2014-04-24 Segher Boessenkool <segher@kernel.crashing.org>
[official-gcc.git] / gcc / ChangeLog
blob6100b75683bb3d5cc6ae97578324312fcc818f5c
1 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
3         PR target/60822
4         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
5         operand 1.
7 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
9         * flag-types.h (enum ivar_visibility): Add.
11 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
13         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
14         function * argument.
16 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
18         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
20 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
21             Tom de Vries  <tom@codesourcery.com>
23         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
24         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
25         reg-note.
26         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
27         * emit-rtl.c (try_split): Same.
29 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
30             Tom de Vries  <tom@codesourcery.com>
32         * common.opt (fuse-caller-save): New option.
34 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
36         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of elements
37         for big-endian.
39 2014-04-24  Richard Biener  <rguenther@suse.de>
41         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
42         during TER and instead use the sepops interface for expanding
43         non-GIMPLE_SINGLE_RHS.
45 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
47         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
48         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
50 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
52         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
53         assembler 64-bit option.
54         * configure: Regenerate.
56 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
58         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
59         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
60         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
61         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
62         (TARGET_CRYPTO): Take TARGET_SIMD into account.
64 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
66         * config/aarch64/aarch64-builtins.c
67         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
68         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
69         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
70         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
71         builtins.
72         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
73         (Vrevsuff): New mode attribute.
75 2014-04-24  Terry Guo  <terry.guo@arm.com>
77         * config/arm/arm.h (machine_function): Define variable
78         after_arm_reorg here.
79         * config/arm/arm.c (after_arm_reorg): Remove the definition.
80         (arm_split_constant): Update the way to access variable
81         after_arm_reorg.
82         (arm_reorg): Ditto.
83         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
85 2014-04-23  Tom de Vries  <tom@codesourcery.com>
87         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
89 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
91         * is-a.h: Update comments to reflect the following changes to the
92         "pointerness" of the API, making the template parameter match the
93         return type, allowing use of is-a.h with typedefs of pointers.
94         (is_a_helper::cast): Return a T rather then a pointer to a T, so
95         that the return type matches the parameter to the is_a_helper.
96         (as_a): Likewise.
97         (dyn_cast): Likewise.
99         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
100         pointer from the is-a.h API.
102         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
103         (is_a_helper <cgraph_node *>::test): ...this, matching change to
104         is-a.h API.
105         (is_a_helper <varpool_node>::test): Likewise, convert to...
106         (is_a_helper <varpool_node *>::test): ...this.
108         (varpool_first_variable): Update for removal of implicit pointer
109         from the is-a.h API.
110         (varpool_next_variable): Likewise.
111         (varpool_first_static_initializer): Likewise.
112         (varpool_next_static_initializer): Likewise.
113         (varpool_first_defined_variable): Likewise.
114         (varpool_next_defined_variable): Likewise.
115         (cgraph_first_defined_function): Likewise.
116         (cgraph_next_defined_function): Likewise.
117         (cgraph_first_function): Likewise.
118         (cgraph_next_function): Likewise.
119         (cgraph_first_function_with_gimple_body): Likewise.
120         (cgraph_next_function_with_gimple_body): Likewise.
121         (cgraph_alias_target): Likewise.
122         (varpool_alias_target): Likewise.
123         (cgraph_function_or_thunk_node): Likewise.
124         (varpool_variable_node): Likewise.
125         (symtab_real_symbol_p): Likewise.
126         * cgraphunit.c (referred_to_p): Likewise.
127         (analyze_functions): Likewise.
128         (handle_alias_pairs): Likewise.
129         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
130         * gimple-ssa.h (gimple_vuse_op): Likewise.
131         (gimple_vdef_op): Likewise.
132         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
133         * gimple.c (gimple_build_asm_1): Likewise.
134         (gimple_build_try): Likewise.
135         (gimple_build_resx): Likewise.
136         (gimple_build_eh_dispatch): Likewise.
137         (gimple_build_omp_for): Likewise.
138         (gimple_omp_for_set_clauses): Likewise.
140         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
141         (is_a_helper <gimple_statement_asm *>::test): ...this.
142         (is_a_helper <gimple_statement_bind>::test): Convert to...
143         (is_a_helper <gimple_statement_bind *>::test): ...this.
144         (is_a_helper <gimple_statement_call>::test): Convert to...
145         (is_a_helper <gimple_statement_call *>::test): ...this.
146         (is_a_helper <gimple_statement_catch>::test): Convert to...
147         (is_a_helper <gimple_statement_catch *>::test): ...this.
148         (is_a_helper <gimple_statement_resx>::test): Convert to...
149         (is_a_helper <gimple_statement_resx *>::test): ...this.
150         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
151         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
152         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
153         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
154         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
155         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
156         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
157         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
158         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
159         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
160         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
161         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
162         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
163         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
164         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
165         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
166         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
167         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
168         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
169         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
170         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
171         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
172         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
173         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
174         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
175         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
176         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
177         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
178         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
179         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
180         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
181         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
182         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
183         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
184         (is_a_helper <gimple_statement_phi>::test): Convert to...
185         (is_a_helper <gimple_statement_phi *>::test): ...this.
186         (is_a_helper <gimple_statement_transaction>::test): Convert to...
187         (is_a_helper <gimple_statement_transaction *>::test): ...this.
188         (is_a_helper <gimple_statement_try>::test): Convert to...
189         (is_a_helper <gimple_statement_try *>::test): ...this.
190         (is_a_helper <gimple_statement_wce>::test): Convert to...
191         (is_a_helper <gimple_statement_wce *>::test): ...this.
192         (is_a_helper <const gimple_statement_asm>::test): Convert to...
193         (is_a_helper <const gimple_statement_asm *>::test): ...this.
194         (is_a_helper <const gimple_statement_bind>::test): Convert to...
195         (is_a_helper <const gimple_statement_bind *>::test): ...this.
196         (is_a_helper <const gimple_statement_call>::test): Convert to...
197         (is_a_helper <const gimple_statement_call *>::test): ...this.
198         (is_a_helper <const gimple_statement_catch>::test): Convert to...
199         (is_a_helper <const gimple_statement_catch *>::test): ...this.
200         (is_a_helper <const gimple_statement_resx>::test): Convert to...
201         (is_a_helper <const gimple_statement_resx *>::test): ...this.
202         (is_a_helper <const gimple_statement_eh_dispatch>::test):
203         Convert to...
204         (is_a_helper <const gimple_statement_eh_dispatch *>::test):
205         ...this.
206         (is_a_helper <const gimple_statement_eh_filter>::test): Convert
207         to...
208         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
209         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
210         Convert to...
211         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
212         ...this.
213         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
214         Convert to...
215         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
216         ...this.
217         (is_a_helper <const gimple_statement_omp_return>::test): Convert
218         to...
219         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
220         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
221         to...
222         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
223         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
224         to...
225         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
226         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
227         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
228         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert
229         to...
230         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
231         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
232         to...
233         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
234         (is_a_helper <const gimple_statement_omp_target>::test): Convert
235         to...
236         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
237         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
238         to...
239         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
240         (is_a_helper <const gimple_statement_omp_single>::test): Convert
241         to...
242         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
243         (is_a_helper <const gimple_statement_omp_teams>::test): Convert
244         to...
245         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
246         (is_a_helper <const gimple_statement_omp_task>::test): Convert
247         to...
248         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
249         (is_a_helper <const gimple_statement_phi>::test): Convert to...
250         (is_a_helper <const gimple_statement_phi *>::test): ...this.
251         (is_a_helper <const gimple_statement_transaction>::test): Convert
252         to...
253         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
254         (is_a_helper <const gimple_statement_with_ops>::test): Convert
255         to...
256         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
257         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
258         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
259         (is_a_helper <const gimple_statement_with_memory_ops>::test):
260         Convert to...
261         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
262         ...this.
263         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert
264         to...
265         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
267         (gimple_use_ops): Update for removal of implicit pointer from the
268         is-a.h API.
269         (gimple_set_use_ops): Likewise.
270         (gimple_vuse): Likewise.
271         (gimple_vdef): Likewise.
272         (gimple_vuse_ptr): Likewise.
273         (gimple_vdef_ptr): Likewise.
274         (gimple_set_vuse): Likewise.
275         (gimple_set_vdef): Likewise.
276         (gimple_omp_return_set_lhs): Likewise.
277         (gimple_omp_return_lhs): Likewise.
278         (gimple_omp_return_lhs_ptr): Likewise.
279         (gimple_call_fntype): Likewise.
280         (gimple_call_set_fntype): Likewise.
281         (gimple_call_set_internal_fn): Likewise.
282         (gimple_call_use_set): Likewise.
283         (gimple_call_clobber_set): Likewise.
284         (gimple_bind_vars): Likewise.
285         (gimple_bind_set_vars): Likewise.
286         (gimple_bind_body_ptr): Likewise.
287         (gimple_bind_set_body): Likewise.
288         (gimple_bind_add_stmt): Likewise.
289         (gimple_bind_block): Likewise.
290         (gimple_bind_set_block): Likewise.
291         (gimple_asm_ninputs): Likewise.
292         (gimple_asm_noutputs): Likewise.
293         (gimple_asm_nclobbers): Likewise.
294         (gimple_asm_nlabels): Likewise.
295         (gimple_asm_input_op): Likewise.
296         (gimple_asm_input_op_ptr): Likewise.
297         (gimple_asm_output_op): Likewise.
298         (gimple_asm_output_op_ptr): Likewise.
299         (gimple_asm_set_output_op): Likewise.
300         (gimple_asm_clobber_op): Likewise.
301         (gimple_asm_set_clobber_op): Likewise.
302         (gimple_asm_label_op): Likewise.
303         (gimple_asm_set_label_op): Likewise.
304         (gimple_asm_string): Likewise.
305         (gimple_catch_types): Likewise.
306         (gimple_catch_types_ptr): Likewise.
307         (gimple_catch_handler_ptr): Likewise.
308         (gimple_catch_set_types): Likewise.
309         (gimple_catch_set_handler): Likewise.
310         (gimple_eh_filter_types): Likewise.
311         (gimple_eh_filter_types_ptr): Likewise.
312         (gimple_eh_filter_failure_ptr): Likewise.
313         (gimple_eh_filter_set_types): Likewise.
314         (gimple_eh_filter_set_failure): Likewise.
315         (gimple_eh_must_not_throw_fndecl): Likewise.
316         (gimple_eh_must_not_throw_set_fndecl): Likewise.
317         (gimple_eh_else_n_body_ptr): Likewise.
318         (gimple_eh_else_e_body_ptr): Likewise.
319         (gimple_eh_else_set_n_body): Likewise.
320         (gimple_eh_else_set_e_body): Likewise.
321         (gimple_try_eval_ptr): Likewise.
322         (gimple_try_cleanup_ptr): Likewise.
323         (gimple_try_set_eval): Likewise.
324         (gimple_try_set_cleanup): Likewise.
325         (gimple_wce_cleanup_ptr): Likewise.
326         (gimple_wce_set_cleanup): Likewise.
327         (gimple_phi_capacity): Likewise.
328         (gimple_phi_num_args): Likewise.
329         (gimple_phi_result): Likewise.
330         (gimple_phi_result_ptr): Likewise.
331         (gimple_phi_set_result): Likewise.
332         (gimple_phi_arg): Likewise.
333         (gimple_phi_set_arg): Likewise.
334         (gimple_resx_region): Likewise.
335         (gimple_resx_set_region): Likewise.
336         (gimple_eh_dispatch_region): Likewise.
337         (gimple_eh_dispatch_set_region): Likewise.
338         (gimple_omp_critical_name): Likewise.
339         (gimple_omp_critical_name_ptr): Likewise.
340         (gimple_omp_critical_set_name): Likewise.
341         (gimple_omp_for_clauses): Likewise.
342         (gimple_omp_for_clauses_ptr): Likewise.
343         (gimple_omp_for_set_clauses): Likewise.
344         (gimple_omp_for_collapse): Likewise.
345         (gimple_omp_for_index): Likewise.
346         (gimple_omp_for_index_ptr): Likewise.
347         (gimple_omp_for_set_index): Likewise.
348         (gimple_omp_for_initial): Likewise.
349         (gimple_omp_for_initial_ptr): Likewise.
350         (gimple_omp_for_set_initial): Likewise.
351         (gimple_omp_for_final): Likewise.
352         (gimple_omp_for_final_ptr): Likewise.
353         (gimple_omp_for_set_final): Likewise.
354         (gimple_omp_for_incr): Likewise.
355         (gimple_omp_for_incr_ptr): Likewise.
356         (gimple_omp_for_set_incr): Likewise.
357         (gimple_omp_for_pre_body_ptr): Likewise.
358         (gimple_omp_for_set_pre_body): Likewise.
359         (gimple_omp_parallel_clauses): Likewise.
360         (gimple_omp_parallel_clauses_ptr): Likewise.
361         (gimple_omp_parallel_set_clauses): Likewise.
362         (gimple_omp_parallel_child_fn): Likewise.
363         (gimple_omp_parallel_child_fn_ptr): Likewise.
364         (gimple_omp_parallel_set_child_fn): Likewise.
365         (gimple_omp_parallel_data_arg): Likewise.
366         (gimple_omp_parallel_data_arg_ptr): Likewise.
367         (gimple_omp_parallel_set_data_arg): Likewise.
368         (gimple_omp_task_clauses): Likewise.
369         (gimple_omp_task_clauses_ptr): Likewise.
370         (gimple_omp_task_set_clauses): Likewise.
371         (gimple_omp_task_child_fn): Likewise.
372         (gimple_omp_task_child_fn_ptr): Likewise.
373         (gimple_omp_task_set_child_fn): Likewise.
374         (gimple_omp_task_data_arg): Likewise.
375         (gimple_omp_task_data_arg_ptr): Likewise.
376         (gimple_omp_task_set_data_arg): Likewise.
377         (gimple_omp_taskreg_clauses): Likewise.
378         (gimple_omp_taskreg_clauses_ptr): Likewise.
379         (gimple_omp_taskreg_set_clauses): Likewise.
380         (gimple_omp_taskreg_child_fn): Likewise.
381         (gimple_omp_taskreg_child_fn_ptr): Likewise.
382         (gimple_omp_taskreg_set_child_fn): Likewise.
383         (gimple_omp_taskreg_data_arg): Likewise.
384         (gimple_omp_taskreg_data_arg_ptr): Likewise.
385         (gimple_omp_taskreg_set_data_arg): Likewise.
386         (gimple_omp_task_copy_fn): Likewise.
387         (gimple_omp_task_copy_fn_ptr): Likewise.
388         (gimple_omp_task_set_copy_fn): Likewise.
389         (gimple_omp_task_arg_size): Likewise.
390         (gimple_omp_task_arg_size_ptr): Likewise.
391         (gimple_omp_task_set_arg_size): Likewise.
392         (gimple_omp_task_arg_align): Likewise.
393         (gimple_omp_task_arg_align_ptr): Likewise.
394         (gimple_omp_task_set_arg_align): Likewise.
395         (gimple_omp_single_clauses): Likewise.
396         (gimple_omp_single_clauses_ptr): Likewise.
397         (gimple_omp_single_set_clauses): Likewise.
398         (gimple_omp_target_clauses): Likewise.
399         (gimple_omp_target_clauses_ptr): Likewise.
400         (gimple_omp_target_set_clauses): Likewise.
401         (gimple_omp_target_child_fn): Likewise.
402         (gimple_omp_target_child_fn_ptr): Likewise.
403         (gimple_omp_target_set_child_fn): Likewise.
404         (gimple_omp_target_data_arg): Likewise.
405         (gimple_omp_target_data_arg_ptr): Likewise.
406         (gimple_omp_target_set_data_arg): Likewise.
407         (gimple_omp_teams_clauses): Likewise.
408         (gimple_omp_teams_clauses_ptr): Likewise.
409         (gimple_omp_teams_set_clauses): Likewise.
410         (gimple_omp_sections_clauses): Likewise.
411         (gimple_omp_sections_clauses_ptr): Likewise.
412         (gimple_omp_sections_set_clauses): Likewise.
413         (gimple_omp_sections_control): Likewise.
414         (gimple_omp_sections_control_ptr): Likewise.
415         (gimple_omp_sections_set_control): Likewise.
416         (gimple_omp_for_set_cond): Likewise.
417         (gimple_omp_for_cond): Likewise.
418         (gimple_omp_atomic_store_set_val): Likewise.
419         (gimple_omp_atomic_store_val): Likewise.
420         (gimple_omp_atomic_store_val_ptr): Likewise.
421         (gimple_omp_atomic_load_set_lhs): Likewise.
422         (gimple_omp_atomic_load_lhs): Likewise.
423         (gimple_omp_atomic_load_lhs_ptr): Likewise.
424         (gimple_omp_atomic_load_set_rhs): Likewise.
425         (gimple_omp_atomic_load_rhs): Likewise.
426         (gimple_omp_atomic_load_rhs_ptr): Likewise.
427         (gimple_omp_continue_control_def): Likewise.
428         (gimple_omp_continue_control_def_ptr): Likewise.
429         (gimple_omp_continue_set_control_def): Likewise.
430         (gimple_omp_continue_control_use): Likewise.
431         (gimple_omp_continue_control_use_ptr): Likewise.
432         (gimple_omp_continue_set_control_use): Likewise.
433         (gimple_transaction_body_ptr): Likewise.
434         (gimple_transaction_label): Likewise.
435         (gimple_transaction_label_ptr): Likewise.
436         (gimple_transaction_set_body): Likewise.
437         (gimple_transaction_set_label): Likewise.
439         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
440         * ipa-inline-analysis.c (inline_write_summary): Likewise.
441         * ipa-ref.c (ipa_record_reference): Likewise.
442         * ipa-reference.c (analyze_function): Likewise.
443         (ipa_reference_write_optimization_summary): Likewise.
444         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
445         (address_taken_from_non_vtable_p): Likewise.
446         (comdat_can_be_unshared_p_1): Likewise.
447         * lto-cgraph.c (lto_output_ref): Likewise.
448         (add_references): Likewise.
449         (compute_ltrans_boundary): Likewise.
450         (output_symtab): Likewise.
451         (input_ref): Likewise.
452         (input_cgraph_1): Likewise.
453         (output_cgraph_opt_summary): Likewise.
454         * lto-streamer-out.c (lto_output): Likewise.
455         (output_symbol_p): Likewise.
456         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
457         (lsei_start_function_in_partition): Likewise.
458         (lsei_next_variable_in_partition): Likewise.
459         (lsei_start_variable_in_partition): Likewise.
460         * symtab.c (insert_to_assembler_name_hash): Likewise.
461         (unlink_from_assembler_name_hash): Likewise.
462         (symtab_unregister_node): Likewise.
463         (symtab_remove_node): Likewise.
464         (dump_symtab_node): Likewise.
465         (verify_symtab_base): Likewise.
466         (verify_symtab_node): Likewise.
467         (symtab_make_decl_local): Likewise.
468         (symtab_alias_ultimate_target): Likewise.
469         (symtab_resolve_alias): Likewise.
470         (symtab_get_symbol_partitioning_class): Likewise.
471         * tree-phinodes.c (allocate_phi_node): Likewise.
472         (reserve_phi_args_for_new_edge): Likewise.
473         (remove_phi_args): Likewise.
474         * varpool.c (varpool_node_for_asm): Likewise.
475         (varpool_remove_unreferenced_decls): Likewise.
477 2014-04-23  Jeff Law  <law@redhat.com>
479         PR tree-optimization/60902
480         * tree-ssa-threadedge.c
481         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
482         invalidate outputs from statements that do not produce useful
483         outputs for threading.
485 2014-04-23 Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
487         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
488         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
489         machine descriptions for Stack Smashing Protector.
491 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
493         * aarch64.md (<optab>_rol<mode>3): New pattern.
494         (<optab>_rolsi3_uxtw): Likewise.
495         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
497 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
499         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
500         (arm_cortex_a12_tune): Likewise.
502 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
504         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
506 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
508         * config/arm/arm.md (arm_rev16si2): New pattern.
509         (arm_rev16si2_alt): Likewise.
510         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
512 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
514        * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
515        (rev16<mode>2_alt): Likewise.
516        * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
517        * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
518        (aarch_rev16_shleft_mask_imm_p): Likewise.
519        (aarch_rev16_p_1): Likewise.
520        (aarch_rev16_p): Likewise.
521        * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
522        (aarch_rev16_shright_mask_imm_p): Likewise.
523        (aarch_rev16_shleft_mask_imm_p): Likewise.
525 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
527        * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
528        * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
529        rev cost.
530        (cortex_a53_extra_costs): Likewise.
531        (cortex_a57_extra_costs): Likewise.
532        * config/arm/arm.c (cortexa9_extra_costs): Likewise.
533        (cortexa7_extra_costs): Likewise.
534        (cortexa8_extra_costs): Likewise.
535        (cortexa12_extra_costs): Likewise.
536        (cortexa15_extra_costs): Likewise.
537        (v7m_extra_costs): Likewise.
538        (arm_new_rtx_costs): Handle BSWAP.
540 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
542        * config/arm/arm.c (cortexa8_extra_costs): New table.
543        (arm_cortex_a8_tune): New tuning struct.
544        * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
546 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
548        * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
550 2014-04-23  Richard Biener  <rguenther@suse.de>
552         * Makefile.in (OBJS): Remove loop-unswitch.o.
553         * loop-unswitch.c: Delete.
554         * tree-pass.h (make_pass_rtl_unswitch): Remove.
555         * passes.def (pass_rtl_unswitch): Likewise.
556         * loop-init.c (gate_rtl_unswitch): Likewise.
557         (rtl_unswitch): Likewise.
558         (pass_data_rtl_unswitch): Likewise.
559         (pass_rtl_unswitch): Likewise.
560         (make_pass_rtl_unswitch): Likewise.
561         * rtl.h (reversed_condition): Likewise.
562         (compare_and_jump_seq): Likewise.
563         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
564         and make static.
565         * loop-unroll.c (compare_and_jump_seq): Likewise.
567 2014-04-23  Richard Biener  <rguenther@suse.de>
569         PR tree-optimization/60903
570         * tree-ssa-loop-im.c (analyze_memory_references): Remove
571         commented code block.
572         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
573         loop flags to newly created BBs and edges.
575 2014-04-23  Nick Clifton  <nickc@redhat.com>
577         * config/msp430/msp430.c (msp430_handle_option): Move function
578         to msp430-common.c
579         (msp430_option_override): Simplify mcu and mcpu option handling.
580         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
581         support for -mhwmult command line option.
582         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
583         -mhwmult command line option.
584         (msp430_hwmult_enabled): Delete.
585         (msp43o_output_labelref): Add support for -mhwmult command line
586         option.
587         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
588         (umulsidi3): Likewise.
589         * config/msp430/msp430.opt (mmcu): Add Report attribute.
590         (mcpu, mlarge, msmall): Likewise.
591         (mhwmult): New option.
592         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
593         prototype.
594         (msp430_is_f5_mcu): Remove prototype.
595         (msp430_use_f5_series_hwmult): Add prototype.
596         * config/msp430/msp430-opts.h: New file.
597         * common/config/msp430: New directory.
598         * common/config/msp430/msp430-common.c: New file.
599         * config.gcc (msp430): Remove target_has_targetm_common.
600         * doc/invoke.texi: Document -mhwmult command line option.
602 2014-04-23  Nick Clifton  <nickc@redhat.com>
604         * config/i386/cygwin.h (ENDFILE_SPEC): Include
605         default-manifest.o if it can be found in the search path.
606         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
608 2014-04-23  Terry Guo  <terry.guo@arm.com>
610         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
612 2014-04-23  Richard Biener  <rguenther@suse.de>
614         PR middle-end/60895
615         * tree-inline.c (declare_return_variable): Use mark_addressable.
617 2014-04-23  Richard Biener  <rguenther@suse.de>
619         PR middle-end/60891
620         * loop-init.c (loop_optimizer_init): Make sure to apply
621         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
623 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
625         PR sanitizer/60275
626         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
627         New options.
628         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
629         if flag_sanitize_undefined_trap_on_error.
630         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
631         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
632         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
633         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
634         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
635         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
636         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
637         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
638         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
639         * ubsan.c (ubsan_instrument_unreachable): Return
640         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
641         (ubsan_expand_null_ifn): Emit __builtin_trap ()
642         if flag_sanitize_undefined_trap_on_error and
643         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
644         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
645         instrument_bool_enum_load): Emit __builtin_trap () if
646         flag_sanitize_undefined_trap_on_error and
647         __builtin_handle_*_abort () if !flag_sanitize_recover.
648         * doc/invoke.texi (-fsanitize-recover,
649         -fsanitize-undefined-trap-on-error): Document.
651 2014-04-22  Christian Bruel  <christian.bruel@st.com>
653         * config/sh/sh.md (mov<mode>): Replace movQIHI.
654          Force immediates to SImode.
656 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
658         * config/nios2/nios2.md (UNSPEC_ROUND): New.
659         (lroundsfsi2): New.
660         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
661         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
662         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
663         (nios2_fpu_insn): Add entry for round.
664         (N2FPU_NO_ERRNO_P): Define.
665         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
666         flag_errno_math.
667         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
669 2014-04-22  Richard Henderson  <rth@redhat.com>
671         * config/aarch64/aarch64 (addti3, subti3): New expanders.
672         (add<GPI>3_compare0): Remove leading * from name.
673         (add<GPI>3_carryin): Likewise.
674         (sub<GPI>3_compare0): Likewise.
675         (sub<GPI>3_carryin): Likewise.
676         (<su_optab>mulditi3): New expander.
677         (multi3): New expander.
678         (madd<GPI>): Remove leading * from name.
680 2014-04-22  Martin Jambor  <mjambor@suse.cz>
682         * cgraphclones.c (cgraph_function_versioning): Copy
683         ipa_transforms_to_apply instead of asserting it is empty.
685 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
687         PR target/60868
688         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
689         on count_exp to get mode.
691 2014-04-22  Andrew Pinski  <apinski@cavium.com>
693         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
694         Handle TLS for ILP32.
695         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
696         (tlsie_small_<mode>): this and handle PTR.
697         (tlsie_small_sidi): New pattern.
698         (tlsle_small): Change to an expand to handle ILP32.
699         (tlsle_small_<mode>): New pattern.
700         (tlsdesc_small): Rename to ...
701         (tlsdesc_small_<mode>): this and handle PTR.
703 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
705         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
707 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
709         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
710         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
711         (aarch64_types_signed_poly_qualifiers): Likewise.
712         (aarch64_types_unsigned_signed_qualifiers): Likewise.
713         (aarch64_types_poly_signed_qualifiers): Likewise.
714         (TYPES_REINTERP_SS): Type macro added.
715         (TYPES_REINTERP_SU): Likewise.
716         (TYPES_REINTERP_SP): Likewise.
717         (TYPES_REINTERP_US): Likewise.
718         (TYPES_REINTERP_PS): Likewise.
719         (aarch64_fold_builtin): New expression folding added.
720         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
721         Declarations removed.
722         (REINTERP_SS): Declarations added.
723         (REINTERP_US): Likewise.
724         (REINTERP_PS): Likewise.
725         (REINTERP_SU): Likewise.
726         (REINTERP_SP): Likewise.
727         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
728         (vreinterpretq_p8_f64): Likewise.
729         (vreinterpret_p16_f64): Likewise.
730         (vreinterpretq_p16_f64): Likewise.
731         (vreinterpret_f32_f64): Likewise.
732         (vreinterpretq_f32_f64): Likewise.
733         (vreinterpret_f64_f32): Likewise.
734         (vreinterpret_f64_p8): Likewise.
735         (vreinterpret_f64_p16): Likewise.
736         (vreinterpret_f64_s8): Likewise.
737         (vreinterpret_f64_s16): Likewise.
738         (vreinterpret_f64_s32): Likewise.
739         (vreinterpret_f64_s64): Likewise.
740         (vreinterpret_f64_u8): Likewise.
741         (vreinterpret_f64_u16): Likewise.
742         (vreinterpret_f64_u32): Likewise.
743         (vreinterpret_f64_u64): Likewise.
744         (vreinterpretq_f64_f32): Likewise.
745         (vreinterpretq_f64_p8): Likewise.
746         (vreinterpretq_f64_p16): Likewise.
747         (vreinterpretq_f64_s8): Likewise.
748         (vreinterpretq_f64_s16): Likewise.
749         (vreinterpretq_f64_s32): Likewise.
750         (vreinterpretq_f64_s64): Likewise.
751         (vreinterpretq_f64_u8): Likewise.
752         (vreinterpretq_f64_u16): Likewise.
753         (vreinterpretq_f64_u32): Likewise.
754         (vreinterpretq_f64_u64): Likewise.
755         (vreinterpret_s64_f64): Likewise.
756         (vreinterpretq_s64_f64): Likewise.
757         (vreinterpret_u64_f64): Likewise.
758         (vreinterpretq_u64_f64): Likewise.
759         (vreinterpret_s8_f64): Likewise.
760         (vreinterpretq_s8_f64): Likewise.
761         (vreinterpret_s16_f64): Likewise.
762         (vreinterpretq_s16_f64): Likewise.
763         (vreinterpret_s32_f64): Likewise.
764         (vreinterpretq_s32_f64): Likewise.
765         (vreinterpret_u8_f64): Likewise.
766         (vreinterpretq_u8_f64): Likewise.
767         (vreinterpret_u16_f64): Likewise.
768         (vreinterpretq_u16_f64): Likewise.
769         (vreinterpret_u32_f64): Likewise.
770         (vreinterpretq_u32_f64): Likewise.
772 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
774         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
775         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
776         (vreinterpret_p8_s8): Likewise.
777         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
778         (vreinterpret_p8_s16): Likewise.
779         (vreinterpret_p8_s32): Likewise.
780         (vreinterpret_p8_s64): Likewise.
781         (vreinterpret_p8_f32): Likewise.
782         (vreinterpret_p8_u8): Likewise.
783         (vreinterpret_p8_u16): Likewise.
784         (vreinterpret_p8_u32): Likewise.
785         (vreinterpret_p8_u64): Likewise.
786         (vreinterpret_p8_p16): Likewise.
787         (vreinterpretq_p8_s8): Likewise.
788         (vreinterpretq_p8_s16): Likewise.
789         (vreinterpretq_p8_s32): Likewise.
790         (vreinterpretq_p8_s64): Likewise.
791         (vreinterpretq_p8_f32): Likewise.
792         (vreinterpretq_p8_u8): Likewise.
793         (vreinterpretq_p8_u16): Likewise.
794         (vreinterpretq_p8_u32): Likewise.
795         (vreinterpretq_p8_u64): Likewise.
796         (vreinterpretq_p8_p16): Likewise.
797         (vreinterpret_p16_s8): Likewise.
798         (vreinterpret_p16_s16): Likewise.
799         (vreinterpret_p16_s32): Likewise.
800         (vreinterpret_p16_s64): Likewise.
801         (vreinterpret_p16_f32): Likewise.
802         (vreinterpret_p16_u8): Likewise.
803         (vreinterpret_p16_u16): Likewise.
804         (vreinterpret_p16_u32): Likewise.
805         (vreinterpret_p16_u64): Likewise.
806         (vreinterpret_p16_p8): Likewise.
807         (vreinterpretq_p16_s8): Likewise.
808         (vreinterpretq_p16_s16): Likewise.
809         (vreinterpretq_p16_s32): Likewise.
810         (vreinterpretq_p16_s64): Likewise.
811         (vreinterpretq_p16_f32): Likewise.
812         (vreinterpretq_p16_u8): Likewise.
813         (vreinterpretq_p16_u16): Likewise.
814         (vreinterpretq_p16_u32): Likewise.
815         (vreinterpretq_p16_u64): Likewise.
816         (vreinterpretq_p16_p8): Likewise.
817         (vreinterpret_f32_s8): Likewise.
818         (vreinterpret_f32_s16): Likewise.
819         (vreinterpret_f32_s32): Likewise.
820         (vreinterpret_f32_s64): Likewise.
821         (vreinterpret_f32_u8): Likewise.
822         (vreinterpret_f32_u16): Likewise.
823         (vreinterpret_f32_u32): Likewise.
824         (vreinterpret_f32_u64): Likewise.
825         (vreinterpret_f32_p8): Likewise.
826         (vreinterpret_f32_p16): Likewise.
827         (vreinterpretq_f32_s8): Likewise.
828         (vreinterpretq_f32_s16): Likewise.
829         (vreinterpretq_f32_s32): Likewise.
830         (vreinterpretq_f32_s64): Likewise.
831         (vreinterpretq_f32_u8): Likewise.
832         (vreinterpretq_f32_u16): Likewise.
833         (vreinterpretq_f32_u32): Likewise.
834         (vreinterpretq_f32_u64): Likewise.
835         (vreinterpretq_f32_p8): Likewise.
836         (vreinterpretq_f32_p16): Likewise.
837         (vreinterpret_s64_s8): Likewise.
838         (vreinterpret_s64_s16): Likewise.
839         (vreinterpret_s64_s32): Likewise.
840         (vreinterpret_s64_f32): Likewise.
841         (vreinterpret_s64_u8): Likewise.
842         (vreinterpret_s64_u16): Likewise.
843         (vreinterpret_s64_u32): Likewise.
844         (vreinterpret_s64_u64): Likewise.
845         (vreinterpret_s64_p8): Likewise.
846         (vreinterpret_s64_p16): Likewise.
847         (vreinterpretq_s64_s8): Likewise.
848         (vreinterpretq_s64_s16): Likewise.
849         (vreinterpretq_s64_s32): Likewise.
850         (vreinterpretq_s64_f32): Likewise.
851         (vreinterpretq_s64_u8): Likewise.
852         (vreinterpretq_s64_u16): Likewise.
853         (vreinterpretq_s64_u32): Likewise.
854         (vreinterpretq_s64_u64): Likewise.
855         (vreinterpretq_s64_p8): Likewise.
856         (vreinterpretq_s64_p16): Likewise.
857         (vreinterpret_u64_s8): Likewise.
858         (vreinterpret_u64_s16): Likewise.
859         (vreinterpret_u64_s32): Likewise.
860         (vreinterpret_u64_s64): Likewise.
861         (vreinterpret_u64_f32): Likewise.
862         (vreinterpret_u64_u8): Likewise.
863         (vreinterpret_u64_u16): Likewise.
864         (vreinterpret_u64_u32): Likewise.
865         (vreinterpret_u64_p8): Likewise.
866         (vreinterpret_u64_p16): Likewise.
867         (vreinterpretq_u64_s8): Likewise.
868         (vreinterpretq_u64_s16): Likewise.
869         (vreinterpretq_u64_s32): Likewise.
870         (vreinterpretq_u64_s64): Likewise.
871         (vreinterpretq_u64_f32): Likewise.
872         (vreinterpretq_u64_u8): Likewise.
873         (vreinterpretq_u64_u16): Likewise.
874         (vreinterpretq_u64_u32): Likewise.
875         (vreinterpretq_u64_p8): Likewise.
876         (vreinterpretq_u64_p16): Likewise.
877         (vreinterpret_s8_s16): Likewise.
878         (vreinterpret_s8_s32): Likewise.
879         (vreinterpret_s8_s64): Likewise.
880         (vreinterpret_s8_f32): Likewise.
881         (vreinterpret_s8_u8): Likewise.
882         (vreinterpret_s8_u16): Likewise.
883         (vreinterpret_s8_u32): Likewise.
884         (vreinterpret_s8_u64): Likewise.
885         (vreinterpret_s8_p8): Likewise.
886         (vreinterpret_s8_p16): Likewise.
887         (vreinterpretq_s8_s16): Likewise.
888         (vreinterpretq_s8_s32): Likewise.
889         (vreinterpretq_s8_s64): Likewise.
890         (vreinterpretq_s8_f32): Likewise.
891         (vreinterpretq_s8_u8): Likewise.
892         (vreinterpretq_s8_u16): Likewise.
893         (vreinterpretq_s8_u32): Likewise.
894         (vreinterpretq_s8_u64): Likewise.
895         (vreinterpretq_s8_p8): Likewise.
896         (vreinterpretq_s8_p16): Likewise.
897         (vreinterpret_s16_s8): Likewise.
898         (vreinterpret_s16_s32): Likewise.
899         (vreinterpret_s16_s64): Likewise.
900         (vreinterpret_s16_f32): Likewise.
901         (vreinterpret_s16_u8): Likewise.
902         (vreinterpret_s16_u16): Likewise.
903         (vreinterpret_s16_u32): Likewise.
904         (vreinterpret_s16_u64): Likewise.
905         (vreinterpret_s16_p8): Likewise.
906         (vreinterpret_s16_p16): Likewise.
907         (vreinterpretq_s16_s8): Likewise.
908         (vreinterpretq_s16_s32): Likewise.
909         (vreinterpretq_s16_s64): Likewise.
910         (vreinterpretq_s16_f32): Likewise.
911         (vreinterpretq_s16_u8): Likewise.
912         (vreinterpretq_s16_u16): Likewise.
913         (vreinterpretq_s16_u32): Likewise.
914         (vreinterpretq_s16_u64): Likewise.
915         (vreinterpretq_s16_p8): Likewise.
916         (vreinterpretq_s16_p16): Likewise.
917         (vreinterpret_s32_s8): Likewise.
918         (vreinterpret_s32_s16): Likewise.
919         (vreinterpret_s32_s64): Likewise.
920         (vreinterpret_s32_f32): Likewise.
921         (vreinterpret_s32_u8): Likewise.
922         (vreinterpret_s32_u16): Likewise.
923         (vreinterpret_s32_u32): Likewise.
924         (vreinterpret_s32_u64): Likewise.
925         (vreinterpret_s32_p8): Likewise.
926         (vreinterpret_s32_p16): Likewise.
927         (vreinterpretq_s32_s8): Likewise.
928         (vreinterpretq_s32_s16): Likewise.
929         (vreinterpretq_s32_s64): Likewise.
930         (vreinterpretq_s32_f32): Likewise.
931         (vreinterpretq_s32_u8): Likewise.
932         (vreinterpretq_s32_u16): Likewise.
933         (vreinterpretq_s32_u32): Likewise.
934         (vreinterpretq_s32_u64): Likewise.
935         (vreinterpretq_s32_p8): Likewise.
936         (vreinterpretq_s32_p16): Likewise.
937         (vreinterpret_u8_s8): Likewise.
938         (vreinterpret_u8_s16): Likewise.
939         (vreinterpret_u8_s32): Likewise.
940         (vreinterpret_u8_s64): Likewise.
941         (vreinterpret_u8_f32): Likewise.
942         (vreinterpret_u8_u16): Likewise.
943         (vreinterpret_u8_u32): Likewise.
944         (vreinterpret_u8_u64): Likewise.
945         (vreinterpret_u8_p8): Likewise.
946         (vreinterpret_u8_p16): Likewise.
947         (vreinterpretq_u8_s8): Likewise.
948         (vreinterpretq_u8_s16): Likewise.
949         (vreinterpretq_u8_s32): Likewise.
950         (vreinterpretq_u8_s64): Likewise.
951         (vreinterpretq_u8_f32): Likewise.
952         (vreinterpretq_u8_u16): Likewise.
953         (vreinterpretq_u8_u32): Likewise.
954         (vreinterpretq_u8_u64): Likewise.
955         (vreinterpretq_u8_p8): Likewise.
956         (vreinterpretq_u8_p16): Likewise.
957         (vreinterpret_u16_s8): Likewise.
958         (vreinterpret_u16_s16): Likewise.
959         (vreinterpret_u16_s32): Likewise.
960         (vreinterpret_u16_s64): Likewise.
961         (vreinterpret_u16_f32): Likewise.
962         (vreinterpret_u16_u8): Likewise.
963         (vreinterpret_u16_u32): Likewise.
964         (vreinterpret_u16_u64): Likewise.
965         (vreinterpret_u16_p8): Likewise.
966         (vreinterpret_u16_p16): Likewise.
967         (vreinterpretq_u16_s8): Likewise.
968         (vreinterpretq_u16_s16): Likewise.
969         (vreinterpretq_u16_s32): Likewise.
970         (vreinterpretq_u16_s64): Likewise.
971         (vreinterpretq_u16_f32): Likewise.
972         (vreinterpretq_u16_u8): Likewise.
973         (vreinterpretq_u16_u32): Likewise.
974         (vreinterpretq_u16_u64): Likewise.
975         (vreinterpretq_u16_p8): Likewise.
976         (vreinterpretq_u16_p16): Likewise.
977         (vreinterpret_u32_s8): Likewise.
978         (vreinterpret_u32_s16): Likewise.
979         (vreinterpret_u32_s32): Likewise.
980         (vreinterpret_u32_s64): Likewise.
981         (vreinterpret_u32_f32): Likewise.
982         (vreinterpret_u32_u8): Likewise.
983         (vreinterpret_u32_u16): Likewise.
984         (vreinterpret_u32_u64): Likewise.
985         (vreinterpret_u32_p8): Likewise.
986         (vreinterpret_u32_p16): Likewise.
987         (vreinterpretq_u32_s8): Likewise.
988         (vreinterpretq_u32_s16): Likewise.
989         (vreinterpretq_u32_s32): Likewise.
990         (vreinterpretq_u32_s64): Likewise.
991         (vreinterpretq_u32_f32): Likewise.
992         (vreinterpretq_u32_u8): Likewise.
993         (vreinterpretq_u32_u16): Likewise.
994         (vreinterpretq_u32_u64): Likewise.
995         (vreinterpretq_u32_p8): Likewise.
996         (vreinterpretq_u32_p16): Likewise.
998 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1000         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
1001         Pattern extended.
1002         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
1003         extended.
1004         (sqabs): Likewise.
1005         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
1006         (vqnegd_s64): Likewise.
1007         (vqabs_s64): Likewise.
1008         (vqabsd_s64): Likewise.
1010 2014-04-22  Richard Henderson  <rth@redhat.com>
1012         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
1013         computation to the top of the loop.
1015 2014-04-22  Renlin  <renlin.li@arm.com>
1016             Jiong Wang  <jiong.wang@arm.com>
1018         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
1019         * config/aarch64/aarch64.c (aarch64_layout_frame)
1020         (aarch64_initial_elimination_offset): Likewise.
1022 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1024         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
1025         Fix indentation.
1027 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
1029         * machmode.h (bitwise_mode_for_mode): Declare.
1030         * stor-layout.h (bitwise_type_for_mode): Likewise.
1031         * stor-layout.c (bitwise_mode_for_mode): New function.
1032         (bitwise_type_for_mode): Likewise.
1033         * builtins.c (fold_builtin_memory_op): Use it instead of
1034         int_mode_for_mode and build_nonstandard_integer_type.
1036 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1038         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
1039         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
1040         (*-*-solaris2*): Simplify.
1041         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
1042         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
1043         *-*-solaris2.9* handling.
1045         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
1046         as bug.
1047         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
1048         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
1049         handling, simplify.
1050         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
1051         * configure: Regenerate.
1053         * config/i386/sol2-9.h: Remove.
1055         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
1056         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
1057         Remove Solaris 9 references.
1059 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
1061         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
1062         (floatuns<GPI:mode><GPF:mode>2): Remove.
1063         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
1064         and floatuns conversions.
1065         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
1066         and floatuns conversions.
1067         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
1068         (w1,w2): New mode attributes for inequal width conversions.
1070 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
1072         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
1073         the output asm format.
1075 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
1077         * config/aarch64/aarch64-simd.md
1078         (aarch64_cm<optab>di): Always split.
1079         (*aarch64_cm<optab>di): New.
1080         (aarch64_cmtstdi): Always split.
1081         (*aarch64_cmtstdi): New.
1083 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1085         PR tree-optimization/60823
1086         * omp-low.c (ipa_simd_modify_function_body): Go through
1087         all SSA_NAMEs and for those refering to vector arguments
1088         which are going to be replaced adjust SSA_NAME_VAR and,
1089         if it is a default definition, change it into a non-default
1090         definition assigned at the beginning of function from new_decl.
1091         (ipa_simd_modify_stmt_ops): Rewritten.
1092         * tree-dfa.c (set_ssa_default_def): When removing default def,
1093         check for NULL loc instead of NULL *loc.
1095 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1097         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
1098         restrictions on core registers for DImode values in Thumb2.
1100 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1102         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
1103         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
1105 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1107         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
1108         (*iordi_notzesidi_di): Likewise.
1109         (*iordi_notsesidi_di): Likewise.
1111 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1113         * config/arm/arm-protos.h (tune_params): New struct members.
1114         * config/arm/arm.c: Initialise tune_params per processor.
1115         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
1116         for speed, based on new tune_params.
1118 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1120         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
1121         added.
1122         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
1123         macro.
1124         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
1125         corrected.
1126         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
1127         * config/aarch64/arm_neon.h (vrnd_f64): Added.
1128         (vrnda_f64): Likewise.
1129         (vrndi_f64): Likewise.
1130         (vrndm_f64): Likewise.
1131         (vrndn_f64): Likewise.
1132         (vrndp_f64): Likewise.
1133         (vrndx_f64): Likewise.
1135 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1137         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
1138         GET_MODE_SIZE argument is enum machine_mode.
1140 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1142         PR target/60910
1143         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
1144         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
1146 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
1148         PR middle-end/60281
1149         * asan.c (asan_emit_stack_protection): Force the base to align to
1150         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
1151         appropriate bits if STRICT_ALIGNMENT.
1152         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
1153         when asan is on.
1154         (expand_used_vars): Leave a space in the stack frame for alignment
1155         if STRICT_ALIGNMENT.
1157 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
1159         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
1160         than a gimple.
1161         (gimple_store_p): Likewise.
1162         (gimple_assign_load_p): Likewise.
1163         (gimple_assign_cast_p): Likewise.
1164         (gimple_clobber_p): Likewise.
1166         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
1167         rather than a gimple.
1168         (gimple_assign_cast_p): Likewise.
1170 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
1172         PR target/60735
1173         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
1174         If mode is DDmode and TARGET_E500_DOUBLE allow move.
1176         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
1177         more debug information for E500 if -mdebug=reg.
1179 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
1181         PR target/60909
1182         * config/i386/i386.c (ix86_expand_builtin)
1183         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
1184         register for target RTX.
1185         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
1187 2014-04-18  Cong Hou  <congh@google.com>
1189         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
1190         the widen-mult pattern by handling two operands with different sizes,
1191         and operands whose size is smaller than half of the result type.
1193 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1195         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
1196         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
1197         (do_estimate_edge_time): Compute it.
1198         * ipa-inline.c (want_inline_small_function_p): Bypass
1199         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
1201 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1203         * ipa-inline.c (spec_rem): New static variable.
1204         (dump_overall_stats): New function.
1205         (dump_inline_stats): New function.
1207 2014-04-18  Richard Henderson  <rth@redhat.com>
1209         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
1210         to GET_MODE_SIZE, not a reg_class_t.
1212 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1214         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
1215         (vsx_xxmrglw_<mode>): Likewise.
1217 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
1219         PR target/60876
1220         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
1221         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
1222         (rs6000_init_hard_regno_mode_ok): Likewise.
1224 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
1226         * ipa-inline.c (inline_small_functions): Account only non-cold
1227         functions.
1228         * doc/invoke.texi (inline-unit-growth): Update documentation.
1230 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
1232         * config/rs6000/rs6000.md (addti3, subti3): New.
1234 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
1236         PR target/60863
1237         * config/i386/i386.c (ix86_expand_clear): Remove outdated
1238         comment.  Check optimize_insn_for_size_p instead of
1239         optimize_insn_for_speed_p.
1241 2014-04-17  Martin Jambor  <mjambor@suse.cz>
1243         * gimple-iterator.c (gsi_start_edge): New function.
1244         * gimple-iterator.h (gsi_start_edge): Declare.
1245         * tree-sra.c (single_non_eh_succ): New function.
1246         (disqualify_ops_if_throwing_stmt): Renamed to
1247         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
1248         having one non-EH successor BB.
1249         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
1250         generate loads into replacements.
1251         (sra_modify_assign): Likewise and and also use the simple path for
1252         such statements.
1253         (sra_modify_function_body): Commit statements on edges.
1255 2014-04-17  Richard Biener  <rguenther@suse.de>
1257         PR middle-end/60849
1258         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
1259         comparison results and add clarifying comment.
1261 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1263         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
1264         (blank_mode): Initialize it.
1265         (emit_mode_size_inline, emit_mode_nunits_inline,
1266         emit_mode_inner_inline): New functions.
1267         (emit_insn_modes_h): Call them and surround their output with
1268         #if GCC_VERSION >= 4001 ... #endif.
1269         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
1270         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
1271         mode_* arrays if the argument is __builtin_constant_p.
1272         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
1273         is enum machine_mode.
1275 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1277         * passes.c (opt_pass::execute): Adjust.
1278         (pass_manager::execute_pass_mode_switching): Likewise.
1279         (early_local_passes::execute): Likewise.
1280         (execute_one_pass): Pass cfun to the pass's execute method.
1281         * tree-pass.h (opt_pass::execute): Add function * argument.
1282         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1283         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1284         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1285         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1286         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1287         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
1288         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
1289         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1290         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1291         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
1292         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
1293         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
1294         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
1295         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
1296         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1297         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1298         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1299         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
1300         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
1301         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1302         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1303         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1304         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1305         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1306         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1307         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1308         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1309         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1310         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1311         Adjust.
1313 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1315         * passes.c (opt_pass::gate): Take function * argument.
1316         (gate_all_early_local_passes): Merge into
1317         (early_local_passes::gate): this.
1318         (gate_all_early_optimizations): Merge into
1319         (all_early_optimizations::gate): this.
1320         (gate_all_optimizations): Mege into
1321         (all_optimizations::gate): this.
1322         (gate_all_optimizations_g): Merge into
1323         (all_optimizations_g::gate): this.
1324         (gate_rest_of_compilation): Mege into
1325         (rest_of_compilation::gate): this.
1326         (gate_postreload): Merge into
1327         (postreload::gate): this.
1328         (dump_one_pass): Pass cfun to the pass's gate method.
1329         (execute_ipa_summary_passes): Likewise.
1330         (execute_one_pass): Likewise.
1331         (ipa_write_summaries_2): Likewise.
1332         (ipa_write_optimization_summaries_1): Likewise.
1333         (ipa_read_summaries_1): Likewise.
1334         (ipa_read_optimization_summaries_1): Likewise.
1335         (execute_ipa_stmt_fixups): Likewise.
1336         * tree-pass.h (opt_pass::gate): Add function * argument.
1337         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
1338         combine-stack-adj.c, combine.c, compare-elim.c,
1339         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1340         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
1341         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
1342         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
1343         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1344         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1345         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1346         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
1347         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
1348         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
1349         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1350         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1351         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
1352         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1353         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1354         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1355         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1356         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1357         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1358         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1359         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1360         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1361         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
1362         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
1363         var-tracking.c, vtable-verify.c, web.c: Adjust.
1365 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1367         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
1368         * configure: Regenerate.
1370 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1372         * passes.c (dump_one_pass): don't check pass->has_gate.
1373         (execute_ipa_summary_passes): Likewise.
1374         (execute_one_pass): Likewise.
1375         (ipa_write_summaries_2): Likewise.
1376         (ipa_write_optimization_summaries_1): Likewise.
1377         (ipa_read_optimization_summaries_1): Likewise.
1378         (execute_ipa_stmt_fixups): Likewise.
1379         * tree-pass.h (pass_data::has_gate): Remove.
1380         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1381         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1382         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1383         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1384         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1385         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
1386         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
1387         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
1388         gimple-low.c, gimple-ssa-isolate-paths.c,
1389         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
1390         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
1391         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1392         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
1393         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
1394         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
1395         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
1396         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
1397         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
1398         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1399         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1400         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1401         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1402         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1403         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1404         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1405         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1406         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1407         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1408         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1409         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1410         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1411         Adjust.
1413 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1415         * pass_manager.h (pass_manager::register_dump_files_1): Remove
1416         declaration.
1417         * passes.c (pass_manager::register_dump_files_1): Merge into
1418         (pass_manager::register_dump_files): this, and remove its handling of
1419         properties since the pass always has the properties anyway.
1420         (pass_manager::pass_manager): Adjust.
1422 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1424         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
1425         * passes.c (pass_manager::register_dump_files_1): Remove dead code
1426         dealing with properties.
1427         (pass_manager::register_dump_files): Adjust.
1429 2014-03-20  Mark Wielaard  <mjw@redhat.com>
1431         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
1432         then represent the bound as normal constant value.
1434 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1436         PR target/60847
1437         Forward port from 4.8 branch
1438         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
1440         * config/i386/bmiintrin.h (_blsi_u32): New.
1441         (_blsi_u64): Ditto.
1442         (_blsr_u32): Ditto.
1443         (_blsr_u64): Ditto.
1444         (_blsmsk_u32): Ditto.
1445         (_blsmsk_u64): Ditto.
1446         (_tzcnt_u32): Ditto.
1447         (_tzcnt_u64): Ditto.
1449 2014-04-17  Kito Cheng  <kito@0xlab.org>
1451         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
1453 2014-04-17  Richard Biener  <rguenther@suse.de>
1455         PR middle-end/60849
1456         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
1457         boolean results for comparisons.
1459 2014-04-17  Richard Biener  <rguenther@suse.de>
1461         PR tree-optimization/60836
1462         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
1463         initial PHI args to be gimple values.
1465 2014-04-17   Richard Biener  <rguenther@suse.de>
1467         PR tree-optimization/60841
1468         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
1469         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
1470         of stmts to SLP build.
1471         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
1472         (vect_analyze_slp): Likewise.
1473         (vect_analyze_slp_instance): Likewise.
1474         (vect_build_slp_tree): Limit overall SLP tree growth.
1475         * tree-vectorizer.h (vect_analyze_data_refs,
1476         vect_analyze_slp): Adjust prototypes.
1478 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1480         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
1481         Silvermont.
1483 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1485         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
1486         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
1487         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
1488         for TARGET_SLOW_PSHUFB
1490 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1492         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
1493         * config/i386/i386.c (intel_cost): Ditto.
1495 2014-04-17  Joey Ye  <joey.ye@arm.com>
1497         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
1499 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1501         * opts.c (common_handle_option): Disable -fipa-reference coorectly
1502         with -fuse-profile.
1504 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1506         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
1507         (type_all_derivations_known_p): New predicate.
1508         (type_all_ctors_visible_p): New predicate.
1509         (type_possibly_instantiated_p): New predicate.
1510         (get_odr_type): Compute all_derivations_known.
1511         (dump_odr_type): Dump the flag.
1512         (maybe_record_type): Cleanup.
1513         (record_target_from_binfo): Add bases_to_consider array;
1514         record bases for types w/o instances and skip CXX destructor.
1515         (possible_polymorphic_call_targets_1): Add bases_to_consider
1516         and consider_construction parameters; check if type may have instance.
1517         (get_polymorphic_call_info): Set maybe_in_construction to true
1518         when we know nothing.
1519         (record_targets_from_bases): Skip CXX destructors; they are
1520         never called for types in construction.
1521         (possible_polymorphic_call_targets): Do not record target when
1522         type may not have instance.
1524 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1526         PR ipa/60854
1527         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
1528         external aliases alive, too.
1530 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
1532         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
1533         definition.
1535 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
1537         * final.c (compute_alignments): Do not apply loop alignment to a block
1538         falling through to the exit.
1540 2014-04-16  Catherine Moore  <clm@codesourcery.com>
1542         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
1543         Adjust constraints for microMIPS store patterns.
1545 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
1547         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
1549 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
1551         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
1552         (append_use): Run at -O0.
1553         (append_vdef): Likewise.
1554         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
1555         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
1557 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
1559         PR tree-optimization/60844
1560         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
1561         (propagate_op_to_single_use, remove_visited_stmt_chain,
1562         linearize_expr, repropagate_negates, reassociate_bb): Use it
1563         instead of gsi_remove.
1565 2014-04-16  Martin Jambor  <mjambor@suse.cz>
1567         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
1568         ipa_transforms_to_apply.
1569         (cgraph_function_versioning): Assert that old_node has empty
1570         ipa_transforms_to_apply.
1571         * trans-mem.c (ipa_tm_create_version): Likewise.
1572         * tree-inline.c (tree_function_versioning): Do not duplicate
1573         ipa_transforms_to_apply.
1575 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1577         PR target/60817
1578         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
1579         x86_64-*-* cases.
1580         Pass necessary as flags on 64-bit Solaris/x86.
1581         Use lowercase relocs for x86_64-*-*.
1582         * configure: Regenerate.
1584 2014-04-15  Jan Hubicka  <jh@suse.cz>
1586         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
1587         (maybe_record_node, likely_target_p): Use it.
1589 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1591         PR target/60839
1592         Revert following patch
1594         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
1596         PR target/60735
1597         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1598         software floating point or no floating point registers, do not
1599         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
1600         in GPRs that occurs after we tested for GPRs that would never be
1601         true.
1603         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1604         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1605         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
1606         specifically allow DDmode, since that does not use the SPE SIMD
1607         instructions.
1609 2014-03-21  Mark Wielaard  <mjw@redhat.com>
1611         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
1612         as unsigned or int depending on type and value used.
1614 2014-04-15  Richard Biener  <rguenther@suse.de>
1616         PR rtl-optimization/56965
1617         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
1618         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
1619         ... here.
1620         * alias.c (true_dependence_1): Do not call
1621         nonoverlapping_component_refs_p.
1622         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
1623         nonoverlapping_component_refs_p.
1624         (indirect_refs_may_alias_p): Likewise.
1626 2014-04-15  Teresa Johnson  <tejohnson@google.com>
1628         * cfg.c (dump_bb_info): Fix flags check.
1629         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
1631 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1633         PR rtl-optimization/60663
1634         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
1635         avoid 0 cost.
1637 2014-04-15  Richard Biener  <rguenther@suse.de>
1639         * lto-streamer.h (LTO_major_version): Bump to 4.
1641 2014-04-15  Richard Biener  <rguenther@suse.de>
1643         * common.opt (lto_partition_model): New enum.
1644         (flto-partition=): Merge separate options with a single with argument,
1645         add -flto-partition=one support.
1646         * flag-types.h (enum lto_partition_model): Declare.
1647         * opts.c (finish_options): Remove duplicate -flto-partition=
1648         option check.
1649         * lto-wrapper.c (run_gcc): Adjust.
1651 2014-04-15  Richard Biener  <rguenther@suse.de>
1653         * alias.c (ncr_compar): New function.
1654         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
1656 2014-04-15  Richard Biener  <rguenther@suse.de>
1658         * alias.c (record_component_aliases): Do not walk BINFOs.
1660 2014-04-15  Richard Biener  <rguenther@suse.de>
1662         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1663         Add struct function argument and adjust.
1664         (find_func_aliases_for_call): Likewise.
1665         (find_func_aliases): Likewise.
1666         (find_func_clobbers): Likewise.
1667         (intra_create_variable_infos): Likewise.
1668         (compute_points_to_sets): Likewise.
1669         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
1671 2014-04-15  Richard Biener  <rguenther@suse.de>
1673         * tree.c (iterative_hash_expr): Use enum tree_code_class
1674         to store TREE_CODE_CLASS.
1675         (tree_block): Likewise.
1676         (tree_set_block): Likewise.
1677         * tree.h (fold_build_pointer_plus_loc): Use
1678         convert_to_ptrofftype_loc.
1680 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
1682         PR plugins/59335
1683         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
1684         added in 4.9.
1686 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
1688         * cfgloop.h (struct loop): Move force_vectorize down.
1689         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
1690         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
1691         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
1692         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
1693         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
1694         * tree-core.h (enum annot_expr_kind): Add new kind values.
1695         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
1696         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
1697         kinds.
1698         * tree.def (ANNOTATE_EXPR): Tweak comment.
1700 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
1702         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
1703         cxa_pure_virtual).
1705 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
1707         * tree.h (TYPE_IDENTIFIER): Declare.
1708         * tree.c (subrange_type_for_debug_p): Use it.
1709         * godump.c (go_format_type): Likewise.
1710         * dwarf2out.c (is_cxx_auto, modified_type_die,
1711         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
1712         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
1714 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
1716         PR lto/60820
1717         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
1719 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
1721         * config/i386/i386.c (examine_argument): Return bool.  Return true if
1722         parameter should be passed in memory.
1723         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
1724         (construct_container): Update calls to examine_argument.
1725         (function_arg_advance_64): Ditto.
1726         (return_in_memory_32): Merge with ix86_return_in_memory.
1727         (return_in_memory_64): Ditto.
1728         (return_in_memory_ms_64): Ditto.
1730 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
1732         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
1733         * coverage.c (coverage_compute_profile_id): Handle externally visible
1734         symbols.
1736 2014-04-14  Martin Jambor  <mjambor@suse.cz>
1738         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
1739         DECL_DISREGARD_INLINE_LIMITS functions.
1741 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
1743         PR target/60827
1744         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
1746 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
1748         PR target/60827
1749         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
1750         optimize_insn_for_speed_p instead of
1751         optimize_function_for_speed_p.
1753 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
1755         * doc/invoke.texi (free): Document AArch64.
1757 2014-04-14  Richard Biener  <rguenther@suse.de>
1759         PR tree-optimization/60042
1760         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
1761         (insert_into_preds_of_block): Do not prevent PHI insertion
1762         for REFERENCE exprs here ...
1763         (eliminate_dom_walker::before_dom_children): ... but prevent
1764         their use here under similar conditions when applied to the
1765         IL after PRE optimizations.
1767 2014-04-14  Richard Biener  <rguenther@suse.de>
1769         * passes.def: Move early points-to after early SRA.
1771 2014-04-14  Richard Biener  <rguenther@suse.de>
1773         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
1774         check for which sign-changes we allow when forwarding
1775         a converted value into a switch.
1777 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
1779         * stor-layout.c (place_field): Finalize non-constant offset for the
1780         field, if any.
1782 2014-04-14  Richard Biener  <rguenther@suse.de>
1784         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
1785         as argument.
1786         (expand_switch_using_bit_tests_p): Likewise.
1787         (process_switch): Compute and pass on speed_p based on the
1788         switch stmt.
1789         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
1790         optimize_bb_for_speed_p.
1792 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
1794         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
1795         * function.h (struct function): Rename has_force_vect_loops into
1796         has_force_vectorize_loops.
1797         * lto-streamer-in.c (input_cfg): Adjust for renaming.
1798         (input_struct_function_base): Likewise.
1799         * lto-streamer-out.c (output_cfg): Likewise.
1800         (output_struct_function_base): Likewise.
1801         * omp-low.c (expand_omp_simd): Likewise.
1802         * tree-cfg.c (move_sese_region_to_fn): Likewise.
1803         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1804         (version_loop_for_if_conversion): Likewise.
1805         (tree_if_conversion): Likewise.
1806         (main_tree_if_conversion): Likewise.
1807         (gate_tree_if_conversion): Likewise.
1808         * tree-inline.c (copy_loops): Likewise.
1809         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
1810         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
1811         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
1812         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
1813         * tree-vectorizer.c (vectorize_loops): Likewise.
1814         * tree-vectorizer.h (unlimited_cost_model): Likewise.
1816 2014-04-14  Richard Biener  <rguenther@suse.de>
1818         PR lto/60720
1819         * lto-streamer-out.c (wrap_refs): New function.
1820         (lto_output): Wrap symbol references in global initializes in
1821         type-preserving MEM_REFs.
1823 2014-04-14  Christian Bruel  <christian.bruel@st.com>
1825         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
1827 2014-04-14  Christian Bruel  <christian.bruel@st.com>
1829         * config/sh/sh.md (setmemqi): New expand pattern.
1830         * config/sh/sh.h (CLEAR_RATIO): Define.
1831         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
1832         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
1834 2014-04-14  Richard Biener  <rguenther@suse.de>
1836         PR middle-end/55022
1837         * fold-const.c (negate_expr_p): Don't negate directional rounding
1838         division.
1839         (fold_negate_expr): Likewise.
1841 2014-04-14  Richard Biener  <rguenther@suse.de>
1843         PR tree-optimization/59817
1844         PR tree-optimization/60453
1845         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
1846         recursion to catch all CHRECs in the scalar evolution and restrict
1847         the predicate for the remains appropriately.
1849 2014-04-12  Catherine Moore  <clm@codesourcery.com>
1851         * config/mips/constraints.md: Add new register constraint "kb".
1852         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
1853         (*movhi_internal): Likewise.
1854         (*movqi_internal): Likewise.
1855         * config/mips/mips.h (M16_STORE_REGS): New register class.
1856         (REG_CLASS_NAMES): Add M16_STORE_REGS.
1857         (REG_CLASS_CONTENTS): Likewise.
1858         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
1860 2014-04-11  Tobias Burnus  <burnus@net-b.de>
1862         PR c/60194
1863         * doc/invoke.texi (-Wformat-signedness): Document it.
1864         (Wformat=2): Mention that this enables -Wformat-signedness.
1866 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
1868         * common/config/epiphany/epiphany-common.c
1869         (epiphany_option_optimization_table): Enable section anchors by
1870         default at -O1 or higher.
1871         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
1872         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
1873         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
1874         carries no extra cost.
1875         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
1876         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
1877         * config/epiphany/predicates.md (memclob_operand): New predicate.
1878         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
1879         Use memclob_operand predicate and X constraint for operand 3.
1881 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
1883         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
1884         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
1885         its operands.
1887 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
1889         PR rtl-optimization/60651
1890         * mode-switching.c (optimize_mode_switching): Make sure to emit
1891         sets of a lower numbered entity before sets of a higher numbered
1892         entity to a mode of the same or lower priority.
1893         When creating a seginfo for a basic block that starts with a code
1894         label, move the insertion point past the code label.
1895         (new_seginfo): Document and enforce requirement that
1896         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
1897         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
1898         * doc/tm.texi: Regenerate.
1900 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
1902         PR target/60811
1903         * config/arc/arc.c (arc_save_restore): Fix assert typo.
1905 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
1907         * BASE-VER: Set to 4.10.0.
1909 2014-04-11  Tobias Burnus  <burnus@net-b.de>
1911         PR other/59055
1912         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
1913         * doc/gcc.texi (Service): Update description in the @menu
1914         * doc/invoke.texi (Option Summary): Remove misplaced and
1915         duplicated @menu.
1917 2014-04-11  Steve Ellcey  <sellcey@mips.com>
1918             Jakub Jelinek  <jakub@redhat.com>
1920         PR middle-end/60556
1921         * expr.c (convert_move): Use emit_store_flag_force instead of
1922         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
1923         argument to it.
1925 2014-04-11  Richard Biener  <rguenther@suse.de>
1927         PR middle-end/60797
1928         * varasm.c (assemble_alias): Avoid endless error reporting
1929         recursion by setting TREE_ASM_WRITTEN.
1931 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1933         * config/s390/s390.md: Add a splitter for NOT rtx.
1935 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
1937         PR rtl-optimization/60663
1938         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
1940 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
1941             Jakub Jelinek  <jakub@redhat.com>
1943         PR lto/60567
1944         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
1945         flag from decl_node to node.
1947 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1949         PR debug/60655
1950         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
1951         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
1952         ameliorating the cases where it can be.
1954 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
1956         Revert
1957         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
1959         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1960         (loadsync_<mode>): Change mode.
1961         (load_quadpti, store_quadpti): New.
1962         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1963         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1964         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
1966 2014-04-09  Cong Hou  <congh@google.com>
1968         PR testsuite/60773
1969         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
1970         documentation.
1972 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1974         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
1975         instead of vnor to exploit possible fusion opportunity in the
1976         future.
1977         (altivec_expand_vec_perm_const_le): Likewise.
1979 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
1981         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1982         (loadsync_<mode>): Change mode.
1983         (load_quadpti, store_quadpti): New.
1984         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1985         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1987 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
1989         PR target/60763
1990         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
1991         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
1992         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
1994 2014-04-08  Richard Biener  <rguenther@suse.de>
1996         PR middle-end/60706
1997         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
1998         a 64bit widest int print double-int similar to on HWI64 hosts.
2000 2014-04-08  Richard Biener  <rguenther@suse.de>
2002         PR tree-optimization/60785
2003         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
2004         default defs properly.
2006 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
2008         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
2009         (Weffc++): Likewise.
2011 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
2013         * ipa-devirt.c (maybe_record_node): When node is not recorded,
2014         set completep to false rather than true.
2016 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
2018         PR target/60504
2019         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
2020         ARM_TARGET2_DWARF_FORMAT.
2022 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
2024         PR target/60609
2025         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
2026         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
2027         ADDR_DIFF_VEC.
2029 2014-04-07  Richard Biener  <rguenther@suse.de>
2031         PR tree-optimization/60766
2032         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
2033         (may_eliminate_iv): Convert cand_value_at result to desired type.
2035 2014-04-07  Jason Merrill  <jason@redhat.com>
2037         PR c++/60731
2038         * common.opt (-fno-gnu-unique): Add.
2039         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
2041 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2043         * haifa-sched.c: Fix outdated function reference and minor
2044         grammar errors in introductory comment.
2046 2014-04-07  Richard Biener  <rguenther@suse.de>
2048         PR middle-end/60750
2049         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
2050         for noreturn calls.
2051         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
2053 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
2055         PR debug/55794
2056         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
2057         size accounting for thunks.
2058         (pa_asm_output_mi_thunk): Use final_start_function() and
2059         final_end_function() to output function start and end directives.
2061 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
2063         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
2064         device specific ISA/ feature information. Remove short_sp and
2065         errata_skip ds.  Add avr_device_specific_features enum to have device
2066         specific info.
2067         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
2068         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
2069         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
2070         updated device specific info.
2071         * config/avr/avr-mcus.def: Merge device specific details to
2072         dev_attribute field.
2073         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
2074         errata_skip.
2075         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
2076         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
2077         assembler if RMW isa supported by current device.
2078         * config/avr/genmultilib.awk: Update as device info structure changed.
2079         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
2081 2014-04-04  Cong Hou  <congh@google.com>
2083         PR tree-optimization/60656
2084         * tree-vect-stmts.c (supportable_widening_operation):
2085         Fix a bug that elements in a vector with vect_used_by_reduction
2086         property are incorrectly reordered when the operation on it is not
2087         consistant with the one in reduction operation.
2089 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
2091         PR rtl-optimization/60155
2092         * gcse.c (record_set_data): New function.
2093         (single_set_gcse): New function.
2094         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
2095         (hoist_code): Likewise.
2096         (get_pressure_class_and_nregs): Likewise.
2098 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
2100         * explow.c (probe_stack_range): Emit a final optimization blockage.
2102 2014-04-04  Anthony Green  <green@moxielogic.com>
2104         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
2105         typos.
2107 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
2109         PR ipa/59626
2110         * lto-cgraph.c (input_overwrite_node): Check that partitioning
2111         flags are set only during streaming.
2112         * ipa.c (process_references, walk_polymorphic_call_targets,
2113         symtab_remove_unreachable_nodes): Drop bodies of always inline
2114         after early inlining.
2115         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
2117 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
2118         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2120         PR debug/60655
2121         * dwarf2out.c (const_ok_for_output_1): Reject expressions
2122         containing a NOT.
2124 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2126         PR bootstrap/60743
2127         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
2128         duration.
2129         (cortex_a53_fdivd): Likewise.
2131 2014-04-04  Martin Jambor  <mjambor@suse.cz>
2133         PR ipa/60640
2134         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
2135         Adjust all callers.
2136         * cgraph.c (clone_of_p): Also return true if thunks match.
2137         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
2138         cgraph_function_or_thunk_node and an obsolete comment.
2139         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
2140         file.
2141         (build_function_decl_skip_args): Likewise.
2142         (set_new_clone_decl_and_node_flags): New function.
2143         (duplicate_thunk_for_node): Likewise.
2144         (redirect_edge_duplicating_thunks): Likewise.
2145         (cgraph_clone_node): New parameter args_to_skip, pass it to
2146         redirect_edge_duplicating_thunks which is called instead of
2147         cgraph_redirect_edge_callee.
2148         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
2149         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
2151 2014-04-04  Jeff Law  <law@redhat.com>
2153         PR target/60657
2154         * config/arm/predicates.md (const_int_I_operand): New predicate.
2155         (const_int_M_operand): Similarly.
2156         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
2157         const_int_operand.
2158         (insv_t2, extv_reg, extzv_t2): Likewise.
2159         (load_multiple_with_writeback): Similarly for const_int_I_operand.
2160         (pop_multiple_with_writeback_and_return): Likewise.
2161         (vfp_pop_multiple_with_writeback): Likewise
2163 2014-04-04  Richard Biener  <rguenther@suse.de>
2165         PR ipa/60746
2166         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
2167         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
2168         non-GIMPLE_LABELs.
2169         * gimplify.h (gimple_add_tmp_var_fn): Declare.
2170         * gimplify.c (gimple_add_tmp_var_fn): New function.
2171         * gimple-expr.h (create_tmp_reg_fn): Declare.
2172         * gimple-expr.c (create_tmp_reg_fn): New function.
2173         * gimple-low.c (record_vars_into): Don't change cfun.
2174         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
2175         code generation without cfun.
2177 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
2179         PR bootstrap/60719
2180         * Makefile.in (install-driver): Fix shell scripting.
2182 2014-04-03  Cong Hou  <congh@google.com>
2184         PR tree-optimization/60505
2185         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
2186         threshold of number of iterations below which no vectorization
2187         will be done.
2188         * tree-vect-loop.c (new_loop_vec_info):
2189         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
2190         * tree-vect-loop.c (vect_analyze_loop_operations):
2191         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
2192         * tree-vect-loop.c (vect_transform_loop):
2193         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
2194         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
2195         of iterations of the loop and see if we should build the epilogue.
2197 2014-04-03  Richard Biener  <rguenther@suse.de>
2199         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
2200         (streamer_tree_cache_create): Adjust.
2201         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
2202         to allow optional nodes array.
2203         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
2204         (streamer_tree_cache_append): Likewise.
2205         (streamer_tree_cache_create): Create nodes array optionally
2206         as specified by parameter.
2207         * lto-streamer-out.c (create_output_block): Avoid maintaining
2208         the node array in the writer cache.
2209         (DFS_write_tree): Remove assertion.
2210         (produce_asm_for_decls): Free the out decl state hash table early.
2211         * lto-streamer-in.c (lto_data_in_create): Adjust for
2212         streamer_tree_cache_create prototype change.
2214 2014-04-03  Richard Biener  <rguenther@suse.de>
2216         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
2217         set TREE_CHAIN to NULL_TREE.
2219 2014-04-03  Richard Biener  <rguenther@suse.de>
2221         PR tree-optimization/60740
2222         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
2223         over all GIMPLE_COND operands.
2225 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
2227         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
2228         (Weffc++): Remove Scott's numbering, merge lists and reference
2229         Wnon-virtual-dtor.
2231 2014-04-03  Nick Clifton  <nickc@redhat.com>
2233         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
2234         properly.
2236 2014-04-03  Martin Jambor  <mjambor@suse.cz>
2238         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
2239         mention gcc_unreachable before failing.
2240         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
2241         removed symbols.
2243 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
2245         PR ipa/60659
2246         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
2247         inconsistent code and instead mark the context inconsistent.
2248         (possible_polymorphic_call_targets): For inconsistent contexts
2249         return empty complete list.
2251 2014-04-02  Anthony Green  <green@moxielogic.com>
2253         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
2254         (extendqisi2, extendhisi2): Define.
2255         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
2256         (WCHAR_TYPE): Change to unsigned int.
2258 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2260         PR tree-optimization/60733
2261         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
2262         insertion point for PHI candidates to be the end of the feeding
2263         block for the PHI argument.
2265 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
2267         PR rtl-optimization/60650
2268         * lra-constraints.c (process_alt_operands): Decrease reject for
2269         earlyclobber matching.
2271 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2273         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
2275 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2277         * config/spu/spu.c (pad_bb): Do not crash when the last
2278         insn is CODE_FOR_blockage.
2280 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2282         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
2283         lies outside the target mode.
2285 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2287         PR target/60735
2288         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
2289         software floating point or no floating point registers, do not
2290         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
2291         in GPRs that occurs after we tested for GPRs that would never be
2292         true.
2294         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
2295         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
2296         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
2297         specifically allow DDmode, since that does not use the SPE SIMD
2298         instructions.
2300 2014-04-02  Richard Biener  <rguenther@suse.de>
2302         PR middle-end/60729
2303         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
2304         MODE_INTs.  Properly use negv_optab.
2305         (expand_abs): Likewise.
2307 2014-04-02  Richard Biener  <rguenther@suse.de>
2309         PR bootstrap/60719
2310         * Makefile.in (install-driver): Guard extra installs with special
2311         names properly.
2313 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
2315         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2316         Document vec_vgbbd.
2318 2014-04-01  Richard Henderson  <rth@redhat.com>
2320         PR target/60704
2321         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
2322         alternative enabled before register allocation.
2324 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
2326         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
2327         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
2328         typo.
2329         (nios2_large_got_address): Remove unneeded 'sym' parameter.
2330         (nios2_got_address): Update nios2_large_got_address call site.
2331         (nios2_delegitimize_address): New function.
2332         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
2333         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
2334         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
2336 2014-04-01  Martin Husemann  <martin@duskware.de>
2338         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
2339         for -mabi=32.
2341 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
2343         PR rtl-optimization/60604
2344         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
2345         check from register_operand.
2346         (register_operand): Redefine in terms of general_operand.
2347         (nonmemory_operand): Use register_operand for the non-constant cases.
2349 2014-04-01  Richard Biener  <rguenther@suse.de>
2351         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
2353 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2355         * doc/invoke.texi (mapp-regs): Clarify.
2357 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
2359         * config/i386/avx512fintrin.h (__v32hi): Define type.
2360         (__v64qi): Likewise.
2361         (_mm512_set1_epi8): Define.
2362         (_mm512_set1_epi16): Define.
2363         (_mm512_set4_epi32): Define.
2364         (_mm512_set4_epi64): Define.
2365         (_mm512_set4_pd): Define.
2366         (_mm512_set4_ps): Define.
2367         (_mm512_setr4_epi64): Define.
2368         (_mm512_setr4_epi32): Define.
2369         (_mm512_setr4_pd): Define.
2370         (_mm512_setr4_ps): Define.
2371         (_mm512_setzero_epi32): Define.
2373 2014-03-31  Martin Jambor  <mjambor@suse.cz>
2375         PR middle-end/60647
2376         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
2377         callsite_arguments_match_p.  Updated all callers.  Also check types of
2378         corresponding formal parameters and actual arguments.
2379         (not_all_callers_have_enough_arguments_p) Renamed to
2380         some_callers_have_mismatched_arguments_p.
2382 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
2384         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
2386 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
2388         PR target/60034
2389         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
2390         section anchor.
2392 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
2394         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
2395         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
2396         Split out
2397         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
2398         Use FMAMODE_NOVF512 mode iterator.
2399         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
2400         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
2401         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
2402         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
2403         Split out
2404         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
2405         Use VF_128_256 mode iterator.
2406         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
2407         Ditto.
2409 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2411         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
2412         static chain if needed.
2414 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
2416         PR target/60697
2417         * lra-constraints.c (index_part_to_reg): New.
2418         (process_address): Use it.
2420 2014-03-27  Jeff Law  <law@redhat.com>
2421             Jakub Jelinek  <jakub@redhat.com>
2423         PR target/60648
2424         * expr.c (do_tablejump): Use simplify_gen_binary rather than
2425         gen_rtx_{PLUS,MULT} to build up the address expression.
2427         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
2428         creating non-canonical RTL.
2430 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2432         PR ipa/60243
2433         * ipa-inline.c (want_inline_small_function_p): Short circuit large
2434         functions; reorganize to make cheap checks first.
2435         (inline_small_functions): Do not estimate growth when dumping;
2436         it is expensive.
2437         * ipa-inline.h (inline_summary): Add min_size.
2438         (growth_likely_positive): New function.
2439         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
2440         (set_cond_stmt_execution_predicate): Cleanup.
2441         (estimate_edge_size_and_time): Compute min_size.
2442         (estimate_calls_size_and_time): Likewise.
2443         (estimate_node_size_and_time): Likewise.
2444         (inline_update_overall_summary): Update min_size.
2445         (do_estimate_edge_time): Likewise.
2446         (do_estimate_edge_size): Update.
2447         (do_estimate_edge_hints): Update.
2448         (growth_likely_positive): New function.
2450 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
2452         PR target/60693
2453         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
2454         also if addr has VOIDmode.
2456 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2458         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
2459         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
2460         Declare extern.
2461         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
2462         instructions as well as AdvancedSIMD loads.
2464 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2466         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
2467         Use crypto_aese type.
2468         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
2469         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
2470         crypto_aese, crypto_aesmc.  Move to types.md.
2471         * config/arm/types.md (crypto_aes): Split into crypto_aese,
2472         crypto_aesmc.
2473         * config/arm/iterators.md (crypto_type): Likewise.
2475 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2477         * cgraph.c: Include expr.h and tree-dfa.h.
2478         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
2479         remove LHS.
2481 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
2483         PR target/60675
2484         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
2485         regs from checking multi-reg pseudos.
2487 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2489         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
2491 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2493         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
2494         if it would clobber the stack pointer, even temporarily.
2496 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
2498         * mode-switching.c: Make small adjustments to the top comment.
2500 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
2502         * config/rs6000/constraints.md (wD constraint): New constraint to
2503         match the constant integer to get the top DImode/DFmode out of a
2504         vector in a VSX register.
2506         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
2507         match the constant integer to get the top DImode/DFmode out of a
2508         vector in a VSX register.
2510         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
2511         for ISA 2.07.
2513         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2514         vbpermq builtins.
2516         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
2517         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
2519         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
2520         Optimize vec_extract of 64-bit values, where the value being
2521         extracted is in the top word, where we can use scalar
2522         instructions.  Add direct move and store support.  Combine the big
2523         endian/little endian vector select load support into a single insn.
2524         (vsx_extract_<mode>_internal1): Likewise.
2525         (vsx_extract_<mode>_internal2): Likewise.
2526         (vsx_extract_<mode>_load): Likewise.
2527         (vsx_extract_<mode>_store): Likewise.
2528         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
2529         combined into vsx_extract_<mode>_load.
2530         (vsx_extract_<mode>_one_le): Likewise.
2532         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
2533         define the top 64-bit vector element.
2535         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
2536         constraint.
2538         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2539         Document vec_vbpermq builtin.
2541         PR target/60672
2542         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
2543         enable use of xxsldwi and xxpermdi builtin functions.
2544         (vec_xxpermdi): Likewise.
2546         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2547         Document use of vec_xxsldwi and vec_xxpermdi builtins.
2549 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
2551         PR rtl-optimization/60650
2552         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
2553         first_p.  Use it.
2554         (find_spills_for): New.
2555         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
2556         Spill all pseudos on the second iteration.
2558 2014-03-27  Marek Polacek  <polacek@redhat.com>
2560         PR c/50347
2561         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
2562         types.
2564 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2566         * config/s390/s390.c (s390_can_use_return_insn): Check for
2567         call-saved FPRs on 31 bit.
2569 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
2571         PR middle-end/60682
2572         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
2573         if they need regimplification, just drop them instead of
2574         calling gimple_regimplify_operands on them.
2576 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2578         PR target/60580
2579         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
2580         (aarch64_frame_pointer_required): Adjust logic.
2581         (aarch64_can_eliminate): Adjust logic.
2582         (aarch64_override_options_after_change): Adjust logic.
2584 2014-03-27  Dehao Chen  <dehao@google.com>
2586         * ipa-inline.c (early_inliner): Update node's inline info.
2588 2014-03-26  Dehao Chen  <dehao@google.com>
2590         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
2591         compiler inserted conditional jumps for NAN float check.
2593 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
2595         * ubsan.h (ubsan_create_data): Change second argument's type
2596         to const location_t *.
2597         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
2598         _("<unknown>").
2599         (ubsan_create_data): Change second argument to const location_t *PLOC.
2600         Create Loc field whenever PLOC is non-NULL.
2601         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
2602         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
2603         callers.
2605         PR other/59545
2606         * real.c (real_to_integer2): Change type of low to UHWI.
2608 2014-03-26  Tobias Burnus  <burnus@net-b.de>
2610         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
2611         (CILK_SELF_SPECS): New define.
2612         (driver_self_specs): Use it.
2614 2014-03-26  Richard Biener  <rguenther@suse.de>
2616         * tree-pretty-print.c (percent_K_format): Implement special
2617         case for LTO and its stripped down BLOCK tree.
2619 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
2621         PR sanitizer/60636
2622         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
2624         * tree-vrp.c (simplify_internal_call_using_ranges): If only
2625         one range is range_int_cst_p, but not both, at least optimize
2626         addition/subtraction of 0 and multiplication by 0 or 1.
2627         * gimple-fold.c (gimple_fold_call): Fold
2628         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
2629         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
2630         INTEGER_CSTs, try to fold at least x * 0 and y - y.
2632 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
2634         PR rtl-optimization/60452
2635         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
2636         <case REG>: Return 1 for invalid offsets from the frame pointer.
2638 2014-03-26  Marek Polacek  <polacek@redhat.com>
2640         PR c/37428
2641         * doc/extend.texi (C Extensions): Mention variable-length arrays in
2642         a structure/union.
2644 2014-03-26  Marek Polacek  <polacek@redhat.com>
2646         PR c/39525
2647         * doc/extend.texi (Designated Inits): Describe what happens to omitted
2648         field members.
2650 2014-03-26  Marek Polacek  <polacek@redhat.com>
2652         PR other/59545
2653         * ira-color.c (update_conflict_hard_regno_costs): Perform the
2654         multiplication in unsigned type.
2656 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
2658         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
2660 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
2662         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
2664 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
2666         PR ipa/60315
2667         * cif-code.def (UNREACHABLE) New code.
2668         * ipa-inline.c (inline_small_functions): Skip edges to
2669         __builtlin_unreachable.
2670         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
2671         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
2672         predicate to __bulitin_unreachable.
2673         (set_cond_stmt_execution_predicate): Fix issue when
2674         invert_tree_comparison returns ERROR_MARK.
2675         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
2676         propagate to inline clones.
2677         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
2678         to unreachable.
2679         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
2680         * cgraphclones.c (cgraph_clone_node): If call destination is already
2681         ureachable, do not redirect it back.
2682         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
2683         unreachable.
2685 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
2687         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
2688         Do not modify inline clones.
2690 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
2692         * config/i386/i386.md (general_sext_operand): New mode attr.
2693         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
2694         don't generate (sign_extend (const_int)).
2695         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
2696         operands[2].  Use We constraint instead of <i> and
2697         <general_sext_operand> predicate instead of <general_operand>.
2698         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
2699         * config/i386/constraints.md (We): New constraint.
2700         * config/i386/predicates.md (x86_64_sext_operand,
2701         sext_operand): New predicates.
2703 2014-03-25  Martin Jambor  <mjambor@suse.cz>
2705         PR ipa/60600
2706         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
2707         inconsistent devirtualizations to __builtin_unreachable.
2709 2014-03-25  Marek Polacek  <polacek@redhat.com>
2711         PR c/35449
2712         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
2714 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
2716         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
2717         order of elements for big-endian.
2719 2014-03-25  Richard Biener  <rguenther@suse.de>
2721         PR middle-end/60635
2722         * gimplify-me.c (gimple_regimplify_operands): Update the
2723         re-gimplifed stmt.
2725 2014-03-25  Martin Jambor  <mjambor@suse.cz>
2727         PR ipa/59176
2728         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
2729         (lto_output_varpool_node): Likewise.
2730         (input_overwrite_node): Likewise.
2731         (input_varpool_node): Likewise.
2733 2014-03-25  Richard Biener  <rguenther@suse.de>
2735         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
2736         (run_gcc): Likewise.
2738 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
2740         * combine.c (simplify_compare_const): Add MODE argument.
2741         Handle mode_width 0 as very large mode_width.
2742         (try_combine, simplify_comparison): Adjust callers.
2744         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
2745         type to avoid signed integer overflow.
2746         * explow.c (plus_constant): Likewise.
2748 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
2750         * doc/generic.texi: Correct typos.
2752 2014-03-24  Tobias Burnus  <burnus@net-b.de>
2754         * doc/invoke.texi (-flto): Expand section about
2755         using static libraries with LTO.
2757 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2759         PR rtl-optimization/60501
2760         * optabs.def (addptr3_optab): New optab.
2761         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
2762         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
2763         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
2765         * lra.c (emit_add3_insn): Use the addptr pattern if available.
2767         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
2769 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
2771         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
2772         _mm512_set1_pd.
2774         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
2775         (_mm256_undefined_ps): Define.
2776         (_mm256_undefined_pd): Define.
2777         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
2778         (_mm_undefined_pd): Define.
2779         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
2780         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
2781         (_mm512_undefined_ps): Define.
2782         (_mm512_undefined_pd): Define.
2783         Use _mm*_undefined_*.
2784         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
2786 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
2788         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
2789         (lshr_simd): DI mode added.
2790         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
2791         (aarch64_ushr_simddi): Likewise.
2792         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
2793         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
2794         (vshrd_n_u64): Likewise.
2796 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2798         * Makefile.in (s-macro_list): Depend on cc1.
2800 2014-03-23  Teresa Johnson  <tejohnson@google.com>
2802         * ipa-utils.c (ipa_print_order): Use specified dump file.
2804 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
2806         PR rtl-optimization/60601
2807         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
2809         * gcc.c (eval_spec_function): Initialize save_growing_value.
2811 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
2813         PR sanitizer/60613
2814         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
2815         code == MINUS_EXPR, never swap op0 with op1.
2817         * toplev.c (init_local_tick): Avoid signed integer multiplication
2818         overflow.
2819         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
2820         shift by first operand's bitsize.
2822 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
2824         PR target/60610
2825         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
2826         redefine to 1 or 0.
2827         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
2828         TARGET_ISA_64BIT_P(x).
2830 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2832         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
2833         pattern for vector nor instead of subtract from splat(-1).
2834         (altivec_expand_vec_perm_const_le): Likewise.
2836 2014-03-21  Richard Henderson  <rth@twiddle.net>
2838         PR target/60598
2839         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
2840         related insns after epilogue_completed.
2842 2014-03-21  Martin Jambor  <mjambor@suse.cz>
2844         PR ipa/59176
2845         * cgraph.h (symtab_node): New flag body_removed.
2846         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
2847         when removing bodies.
2848         * symtab.c (dump_symtab_base): Dump body_removed flag.
2849         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
2850         had their bodies removed.
2852 2014-03-21  Martin Jambor  <mjambor@suse.cz>
2854         PR ipa/60419
2855         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
2856         in the border.
2858 2014-03-21  Richard Biener  <rguenther@suse.de>
2860         PR tree-optimization/60577
2861         * tree-core.h (struct tree_base): Document nothrow_flag use
2862         in VAR_DECL_NONALIASED.
2863         * tree.h (VAR_DECL_NONALIASED): New.
2864         (may_be_aliased): Adjust.
2865         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
2867 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
2869         * expr.c (expand_expr_real_1): Remove outdated comment.
2871 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
2873         PR middle-end/60597
2874         * ira.c (adjust_cleared_regs): Call copy_rtx on
2875         *reg_equiv[REGNO (loc)].src_p before passing it to
2876         simplify_replace_fn_rtx.
2878         PR target/60568
2879         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
2880         into CONST, put pic register as first operand of PLUS.  Use
2881         gen_const_mem for both 32-bit and 64-bit PIC got loads.
2883 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2885         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
2887 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
2889         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
2890         around for store forwarding issue in the FPU on the UT699.
2891         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
2892         loads and operations if -mfix-ut699 is specified.
2893         (divtf3_hq): Tweak attribute.
2894         (sqrttf2_hq): Likewise.
2896 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
2898         * calls.c (store_one_arg): Remove incorrect const qualification on the
2899         type of the temporary.
2900         * cfgexpand.c (expand_return): Likewise.
2901         * expr.c (expand_constructor): Likewise.
2902         (expand_expr_real_1): Likewise.
2904 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2906         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
2907         of parts.
2909 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
2911         PR target/60039
2912         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
2914 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
2916         * config/arm/aarch-common-protos.h
2917         (alu_cost_table): Fix spelling of "extend".
2918         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
2920 2014-03-19  Richard Biener  <rguenther@suse.de>
2922         PR middle-end/60553
2923         * tree-core.h (tree_type_common): Re-order pointer members
2924         to reduce recursion depth during GC walks.
2926 2014-03-19  Marek Polacek  <polacek@redhat.com>
2928         PR sanitizer/60569
2929         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
2930         before accessing it.
2932 2014-03-19  Richard Biener  <rguenther@suse.de>
2934         PR lto/59543
2935         * lto-streamer-in.c (input_function): In WPA stage do not drop
2936         debug stmts.
2938 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
2940         PR tree-optimization/60559
2941         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
2942         with build_zero_cst assignment.
2944 2014-03-18  Kai Tietz  <ktietz@redhat.com>
2946         PR rtl-optimization/56356
2947         * sdbout.c (sdbout_parms): Verify that parms'
2948         incoming argument is valid.
2949         (sdbout_reg_parms): Likewise.
2951 2014-03-18  Richard Henderson  <rth@redhat.com>
2953         PR target/60562
2954         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
2955         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
2956         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
2958 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
2960         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
2961         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
2962         Italicize plugin event names in description.  Explain that
2963         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
2964         Remind that no GCC functions should be called after PLUGIN_FINISH.
2965         Explain what pragmas with expansion are.
2967 2014-03-18  Martin Liska  <mliska@suse.cz>
2969         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
2970         gimple call statement is update.
2971         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
2972         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
2974 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
2976         PR sanitizer/60557
2977         * ubsan.c (ubsan_instrument_unreachable): Call
2978         initialize_sanitizer_builtins.
2979         (ubsan_pass): Likewise.
2981         PR sanitizer/60535
2982         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
2983         varpool_finalize_decl instead of rest_of_decl_compilation.
2985 2014-03-18  Richard Biener  <rguenther@suse.de>
2987         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
2988         by using bitmap_and_compl instead of bitmap_and_compl_into.
2989         (df_rd_transfer_function): Likewise.
2991 2014-03-18  Richard Biener  <rguenther@suse.de>
2993         * doc/lto.texi (fresolution): Fix typo.
2995 2014-03-18  Richard Biener  <rguenther@suse.de>
2997         * doc/invoke.texi (flto): Update for changes in 4.9.
2999 2014-03-18  Richard Biener  <rguenther@suse.de>
3001         * doc/loop.texi: Remove section on the removed lambda framework.
3002         Update loop docs with recent changes in preserving loop structure.
3004 2014-03-18  Richard Biener  <rguenther@suse.de>
3006         * doc/lto.texi (-fresolution): Document.
3008 2014-03-18  Richard Biener  <rguenther@suse.de>
3010         * doc/contrib.texi: Adjust my name.
3012 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3014         PR ipa/58721
3015         * internal-fn.c: Include diagnostic-core.h.
3016         (expand_BUILTIN_EXPECT): New function.
3017         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
3018         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
3019         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
3020         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
3021         IFN_BUILTIN_EXPECT.
3022         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
3023         Revert 3 argument __builtin_expect code.
3024         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
3025         * gimple-fold.c (gimple_fold_call): Likewise.
3026         * tree.h (fold_builtin_expect): New prototype.
3027         * builtins.c (build_builtin_expect_predicate): Add predictor
3028         argument, if non-NULL, create 3 argument __builtin_expect.
3029         (fold_builtin_expect): No longer static.  Add ARG2 argument,
3030         pass it through to build_builtin_expect_predicate.
3031         (fold_builtin_2): Adjust caller.
3032         (fold_builtin_3): Handle BUILT_IN_EXPECT.
3033         * internal-fn.def (BUILTIN_EXPECT): New.
3035 2014-03-18  Tobias Burnus  <burnus@net-b.de>
3037         PR ipa/58721
3038         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
3039         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
3040         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
3042 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
3044         PR ipa/58721
3045         * predict.c (combine_predictions_for_bb): Fix up formatting.
3046         (expr_expected_value_1, expr_expected_value): Add predictor argument,
3047         fill what it points to if non-NULL.
3048         (tree_predict_by_opcode): Adjust caller, use the predictor.
3049         * predict.def (PRED_COMPARE_AND_SWAP): Add.
3051 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
3053         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
3054         proper constant for the store mode.
3056 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
3058         * symtab.c (change_decl_assembler_name): Fix transparent alias
3059         chain construction.
3061 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
3063         * config/aarch64/aarch64.c: Correct the comments about the
3064         aarch64 stack layout.
3066 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
3068         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
3069         check for GF_OMP_FOR_KIND_FOR.
3071 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
3073         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
3074         ymm and zmm register names.
3076 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
3078         PR target/60516
3079         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
3080         note creation for the 2010-08-31 changes.
3082 2014-03-17  Marek Polacek  <polacek@redhat.com>
3084         PR middle-end/60534
3085         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
3086         as -fno-tree-loop-vectorize.
3087         (expand_omp_simd): Likewise.
3089 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
3091         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
3092         (eligible_for_call_delay): New prototype.
3093         * config/sparc/sparc.c (tls_call_delay): Rename into...
3094         (eligible_for_call_delay): ...this.  Return false if the instruction
3095         cannot be put in the delay slot of a branch.
3096         (eligible_for_restore_insn): Simplify.
3097         (eligible_for_return_delay): Return false if the instruction cannot be
3098         put in the delay slot of a branch and simplify.
3099         (eligible_for_sibcall_delay): Return false if the instruction cannot be
3100         put in the delay slot of a branch.
3101         * config/sparc/sparc.md (fix_ut699): New attribute.
3102         (tls_call_delay): Delete.
3103         (in_call_delay): Reimplement.
3104         (eligible_for_sibcall_delay): Rename into...
3105         (in_sibcall_delay): ...this.
3106         (eligible_for_return_delay): Rename into...
3107         (in_return_delay): ...this.
3108         (in_branch_delay): Reimplement.
3109         (in_uncond_branch_delay): Delete.
3110         (in_annul_branch_delay): Delete.
3112 2014-03-14  Richard Henderson  <rth@redhat.com>
3114         PR target/60525
3115         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
3116         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
3117         (*floathi<X87MODEF>2_i387_with_temp): Remove.
3118         (floathi splitters): Remove.
3119         (float<SWI48x>xf2): New pattern.
3120         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
3121         code that tried to handle DImode for 32-bit, but which was excluded
3122         by the pattern's condition.  Drop allocation of stack temporary.
3123         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
3124         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
3125         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
3126         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
3127         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
3128         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
3129         (*float<SWI48><MODEF>2_sse_interunit): Remove.
3130         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
3131         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
3132         (*float<SWI48x><X87MODEF>2_i387): Remove.
3133         (all float _with_temp splitters): Remove.
3134         (*float<SWI48x><MODEF>2_i387): New pattern.
3135         (*float<SWI48><MODEF>2_sse): New pattern.
3136         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
3137         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
3139 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
3140             Marek Polacek  <polacek@redhat.com>
3142         PR middle-end/60484
3143         * common.opt (dump_base_name_prefixed): New Variable.
3144         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
3145         if x_dump_base_name_prefixed is already set, set it at the end.
3147 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
3149         PR rtl-optimization/60508
3150         * lra-constraints.c (get_reload_reg): Add new parameter
3151         in_subreg_p.
3152         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
3153         Pass the new parameter values.
3155 2014-03-14  Richard Biener  <rguenther@suse.de>
3157         * common.opt: Revert unintented changes from r205065.
3158         * opts.c: Likewise.
3160 2014-03-14  Richard Biener  <rguenther@suse.de>
3162         PR middle-end/60518
3163         * cfghooks.c (split_block): Properly adjust all loops the
3164         block was a latch of.
3166 2014-03-14  Martin Jambor  <mjambor@suse.cz>
3168         PR lto/60461
3169         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
3170         and simplify it.
3172 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
3174         PR target/59396
3175         * config/avr/avr.c (avr_set_current_function): Pass function name
3176         through default_strip_name_encoding before sanity checking instead
3177         of skipping the first char of the assembler name.
3179 2014-03-13  Richard Henderson  <rth@redhat.com>
3181         PR debug/60438
3182         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
3183         (ix86_force_to_memory, ix86_free_from_memory): Remove.
3184         * config/i386/i386-protos.h: Likewise.
3185         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
3186         in the expander instead of a splitter.
3187         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
3188         any possibility of requiring a memory.
3189         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
3190         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
3191         (fp branch splitters): Update for ix86_split_fp_branch.
3192         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
3193         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
3194         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
3195         (*fop_<MODEF>_2_i387): Remove f/r alternative.
3196         (*fop_<MODEF>_3_i387): Likewise.
3197         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
3198         (splitters for the fop_* register patterns): Remove.
3199         (fscalexf4_i387): Rename from *fscalexf4_i387.
3200         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
3202 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3204         PR tree-optimization/59779
3205         * tree-dfa.c (get_ref_base_and_extent): Use double_int
3206         type for bitsize and maxsize instead of HOST_WIDE_INT.
3208 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
3210         PR rtl-optimization/57320
3211         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
3212         the CFG after thread_prologue_and_epilogue_insns.
3214 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
3216         PR rtl-optimization/57189
3217         * lra-constraints.c (process_alt_operands): Disfavor spilling
3218         vector pseudos.
3220 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
3222         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
3224 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3226         PR tree-optimization/59025
3227         PR middle-end/60418
3228         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
3229         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
3231 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
3233         PR target/60486
3234         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
3235         calls of avr_out_plus_1.
3237 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
3239         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
3240         BB's single pred and update the father loop's latch info later.
3242 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
3244         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
3245         (VEC_M): Likewise.
3246         (VEC_N): Likewise.
3247         (VEC_R): Likewise.
3248         (VEC_base): Likewise.
3249         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
3250         registers, we need to swap double words in little endian mode.
3252         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
3253         to be a container mode for 128-bit integer operations added in ISA
3254         2.07.  Unlike TImode and PTImode, the preferred register set is
3255         the Altivec/VMX registers for the 128-bit operations.
3257         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
3258         declarations.
3259         (rs6000_split_128bit_ok_p): Likewise.
3261         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
3262         macros for creating ISA 2.07 normal and overloaded builtin
3263         functions with 3 arguments.
3264         (BU_P8V_OVERLOAD_3): Likewise.
3265         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
3266         for use as overloaded functions.
3267         (VPERM_1TI_UNS): Likewise.
3268         (VSEL_1TI): Likewise.
3269         (VSEL_1TI_UNS): Likewise.
3270         (ST_INTERNAL_1ti): Likewise.
3271         (LD_INTERNAL_1ti): Likewise.
3272         (XXSEL_1TI): Likewise.
3273         (XXSEL_1TI_UNS): Likewise.
3274         (VPERM_1TI): Likewise.
3275         (VPERM_1TI_UNS): Likewise.
3276         (XXPERMDI_1TI): Likewise.
3277         (SET_1TI): Likewise.
3278         (LXVD2X_V1TI): Likewise.
3279         (STXVD2X_V1TI): Likewise.
3280         (VEC_INIT_V1TI): Likewise.
3281         (VEC_SET_V1TI): Likewise.
3282         (VEC_EXT_V1TI): Likewise.
3283         (EQV_V1TI): Likewise.
3284         (NAND_V1TI): Likewise.
3285         (ORC_V1TI): Likewise.
3286         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
3287         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
3288         overloaded builtin.
3289         (VADDUQM): Likewise.
3290         (VSUBCUQ): Likewise.
3291         (VADDEUQM): Likewise.
3292         (VADDECUQ): Likewise.
3293         (VSUBEUQM): Likewise.
3294         (VSUBECUQ): Likewise.
3296         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
3297         __int128_t and __uint128_t types.
3298         (__uint128_type): Likewise.
3299         (altivec_categorize_keyword): Add support for vector __int128_t,
3300         vector __uint128_t, vector __int128, and vector unsigned __int128
3301         as a container type for TImode operations that need to be done in
3302         VSX/Altivec registers.
3303         (rs6000_macro_to_expand): Likewise.
3304         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
3305         to support 128-bit integer instructions vaddcuq, vadduqm,
3306         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
3307         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
3309         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
3310         for V1TImode, and set up preferences to use VSX/Altivec registers.
3311         Setup VSX reload handlers.
3312         (rs6000_debug_reg_global): Likewise.
3313         (rs6000_init_hard_regno_mode_ok): Likewise.
3314         (rs6000_preferred_simd_mode): Likewise.
3315         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
3316         (easy_altivec_constant): Likewise.
3317         (output_vec_const_move): Likewise.
3318         (rs6000_expand_vector_set): Convert V1TImode set and extract to
3319         simple move.
3320         (rs6000_expand_vector_extract): Likewise.
3321         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
3322         addressing.
3323         (rs6000_const_vec): Add support for V1TImode.
3324         (rs6000_emit_le_vsx_load): Swap double words when loading or
3325         storing TImode/V1TImode.
3326         (rs6000_emit_le_vsx_store): Likewise.
3327         (rs6000_emit_le_vsx_move): Likewise.
3328         (rs6000_emit_move): Add support for V1TImode.
3329         (altivec_expand_ld_builtin): Likewise.
3330         (altivec_expand_st_builtin): Likewise.
3331         (altivec_expand_vec_init_builtin): Likewise.
3332         (altivec_expand_builtin): Likewise.
3333         (rs6000_init_builtins): Add support for V1TImode type.  Add
3334         support for ISA 2.07 128-bit integer builtins.  Define type names
3335         for the VSX/Altivec vector types.
3336         (altivec_init_builtins): Add support for overloaded vector
3337         functions with V1TImode type.
3338         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
3339         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
3340         external function.
3341         (rs6000_split_128bit_ok_p): Likewise.
3342         (rs6000_handle_altivec_attribute): Create V1TImode from vector
3343         __int128_t and vector __uint128_t.
3345         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
3346         and mode attributes.
3347         (VSX_M): Likewise.
3348         (VSX_M2): Likewise.
3349         (VSm): Likewise.
3350         (VSs): Likewise.
3351         (VSr): Likewise.
3352         (VSv): Likewise.
3353         (VS_scalar): Likewise.
3354         (VS_double): Likewise.
3355         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
3357         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
3358         we support the ISA 2.07 128-bit integer arithmetic instructions.
3359         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
3360         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
3361         and TImode types for use with the builtin functions.
3362         (V1TI_type_node): Likewise.
3363         (unsigned_V1TI_type_node): Likewise.
3364         (intTI_type_internal_node): Likewise.
3365         (uintTI_type_internal_node): Likewise.
3367         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
3368         128-bit builtin functions.
3369         (UNSPEC_VADDEUQM): Likewise.
3370         (UNSPEC_VADDECUQ): Likewise.
3371         (UNSPEC_VSUBCUQ): Likewise.
3372         (UNSPEC_VSUBEUQM): Likewise.
3373         (UNSPEC_VSUBECUQ): Likewise.
3374         (VM): Add V1TImode to vector mode iterators.
3375         (VM2): Likewise.
3376         (VI_unit): Likewise.
3377         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
3378         (altivec_vaddcuq): Likewise.
3379         (altivec_vsubuqm): Likewise.
3380         (altivec_vsubcuq): Likewise.
3381         (altivec_vaddeuqm): Likewise.
3382         (altivec_vaddecuq): Likewise.
3383         (altivec_vsubeuqm): Likewise.
3384         (altivec_vsubecuq): Likewise.
3386         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
3387         mode iterators.
3388         (BOOL_128): Likewise.
3389         (BOOL_REGS_OUTPUT): Likewise.
3390         (BOOL_REGS_OP1): Likewise.
3391         (BOOL_REGS_OP2): Likewise.
3392         (BOOL_REGS_UNARY): Likewise.
3393         (BOOL_REGS_AND_CR0): Likewise.
3395         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
3396         128-bit integer builtin support.
3397         (vec_vadduqm): Likewise.
3398         (vec_vaddecuq): Likewise.
3399         (vec_vaddeuqm): Likewise.
3400         (vec_vsubecuq): Likewise.
3401         (vec_vsubeuqm): Likewise.
3402         (vec_vsubcuq): Likewise.
3403         (vec_vsubuqm): Likewise.
3405         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3406         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
3407         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
3408         128-bit integer add/subtract to ISA 2.07.
3410 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
3412         * config/arc/arc.c (arc_predicate_delay_insns):
3413         Fix third argument passed to conditionalize_nonjump.
3415 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
3417         * config/aarch64/aarch64-builtins.c
3418         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
3419         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
3420         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
3421         instead of __builtin_lfloor.
3422         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
3424 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3426         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
3427         (tree_ssa_ifcombine_bb_1): New function.
3428         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
3429         is an empty forwarder block to then_bb or vice versa and then_bb
3430         and else_bb are effectively swapped.
3432 2014-03-12  Christian Bruel  <christian.bruel@st.com>
3434         PR target/60264
3435         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
3436         REG_CFA_DEF_CFA note.
3437         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
3438         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
3440 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
3442         PR tree-optimization/60454
3443         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
3445 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3447         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
3448         Do not define target_cpu_default2 to generic.
3449         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
3450         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
3451         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
3453 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3454             Marc Glisse  <marc.glisse@inria.fr>
3456         PR tree-optimization/60502
3457         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
3458         instead of build_low_bits_mask.
3460 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3462         PR middle-end/60482
3463         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
3464         if there are multiple uses, but op doesn't live on E edge.
3465         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
3466         clobber stmts before __builtin_unreachable.
3468 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
3470         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
3471         hard_frame_pointer_rtx.
3472         * cse.c (cse_insn): Remove volatile check.
3473         * cselib.c (cselib_process_insn): Likewise.
3474         * dse.c (scan_insn): Likewise.
3476 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3478         * config/arc/arc.c (conditionalize_nonjump): New function,
3479         broken out of ...
3480         (arc_ifcvt): ... this.
3481         (arc_predicate_delay_insns): Use it.
3483 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3485         * config/arc/predicates.md (extend_operand): During/after reload,
3486         allow const_int_operand.
3487         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
3488         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
3489         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
3490         to "i".
3491         (umulsi3_highpart_i): Likewise.
3493 2014-03-11  Richard Biener  <rguenther@suse.de>
3495         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
3496         Add asserts to guard possible wrong-code bugs.
3498 2014-03-11  Richard Biener  <rguenther@suse.de>
3500         PR tree-optimization/60429
3501         PR tree-optimization/60485
3502         * tree-ssa-structalias.c (set_union_with_increment): Properly
3503         take into account all fields that overlap the shifted vars.
3504         (do_sd_constraint): Likewise.
3505         (do_ds_constraint): Likewise.
3506         (get_constraint_for_ptr_offset): Likewise.
3508 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
3510         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
3511         (nios2_compute_frame_layout):
3512         Add calculation of cfun->machine->fp_save_offset.
3513         (nios2_expand_prologue): Correct setting of frame pointer register
3514         in prologue.
3515         (nios2_expand_epilogue): Update recovery of stack pointer from
3516         frame pointer accordingly.
3517         (nios2_initial_elimination_offset): Update calculation of offset
3518         for eliminating to HARD_FRAME_POINTER_REGNUM.
3520 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
3522         PR ipa/60457
3523         * ipa.c (symtab_remove_unreachable_nodes): Don't call
3524         cgraph_get_create_node on VAR_DECLs.
3526 2014-03-10  Richard Biener  <rguenther@suse.de>
3528         PR middle-end/60474
3529         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
3531 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
3533         * config/vms/vms.opt (vms_float_format): New variable.
3535 2014-03-08  Tobias Burnus  <burnus@net-b.de>
3537         * doc/invoke.texi (-fcilkplus): Update implementation status.
3539 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
3540             Richard Biener  <rguenther@suse.de>
3542         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
3543         consistently accross all TUs.
3544         (run_gcc): Enable -fshort-double automatically at link at link-time
3545         and disallow override.
3547 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
3549         PR target/58271
3550         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
3551         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
3552         if they can't be used.
3554 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3556         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
3557         for Solaris 11/x86 ld.
3558         * configure: Regenerate.
3560 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3562         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
3563         (LIB_TLS_SPEC): Save as ld_tls_libs.
3564         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
3565         (HAVE_AS_IX86_TLSLDM): New test.
3566         * configure, config.in: Regenerate.
3567         * config/i386/i386.c (legitimize_tls_address): Fall back to
3568         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
3569         cannot support TLS_MODEL_LOCAL_DYNAMIC.
3570         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
3571         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
3573 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
3575         * common.opt (fira-loop-pressure): Mark as optimization.
3577 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
3579         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
3580         an OpenMP mappable type.
3582 2014-03-06  Matthias Klose  <doko@ubuntu.com>
3584         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
3585         MULTILIB_OSDIRNAMES is not defined.
3587 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
3588             Meador Inge  <meadori@codesourcery.com>
3590         PR target/58595
3591         * config/arm/arm.c (arm_tls_symbol_p): Remove.
3592         (arm_legitimize_address): Call legitimize_tls_address for any
3593         arm_tls_referenced_p expression, handle constant addend.  Call it
3594         before testing for !TARGET_ARM.
3595         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
3597 2014-03-06  Richard Biener  <rguenther@suse.de>
3599         PR middle-end/60445
3600         PR lto/60424
3601         PR lto/60427
3602         Revert
3603         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
3605         * tree-streamer.c (record_common_node): Assert we don't record
3606         nodes with type double.
3607         (preload_common_node): Skip type double, complex double and double
3608         pointer since it is now frontend dependent due to fshort-double option.
3610 2014-03-06  Richard Biener  <rguenther@suse.de>
3612         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
3613         or -fno-lto is specified and the linker has full plugin support.
3614         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
3615         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
3616         * lto-wrapper.c (merge_and_complain): Merge compile-time
3617         optimization levels.
3618         (run_gcc): And pass it through to the link options.
3620 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
3622         PR debug/60381
3623         Revert:
3624         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
3625         PR debug/59992
3626         * cselib.c (remove_useless_values): Skip to avoid quadratic
3627         behavior if the condition moved from...
3628         (cselib_process_insn): ... here holds.
3630 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
3632         PR plugins/59335
3633         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
3634         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
3636         PR plugins/59335
3637         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
3638         (TM_H): Add x86-tune.def.
3640 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3642         * config/aarch64/aarch64.c (generic_tunings):
3643         Use cortexa57_extra_costs.
3645 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
3647         PR lto/60404
3648         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
3649         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
3650         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
3651         cost for in_lto_p.
3653 2014-03-04  Heiher  <r@hev.cc>
3655         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
3656         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
3658 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
3660         * config/i386/predicates.md (const2356_operand): Change to ...
3661         (const2367_operand): ... this.
3662         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
3663         const2367_operand.
3664         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3665         (*avx512pf_scatterpf<mode>sf): Ditto.
3666         (avx512pf_scatterpf<mode>df): Ditto.
3667         (*avx512pf_scatterpf<mode>df_mask): Ditto.
3668         (*avx512pf_scatterpf<mode>df): Ditto.
3669         * config/i386/i386.c (ix86_expand_builtin): Update
3670         incorrect hint operand error message.
3672 2014-03-04  Richard Biener  <rguenther@suse.de>
3674         * lto-section-in.c (lto_get_section_data): Fix const cast.
3676 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
3678         * tree-streamer.c (record_common_node): Assert we don't record
3679         nodes with type double.
3680         (preload_common_node): Skip type double, complex double and double
3681         pointer since it is now frontend dependent due to fshort-double option.
3683 2014-03-04  Richard Biener  <rguenther@suse.de>
3685         PR lto/60405
3686         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
3687         (lto_input_toplevel_asms): Likewise.
3688         * lto-section-in.c (lto_get_section_data): Instead do it here
3689         for every section.
3691 2014-03-04  Richard Biener  <rguenther@suse.de>
3693         PR tree-optimization/60382
3694         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
3695         dead PHIs a reduction.
3697 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
3699         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
3700         hint value.
3701         (_mm_prefetch): Move out of GCC target("sse") pragma.
3702         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
3703         GCC target("prfchw") pragma.
3704         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
3705         for locality <= 2.
3706         * config/i386/i386.c (ix86_option_override_internal): Enable
3707         -mprfchw with -mprefetchwt1.
3709 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
3711         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
3712         Mark as varying.
3714 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
3716         * opts.h (CL_PCH_IGNORE): Define.
3717         * targhooks.c (option_affects_pch_p):
3718         Return false for options that have CL_PCH_IGNORE set.
3719         * opt-functions.awk: Process PchIgnore.
3720         * doc/options.texi: Document PchIgnore.
3722         * config/arc/arc.opt (misize): Add PchIgnore property.
3724 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3726         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
3727         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
3728         constraint on constants to permit them being loaded into
3729         GENERAL_REGS or BASE_REGS.
3731 2014-03-03  Nick Clifton  <nickc@redhat.com>
3733         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
3734         anti-cacnonical alternatives.
3735         (negandhi3_real): New pattern.
3736         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
3738 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
3740         * config/avr/avr-mcus.def: Remove atxmega16x1.
3741         * config/avr/avr-tables.opt: Regenerate.
3742         * config/avr/t-multilib: Regenerate.
3743         * doc/avr-mmcu.texi: Regenerate.
3745 2014-03-03  Tobias Grosser  <tobias@grosser.es>
3746             Mircea Namolaru  <mircea.namolaru@inria.fr>
3748         PR tree-optimization/58028
3749         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
3750         scalar dimensions.
3752 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3754         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
3755         not handled by recognizers.
3757 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
3759         PR middle-end/60175
3760         * function.c (expand_function_end): Don't emit
3761         clobber_return_register sequence if clobber_after is a BARRIER.
3762         * cfgexpand.c (construct_exit_block): Append instructions before
3763         return_label to prev_bb.
3765 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3767         * config/rs6000/constraints.md: Document reserved use of "wc".
3769 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
3771         PR ipa/60150
3772         * ipa.c (function_and_variable_visibility): When dissolving comdat
3773         group, also set all symbols to local.
3775 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
3777         PR ipa/60306
3779         Revert:
3780         2013-12-14   Jan Hubicka  <jh@suse.cz>
3781         PR middle-end/58477
3782         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
3784 2014-03-02  Jon Beniston  <jon@beniston.com>
3786         PR bootstrap/48230
3787         PR bootstrap/50927
3788         PR bootstrap/52466
3789         PR target/46898
3790         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
3791           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
3792         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
3793         (simple_return, *simple_return): New patterns
3794         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
3795         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
3797 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
3799         * dwarf2out.c (gen_subprogram_die): Tidy.
3801 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
3803         PR target/60071
3804         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
3805         (*mov_t_msb_neg_negc): ... this new insn.
3807 2014-02-28  Jason Merrill  <jason@redhat.com>
3809         PR c++/58678
3810         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
3811         function.
3813 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
3815         PR c++/60314
3816         * dwarf2out.c (decltype_auto_die): New static.
3817         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
3818         (gen_type_die_with_usage): Handle 'decltype(auto)'.
3819         (is_cxx_auto): Likewise.
3821 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
3823         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
3824         we are not using general regs only.
3826 2014-02-28  Richard Biener  <rguenther@suse.de>
3828         PR target/60280
3829         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
3830         previous fix and only allow to remove trivial pre-headers
3831         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
3832         (remove_forwarder_block): Properly update the latch of a loop.
3834 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
3836         PR debug/59992
3837         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
3838         (cselib_preserved_hash_table): New.
3839         (preserve_constants_and_equivs): Move preserved vals to it.
3840         (cselib_find_slot): Look it up first.
3841         (cselib_init): Initialize it.
3842         (cselib_finish): Release it.
3843         (dump_cselib_table): Dump it.
3845 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
3847         PR debug/59992
3848         * cselib.c (remove_useless_values): Skip to avoid quadratic
3849         behavior if the condition moved from...
3850         (cselib_process_insn): ... here holds.
3852 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
3854         PR debug/57232
3855         * var-tracking.c (vt_initialize): Apply the same condition to
3856         preserve the CFA base value.
3858 2014-02-28  Joey Ye  <joey.ye@arm.com>
3860         PR target/PR60169
3861         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
3862         if reload in progress or completed.
3864 2014-02-28  Tobias Burnus  <burnus@net-b.de>
3866         PR middle-end/60147
3867         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
3868         NAMELIST_DECL.
3870 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
3872         * doc/tm.texi.in (Condition Code Status): Update documention for
3873         relative locations of cc0-setter and cc0-user.
3875 2014-02-27  Jeff Law  <law@redhat.com>
3877         PR rtl-optimization/52714
3878         * combine.c (try_combine): When splitting an unrecognized PARALLEL
3879         into two independent simple sets, if I3 is a jump, ensure the
3880         pattern we place into I3 is a (set (pc) ...).
3882 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
3883             Jeff Law  <law@redhat.com>
3885         PR rtl-optimization/49847
3886         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
3887         are in different blocks.
3888         * doc/tm.texi (Condition Code Status): Update documention for
3889         relative locations of cc0-setter and cc0-user.
3891 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
3893         PR target/59222
3894         * lra.c (lra_emit_add): Check SUBREG too.
3896 2014-02-27  Andreas Schwab  <schwab@suse.de>
3898         * config/m68k/m68k.c (m68k_option_override): Disable
3899         -flive-range-shrinkage for classic m68k.
3900         (m68k_override_options_after_change): Likewise.
3902 2014-02-27  Marek Polacek  <polacek@redhat.com>
3904         PR middle-end/59223
3905         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
3906         -Wmaybe-uninitialized.
3908 2014-02-27  Alan Modra  <amodra@gmail.com>
3910         PR target/57936
3911         * reload1.c (emit_input_reload_insns): When reload_override_in,
3912         set old to rl->in_reg when rl->in_reg is a subreg.
3914 2014-02-26  Richard Biener  <rguenther@suse.de>
3916         PR bootstrap/60343
3917         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
3919 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
3921         * common/config/i386/predicates.md (const1256_operand): Remove.
3922         (const2356_operand): New.
3923         (const_1_to_2_operand): Remove.
3924         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3925         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3926         (*avx512pf_gatherpf<mode>sf): Ditto.
3927         (avx512pf_gatherpf<mode>df): Ditto.
3928         (*avx512pf_gatherpf<mode>df_mask): Ditto.
3929         (*avx512pf_gatherpf<mode>df): Ditto.
3930         (avx512pf_scatterpf<mode>sf): Ditto.
3931         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3932         (*avx512pf_scatterpf<mode>sf): Ditto.
3933         (avx512pf_scatterpf<mode>df): Ditto.
3934         (*avx512pf_scatterpf<mode>df_mask): Ditto.
3935         (*avx512pf_scatterpf<mode>df): Ditto.
3936         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
3938 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
3940         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
3941         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
3942         (_mm512_mask_testn_epi64_mask): Move to ...
3943         * config/i386/avx512cdintrin.h: Here.
3944         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
3945         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
3946         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
3947         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
3948         TARGET_AVX512F from TARGET_AVX512CD.
3950 2014-02-26  Richard Biener  <rguenther@suse.de>
3952         PR ipa/60327
3953         * ipa.c (walk_polymorphic_call_targets): Properly guard
3954         call to inline_update_overall_summary.
3956 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
3958         PR target/60280
3959         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
3960         and latches only if requested.  Fix latch if it is removed.
3961         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
3962         LOOPS_HAVE_PREHEADERS.
3964 2014-02-25  Andrew Pinski  <apinski@cavium.com>
3966         * builtins.c (expand_builtin_thread_pointer): Create a new target
3967         when the target is NULL.
3969 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
3971         PR rtl-optimization/60317
3972         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3973         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3974         * lra-assigns.c: Include params.h.
3975         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
3976         other reload pseudos considerations.
3978 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3980         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
3981         to use canonical form for nor<mode>3.
3983 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3985         PR target/55426
3986         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
3987         conversions.
3989 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
3991         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
3992         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
3993         (ix86_handle_option): Handle OPT_mprefetchwt1.
3994         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
3995         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3996         PREFETCHWT1 CPUID.
3997         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3998         OPTION_MASK_ISA_PREFETCHWT1.
3999         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
4000         (PTA_PREFETCHWT1): New.
4001         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
4002         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
4003         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
4004         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
4005         (*prefetch_avx512pf_<mode>_: Change into ...
4006         (*prefetch_prefetchwt1_<mode>: This.
4007         * config/i386/i386.opt (mprefetchwt1): New.
4008         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
4009         (_mm_prefetch): Handle intent to write.
4010         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
4012 2014-02-25  Richard Biener  <rguenther@suse.de>
4014         PR middle-end/60291
4015         * emit-rtl.c (mem_attrs_htab): Remove.
4016         (mem_attrs_htab_hash): Likewise.
4017         (mem_attrs_htab_eq): Likewise.
4018         (set_mem_attrs): Always allocate new mem-attrs when something changed.
4019         (init_emit_once): Do not allocate mem_attrs_htab.
4021 2014-02-25  Richard Biener  <rguenther@suse.de>
4023         PR lto/60319
4024         * lto-opts.c (lto_write_options): Output non-explicit conservative
4025         -fwrapv, -fno-trapv and -fno-strict-overflow.
4026         * lto-wrapper.c (merge_and_complain): Handle merging those options.
4027         (run_gcc): And pass them through.
4029 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4031         * sel-sched.c (calculate_new_fences): New parameter ptime.
4032         Calculate it as a maximum over all fence cycles.
4033         (sel_sched_region_2): Adjust the call to calculate_new_fences.
4034         Print the final schedule timing when sched_verbose.
4036 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4038         PR rtl-optimization/60292
4039         * sel-sched.c (fill_vec_av_set): Do not reset target availability
4040         bit fot the fence instruction.
4042 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
4044         * calls.h: Fix typo in comment.
4046 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
4048         * config/pa/pa.c (pa_output_move_double): Don't valididate when
4049         adjusting offsetable addresses.
4051 2014-02-24  Guozhi Wei  <carrot@google.com>
4053         * sparseset.h (sparseset_pop): Fix the wrong index.
4055 2014-02-24  Walter Lee  <walt@tilera.com>
4057         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
4058         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
4059         triplet.
4060         * common/config/tilegx/tilegx-common.c
4061         (TARGET_DEFAULT_TARGET_FLAGS): Define.
4062         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
4063         (LINK_SPEC): Ditto.
4064         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
4065         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
4066         (tilegx_gimplify_va_arg_expr): Handle big endian.
4067         (tilegx_expand_unaligned_load): Ditto.
4068         (tilegx_expand_unaligned_store): Ditto.
4069         (TARGET_RETURN_IN_MSB): New.
4070         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
4071         (TARGET_ENDIAN_DEFAULT): New.
4072         (TARGET_BIG_ENDIAN): Handle big endian.
4073         (BYTES_BIG_ENDIAN): Ditto.
4074         (WORDS_BIG_ENDIAN): Ditto.
4075         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
4076         (ENDIAN_SPEC): New.
4077         (EXTRA_SPECS): New.
4078         * config/tilegx/tilegx.md (extv): Handle big endian.
4079         (extzv): Ditto.
4080         (insn_st<n>): Ditto.
4081         (insn_st<n>_add<bitsuffix>): Ditto.
4082         (insn_stnt<n>): Ditto.
4083         (insn_stnt<n>_add<bitsuffix>):Ditto.
4084         (vec_interleave_highv8qi): Handle big endian.
4085         (vec_interleave_highv8qi_be): New.
4086         (vec_interleave_highv8qi_le): New.
4087         (insn_v1int_h): Handle big endian.
4088         (vec_interleave_lowv8qi): Handle big endian.
4089         (vec_interleave_lowv8qi_be): New.
4090         (vec_interleave_lowv8qi_le): New.
4091         (insn_v1int_l): Handle big endian.
4092         (vec_interleave_highv4hi): Handle big endian.
4093         (vec_interleave_highv4hi_be): New.
4094         (vec_interleave_highv4hi_le): New.
4095         (insn_v2int_h): Handle big endian.
4096         (vec_interleave_lowv4hi): Handle big endian.
4097         (vec_interleave_lowv4hi_be): New.
4098         (vec_interleave_lowv4hi_le): New.
4099         (insn_v2int_l): Handle big endian.
4100         (vec_interleave_highv2si): Handle big endian.
4101         (vec_interleave_highv2si_be): New.
4102         (vec_interleave_highv2si_le): New.
4103         (insn_v4int_h): Handle big endian.
4104         (vec_interleave_lowv2si): Handle big endian.
4105         (vec_interleave_lowv2si_be): New.
4106         (vec_interleave_lowv2si_le): New.
4107         (insn_v4int_l): Handle big endian.
4108         * config/tilegx/tilegx.opt (mbig-endian): New option.
4109         (mlittle-endian): New option.
4110         * doc/install.texi: Document tilegxbe-linux.
4111         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
4113 2014-02-24  Martin Jambor  <mjambor@suse.cz>
4115         PR ipa/60266
4116         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
4117         there are no parameter descriptors.
4119 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
4121         PR rtl-optimization/60268
4122         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
4123         initialization to ...
4124         (sched_rgn_init): ... here.
4125         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
4127 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4129         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
4130         names.
4132 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
4134         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
4135         definition.
4137 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4139         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
4140         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
4142 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4144         * config/microblaze/predicates.md: Add cmp_op predicate.
4145         * config/microblaze/microblaze.md: Add branch_compare instruction
4146         which uses cmp_op predicate and emits cmp insn before branch.
4147         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
4148         to microblaze_expand_conditional_branch and consolidate logic.
4149         (microblaze_expand_conditional_branch): emit branch_compare
4150         insn instead of handling cmp op separate from branch insn.
4152 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4154         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
4155         to permit subregs.
4157 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4159         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
4160         define_insn with define_expand and new define_insn
4161         *altivec_lve<VI_char>x_internal.
4162         (altivec_stve<VI_char>x): Replace define_insn with define_expand
4163         and new define_insn *altivec_stve<VI_char>x_internal.
4164         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
4165         prototype.
4166         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
4167         lve*x built-ins.
4168         (altivec_expand_stvex_be): New function.
4170 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
4172         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
4173         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
4174         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
4175         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
4177 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
4179         PR target/60298
4180         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
4181         instead of emit_move_insn.
4183 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4185         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
4186         vspltw with vsldoi.
4187         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
4188         gen_altivec_vsumsws.
4190 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4192         * config/rs6000/altivec.md (altivec_lvxl): Rename as
4193         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
4194         (altivec_lvxl_<mode>): New define_expand incorporating
4195         -maltivec=be semantics where needed.
4196         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
4197         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
4198         semantics where needed.
4199         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
4200         (altivec_stvx_<mode>): New define_expand incorporating
4201         -maltivec=be semantics where needed.
4202         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
4203         VM2 iterator instead of V4SI.
4204         (altivec_stvxl_<mode>): New define_expand incorporating
4205         -maltivec=be semantics where needed.
4206         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
4207         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
4208         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
4209         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
4210         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
4211         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
4212         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
4213         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
4214         ALTIVEC_BUILTIN_STVXL.
4215         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
4216         (altivec_expand_stvx_be): Likewise.
4217         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
4218         (altivec_expand_lvx_be): Likewise.
4219         (altivec_expand_stvx_be): Likewise.
4220         (altivec_expand_builtin): Add cases for
4221         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
4222         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
4223         (altivec_init_builtins): Add definitions for
4224         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
4225         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
4227 2014-02-21  Catherine Moore  <clm@codesourcery.com>
4229         * doc/invoke.texi (mvirt, mno-virt): Document.
4230         * config/mips/mips.opt (mvirt): New option.
4231         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
4233 2014-02-21  Richard Biener  <rguenther@suse.de>
4235         PR tree-optimization/60276
4236         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
4237         (STMT_VINFO_MIN_NEG_DIST): New macro.
4238         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
4239         STMT_VINFO_MIN_NEG_DIST.
4240         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
4241         made for negative dependence distances still hold.
4243 2014-02-21  Richard Biener  <rguenther@suse.de>
4245         PR middle-end/60291
4246         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
4247         DECL_INITIAL for globals not in the current function context.
4249 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4251         PR tree-optimization/56490
4252         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
4253         * tree-ssa-uninit.c: Include params.h.
4254         (compute_control_dep_chain): Add num_calls argument, return false
4255         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
4256         num_calls to recursive call.
4257         (find_predicates): Change dep_chain into normal array,
4258         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
4259         variable and adjust compute_control_dep_chain caller.
4260         (find_def_preds): Likewise.
4262 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
4264         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
4265         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
4267 2014-02-21  Nick Clifton  <nickc@redhat.com>
4269         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
4270         (pushhi1): Likewise.
4271         (popqi1): Add mode to pre_dec.
4272         (pophi1): Likewise.
4274 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4276         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
4277         mode for mask of V8SFmode permutation.
4279 2014-02-20  Richard Henderson  <rth@redhat.com>
4281         PR c++/60272
4282         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
4283         a new pseudo for OLDVAL.
4285 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
4287         PR target/57896
4288         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
4289         gen_reg_rtx if d->testing_p.
4290         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
4291         if d->testing_p and we will certainly return true.
4292         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
4293         if d->testing_p.
4295 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
4297         * emit-rtl.c (gen_reg_rtx): Assert that
4298         crtl->emit.regno_pointer_align_length is non-zero.
4300 2014-02-20  Richard Henderson  <rth@redhat.com>
4302         PR c++/60272
4303         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
4304         on failure the store back into EXPECT.
4306 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
4307             Sandra Loosemore  <sandra@codesourcery.com>
4309         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
4310         * config/nios2/nios2.c (nios2_function_profiler): Add
4311         -fPIC (flag_pic == 2) support.
4312         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
4313         (nios2_large_offset_p): New function.
4314         (nios2_unspec_reloc_p): Move up position, update to use
4315         nios2_large_offset_p.
4316         (nios2_unspec_address): Remove function.
4317         (nios2_unspec_offset): New function.
4318         (nios2_large_got_address): New function.
4319         (nios2_got_address): Add large offset support.
4320         (nios2_legitimize_tls_address): Update usage of removed and new
4321         functions.
4322         (nios2_symbol_binds_local_p): New function.
4323         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
4324         (nios2_legitimize_address): Update to use nios2_large_offset_p.
4325         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
4326         (nios2_print_operand): Merge H/L processing, add hiadj/lo
4327         processing for (const (unspec ...)).
4328         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
4330 2014-02-20  Richard Biener  <rguenther@suse.de>
4332         * tree-cfg.c (replace_uses_by): Mark altered BBs before
4333         doing the substitution.
4334         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
4336 2014-02-20  Martin Jambor  <mjambor@suse.cz>
4338         PR ipa/55260
4339         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
4340         info when checking whether lattices are bottom.
4342 2014-02-20  Richard Biener  <rguenther@suse.de>
4344         PR middle-end/60221
4345         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
4346         regions at -O0.
4348 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
4350         PR ipa/58555
4351         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
4352         parameter specifying the scaling.
4353         (inline_call): Update.
4354         (want_inline_recursively): Guard division by zero.
4355         (recursive_inlining): Update.
4356         * ipa-inline.h (clone_inlined_nodes): Update.
4358 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4360         PR target/60204
4361         * config/i386/i386.c (classify_argument): Pass structures of size
4362         64 bytes or less in register.
4364 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4365             Kirill Yukhin  <kirill.yukhin@intel.com>
4367         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
4368         (_mm_rcp28_round_ss): Ditto.
4369         (_mm_rsqrt28_round_sd): Ditto.
4370         (_mm_rsqrt28_round_ss): Ditto.
4371         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
4372         (_mm_rcp14_round_ss): Ditto.
4373         (_mm_rsqrt14_round_sd): Ditto.
4374         (_mm_rsqrt14_round_ss): Ditto.
4375         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
4376         the first input operand, get rid of match_dup.
4377         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
4378         attribute to sse.
4379         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4380         Ditto.
4381         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
4382         operand as the first input operand, set type attribute.
4383         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4384         Set type attribute.
4385         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
4386         operand as the first input operand, set type attribute.
4388 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4390         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
4391         bit of zero.
4393 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
4395         PR target/60207
4396         * config/i386/i386.c (construct_container): Remove TFmode check
4397         for X86_64_INTEGER_CLASS.
4399 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
4401         PR target/59794
4402         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
4403         only when -Wpsabi is enabled.
4405 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
4407          PR target/59799
4408         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
4409         passing arrays in registers are the same as for structs, so remove the
4410         special case for them.
4412 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
4414         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
4415         destination type, extract only the valid bits if the source type is not
4416         integral and has a different mode.
4418 2014-02-19  Richard Biener  <rguenther@suse.de>
4420         PR ipa/60243
4421         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
4422         for all calls.
4424 2014-02-19  Richard Biener  <rguenther@suse.de>
4426         PR ipa/60243
4427         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
4428         (ipa_modify_call_arguments): Emit an argument load explicitely and
4429         preserve virtual SSA form there and for the replacement call.
4430         Do not update SSA form nor free dominance info.
4432 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4434         * ipa.c (function_and_variable_visibility): Also clear WEAK
4435         flag when disolving COMDAT_GROUP.
4437 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4439         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
4440         * ipa-prop.c (ipa_set_jf_known_type): Return early when
4441         not devirtualizing.
4442         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
4443         do more sanity checks.
4444         (detect_type_change): Return true when giving up early.
4445         (compute_complex_assign_jump_func): Fix type parameter of
4446         ipa_set_ancestor_jf.
4447         (compute_complex_ancestor_jump_func): Likewise.
4448         (update_jump_functions_after_inlining): Fix updating of
4449         ancestor function.
4450         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
4452 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4454         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
4455         inline clones when edge disappears.
4457 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4459         PR target/60203
4460         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
4461         Split 64-bit moves into 2 patterns.  Do not allow the use of
4462         direct move for TDmode in little endian, since the decimal value
4463         has little endian bytes within a word, but the 64-bit pieces are
4464         ordered in a big endian fashion, and normal subreg's of TDmode are
4465         not allowed.
4466         (mov<mode>_64bit_dm): Likewise.
4467         (movtd_64bit_nodm): Likewise.
4469 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
4471         PR tree-optimization/60174
4472         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
4473         statement of an SSA_NAME that occurs in an abnormal PHI node.
4475 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
4477         PR sanitizer/60142
4478         * final.c (SEEN_BB): Remove.
4479         (SEEN_NOTE, SEEN_EMITTED): Renumber.
4480         (final_scan_insn): Don't force_source_line on second
4481         NOTE_INSN_BASIC_BLOCK.
4483 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
4485         PR target/60205
4486         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
4487         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
4488         (type_natural_mode): Warn ABI change when %zmm register is not
4489         available for AVX512F vector value passing.
4491 2014-02-18  Kai Tietz  <ktietz@redhat.com>
4493         PR target/60193
4494         * config/i386/i386.c (ix86_expand_prologue): Use value in
4495         rax register as displacement when restoring %r10 or %rax.
4496         Fix wrong offset when restoring both registers.
4498 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
4500         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
4501         assertion with conditional return.
4503 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
4504             Uros Bizjak  <ubizjak@gmail.com>
4506         PR driver/60233
4507         * config/i386/driver-i386.c (host_detect_local_cpu): If
4508         YMM state is not saved by the OS, also clear has_f16c.  Move
4509         CPUID 0x80000001 handling before YMM state saving checking.
4511 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
4513         PR rtl-optimization/58960
4514         * haifa-sched.c (alloc_global_sched_pressure_data): New,
4515         factored out from ...
4516         (sched_init): ... here.
4517         (free_global_sched_pressure_data): New, factored out from ...
4518         (sched_finish): ... here.
4519         * sched-int.h (free_global_sched_pressure_data): Declare.
4520         * sched-rgn.c (nr_regions_initial): New static global.
4521         (haifa_find_rgns): Initialize it.
4522         (schedule_region): Disable sched-pressure for the newly
4523         generated regions.
4525 2014-02-17  Richard Biener  <rguenther@suse.de>
4527         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
4528         release SSA defs of pattern stmts.
4530 2014-02-17  Richard Biener  <rguenther@suse.de>
4532         * tree-inline.c (expand_call_inline): Release the virtual
4533         operand defined by the call we are about to inline.
4535 2014-02-17  Richard Biener  <rguenther@suse.de>
4537         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
4539 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
4540             Ilya Tocar  <ilya.tocar@intel.com>
4542         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
4543         arguments order in builtin.
4544         (_mm512_permutexvar_epi64): Ditto.
4545         (_mm512_mask_permutexvar_epi64): Ditto
4546         (_mm512_maskz_permutexvar_epi32): Ditto
4547         (_mm512_permutexvar_epi32): Ditto
4548         (_mm512_mask_permutexvar_epi32): Ditto
4550 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4552         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
4553         (p8_vmrgow): Likewise.
4555 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4557         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
4558         endian targets.
4560 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
4562         PR target/60203
4563         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
4564         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
4565         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
4566         using direct move instructions on ISA 2.07.  Also adjust
4567         instruction length for 64-bit.
4568         (mov<mode>_64bit, TFmode/TDmode): Likewise.
4569         (mov<mode>_32bit, TFmode/TDmode): Likewise.
4571 2014-02-15  Alan Modra  <amodra@gmail.com>
4573         PR target/58675
4574         PR target/57935
4575         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
4576         find_replacement on parts of insn rtl that might be reloaded.
4578 2014-02-15  Richard Biener  <rguenther@suse.de>
4580         PR tree-optimization/60183
4581         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
4582         (tree_ssa_phiprop): Calculate and free post-dominators.
4584 2014-02-14  Jeff Law  <law@redhat.com>
4586         PR rtl-optimization/60131
4587         * ree.c (get_extended_src_reg): New function.
4588         (combine_reaching_defs): Use it rather than assuming location of REG.
4589         (find_and_remove_re): Verify first operand of extension is
4590         a REG before adding the insns to the copy list.
4592 2014-02-14  Roland McGrath  <mcgrathr@google.com>
4594         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
4595         * configure: Regenerated.
4596         * config.in: Regenerated.
4597         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
4598         instead of ASM_SHORT.
4600 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
4601             Richard Earnshaw  <rearnsha@arm.com>
4603         PR rtl-optimization/59535
4604         * lra-constraints.c (process_alt_operands): Encourage alternative
4605         when unassigned pseudo class is superset of the alternative class.
4606         (inherit_reload_reg): Don't inherit when optimizing for code size.
4607         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
4608         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
4609         modes not less than 4 for Thumb1.
4611 2014-02-14  Kyle McMartin  <kyle@redhat.com>
4613         PR pch/60010
4614         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
4616 2014-02-14  Richard Biener  <rguenther@suse.de>
4618         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
4619         (get_frame_arg): Drop the assert with langhook types_compatible_p.
4620         Do not strip INDIRECT_REFs.
4622 2014-02-14  Richard Biener  <rguenther@suse.de>
4624         PR lto/60179
4625         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
4626         DECL_FUNCTION_SPECIFIC_TARGET.
4627         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
4628         * tree-streamer-out.c (pack_ts_target_option): Remove.
4629         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
4630         (write_ts_function_decl_tree_pointers): Do not stream
4631         DECL_FUNCTION_SPECIFIC_TARGET.
4632         * tree-streamer-in.c (unpack_ts_target_option): Remove.
4633         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
4634         (lto_input_ts_function_decl_tree_pointers): Do not stream
4635         DECL_FUNCTION_SPECIFIC_TARGET.
4637 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
4639         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
4640         (get_initial_def_for_induction, vectorizable_induction): Ignore
4641         debug stmts when looking for exit_phi.
4642         (vectorizable_live_operation): Fix up condition.
4644 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4646         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
4647         nreverse() because it changes the content of original tree list.
4649 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4651         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
4652         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
4654 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4656         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
4657         GNU coding standards.
4659 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
4661         PR debug/60152
4662         * dwarf2out.c (gen_subprogram_die): Don't call
4663         add_calling_convention_attribute if subr_die is old_die.
4665 2014-02-13  Sharad Singhai  <singhai@google.com>
4667         * doc/optinfo.texi: Fix order of nodes.
4669 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
4671         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
4672         operands[2], not operands[3].
4674 2014-02-13  Richard Biener  <rguenther@suse.de>
4676         PR bootstrap/59878
4677         * doc/install.texi (ISL): Update recommended version to 0.12.2,
4678         mention the possibility of an in-tree build.
4679         (CLooG): Update recommended version to 0.18.1, mention the
4680         possibility of an in-tree build and clarify that the ISL
4681         bundled with CLooG does not work.
4683 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
4685         PR target/43546
4686         * expr.c (compress_float_constant): If x is a hard register,
4687         extend into a pseudo and then move to x.
4689 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4691         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
4692         caused by bad second argument to warning_at() with -mhotpatch and
4693         nested functions (e.g. with gfortran).
4695 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
4697         * opts.c (option_name): Remove "enabled by default" rider.
4699 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
4701         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
4703 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
4704             Uros Bizjak  <ubizjak@gmail.com>
4706         PR target/60151
4707         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
4708         * configure: Regenerated.
4710 2014-02-12  Richard Biener  <rguenther@suse.de>
4712         * vec.c (vec_prefix::calculate_allocation): Move as
4713         inline variant to vec.h.
4714         (vec_prefix::calculate_allocation_1): New out-of-line version.
4715         * vec.h (vec_prefix::calculate_allocation_1): Declare.
4716         (vec_prefix::m_has_auto_buf): Rename to ...
4717         (vec_prefix::m_using_auto_storage): ... this.
4718         (vec_prefix::calculate_allocation): Inline the easy cases
4719         and dispatch to calculate_allocation_1 which doesn't need the
4720         prefix address.
4721         (va_heap::reserve): Use gcc_checking_assert.
4722         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
4723         m_using_auto_storage.
4724         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
4725         member and adjust.
4726         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
4727         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
4728         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
4730 2014-02-12  Richard Biener  <rguenther@suse.de>
4732         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
4733         when we found a dependence.
4735 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
4737         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
4738         common code...
4739         (maybe_fold_stmt): ... into this new function.
4740         * omp-low.c (lower_omp): Update comment.
4742         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
4743         last use.
4745         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
4746         dereference.
4748 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
4750         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
4751         identifiers in comments.
4752         (cortexa53_extra_costs): Likewise.
4753         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
4754         (cortexa7_extra_costs): Likewise.
4755         (cortexa12_extra_costs): Likewise.
4756         (cortexa15_extra_costs): Likewise.
4757         (v7m_extra_costs): Likewise.
4759 2014-02-12  Richard Biener  <rguenther@suse.de>
4761         PR middle-end/60092
4762         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
4763         of posix_memalign being successful.
4764         (lower_stmt): Restrict lowering of posix_memalign to when
4765         -ftree-bit-ccp is enabled.
4767 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
4769         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
4770         arg_loc.
4771         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
4773 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
4775         PR rtl-optimization/60116
4776         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
4777         other_insn once the combination has been validated.
4779 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
4781         PR lto/59468
4782         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
4783         and wrapper.
4784         * ipa-devirt.c: Include demangle.h
4785         (odr_violation_reported): New static variable.
4786         (add_type_duplicate): Update odr_violations.
4787         (maybe_record_node): Add completep parameter; update it.
4788         (record_target_from_binfo): Add COMPLETEP parameter;
4789         update it as needed.
4790         (possible_polymorphic_call_targets_1): Likewise.
4791         (struct polymorphic_call_target_d): Add nonconstruction_targets;
4792         rename FINAL to COMPLETE.
4793         (record_targets_from_bases): Sanity check we found the binfo;
4794         fix COMPLETEP updating.
4795         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
4796         parameter, fix computing of COMPLETEP.
4797         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
4798         at LTO time do demangling.
4799         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
4800         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
4801         parameter.
4802         (gimple_get_virt_method_for_binfo): Likewise.
4803         * gimple-fold.h (gimple_get_virt_method_for_binfo,
4804         gimple_get_virt_method_for_vtable): Update prototypes.
4806 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
4808         PR target/49008
4809         * genautomata.c (add_presence_absence): Fix typo with
4810         {final_}presence_list.
4812 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
4814         PR target/60137
4815         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
4816         for VSX/Altivec vectors that land in GPR registers.
4818 2014-02-11  Richard Henderson  <rth@redhat.com>
4819             Jakub Jelinek  <jakub@redhat.com>
4821         PR debug/59776
4822         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
4823         around drhs if type conversion to lacc->type is not useless.
4825 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4827         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
4828         tuning struct.
4829         (cortex-a57.cortex-a53): Likewise.
4830         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
4832 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4834         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
4835         arm_restrict_it.
4837 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
4839         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
4840         add_options_for_arm_vfp3.
4842 2014-02-11  Jeff Law  <law@redhat.com>
4844         PR middle-end/54041
4845         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
4846         object with an undesirable mode.
4848 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4850         PR libgomp/60107
4851         * config/i386/sol2-9.h: New file.
4852         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
4853         *-*-solaris2.9*): Use it.
4855 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
4857         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
4858         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
4860 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
4862         * config/microblaze/microblaze.c: Extend mcpu version format
4864 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
4866         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
4868 2014-02-10  Richard Henderson  <rth@redhat.com>
4870         PR target/59927
4871         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
4872         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
4873         ms-abi vs -mno-accumulate-outgoing-args.
4874         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
4875         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
4876         respect to ms-abi.
4878 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4880         PR middle-end/60080
4881         * cfgexpand.c (expand_asm_operands): Attach source location to
4882         ASM_INPUT rtx objects.
4883         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
4885 2014-02-10  Nick Clifton  <nickc@redhat.com>
4887         * config/mn10300/mn10300.c (popcount): New function.
4888         (mn10300_expand_prologue): Include saved registers in stack usage
4889         count.
4891 2014-02-10  Jeff Law  <law@redhat.com>
4893         PR middle-end/52306
4894         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
4895         when changing the SET_DEST of a prior insn to avoid an input reload.
4897 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4899         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
4900         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
4901         -mcall-openbsd, or -mcall-linux.
4902         (CC1_ENDIAN_BIG_SPEC): Remove.
4903         (CC1_ENDIAN_LITTLE_SPEC): Remove.
4904         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4905         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
4906         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
4907         and %cc1_endian_default.
4908         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4910 2014-02-10  Richard Biener  <rguenther@suse.de>
4912         PR tree-optimization/60115
4913         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
4914         MEM_REF handling.  Properly verify that the accesses are not
4915         out of the objects bound.
4917 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4919         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
4920         coretex to cortex.
4922 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
4924         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
4925         proper constants and fix formatting.
4926         (possible_polymorphic_call_targets): Fix formatting.
4928 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
4929             Ilya Tocar  <ilya.tocar@intel.com>
4931         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
4932         (_mm512_loadu_epi32): Renamed into...
4933         (_mm512_loadu_si512): This.
4934         (_mm512_storeu_epi32): Renamed into...
4935         (_mm512_storeu_si512): This.
4936         (_mm512_maskz_ceil_ps): Removed.
4937         (_mm512_maskz_ceil_pd): Ditto.
4938         (_mm512_maskz_floor_ps): Ditto.
4939         (_mm512_maskz_floor_pd): Ditto.
4940         (_mm512_floor_round_ps): Ditto.
4941         (_mm512_floor_round_pd): Ditto.
4942         (_mm512_ceil_round_ps): Ditto.
4943         (_mm512_ceil_round_pd): Ditto.
4944         (_mm512_mask_floor_round_ps): Ditto.
4945         (_mm512_mask_floor_round_pd): Ditto.
4946         (_mm512_mask_ceil_round_ps): Ditto.
4947         (_mm512_mask_ceil_round_pd): Ditto.
4948         (_mm512_maskz_floor_round_ps): Ditto.
4949         (_mm512_maskz_floor_round_pd): Ditto.
4950         (_mm512_maskz_ceil_round_ps): Ditto.
4951         (_mm512_maskz_ceil_round_pd): Ditto.
4952         (_mm512_expand_pd): Ditto.
4953         (_mm512_expand_ps): Ditto.
4954         * config/i386/i386.c (ix86_builtins): Remove
4955         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
4956         (bdesc_args): Ditto.
4957         * config/i386/predicates.md (const1256_operand): New.
4958         (const_1_to_2_operand): Ditto.
4959         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4960         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4961         (*avx512pf_gatherpf<mode>sf): Ditto.
4962         (avx512pf_gatherpf<mode>df): Ditto.
4963         (*avx512pf_gatherpf<mode>df_mask): Ditto.
4964         (*avx512pf_gatherpf<mode>df): Ditto.
4965         (avx512pf_scatterpf<mode>sf): Ditto.
4966         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4967         (*avx512pf_scatterpf<mode>sf): Ditto.
4968         (avx512pf_scatterpf<mode>df): Ditto.
4969         (*avx512pf_scatterpf<mode>df_mask): Ditto.
4970         (*avx512pf_scatterpf<mode>df): Ditto.
4971         (avx512f_expand<mode>): Removed.
4972         (<shift_insn><mode>3<mask_name>): Change predicate type.
4974 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
4976         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
4977         not at the end of datarefs vector use ordered_remove to avoid
4978         reordering datarefs vector.
4980         PR c/59984
4981         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
4982         mark local addressable non-static vars as GOVD_PRIVATE
4983         instead of GOVD_LOCAL.
4984         * omp-low.c (lower_omp_for): Move gimple_bind_vars
4985         and BLOCK_VARS of gimple_bind_block to new_stmt rather
4986         than copying them.
4988         PR middle-end/60092
4989         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
4990         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
4991         assume_aligned or alloc_align attributes.
4992         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
4993         arguments.  Handle also assume_aligned and alloc_align attributes.
4994         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
4995         calls to functions with assume_aligned or alloc_align attributes.
4996         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
4998 2014-02-08  Terry Guo  <terry.guo@arm.com>
5000         * doc/invoke.texi: Document ARM -march=armv7e-m.
5002 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5004         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
5005         flag on __cilkrts_rethrow builtin.
5007         PR ipa/60026
5008         * ipa-cp.c (determine_versionability): Fail at -O0
5009         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
5010         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
5012         Revert:
5013         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5015         PR ipa/60026
5016         * tree-inline.c (copy_forbidden): Fail for
5017         __attribute__((optimize (0))) functions.
5019 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5021         * varpool.c: Include pointer-set.h.
5022         (varpool_remove_unreferenced_decls): Variables in other partitions
5023         will not be output; be however careful to not lose information
5024         about partitioning.
5026 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5028         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
5029         lookup in the vtable constructor.
5031 2014-02-07  Jeff Law  <law@redhat.com>
5033         PR target/40977
5034         * config/m68k/m68k.md (ashldi_extsi): Turn into a
5035         define_insn_and_split.
5037         * ipa-inline.c (inline_small_functions): Fix typos.
5039 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5041         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
5042         (s390_can_use_return_insn): Declare.
5043         * config/s390/s390.h (EPILOGUE_USES): Define.
5044         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
5045         instructions.
5046         (s390_chunkify_start): Handle return JUMP_LABELs.
5047         (s390_early_mach): Emit a main_pool instruction on the entry edge.
5048         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
5049         (s390_can_use_return_insn): New functions.
5050         (s390_fix_long_loop_prediction): Handle conditional returns.
5051         (TARGET_SET_UP_BY_PROLOGUE): Define.
5052         * config/s390/s390.md (ANY_RETURN): New code iterator.
5053         (*creturn, *csimple_return, return, simple_return): New patterns.
5055 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5057         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
5058         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
5059         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
5060         REG_CFA_RESTORE list when deciding not to restore a register.
5062 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5064         * config/s390/s390.c: Include tree-pass.h and context.h.
5065         (s390_early_mach): New function, split out from...
5066         (s390_emit_prologue): ...here.
5067         (pass_data_s390_early_mach): New pass structure.
5068         (pass_s390_early_mach): New class.
5069         (s390_option_override): Create and register early_mach pass.
5070         Move to end of file.
5072 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5074         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
5075         to match for the exit block.
5077 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5079         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
5080         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
5081         Reject misaligned operands.
5083 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5085         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
5087 2014-02-07  Richard Biener  <rguenther@suse.de>
5089         PR middle-end/60092
5090         * gimple-low.c (lower_builtin_posix_memalign): New function.
5091         (lower_stmt): Call it to lower posix_memalign in a way
5092         to make alignment info accessible.
5094 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
5096         PR c++/60082
5097         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
5098         __builtin_setjmp_receiver.
5100 2014-02-07  Richard Biener  <rguenther@suse.de>
5102         PR middle-end/60092
5103         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
5104         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
5105         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
5106         Handle BUILT_IN_POSIX_MEMALIGN.
5107         (find_func_clobbers): Likewise.
5108         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5109         (call_may_clobber_ref_p_1): Likewise.
5111 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5113         PR ipa/59918
5114         * ipa-devirt.c (record_target_from_binfo): Remove overactive
5115         sanity check.
5117 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5119         PR ipa/59469
5120         * lto-cgraph.c (lto_output_node): Use
5121         symtab_get_symbol_partitioning_class.
5122         (lto_output_varpool_node): likewise.
5123         (symtab_get_symbol_partitioning_class): Move here from
5124         lto/lto-partition.c
5125         * cgraph.h (symbol_partitioning_class): Likewise.
5126         (symtab_get_symbol_partitioning_class): Declare.
5128 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5130         * ggc.h (ggc_internal_cleared_alloc): New macro.
5131         * vec.h (vec_safe_copy): Handle memory stats.
5132         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
5133         * target-globals.c (save_target_globals): Likewise.
5135 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5137         PR target/60077
5138         * expr.c (emit_move_resolve_push): Export; be bit more selective
5139         on when to clear alias set.
5140         * expr.h (emit_move_resolve_push): Declare.
5141         * function.h (struct function): Add tail_call_marked.
5142         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
5143         * config/i386/i386-protos.h (ix86_expand_push): Remove.
5144         * config/i386/i386.md (TImode move expander): De not call
5145         ix86_expand_push.
5146         (FP push expanders): Preserve memory attributes.
5147         * config/i386/sse.md (push<mode>1): Remove.
5148         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
5149         (ix86_expand_push): Remove.
5150         * config/i386/mmx.md (push<mode>1): Remove.
5152 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5154         PR rtl-optimization/60030
5155         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
5156         lopart with paradoxical subreg before shifting it up by hprec.
5158 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5160         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
5161         Remove extra newline at end of file.
5162         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
5163         (arm_issue_rate): Handle cortexa57.
5164         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
5165         (cortex-a57.cortex-a53): Likewise.
5167 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5169         PR target/59575
5170         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
5171         don't record in REG_FRAME_RELATED_EXPR registers not set in that
5172         bitmask.
5173         (arm_expand_prologue): Adjust all callers.
5174         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
5175         info, registers also at the lowest numbered registers side.  Use
5176         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
5177         XEXP.
5179         PR debug/59992
5180         * var-tracking.c (adjust_mems): Before adding a SET to
5181         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
5183 2014-02-06  Alan Modra  <amodra@gmail.com>
5185         PR target/60032
5186         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
5187         change SDmode to DDmode when lra_in_progress.
5189 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5191         PR middle-end/59150
5192         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
5193         free_data_ref on the dr first, and before goto again also set dr
5194         to the next dr.  For simd_lane_access, free old datarefs[i] before
5195         overwriting it.  For get_vectype_for_scalar_type failure, don't
5196         free_data_ref if simd_lane_access.
5198         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
5200         PR target/60062
5201         * tree.h (opts_for_fn): New inline function.
5202         (opt_for_fn): Define.
5203         * config/i386/i386.c (ix86_function_regparm): Use
5204         opt_for_fn (decl, optimize) instead of optimize.
5206 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5208         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
5209         for SYMBOL_REF in large memory model.
5211 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5213         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
5214         and crypto support.
5215         (cortex-a57): Likewise.
5216         (cortex-a57.cortex-a53): Likewise.
5218 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
5219             Kugan Vivekanandarajah  <kuganv@linaro.org>
5221         * config/arm/arm.c (arm_vector_alignment_reachable): Check
5222         unaligned_access.
5223         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
5225 2014-02-06  Richard Biener  <rguenther@suse.de>
5227         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
5228         set_loop_copy and initialize_original_copy_tables.
5230 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
5232         * config/aarch64/aarch64-simd.md
5233         (aarch64_ashr_simddi): Change QI to SI.
5235 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5236             Jakub Jelinek  <jakub@redhat.com>
5238         PR middle-end/60013
5239         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
5240         of the dataflow.
5242 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5244         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
5245         CODE_FOR_altivec_vpku[hw]um to
5246         CODE_FOR_altivec_vpku[hw]um_direct.
5247         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
5248         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
5249         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
5250         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
5252 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5254         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
5255         generation for -maltivec=be.
5256         (altivec_vsumsws): Simplify redundant test.
5258 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5260         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
5261         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
5262         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
5263         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
5264         gen_altivec_vpkuwum.
5265         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
5266         BYTES_BIG_ENDIAN.
5267         (altivec_vpks<VI_char>ss): Likewise.
5268         (altivec_vpks<VI_char>us): Likewise.
5269         (altivec_vpku<VI_char>us): Likewise.
5270         (altivec_vpku<VI_char>um): Likewise.
5271         (altivec_vpku<VI_char>um_direct): New (copy of
5272         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
5273         internal use).
5274         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
5275         target is little endian and -maltivec=be is not specified.
5276         (*altivec_vupkhs<VU_char>_direct): New (copy of
5277         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
5278         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
5279         target is little endian and -maltivec=be is not specified.
5280         (*altivec_vupkls<VU_char>_direct): New (copy of
5281         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
5282         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
5283         little endian and -maltivec=be is not specified.
5284         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
5285         little endian and -maltivec=be is not specified.
5287 2014-02-05  Richard Henderson  <rth@redhat.com>
5289         PR debug/52727
5290         * combine-stack-adj.c: Revert r206943.
5291         * sched-int.h (struct deps_desc): Add last_args_size.
5292         * sched-deps.c (init_deps): Initialize it.
5293         (sched_analyze_insn): Add OUTPUT dependencies between insns that
5294         contain REG_ARGS_SIZE notes.
5296 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5298         * lto-cgraph.c (asm_nodes_output): Make global.
5299         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
5300         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
5301         (driver_handle_option): Handle OPT_fwpa.
5303 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
5305         PR ipa/59947
5306         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
5307         a comment typo and formatting issue.  If odr_hash hasn't been
5308         created, return vNULL and set *completep to false.
5310         PR middle-end/57499
5311         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
5312         bb with no successors.
5314 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
5316         PR target/59718
5317         * doc/invoke.texi (-march): Clarify documentation for ARM.
5318         (-mtune): Likewise.
5319         (-mcpu): Likewise.
5321 2014-02-05  Richard Biener  <rguenther@suse.de>
5323         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
5324         when not vectorizing because of too many alias checks.
5325         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
5326         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
5328 2014-02-05  Nick Clifton  <nickc@redhat.com>
5330         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
5331         accept extended registers in any mode when compiling for the MN10300.
5333 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
5335         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
5336         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
5337         sanitization attributes.
5338         (can_inline_edge_p): Likewise.
5339         (sanitize_attrs_match_for_inline_p): New function.
5341 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5343         * ipa-prop.c (detect_type_change): Shor circuit testing of
5344         type changes on THIS pointer.
5346 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
5348         PR target/59777
5349         * config/pa/pa.c (legitimize_tls_address): Return original address
5350         if not passed a SYMBOL_REF rtx.
5351         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
5352         addresses.
5353         (pa_emit_move_sequence): Simplify TLS source operands.
5354         (pa_legitimate_constant_p): Reject all TLS constants.
5355         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
5356         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
5358 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5360         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
5361         groups when we know they are controlled by LTO.
5362         * varasm.c (default_binds_local_p_1): If object is in other partition,
5363         it will be resolved locally.
5365 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5367         * config/host-linux.c (linux_gt_pch_use_address): Don't
5368         use SSIZE_MAX because it is not always defined.
5370 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
5372         PR bootstrap/59913
5373         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
5374         threshold for pseudo splitting.
5375         (update_ebb_live_info): Process call argument hard registers and
5376         hard registers from insn definition too.
5377         (max_small_class_regs_num): New constant.
5378         (inherit_in_ebb): Update live hard regs through EBBs.  Update
5379         reloads_num only for small register classes.  Don't split for
5380         outputs of jumps.
5382 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
5384         PR ipa/60058
5385         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
5386         is non-null.
5388 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5390         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
5391         visibility is safe.
5393 2014-02-04  Marek Polacek  <polacek@redhat.com>
5395         * gdbinit.in (pel): Define.
5397 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5399         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
5400         behavior.
5402 2014-02-04  Richard Biener  <rguenther@suse.de>
5404         PR lto/59723
5405         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
5406         in function context local.
5407         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
5408         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
5409         similar to LTO_imported_decl_ref.
5411 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5413         PR tree-optimization/60002
5414         * cgraphclones.c (build_function_decl_skip_args): Clear
5415         DECL_LANG_SPECIFIC.
5417         PR tree-optimization/60023
5418         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
5419         false to gsi_replace.
5420         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
5421         has been in some EH region and vec_stmt could throw, add
5422         vec_stmt into the same EH region.
5423         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
5424         has no lhs, ignore it.
5425         * internal-fn.c (expand_MASK_LOAD): Likewise.
5427         PR ipa/60026
5428         * tree-inline.c (copy_forbidden): Fail for
5429         __attribute__((optimize (0))) functions.
5431         PR other/58712
5432         * omp-low.c (simd_clone_struct_copy): If from->inbranch
5433         is set, copy one less argument.
5434         (expand_simd_clones): Don't subtract clone_info->inbranch
5435         from simd_clone_struct_alloc argument.
5437         PR rtl-optimization/57915
5438         * recog.c (simplify_while_replacing): If all unary/binary/relational
5439         operation arguments are constant, attempt to simplify those.
5441         PR middle-end/59261
5442         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
5443         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
5445 2014-02-04  Richard Biener  <rguenther@suse.de>
5447         PR tree-optimization/60012
5448         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
5449         TBAA disambiguation to all DDRs.
5451 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5453         PR target/59788
5454         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
5455         (LINK_SPEC): Use it for -shared, -shared-libgcc.
5457 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5459         PR ipa/59882
5460         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
5462 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5464         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
5465         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
5467 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5469         PR ipa/59831
5470         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
5471         to figure out targets of polymorphic calls with known decl.
5472         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5473         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
5474         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
5475         (get_polymorphic_call_info): ... here.
5476         (get_polymorphic_call_info_from_invariant): New function.
5478 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5480         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
5481         lookup via vtable pointer; check for type consistency
5482         and turn inconsitent facts into UNREACHABLE.
5483         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5484         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
5485         type inconsistent querries; return UNREACHABLE instead.
5487 2014-02-03  Richard Henderson  <rth@twiddle.net>
5489         PR tree-opt/59924
5490         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
5491         already processed this node.
5492         (normalize_one_pred_1): Pass along mark_set.
5493         (normalize_one_pred): Create and destroy a pointer_set_t.
5494         (normalize_one_pred_chain): Likewise.
5496 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
5498         PR gcov-profile/58602
5499         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
5501 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5503         PR ipa/59831
5504         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
5505         -fno-devirtualize; try to devirtualize by the knowledge of
5506         virtual table pointer given by aggregate propagation.
5507         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5508         (ipa_print_node_jump_functions): Dump also offset that
5509         is relevant for polymorphic calls.
5510         (determine_known_aggregate_parts): Add arg_type parameter; use it
5511         instead of determining the type from pointer type.
5512         (ipa_compute_jump_functions_for_edge): Update call of
5513         determine_known_aggregate_parts.
5514         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
5515         (gimple_get_virt_method_for_binfo): ... here; simplify using
5516         vtable_pointer_value_to_vtable.
5517         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
5518         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
5519         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
5520         (vtable_pointer_value_to_vtable): Break out from ...; handle also
5521         POINTER_PLUS_EXPR.
5522         (vtable_pointer_value_to_binfo): ... here.
5523         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
5525 2014-02-03  Teresa Johnson  <tejohnson@google.com>
5527         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
5528         redef of outer loop index variable.
5530 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
5532         PR c++/53017
5533         PR c++/59211
5534         * doc/extend.texi (Function Attributes): Typo.
5536 2014-02-03  Cong Hou  <congh@google.com>
5538         PR tree-optimization/60000
5539         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
5540         if the vectorized statement is a store.  A store statement can only
5541         appear at the end of pattern statements.
5543 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
5545         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
5546         (ix86_option_override_internal): Default long double to 64-bit for
5547         32-bit Bionic and to 128-bit for 64-bit Bionic.
5549         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
5550         TARGET_LONG_DOUBLE_128 is true.
5551         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
5553         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
5554         (mlong-double-64): Negate -mlong-double-128.
5555         (mlong-double-128): New option.
5557         * config/i386/i386-c.c (ix86_target_macros): Define
5558         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
5560         * doc/invoke.texi: Document -mlong-double-128.
5562 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
5564         PR rtl-optimization/60024
5565         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
5567 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
5569         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
5571 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
5573         PR rtl-optimization/57662
5574         * sel-sched.c (code_motion_path_driver): Do not mark already not
5575         existing blocks in the visiting bitmap.
5577 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
5579         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
5580         on the insn being emitted.
5582 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
5583             Will Deacon  <will.deacon@arm.com>
5585         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
5587 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5589         * config/arm/arm-tables.opt: Regenerate.
5591 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5593         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
5594         for vector types other than V16QImode.
5595         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
5596         define_expand, and call altivec_expand_vec_perm_le when producing
5597         code with little endian element order.
5598         (*altivec_vperm_<mode>_internal): New insn having previous
5599         behavior of altivec_vperm_<mode>.
5600         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
5601         altivec_expand_vec_perm_le when producing code with little endian
5602         element order.
5603         (*altivec_vperm_<mode>_uns_internal): New insn having previous
5604         behavior of altivec_vperm_<mode>_uns.
5606 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5608         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
5609         (altivec_vsumsws): Add handling for -maltivec=be with a little
5610         endian target.
5611         (altivec_vsumsws_direct): New.
5612         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
5613         gen_altivec_vsumsws.
5615 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
5617         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
5618         vtable_pointer_value_to_binfo): New functions.
5619         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
5620         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
5622 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
5624         * config/nios2/nios2.md (load_got_register): Initialize GOT
5625         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
5626         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
5628 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
5630         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
5631         preserverd by passthrough, do not propagate the type.
5633 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5635         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
5636         (mips_atomic_assign_expand_fenv): New function.
5637         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
5639 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5641         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
5642         (__builtin_mips_set_fcsr): Likewise.
5643         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
5644         MIPS_USI_FTYPE_VOID.
5645         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
5646         (mips16_expand_set_fcsr): Likewise.
5647         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
5648         (mips16_set_fcsr_stub): Likewise.
5649         (mips16_get_fcsr_one_only_stub): New class.
5650         (mips16_set_fcsr_one_only_stub): Likewise.
5651         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
5652         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
5653         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
5654         (hard_float): New availability predicate.
5655         (mips_builtins): Add get_fcsr and set_fcsr.
5656         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
5657         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
5658         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
5659         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
5660         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
5661         patterns.
5663 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5665         * config/mips/mips.c (mips_one_only_stub): New class.
5666         (mips_need_mips16_rdhwr_p): Replace with...
5667         (mips16_rdhwr_stub): ...this new variable.
5668         (mips16_stub_call_address): New function.
5669         (mips16_rdhwr_one_only_stub): New class.
5670         (mips_expand_thread_pointer): Use mips16_stub_call_address.
5671         (mips_output_mips16_rdhwr): Delete.
5672         (mips_finish_stub): New function.
5673         (mips_code_end): Use it to handle rdhwr stubs.
5675 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
5677         PR target/60017
5678         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
5679         when calculating size of integer atomic types.
5681 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
5683         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
5685 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
5687         PR tree-optimization/60003
5688         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
5689         * profile.c (branch_prob): Use gimple_call_builtin_p
5690         to check for BUILT_IN_SETJMP_RECEIVER.
5691         * tree-inline.c (copy_bb): Call notice_special_calls.
5693 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
5695         PR bootstrap/59985
5696         * lra-constraints.c (process_alt_operands): Update reload_sum only
5697         on the first pass.
5699 2014-01-31  Richard Henderson  <rth@redhat.com>
5701         PR middle-end/60004
5702         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
5703         until after else_eh is processed.
5705 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
5707         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
5708         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
5709         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
5710         in smmintrin.h, remove them.
5711         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
5712         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
5713         * config/i386/i386.md (ROUND_SAE): Fix value.
5714         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
5715         (const48_operand): New.
5716         * config/i386/subst.md (round), (round_expand): Use
5717         const_4_or_8_to_11_operand.
5718         (round_saeonly), (round_saeonly_expand): Use const48_operand.
5720 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
5722         * config/i386/constraints.md (Yk): Swap meaning with k.
5723         * config/i386/i386.md (movhi_internal): Change Yk to k.
5724         (movqi_internal): Ditto.
5725         (*k<logic><mode>): Ditto.
5726         (*andhi_1): Ditto.
5727         (*andqi_1): Ditto.
5728         (kandn<mode>): Ditto.
5729         (*<code>hi_1): Ditto.
5730         (*<code>qi_1): Ditto.
5731         (kxnor<mode>): Ditto.
5732         (kortestzhi): Ditto.
5733         (kortestchi): Ditto.
5734         (kunpckhi): Ditto.
5735         (*one_cmplhi2_1): Ditto.
5736         (*one_cmplqi2_1): Ditto.
5737         * config/i386/sse.md (): Change k to Yk.
5738         (avx512f_load<mode>_mask): Ditto.
5739         (avx512f_blendm<mode>): Ditto.
5740         (avx512f_store<mode>_mask): Ditto.
5741         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
5742         (avx512f_storedqu<mode>_mask): Ditto.
5743         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
5744         Ditto.
5745         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
5746         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
5747         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
5748         (avx512f_maskcmp<mode>3): Ditto.
5749         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
5750         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
5751         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
5752         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
5753         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
5754         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
5755         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
5756         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
5757         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
5758         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
5759         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
5760         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
5761         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
5762         (vec_extract_lo_<mode>_maskm): Ditto.
5763         (vec_extract_hi_<mode>_maskm): Ditto.
5764         (avx512f_vternlog<mode>_mask): Ditto.
5765         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
5766         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
5767         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
5768         (avx512f_<code>v8div16qi2_mask): Ditto.
5769         (avx512f_<code>v8div16qi2_mask_store): Ditto.
5770         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
5771         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
5772         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
5773         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
5774         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5775         (*avx512pf_gatherpf<mode>df_mask): Ditto.
5776         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5777         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5778         (avx512cd_maskb_vec_dupv8di): Ditto.
5779         (avx512cd_maskw_vec_dupv16si): Ditto.
5780         (avx512f_vpermi2var<mode>3_maskz): Ditto.
5781         (avx512f_vpermi2var<mode>3_mask): Ditto.
5782         (avx512f_vpermi2var<mode>3_mask): Ditto.
5783         (avx512f_vpermt2var<mode>3_maskz): Ditto.
5784         (*avx512f_gathersi<mode>): Ditto.
5785         (*avx512f_gathersi<mode>_2): Ditto.
5786         (*avx512f_gatherdi<mode>): Ditto.
5787         (*avx512f_gatherdi<mode>_2): Ditto.
5788         (*avx512f_scattersi<mode>): Ditto.
5789         (*avx512f_scatterdi<mode>): Ditto.
5790         (avx512f_compress<mode>_mask): Ditto.
5791         (avx512f_compressstore<mode>_mask): Ditto.
5792         (avx512f_expand<mode>_mask): Ditto.
5793         * config/i386/subst.md (mask): Change k to Yk.
5794         (mask_scalar_merge): Ditto.
5795         (sd): Ditto.
5797 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
5799         * doc/extend.texi (Vector Extensions): Document ?: in C++.
5801 2014-01-31  Richard Biener  <rguenther@suse.de>
5803         PR middle-end/59990
5804         * builtins.c (fold_builtin_memory_op): Make sure to not
5805         use a floating-point mode or a boolean or enumeral type for
5806         the copy operation.
5808 2014-01-30  DJ Delorie  <dj@redhat.com>
5810         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
5811         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
5812         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
5813         whenever main() has an epilogue.
5815 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5817         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
5818         unused variable "field".
5819         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
5820         (vsx_mergeh_<mode>): Likewise.
5821         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
5822         (altivec_vmrghh): Likewise.
5823         (altivec_vmrghw): Likewise.
5824         (altivec_vmrglb): Likewise.
5825         (altivec_vmrglh): Likewise.
5826         (altivec_vmrglw): Likewise.
5827         (altivec_vspltb): Add missing uses.
5828         (altivec_vsplth): Likewise.
5829         (altivec_vspltw): Likewise.
5830         (altivec_vspltsf): Likewise.
5832 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
5834         PR target/59923
5835         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
5836         frame related instructions.
5838 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
5840         PR rtl-optimization/59959
5841         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
5842         any reload of register whose subreg is invalid.
5844 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
5846         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
5847         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
5848         Add missing return type - void.
5850 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5852         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
5853         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
5854         remove element index adjustment for endian (now handled in vsx.md
5855         and altivec.md).
5856         (altivec_expand_vec_perm_const): Use
5857         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
5858         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
5859         (vsx_xxspltw_<mode>): Adjust element index for little endian.
5860         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
5861         define_expand and a new define_insn *altivec_vspltb_internal;
5862         adjust for -maltivec=be on a little endian target.
5863         (altivec_vspltb_direct): New.
5864         (altivec_vsplth): Divide into a define_expand and a new
5865         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
5866         little endian target.
5867         (altivec_vsplth_direct): New.
5868         (altivec_vspltw): Divide into a define_expand and a new
5869         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
5870         little endian target.
5871         (altivec_vspltw_direct): New.
5872         (altivec_vspltsf): Divide into a define_expand and a new
5873         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
5874         a little endian target.
5876 2014-01-30  Richard Biener  <rguenther@suse.de>
5878         PR tree-optimization/59993
5879         * tree-ssa-forwprop.c (associate_pointerplus): Check we
5880         can propagate form the earlier stmt and avoid the transform
5881         when the intermediate result is needed.
5883 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
5885         * README.Portability: Fix typo.
5887 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
5889         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
5890         comparison_operator with ordered_comparison_operator.
5892 2014-01-30  Nick Clifton  <nickc@redhat.com>
5894         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
5895         Rename to mn10300_store_multiple_regs.
5896         * config/mn10300/mn10300.c: Likewise.
5897         * config/mn10300/mn10300.md (store_movm): Fix typo: call
5898         store_multiple_regs.
5899         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
5900         Call mn10300_store_multiple_regs.
5902 2014-01-30  Nick Clifton  <nickc@redhat.com>
5903             DJ Delorie  <dj@redhat.com>
5905         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
5906         %fp 2 to keep registers after it properly word-aligned.
5907         (rl78_alloc_physical_registers_umul): Handle the case where both
5908         input operands are the same.
5910 2014-01-30  Richard Biener  <rguenther@suse.de>
5912         PR tree-optimization/59903
5913         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
5914         check properly.
5916 2014-01-30  Jason Merrill  <jason@redhat.com>
5918         PR c++/59633
5919         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
5921         PR c++/59645
5922         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
5924 2014-01-30  Richard Biener  <rguenther@suse.de>
5926         PR tree-optimization/59951
5927         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
5929 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
5931         PR target/59784
5932         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
5933         SFmode to DFmode case.
5935 2014-01-29  DJ Delorie  <dj@redhat.com>
5937         * config/msp430/msp430.opt (-minrt): New.
5938         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
5939         if -minrt given.
5940         (ENDFILE_SPEC): Likewise.
5942 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
5944         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
5945         (estimate_function_body_sizes): Use it.
5947 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
5949         PR c++/58561
5950         * dwarf2out.c (is_cxx_auto): New.
5951         (is_base_type): Use it.
5952         (gen_type_die_with_usage): Likewise.
5954 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5956         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
5957         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
5958         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
5959         -maltivec=be with LE targets.
5960         (vsx_mergeh_<mode>): Likewise.
5961         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
5962         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
5963         (altivec_vmrghb): Replace with define_expand and new
5964         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
5965         (altivec_vmrghb_direct): New define_insn.
5966         (altivec_vmrghh): Replace with define_expand and new
5967         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
5968         (altivec_vmrghh_direct): New define_insn.
5969         (altivec_vmrghw): Replace with define_expand and new
5970         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
5971         (altivec_vmrghw_direct): New define_insn.
5972         (*altivec_vmrghsf): Adjust for endianness.
5973         (altivec_vmrglb): Replace with define_expand and new
5974         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
5975         (altivec_vmrglb_direct): New define_insn.
5976         (altivec_vmrglh): Replace with define_expand and new
5977         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
5978         (altivec_vmrglh_direct): New define_insn.
5979         (altivec_vmrglw): Replace with define_expand and new
5980         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
5981         (altivec_vmrglw_direct): New define_insn.
5982         (*altivec_vmrglsf): Adjust for endianness.
5983         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5984         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5985         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5986         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5987         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5988         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5989         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5990         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5992 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5994         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
5995         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
5996         whitespace.
5998 2014-01-29  Richard Biener  <rguenther@suse.de>
6000         PR tree-optimization/58742
6001         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
6002         associate_pointerplus_align.
6003         (associate_pointerplus_diff): New function.
6004         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
6005         and associate_pointerplus_diff.
6007 2014-01-29  Richard Biener  <rguenther@suse.de>
6009         * lto-streamer.h (LTO_major_version): Bump to 3.
6010         (LTO_minor_version): Reset to 0.
6012 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
6014         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
6015         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
6016         (arm_file_start): Generate correct asm header for armv7ve.
6017         * config/arm/bpabi.h: Add multilib support for armv7ve.
6018         * config/arm/driver-arm.c: Change the architectures of cortex-a7
6019         and cortex-a15 to armv7ve.
6020         * config/arm/t-aprofile: Add multilib support for armv7ve.
6021         * doc/invoke.texi: Document -march=armv7ve.
6023 2014-01-29  Richard Biener  <rguenther@suse.de>
6025         PR tree-optimization/58742
6026         * tree-ssa-forwprop.c (associate_plusminus): Return true
6027         if we changed sth, defer EH cleanup to ...
6028         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
6029         (simplify_mult): New function.
6031 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6033         PR middle-end/59917
6034         PR tree-optimization/59920
6035         * tree.c (build_common_builtin_nodes): Remove
6036         __builtin_setjmp_dispatcher initialization.
6037         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
6038         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
6039         instead of gsi_after_labels + manually skipping debug stmts.
6040         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
6041         ignore bbs with IFN_ABNORMAL_DISPATCHER.
6042         * tree-inline.c (copy_edges_for_bb): Remove
6043         can_make_abnormal_goto argument, instead add abnormal_goto_dest
6044         argument.  Ignore computed_goto_p stmts.  Don't call
6045         make_abnormal_goto_edges.  If a call might need abnormal edges
6046         for non-local gotos, see if it already has an edge to
6047         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
6048         with true argument, don't do anything then, otherwise add
6049         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
6050         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
6051         caller.
6052         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
6053         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
6054         (lower_stmt): Don't set data->calls_builtin_setjmp.
6055         (lower_builtin_setjmp): Adjust comment.
6056         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
6057         * tree-cfg.c (found_computed_goto): Remove.
6058         (factor_computed_gotos): Remove.
6059         (make_goto_expr_edges): Return bool, true for computed gotos.
6060         Don't call make_abnormal_goto_edges.
6061         (build_gimple_cfg): Don't set found_computed_goto, don't call
6062         factor_computed_gotos.
6063         (computed_goto_p): No longer static.
6064         (make_blocks): Don't set found_computed_goto.
6065         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
6066         (make_edges): If make_goto_expr_edges returns true, push bb
6067         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
6068         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
6069         vector.  Record mapping between bbs and OpenMP regions if there
6070         are any, adjust make_gimple_omp_edges caller.  Call
6071         handle_abnormal_edges.
6072         (make_abnormal_goto_edges): Remove.
6073         * tree-cfg.h (make_abnormal_goto_edges): Remove.
6074         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
6075         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
6076         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
6077         * internal-fn.def (ABNORMAL_DISPATCHER): New.
6078         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
6079         filling *region also set *region_idx to (*region)->entry->index.
6081         PR other/58712
6082         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
6083         For REGs set ORIGINAL_REGNO.
6085 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
6087         * doc/md.texi: Mention that a target shouldn't implement
6088         vec_widen_(s|u)mul_even/odd pair if it is less efficient
6089         than hi/lo pair.
6091 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6093         PR tree-optimization/59594
6094         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
6095         a copy of the datarefs vector rather than the vector itself.
6097 2014-01-28  Jason Merrill  <jason@redhat.com>
6099         PR c++/53756
6100         * dwarf2out.c (auto_die): New static.
6101         (gen_type_die_with_usage): Handle C++1y 'auto'.
6102         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
6103         on definition.
6105 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
6107         PR target/59672
6108         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
6109         (SPEC_X32): Likewise.
6110         (SPEC_64): Likewise.
6111         * config/i386/i386.c (ix86_option_override_internal): Turn off
6112         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
6113         for TARGET_16BIT.
6114         (x86_file_start): Output .code16gcc for TARGET_16BIT.
6115         * config/i386/i386.h (TARGET_16BIT): New macro.
6116         (TARGET_16BIT_P): Likewise.
6117         * config/i386/i386.opt: Add m16.
6118         * doc/invoke.texi: Document -m16.
6120 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
6122         PR preprocessor/59935
6123         * input.c (location_get_source_line): Bail out on when line number
6124         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
6126 2014-01-28  Richard Biener  <rguenther@suse.de>
6128         PR tree-optimization/58742
6129         * tree-ssa-forwprop.c (associate_plusminus): Handle
6130         pointer subtraction of the form (T)(P + A) - (T)P.
6132 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6134         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
6135         at const_int_cost.
6137 2014-01-28  Richard Biener  <rguenther@suse.de>
6139         Revert
6140         2014-01-28  Richard Biener  <rguenther@suse.de>
6142         PR rtl-optimization/45364
6143         PR rtl-optimization/59890
6144         * var-tracking.c (local_get_addr_clear_given_value): Handle
6145         already cleared slot.
6146         (val_reset): Handle not allocated local_get_addr_cache.
6147         (vt_find_locations): Use post-order on the inverted CFG.
6149 2014-01-28  Richard Biener  <rguenther@suse.de>
6151         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
6153 2014-01-28  Richard Biener  <rguenther@suse.de>
6155         PR rtl-optimization/45364
6156         PR rtl-optimization/59890
6157         * var-tracking.c (local_get_addr_clear_given_value): Handle
6158         already cleared slot.
6159         (val_reset): Handle not allocated local_get_addr_cache.
6160         (vt_find_locations): Use post-order on the inverted CFG.
6162 2014-01-28  Alan Modra  <amodra@gmail.com>
6164         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
6165         * configure.ac <recursive call for build != host>: Define
6166         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
6167         and LD_FOR_BUILD too.
6168         * configure: Regenerate.
6170 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
6172         * config/i386/i386.c (get_builtin_code_for_version): Separate
6173         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
6174         Broadwell from Haswell.
6176 2014-01-27  Steve Ellcey  <sellcey@mips.com>
6178         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
6179         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
6180         * config/mips/mips.c (mips_option_override): Change setting
6181         of TARGET_DSP.
6182         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
6183         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
6184         Change from Mask to Var.
6186 2014-01-27  Jeff Law  <law@redhat.com>
6188         * ipa-inline.c (inline_small_functions): Fix typo.
6190 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
6192         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
6193         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
6194         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
6195         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
6196         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
6197         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
6198         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
6199         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
6200         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
6201         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
6202         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
6203         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
6204         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
6205         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
6206         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
6207         (_mm512_storeu_epi64): Ditto.
6208         (_mm512_cmpge_epi32_mask): Ditto.
6209         (_mm512_cmpge_epu32_mask): Ditto.
6210         (_mm512_cmpge_epi64_mask): Ditto.
6211         (_mm512_cmpge_epu64_mask): Ditto.
6212         (_mm512_cmple_epi32_mask): Ditto.
6213         (_mm512_cmple_epu32_mask): Ditto.
6214         (_mm512_cmple_epi64_mask): Ditto.
6215         (_mm512_cmple_epu64_mask): Ditto.
6216         (_mm512_cmplt_epi32_mask): Ditto.
6217         (_mm512_cmplt_epu32_mask): Ditto.
6218         (_mm512_cmplt_epi64_mask): Ditto.
6219         (_mm512_cmplt_epu64_mask): Ditto.
6220         (_mm512_cmpneq_epi32_mask): Ditto.
6221         (_mm512_cmpneq_epu32_mask): Ditto.
6222         (_mm512_cmpneq_epi64_mask): Ditto.
6223         (_mm512_cmpneq_epu64_mask): Ditto.
6224         (_mm512_expand_pd): Ditto.
6225         (_mm512_expand_ps): Ditto.
6226         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
6227         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
6228         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
6229         * config/i386/i386.c (ix86_builtins): Add
6230         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
6231         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
6232         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
6233         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
6234         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
6235         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
6236         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
6237         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
6238         IX86_BUILTIN_PMOVUSQW512_MEM.
6239         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
6240         __builtin_ia32_pmovsqd512mem_mask,
6241         __builtin_ia32_pmovqd512mem_mask,
6242         __builtin_ia32_pmovusqw512mem_mask,
6243         __builtin_ia32_pmovsqw512mem_mask,
6244         __builtin_ia32_pmovqw512mem_mask,
6245         __builtin_ia32_pmovusdw512mem_mask,
6246         __builtin_ia32_pmovsdw512mem_mask,
6247         __builtin_ia32_pmovdw512mem_mask,
6248         __builtin_ia32_pmovqb512mem_mask,
6249         __builtin_ia32_pmovusqb512mem_mask,
6250         __builtin_ia32_pmovsqb512mem_mask,
6251         __builtin_ia32_pmovusdb512mem_mask,
6252         __builtin_ia32_pmovsdb512mem_mask,
6253         __builtin_ia32_pmovdb512mem_mask.
6254         (bdesc_args): Add __builtin_ia32_expanddf512,
6255         __builtin_ia32_expandsf512.
6256         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
6257         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
6258         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
6259         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
6260         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
6261         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
6262         (avx512f_<code>v8div16qi2_mask_store): This.
6263         (avx512f_expand<mode>): New.
6265 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
6267         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
6268         New.
6269         (_mm512_mask_prefetch_i64gather_pd): Ditto.
6270         (_mm512_prefetch_i32scatter_pd): Ditto.
6271         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6272         (_mm512_prefetch_i64scatter_pd): Ditto.
6273         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6274         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
6275         (_mm512_mask_prefetch_i64gather_ps): Ditto.
6276         (_mm512_prefetch_i32scatter_ps): Ditto.
6277         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6278         (_mm512_prefetch_i64scatter_ps): Ditto.
6279         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6280         * config/i386/i386-builtin-types.def: Define
6281         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
6282         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
6283         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
6284         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
6285         IX86_BUILTIN_SCATTERPFQPD.
6286         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
6287         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
6288         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
6289         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
6290         __builtin_ia32_scatterpfqps.
6291         (ix86_expand_builtin): Expand new built-ins.
6292         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
6293         fix memory access data type.
6294         (*avx512pf_gatherpf<mode>_mask): Ditto.
6295         (*avx512pf_gatherpf<mode>): Ditto.
6296         (avx512pf_scatterpf<mode>): Ditto.
6297         (*avx512pf_scatterpf<mode>_mask): Ditto.
6298         (*avx512pf_scatterpf<mode>): Ditto.
6299         (GATHER_SCATTER_SF_MEM_MODE): New.
6300         (avx512pf_gatherpf<mode>df): Ditto.
6301         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6302         (*avx512pf_scatterpf<mode>df): Ditto.
6304 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
6306         PR bootstrap/59934
6307         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
6308         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
6309         reached.
6311 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6313         * common/config/arm/arm-common.c
6314         (arm_rewrite_mcpu): Handle multiple names.
6315         * config/arm/arm.h
6316         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6318 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6320         * gimple-builder.h (create_gimple_tmp): Delete.
6322 2014-01-27  Christian Bruel  <christian.bruel@st.com>
6324         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
6325         words comparisons.
6327 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
6329         * config/pa/pa.md (call): Generate indirect long calls to non-local
6330         functions when outputing 32-bit code.
6331         (call_value): Likewise except for special call to buggy powf function.
6333         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
6334         portable runtime and PIC indirect calls.
6335         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
6336         and PIC call sequences.  Use ldo instead of blr to set return register
6337         in PIC call sequence.
6339 2014-01-25  Walter Lee  <walt@tilera.com>
6341         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
6342         avoid clobbering a live register.
6344 2014-01-25  Walter Lee  <walt@tilera.com>
6346         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
6347         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
6348         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
6349         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
6351 2014-01-25  Walter Lee  <walt@tilera.com>
6353         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
6354         arguments on even registers.
6355         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
6356         STACK_BOUNDARY.
6357         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
6358         (BIGGEST_ALIGNMENT): Ditto.
6359         (BIGGEST_FIELD_ALIGNMENT): Ditto.
6361 2014-01-25  Walter Lee  <walt@tilera.com>
6363         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
6364         insns before bundling.
6365         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
6367 2014-01-25  Walter Lee  <walt@tilera.com>
6369         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
6370         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
6371         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
6373 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6375         * config/mips/constraints.md (kl): Delete.
6376         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
6377         define expands, using...
6378         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
6379         instructions for MIPS16.
6380         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
6381         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
6383 2014-01-25  Walter Lee  <walt@tilera.com>
6385         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
6386         (clzdi2): Ditto.
6387         (ffsdi2): Ditto.
6389 2014-01-25  Walter Lee  <walt@tilera.com>
6391         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
6392         (TARGET_EXPAND_TO_RTL_HOOK): Define.
6394 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6396         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
6397         Handle XOR.
6399 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
6401         * print-rtl.c (in_call_function_usage): New var.
6402         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
6403         EXPR_LIST mode as mode and not as reg note name.
6405         PR middle-end/59561
6406         * cfgloopmanip.c (copy_loop_info): If
6407         loop->warned_aggressive_loop_optimizations, make sure
6408         the flag is set in target loop too.
6410 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6412         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
6413         flag_cilkplus.
6414         * builtins.def: Likewise.
6415         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
6416         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
6417         * ira.c (ira_setup_eliminable_regset): Likewise.
6418         * omp-low.c (gate_expand_omp): Likewise.
6419         (execute_lower_omp): Likewise.
6420         (diagnose_sb_0): Likewise.
6421         (gate_diagnose_omp_blocks): Likewise.
6422         (simd_clone_clauses_extract): Likewise.
6423         (gate): Likewise.
6425 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6427         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
6428         correction for little endian...
6429         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
6430         here.
6432 2014-01-24  Jeff Law  <law@redhat.com>
6434         PR tree-optimization/59919
6435         * tree-vrp.c (find_assert_locations_1): Do not register asserts
6436         for non-returning calls.
6438 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
6440         * common/config/aarch64/aarch64-common.c
6441         (aarch64_rewrite_mcpu): Handle multiple names.
6442         * config/aarch64/aarch64.h
6443         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6445 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
6447         * input.c (add_file_to_cache_tab): Handle the case where fopen
6448         returns NULL.
6450 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
6452         PR target/59929
6453         * config/i386/i386.md (pushsf splitter): Get stack adjustment
6454         from push operand if code of push isn't PRE_DEC.
6456 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6458         PR target/59909
6459         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
6460         -mquad-memory-atomic.  Update -mquad-memory documentation to say
6461         it is only used for non-atomic loads/stores.
6463         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
6464         -mquad-memory or -mquad-memory-atomic switches.
6466         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
6467         -mquad-memory-atomic to ISA 2.07 support.
6469         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
6470         to separate support of normal quad word memory operations (ldq, stq)
6471         from the atomic quad word memory operations.
6473         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
6474         support to separate non-atomic quad word operations from atomic
6475         quad word operations.  Disable non-atomic quad word operations in
6476         little endian mode so that we don't have to swap words after the
6477         load and before the store.
6478         (quad_load_store_p): Add comment about atomic quad word support.
6479         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
6480         options printed with -mdebug=reg.
6482         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
6483         -mquad-memory-atomic as the test for whether we have quad word
6484         atomic instructions.
6485         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
6486         or -mp8-vector are used, allow byte/half-word atomic operations.
6488         * config/rs6000/sync.md (load_lockedti): Insure that the address
6489         is a proper indexed or indirect address for the lqarx instruction.
6490         On little endian systems, swap the hi/lo registers after the lqarx
6491         instruction.
6492         (load_lockedpti): Use indexed_or_indirect_operand predicate to
6493         insure the address is valid for the lqarx instruction.
6494         (store_conditionalti): Insure that the address is a proper indexed
6495         or indirect address for the stqcrx. instruction.  On little endian
6496         systems, swap the hi/lo registers before doing the stqcrx.
6497         instruction.
6498         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
6499         insure the address is valid for the stqcrx. instruction.
6501         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
6502         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
6503         type of quad memory support is available.
6505 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
6507         PR regression/59915
6508         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
6509         there is a danger of looping.
6511 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
6513         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6514         force flag_ira_loop_pressure if set via command line.
6516 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6518         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
6519         (ashr_simd): New builtin handling DI mode.
6520         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
6521         (aarch64_sshr_simddi): New match pattern.
6522         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
6523         (vshrd_n_s64): Likewise.
6524         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
6526 2014-01-23  Nick Clifton  <nickc@redhat.com>
6528         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
6529         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
6530         favour of mcu specific scripts.
6531         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
6532         430x multilibs.
6534 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
6535             Alex Velenko  <Alex.Velenko@arm.com>
6537         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
6538         (vaddv_s16): Likewise.
6539         (vaddv_s32): Likewise.
6540         (vaddv_u8): Likewise.
6541         (vaddv_u16): Likewise.
6542         (vaddv_u32): Likewise.
6543         (vaddvq_s8): Likewise.
6544         (vaddvq_s16): Likewise.
6545         (vaddvq_s32): Likewise.
6546         (vaddvq_s64): Likewise.
6547         (vaddvq_u8): Likewise.
6548         (vaddvq_u16): Likewise.
6549         (vaddvq_u32): Likewise.
6550         (vaddvq_u64): Likewise.
6551         (vaddv_f32): Likewise.
6552         (vaddvq_f32): Likewise.
6553         (vaddvq_f64): Likewise.
6554         (vmaxv_f32): Likewise.
6555         (vmaxv_s8): Likewise.
6556         (vmaxv_s16): Likewise.
6557         (vmaxv_s32): Likewise.
6558         (vmaxv_u8): Likewise.
6559         (vmaxv_u16): Likewise.
6560         (vmaxv_u32): Likewise.
6561         (vmaxvq_f32): Likewise.
6562         (vmaxvq_f64): Likewise.
6563         (vmaxvq_s8): Likewise.
6564         (vmaxvq_s16): Likewise.
6565         (vmaxvq_s32): Likewise.
6566         (vmaxvq_u8): Likewise.
6567         (vmaxvq_u16): Likewise.
6568         (vmaxvq_u32): Likewise.
6569         (vmaxnmv_f32): Likewise.
6570         (vmaxnmvq_f32): Likewise.
6571         (vmaxnmvq_f64): Likewise.
6572         (vminv_f32): Likewise.
6573         (vminv_s8): Likewise.
6574         (vminv_s16): Likewise.
6575         (vminv_s32): Likewise.
6576         (vminv_u8): Likewise.
6577         (vminv_u16): Likewise.
6578         (vminv_u32): Likewise.
6579         (vminvq_f32): Likewise.
6580         (vminvq_f64): Likewise.
6581         (vminvq_s8): Likewise.
6582         (vminvq_s16): Likewise.
6583         (vminvq_s32): Likewise.
6584         (vminvq_u8): Likewise.
6585         (vminvq_u16): Likewise.
6586         (vminvq_u32): Likewise.
6587         (vminnmv_f32): Likewise.
6588         (vminnmvq_f32): Likewise.
6589         (vminnmvq_f64): Likewise.
6591 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
6593         * config/aarch64/aarch64-simd.md
6594         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
6595         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
6596         (*aarch64_mul3_elt<mode>): Likewise.
6597         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
6598         (*aarch64_mul3_elt_to_64v2df): Likewise.
6599         (*aarch64_mla_elt<mode>): Likewise.
6600         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
6601         (*aarch64_mls_elt<mode>): Likewise.
6602         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
6603         (*aarch64_fma4_elt<mode>): Likewise.
6604         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
6605         (*aarch64_fma4_elt_to_64v2df): Likewise.
6606         (*aarch64_fnma4_elt<mode>): Likewise.
6607         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6608         (*aarch64_fnma4_elt_to_64v2df): Likewise.
6609         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
6610         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
6611         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6612         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6613         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6614         (aarch64_sqdmull_lane<mode>_internal): Likewise.
6615         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6617 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6619         * config/aarch64/aarch64-simd.md
6620         (aarch64_be_checked_get_lane<mode>): New define_expand.
6621         * config/aarch64/aarch64-simd-builtins.def
6622         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
6623         New builtin definition.
6624         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
6625         Use new safe be builtin.
6627 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6629         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
6630         New define_insn.
6631         (aarch64_be_st1<mode>): Likewise.
6632         (aarch_ld1<VALL:mode>): Define_expand modified.
6633         (aarch_st1<VALL:mode>): Likewise.
6634         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
6635         (UNSPEC_ST1): Likewise.
6637 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
6639         * config/microblaze/microblaze.md: Add trap insn and attribute
6641 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
6643         PR preprocessor/58580
6644         * input.h (location_get_source_line): Take an additional line_size
6645         parameter.
6646         (void diagnostics_file_cache_fini): Declare new function.
6647         * input.c (struct fcache): New type.
6648         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
6649         New static constants.
6650         (diagnostic_file_cache_init, total_lines_num)
6651         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
6652         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
6653         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
6654         (get_next_line, read_next_line, goto_next_line, read_line_num):
6655         New static function definitions.
6656         (diagnostic_file_cache_fini): New function.
6657         (location_get_source_line): Take an additional output line_len
6658         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
6659         read_line_num.
6660         * diagnostic.c (diagnostic_finish): Call
6661         diagnostic_file_cache_fini.
6662         (adjust_line): Take an additional input parameter for the length
6663         of the line, rather than calculating it with strlen.
6664         (diagnostic_show_locus): Adjust the use of
6665         location_get_source_line and adjust_line with respect to their new
6666         signature.  While displaying a line now, do not stop at the first
6667         null byte.  Rather, display the zero byte as a space and keep
6668         going until we reach the size of the line.
6669         * Makefile.in: Add vec.o to OBJS-libcommon
6671 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
6672             Ilya Tocar     <ilya.tocar@intel.com>
6674         * config/i386/avx512fintrin.h (_mm512_kmov): New.
6675         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
6676         (__builtin_ia32_kmov16): Ditto.
6677         * config/i386/i386.md (UNSPEC_KMOV): New.
6678         (kmovw): Ditto.
6680 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
6682         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
6683         (_mm512_storeu_si512): Ditto.
6685 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
6687         PR target/52125
6688         * rtl.h (get_referenced_operands): Declare.
6689         * recog.c (get_referenced_operands): New function.
6690         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
6691         operands have been referenced when recording LO_SUM references.
6693 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
6695         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
6697 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
6699         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
6700         Enable for generic and recent AMD targets.
6702 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
6704         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
6705         ARG_SIZE note when adjustment was eliminated.
6707 2014-01-22  Jeff Law  <law@redhat.com>
6709         PR tree-optimization/59597
6710         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
6711         in file.  Accept new argument REGISTERING and use it to modify
6712         dump output appropriately.
6713         (register_jump_thread): Corresponding changes.
6714         (mark_threaded_blocks): Reinstate code to cancel unprofitable
6715         thread paths involving joiner blocks.  Add code to dump cancelled
6716         jump threading paths.
6718 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
6720         PR rtl-optimization/59477
6721         * lra-constraints.c (inherit_in_ebb): Process call for living hard
6722         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
6724 2014-01-22  Tom Tromey  <tromey@redhat.com>
6726         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
6727         PARAMS.
6728         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
6730 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
6732         PR rtl-optimization/59896
6733         * lra-constraints.c (process_alt_operands): Check unused note for
6734         matched operands of insn with no output reloads.
6736 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
6738         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
6739         (mips_move_from_gpr_cost): Likewise.
6741 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
6743         PR rtl-optimization/59858
6744         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
6745         ira_class_hard_regs_num.
6746         (process_alt_operands): Increase reject for dying matched operand.
6748 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
6750         PR target/59003
6751         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
6752         smaller than size, perform several stores or loads and stores
6753         at dst + count - size to store or copy all of size bytes, rather
6754         than just last modesize bytes.
6756 2014-01-20  DJ Delorie  <dj@redhat.com>
6758         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
6759         that CLOBBERs are REGs before propogating their values.
6761 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
6763         PR middle-end/59789
6764         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
6765         (cgraph_inline_failed_type): New function.
6766         * cgraph.h (DEFCIFCODE): Add type.
6767         (cgraph_inline_failed_type_t): New enum.
6768         (cgraph_inline_failed_type): New prototype.
6769         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
6770         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
6771         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
6772         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
6773         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
6774         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
6775         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
6776         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
6777         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
6778         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
6779         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
6780         OPTIMIZATION_MISMATCH.
6781         * tree-inline.c (expand_call_inline): Emit errors during
6782         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
6784 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
6786         PR target/59685
6787         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
6788         mode attribute in insn output.
6790 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
6792         * output.h (output_constant): Delete.
6793         * varasm.c (output_constant): Make private.
6795 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
6797         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
6799 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
6801         PR middle-end/59860
6802         * tree.h (fold_builtin_strcat): New prototype.
6803         * builtins.c (fold_builtin_strcat): No longer static.  Add len
6804         argument, if non-NULL, don't call c_strlen.  Optimize
6805         directly into __builtin_memcpy instead of __builtin_strcpy.
6806         (fold_builtin_2): Adjust fold_builtin_strcat caller.
6807         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
6809 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
6811         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6812         for SImode_address_operand operands, having only a REG argument.
6814 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6816         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
6817         loader name using mbig-endian.
6818         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
6820 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
6822         * doc/invoke.texi (-march): Clarify documentation for AArch64.
6823         (-mtune): Likewise.
6824         (-mcpu): Likewise.
6826 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
6828         * config/aarch64/aarch64-protos.h
6829         (aarch64_cannot_change_mode_class_ptr): Declare.
6830         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
6831         aarch64_cannot_change_mode_class_ptr): New.
6832         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
6833         backend hook aarch64_cannot_change_mode_class.
6835 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
6837         * common/config/aarch64/aarch64-common.c
6838         (aarch64_handle_option): Don't handle any option order logic here.
6839         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
6840         selected_cpu, warn on architecture version mismatch.
6841         (aarch64_override_options): Fix parsing order for option strings.
6843 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6844             Iain Sandoe  <iain@codesourcery.com>
6846         PR bootstrap/59496
6847         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
6848         warning.  Amend comment to reflect current functionality.
6850 2014-01-20  Richard Biener  <rguenther@suse.de>
6852         PR middle-end/59860
6853         * builtins.c (fold_builtin_strcat): Remove case better handled
6854         by tree-ssa-strlen.c.
6856 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
6858         * config/aarch64/aarch64.opt
6859         (mcpu, march, mtune): Make case-insensitive.
6861 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
6863         PR target/59880
6864         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6865         if operands[1] is a REG or ZERO_EXTEND of a REG.
6867 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
6869         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
6871 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
6873         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
6874         long non-pic millicode calls.
6876 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6878         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
6880 2014-01-19  Kito Cheng  <kito@0xlab.org>
6882         * builtins.c (expand_movstr): Check movstr expand done or fail.
6884 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
6885             H.J. Lu  <hongjiu.lu@intel.com>
6887         PR target/59379
6888         * config/i386/i386.md (*lea<mode>): Zero-extend return register
6889         to DImode for zero-extended addresses.
6891 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
6893         PR rtl-optimization/57763
6894         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
6895         on the new indirect jump_insn and increment LABEL_NUSES (label).
6897 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
6899         PR bootstrap/59580
6900         PR bootstrap/59583
6901         * config.gcc (x86_archs): New variable.
6902         (x86_64_archs): Likewise.
6903         (x86_cpus): Likewise.
6904         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
6905         --with-arch/--with-cpu= options.
6906         Support --with-arch=/--with-cpu={nehalem,westmere,
6907         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
6909 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
6911         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
6912         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
6914 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
6916         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
6918 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
6920         PR target/58944
6921         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
6922         clear cpp_get_options (parse_in)->warn_unused_macros for
6923         ix86_target_macros_internal with cpp_define.
6925 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
6927         * jump.c (delete_related_insns): Keep (use (insn))s.
6928         * reorg.c (redundant_insn): Check for barriers too.
6930 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
6932         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
6934 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
6936         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
6937         call to $$dyncall when TARGET_LONG_CALLS is true.
6939 2014-01-17  Jeff Law  <law@redhat.com>
6941         * ree.c (combine_set_extension): Temporarily disable test for
6942         changing number of hard registers.
6944 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
6946         PR middle-end/58125
6947         * ipa-inline-analysis.c (inline_free_summary):
6948         Do not free summary of aliases.
6950 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
6952         PR middle-end/59706
6953         * gimplify.c (gimplify_expr): Use create_tmp_var
6954         instead of create_tmp_var_raw.  If cond doesn't have
6955         integral type, don't add the IFN_ANNOTATE builtin at all.
6957 2014-01-17  Martin Jambor  <mjambor@suse.cz>
6959         PR ipa/59736
6960         * ipa-cp.c (prev_edge_clone): New variable.
6961         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
6962         Also resize prev_edge_clone vector.
6963         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
6964         (ipcp_edge_removal_hook): New function.
6965         (ipcp_driver): Register ipcp_edge_removal_hook.
6967 2014-01-17  Andrew Pinski  <apinski@cavium.com>
6968             Steve Ellcey  <sellcey@mips.com>
6970         PR target/59462
6971         * config/mips/mips.c (mips_print_operand): Check operand mode instead
6972         of operator mode.
6974 2014-01-17  Jeff Law  <law@redhat.com>
6976         PR middle-end/57904
6977         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
6978         so that pass_ccp runs first.
6980 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
6982         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
6983         (ix86_adjust_cost): Use !TARGET_XXX.
6984         (do_reorder_for_imul): Likewise.
6985         (swap_top_of_ready_list): Likewise.
6986         (ix86_sched_reorder): Likewise.
6988 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
6990         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6991         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
6992         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
6993         (intel_memset): New.  Duplicate slm_memset.
6994         (intel_cost): New.  Duplicate slm_cost.
6995         (m_INTEL): New macro.
6996         (processor_target_table): Add "intel".
6997         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
6998         with PROCESSOR_INTEL for "intel".
6999         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
7000         PROCESSOR_SILVERMONT.
7001         (ix86_issue_rate): Likewise.
7002         (ix86_adjust_cost): Likewise.
7003         (ia32_multipass_dfa_lookahead): Likewise.
7004         (swap_top_of_ready_list): Likewise.
7005         (ix86_sched_reorder): Likewise.
7006         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
7007         instead of TARGET_OPT_AGU.
7008         * config/i386/i386.h (TARGET_INTEL): New.
7009         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
7010         (processor_type): Add PROCESSOR_INTEL.
7011         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
7012         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
7014 2014-01-17  Marek Polacek  <polacek@redhat.com>
7016         PR c/58346
7017         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
7018         size is zero.
7020 2014-01-17  Richard Biener  <rguenther@suse.de>
7022         PR tree-optimization/46590
7023         * opts.c (default_options_table): Add entries for
7024         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
7025         all enabled at -O1 but not for -Og.
7026         * common.opt (fbranch-count-reg): Remove Init(1).
7027         (fmove-loop-invariants): Likewise.
7028         (ftree-pta): Likewise.
7030 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7032         * config/i386/i386.c (ix86_data_alignment): For compatibility with
7033         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
7034         decls to at least the GCC 4.8 used alignments.
7036         PR fortran/59440
7037         * tree-nested.c (convert_nonlocal_reference_stmt,
7038         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
7039         of GIMPLE_BIND stmts, adjust associated decls.
7041 2014-01-17  Richard Biener  <rguenther@suse.de>
7043         PR tree-optimization/46590
7044         * vec.h (vec<>::bseach): New member function implementing
7045         binary search according to C89 bsearch.
7046         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
7047         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
7048         bitmap pointer again.  Make accesses_in_loop a flat array.
7049         (mem_ref_obstack): New global.
7050         (outermost_indep_loop): Adjust for mem_ref->stored changes.
7051         (mark_ref_stored): Likewise.
7052         (ref_indep_loop_p_2): Likewise.
7053         (set_ref_stored_in_loop): New helper function.
7054         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
7055         (memref_free): Adjust.
7056         (record_mem_ref_loc): Simplify.
7057         (gather_mem_refs_stmt): Adjust.
7058         (sort_locs_in_loop_postorder_cmp): New function.
7059         (analyze_memory_references): Sort accesses_in_loop after
7060         loop postorder number.
7061         (find_ref_loc_in_loop_cmp): New function.
7062         (for_all_locs_in_loop): Find relevant cluster of locs in
7063         accesses_in_loop and iterate without recursion.
7064         (execute_sm): Avoid uninit warning.
7065         (struct ref_always_accessed): Simplify.
7066         (ref_always_accessed::operator ()): Likewise.
7067         (ref_always_accessed_p): Likewise.
7068         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
7069         loop postorder numbers here.
7070         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
7071         numbers.
7073 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7075         PR c++/57945
7076         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
7077         on decls for which assemble_alias has been called.
7079 2014-01-17  Nick Clifton  <nickc@redhat.com>
7081         * config/msp430/msp430.opt: (mcpu): New option.
7082         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
7083         (msp430_option_override): Parse target_cpu.  If the MCU name
7084         matches a generic string, clear target_mcu.
7085         (msp430_attr): Allow numeric interrupt values up to 63.
7086         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
7087         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
7088         option.
7089         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
7090         Add mcpu matches.
7091         * config/msp430/msp430.md (popm): Use %J rather than %I.
7092         (addsi3): Use msp430_nonimmediate_operand for operand 2.
7093         (addhi_cy_i): Use immediate_operand for operand 2.
7094         * doc/invoke.texi: Document -mcpu option.
7096 2014-01-17  Richard Biener  <rguenther@suse.de>
7098         PR rtl-optimization/38518
7099         * df.h (df_analyze_loop): Declare.
7100         * df-core.c: Include cfgloop.h.
7101         (df_analyze_1): Split out main part of df_analyze.
7102         (df_analyze): Adjust.
7103         (loop_inverted_post_order_compute): New function.
7104         (loop_post_order_compute): Likewise.
7105         (df_analyze_loop): New function avoiding whole-function
7106         postorder computes.
7107         * loop-invariant.c (find_defs): Use df_analyze_loop.
7108         (find_invariants): Adjust.
7109         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
7111 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
7113         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
7114         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
7116 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
7118         * ipa-ref.c (ipa_remove_stmt_references): Fix references
7119         traversal when removing references.
7121 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
7123         PR ipa/59775
7124         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
7126 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
7128         PR middle-end/56791
7129         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
7130         pushing a reload for an autoinc when we had previously reloaded an
7131         inner part of the address.
7133 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
7135         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
7136         field.
7137         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
7138         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
7139         when not giving up or versioning for alias only because of
7140         loop->safelen.
7141         (vect_analyze_data_ref_dependences): Set to true.
7142         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
7143         is a GIMPLE_PHI.
7144         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
7145         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
7146         to the condition.
7148         PR middle-end/58344
7149         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
7151         PR target/59839
7152         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
7153         operand 0 predicate for gathers, use a new pseudo as subtarget.
7155 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7157         PR middle-end/59609
7158         * lra-constraints.c (process_alt_operands): Add printing debug info.
7159         Check absence of input/output reloads for matched operands too.
7161 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7163         PR rtl-optimization/59835
7164         * ira.c (ira_init_register_move_cost): Increase cost for
7165         impossible modes.
7167 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
7169         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
7171 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
7173         PR target/59780
7174         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
7175         non-register objects.  Use gen_(high/low)part more consistently.
7176         Fix assertions.
7178 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
7180         PR target/59844
7181         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
7182         endian support, remove tests for WORDS_BIG_ENDIAN.
7183         (p8_mfvsrd_3_<mode>): Likewise.
7184         (reload_gpr_from_vsx<mode>): Likewise.
7185         (reload_gpr_from_vsxsf): Likewise.
7186         (p8_mfvsrd_4_disf): Likewise.
7188 2014-01-16  Richard Biener  <rguenther@suse.de>
7190         PR rtl-optimization/46590
7191         * lcm.c (compute_antinout_edge): Use postorder iteration.
7192         (compute_laterin): Use inverted postorder iteration.
7194 2014-01-16  Nick Clifton  <nickc@redhat.com>
7196         PR middle-end/28865
7197         * varasm.c (output_constant): Return the number of bytes actually
7198         emitted.
7199         (output_constructor_array_range): Update the field size with the
7200         number of bytes emitted by output_constant.
7201         (output_constructor_regular_field): Likewise.  Also do not
7202         complain if the total number of bytes emitted is now greater
7203         than the expected fieldpos.
7204         * output.h (output_constant): Update prototype and descriptive comment.
7206 2014-01-16  Marek Polacek  <polacek@redhat.com>
7208         PR middle-end/59827
7209         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
7210         it is error_mark_node.
7212 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
7214         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
7215         VALID_AVX256_REG_OR_OI_MODE.
7217 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
7219         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
7220         current procedure should be profiled.
7222 2014-01-15  Andrew Pinski  <apinski@cavium.com>
7224         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
7225         of moving from/to the STACK_REG register class.
7227 2014-01-15  Richard Henderson  <rth@redhat.com>
7229         PR debug/54694
7230         * reginfo.c (global_regs_decl): Globalize.
7231         * rtl.h (global_regs_decl): Declare.
7232         * ira.c (do_reload): Diagnose frame_pointer_needed and it
7233         reserved via global_regs.
7235 2014-01-15  Teresa Johnson  <tejohnson@google.com>
7237         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
7239 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7241         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
7242         and vmulosh rather than call gen_vec_widen_smult_*.
7243         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
7244         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
7245         (vec_widen_smult_even_v16qi): Likewise.
7246         (vec_widen_umult_even_v8hi): Likewise.
7247         (vec_widen_smult_even_v8hi): Likewise.
7248         (vec_widen_umult_odd_v16qi): Likewise.
7249         (vec_widen_smult_odd_v16qi): Likewise.
7250         (vec_widen_umult_odd_v8hi): Likewise.
7251         (vec_widen_smult_odd_v8hi): Likewise.
7252         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
7253         vmuloub rather than call gen_vec_widen_umult_*.
7254         (vec_widen_umult_lo_v16qi): Likewise.
7255         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
7256         vmulosb rather than call gen_vec_widen_smult_*.
7257         (vec_widen_smult_lo_v16qi): Likewise.
7258         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
7259         rather than call gen_vec_widen_umult_*.
7260         (vec_widen_umult_lo_v8hi): Likewise.
7261         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
7262         rather than call gen_vec_widen_smult_*.
7263         (vec_widen_smult_lo_v8hi): Likewise.
7265 2014-01-15  Jeff Law  <law@redhat.com>
7267         PR tree-optimization/59747
7268         * ree.c (find_and_remove_re): Properly handle case where a second
7269         eliminated extension requires widening a copy created for elimination
7270         of a prior extension.
7271         (combine_set_extension): Ensure that the number of hard regs needed
7272         for a destination register does not change when we widen it.
7274 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7276         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
7277         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
7278         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
7279         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
7280         (avr-*-rtems*): Likewise.
7281         (bfin*-rtems*): Likewise.
7282         (moxie-*-rtems*): Likewise.
7283         (h8300-*-rtems*): Likewise.
7284         (i[34567]86-*-rtems*): Likewise.
7285         (lm32-*-rtems*): Likewise.
7286         (m32r-*-rtems*): Likewise.
7287         (m68k-*-rtems*): Likewise.
7288         (microblaze*-*-rtems*): Likewise.
7289         (mips*-*-rtems*): Likewise.
7290         (powerpc-*-rtems*): Likewise.
7291         (sh-*-rtems*): Likewise.
7292         (sparc-*-rtems*): Likewise.
7293         (sparc64-*-rtems*): Likewise.
7294         (v850-*-rtems*): Likewise.
7295         (m32c-*-rtems*): Likewise.
7297 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
7299         PR rtl-optimization/59511
7300         * ira.c (ira_init_register_move_cost): Use memory costs for some
7301         cases of register move cost calculations.
7302         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
7303         instead of BB frequency.
7304         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
7305         * lra-assigns.c (find_hard_regno_for): Ditto.
7307 2014-01-15  Richard Biener  <rguenther@suse.de>
7309         PR tree-optimization/59822
7310         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
7311         (vectorizable_load): Use it to hoist defs of uses of invariant
7312         loads out of the loop.
7314 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
7315             Kugan Vivekanandarajah  <kuganv@linaro.org>
7317         PR target/59695
7318         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
7319         truncation.
7321 2014-01-15  Richard Biener  <rguenther@suse.de>
7323         PR rtl-optimization/59802
7324         * lcm.c (compute_available): Use inverted postorder to seed
7325         the initial worklist.
7327 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7329         PR target/59803
7330         * config/s390/s390.c (s390_preferred_reload_class): Don't return
7331         ADDR_REGS for invalid symrefs in non-PIC code.
7333 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
7335         PR other/58712
7336         * builtins.c (determine_block_size): Initialize *probable_max_size
7337         even if len_rtx is CONST_INT.
7339 2014-01-14  Andrew Pinski  <apinski@cavium.com>
7341         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
7342         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
7343         (cortexa53_tunings): Likewise.
7344         (aarch64_sched_issue_rate): New function.
7345         (TARGET_SCHED_ISSUE_RATE): Define.
7347 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7349         * ira-costs.c (find_costs_and_classes): Add missed
7350         ira_init_register_move_cost_if_necessary.
7352 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7354         PR target/59787
7355         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
7357 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
7359         PR target/59794
7360         * config/i386/i386.c (type_natural_mode): Add a bool parameter
7361         to indicate if type is used for function return value.  Warn ABI
7362         change if the vector mode isn't available for function return value.
7363         (ix86_function_arg_advance): Pass false to type_natural_mode.
7364         (ix86_function_arg): Likewise.
7365         (ix86_gimplify_va_arg): Likewise.
7366         (function_arg_32): Don't warn ABI change.
7367         (ix86_function_value): Pass true to type_natural_mode.
7368         (ix86_return_in_memory): Likewise.
7369         (ix86_struct_value_rtx): Removed.
7370         (TARGET_STRUCT_VALUE_RTX): Likewise.
7372 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7374         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
7375         converting a conditional jump into a conditional return.
7377 2014-01-14  Richard Biener  <rguenther@suse.de>
7379         PR tree-optimization/58921
7380         PR tree-optimization/59006
7381         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
7382         hoisting invariant stmts.
7383         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
7384         invariant loads on the preheader edge if possible.
7386 2014-01-14  Joey Ye  <joey.ye@arm.com>
7388         * doc/plugin.texi (Building GCC plugins): Update to C++.
7390 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
7392         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
7393         (_mm_rcp28_round_ss): Ditto.
7394         (_mm_rsqrt28_round_sd): Ditto.
7395         (_mm_rsqrt28_round_ss): Ditto.
7396         (_mm_rcp28_sd): Ditto.
7397         (_mm_rcp28_ss): Ditto.
7398         (_mm_rsqrt28_sd): Ditto.
7399         (_mm_rsqrt28_ss): Ditto.
7400         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
7401         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
7402         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
7403         (IX86_BUILTIN_RCP28SD): Ditto.
7404         (IX86_BUILTIN_RCP28SS): Ditto.
7405         (IX86_BUILTIN_RSQRT28SD): Ditto.
7406         (IX86_BUILTIN_RSQRT28SS): Ditto.
7407         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
7408         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
7409         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
7410         (ix86_expand_special_args_builtin): Expand new FTYPE.
7411         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
7412         (srcp14<mode>): Make insn unary.
7413         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
7414         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
7415         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
7416         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
7417         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
7418         Fix rounding: make it SAE only.
7419         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
7420         Ditto.
7421         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
7422         Ditto.
7423         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
7424         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
7425         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
7426         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
7427         (round_saeonly_mask_scalar_operand4): Ditto.
7428         (round_saeonly_mask_scalar_op3): Ditto.
7429         (round_saeonly_mask_scalar_op4): Ditto.
7431 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7433         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7434         Implement -maltivec=be for vec_insert and vec_extract.
7436 2014-01-10  DJ Delorie  <dj@redhat.com>
7438         * config/msp430/msp430.md (call_internal): Don't allow memory
7439         references with SP as the base register.
7440         (call_value_internal): Likewise.
7441         * config/msp430/constraints.md (Yc): New.  For memory references
7442         that don't use SP as a base register.
7444         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
7445         "an integer without a # prefix"
7446         * config/msp430/msp430.md (epilogue_helper): Use it.
7448 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
7450         PR target/59617
7451         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
7452         AVX512F gather builtins.
7453         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
7454         on gather decls with INTEGER_TYPE masktype.
7455         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
7456         directly into the builtin rather than hoisting it before loop.
7458         PR tree-optimization/59387
7459         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
7460         (scev_const_prop): If folded_casts and type has undefined overflow,
7461         use force_gimple_operand instead of force_gimple_operand_gsi and
7462         for each added stmt if it is assign with
7463         arith_code_with_undefined_signed_overflow, call
7464         rewrite_to_defined_overflow.
7465         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
7466         gimple-fold.h instead.
7467         (arith_code_with_undefined_signed_overflow,
7468         rewrite_to_defined_overflow): Moved to ...
7469         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
7470         rewrite_to_defined_overflow): ... here.  No longer static.
7471         Include gimplify-me.h.
7472         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
7473         rewrite_to_defined_overflow): New prototypes.
7475 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7477         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
7479 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
7481         * builtins.c (get_object_alignment_2): Minor tweak.
7482         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
7484 2014-01-13  Christian Bruel  <christian.bruel@st.com>
7486         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
7487         optimized non constant lengths.
7489 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
7491         PR libgomp/59194
7492         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
7493         load as __atomic_load_N if possible.
7495 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
7497         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
7498         target parameter.
7499         (rs6000_expand_builtin): Adjust call.
7501 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
7503         PR target/58115
7504         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
7505         * config/rs6000/rs6000.c: Include target-globals.h.
7506         (rs6000_set_current_function): Instead of doing target_reinit
7507         unconditionally, use save_target_globals_default_opts and
7508         restore_target_globals.
7510         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
7511         FPSCR.
7512         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
7513         (rs6000_expand_builtin): Handle mffs and mtfsf.
7514         (rs6000_init_builtins): Define mffs and mtfsf.
7515         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
7516         (rs6000_mffs): New pattern.
7517         (rs6000_mtfsf): New pattern.
7519 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
7521         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
7522         Start narrowing with START.  Apply candidate-use pair
7523         and check overall cost in narrowing.
7524         (iv_ca_prune): Pass new argument.
7526 2014-01-10  Jeff Law  <law@redhat.com>
7528         PR middle-end/59743
7529         * ree.c (combine_reaching_defs): Ensure the defining statement
7530         occurs before the extension when optimizing extensions with
7531         different source and destination hard registers.
7533 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
7535         PR ipa/58585
7536         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
7537         vtables into the type inheritance graph.
7539 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7541         PR rtl-optimization/59754
7542         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
7543         modes in the REGNO != REGNO case.
7545 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7547         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
7549 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7551         PR tree-optimization/59745
7552         * tree-predcom.c (tree_predictive_commoning_loop): Call
7553         free_affine_expand_cache if giving up because components is NULL.
7555         * target-globals.c (save_target_globals): Allocate < 4KB structs using
7556         GC in payload of target_globals struct instead of allocating them on
7557         the heap and the larger structs separately using GC.
7558         * target-globals.h (struct target_globals): Make regs, hard_regs,
7559         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
7560         of GTY((skip)) and change type to void *.
7561         (reset_target_globals): Cast loads from those fields to corresponding
7562         types.
7564 2014-01-10  Steve Ellcey  <sellcey@mips.com>
7566         PR plugins/59335
7567         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
7568         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
7569         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
7571 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
7573         PR target/59744
7574         * aarch64-modes.def (CC_Zmode): New flags mode.
7575         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
7576         represents an equality.
7577         (aarch64_get_condition_code): Handle CC_Zmode.
7578         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
7580 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7582         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
7583         extraction in good case.
7585 2014-01-10  Richard Biener  <rguenther@suse.de>
7587         PR tree-optimization/59374
7588         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
7589         checking after SLP discovery.  Mark stmts not participating
7590         in any SLP instance properly.
7592 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7594         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
7595         when handling a SET rtx.
7597 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7599         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
7600         (cortex-a57): Likewise.
7601         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
7603 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7605         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
7606         non-iwmmxt builtins.
7608 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
7610         PR ipa/58252
7611         PR ipa/59226
7612         * ipa-devirt.c record_target_from_binfo): Take as argument
7613         stack of binfos and lookup matching one for virtual inheritance.
7614         (possible_polymorphic_call_targets_1): Update.
7616 2014-01-10  Huacai Chen  <chenhc@lemote.com>
7618         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
7619         kernel strings for Loongson-2E/2F/3A.
7621 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7623         PR middle-end/59670
7624         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
7625         is_gimple_call before calling gimple_call_internal_p.
7627 2014-01-09  Steve Ellcey  <sellcey@mips.com>
7629         * Makefile.in (TREE_FLOW_H): Remove.
7630         (TREE_SSA_H): Add file names from tree-flow.h.
7631         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
7632         * tree.h: Remove tree-flow.h reference.
7633         * hash-table.h: Remove tree-flow.h reference.
7634         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
7635         reference with tree-ssa-loop.h.
7637 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7639         * doc/invoke.texi: Add -maltivec={be,le} options, and document
7640         default element-order behavior for -maltivec.
7641         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
7642         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
7643         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
7644         when targeting big endian, at least for now.
7645         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
7647 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
7649         PR middle-end/47735
7650         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
7651         var satisfies use_register_for_decl, just take into account type
7652         alignment, rather than decl alignment.
7654         PR tree-optimization/59622
7655         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
7656         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
7657         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
7658         Don't devirtualize for inplace at all.  For targets.length () == 1,
7659         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
7661 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
7663         * config/i386/i386.md (cpu): Remove the unused btver1.
7665 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
7667         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
7669 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
7671         PR target/58115
7672         * tree-core.h (struct target_globals): New forward declaration.
7673         (struct tree_target_option): Add globals field.
7674         * tree.h (TREE_TARGET_GLOBALS): Define.
7675         (prepare_target_option_nodes_for_pch): New prototype.
7676         * target-globals.h (struct target_globals): Define even if
7677         !SWITCHABLE_TARGET.
7678         * tree.c (prepare_target_option_node_for_pch,
7679         prepare_target_option_nodes_for_pch): New functions.
7680         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
7681         * config/i386/i386.c: Include target-globals.h.
7682         (ix86_set_current_function): Instead of doing target_reinit
7683         unconditionally, use save_target_globals_default_opts and
7684         restore_target_globals.
7686 2014-01-09  Richard Biener  <rguenther@suse.de>
7688         PR tree-optimization/59715
7689         * tree-cfg.h (split_critical_edges): Declare.
7690         * tree-cfg.c (split_critical_edges): Export.
7691         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
7693 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
7695         * cfgexpand.c (expand_stack_vars): Optionally disable
7696         asan stack protection.
7697         (expand_used_vars): Likewise.
7698         (partition_stack_vars): Likewise.
7699         * asan.c (asan_emit_stack_protection): Optionally disable
7700         after return stack usage.
7701         (instrument_derefs): Optionally disable memory access instrumentation.
7702         (instrument_builtin_call): Likewise.
7703         (instrument_strlen_call): Likewise.
7704         (asan_protect_global): Optionally disable global variables protection.
7705         * doc/invoke.texi: Added doc for new options.
7706         * params.def: Added new options.
7707         * params.h: Likewise.
7709 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
7711         PR rtl-optimization/59724
7712         * ifcvt.c (cond_exec_process_if_block): Don't call
7713         flow_find_head_matching_sequence with 0 longest_match.
7714         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
7715         non-active insns if !stop_after.
7716         (try_head_merge_bb): Revert 2014-01-07 changes.
7718 2014-01-08  Jeff Law  <law@redhat.com>
7720         * ree.c (get_sub_rtx): New function, extracted from...
7721         (merge_def_and_ext): Here.
7722         (combine_reaching_defs): Use get_sub_rtx.
7724 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
7726         * cgraph.h (varpool_variable_node): Do not choke on null node.
7728 2014-01-08  Catherine Moore  <clm@codesourcery.com>
7730         * config/mips/mips.md (simple_return): Attempt to use JRC
7731         for microMIPS.
7732         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
7734 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
7736         PR rtl-optimization/59137
7737         * reorg.c (steal_delay_list_from_target): Call update_block for
7738         elided insns.
7739         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
7741 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7743         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
7744         two duplicate entries.
7746 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
7748         Revert:
7749         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
7751         * config/mips/mips.c (mips_truncated_op_cost): New function.
7752         (mips_rtx_costs): Adjust test for BADDU.
7753         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
7755         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
7757         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
7758         (*baddu_si): ...this new pattern.
7760 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
7762         PR ipa/59722
7763         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
7765 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7767         PR middle-end/57748
7768         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
7769         inner_reference_p.
7770         (expand_expr, expand_normal): Adjust.
7771         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
7772         inner_reference_p. Use inner_reference_p to expand inner references.
7773         (store_expr): Adjust.
7774         * cfgexpand.c (expand_call_stmt): Adjust.
7776 2014-01-08  Rong Xu  <xur@google.com>
7778         * gcov-io.c (gcov_var): Move from gcov-io.h.
7779         (gcov_position): Ditto.
7780         (gcov_is_error): Ditto.
7781         (gcov_rewrite): Ditto.
7782         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
7783         only part to libgcc/libgcov.h.
7785 2014-01-08  Marek Polacek  <polacek@redhat.com>
7787         PR middle-end/59669
7788         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
7790 2014-01-08  Marek Polacek  <polacek@redhat.com>
7792         PR sanitizer/59667
7793         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
7795 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
7797         PR rtl-optimization/59649
7798         * stor-layout.c (get_mode_bounds): For BImode return
7799         0 and STORE_FLAG_VALUE.
7801 2014-01-08  Richard Biener  <rguenther@suse.de>
7803         PR middle-end/59630
7804         * gimple.h (is_gimple_builtin_call): Remove.
7805         (gimple_builtin_call_types_compatible_p): New.
7806         (gimple_call_builtin_p): New overload.
7807         * gimple.c (is_gimple_builtin_call): Remove.
7808         (validate_call): Rename to ...
7809         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
7810         check return types.
7811         (validate_type): New static function.
7812         (gimple_call_builtin_p): New overload and adjust.
7813         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
7814         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
7815         (gimple_fold_stmt_to_constant_1): Likewise.
7816         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
7818 2014-01-08  Richard Biener  <rguenther@suse.de>
7820         PR middle-end/59471
7821         * gimplify.c (gimplify_expr): Gimplify register-register type
7822         VIEW_CONVERT_EXPRs to separate stmts.
7824 2014-01-07  Jeff Law  <law@redhat.com>
7826         PR middle-end/53623
7827         * ree.c (combine_set_extension): Handle case where source
7828         and destination registers in an extension insn are different.
7829         (combine_reaching_defs): Allow source and destination registers
7830         in extension to be different under limited circumstances.
7831         (add_removable_extension): Remove restriction that the
7832         source and destination registers in the extension are the same.
7833         (find_and_remove_re): Emit a copy from the extension's
7834         destination to its source after the defining insn if
7835         the source and destination registers are different.
7837         PR middle-end/59285
7838         * ifcvt.c (merge_if_block): If we are merging a block with more than
7839         one successor with a block with no successors, remove any BARRIER
7840         after the second block.
7842 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
7844         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
7846 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
7848         PR target/59652
7849         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
7850         for 14-bit register offsets when INT14_OK_STRICT is false.
7852 2014-01-07  Roland Stigge  <stigge@antcom.de>
7853             Michael Meissner  <meissner@linux.vnet.ibm.com>
7855         PR 57386/target
7856         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
7857         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
7859 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
7861         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
7862         -mcpu.
7864 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
7866         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
7867         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
7868         rtx is const0_rtx or not.
7870 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
7872         PR target/58115
7873         * target-globals.c (save_target_globals): Remove this_fn_optab
7874         handling.
7875         * toplev.c: Include optabs.h.
7876         (target_reinit): Temporarily restore the global options if another
7877         set of options are in force.
7879 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
7881         PR rtl-optimization/58668
7882         * cfgcleanup.c (flow_find_cross_jump): Don't count
7883         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
7884         to determine what is counted.
7885         (flow_find_head_matching_sequence): Use active_insn_p to determine
7886         what is counted.
7887         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
7888         counting change.
7889         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
7890         determine what is counted.
7892         PR tree-optimization/59643
7893         * tree-predcom.c (split_data_refs_to_components): If one dr is
7894         read and one write, determine_offset fails and the write isn't
7895         in the bad component, just put the read into the bad component.
7897 2014-01-07  Mike Stump  <mikestump@comcast.net>
7898             Jakub Jelinek  <jakub@redhat.com>
7900         PR pch/59436
7901         * tree-core.h (struct tree_optimization_option): Change optabs
7902         type from unsigned char * to void *.
7903         * optabs.c (init_tree_optimization_optabs): Adjust
7904         TREE_OPTIMIZATION_OPTABS initialization.
7906 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
7908         PR target/59644
7909         * config/i386/i386.h (struct machine_function): Add
7910         no_drap_save_restore field.
7911         * config/i386/i386.c (ix86_save_reg): Use
7912         !cfun->machine->no_drap_save_restore instead of
7913         crtl->stack_realign_needed.
7914         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
7915         this function clears frame_pointer_needed.  Set
7916         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
7917         and DRAP reg is needed.
7919 2014-01-06  Marek Polacek  <polacek@redhat.com>
7921         PR c/57773
7922         * doc/implement-c.texi: Mention that other integer types are
7923         permitted as bit-field types in strictly conforming mode.
7925 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
7927         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
7928         is newly allocated.
7930 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
7932         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
7934 2014-01-06  Martin Jambor  <mjambor@suse.cz>
7936         PR ipa/59008
7937         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
7938         to int.
7939         * ipa-prop.c (ipa_print_node_params): Fix indentation.
7941 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
7943         PR debug/59350
7944         PR debug/59510
7945         * var-tracking.c (add_stores): Preserve the value of the source even if
7946         we don't record the store.
7948 2014-01-06  Terry Guo  <terry.guo@arm.com>
7950         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
7952 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
7954         PR bootstrap/59541
7955         * config/darwin.c (darwin_function_section): Adjust return values to
7956         correspond to optimisation changes made in r206070.
7958 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
7960         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
7961         from prefetch_block tune setting.
7962         (nocona_cost): Correct size of prefetch block to 64.
7964 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
7966         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
7967         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
7968         used to save the static chain register in the computation of the offset
7969         from which the FP registers need to be restored.
7971 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
7973         PR tree-optimization/59519
7974         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
7975         ICE if get_current_def (current_new_name) is already non-NULL, as long
7976         as it is a phi result of some other phi in *new_exit_bb that has
7977         the same argument.
7979         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
7980         or vmovdqu* for misaligned_operand.
7981         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
7982         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
7983         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
7984         aligned_mem for AVX512F masked aligned load and store builtins and for
7985         non-temporal moves.
7987 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
7989         PR tree-optimization/59651
7990         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
7991         Address range for negative step should be added by TYPE_SIZE_UNIT.
7993 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
7995         * config/m68k/m68k.c (handle_move_double): Handle pushes with
7996         overlapping registers also for registers other than the stack pointer.
7998 2014-01-03  Marek Polacek  <polacek@redhat.com>
8000         PR other/59661
8001         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
8002         __builtin_FILE.
8004 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
8006         PR target/59625
8007         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
8008         asm goto as jump.
8010         * config/i386/i386.md (MODE_SIZE): New mode attribute.
8011         (push splitter): Use <P:MODE_SIZE> instead of
8012         GET_MODE_SIZE (<P:MODE>mode).
8013         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
8014         (mov -1, reg peephole2): Likewise.
8015         * config/i386/sse.md (*mov<mode>_internal,
8016         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
8017         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
8018         *<code><mode>3, *andnot<mode>3<mask_name>,
8019         <mask_codefor><code><mode>3<mask_name>): Likewise.
8020         * config/i386/subst.md (mask_mode512bit_condition,
8021         sd_mask_mode512bit_condition): Likewise.
8023 2014-01-02  Xinliang David Li  <davidxl@google.com>
8025         PR tree-optimization/59303
8026         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
8027         (dump_predicates): Better output format.
8028         (pred_equal_p): New function.
8029         (is_neq_relop_p): Ditto.
8030         (is_neq_zero_form_p): Ditto.
8031         (pred_expr_equal_p): Ditto.
8032         (pred_neg_p): Ditto.
8033         (simplify_pred): Ditto.
8034         (simplify_preds_2): Ditto.
8035         (simplify_preds_3): Ditto.
8036         (simplify_preds_4): Ditto.
8037         (simplify_preds): Ditto.
8038         (push_pred): Ditto.
8039         (push_to_worklist): Ditto.
8040         (get_pred_info_from_cmp): Ditto.
8041         (is_degenerated_phi): Ditto.
8042         (normalize_one_pred_1): Ditto.
8043         (normalize_one_pred): Ditto.
8044         (normalize_one_pred_chain): Ditto.
8045         (normalize_preds): Ditto.
8046         (normalize_cond_1): Remove function.
8047         (normalize_cond): Ditto.
8048         (is_gcond_subset_of): Ditto.
8049         (is_subset_of_any): Ditto.
8050         (is_or_set_subset_of): Ditto.
8051         (is_and_set_subset_of): Ditto.
8052         (is_norm_cond_subset_of): Ditto.
8053         (pred_chain_length_cmp): Ditto.
8054         (convert_control_dep_chain_into_preds): Type change.
8055         (find_predicates): Ditto.
8056         (find_def_preds): Ditto.
8057         (destroy_predicates_vecs): Ditto.
8058         (find_matching_predicates_in_rest_chains): Ditto.
8059         (use_pred_not_overlap_with_undef_path_pred): Ditto.
8060         (is_pred_expr_subset): Ditto.
8061         (is_pred_chain_subset_of): Ditto.
8062         (is_included_in): Ditto.
8063         (is_superset_of): Ditto.
8065 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8067         Update copyright years.
8069 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8071         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
8072         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
8073         config/arc/arc.md, config/arc/arc.opt,
8074         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
8075         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
8076         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
8077         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
8078         config/linux-protos.h, config/linux.c, config/winnt-c.c,
8079         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
8080         vtable-verify.c, vtable-verify.h: Use the standard form for the
8081         copyright notice.
8083 2014-01-02  Tobias Burnus  <burnus@net-b.de>
8085         * gcc.c (process_command): Update copyright notice dates.
8086         * gcov-dump.c: Ditto.
8087         * gcov.c: Ditto.
8088         * doc/cpp.texi: Bump @copying's copyright year.
8089         * doc/cppinternals.texi: Ditto.
8090         * doc/gcc.texi: Ditto.
8091         * doc/gccint.texi: Ditto.
8092         * doc/gcov.texi: Ditto.
8093         * doc/install.texi: Ditto.
8094         * doc/invoke.texi: Ditto.
8096 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8098         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
8100 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
8102         * config/i386/sse.md (*mov<mode>_internal): Guard
8103         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
8105         PR rtl-optimization/59647
8106         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
8107         new_rtx into UNSIGNED_FLOAT rtxes.
8109 Copyright (C) 2014 Free Software Foundation, Inc.
8111 Copying and distribution of this file, with or without modification,
8112 are permitted in any medium without royalty provided the copyright
8113 notice and this notice are preserved.