps.
17467         (umips_swm_mask, umips_swm_encoding): New.
17468         (umips_build_save_restore): New function.
17469         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17470         (was_mips16_p): Remove.
17471         (old_compression_mode): New.
17472         (mips_set_compression_mode): New function.
17473         (mips_set_current_function): Add microMIPS support.
17474         (mips_option_override): Likewise.
17475         (umips_save_restore_pattern_p): New function.
17476         (umips_output_save_restore): New function.
17477         (umips_load_store_pair_p_1): New function.
17478         (umips_load_store_pair_p): New function.
17479         (umips_output_load_store_pair_1): New function.
17480         (umips_output_load_store_pair): New function.
17481         (umips_movep_target_p) New function.
17482         (mips_prepare_pch_save): Add microMIPS support.
17483         * config/mips/mips.h (TARGET_COMPRESSION): New.
17484         (TARGET_CPU_CPP_BUILTINS): Update macro
17485         to use new compression flags and to support microMIPS.
17486         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17487         (MIPS_ARCH_FLOAT_SPEC): Likewise.
17488         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17489         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17490         (ASM_SPEC): Support mmicromips and mno-micromips.
17491         (M16STORE_REG_P): New macro.
17492         (MIPS_CALL): Support TARGET_MICROMIPS.
17493         (MICROMIPS_J): New macro.
17494         (mips_base_mips16): Rename this...
17495         (mips_base_compression_flags): ...to this.
17496         (UMIPS_12BIT_OFFSET_P): New macro.
17497         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17498         (MULTILIB_DIRNAMES): Likewise.
17500 2013-02-25  Tom de Vries  <tom@codesourcery.com>
17502         PR rtl-optimization/56131
17503         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
17504         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
17505         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
17507 2013-02-25  Tobias Burnus  <burnus@net-b.de>
17509         * doc/invoke.texi (-fsanitize=): Move from optimization
17510         to debugging options.
17512 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
17514         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
17516 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
17517             Alexander Monakov  <amonakov@ispras.ru>
17519         PR middle-end/56077
17520         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
17521         flush pending lists also on non-jumps.  Adjust comment.
17523 2013-02-24  Catherine Moore  <clm@codesourcery.com>
17524             Maciej W. Rozycki  <macro@codesourcery.com>
17525             Tom de Vries  <tom@codesourcery.com>
17526             Nathan Sidwell  <nathan@codesourcery.com>
17527             Iain Sandoe  <iain@codesourcery.com>
17528             Nathan Froyd  <froydnj@codesourcery.com>
17529             Chao-ying Fu  <fu@mips.com>
17531         * doc/extend.texi: (micromips, nomicromips, nocompression):
17532         Document new function attributes.
17533         * doc/invoke.texi (minterlink-compressed, mmicromips,
17534         m14k, m14ke, m14kec): Document new options.
17535         (minterlink-mips16): Update documentation.
17536         * doc/md.texi (ZC, ZD): Document new constraints.
17537         * configure.ac (gcc_cv_as_micromips): Check if linker
17538         supports the .set micromips directive.
17539         * configure: Regenerate.
17540         * config.in: Regenerate.
17541         * config/mips/mips-tables.opt: Regenerate.
17542         * config/mips/micromips.md: New file.
17543         * constraints.md (ZC, AD): New constraints.
17544         * config/mips/predicates.md (movep_src_register): New predicate.
17545         (movep_src_operand): New predicate.
17546         (non_volatile_mem_operand): New predicate.
17547         * config/mips/mips.md (multimem): New type.
17548         (length): Differentiate between 17-bit and 18-bit branch offsets.
17549         (MOVEP1, MOVEP2): New mode iterator.
17550         (mov_<load>l): Use ZC constraint.
17551         (mov_<load>r): Likewise.
17552         (mov_<store>l): Likewise.
17553         (mov_<store>r): Likewise.
17554         (*branch_equality<mode>_inverted): Add microMIPS support.
17555         (*branch_equality<mode>): Likewise.
17556         (*jump_absolute): Likewise.
17557         (indirect_jump_<mode>): Likewise.
17558         (tablejump_<mode>): Likewise.
17559         (<optab>_internal): Likewise.
17560         (sibcall_internal): Likewise.
17561         (sibcall_value_internal): Likewise.
17562         (prefetch): Use constraint ZD.
17563         * config/mips/mips.opt (minterlink-compressed): New option.
17564         (minterlink-mips16): Now an alias for minterlink-compressed.
17565         (mmicromips): New option.
17566         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17567         (compare_and_swap_12): Likewise.
17568         (sync_add<mode>): Likewise.
17569         (sync_<optab>_12): Likewise.
17570         (sync_old_<optab>_12): Likewise.
17571         (sync_new_<optab>_12): Likewise.
17572         (sync_nand_12): Likewise.
17573         (sync_old_nand_12): Likewise.
17574         (sync_new_nand_12): Likewise.
17575         (sync_sub<mode>): Likewise.
17576         (sync_old_add<mode>): Likewise.
17577         (sync_old_sub<mode>): Likewise.
17578         (sync_new_add<mode>): Likewise.
17579         (sync_new_sub<mode>): Likewise.
17580         (sync_<optab><mode>): Likewise.
17581         (sync_old_<optab><mode>): Likewise.
17582         (sync_new_<optab><mode>): Likewise.
17583         (sync_nand<mode>): Likewise.
17584         (sync_old_nand<mode>): Likewise.
17585         (sync_new_nand<mode>): Likewise.
17586         (sync_lock_test_and_set<mode>): Likewise.
17587         (test_and_set_12): Likewise.
17588         (atomic_compare_and_swap<mode>): Likewise.
17589         (atomic_exchange<mode>_llsc): Likewise.
17590         (atomic_fetch_add<mode>_llsc): Likewise.
17591         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17592         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17593         (umips_save_restore_pattern_p): Likewise.
17594         (umips_load_store_pair_p): Likewise.
17595         (umips_output_load_store_pair): Likewise.
17596         (umips_movep_target_p): Likewise.
17597         (umips_12bit_offset_address_p): Likewise.
17598         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17599         (mips_base_mips16): Rename this...
17600         (mips_base_compression_flags): ...to this. Update all uses.
17601         (mips_attribute_table): Add micromips, nomicromips and nocompression.
17602         (mips_mips16_decl_p): Delete.
17603         (mips_nomips16_decl_p): Delete.
17604         (mips_get_compress_on_flags): New function.
17605         (mips_get_compress_off_flags): New function.
17606         (mips_get_compress_mode): New function.
17607         (mips_get_compress_on_name): New function.
17608         (mips_get_compress_off_name): New function.
17609         (mips_insert_attributes): Support multiple compression types.
17610         (mips_merge_decl_attributes): Likewise.
17611         (umips_12bit_offset_address_p): New function.
17612         (mips_start_function_definition): Emit .set micromips directive.
17613         (mips_call_may_need_jalx_p): New function.
17614         (mips_function_ok_for_sibcall): Add microMIPS support.
17615         (mips_print_operand_punctuation): Support short delay slots and
17616         compact jumps.
17617         (umips_swm_mask, umips_swm_encoding): New.
17618         (umips_build_save_restore): New function.
17619         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17620         (was_mips16_p): Remove.
17621         (old_compression_mode): New.
17622         (mips_set_compression_mode): New function.
17623         (mips_set_current_function): Add microMIPS support.
17624         (mips_option_override): Likewise.
17625         (umips_save_restore_pattern_p): New function.
17626         (umips_output_save_restore): New function.
17627         (umips_load_store_pair_p_1): New function.
17628         (umips_load_store_pair_p): New function.
17629         (umips_output_load_store_pair_1): New function.
17630         (umips_output_load_store_pair): New function.
17631         (umips_movep_target_p) New function.
17632         (mips_prepare_pch_save): Add microMIPS support.
17633         * config/mips/mips.h (TARGET_COMPRESSION): New.
17634         (TARGET_CPU_CPP_BUILTINS): Update macro
17635         to use new compression flags and to support microMIPS.
17636         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17637         (MIPS_ARCH_FLOAT_SPEC): Likewise.
17638         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17639         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17640         (ASM_SPEC): Support mmicromips and mno-micromips.
17641         (M16STORE_REG_P): New macro.
17642         (MIPS_CALL): Support TARGET_MICROMIPS.
17643         (MICROMIPS_J): New macro.
17644         (mips_base_mips16): Rename this...
17645         (mips_base_compression_flags): ...to this.
17646         (UMIPS_12BIT_OFFSET_P): New macro.
17647         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17648         (MULTILIB_DIRNAMES): Likewise.
17650 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
17652         PR target/52555
17653         * target-globals.c (save_target_globals): For init_reg_sets and
17654         target_reinit remporarily set this_fn_optabs to this_target_optabs.
17656 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
17658         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
17659         * config/aarch64/t-aarch64
17660         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
17662 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
17664         PR inline-asm/56148
17665         * lra-constraints.c (process_alt_operands): Reload operand
17666         conflicting with earlier clobber only if no more other conflicting
17667         operands.
17669 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
17671         PR sanitizer/56393
17672         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
17673         if not linking a shared library.
17675 2013-02-22  Seth LaForge  <sethml@google.com>
17677         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
17679 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
17681         * config/arm/arm.md (split for extendsidi): Update condition.
17682         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
17683         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
17684         (qhs_zextenddi_cstr): Likewise.
17686 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
17688         PR middle-end/56420
17689         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
17690         avoid signed wrapping.
17691         (expand_mult): Handle properly multiplication by
17692         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
17693         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
17694         in the compiler if coeff is HOST_WIDE_INT_MIN.
17695         (expand_divmod): Don't make ext_op1 static, change it's type to
17696         uhwi.  Avoid undefined behavior in -INTVAL (op1).
17698         PR rtl-optimization/50339
17699         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
17700         field.
17701         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
17702         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
17703         into splitting_ashiftrt field.
17704         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
17705         ASHIFTRT.
17706         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
17707         choices.
17709 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
17711         PR middle-end/56108
17712         * trans-mem.c (execute_tm_mark): Do not expand transactions that
17713         are sure to go irrevocable.
17715 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
17717         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
17718         scalars are valid operands.
17720 2013-02-21  Martin Jambor  <mjambor@suse.cz>
17722         PR tree-optimization/56310
17723         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
17724         only matching indices and non-negative final offsets.
17725         (intersect_aggregates_with_edge): Pass src_idx to
17726         agg_replacements_to_vector.  Pass src_idx insstead of index to
17727         intersect_with_agg_replacements.
17729 2013-02-21  Martin Jambor  <mjambor@suse.cz>
17731         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
17732         instead of hard-wired defaults.
17734 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
17736         * doc/invoke.texi (MIPS Options): Update documentation of the
17737         floating-point multiply-accumulate instruction restrictions.
17739 2013-02-21  Kostya Serebryany  <kcc@google.com>
17741         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
17742         asan_shadow_offset on x86_64 linux.
17744 2013-02-21  Richard Biener  <rguenther@suse.de>
17746         PR tree-optimization/56415
17747         Revert
17748         2013-02-11  Richard Biener  <rguenther@suse.de>
17750         PR tree-optimization/56273
17751         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17752         first VRP run.
17754 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
17756         PR bootstrap/56258
17757         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
17758         instead of @itemx.
17760         PR inline-asm/56405
17761         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
17762         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
17764 2013-02-20  Jan Hubicka  <jh@suse.cz>
17766         PR tree-optimization/56265
17767         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
17768         when target is referenced for first time.
17770 2013-02-20  Richard Biener  <rguenther@suse.de>
17772         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
17773         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
17774         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
17775         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
17776         not return anything.
17777         (rename_ssa_copies): Do not remove unused locals.
17778         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
17779         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
17780         * passes.c (execute_function_todo): Do not schedule unused locals
17781         removal if cleanup_tree_cfg did something.
17782         * tree-ssa-live.c (remove_unused_locals): Dump statistics
17783         about the number of removed locals.
17785 2013-02-20  Richard Biener  <rguenther@suse.de>
17787         PR tree-optimization/56398
17788         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
17790 2013-02-20  Martin Jambor  <mjambor@suse.cz>
17792         PR tree-optimization/55334
17793         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
17794         restricted pointers to arrays.
17796 2013-02-20  Richard Biener  <rguenther@suse.de>
17797             Jakub Jelinek  <jakub@redhat.com>
17799         PR tree-optimization/56396
17800         * tree-ssa-ccp.c (n_const_val): New static variable.
17801         (get_value): Return NULL for SSA names we don't have a lattice
17802         entry for.
17803         (ccp_initialize): Initialize n_const_val.
17804         * tree-ssa-copy.c (n_copy_of): New static variable.
17805         (init_copy_prop): Initialize n_copy_of.
17806         (get_value): Return NULL_TREE for SSA names we don't have a
17807         lattice entry for.
17809 2013-02-20  Martin Jambor  <mjambor@suse.cz>
17811         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
17813 2013-02-20  Richard Biener  <rguenther@suse.de>
17815         * genpreds.c (write_lookup_constraint): Do not compare first
17816         letter of the constraint again.
17818 2013-02-20  Richard Biener  <rguenther@suse.de>
17820         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
17821         and ceil_log2.
17822         (get_use_iv_cost): Terminate hashtable walk when coming across
17823         an empty entry.
17825 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
17827         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
17828         reassociation for avx2 targets.
17830 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17832         * config/microblaze/microblaze.c: microblaze_has_clz = 0
17833         Add version check for v8.10.a to enable microblaze_has_clz
17834         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
17835         version and TARGET_PATTERN_COMPARE check
17836         * config/microblaze/microblaze.md: New clzsi2 instruction
17838 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17840         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
17841         function before branching.
17843 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17845         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
17846         DUMP_INSN_RTX_UID.
17847         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
17849 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
17851         PR middle-end/55889
17852         * sel-sched.c: Include ira.h.
17853         (implicit_clobber_conflict_p): New function.
17854         (moveup_expr): Use it.
17855         * Makefile.in (sel-sched.o): Depend on ira.h.
17857 2013-02-19  Richard Biener  <rguenther@suse.de>
17859         PR tree-optimization/56384
17860         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
17861         (vn_hash_type): Split out from ...
17862         (vn_hash_constant_with_type): ... here.
17863         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
17864         (vn_phi_eq): Compare types from vn_phi_s structure.
17865         (vn_phi_lookup): Populate vn_phi_s type.
17866         (vn_phi_insert): Likewise.
17868 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
17870         PR tree-optimization/56350
17871         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
17872         if haven't found reduction or nested cycle operand, rather than
17873         asserting we must find it.
17875         PR tree-optimization/56381
17876         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
17877         to fold_build3.
17879 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
17880             Jakub Jelinek  <jakub@redhat.com>
17882         PR target/52555
17883         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
17884         (swap_optab_enable): Same.
17885         (init_all_optabs): Use argument instead of global.
17886         * tree.h (struct tree_optimization_option): New field target_optabs.
17887         * expr.h (init_all_optabs): Add argument to prototype.
17888         (TREE_OPTIMIZATION_OPTABS): New.
17889         (save_optabs_if_changed): Protoize.
17890         * optabs.h: Declare this_fn_optabs.
17891         * optabs.c (save_optabs_if_changed): New.
17892         Declare this_fn_optabs.
17893         (init_optabs): Add argument to init_all_optabs() call.
17894         * function.c (invoke_set_current_function_hook): Handle per
17895         function optabs.
17896         * function.h (struct function): New field optabs.
17897         * config/mips/mips.c (mips_set_mips16_mode): Handle when
17898         optimization_current_node has changed.
17899         * target-globals.h (save_target_globals_default_opts): Protoize.
17900         * target-globals.c (save_target_globals_default_opts): New.
17902 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17904         PR target/56347
17905         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
17906         registers %fr12 and %fr12R as call used.
17908         PR target/56214
17909         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
17910         and HImode, require all displacements to be an integer multiple of
17911         their mode size.
17912         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
17913         only allow QImode and HImode when reload is in progress and strict is
17914         true.  Likewise for symbolic addresses.  Use base14_operand to check
17915         displacements in REG+BASE addresses.
17917 2013-02-18  Richard Biener  <rguenther@suse.de>
17919         PR tree-optimization/56366
17920         * tree-vect-loop.c (get_initial_def_for_induction): Properly
17921         handle sign-conversion of outer-loop initial induction value.
17923 2013-02-18  Richard Biener  <rguenther@suse.de>
17925         PR middle-end/56349
17926         * cfghooks.c (merge_blocks): If we merge a latch into another
17927         block adjust references to it.
17928         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
17929         (verify_loop_structure): Verify that a recorded latch is in fact
17930         a latch.
17932 2013-02-18  Richard Biener  <rguenther@suse.de>
17934         PR tree-optimization/56321
17935         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
17936         order SSA name release and virtual operand unlinking.
17938 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17940         * config/microblaze/microblaze.md (save_stack_block): Define.
17941         (restore_stack_block): Likewise.
17943 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17945         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17946         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17947         * config/microblaze/microblaze.c (microblaze_option_override):
17948         Bail out early for PIC modes when target does not support PIC.
17950 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17952         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17953         Replace with a microblaze version.
17954         (microblaze_trampoline_init): Adapt for microblaze.
17955         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17956         microblaze.
17958 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
17959             Dodji Seketeli  <dodji@redhat.com>
17961         PR asan/56330
17962         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17963         (instrument_mem_region_access): Do not forget to always put
17964         instrumentation of the of 'base' and 'base + len' in a "if (len !=
17965         0) statement, even for cases where either 'base' or 'base + len'
17966         are not instrumented -- because they have been previously
17967         instrumented.  Simplify the logic by putting all the statements
17968         instrument 'base + len' inside a sequence, and then insert that
17969         sequence right before the current insertion point.  Then, to
17970         instrument 'base + len', just get an iterator on that statement.
17971         And do not forget to update the pointer to iterator the function
17972         received as argument.
17974 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
17976         PR rtl-optimization/56348
17977         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17979 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
17981         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17982         (clean_graph_dump_file): Pass base to start_graph_dump.
17984 2013-02-14  Richard Henderson  <rth@redhat.com>
17986         PR target/55941
17987         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17989 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
17991         * collect2-aix.h: Define F_LOADONLY.
17993 2013-02-14  Richard Biener  <rguenther@suse.de>
17995         PR lto/50494
17996         * varasm.c (output_constant_def_1): Get the decl representing
17997         the constant as argument.
17998         (output_constant_def): Wrap output_constant_def_1.
17999         (make_decl_rtl): Use output_constant_def_1 with the decl
18000         representing the constant.
18001         (build_constant_desc): Optionally re-use a decl already
18002         representing the constant.
18003         (tree_output_constant_def): Adjust.
18005 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
18007         Fix an asan crash
18008         * asan.c (instrument_builtin_call):  Really put the length of the
18009         second source argument into src1_len.
18011 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
18013         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
18014         argument.  If it is false, don't create edge from then_bb to
18015         fallthru_bb.
18016         (insert_if_then_before_iter): Pass true to it.
18017         (build_check_stmt): Pass false to it.
18018         (transform_statements): Flush hash table only on extended basic
18019         block boundaries, rather than at the beginning of every bb.
18020         Don't flush hash table on nonfreeing_call_p calls.
18021         * tree-flow.h (nonfreeing_call_p): New prototype.
18022         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
18024 2013-02-13  David S. Miller  <davem@davemloft.net>
18026         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
18028 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
18030         PR target/56184
18031         * ira.c (max_regno_before_ira): Move from ...
18032         (ira): ... here.
18033         (fix_reg_equiv_init): Use max_regno_before_ira instead of
18034         vec_safe_length.
18036 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
18038         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
18040 2013-02-13  Richard Biener  <rguenther@suse.de>
18042         PR lto/56295
18043         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
18044         globals in MEM_REFs.
18046 2013-02-13  Richard Biener  <rguenther@suse.de>
18048         * loop-init.c (loop_optimizer_init): Clear loop state when
18049         re-initializing preserved loops.
18050         * loop-unswitch.c (unswitch_single_loop): Return whether
18051         we unswitched the loop.  Do not verify loop state here.
18052         (unswitch_loops): When we unswitched a loop discover new loops.
18054 2013-02-13  Kostya Serebryany  <kcc@google.com>
18056         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
18057         on x86_64 linux.
18058         * sanitizer.def: Rename __asan_init to __asan_init_v1.
18060 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
18062         Avoid instrumenting duplicated memory access in the same basic block
18063         * Makefile.in (asan.o): Add new dependency on hash-table.h
18064         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
18065         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
18066         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
18067         (free_mem_ref_resources, has_mem_ref_been_instrumented)
18068         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
18069         (get_mem_ref_of_assignment): New functions.
18070         (get_mem_refs_of_builtin_call): Extract from
18071         instrument_builtin_call and tweak a little bit to make it fit with
18072         the new signature.
18073         (instrument_builtin_call): Use the new
18074         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
18075         of is_gimple_builtin_call.
18076         (instrument_derefs, instrument_mem_region_access): Insert the
18077         instrumented memory reference into the hash table.
18078         (maybe_instrument_assignment): Renamed instrument_assignment into
18079         this, and change it to advance the iterator when instrumentation
18080         actually happened and return true in that case.  This makes it
18081         homogeneous with maybe_instrument_assignment, and thus give a
18082         chance to callers to be more 'regular'.
18083         (transform_statements): Clear the memory reference hash table
18084         whenever we enter a new BB, when we cross a function call, or when
18085         we are done transforming statements.  Use
18086         maybe_instrument_assignment instead of instrumentation.  No more
18087         need to special case maybe_instrument_assignment and advance the
18088         iterator after calling it; it's now handled just like
18089         maybe_instrument_call.  Update comment.
18091 2013-02-13  Richard Biener  <rguenther@suse.de>
18093         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
18094         Fix loop discovery code.
18096 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
18098         PR inline-asm/56148
18099         * lra-constraints.c (process_alt_operands): Match early clobber
18100         operand with itself.  Check conflicts with earlyclobber only if
18101         the operand is not reloaded.  Prefer to reload conflicting operand
18102         if earlyclobber and matching operands are the same.
18104 2013-02-12  Richard Biener  <rguenther@suse.de>
18106         PR lto/56297
18107         * lto-streamer-out.c (write_symbol): Do not output symbols
18108         for hard register variables.
18110 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
18112         PR target/54222
18113         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
18114         (umulsidi3_insn, mulsidi3_insn): New insns.
18116 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
18118         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
18119         (struct tune_params): Add vec_costs field.
18120         * config/arm/arm.c (arm_builtin_vectorization_cost)
18121         (arm_add_stmt_cost): New functions.
18122         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
18123         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
18124         (arm_default_vec_cost): New struct of type cpu_vec_costs.
18125         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
18126         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
18127         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
18128         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
18130 2013-02-12  Richard Biener  <rguenther@suse.de>
18132         PR lto/56295
18133         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
18134         decls again if possible.
18136 2013-02-12  Richard Biener  <rguenther@suse.de>
18138         PR middle-end/56288
18139         * tree-ssa.c (verify_ssa_name): Fix check, move
18140         SSA_NAME_IN_FREE_LIST check up.
18142 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
18143             Steven Bosscher   <steven@gcc.gnu.org>
18145         PR rtl-optimization/56151
18146         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
18147         equal to op0 or op1, and last_insn pattern is CODE operation
18148         with MEM dest and one of the operands matches that MEM.
18150 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
18152         * doc/extend.texi: Document Function Multiversioning and "default"
18153         parameter string to target attribute.
18154         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
18155         target attribute parameter is "default".
18156         (ix86_compare_version_priority): Remove checks for target attribute.
18157         (ix86_mangle_function_version_assembler_name): Change error to sorry.
18158         Remove check for target attribute equal to NULL. Add assert.
18159         (ix86_generate_version_dispatcher_body): Change error to sorry.
18161 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
18162             Jack Howarth  <howarth@bromo.med.uc.edu>
18163             Patrick Marlier  <patrick.marlier@gmail.com>
18165         PR libitm/55693
18166         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
18167         define ENDFILE_SPEC as TM_DESTRUCTOR.
18168         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
18170 2013-02-11  Alexander Potapenko  <glider@google.com>
18171             Jack Howarth  <howarth@bromo.med.uc.edu>
18172             Jakub Jelinek  <jakub@redhat.com>
18174         PR sanitizer/55617
18175         * config/darwin.c (cdtor_record): Rename ctor_record.
18176         (sort_cdtor_records): Rename sort_ctor_records.
18177         (finalize_dtors): New routine to sort destructors by
18178         priority before use in assemble_integer.
18179         (machopic_asm_out_destructor): Use finalize_dtors if needed.
18181 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
18183         PR rtl-optimization/56275
18184         * simplify-rtx.c (avoid_constant_pool_reference): Check that
18185         offset is non-negative and less than cmode size before
18186         calling simplify_subreg.
18188 2013-02-11  Richard Biener  <rguenther@suse.de>
18190         PR tree-optimization/56264
18191         * cfgloop.h (fix_loop_structure): Adjust prototype.
18192         * loop-init.c (fix_loop_structure): Return the number of
18193         newly discovered loops.
18194         * tree-cfgcleanup.c (repair_loop_structures): When new loops
18195         are discovered, do a full loop-closed SSA rewrite.
18197 2013-02-11  Richard Biener  <rguenther@suse.de>
18199         PR tree-optimization/56273
18200         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
18201         first VRP run.
18202         (check_array_ref): Fix missing newline in dumps.
18203         (search_for_addr_array): Likewise.
18205 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
18207         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
18209 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
18211         PR target/56256
18212         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
18214 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
18216         PR rtl-optimization/56246
18217         * lra-constraints.c (simplify_operand_subreg): Try to reuse
18218         reload pseudo.
18219         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
18220         constraints are satisfied.
18222 2013-02-08  Jeff Law  <law@redhat.com>
18224         PR debug/53948
18225         * emit-rtl.c (reg_is_parm_p): New function.
18226         * regs.h (reg_is_parm_p): New prototype.
18227         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
18228         callee-clobbered registers.
18230 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
18232         PR target/56043
18233         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
18234         If there is no implicit builtin declaration, just return NULL.
18236 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
18238         * config/i386/sse.md (FMAMODEM): New mode iterator.
18239         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
18240         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
18242 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
18244         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
18245         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
18246         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
18248 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18250         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
18251         (microblaze*-*-elf): Likewise.
18252         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
18253         LINK_SPEC.
18254         * config/microblaze/microblaze-c.c: Add builtin defines for
18255         _LITTLE_ENDIAN and _BIG_ENDIAN.
18256         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
18257         add to TARGET_DEFAULT flags.
18258         Expand ASM_SPEC and LINK_SPEC.
18259         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
18260         * config/microblaze/microblaze.md: Update extendsidi2 and
18261         movdi_internal instructions to use low-order / high-order reg
18262         print_operands.
18263         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
18264         options and inversemask / mask of LITTLE_ENDIAN.
18265         * config/microblaze/t-microblaze: Expand multilib options to
18266         include mlittle-endian (le) and update exceptions patterns.
18268 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
18270         PR rtl-optimization/56195
18271         * lra-constraints.c (get_reload_reg): Don't reuse regs
18272         if they have smaller mode than requested, if they have
18273         wider mode than requested, try to return a SUBREG.
18275         PR tree-optimization/56250
18276         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
18277         if type is unsigned and code isn't MULT_EXPR.
18279 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
18281         PR tree-optimization/56064
18282         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
18283         bits according to mode.
18284         * fixed-value.h (fixed_from_double_int)
18285         (const_fixed_from_double_int): Adjust comments.
18287 2013-02-08  Richard Biener  <rguenther@suse.de>
18289         PR lto/56231
18290         * lto-streamer.h (struct data_in): Remove current_file, current_line
18291         and current_col members.
18292         * lto-streamer-out.c (lto_output_location): Stream changed bits
18293         en-block for efficiency.
18294         * lto-streamer-in.c (clear_line_info): Remove.
18295         (lto_input_location): Cache current file, line and column
18296         globally via local statics.  Read changed bits en-block.
18297         (input_function): Do not call clear_line_info.
18298         (lto_read_body): Likewise.
18299         (lto_input_toplevel_asms): Likewise.
18301 2013-02-08  Michael Matz  <matz@suse.de>
18303         PR tree-optimization/52448
18304         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
18305         (nt_call_phase): New static.
18306         (add_or_mark_expr): Only mark accesses with newer phase than any
18307         call seen.
18308         (nonfreeing_call_p): New.
18309         (nt_init_block): Update nt_call_phase, mark blocks as visited.
18310         (nt_fini_block): Keep blocks marked as visited.
18311         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
18313 2013-02-08  Richard Biener  <rguenther@suse.de>
18315         * ira.c (ira): Free broken dominator information.
18317 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
18319         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
18321 2013-02-08  Marek Polacek  <polacek@redhat.com>
18323         * cfgloop.c (verify_loop_structure): Add more checking of headers.
18325 2013-02-08  Richard Biener  <rguenther@suse.de>
18327         PR middle-end/56181
18328         * cfgloop.h (flow_loops_find): Adjust.
18329         (bb_loop_header_p): Declare.
18330         * cfgloop.c (bb_loop_header_p): New function split out from ...
18331         (flow_loops_find): ... here.  Adjust function signature,
18332         support incremental loop structure update.
18333         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
18334         * cfgloopmanip.c (fix_loop_structure): Move ...
18335         * loop-init.c (fix_loop_structure): ... here.
18336         (apply_loop_flags): Split out from ...
18337         (loop_optimizer_init): ... here.
18338         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
18339         in incremental mode, only remove dead loops here.
18341 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
18343         PR target/54222
18344         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
18345         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
18346         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
18347         (*round<mode>3.libgcc): New insns for fixed-modes.
18348         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
18349         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
18350         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
18351         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
18352         implementations.  Define to __builtin_avr_absFX,
18353         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
18354         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
18355         __builtin_avr_countlsFX, respectively.
18356         * config/avr/avr-c.c (target.h): Include it.
18357         (enum avr_builtin_id): New enum.
18358         (avr_resolve_overloaded_builtin): New static function.
18359         (avr_register_target_pragmas): Use it to set
18360         targetm.resolve_overloaded_builtin.
18361         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
18362         tree nodes used by DEF_BUILTIN.
18363         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
18364         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
18365         <AVR_BUILTIN_xxBITS>: Same.
18367 2013-02-08  Richard Biener  <rguenther@suse.de>
18369         * cfgloop.c (verify_loop_structure): Properly handle
18370         a loop exiting to another loop header.
18371         * ira-int.h (ira_loops): Remove.
18372         * ira.c (ira_loops): Remove.
18373         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
18374         (do_reload): Use loop_optimizer_finalize.
18375         * ira-build.c (create_loop_tree_nodes): Use get_loops and
18376         number_of_loops to access the loop tree.
18377         (more_one_region_p): Likewise.
18378         (finish_loop_tree_nodes): Likewise.
18379         (rebuild_regno_allocno_maps): Likewise.
18380         (mark_loops_for_removal): Likewise.
18381         (mark_all_loops_for_removal): Likewise.
18382         (remove_unnecessary_regions): Likewise.
18383         (ira_build): Likewise.
18384         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18386 2013-02-08  Richard Biener  <rguenther@suse.de>
18388         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
18389         * ipa-pure-const.c (analyze_function): Avoid calling
18390         mark_irreducible_loops twice.
18391         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
18393 2013-02-07  David S. Miller  <davem@davemloft.net>
18395         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
18396         on 'reg'.
18397         * var-tracking.c (vt_add_function_parameter): Test the presence of
18398         HAVE_window_save properly and do not remap argument registers when
18399         we have a leaf function.
18401 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
18403         PR bootstrap/56227
18404         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
18405         instead of "ll".
18406         * config/i386/i386.c (ix86_print_operand): Ditto.
18408 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
18410         * lra-constraints.c (process_alt_operands): Fix recently added comment.
18412 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
18414         PR rtl-optimization/56225
18415         * lra-constraints.c (process_alt_operands): Check that reload hard
18416         reg can hold value for strict_low_part.
18418 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
18420         PR debug/56154
18421         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
18422         dwarf2out_end_function.
18423         (in_first_function_p, maybe_at_text_label_p,
18424         first_loclabel_num_not_at_text_label): New variables.
18425         (dwarf2out_var_location): In the first function find out
18426         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
18427         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
18428         functions.
18430 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
18432         PR rtl-optimization/56178
18433         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
18434         SUBREG of a register.  Tidy up related block of code.
18435         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
18436         note if the source is a register or a SUBREG of a register.
18438 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
18440         PR target/56228
18441         * config/rs6000/rs6000.md (ptrm): New mode attr.
18442         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
18443         call_value_indirect_aix<pttrsize>,
18444         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
18445         m in constraints.
18447 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
18449         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
18450         if -bnortl. Convert to strcmp and strncmp.
18452 2013-02-07  Alan Modra  <amodra@gmail.com>
18454         PR target/54009
18455         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
18456         addresses won't wrap when offsetting.
18457         (rs6000_secondary_reload): Provide secondary reloads needed for
18458         wrapping LO_SUM addresses.
18460 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
18462         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
18463         MACH, just __MACH__.
18465 2013-02-06  Richard Biener  <rguenther@suse.de>
18467         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
18468         instead of calling fix_loop_structure.
18470 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
18472         PR middle-end/56217
18473         * omp-low.c (use_pointer_for_field): Return false if
18474         lower_send_shared_vars doesn't generate any copy-out code.
18476 2013-02-06  Tom de Vries  <tom@codesourcery.com>
18478         PR rtl-optimization/56131
18479         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
18480         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
18481         of the label is NULL.  Add comment.
18483 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
18485         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
18487         PR sanitizer/55374
18488         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
18489         (STATIC_LIBTSAN_LIBS): Likewise.
18490         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
18491         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
18492         is defined, don't add anything else beyond that.
18493         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
18494         (LINK_COMMAND_SPEC): Use them.
18496         PR tree-optimization/56205
18497         * tree-stdarg.c (check_all_va_list_escapes): Return true if
18498         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
18499         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
18501 2013-02-05  Richard Biener  <rguenther@suse.de>
18503         PR tree-optimization/53342
18504         PR tree-optimization/53185
18505         * tree-vectorizer.h (vect_check_strided_load): Remove.
18506         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
18507         not disallow peeling for vectorized strided loads.
18508         (vect_check_strided_load): Make static and simplify.
18509         (vect_analyze_data_refs): Adjust.
18510         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
18511         correctly when vectorizing strided loads.
18513 2013-02-05  Richard Biener  <rguenther@suse.de>
18515         * doc/install.texi: Refer to ISL, not PPL.
18517 2013-02-05  Jan Hubicka  <jh@suse.cz>
18519         PR tree-optimization/55789
18520         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
18522 2013-02-05  Jan Hubicka  <jh@suse.cz>
18524         PR tree-optimization/55789
18525         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
18526         the dead call anyway.
18528 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
18530         PR sanitizer/55374
18531         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
18533 2013-02-04  Alexander Potapenko  <glider@google.com>
18534             Jack Howarth  <howarth@bromo.med.uc.edu>
18535             Jakub Jelinek  <jakub@redhat.com>
18537         PR sanitizer/55617
18538         * config/darwin.c (sort_ctor_records): Stabilized qsort
18539         on constructor priority by using original position.
18540         (finalize_ctors): New routine to sort constructors by
18541         priority before use in assemble_integer.
18542         (machopic_asm_out_constructor): Use finalize_ctors if needed.
18544 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
18546         PR libstdc++/54314
18547         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
18548         about visibility on artificial decls.
18549         * config/sol2.c (solaris_assemble_visibility): Likewise.
18551 2013-02-04  Kai Tietz  <ktietz@redhat.com>
18553         PR target/56186
18554         * config/i386/i386.c (function_value_ms_64): Add additional valtype
18555         argument and improve checking of return-argument types for 16-byte
18556         modes.
18557         (ix86_function_value_1): Add additional valtype argument on call
18558         of function_value_64.
18559         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
18560         handling infunction_value_64 function.
18562 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
18564         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
18566 2013-02-04  Richard Biener  <rguenther@suse.de>
18568         PR tree-optimization/56188
18569         * tree-ssa-structalias.c (label_visit): Consider case with
18570         initially non-empty points-to set.
18571         (perform_var_substitution): Dump node mapping and clean up.
18573 2013-02-04  Richard Guenther  <rguenther@suse.de>
18575         PR lto/56168
18576         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
18577         node prevail as last resort.
18578         (lto_symtab_merge_decls): Remove guard on LTRANS here.
18579         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
18581 2013-02-04  Richard Biener  <rguenther@suse.de>
18583         PR tree-optimization/56113
18584         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
18585         Merge into ...
18586         (equiv_class_lookup_or_add): ... this.
18587         (label_visit): Adjust and fix error in previous patch.
18588         (perform_var_substitution): Adjust.
18590 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
18592         * config/sh/divtab.c: Fix formatting and comments throughout the file.
18593         * config/sh/sh4-300.md: Likewise.
18594         * config/sh/sh4a.md: Likewise.
18595         * config/sh/constraints.md: Likewise.
18596         * config/sh/sh.md: Likewise.
18597         * config/sh/netbsd-elf.h: Likewise.
18598         * config/sh/predicates.md: Likewise.
18599         * config/sh/sh-protos.h: Likewise.
18600         * config/sh/ushmedia.h: Likewise.
18601         * config/sh/linux.h: Likewise.
18602         * config/sh/sh.c: Likewise.
18603         * config/sh/superh.h: Likewise.
18604         * config/sh/elf.h: Likewise.
18605         * config/sh/sh4.md: Likewise.
18606         * config/sh/sh.h: Likewise.
18608 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18610         * config/pa/constraints.md: Adjust unused letters.  Change "T"
18611         constraint to match_test floating_point_store_memory_operand().
18612         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
18613         (base14_operand): New.
18614         (floating_point_store_memory_operand): New.
18615         (integer_store_memory_operand): Revise to use base14_operand and
18616         reg_plus_base_memory_operand.
18617         (move_dest_operand): Allow symbolic_memory_operands.
18618         (symbolic_memory_operand): Check for LO_SOM.
18619         (symbolic_operand): Change default case to break.
18620         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
18621         CONST_DOUBLE values to be reloaded by putting them into memory when
18622         the destination is a floating point register.
18623         (movdf): Remove code to handle CONST_DOUBLE.
18624         (movsf): Likewise.
18625         (reload_indf_r1): New.
18626         (reload_insf_r1): New.
18627         Consistently use "Q" and "T" constraints with integer and floating
18628         point move instructions, respectively.
18629         (movdi): Remove FAIL.
18630         Change predicate for source operand unamed DImode move from
18631         general_operand to move_src_operand.
18632         (umulsidi3): Change predicate for destination operand to
18633         register_operand.
18634         Likewise for similar unamed patterns.
18635         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
18636         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
18637         (hppa_legitimize_address): Simplify mask calculation.
18638         (pa_emit_move_sequence): Revised handling of secondary reloads from
18639         REG+D addresses for floating point loads and stores.  Directly handle
18640         loading CONST0_RTX (mode) to a floating point register.
18641         (pa_secondary_reload): Handle reloading DF and SFmode constant values
18642         to floating point registers.  Don't restrict secondary reloads to
18643         floating point registers to integer modes.  Revise some comments and
18644         cleanup some code.
18645         (TARGET_LEGITIMATE_ADDRESS_P): Define.
18646         (pa_legitimate_address_p): New.
18647         (pa_legitimize_reload_address): New.
18648         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
18649         (STRICT_REG_OK_FOR_BASE_P): New.
18650         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
18651         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
18653 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
18654             Andrew Dixie  <andrewd@gentrack.com>
18656         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
18657         flag set.
18659 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
18661         * expmed.c (extract_bit_field_1): Pass the full width of the
18662         structure to get_best_reg_extraction_insn.
18664 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
18666         PR target/54601
18667         * configure.ac (use_cxa_atexit): Add AIX.
18668         * configure: Regenerate.
18670         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
18672 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
18674         PR debug/54793
18675         * final.c (need_profile_function): New variable.
18676         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
18677         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
18678         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
18679         notes, targetm.asm_out.function_prologue doesn't emit anything,
18680         HAVE_prologue and profiler should be emitted before prologue,
18681         set need_profile_function instead of emitting it.
18682         (final_scan_insn): If need_profile_function, emit
18683         profile_function on the first NOTE_INSN_BASIC_BLOCK or
18684         NOTE_INSN_FUNCTION_BEG note.
18686 2013-02-01  Richard Henderson  <rth@redhat.com>
18688         * config/rs6000/rs6000.md (smulditi3): New.
18689         (umulditi3): New.
18691         * config/alpha/alpha.md (umulditi3): New.
18693 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
18695         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
18696         (ASM_OUTPUT_ALIGNED_LOCAL): New.
18698 2013-02-01  Richard Biener  <rguenther@suse.de>
18700         PR tree-optimization/56113
18701         * tree-ssa-structalias.c (label_visit): Reduce work for
18702         single-predecessor nodes.
18704 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
18706         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
18707         range isn't testing for zero.
18709 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
18711         PR middle-end/56113
18712         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
18714 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
18715             Nick Clifton  <nickc@redhat.com>
18717         * config/v850/constraints.md (Q): Define as a memory constraint.
18718         * config/v850/predicates.md (label_ref_operand): New predicate.
18719         (e3v5_shift_operand): New predicate.
18720         (ior_operator): New predicate.
18721         * config/v850/t-v850: Add e3v5 multilib.
18722         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
18723         (v850_gen_movdi): Prototype.
18724         * config/v850/v850.c: Add support for e3v5 architecture.
18725         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
18726         TARGET_V850E_UP.
18727         (construct_save_jarl): Add e3v5 long JARL support.
18728         (v850_adjust_insn_length): New function.  Adjust length of call
18729         insns when using e3v5 instructions.
18730         (v850_gen_movdi): New function: Generate instructions to move a
18731         DImode value.
18732         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
18733         (CPP_SPEC): Define __v850e3v5__ as appropriate.
18734         (TARGET_USE_FPU): Enable for e3v5.
18735         (CONST_OK_FOR_W): New macro.
18736         (ADJUST_INSN_LENGTH): Define.
18737         * config/v850/v850.md (UNSPEC_LOOP): Define.
18738         (attr cpu): Add v850e3v5.
18739         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
18740         (movdi): New pattern.
18741         (movdi_internal): New pattern.
18742         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
18743         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
18744         (cstoresf4): Likewise.
18745         (cstoredf4): Likewise.
18746         (insv): New pattern.
18747         (rotlso3_a): New pattern.
18748         (rotlsi3_b): New pattern
18749         (rotlsi3_v850e3v5): New pattern.
18750         (doloop_begin): New pattern.
18751         (fix_loop_counter): New pattern.
18752         (doloop_end): New pattern.
18753         (branch_normal): Add e3v5 long branch support.
18754         (branch_invert): Likewise.
18755         (branch_z_normal): Likewise.
18756         (branch_z_invert): Likewise.
18757         (branch_nz_normal): Likewise.
18758         (branch_nz_invert): Likewise.
18759         (call_internal_short): Add e3v5 register-indirect JARL support.
18760         (call_internal_long): Likewise.
18761         (call_value_internal_short): Likewise.
18762         (call_value_internal_long): Likewise.
18763         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
18764         (mloop): New option.
18765         * config.gcc: Add support for configuring v840e3v5 target.
18766         * doc/invoke.texi: Document new v850 specific command line options.
18768 2013-01-31  Paul Koning  <ni1d@arrl.net>
18770         PR debug/55059
18771         PR debug/54508
18772         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
18773         children if parent is a class.
18774         (prune_unused_types_prune): Don't add DW_AT_declaration.
18776 2013-01-31  Richard Biener  <rguenther@suse.de>
18778         PR tree-optimization/56157
18779         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
18780         match up operand with SLP child.
18782 2013-01-31  Jason Merrill  <jason@redhat.com>
18784         PR debug/54410
18785         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
18786         parameters the first time.
18787         (gen_scheduled_generic_parms_dies): Check completeness here.
18789 2013-01-31  Richard Biener  <rguenther@suse.de>
18791         PR middle-end/53073
18792         * common.opt (faggressive-loop-optimizations): New flag,
18793         enabled by default.
18794         * doc/invoke.texi (faggressive-loop-optimizations): Document.
18795         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
18796         infer_loop_bounds_from_undefined by it.
18798 2013-01-31  Richard Biener  <rguenther@suse.de>
18800         PR tree-optimization/56150
18801         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
18802         visit virtual operands.
18803         (find_uses_to_rename_bb): Likewise.
18805 2013-01-31  Richard Biener  <rguenther@suse.de>
18807         PR tree-optimization/56150
18808         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
18809         mixed store non-store stmts.
18811 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
18813         PR sanitizer/55374
18814         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
18815         LIBASAN_EARLY_SPEC is defined.
18816         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
18817         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
18818         before %o.
18819         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
18821         PR c++/55742
18822         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
18823         invalid args instead of ICEing on it.
18824         (ix86_valid_target_attribute_tree): Return error_mark_node if
18825         ix86_valid_target_attribute_inner_p failed.
18826         (ix86_valid_target_attribute_p): Return false only if
18827         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
18828         target("default") attribute.
18829         (sorted_attr_string): Change argument from const char * to tree,
18830         merge in all target attribute arguments rather than just one.
18831         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
18832         instead of free.  Avoid using strcat.
18833         (ix86_mangle_function_version_assembler_name): Mangle
18834         target("default") as if no target attribute is present.  Adjust
18835         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
18836         instead of xmalloc and XDELETEVEC instead of free.
18837         (ix86_function_versions): Don't return true if one of the decls
18838         doesn't have target attribute.  If they don't and one of the decls
18839         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
18840         sorted_attr_string caller.  Use XDELETEVEC instead of free.
18841         (ix86_supports_function_versions): Remove.
18842         (make_name): Fix up formatting.
18843         (make_dispatcher_decl): Remove resolver_name and its initialization.
18844         Avoid leaking memory.
18845         (is_function_default_version): Return true if there is
18846         target("default") attribute rather than no target attribute at all.
18847         (make_resolver_func): Avoid leaking memory.
18848         (ix86_generate_version_dispatcher_body): Likewise.
18849         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
18850         * target.def (supports_function_versions): Remove.
18851         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
18852         * doc/tm.texi: Regenerated.
18854 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
18856         PR rtl-optimization/56144
18857         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
18858         for values with side effects.
18860 2013-01-30  Richard Biener  <rguenther@suse.de>
18862         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
18863         (sparseset_pop): Likewise.
18864         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
18865         to be able to use quick_push in the worker loop.
18867 2013-01-30  Marek Polacek  <polacek@redhat.com>
18869         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
18871 2013-01-30  Richard Biener  <rguenther@suse.de>
18873         PR lto/56147
18874         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
18876 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
18878         PR tree-optimization/56064
18879         * fixed-value.c (fixed_from_double_int): New function.
18880         * fixed-value.h (fixed_from_double_int): New prototype.
18881         (const_fixed_from_double_int): New static inline function.
18882         * fold-const.c (native_interpret_fixed): New static function.
18883         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
18884         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
18885         (native_encode_fixed): New static function.
18886         (native_encode_expr) <FIXED_CST>: Use it.
18887         (native_interpret_int): Move double_int worker code to...
18888         * double-int.c (double_int::from_buffer): ...this new static method.
18889         * double-int.h (double_int::from_buffer): Prototype it.
18891 2013-01-30  Richard Biener  <rguenther@suse.de>
18893         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
18894         New pointer-map and obstack.
18895         (init_alias_vars): Allocate pointer-map and obstack.
18896         (delete_points_to_sets): Free them.
18897         (find_what_var_points_to): Cache result.
18898         (find_what_p_points_to): Adjust for changed interface of
18899         find_what_var_points_to.
18900         (compute_points_to_sets): Likewise.
18901         (ipa_pta_execute): Likewise.
18903 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
18905         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
18906         * configure: Regenerate.
18907         * config.in: Regenerate.
18908         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
18909         #nobits/#progbits if supported.
18911 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
18913         PR target/56121
18914         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
18915         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
18916         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
18918 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18920         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
18921         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
18923 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18925         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
18926         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
18928 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18930         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
18931         declaration.
18932         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
18933         * config/arm/cortex-a7.md: New bypasses using
18934         arm_mac_accumulator_is_result.
18936 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18938         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
18939         (cortex_a7_neon_mla): Likewise.
18940         (cortex_a7_fpfmad): New reservation.
18941         (cortex_a7_fpmacs): Use ffmas and update required units.
18942         (cortex_a7_fpmuld): Update required units and latency.
18943         (cortex_a7_fpmacd): Likewise.
18944         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18945         (cortex_a7_neon). Likewise.
18946         (bypass) Update participating units.
18948 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
18950         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18951         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18952         from fmac to ffma.
18953         * config/arm/vfp11.md (vfp_farith): Use ffmas.
18954         (vfp_fmul): Use ffmad.
18955         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18956         (cortex_r4_fmacd): Use ffmad.
18957         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18958         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
18959         (cortex_a9_fmacd): Use ffmad.
18960         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18961         (cortex_a8_vfp_macd): Use ffmad.
18962         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18963         (cortex_a5_fpmacd): Use ffmad.
18964         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18965         (cortex_a15_vfp_macd): Use ffmad.
18966         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18968 2013-01-29  Jason Merrill  <jason@redhat.com>
18970         PR libstdc++/54314
18971         * varasm.c (default_assemble_visibility): Don't warn about
18972         visibility on artificial decls.
18974 2013-01-29  Richard Biener  <rguenther@suse.de>
18976         PR tree-optimization/56113
18977         * tree-ssa-structalias.c (equiv_class_lookup): Also return
18978         the bitmap leader.
18979         (label_visit): Free duplicate bitmaps and record the leader instead.
18980         (perform_var_substitution): Adjust.
18982 2013-01-29  Richard Biener  <rguenther@suse.de>
18984         PR tree-optimization/55270
18985         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18986         the CFG, schedule loops for fixup.
18988 2013-01-29  Nick Clifton  <nickc@redhat.com>
18990         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18991         SP_REG.
18993 2013-01-28  Leif Ekblad  <leif@rdos.net>
18995         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18996         * config/i386/i386.h (TARGET_RDOS): New macro.
18997         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18998         * config/i386/i386.c (ix86_option_override_internal): For 64bit
18999         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
19000         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
19001         DEFAULT_LARGE_SECTION_THRESHOLD.
19002         * config/i386/i386.md (R14_REG, R15_REG): New constants.
19003         * config/i386/rdos.h: New file.
19004         * config/i386/rdos64.h: New file.
19006 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
19008         PR other/54814
19009         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
19010         TEST_HARD_REG_BIT.
19012 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
19014         PR rtl-optimization/56117
19015         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
19016         call cselib_lookup_from_insn on the MEM before calling
19017         add_insn_mem_dependence.
19019 2013-01-28  Richard Biener  <rguenther@suse.de>
19021         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
19022         to a stmt that didn't have one.
19023         (copy_phis_for_bb): Likewise for PHI arguments.
19024         (copy_debug_stmt): Likewise for debug stmts.
19026 2013-01-28  Richard Biener  <rguenther@suse.de>
19028         PR tree-optimization/56034
19029         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
19030         (partition_builtin_p): Adjust.
19031         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
19032         it is the last partition.
19033         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
19034         up the vertex for the definition.
19035         (classify_partition): Classify whether a partition is a
19036         PKIND_REDUCTION, thus has uses outside of the loop.
19037         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
19038         Merge all PKIND_REDUCTION partitions into the last partition.
19039         (tree_loop_distribution): Seed partitions from reductions as well.
19041 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
19043         PR tree-optimization/56125
19044         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
19045         pow(x,c) into sqrt(x) * powi(x, n/2) or
19046         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
19047         optimizing for size.
19048         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
19049         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
19050         integer.
19052         PR tree-optimization/56094
19053         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
19054         to UNKNOWN_LOCATION while gimplifying expr.
19056 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
19058         PR target/56114
19059         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
19060         operand 0 in movabs insn template for -masm=intel asm alternative.
19061         (*movabs<mode>_2): Ditto for operand 1.
19063 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
19065         PR target/54663
19066         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
19067         of microblaze-c.o
19069 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19071         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
19072         tm_file.
19074 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19076         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
19077         Undef to avoid warning.
19079 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
19081         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
19082         * configure: Regenerate.
19084 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
19086         PR tree-optimization/56098
19087         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
19088         for stmts with volatile ops.
19089         (cond_store_replacement): Don't optimize if assign has volatile ops.
19090         (cond_if_else_store_replacement_1): Don't optimize if either
19091         then_assign or else_assign have volatile ops.
19092         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
19093         volatile ops.
19095 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
19097         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
19099 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
19101         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
19102         missing ':' in asm example.
19104 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
19106         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
19107         entries into lane and laneq entries.
19108         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
19109         Remove AdvSIMD scalar modes.
19110         (aarch64_sq<r>dmulh_laneq<mode>): New.
19111         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
19112         modes.
19113         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
19114         builtin implementations to relfect changes in RTL in aarch64-simd.md.
19115         * config/aarch64/iterators.md (VCOND): New.
19116         (VCONQ): New.
19118 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
19120         PR target/54222
19121         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
19122         Add NULL LIBNAME argument to existing definitions.
19123         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
19124         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
19125         * config/avr/avr.c (DEF_BUILTIN): Same.
19126         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
19127         (avr_expand_builtin): Expand to a vanilla call if a libgcc
19128         implementation is available (DECL_ASSEMBLER_NAME is set).
19129         (avr_fold_absfx): New static function.
19130         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
19131         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
19132         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
19133         AVR_BUILTIN_ABSLLK.
19134         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
19135         (abshk, absk, abslk, absllk): Provide as static inline functions.
19137 2013-01-25  Marek Polacek  <polacek@redhat.com>
19139         PR tree-optimization/56035
19140         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
19142 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
19144         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
19145         (*movtf_internal_rex64): Add (!o,C) alternative
19146         (*movxf_internal_rex64): Ditto.
19147         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
19149 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
19151         * doc/invoke.texi: fix typo.
19152         * doc/objc.texi: fix typo.
19154 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
19156         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
19157         for the first two alternatives.
19159 2013-01-24  Diego Novillo  <dnovillo@google.com>
19161         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
19162         (ggc-zone.o): Remove.
19163         * configure.ac: Remove option --with-gc.
19164         * configure: Re-generate.
19165         * doc/install.texi: Remove documentation for --with-gc.
19166         * gengtype.c (write_enum_defn): Remove.  Update all users.
19167         (write_Types_process_field): Remove generation of gt_e_* argument.
19168         (output_type_enum): Remove.  Update all users.
19169         (write_enum_defn): Remove.  Update all users.
19170         (enum alloc_zone): Remove.  Update all users.
19171         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
19172         * ggc-common.c (ggc_splay_alloc): Remove first argument.
19173         Update all callers.
19174         (struct ptr_data): Remove field TYPE.  Update all users.
19175         (gt_pch_note_object): Remove argument TYPE.  Update all users.
19176         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
19177         Update all users.
19178         * ggc-none.c (ggc_alloc_typed_stat): Remove.
19179         (struct alloc_zone): Remove.
19180         (ggc_internal_alloc_zone_stat): Remove.
19181         (ggc_internal_cleared_alloc_zone_stat): Remove.
19182         * ggc-page.c (ggc_alloc_typed_stat): Remove.
19183         (ggc_pch_count_object): Remove last argument.  Update all users.
19184         (ggc_pch_alloc_object): Remove last argument.  Update all users.
19185         (struct alloc_zone): Remove.
19186         * ggc-zone.c: Remove.
19187         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
19188         (struct alloc_zone): Remove.
19189         (ggc_alloc_typed_stat): Remove.
19190         (ggc_alloc_typed): Remove.
19191         (ggc_splay_alloc): Remove first argument.
19192         (rtl_zone): Remove.  Update all users.
19193         (tree_zone): Remove.  Update all users.
19194         (tree_id_zone): Remove.  Update all users.
19195         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
19196         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
19197         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
19198         * tree-ssanames.c: Remove references to zone allocator in comments.
19200 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
19202         * config/avr/avr.c (avr_out_fract): Make register numbers that
19203         might be outside of source operand signed.
19205 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
19207         * config/i386/constraints.md (Yf): New constraint.
19208         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
19209         of f constraint to conditionaly disable x87 register preferences.
19210         (*movdf_internal): Ditto.
19211         (*movsf_internal): Ditto.
19213 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
19215         PR inline-asm/55934
19216         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
19217         that have operands with impossible constraints.
19218         Add a FIXME for a speed-up opportunity.
19219         * lra-constraints.c (process_alt_operands): Verify that a class
19220         selected from constraints on asms is valid for the operand mode.
19221         (curr_insn_transform): Remove incorrect comment.
19223 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
19225         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
19226         TOC operand is a valid symbol ref in the constant pool.
19228 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19230         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
19232 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
19234         PR target/54222
19235         * config/avr/stdfix.h: New file.
19236         * t-avr (stdfix-gcc.h): New rule to build it.
19237         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
19239 2013-01-23  Kostya Serebryany  <kcc@google.com>
19241         * config/darwin.h: remove dependency on
19242         CoreFoundation (asan on Mac OS).
19244 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
19246         PR target/49069
19247         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
19248         instead of cmpdi_operand for first comparison operand.
19249         Don't assert that comparison operands aren't both constants.
19251 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
19253         * doc/install.texi (Downloading the Source): Update references to
19254         downloading separate components.
19256 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
19258         * doc/extend.texi (__int128): Improve grammar.
19260 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
19262         PR target/56028
19263         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
19264         alternative to (o,r).
19265         (*movdi_internal_rex64): Remove (!o,n) alternative.
19266         (DImode immediate->memory splitter): Remove.
19267         (DImode immediate->memory peephole2): Remove.
19268         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
19269         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
19270         alternative to (!o,*r).
19271         (*movtf_internal_sse): New pattern.
19272         (*movxf_internal_rex64): New pattern.
19273         (*movxf_internal): Disable for TARGET_64BIT.
19274         (*movdf_internal_rex64): Remove (!o,F) alternative.
19276 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
19278         PR middle-end/56074
19279         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
19280         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
19281         * tree-vect-loop-manip.c (find_loop_location): Also ignore
19282         stmt locations where LOCATION_LOCUS of the stmt location is
19283         UNKNOWN_LOCATION or BUILTINS_LOCATION.
19285         PR target/55686
19286         * config/i386/i386.md (UNSPEC_STOS): New.
19287         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
19288         *strsetqi_1): Add UNSPEC_STOS.
19290 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
19292         PR c++/56067
19293         * doc/invoke.texi: Remove left over -Wsynth example.
19295 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
19297         PR tree-optimization/56051
19298         * fold-const.c (fold_binary_loc): Don't fold
19299         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
19300         a narrowing conversion, or widening conversion from signed
19301         to unsigned.
19303 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
19305         PR rtl-optimization/56023
19306         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
19307         dependent on debug instruction.
19309 2013-01-21  Martin Jambor  <mjambor@suse.cz>
19311         PR middle-end/56022
19312         * function.c (allocate_struct_function): Call
19313         invoke_set_current_function_hook earlier.
19315 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
19317         * reload1.c (init_reload): Only initialize reload_obstack
19318         during the first call.
19320 2013-01-21  Marek Polacek  <polacek@redhat.com>
19322         * cfgloop.c (verify_loop_structure): Fix up grammar.
19324 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
19326         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
19327         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
19329 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19331         PR target/56058
19332         * config/arm/marvell-pj4.md: Update copyright year.
19333         Fix up use of alu to alu_reg and simple_alu_imm.
19335 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
19337         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
19339 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
19341         PR target/55433
19342         * lra-constraints.c (curr_insn_transform): Don't reuse original
19343         insn for secondary memory move when memory mode should be different.
19345 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19347         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
19348         atomic_storedi_1): New patterns.
19350 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
19352         btver2 pipeline descriptions.
19353         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
19354         descriptions.
19355         * config/i386/i386.md (btver2_decode): New type attributes.
19356         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
19357         type attributes.
19358         * config/i386/btver2.md: New file describing btver2 pipelines.
19360 2013-01-19  Andrew Pinski  <apinski@cavium.com>
19362         PR tree-optimization/52631
19363         * tree-ssa-sccvn (visit_use): Before looking up the original
19364         statement, try looking up the simplified expression.
19366 2013-01-19  Anthony Green  <green@moxielogic.com>
19368         * config/moxie/moxie.c (moxie_expand_prologue): Set
19369         current_function_static_stack_size.
19371 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
19373         PR tree-optimization/56029
19374         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
19375         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
19377 2013-01-18  Sharad Singhai  <singhai@google.com>
19379         PR tree-optimization/55995
19380         * dumpfile.c (dump_loc): Print location only if available.
19381         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
19383 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
19385         PR target/55433
19386         * lra-constraints.c (curr_insn_transform): Reuse original insn for
19387         secondary memory move.
19388         (inherit_reload_reg): Use rclass instead of cl for
19389         check_secondary_memory_needed_p.
19391 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
19393         PR middle-end/56015
19394         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
19395         the case where writing real complex part of target modifies op1.
19397 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
19399         * config/aarch64/aarch64-simd.md
19400         (aarch64_vcond_internal<mode>): Handle unordered cases.
19401         * config/aarch64/iterators.md (v_cmp_result): New.
19403 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
19404             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19406         * config/arm/marvell-pj4.md: New file.
19407         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
19408         * config/arm/arm.md (generic_sched): Add marvell_pj4.
19409         (generic_vfp): Likewise.
19410         * config/arm/arm-cores.def: Add marvell-pj4.
19411         * config/arm/arm-tune.md: Regenerate.
19412         * config/arm/arm-tables.opt: Regenerate.
19413         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
19414         * doc/invoke.texi: Document marvell-pj4.
19416 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
19418         * config/aarch64/arm_neon.h: Map scalar types to standard types.
19420 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
19422         PR debug/54114
19423         PR debug/54402
19424         PR debug/49888
19425         * var-tracking.c (negative_power_of_two_p): New.
19426         (global_get_addr_cache, local_get_addr_cache): New.
19427         (get_addr_from_global_cache, get_addr_from_local_cache): New.
19428         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
19429         heading comment.
19430         (vt_stack_offset_p): Remove.
19431         (vt_canon_true_dep): Always canonicalize loc's address.
19432         (clobber_overlapping_mems): Make sure we have a MEM.
19433         (local_get_addr_clear_given_value): New.
19434         (val_reset): Clear local cached entries.
19435         (compute_bb_dataflow): Create and release the local cache.
19436         Disable duplicate MEMs clobbering.
19437         (emit_notes_in_bb): Clobber MEMs likewise.
19438         (vt_emit_notes): Create and release the local cache.
19439         (vt_initialize, vt_finalize): Create and release the global
19440         cache, respectively.
19441         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
19443 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
19445         PR libmudflap/53359
19446         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
19447         not found in the symtab.
19449 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
19451         PR debug/56006
19452         PR rtl-optimization/55547
19453         PR rtl-optimization/53827
19454         PR debug/53671
19455         PR debug/49888
19456         * alias.c (offset_overlap_p): New, factored out of...
19457         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
19458         the conservative special case for symbolic constants.  Don't
19459         adjust zero sizes on alignment.
19461 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
19463         PR rtl-optimization/52573
19464         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
19465         REG_UNUSED for the same register.
19467 2013-01-17  Richard Biener  <rguenther@suse.de>
19468             Marek Polacek  <polacek@redhat.com>
19470         PR rtl-optimization/55833
19471         * loop-unswitch.c (unswitch_loops): Move loop verification...
19472         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
19473         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
19474         Set it to true when we're removing a loop from hierarchy tree in
19475         an irreducible region.
19476         (fix_bb_placements): Adjust caller.
19477         (fix_loop_placements): Likewise.
19479 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
19481         * config/avr/builtins.def (DEF_BUILTIN): Factor out
19482         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
19483         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
19484         Remove ID.  Adjust comments.
19485         * config/avr/avr-c.c (avr_builtin_name): Remove.
19486         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
19487         * config/avr/avr.c (avr_tolower): New static function.
19488         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
19489         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
19490         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
19491         default expansion.
19493 2013-01-17  Jan Hubicka  <jh@suse.cz>
19495         PR tree-optimization/55273
19496         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
19498 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
19500         PR target/55981
19501         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
19502         store through atomic_store<mode>_1.
19503         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
19505 2013-01-17  Martin Jambor  <mjambor@suse.cz>
19507         PR tree-optimizations/55264
19508         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
19509         for virtual methods.
19510         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
19511         virtual methods before inlining is over.
19512         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
19513         virtual functions.
19514         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
19515         non-virtual.
19517 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
19519         PR rtl-optimization/56005
19520         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
19521         pending reads for prefetch.
19523 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
19525         * config/aarch64/aarch64.md
19526         (*cstoresi_neg_uxtw): New pattern.
19527         (*cmovsi_insn_uxtw): New pattern.
19528         (*<optab>si3_uxtw): New pattern.
19529         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
19530         (*<optab>si3_insn_uxtw): New pattern.
19531         (*bswapsi2_uxtw): New pattern.
19533 2013-01-16  Richard Biener  <rguenther@suse.de>
19535         * tree-inline.c (tree_function_versioning): Remove set but
19536         never used variable.
19538 2013-01-16  Richard Biener  <rguenther@suse.de>
19540         PR tree-optimization/55964
19541         * tree-flow.h (rename_variables_in_loop): Remove.
19542         (rename_variables_in_bb): Likewise.
19543         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
19544         (copy_loop_before): Adjust and delete update-ssa status.
19545         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
19546         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
19547         (rename_variables_in_loop): Remove.
19548         (slpeel_update_phis_for_duplicate_loop): Likewise.
19549         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
19550         use available cfg machinery instead of duplicating it.
19551         Update PHI nodes and perform poor-mans SSA update here.
19552         (slpeel_tree_peel_loop_to_edge): Adjust.
19554 2013-01-16  Richard Biener  <rguenther@suse.de>
19556         PR tree-optimization/54767
19557         PR tree-optimization/53465
19558         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
19559         (vrp_visit_phi_node): For PHI arguments coming via backedges
19560         drop all symbolical range information.
19561         (execute_vrp): Compute backedges.
19563 2013-01-16  Richard Biener  <rguenther@suse.de>
19565         * doc/install.texi: Update CLooG and ISL requirements to
19566         0.18.0 and 0.11.1.
19568 2013-01-16  Christian Bruel  <christian.bruel@st.com>
19570         PR target/55301
19571         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
19572         (broken_move): Handle UNSPECV_SP_SWITCH_B.
19573         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
19575 2013-01-16  DJ Delorie  <dj@redhat.com>
19577         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
19578         (UNSPECV_SP_SWITCH_E): New.
19579         (sp_switch_1): Change to an unspec.
19580         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
19581         replace $r15.
19583 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
19585         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
19586         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
19587         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
19588         (expand_mem_thread_fence): Ditto.
19589         (expand_mem_signal_fence): Ditto.
19590         (expand_atomic_load): Ditto.
19591         (expand_atomic_store): Ditto.
19593 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
19595         PR rtl-optimization/55547
19596         PR rtl-optimization/53827
19597         PR debug/53671
19598         PR debug/49888
19599         * alias.c (memrefs_conflict_p): Set sizes to negative after
19600         AND adjustments.
19602 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
19604         PR target/55940
19605         * function.c (thread_prologue_and_epilogue_insns): Always
19606         add crtl->drap_reg to set_up_by_prologue.set, even if
19607         stack_realign_drap is false.
19609 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
19611         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
19612         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
19613         *call): Fix indention.
19615 2013-01-15  Tom de Vries  <tom@codesourcery.com>
19617         PR target/55876
19618         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
19619         Update comment.
19621 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
19623         PR rtl-optimization/55153
19624         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
19626 2013-01-15  Martin Jambor  <mjambor@suse.cz>
19628         PR tree-optimization/55920
19629         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
19630         accesses as grp_to_be_debug_replaced.
19632 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
19634         PR tree-optimization/55920
19635         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
19636         there is non-useless type conversion needed from debug rhs to lhs,
19637         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
19639 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
19640             Mikael Pettersson  <mikpe@it.uu.se>
19642         PR target/43961
19643         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
19644         Thumb.
19645         (ASM_OUTPUT_CASE_LABEL): Remove.
19646         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
19647         * final.c (shorten_branches): Update alignment of labels before
19648         jump tables if CASE_VECTOR_SHORTEN_MODE.
19650 2013-01-15  Richard Biener  <rguenther@suse.de>
19652         PR bootstrap/55961
19653         * system.h: Do not include gmp.h for building host tools.
19655 2013-01-15  Richard Biener  <rguenther@suse.de>
19657         PR middle-end/55882
19658         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
19659         account for bitpos when computing alignment.
19661 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
19663         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
19664         (ix86_target_macros_internal): Likewise.
19666         * config/i386/i386.c (m_CORE2I7): Removed.
19667         (m_CORE_HASWELL): New macro.
19668         (m_CORE_ALL): Likewise.
19669         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
19670         (initial_ix86_arch_features): Likewise.
19671         (processor_target_table): Initializations for Core avx2.
19672         (cpu_names): New names "core-avx2".
19673         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
19674         PROCESSOR_CORE_HASWELL.
19675         (ix86_issue_rate): New case.
19676         (ia32_multipass_dfa_lookahead): Likewise.
19677         (ix86_sched_init_global): Likewise.
19679         * config/i386/i386.h (TARGET_HASWELL): New macro.
19680         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
19681         (processor_type): New PROCESSOR_HASWELL.
19683 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
19685         PR tree-optimization/55955
19686         * tree-vect-loop.c (vectorizable_reduction): Give up early on
19687         *SHIFT_EXPR and *ROTATE_EXPR codes.
19689         PR tree-optimization/48766
19690         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
19691         -ftrapv disable -fwrapv.
19693 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
19695         PR target/55974
19696         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
19697         etc. to 1 and not to __flash.
19698         Use LL suffix for __INT24_MAX__ with -mint8.
19699         Use ULL suffix for __UINT24_MAX__ with -mint8.
19701 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
19703         * config/avr/avr-arch.h
19704         (struct base_arch_s): Use typedef avr_arch_t instead.
19705         (struct arch_info_s): Use typedef avr_arch_info_t instead.
19706         (struct mcu_type_s): Use typedef avr_mcu_t instead.
19707         * config/avr/avr.c: Same.
19708         * config/avr/avr-devices.c: Same.
19709         * config/avr/driver-avr.c: Same.
19710         * config/avr/gen-avr-mmcu-texi.c: Same.
19711         * config/avr/avr-mcus.def: Adjust comment.
19713 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
19715         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
19716         * config/aarch64/iterators.md (VALLDI): New.
19718 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
19719             Andi Kleen  <ak@linux.intel.com>
19721         PR target/55948
19722         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
19723         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
19724         memmodel flag.
19726 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
19728         * config/avr/avr-stdint.h: Remove trailing blanks.
19729         * config/avr/avr-log.h: Same.
19730         * config/avr/avr-arch.h: Same.
19731         * config/avr/avr-devices.c: Same.
19732         * config/avr/avr-dimode.md: Same.
19733         * config/avr/predicates.md: Same.
19734         * config/avr/avr-c.c: Same.  And fix typo.
19736         * config/avr/avr-protos.h: Same.  And:
19737         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
19738         (init_cumulative_args): Rename to avr_init_cumulative_args.
19739         (expand_prologue): Rename to avr_expand_prologue.
19740         (expand_epilogue): Rename to avr_expand_epilogue.
19741         (adjust_insn_length): Rename to avr_adjust_insn_length.
19742         (notice_update_cc): Rename to avr_notice_update_cc.
19743         (final_prescan_insn): Rename to avr_final_prescan_insn.
19744         * config/avr/avr.c: Same.
19745         * config/avr/avr.h: Same.
19746         * config/avr/avr.md: Remove trailing blanks.
19747         (prologue): Use avr_expand_prologue.
19748         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
19750 2013-01-14  Richard Biener  <rguenther@suse.de>
19752         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
19753         verify_location, collect_subblocks): New functions.
19754         (verify_gimple_in_cfg): Verify that locations only reference
19755         BLOCKs in the functions BLOCK tree.
19757 2013-01-14  Richard Biener  <rguenther@suse.de>
19759         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
19760         PHI argument.
19761         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
19762         unshare reference.
19763         (insert_out_of_ssa_copy_on_edge): Likewise.
19764         (rewrite_close_phi_out_of_ssa): Likewise.
19765         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
19766         debug expressions.
19767         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
19768         propagated constants.
19769         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
19770         can not be shared.
19772 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
19774         * config/avr/avr-modes.def: Add GPL copyright notice.
19776 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
19778         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
19779         MEMMODEL_MASK to determine memory model.
19780         (atomic_store<mode>): Ditto from operands[2].
19781         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
19783 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
19785         PR fortran/55935
19786         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
19787         (fold_gimple_assign): Don't call unshare_expr here.
19788         (fold_ctor_reference): Call unshare_expr.
19790 2013-01-13  Terry Guo  <terry.guo@arm.com>
19792         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
19793         * doc/fragments.texi: Document MULTILIB_REUSE.
19794         * gcc.c (multilib_reuse): New internal spec.
19795         (set_multilib_dir): Also search multilib from multilib_reuse.
19796         * genmultilib (tmpmultilib3): Refactor code.
19797         (tmpmultilib4): Ditto.
19798         (multilib_reuse): New multilib argument.
19800 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
19802         * Makefile.in: Update copyright.
19804 2013-01-12  Tom de Vries  <tom@codesourcery.com>
19806         PR middle-end/55890
19807         * calls.c (expand_call): Check if arg_nr is valid.
19809 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
19811         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
19812         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
19813         documentation.  Add missing '__' in front of
19814         __builtin_ia32_packssdw256.
19816 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19818         PR target/55719
19819         * config/s390/s390.c (s390_preferred_reload_class): Do not return
19820         NO_REGS for larl operands.
19821         (s390_reload_larl_operand): Use s390_load_address instead of
19822         emit_move_insn.
19824 2013-01-11  Richard Biener  <rguenther@suse.de>
19826         * tree-cfg.c (verify_node_sharing_1): Split out from ...
19827         (verify_node_sharing): ... here.
19828         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
19830 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
19832         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
19833         Substitute TREECHECKING.
19834         * configure: Regenerate.
19835         * Makefile.in (TREECHECKING): New.
19837 2013-01-11  Richard Guenther  <rguenther@suse.de>
19839         PR tree-optimization/44061
19840         * tree-vrp.c (extract_range_basic): Compute zero as
19841         value-range for __builtin_constant_p of function parameters.
19843 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
19845         Update copyright years.
19847 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
19849         PR rtl-optimization/55672
19850         * lra-eliminations.c (mark_not_eliminable): Permit addition with
19851         const to be eliminable.
19853 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
19855         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
19856         * configure: Regenerate.
19858 2013-01-10  Richard Biener  <rguenther@suse.de>
19860         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
19862 2013-01-10  Richard Biener  <rguenther@suse.de>
19864         PR bootstrap/55792
19865         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
19866         locations for virtual PHI arguments.
19867         (rewrite_update_phi_arguments): Likewise.
19869 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
19871         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
19872         on to assembler.
19874 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
19876         PR tree-optimization/55921
19877         * tree-complex.c (expand_complex_asm): New function.
19878         (expand_complex_operations_1): Call it for GIMPLE_ASM.
19880 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19882         PR target/55718
19883         * config/s390/s390.c (s390_symref_operand_p)
19884         (s390_loadrelative_operand_p): Merge the two functions.
19885         (s390_check_qrst_address, print_operand_address): Add parameters
19886         to s390_loadrelative_operand_p invokation.
19887         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
19888         (s390_reload_larl_operand, s390_secondary_reload): Use
19889         s390_loadrelative_operand_p instead of s390_symref_operand_p.
19890         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
19892 2013-01-09  Mike Stump  <mikestump@comcast.net>
19894         * dse.c (record_store): Remove unnecessary assert.
19896 2013-01-09  Jan Hubicka  <jh@suse.cz>
19898         PR tree-optimization/55569
19899         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
19900         * cfgloop.h (scale_loop_profile): Likewise.
19902 2013-01-09  Jan Hubicka  <jh@suse.cz>
19904         PR lto/45375
19905         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
19906         functions.
19907         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
19909 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
19911         PR middle-end/55114
19912         * expr.h (maybe_emit_group_store): Declare.
19913         * expr.c (maybe_emit_group_store): New function.
19914         * builtins.c (expand_builtin_int_roundingfn): Call it.
19915         (expand_builtin_int_roundingfn_2): Likewise.
19917 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
19919         PR rtl-optimization/55829
19920         * lra-constraints.c (match_reload): Add code for absent output.
19921         (curr_insn_transform): Add code for reloads of matched inputs
19922         without output.
19924 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
19926         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
19927         attribute of movddup insn to DF.
19928         (*vec_interleave_lowv2df): Ditto.
19929         (vec_dupv2df): Ditto.
19931 2013-01-09  Jan Hubicka  <jh@suse.cz>
19933         PR tree-optimiation/55875
19934         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
19935         EVERY_ITERATION parameter.
19936         (number_of_iterations_exit): Check if exit is executed every iteration.
19937         (idx_infer_loop_bounds): Similarly here.
19938         (n_of_executions_at_most): Simplify
19939         to only test for cases where statement is dominated by the
19940         particular bound; handle correctly the "postdominance" test.
19941         (scev_probably_wraps_p): Use max loop iterations info
19942         as a global bound first.
19944 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
19945             Nick Clifton  <nickc@redhat.com>
19947         * config/v850/v850.md (cbranchsf4): New pattern.
19948         (cstoresf4): New pattern.
19949         (cbranchdf4): New pattern.
19950         (cstoredf4): New pattern.
19951         (movsicc): Disallow floating point comparisons.
19952         (cmpsf_le_insn): Fix order of operators.
19953         (cmpsf_lt_insn): Likewise.
19954         (cmpsf_eq_insn): Likewise.
19955         (cmpdf_le_insn): Likewise.
19956         (cmpdf_lt_insn): Likewise.
19957         (cmpdf_eq_insn): Likewise.
19958         (cmpsf_ge_insn): Use LE comparison.
19959         (cmpdf_ge_insn): Likewise.
19960         (cmpsf_gt_insn): Use LT comparison.
19961         (cmpdf_gt_insn): Likewise.
19962         (cmpsf_ne_insn): Delete pattern.
19963         (cmpdf_ne_insn): Delete pattern.
19964         * config/v850/v850.c (v850_gen_float_compare): Use
19965         gen_cmpdf_eq_insn for NE comparison.
19966         (v850_float_z_comparison_operator)
19967         (v850_float_nz_comparison_operator): Move from here ...
19968         * config/v850/predicates.md: ... to here.  Move GT and GE
19969         comparisons into v850_float_z_comparison_operator.
19970         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19971         Delete prototype.
19972         (v850_float_nz_comparison_operator): Likewise.
19974 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19976         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19977         with calls to gen_insvsi/gen_insvdi.
19979 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
19981         * config/i386/i386.c (initial_ix86_tune_features): Set up
19982         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19984 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
19985             Jakub Jelinek  <jakub@redhat.com>
19987         PR tree-optimization/48189
19988         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19989         If nitercst is 0, don't predict the exit edge.
19991 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19993         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19994         in asm_fprintf with reg_names.
19995         (aarch64_print_operand_address): Likewise.
19996         (aarch64_return_addr): Likewise.
19997         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19999 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20001         * config/pa/pa.h (VAL_U6_BITS_P): Define.
20002         (INT_U6_BITS): Likewise.
20003         * config/pa/predicates.md (uint6_operand): New predicate.
20004         (shift5_operand, shift6_operand): Likewise.
20005         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
20006         arith32_operand.
20007         (lshrdi3): Use shift6_operand.
20008         (shrpsi4, shrpdi4): New insn patterns.
20009         (extzv): Delete expander.
20010         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
20011         predicates in unamed zero extract patterns.  Tighten common constraint.
20012         (extv): Delete expander.
20013         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
20014         predicates in unamed sign extract patterns.  Tighten common constraint.
20015         (insv): Delete expander.
20016         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
20017         predicates in unamed insert patterns.  Tighten common constraint.
20018         Change uint32_operand predicate to uint6_operand predicate in unamed
20019         DImode pattern to insert constant values of type 1...1xxxx.
20021 2013-01-04  Jan Hubicka  <jh@suse.cz>
20023         PR tree-optimization/55823
20024         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
20025         issue.
20027 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
20028             Uros Bizjak  <ubizjak@gmail.com>
20030         PR rtl-optimization/55845
20031         * df-problems.c (can_move_insns_across): Stop scanning at
20032         volatile_insn_p source instruction or give up if
20033         across_from .. across_to range contains any volatile_insn_p
20034         instructions.
20036 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
20038         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
20039         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
20040         Declare.
20041         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
20042         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
20044 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
20046         PR fortran/55341
20047         * asan.c (asan_clear_shadow): New function.
20048         (asan_emit_stack_protection): Use it.
20050 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
20052         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
20053         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
20054         with tab instead of space.
20056 2013-01-08  Nick Clifton  <nickc@redhat.com>
20058         * config/rl78/rl78.c (rl78_expand_prologue): Always select
20059         register bank 0 at the start of an interrupt handler.
20060         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
20061         MDBH registers.
20063 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
20065         * config/aarch64/aarch64-simd.md
20066         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
20067         (aarch64_simd_bsl): Likewise.
20068         (aarch64_vcond_internal<mode>): Likewise.
20069         (vcond<mode><mode>): Likewise.
20070         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
20071         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
20073 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
20075         * config/aarch64/aarch64-builtins.c
20076         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
20078 2013-01-08  Martin Jambor  <mjambor@suse.cz>
20080         PR debug/55579
20081         * tree-sra.c (analyze_access_subtree): Return true also after
20082         potentially creating a debug-only replacement.
20084 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
20086         PR middle-end/55890
20087         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
20089         PR tree-optimization/54120
20090         * tree-vrp.c (range_fits_type_p): Don't allow
20091         src_precision < precision from signed vr to unsigned_p
20092         if vr->min or vr->max is negative.
20093         (simplify_float_conversion_using_ranges): Test can_float_p
20094         against CODE_FOR_nothing.
20096 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
20097             Richard Biener  <rguenther@suse.de>
20099         PR middle-end/55851
20100         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
20101         types instead of just INTEGER_TYPE types.
20103 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
20105         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
20106         TF_SIZE): Define.
20108 2013-01-07  Steve Ellcey  <sellcey@mips.com>
20110         PR target/42661
20111         * config/mips/mips.opt: Change mad to mmad to match documentation.
20113 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
20115         PR target/55897
20116         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
20117         .progmemx.data now.
20119 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
20121         PR target/55897
20122         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
20123         (avr_addrspace_t): Add .section_name field.
20124         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
20125         array size.
20126         (avr_addrspace): Same.  Initialize .section_name.  Remove last
20127         NULL entry.  Put __memx into .progmemx.data.
20128         (progmem_section_prefix): Remove.
20129         (avr_asm_init_sections): No need to initialize progmem_section.
20130         (avr_asm_named_section): Use avr_addrspace[].section_name to get
20131         section name prefix.
20132         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
20133         retrieve the progmem section.
20134         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
20135         boundary to run over avr_addrspace[].
20136         (avr_register_target_pragmas): Ditto.
20138 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
20140         * varasm.c (output_constant_def_contents): For asan_protect_global
20141         protected strings, adjust DECL_ALIGN if needed, before testing for
20142         anchored symbols.
20143         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
20144         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
20145         normal decls.
20146         (output_object_block): For asan protected decls, emit asan padding
20147         after their contents.
20148         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
20149         (asan_finish_file): Test it here instead.
20151 2013-01-07  Nick Clifton  <nickc@redhat.com>
20152             Matthias Klose  <doko@debian.org>
20153             Doug Kwan  <dougkwan@google.com>
20154             H.J. Lu  <hongjiu.lu@intel.com>
20156         PR driver/55470
20157         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
20159         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
20161         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
20163         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
20165         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
20167 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
20169         PR target/54461
20170         * doc/install.texi (Cross-Compiler-Specific Options): Document
20171         --with-avrlibc.
20173 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
20175         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
20176         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
20177         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
20178         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
20179         vqmovun_high_s64): Fix source operand number and update copyright.
20181 2013-01-07  Richard Biener  <rguenther@suse.de>
20183         PR middle-end/55890
20184         * gimple.h (gimple_call_builtin_p): New overload.
20185         * gimple.c (validate_call): New function.
20186         (gimple_call_builtin_p): Likewise.
20187         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20188         Use gimple_call_builtin_p.
20189         (find_func_clobbers): Likewise.
20190         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
20191         (strlen_optimize_stmt): Likewise.
20193 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
20195         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
20196         (vld1q_dup_*): Likewise.
20197         (vld1_*): Likewise.
20198         (vld1q_*): Likewise.
20199         (vld1_lane_*): Likewise.
20200         (vld1q_lane_*): Likewise.
20202 2013-01-07  Richard Biener  <rguenther@suse.de>
20204         * lto-streamer.h (LTO_minor_version): Bump to 2.
20206 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
20208         * config/aarch64/aarch64-protos.h
20209         (aarch64_const_double_zero_rtx_p): Rename to...
20210         (aarch64_float_const_zero_rtx_p): ...this.
20211         (aarch64_float_const_representable_p): New.
20212         (aarch64_output_simd_mov_immediate): Likewise.
20213         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
20214         move immediate case.
20215         * config/aarch64/aarch64.c
20216         (aarch64_const_double_zero_rtx_p): Rename to...
20217         (aarch64_float_const_zero_rtx_p): ...this.
20218         (aarch64_print_operand): Allow printing of new constants.
20219         (aarch64_valid_floating_const): New.
20220         (aarch64_legitimate_constant_p): Check for valid floating-point
20221         constants.
20222         (aarch64_simd_valid_immediate): Likewise.
20223         (aarch64_vect_float_const_representable_p): New.
20224         (aarch64_float_const_representable_p): Likewise.
20225         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
20226         (aarch64_output_simd_mov_immediate): New.
20227         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
20228         (*movdf_aarch64): Likewise.
20229         * config/aarch64/constraints.md (Ufc): New.
20230         (Y): call aarch64_float_const_zero_rtx.
20231         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
20233 2013-01-07  Richard Biener  <rguenther@suse.de>
20235         PR tree-optimization/55888
20236         PR tree-optimization/55862
20237         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
20238         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
20239         not if it is contained therein.
20241 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
20243         * config/avr/t-avr: Typo.
20245 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
20247         PR55243
20248         * config/avr/t-avr: Don't automatically rebuild
20249         $(srcdir)/config/avr/t-multilib
20250         $(srcdir)/config/avr/avr-tables.opt
20251         $(srcdir)/doc/avr-mmcu.texi
20252         (avr-mcus): New phony target to build them on request.
20253         (s-avr-mlib, s-avr-mmcu-texi): Remove.
20254         * avr/avr-mcus.def: Adjust comments.
20256 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
20258         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
20260 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
20262         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
20264 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
20266         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
20268 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
20270         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
20271         to generate profiling.
20272         * config/rs6000/aix64.h (LIB_SPEC): Same.
20274 2013-01-04  Andrew Pinski  <apinski@cavium.com>
20276         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
20277         New function.
20278         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
20280 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
20282         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
20283         unconditionally.
20284         (ix86_expand_move): Ditto.
20285         (ix86_zero_extend_to_Pmode): Ditto.
20286         (ix86_expand_call): Ditto.
20287         (ix86_expand_special_args_builtin): Ditto.
20288         (ix86_expand_builtin): Ditto.
20290 2013-01-04  Richard Biener  <rguenther@suse.de>
20292         PR tree-optimization/55862
20293         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
20294         translating them through PHI nodes.
20296 2013-01-04  Martin Jambor  <mjambor@suse.cz>
20298         PR tree-optimization/55755
20299         * tree-sra.c (sra_modify_assign): Do not check that an access has no
20300         children when trying to avoid producing a VIEW_CONVERT_EXPR.
20302 2013-01-04  Marek Polacek  <polacek@redhat.com>
20304         PR middle-end/55859
20305         * opts.c (default_options_optimization): Clarify error message.
20307 2013-01-04  Richard Biener  <rguenther@suse.de>
20309         PR middle-end/55863
20310         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
20311         reassociation.
20313 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20315         PR target/53789
20316         * config/pa/pa.md (movsi): Revert previous change.
20317         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
20318         references.
20320 2013-01-03  Richard Henderson  <rth@redhat.com>
20322         * config/i386/i386.c (ix86_expand_move): Always assign to op1
20323         after eliminating TLS symbols.
20325 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
20327         PR bootstrap/50167
20328         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
20329         * graphite-poly.c (debug_gmp_value): Likewise.
20331 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
20333         PR target/55712
20334         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
20335         selected code model, define __code_mode_small__, __code_model_medium__,
20336         __code_model_large__, __code_model_32__ or __code_model_kernel__.
20337         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
20338         xchg temporary register with %k.  Declare temporary register as
20339         early clobbered.
20340         [__x86_64__]: For medium and large code models, preserve %rbx register.
20342 2013-01-03  Richard Biener  <rguenther@suse.de>
20344         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
20345         (dump_subscript): Adjust.
20346         (finalize_ddr_dependent): Do not dump redundant info.
20347         (analyze_siv_subscript): Adjust.
20348         (subscript_dependence_tester): Likewise.
20349         (compute_affine_dependence): Likewise.
20351 2013-01-03  Richard Biener  <rguenther@suse.de>
20353         Revert
20354         2013-01-03  Richard Biener  <rguenther@suse.de>
20356         PR tree-optimization/55857
20357         * tree-vect-stmts.c (vectorizable_load): Do not setup
20358         re-alignment for invariant loads.
20360         2013-01-02  Richard Biener  <rguenther@suse.de>
20362         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
20363         invariant load do not generate a vector load from the scalar location.
20365 2013-01-03  Richard Biener  <rguenther@suse.de>
20367         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
20368         for not vectorizing.
20369         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
20370         not build INDIRECT_REFs, call get_name once only.
20371         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
20372         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
20374 2013-01-03  Richard Biener  <rguenther@suse.de>
20376         PR tree-optimization/55857
20377         * tree-vect-stmts.c (vectorizable_load): Do not setup
20378         re-alignment for invariant loads.
20380 2013-01-03  Richard Biener  <rguenther@suse.de>
20382         PR lto/55848
20383         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
20384         prefer a built-in decl.
20386 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
20388         * gcc.c (process_command): Update copyright notice dates.
20389         * gcov.c (print_version): Likewise.
20390         * gcov-dump.c (print_version): Likewise.
20392         PR rtl-optimization/55838
20393         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
20394         iv0.step, iv1.step and step.
20396 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
20397             Marc Glisse  <marc.glisse@inria.fr>
20399         PR tree-optimization/55832
20400         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
20401         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
20402         integer_{one,zero}_node.
20404 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
20406         PR debug/54402
20407         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
20408         * var-tracking.c (reverse_op): Don't add reverse ops to
20409         VALUEs that have already
20410         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
20412 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
20414         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
20416 2013-01-02  Teresa Johnson  <tejohnson@google.com>
20418         * dumpfile.c (dump_loc): Print filename with location.
20419         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
20420         new location_t parameter to emit complete unroll message with
20421         new dump framework.
20422         (canonicalize_loop_induction_variables): Compute loops location
20423         and pass to try_unroll_loop_completely.
20424         * loop-unroll.c (report_unroll_peel): New function.
20425         (peel_loops_completely): Use new dump format with location
20426         for main dumpfile message, and invoke report_unroll_peel on success.
20427         (decide_unrolling_and_peeling): Ditto.
20428         (decide_peel_once_rolling): Remove old dumpfile message subsumed
20429         by report_unroll_peel.
20430         (decide_peel_completely): Ditto.
20431         (decide_unroll_constant_iterations): Ditto.
20432         (decide_unroll_runtime_iterations): Ditto.
20433         (decide_peel_simple): Ditto.
20434         (decide_unroll_stupid): Ditto.
20435         * cfgloop.c (get_loop_location): New function.
20436         * cfgloop.h (get_loop_location): Declare.
20438 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
20440         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
20441         NULL.
20443 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20445         PR middle-end/55198
20446         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
20447         BLKmode objects when EXPAND_MEMORY is specified.
20449 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
20451         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
20452         in loop predicate.
20453         (fold_builtin_cpu): Do not share cpu model decls across statements.
20455 2013-01-02  Jason Merrill  <jason@redhat.com>
20457         PR c++/55804
20458         * tree.c (build_array_type_1): Revert earlier change.
20460 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
20462         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
20463         "cortex-a57".
20464         * config/aarch64/aarch64-tune.md: Re-generate.
20466 2013-01-02  Richard Biener  <rguenther@suse.de>
20468         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
20469         invariant load do not generate a vector load from the scalar location.
20471 2013-01-02  Richard Biener  <rguenther@suse.de>
20473         PR bootstrap/55784
20474         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
20475         * configure: Regenerate.
20477 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
20479         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
20480         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
20481         (expand_builtin_int_roundingfn_2): Keep the original target around
20482         for the fallback case.
20484 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
20486         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
20487         to be clear for sign changes.
20489 2013-01-01  Jan Hubicka  <jh@suse.cz>
20491         * ipa-inline-analysis.c: Fix formatting.
20493 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
20495         PR tree-optimization/55831
20496         * tree-vect-loop.c (get_initial_def_for_induction): Use
20497         gsi_after_labels instead of gsi_start_bb.
20499 Copyright (C) 2013 Free Software Foundation, Inc.
20501 Copying and distribution of this file, with or without modification,
20502 are permitted in any medium without royalty provided the copyright
20503 notice and this notice are preserved.