]> git.ipfire.org Git - thirdparty/gcc.git/blame - gcc/ChangeLog
2013-08-06 Martin Jambor <mjambor@suse.cz>
[thirdparty/gcc.git] / gcc / ChangeLog
CommitLineData
1f7747bd 12013-08-06 Martin Jambor <mjambor@suse.cz>
2
3 PR fortran/57987
4 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
5 is not re-finalized. Rename second parameter to no_collect.
6
54dfdbf2 72013-08-06 Martin Jambor <mjambor@suse.cz>
8
9 PR middle-end/58041
10 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
11 MEM_REF has proper alignment information.
12
3d953cb1 132013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
14
15 PR other/12081
16 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
17 class insn_gen_fn.
18 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
19 rtx (*) (rtx, ...) with insn_gen_fn.
20 * genoutput.c (output_insn_data): Cast gen_? function pointers to
21 insn_gen_fn::stored_funcptr. Add initializer braces.
22
bde4aa78 232013-08-05 David Malcolm <dmalcolm@redhat.com>
24
25 Rewrite how instances of passes are cloned to remove assumptions
26 about their sizes (thus allowing pass subclasses to have
27 additional data fields, albeit non-GC-managed ones at this point).
28
29 * passes.c (make_pass_instance): Now that passes have clone
30 methods, rewrite this function to eliminate XNEW and memcpy
31 calls that used hardcoded sizes. Since this function no longer
32 creates pass instances, rename it to...
33 (add_pass_instance): ...this. Document the old way that passes
34 were numbered and flagged, and rework this function to continue
35 using it.
36 (next_pass_1): Add an initial_pass argument for use by
37 add_pass_instance.
38 (position_pass): When adding multiple instances of a pass, use
39 the pass's clone method, rather than relying on the XNEW/memcpy
40 within the former make_pass_instance (now add_pass_instance).
41 (pass_manager::pass_manager): When invoking next_pass_1, also
42 supply the initial instance of the current pass within the
43 pass manager.
44
cbe8bda8 452013-08-05 David Malcolm <dmalcolm@redhat.com>
46
47 This is the automated part of the conversion of passes from C
48 structs to C++ classes.
49
50 Patch autogenerated by refactor_passes.py from
51 https://github.com/davidmalcolm/gcc-refactoring-scripts
52 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
53
54 * asan.c (pass_asan): Convert from a global struct to a subclass of
55 gimple_opt_pass along with...
56 (pass_data_asan): ...new pass_data instance and...
57 (make_pass_asan): ...new function.
58 (pass_asan_O0): Convert from a global struct to a subclass of
59 gimple_opt_pass along with...
60 (pass_data_asan_O0): ...new pass_data instance and...
61 (make_pass_asan_O0): ...new function.
62 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
63 subclass of rtl_opt_pass along with...
64 (pass_data_inc_dec): ...new pass_data instance and...
65 (make_pass_inc_dec): ...new function.
66 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
67 a subclass of rtl_opt_pass along with...
68 (pass_data_reorder_blocks): ...new pass_data instance and...
69 (make_pass_reorder_blocks): ...new function.
70 (pass_duplicate_computed_gotos): Convert from a global struct to a
71 subclass of rtl_opt_pass along with...
72 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
73 (make_pass_duplicate_computed_gotos): ...new function.
74 (pass_partition_blocks): Convert from a global struct to a subclass of
75 rtl_opt_pass along with...
76 (pass_data_partition_blocks): ...new pass_data instance and...
77 (make_pass_partition_blocks): ...new function.
78 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
79 struct to a subclass of rtl_opt_pass along with...
80 (pass_data_branch_target_load_optimize1): ...new pass_data instance
81 and...
82 (make_pass_branch_target_load_optimize1): ...new function.
83 (pass_branch_target_load_optimize2): Convert from a global struct to a
84 subclass of rtl_opt_pass along with...
85 (pass_data_branch_target_load_optimize2): ...new pass_data instance
86 and...
87 (make_pass_branch_target_load_optimize2): ...new function.
88 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
89 of rtl_opt_pass along with...
90 (pass_data_jump): ...new pass_data instance and...
91 (make_pass_jump): ...new function.
92 (pass_jump2): Convert from a global struct to a subclass of
93 rtl_opt_pass along with...
94 (pass_data_jump2): ...new pass_data instance and...
95 (make_pass_jump2): ...new function.
96 * cfgexpand.c (pass_expand): Convert from a global struct to a
97 subclass of rtl_opt_pass along with...
98 (pass_data_expand): ...new pass_data instance and...
99 (make_pass_expand): ...new function.
100 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
101 of rtl_opt_pass along with...
102 (pass_data_free_cfg): ...new pass_data instance and...
103 (make_pass_free_cfg): ...new function.
104 (pass_into_cfg_layout_mode): Convert from a global struct to a
105 subclass of rtl_opt_pass along with...
106 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
107 (make_pass_into_cfg_layout_mode): ...new function.
108 (pass_outof_cfg_layout_mode): Convert from a global struct to a
109 subclass of rtl_opt_pass along with...
110 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
111 (make_pass_outof_cfg_layout_mode): ...new function.
112 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
113 struct to a subclass of gimple_opt_pass along with...
114 (pass_data_build_cgraph_edges): ...new pass_data instance and...
115 (make_pass_build_cgraph_edges): ...new function.
116 (pass_rebuild_cgraph_edges): Convert from a global struct to a
117 subclass of gimple_opt_pass along with...
118 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
119 (make_pass_rebuild_cgraph_edges): ...new function.
120 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
121 subclass of gimple_opt_pass along with...
122 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
123 and...
124 (make_pass_remove_cgraph_callee_edges): ...new function.
125 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
126 struct to a subclass of rtl_opt_pass along with...
127 (pass_data_stack_adjustments): ...new pass_data instance and...
128 (make_pass_stack_adjustments): ...new function.
129 * combine.c (pass_combine): Convert from a global struct to a subclass
130 of rtl_opt_pass along with...
131 (pass_data_combine): ...new pass_data instance and...
132 (make_pass_combine): ...new function.
133 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
134 global struct to a subclass of rtl_opt_pass along with...
135 (pass_data_compare_elim_after_reload): ...new pass_data instance
136 and...
137 (make_pass_compare_elim_after_reload): ...new function.
138 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
139 of rtl_opt_pass along with...
140 (pass_data_rtl_cprop): ...new pass_data instance and...
141 (make_pass_rtl_cprop): ...new function.
142 * cse.c (pass_cse): Convert from a global struct to a subclass of
143 rtl_opt_pass along with...
144 (pass_data_cse): ...new pass_data instance and...
145 (make_pass_cse): ...new function.
146 (pass_cse2): Convert from a global struct to a subclass of
147 rtl_opt_pass along with...
148 (pass_data_cse2): ...new pass_data instance and...
149 (make_pass_cse2): ...new function.
150 (pass_cse_after_global_opts): Convert from a global struct to a
151 subclass of rtl_opt_pass along with...
152 (pass_data_cse_after_global_opts): ...new pass_data instance and...
153 (make_pass_cse_after_global_opts): ...new function.
154 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
155 of rtl_opt_pass along with...
156 (pass_data_ud_rtl_dce): ...new pass_data instance and...
157 (make_pass_ud_rtl_dce): ...new function.
158 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
159 rtl_opt_pass along with...
160 (pass_data_fast_rtl_dce): ...new pass_data instance and...
161 (make_pass_fast_rtl_dce): ...new function.
162 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
163 a subclass of rtl_opt_pass along with...
164 (pass_data_df_initialize_opt): ...new pass_data instance and...
165 (make_pass_df_initialize_opt): ...new function.
166 (pass_df_initialize_no_opt): Convert from a global struct to a
167 subclass of rtl_opt_pass along with...
168 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
169 (make_pass_df_initialize_no_opt): ...new function.
170 (pass_df_finish): Convert from a global struct to a subclass of
171 rtl_opt_pass along with...
172 (pass_data_df_finish): ...new pass_data instance and...
173 (make_pass_df_finish): ...new function.
174 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
175 rtl_opt_pass along with...
176 (pass_data_rtl_dse1): ...new pass_data instance and...
177 (make_pass_rtl_dse1): ...new function.
178 (pass_rtl_dse2): Convert from a global struct to a subclass of
179 rtl_opt_pass along with...
180 (pass_data_rtl_dse2): ...new pass_data instance and...
181 (make_pass_rtl_dse2): ...new function.
182 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
183 subclass of rtl_opt_pass along with...
184 (pass_data_dwarf2_frame): ...new pass_data instance and...
185 (make_pass_dwarf2_frame): ...new function.
186 * except.c (pass_set_nothrow_function_flags): Convert from a global
187 struct to a subclass of rtl_opt_pass along with...
188 (pass_data_set_nothrow_function_flags): ...new pass_data instance
189 and...
190 (make_pass_set_nothrow_function_flags): ...new function.
191 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
192 subclass of rtl_opt_pass along with...
193 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
194 and...
195 (make_pass_convert_to_eh_region_ranges): ...new function.
196 * final.c (pass_compute_alignments): Convert from a global struct to a
197 subclass of rtl_opt_pass along with...
198 (pass_data_compute_alignments): ...new pass_data instance and...
199 (make_pass_compute_alignments): ...new function.
200 (pass_final): Convert from a global struct to a subclass of
201 rtl_opt_pass along with...
202 (pass_data_final): ...new pass_data instance and...
203 (make_pass_final): ...new function.
204 (pass_shorten_branches): Convert from a global struct to a subclass of
205 rtl_opt_pass along with...
206 (pass_data_shorten_branches): ...new pass_data instance and...
207 (make_pass_shorten_branches): ...new function.
208 (pass_clean_state): Convert from a global struct to a subclass of
209 rtl_opt_pass along with...
210 (pass_data_clean_state): ...new pass_data instance and...
211 (make_pass_clean_state): ...new function.
212 * function.c (pass_instantiate_virtual_regs): Convert from a global
213 struct to a subclass of rtl_opt_pass along with...
214 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
215 (make_pass_instantiate_virtual_regs): ...new function.
216 (pass_leaf_regs): Convert from a global struct to a subclass of
217 rtl_opt_pass along with...
218 (pass_data_leaf_regs): ...new pass_data instance and...
219 (make_pass_leaf_regs): ...new function.
220 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
221 subclass of rtl_opt_pass along with...
222 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
223 and...
224 (make_pass_thread_prologue_and_epilogue): ...new function.
225 (pass_match_asm_constraints): Convert from a global struct to a
226 subclass of rtl_opt_pass along with...
227 (pass_data_match_asm_constraints): ...new pass_data instance and...
228 (make_pass_match_asm_constraints): ...new function.
229 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
230 subclass of rtl_opt_pass along with...
231 (pass_data_rtl_fwprop): ...new pass_data instance and...
232 (make_pass_rtl_fwprop): ...new function.
233 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
234 rtl_opt_pass along with...
235 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
236 (make_pass_rtl_fwprop_addr): ...new function.
237 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
238 rtl_opt_pass along with...
239 (pass_data_rtl_pre): ...new pass_data instance and...
240 (make_pass_rtl_pre): ...new function.
241 (pass_rtl_hoist): Convert from a global struct to a subclass of
242 rtl_opt_pass along with...
243 (pass_data_rtl_hoist): ...new pass_data instance and...
244 (make_pass_rtl_hoist): ...new function.
245 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
246 subclass of gimple_opt_pass along with...
247 (pass_data_lower_cf): ...new pass_data instance and...
248 (make_pass_lower_cf): ...new function.
249 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
250 from a global struct to a subclass of gimple_opt_pass along with...
251 (pass_data_strength_reduction): ...new pass_data instance and...
252 (make_pass_strength_reduction): ...new function.
253 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
254 of rtl_opt_pass along with...
255 (pass_data_rtl_ifcvt): ...new pass_data instance and...
256 (make_pass_rtl_ifcvt): ...new function.
257 (pass_if_after_combine): Convert from a global struct to a subclass of
258 rtl_opt_pass along with...
259 (pass_data_if_after_combine): ...new pass_data instance and...
260 (make_pass_if_after_combine): ...new function.
261 (pass_if_after_reload): Convert from a global struct to a subclass of
262 rtl_opt_pass along with...
263 (pass_data_if_after_reload): ...new pass_data instance and...
264 (make_pass_if_after_reload): ...new function.
265 * init-regs.c (pass_initialize_regs): Convert from a global struct to
266 a subclass of rtl_opt_pass along with...
267 (pass_data_initialize_regs): ...new pass_data instance and...
268 (make_pass_initialize_regs): ...new function.
269 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
270 of ipa_opt_pass_d along with...
271 (pass_data_ipa_cp): ...new pass_data instance and...
272 (make_pass_ipa_cp): ...new function.
273 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
274 global struct to a subclass of gimple_opt_pass along with...
275 (pass_data_inline_parameters): ...new pass_data instance and...
276 (make_pass_inline_parameters): ...new function.
277 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
278 subclass of gimple_opt_pass along with...
279 (pass_data_early_inline): ...new pass_data instance and...
280 (make_pass_early_inline): ...new function.
281 (pass_ipa_inline): Convert from a global struct to a subclass of
282 ipa_opt_pass_d along with...
283 (pass_data_ipa_inline): ...new pass_data instance and...
284 (make_pass_ipa_inline): ...new function.
285 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
286 struct to a subclass of gimple_opt_pass along with...
287 (pass_data_local_pure_const): ...new pass_data instance and...
288 (make_pass_local_pure_const): ...new function.
289 (pass_ipa_pure_const): Convert from a global struct to a subclass of
290 ipa_opt_pass_d along with...
291 (pass_data_ipa_pure_const): ...new pass_data instance and...
292 (make_pass_ipa_pure_const): ...new function.
293 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
294 to a subclass of ipa_opt_pass_d along with...
295 (pass_data_ipa_reference): ...new pass_data instance and...
296 (make_pass_ipa_reference): ...new function.
297 * ipa-split.c (pass_split_functions): Convert from a global struct to
298 a subclass of gimple_opt_pass along with...
299 (pass_data_split_functions): ...new pass_data instance and...
300 (make_pass_split_functions): ...new function.
301 (pass_feedback_split_functions): Convert from a global struct to a
302 subclass of gimple_opt_pass along with...
303 (pass_data_feedback_split_functions): ...new pass_data instance and...
304 (make_pass_feedback_split_functions): ...new function.
305 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
306 global struct to a subclass of simple_ipa_opt_pass along with...
307 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
308 instance and...
309 (make_pass_ipa_function_and_variable_visibility): ...new function.
310 (pass_ipa_free_inline_summary): Convert from a global struct to a
311 subclass of simple_ipa_opt_pass along with...
312 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
313 (make_pass_ipa_free_inline_summary): ...new function.
314 (pass_ipa_whole_program_visibility): Convert from a global struct to a
315 subclass of ipa_opt_pass_d along with...
316 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
317 and...
318 (make_pass_ipa_whole_program_visibility): ...new function.
319 (pass_ipa_profile): Convert from a global struct to a subclass of
320 ipa_opt_pass_d along with...
321 (pass_data_ipa_profile): ...new pass_data instance and...
322 (make_pass_ipa_profile): ...new function.
323 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
324 ipa_opt_pass_d along with...
325 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
326 (make_pass_ipa_cdtor_merge): ...new function.
327 * ira.c (pass_ira): Convert from a global struct to a subclass of
328 rtl_opt_pass along with...
329 (pass_data_ira): ...new pass_data instance and...
330 (make_pass_ira): ...new function.
331 (pass_reload): Convert from a global struct to a subclass of
332 rtl_opt_pass along with...
333 (pass_data_reload): ...new pass_data instance and...
334 (make_pass_reload): ...new function.
335 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
336 subclass of rtl_opt_pass along with...
337 (pass_data_cleanup_barriers): ...new pass_data instance and...
338 (make_pass_cleanup_barriers): ...new function.
339 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
340 of rtl_opt_pass along with...
341 (pass_data_loop2): ...new pass_data instance and...
342 (make_pass_loop2): ...new function.
343 (pass_rtl_loop_init): Convert from a global struct to a subclass of
344 rtl_opt_pass along with...
345 (pass_data_rtl_loop_init): ...new pass_data instance and...
346 (make_pass_rtl_loop_init): ...new function.
347 (pass_rtl_loop_done): Convert from a global struct to a subclass of
348 rtl_opt_pass along with...
349 (pass_data_rtl_loop_done): ...new pass_data instance and...
350 (make_pass_rtl_loop_done): ...new function.
351 (pass_rtl_move_loop_invariants): Convert from a global struct to a
352 subclass of rtl_opt_pass along with...
353 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
354 (make_pass_rtl_move_loop_invariants): ...new function.
355 (pass_rtl_unswitch): Convert from a global struct to a subclass of
356 rtl_opt_pass along with...
357 (pass_data_rtl_unswitch): ...new pass_data instance and...
358 (make_pass_rtl_unswitch): ...new function.
359 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
360 subclass of rtl_opt_pass along with...
361 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
362 and...
363 (make_pass_rtl_unroll_and_peel_loops): ...new function.
364 (pass_rtl_doloop): Convert from a global struct to a subclass of
365 rtl_opt_pass along with...
366 (pass_data_rtl_doloop): ...new pass_data instance and...
367 (make_pass_rtl_doloop): ...new function.
368 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
369 a subclass of rtl_opt_pass along with...
370 (pass_data_lower_subreg): ...new pass_data instance and...
371 (make_pass_lower_subreg): ...new function.
372 (pass_lower_subreg2): Convert from a global struct to a subclass of
373 rtl_opt_pass along with...
374 (pass_data_lower_subreg2): ...new pass_data instance and...
375 (make_pass_lower_subreg2): ...new function.
376 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
377 struct to a subclass of ipa_opt_pass_d along with...
378 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
379 (make_pass_ipa_lto_gimple_out): ...new function.
380 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
381 of ipa_opt_pass_d along with...
382 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
383 (make_pass_ipa_lto_finish_out): ...new function.
384 * mode-switching.c (pass_mode_switching): Convert from a global struct
385 to a subclass of rtl_opt_pass along with...
386 (pass_data_mode_switching): ...new pass_data instance and...
387 (make_pass_mode_switching): ...new function.
388 * modulo-sched.c (pass_sms): Convert from a global struct to a
389 subclass of rtl_opt_pass along with...
390 (pass_data_sms): ...new pass_data instance and...
391 (make_pass_sms): ...new function.
392 * omp-low.c (pass_expand_omp): Convert from a global struct to a
393 subclass of gimple_opt_pass along with...
394 (pass_data_expand_omp): ...new pass_data instance and...
395 (make_pass_expand_omp): ...new function.
396 (pass_lower_omp): Convert from a global struct to a subclass of
397 gimple_opt_pass along with...
398 (pass_data_lower_omp): ...new pass_data instance and...
399 (make_pass_lower_omp): ...new function.
400 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
401 of gimple_opt_pass along with...
402 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
403 (make_pass_diagnose_omp_blocks): ...new function.
404 * passes.c (pass_early_local_passes): Convert from a global struct to
405 a subclass of simple_ipa_opt_pass along with...
406 (pass_data_early_local_passes): ...new pass_data instance and...
407 (make_pass_early_local_passes): ...new function.
408 (pass_all_early_optimizations): Convert from a global struct to a
409 subclass of gimple_opt_pass along with...
410 (pass_data_all_early_optimizations): ...new pass_data instance and...
411 (make_pass_all_early_optimizations): ...new function.
412 (pass_all_optimizations): Convert from a global struct to a subclass
413 of gimple_opt_pass along with...
414 (pass_data_all_optimizations): ...new pass_data instance and...
415 (make_pass_all_optimizations): ...new function.
416 (pass_all_optimizations_g): Convert from a global struct to a subclass
417 of gimple_opt_pass along with...
418 (pass_data_all_optimizations_g): ...new pass_data instance and...
419 (make_pass_all_optimizations_g): ...new function.
420 (pass_rest_of_compilation): Convert from a global struct to a subclass
421 of rtl_opt_pass along with...
422 (pass_data_rest_of_compilation): ...new pass_data instance and...
423 (make_pass_rest_of_compilation): ...new function.
424 (pass_postreload): Convert from a global struct to a subclass of
425 rtl_opt_pass along with...
426 (pass_data_postreload): ...new pass_data instance and...
427 (make_pass_postreload): ...new function.
428 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
429 subclass of rtl_opt_pass along with...
430 (pass_data_gcse2): ...new pass_data instance and...
431 (make_pass_gcse2): ...new function.
432 * postreload.c (pass_postreload_cse): Convert from a global struct to
433 a subclass of rtl_opt_pass along with...
434 (pass_data_postreload_cse): ...new pass_data instance and...
435 (make_pass_postreload_cse): ...new function.
436 * predict.c (pass_profile): Convert from a global struct to a subclass
437 of gimple_opt_pass along with...
438 (pass_data_profile): ...new pass_data instance and...
439 (make_pass_profile): ...new function.
440 (pass_strip_predict_hints): Convert from a global struct to a subclass
441 of gimple_opt_pass along with...
442 (pass_data_strip_predict_hints): ...new pass_data instance and...
443 (make_pass_strip_predict_hints): ...new function.
444 * recog.c (pass_peephole2): Convert from a global struct to a subclass
445 of rtl_opt_pass along with...
446 (pass_data_peephole2): ...new pass_data instance and...
447 (make_pass_peephole2): ...new function.
448 (pass_split_all_insns): Convert from a global struct to a subclass of
449 rtl_opt_pass along with...
450 (pass_data_split_all_insns): ...new pass_data instance and...
451 (make_pass_split_all_insns): ...new function.
452 (pass_split_after_reload): Convert from a global struct to a subclass
453 of rtl_opt_pass along with...
454 (pass_data_split_after_reload): ...new pass_data instance and...
455 (make_pass_split_after_reload): ...new function.
456 (pass_split_before_regstack): Convert from a global struct to a
457 subclass of rtl_opt_pass along with...
458 (pass_data_split_before_regstack): ...new pass_data instance and...
459 (make_pass_split_before_regstack): ...new function.
460 (pass_split_before_sched2): Convert from a global struct to a subclass
461 of rtl_opt_pass along with...
462 (pass_data_split_before_sched2): ...new pass_data instance and...
463 (make_pass_split_before_sched2): ...new function.
464 (pass_split_for_shorten_branches): Convert from a global struct to a
465 subclass of rtl_opt_pass along with...
466 (pass_data_split_for_shorten_branches): ...new pass_data instance
467 and...
468 (make_pass_split_for_shorten_branches): ...new function.
469 * ree.c (pass_ree): Convert from a global struct to a subclass of
470 rtl_opt_pass along with...
471 (pass_data_ree): ...new pass_data instance and...
472 (make_pass_ree): ...new function.
473 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
474 subclass of rtl_opt_pass along with...
475 (pass_data_stack_regs): ...new pass_data instance and...
476 (make_pass_stack_regs): ...new function.
477 (pass_stack_regs_run): Convert from a global struct to a subclass of
478 rtl_opt_pass along with...
479 (pass_data_stack_regs_run): ...new pass_data instance and...
480 (make_pass_stack_regs_run): ...new function.
481 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
482 subclass of rtl_opt_pass along with...
483 (pass_data_cprop_hardreg): ...new pass_data instance and...
484 (make_pass_cprop_hardreg): ...new function.
485 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
486 subclass of rtl_opt_pass along with...
487 (pass_data_reginfo_init): ...new pass_data instance and...
488 (make_pass_reginfo_init): ...new function.
489 * regmove.c (pass_regmove): Convert from a global struct to a subclass
490 of rtl_opt_pass along with...
491 (pass_data_regmove): ...new pass_data instance and...
492 (make_pass_regmove): ...new function.
493 * regrename.c (pass_regrename): Convert from a global struct to a
494 subclass of rtl_opt_pass along with...
495 (pass_data_regrename): ...new pass_data instance and...
496 (make_pass_regrename): ...new function.
497 * reorg.c (pass_delay_slots): Convert from a global struct to a
498 subclass of rtl_opt_pass along with...
499 (pass_data_delay_slots): ...new pass_data instance and...
500 (make_pass_delay_slots): ...new function.
501 (pass_machine_reorg): Convert from a global struct to a subclass of
502 rtl_opt_pass along with...
503 (pass_data_machine_reorg): ...new pass_data instance and...
504 (make_pass_machine_reorg): ...new function.
505 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
506 of rtl_opt_pass along with...
507 (pass_data_sched): ...new pass_data instance and...
508 (make_pass_sched): ...new function.
509 (pass_sched2): Convert from a global struct to a subclass of
510 rtl_opt_pass along with...
511 (pass_data_sched2): ...new pass_data instance and...
512 (make_pass_sched2): ...new function.
513 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
514 to a subclass of rtl_opt_pass along with...
515 (pass_data_stack_ptr_mod): ...new pass_data instance and...
516 (make_pass_stack_ptr_mod): ...new function.
517 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
518 to a subclass of rtl_opt_pass along with...
519 (pass_data_rtl_store_motion): ...new pass_data instance and...
520 (make_pass_rtl_store_motion): ...new function.
521 * tracer.c (pass_tracer): Convert from a global struct to a subclass
522 of gimple_opt_pass along with...
523 (pass_data_tracer): ...new pass_data instance and...
524 (make_pass_tracer): ...new function.
525 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
526 to a subclass of gimple_opt_pass along with...
527 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
528 (make_pass_diagnose_tm_blocks): ...new function.
529 (pass_lower_tm): Convert from a global struct to a subclass of
530 gimple_opt_pass along with...
531 (pass_data_lower_tm): ...new pass_data instance and...
532 (make_pass_lower_tm): ...new function.
533 (pass_tm_init): Convert from a global struct to a subclass of
534 gimple_opt_pass along with...
535 (pass_data_tm_init): ...new pass_data instance and...
536 (make_pass_tm_init): ...new function.
537 (pass_tm_mark): Convert from a global struct to a subclass of
538 gimple_opt_pass along with...
539 (pass_data_tm_mark): ...new pass_data instance and...
540 (make_pass_tm_mark): ...new function.
541 (pass_tm_edges): Convert from a global struct to a subclass of
542 gimple_opt_pass along with...
543 (pass_data_tm_edges): ...new pass_data instance and...
544 (make_pass_tm_edges): ...new function.
545 (pass_tm_memopt): Convert from a global struct to a subclass of
546 gimple_opt_pass along with...
547 (pass_data_tm_memopt): ...new pass_data instance and...
548 (make_pass_tm_memopt): ...new function.
549 (pass_ipa_tm): Convert from a global struct to a subclass of
550 simple_ipa_opt_pass along with...
551 (pass_data_ipa_tm): ...new pass_data instance and...
552 (make_pass_ipa_tm): ...new function.
553 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
554 subclass of gimple_opt_pass along with...
555 (pass_data_call_cdce): ...new pass_data instance and...
556 (make_pass_call_cdce): ...new function.
557 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
558 subclass of gimple_opt_pass along with...
559 (pass_data_build_cfg): ...new pass_data instance and...
560 (make_pass_build_cfg): ...new function.
561 (pass_split_crit_edges): Convert from a global struct to a subclass of
562 gimple_opt_pass along with...
563 (pass_data_split_crit_edges): ...new pass_data instance and...
564 (make_pass_split_crit_edges): ...new function.
565 (pass_warn_function_return): Convert from a global struct to a
566 subclass of gimple_opt_pass along with...
567 (pass_data_warn_function_return): ...new pass_data instance and...
568 (make_pass_warn_function_return): ...new function.
569 (pass_warn_function_noreturn): Convert from a global struct to a
570 subclass of gimple_opt_pass along with...
571 (pass_data_warn_function_noreturn): ...new pass_data instance and...
572 (make_pass_warn_function_noreturn): ...new function.
573 (pass_warn_unused_result): Convert from a global struct to a subclass
574 of gimple_opt_pass along with...
575 (pass_data_warn_unused_result): ...new pass_data instance and...
576 (make_pass_warn_unused_result): ...new function.
577 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
578 a subclass of gimple_opt_pass along with...
579 (pass_data_merge_phi): ...new pass_data instance and...
580 (make_pass_merge_phi): ...new function.
581 * tree-complex.c (pass_lower_complex): Convert from a global struct to
582 a subclass of gimple_opt_pass along with...
583 (pass_data_lower_complex): ...new pass_data instance and...
584 (make_pass_lower_complex): ...new function.
585 (pass_lower_complex_O0): Convert from a global struct to a subclass of
586 gimple_opt_pass along with...
587 (pass_data_lower_complex_O0): ...new pass_data instance and...
588 (make_pass_lower_complex_O0): ...new function.
589 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
590 subclass of gimple_opt_pass along with...
591 (pass_data_lower_eh): ...new pass_data instance and...
592 (make_pass_lower_eh): ...new function.
593 (pass_refactor_eh): Convert from a global struct to a subclass of
594 gimple_opt_pass along with...
595 (pass_data_refactor_eh): ...new pass_data instance and...
596 (make_pass_refactor_eh): ...new function.
597 (pass_lower_resx): Convert from a global struct to a subclass of
598 gimple_opt_pass along with...
599 (pass_data_lower_resx): ...new pass_data instance and...
600 (make_pass_lower_resx): ...new function.
601 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
602 of gimple_opt_pass along with...
603 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
604 (make_pass_lower_eh_dispatch): ...new function.
605 (pass_cleanup_eh): Convert from a global struct to a subclass of
606 gimple_opt_pass along with...
607 (pass_data_cleanup_eh): ...new pass_data instance and...
608 (make_pass_cleanup_eh): ...new function.
609 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
610 to a subclass of simple_ipa_opt_pass along with...
611 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
612 (make_pass_ipa_lower_emutls): ...new function.
613 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
614 a subclass of gimple_opt_pass along with...
615 (pass_data_if_conversion): ...new pass_data instance and...
616 (make_pass_if_conversion): ...new function.
617 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
618 subclass of gimple_opt_pass along with...
619 (pass_data_build_ssa): ...new pass_data instance and...
620 (make_pass_build_ssa): ...new function.
621 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
622 global struct to a subclass of gimple_opt_pass along with...
623 (pass_data_loop_distribution): ...new pass_data instance and...
624 (make_pass_loop_distribution): ...new function.
625 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
626 subclass of gimple_opt_pass along with...
627 (pass_data_mudflap_1): ...new pass_data instance and...
628 (make_pass_mudflap_1): ...new function.
629 (pass_mudflap_2): Convert from a global struct to a subclass of
630 gimple_opt_pass along with...
631 (pass_data_mudflap_2): ...new pass_data instance and...
632 (make_pass_mudflap_2): ...new function.
633 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
634 subclass of gimple_opt_pass along with...
635 (pass_data_mudflap_1): ...new pass_data instance and...
636 (make_pass_mudflap_1): ...new function.
637 (pass_mudflap_2): Convert from a global struct to a subclass of
638 gimple_opt_pass along with...
639 (pass_data_mudflap_2): ...new pass_data instance and...
640 (make_pass_mudflap_2): ...new function.
641 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
642 gimple_opt_pass along with...
643 (pass_data_nrv): ...new pass_data instance and...
644 (make_pass_nrv): ...new function.
645 (pass_return_slot): Convert from a global struct to a subclass of
646 gimple_opt_pass along with...
647 (pass_data_return_slot): ...new pass_data instance and...
648 (make_pass_return_slot): ...new function.
649 * tree-object-size.c (pass_object_sizes): Convert from a global struct
650 to a subclass of gimple_opt_pass along with...
651 (pass_data_object_sizes): ...new pass_data instance and...
652 (make_pass_object_sizes): ...new function.
653 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
654 global struct to a subclass of gimple_opt_pass along with...
655 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
656 and...
657 (make_pass_cleanup_cfg_post_optimizing): ...new function.
658 (pass_fixup_cfg): Convert from a global struct to a subclass of
659 gimple_opt_pass along with...
660 (pass_data_fixup_cfg): ...new pass_data instance and...
661 (make_pass_fixup_cfg): ...new function.
662 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
663 (make_pass_mudflap_1): ...new function.
664 (pass_mudflap_2): Replace declaration with that of...
665 (make_pass_mudflap_2): ...new function.
666 (pass_asan): Replace declaration with that of...
667 (make_pass_asan): ...new function.
668 (pass_asan_O0): Replace declaration with that of...
669 (make_pass_asan_O0): ...new function.
670 (pass_tsan): Replace declaration with that of...
671 (make_pass_tsan): ...new function.
672 (pass_tsan_O0): Replace declaration with that of...
673 (make_pass_tsan_O0): ...new function.
674 (pass_lower_cf): Replace declaration with that of...
675 (make_pass_lower_cf): ...new function.
676 (pass_refactor_eh): Replace declaration with that of...
677 (make_pass_refactor_eh): ...new function.
678 (pass_lower_eh): Replace declaration with that of...
679 (make_pass_lower_eh): ...new function.
680 (pass_lower_eh_dispatch): Replace declaration with that of...
681 (make_pass_lower_eh_dispatch): ...new function.
682 (pass_lower_resx): Replace declaration with that of...
683 (make_pass_lower_resx): ...new function.
684 (pass_build_cfg): Replace declaration with that of...
685 (make_pass_build_cfg): ...new function.
686 (pass_early_tree_profile): Replace declaration with that of...
687 (make_pass_early_tree_profile): ...new function.
688 (pass_cleanup_eh): Replace declaration with that of...
689 (make_pass_cleanup_eh): ...new function.
690 (pass_sra): Replace declaration with that of...
691 (make_pass_sra): ...new function.
692 (pass_sra_early): Replace declaration with that of...
693 (make_pass_sra_early): ...new function.
694 (pass_early_ipa_sra): Replace declaration with that of...
695 (make_pass_early_ipa_sra): ...new function.
696 (pass_tail_recursion): Replace declaration with that of...
697 (make_pass_tail_recursion): ...new function.
698 (pass_tail_calls): Replace declaration with that of...
699 (make_pass_tail_calls): ...new function.
700 (pass_tree_loop): Replace declaration with that of...
701 (make_pass_tree_loop): ...new function.
702 (pass_tree_loop_init): Replace declaration with that of...
703 (make_pass_tree_loop_init): ...new function.
704 (pass_lim): Replace declaration with that of...
705 (make_pass_lim): ...new function.
706 (pass_tree_unswitch): Replace declaration with that of...
707 (make_pass_tree_unswitch): ...new function.
708 (pass_predcom): Replace declaration with that of...
709 (make_pass_predcom): ...new function.
710 (pass_iv_canon): Replace declaration with that of...
711 (make_pass_iv_canon): ...new function.
712 (pass_scev_cprop): Replace declaration with that of...
713 (make_pass_scev_cprop): ...new function.
714 (pass_empty_loop): Replace declaration with that of...
715 (make_pass_empty_loop): ...new function.
716 (pass_record_bounds): Replace declaration with that of...
717 (make_pass_record_bounds): ...new function.
718 (pass_graphite): Replace declaration with that of...
719 (make_pass_graphite): ...new function.
720 (pass_graphite_transforms): Replace declaration with that of...
721 (make_pass_graphite_transforms): ...new function.
722 (pass_if_conversion): Replace declaration with that of...
723 (make_pass_if_conversion): ...new function.
724 (pass_loop_distribution): Replace declaration with that of...
725 (make_pass_loop_distribution): ...new function.
726 (pass_vectorize): Replace declaration with that of...
727 (make_pass_vectorize): ...new function.
728 (pass_slp_vectorize): Replace declaration with that of...
729 (make_pass_slp_vectorize): ...new function.
730 (pass_complete_unroll): Replace declaration with that of...
731 (make_pass_complete_unroll): ...new function.
732 (pass_complete_unrolli): Replace declaration with that of...
733 (make_pass_complete_unrolli): ...new function.
734 (pass_parallelize_loops): Replace declaration with that of...
735 (make_pass_parallelize_loops): ...new function.
736 (pass_loop_prefetch): Replace declaration with that of...
737 (make_pass_loop_prefetch): ...new function.
738 (pass_iv_optimize): Replace declaration with that of...
739 (make_pass_iv_optimize): ...new function.
740 (pass_tree_loop_done): Replace declaration with that of...
741 (make_pass_tree_loop_done): ...new function.
742 (pass_ch): Replace declaration with that of...
743 (make_pass_ch): ...new function.
744 (pass_ccp): Replace declaration with that of...
745 (make_pass_ccp): ...new function.
746 (pass_phi_only_cprop): Replace declaration with that of...
747 (make_pass_phi_only_cprop): ...new function.
748 (pass_build_ssa): Replace declaration with that of...
749 (make_pass_build_ssa): ...new function.
750 (pass_build_alias): Replace declaration with that of...
751 (make_pass_build_alias): ...new function.
752 (pass_build_ealias): Replace declaration with that of...
753 (make_pass_build_ealias): ...new function.
754 (pass_dominator): Replace declaration with that of...
755 (make_pass_dominator): ...new function.
756 (pass_dce): Replace declaration with that of...
757 (make_pass_dce): ...new function.
758 (pass_dce_loop): Replace declaration with that of...
759 (make_pass_dce_loop): ...new function.
760 (pass_cd_dce): Replace declaration with that of...
761 (make_pass_cd_dce): ...new function.
762 (pass_call_cdce): Replace declaration with that of...
763 (make_pass_call_cdce): ...new function.
764 (pass_merge_phi): Replace declaration with that of...
765 (make_pass_merge_phi): ...new function.
766 (pass_split_crit_edges): Replace declaration with that of...
767 (make_pass_split_crit_edges): ...new function.
768 (pass_pre): Replace declaration with that of...
769 (make_pass_pre): ...new function.
770 (pass_profile): Replace declaration with that of...
771 (make_pass_profile): ...new function.
772 (pass_strip_predict_hints): Replace declaration with that of...
773 (make_pass_strip_predict_hints): ...new function.
774 (pass_lower_complex_O0): Replace declaration with that of...
775 (make_pass_lower_complex_O0): ...new function.
776 (pass_lower_complex): Replace declaration with that of...
777 (make_pass_lower_complex): ...new function.
778 (pass_lower_vector): Replace declaration with that of...
779 (make_pass_lower_vector): ...new function.
780 (pass_lower_vector_ssa): Replace declaration with that of...
781 (make_pass_lower_vector_ssa): ...new function.
782 (pass_lower_omp): Replace declaration with that of...
783 (make_pass_lower_omp): ...new function.
784 (pass_diagnose_omp_blocks): Replace declaration with that of...
785 (make_pass_diagnose_omp_blocks): ...new function.
786 (pass_expand_omp): Replace declaration with that of...
787 (make_pass_expand_omp): ...new function.
788 (pass_expand_omp_ssa): Replace declaration with that of...
789 (make_pass_expand_omp_ssa): ...new function.
790 (pass_object_sizes): Replace declaration with that of...
791 (make_pass_object_sizes): ...new function.
792 (pass_strlen): Replace declaration with that of...
793 (make_pass_strlen): ...new function.
794 (pass_fold_builtins): Replace declaration with that of...
795 (make_pass_fold_builtins): ...new function.
796 (pass_stdarg): Replace declaration with that of...
797 (make_pass_stdarg): ...new function.
798 (pass_early_warn_uninitialized): Replace declaration with that of...
799 (make_pass_early_warn_uninitialized): ...new function.
800 (pass_late_warn_uninitialized): Replace declaration with that of...
801 (make_pass_late_warn_uninitialized): ...new function.
802 (pass_cse_reciprocals): Replace declaration with that of...
803 (make_pass_cse_reciprocals): ...new function.
804 (pass_cse_sincos): Replace declaration with that of...
805 (make_pass_cse_sincos): ...new function.
806 (pass_optimize_bswap): Replace declaration with that of...
807 (make_pass_optimize_bswap): ...new function.
808 (pass_optimize_widening_mul): Replace declaration with that of...
809 (make_pass_optimize_widening_mul): ...new function.
810 (pass_warn_function_return): Replace declaration with that of...
811 (make_pass_warn_function_return): ...new function.
812 (pass_warn_function_noreturn): Replace declaration with that of...
813 (make_pass_warn_function_noreturn): ...new function.
814 (pass_cselim): Replace declaration with that of...
815 (make_pass_cselim): ...new function.
816 (pass_phiopt): Replace declaration with that of...
817 (make_pass_phiopt): ...new function.
818 (pass_forwprop): Replace declaration with that of...
819 (make_pass_forwprop): ...new function.
820 (pass_phiprop): Replace declaration with that of...
821 (make_pass_phiprop): ...new function.
822 (pass_tree_ifcombine): Replace declaration with that of...
823 (make_pass_tree_ifcombine): ...new function.
824 (pass_dse): Replace declaration with that of...
825 (make_pass_dse): ...new function.
826 (pass_nrv): Replace declaration with that of...
827 (make_pass_nrv): ...new function.
828 (pass_rename_ssa_copies): Replace declaration with that of...
829 (make_pass_rename_ssa_copies): ...new function.
830 (pass_sink_code): Replace declaration with that of...
831 (make_pass_sink_code): ...new function.
832 (pass_fre): Replace declaration with that of...
833 (make_pass_fre): ...new function.
834 (pass_check_data_deps): Replace declaration with that of...
835 (make_pass_check_data_deps): ...new function.
836 (pass_copy_prop): Replace declaration with that of...
837 (make_pass_copy_prop): ...new function.
838 (pass_vrp): Replace declaration with that of...
839 (make_pass_vrp): ...new function.
840 (pass_uncprop): Replace declaration with that of...
841 (make_pass_uncprop): ...new function.
842 (pass_return_slot): Replace declaration with that of...
843 (make_pass_return_slot): ...new function.
844 (pass_reassoc): Replace declaration with that of...
845 (make_pass_reassoc): ...new function.
846 (pass_rebuild_cgraph_edges): Replace declaration with that of...
847 (make_pass_rebuild_cgraph_edges): ...new function.
848 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
849 (make_pass_remove_cgraph_callee_edges): ...new function.
850 (pass_build_cgraph_edges): Replace declaration with that of...
851 (make_pass_build_cgraph_edges): ...new function.
852 (pass_local_pure_const): Replace declaration with that of...
853 (make_pass_local_pure_const): ...new function.
854 (pass_tracer): Replace declaration with that of...
855 (make_pass_tracer): ...new function.
856 (pass_warn_unused_result): Replace declaration with that of...
857 (make_pass_warn_unused_result): ...new function.
858 (pass_diagnose_tm_blocks): Replace declaration with that of...
859 (make_pass_diagnose_tm_blocks): ...new function.
860 (pass_lower_tm): Replace declaration with that of...
861 (make_pass_lower_tm): ...new function.
862 (pass_tm_init): Replace declaration with that of...
863 (make_pass_tm_init): ...new function.
864 (pass_tm_mark): Replace declaration with that of...
865 (make_pass_tm_mark): ...new function.
866 (pass_tm_memopt): Replace declaration with that of...
867 (make_pass_tm_memopt): ...new function.
868 (pass_tm_edges): Replace declaration with that of...
869 (make_pass_tm_edges): ...new function.
870 (pass_split_functions): Replace declaration with that of...
871 (make_pass_split_functions): ...new function.
872 (pass_feedback_split_functions): Replace declaration with that of...
873 (make_pass_feedback_split_functions): ...new function.
874 (pass_strength_reduction): Replace declaration with that of...
875 (make_pass_strength_reduction): ...new function.
876 (pass_ipa_lower_emutls): Replace declaration with that of...
877 (make_pass_ipa_lower_emutls): ...new function.
878 (pass_ipa_function_and_variable_visibility): Replace declaration with
879 that of...
880 (make_pass_ipa_function_and_variable_visibility): ...new function.
881 (pass_ipa_tree_profile): Replace declaration with that of...
882 (make_pass_ipa_tree_profile): ...new function.
883 (pass_early_local_passes): Replace declaration with that of...
884 (make_pass_early_local_passes): ...new function.
885 (pass_ipa_whole_program_visibility): Replace declaration with that
886 of...
887 (make_pass_ipa_whole_program_visibility): ...new function.
888 (pass_ipa_lto_gimple_out): Replace declaration with that of...
889 (make_pass_ipa_lto_gimple_out): ...new function.
890 (pass_ipa_increase_alignment): Replace declaration with that of...
891 (make_pass_ipa_increase_alignment): ...new function.
892 (pass_ipa_inline): Replace declaration with that of...
893 (make_pass_ipa_inline): ...new function.
894 (pass_ipa_free_lang_data): Replace declaration with that of...
895 (make_pass_ipa_free_lang_data): ...new function.
896 (pass_ipa_free_inline_summary): Replace declaration with that of...
897 (make_pass_ipa_free_inline_summary): ...new function.
898 (pass_ipa_cp): Replace declaration with that of...
899 (make_pass_ipa_cp): ...new function.
900 (pass_ipa_reference): Replace declaration with that of...
901 (make_pass_ipa_reference): ...new function.
902 (pass_ipa_pure_const): Replace declaration with that of...
903 (make_pass_ipa_pure_const): ...new function.
904 (pass_ipa_pta): Replace declaration with that of...
905 (make_pass_ipa_pta): ...new function.
906 (pass_ipa_lto_finish_out): Replace declaration with that of...
907 (make_pass_ipa_lto_finish_out): ...new function.
908 (pass_ipa_tm): Replace declaration with that of...
909 (make_pass_ipa_tm): ...new function.
910 (pass_ipa_profile): Replace declaration with that of...
911 (make_pass_ipa_profile): ...new function.
912 (pass_ipa_cdtor_merge): Replace declaration with that of...
913 (make_pass_ipa_cdtor_merge): ...new function.
914 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
915 of...
916 (make_pass_cleanup_cfg_post_optimizing): ...new function.
917 (pass_init_datastructures): Replace declaration with that of...
918 (make_pass_init_datastructures): ...new function.
919 (pass_fixup_cfg): Replace declaration with that of...
920 (make_pass_fixup_cfg): ...new function.
921 (pass_expand): Replace declaration with that of...
922 (make_pass_expand): ...new function.
923 (pass_instantiate_virtual_regs): Replace declaration with that of...
924 (make_pass_instantiate_virtual_regs): ...new function.
925 (pass_rtl_fwprop): Replace declaration with that of...
926 (make_pass_rtl_fwprop): ...new function.
927 (pass_rtl_fwprop_addr): Replace declaration with that of...
928 (make_pass_rtl_fwprop_addr): ...new function.
929 (pass_jump): Replace declaration with that of...
930 (make_pass_jump): ...new function.
931 (pass_jump2): Replace declaration with that of...
932 (make_pass_jump2): ...new function.
933 (pass_lower_subreg): Replace declaration with that of...
934 (make_pass_lower_subreg): ...new function.
935 (pass_cse): Replace declaration with that of...
936 (make_pass_cse): ...new function.
937 (pass_fast_rtl_dce): Replace declaration with that of...
938 (make_pass_fast_rtl_dce): ...new function.
939 (pass_ud_rtl_dce): Replace declaration with that of...
940 (make_pass_ud_rtl_dce): ...new function.
941 (pass_rtl_dce): Replace declaration with that of...
942 (make_pass_rtl_dce): ...new function.
943 (pass_rtl_dse1): Replace declaration with that of...
944 (make_pass_rtl_dse1): ...new function.
945 (pass_rtl_dse2): Replace declaration with that of...
946 (make_pass_rtl_dse2): ...new function.
947 (pass_rtl_dse3): Replace declaration with that of...
948 (make_pass_rtl_dse3): ...new function.
949 (pass_rtl_cprop): Replace declaration with that of...
950 (make_pass_rtl_cprop): ...new function.
951 (pass_rtl_pre): Replace declaration with that of...
952 (make_pass_rtl_pre): ...new function.
953 (pass_rtl_hoist): Replace declaration with that of...
954 (make_pass_rtl_hoist): ...new function.
955 (pass_rtl_store_motion): Replace declaration with that of...
956 (make_pass_rtl_store_motion): ...new function.
957 (pass_cse_after_global_opts): Replace declaration with that of...
958 (make_pass_cse_after_global_opts): ...new function.
959 (pass_rtl_ifcvt): Replace declaration with that of...
960 (make_pass_rtl_ifcvt): ...new function.
961 (pass_into_cfg_layout_mode): Replace declaration with that of...
962 (make_pass_into_cfg_layout_mode): ...new function.
963 (pass_outof_cfg_layout_mode): Replace declaration with that of...
964 (make_pass_outof_cfg_layout_mode): ...new function.
965 (pass_loop2): Replace declaration with that of...
966 (make_pass_loop2): ...new function.
967 (pass_rtl_loop_init): Replace declaration with that of...
968 (make_pass_rtl_loop_init): ...new function.
969 (pass_rtl_move_loop_invariants): Replace declaration with that of...
970 (make_pass_rtl_move_loop_invariants): ...new function.
971 (pass_rtl_unswitch): Replace declaration with that of...
972 (make_pass_rtl_unswitch): ...new function.
973 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
974 (make_pass_rtl_unroll_and_peel_loops): ...new function.
975 (pass_rtl_doloop): Replace declaration with that of...
976 (make_pass_rtl_doloop): ...new function.
977 (pass_rtl_loop_done): Replace declaration with that of...
978 (make_pass_rtl_loop_done): ...new function.
979 (pass_web): Replace declaration with that of...
980 (make_pass_web): ...new function.
981 (pass_cse2): Replace declaration with that of...
982 (make_pass_cse2): ...new function.
983 (pass_df_initialize_opt): Replace declaration with that of...
984 (make_pass_df_initialize_opt): ...new function.
985 (pass_df_initialize_no_opt): Replace declaration with that of...
986 (make_pass_df_initialize_no_opt): ...new function.
987 (pass_reginfo_init): Replace declaration with that of...
988 (make_pass_reginfo_init): ...new function.
989 (pass_inc_dec): Replace declaration with that of...
990 (make_pass_inc_dec): ...new function.
991 (pass_stack_ptr_mod): Replace declaration with that of...
992 (make_pass_stack_ptr_mod): ...new function.
993 (pass_initialize_regs): Replace declaration with that of...
994 (make_pass_initialize_regs): ...new function.
995 (pass_combine): Replace declaration with that of...
996 (make_pass_combine): ...new function.
997 (pass_if_after_combine): Replace declaration with that of...
998 (make_pass_if_after_combine): ...new function.
999 (pass_ree): Replace declaration with that of...
1000 (make_pass_ree): ...new function.
1001 (pass_partition_blocks): Replace declaration with that of...
1002 (make_pass_partition_blocks): ...new function.
1003 (pass_match_asm_constraints): Replace declaration with that of...
1004 (make_pass_match_asm_constraints): ...new function.
1005 (pass_regmove): Replace declaration with that of...
1006 (make_pass_regmove): ...new function.
1007 (pass_split_all_insns): Replace declaration with that of...
1008 (make_pass_split_all_insns): ...new function.
1009 (pass_fast_rtl_byte_dce): Replace declaration with that of...
1010 (make_pass_fast_rtl_byte_dce): ...new function.
1011 (pass_lower_subreg2): Replace declaration with that of...
1012 (make_pass_lower_subreg2): ...new function.
1013 (pass_mode_switching): Replace declaration with that of...
1014 (make_pass_mode_switching): ...new function.
1015 (pass_sms): Replace declaration with that of...
1016 (make_pass_sms): ...new function.
1017 (pass_sched): Replace declaration with that of...
1018 (make_pass_sched): ...new function.
1019 (pass_ira): Replace declaration with that of...
1020 (make_pass_ira): ...new function.
1021 (pass_reload): Replace declaration with that of...
1022 (make_pass_reload): ...new function.
1023 (pass_clean_state): Replace declaration with that of...
1024 (make_pass_clean_state): ...new function.
1025 (pass_branch_prob): Replace declaration with that of...
1026 (make_pass_branch_prob): ...new function.
1027 (pass_value_profile_transformations): Replace declaration with that
1028 of...
1029 (make_pass_value_profile_transformations): ...new function.
1030 (pass_postreload_cse): Replace declaration with that of...
1031 (make_pass_postreload_cse): ...new function.
1032 (pass_gcse2): Replace declaration with that of...
1033 (make_pass_gcse2): ...new function.
1034 (pass_split_after_reload): Replace declaration with that of...
1035 (make_pass_split_after_reload): ...new function.
1036 (pass_branch_target_load_optimize1): Replace declaration with that
1037 of...
1038 (make_pass_branch_target_load_optimize1): ...new function.
1039 (pass_thread_prologue_and_epilogue): Replace declaration with that
1040 of...
1041 (make_pass_thread_prologue_and_epilogue): ...new function.
1042 (pass_stack_adjustments): Replace declaration with that of...
1043 (make_pass_stack_adjustments): ...new function.
1044 (pass_peephole2): Replace declaration with that of...
1045 (make_pass_peephole2): ...new function.
1046 (pass_if_after_reload): Replace declaration with that of...
1047 (make_pass_if_after_reload): ...new function.
1048 (pass_regrename): Replace declaration with that of...
1049 (make_pass_regrename): ...new function.
1050 (pass_cprop_hardreg): Replace declaration with that of...
1051 (make_pass_cprop_hardreg): ...new function.
1052 (pass_reorder_blocks): Replace declaration with that of...
1053 (make_pass_reorder_blocks): ...new function.
1054 (pass_branch_target_load_optimize2): Replace declaration with that
1055 of...
1056 (make_pass_branch_target_load_optimize2): ...new function.
1057 (pass_leaf_regs): Replace declaration with that of...
1058 (make_pass_leaf_regs): ...new function.
1059 (pass_split_before_sched2): Replace declaration with that of...
1060 (make_pass_split_before_sched2): ...new function.
1061 (pass_compare_elim_after_reload): Replace declaration with that of...
1062 (make_pass_compare_elim_after_reload): ...new function.
1063 (pass_sched2): Replace declaration with that of...
1064 (make_pass_sched2): ...new function.
1065 (pass_stack_regs): Replace declaration with that of...
1066 (make_pass_stack_regs): ...new function.
1067 (pass_stack_regs_run): Replace declaration with that of...
1068 (make_pass_stack_regs_run): ...new function.
1069 (pass_df_finish): Replace declaration with that of...
1070 (make_pass_df_finish): ...new function.
1071 (pass_compute_alignments): Replace declaration with that of...
1072 (make_pass_compute_alignments): ...new function.
1073 (pass_duplicate_computed_gotos): Replace declaration with that of...
1074 (make_pass_duplicate_computed_gotos): ...new function.
1075 (pass_variable_tracking): Replace declaration with that of...
1076 (make_pass_variable_tracking): ...new function.
1077 (pass_free_cfg): Replace declaration with that of...
1078 (make_pass_free_cfg): ...new function.
1079 (pass_machine_reorg): Replace declaration with that of...
1080 (make_pass_machine_reorg): ...new function.
1081 (pass_cleanup_barriers): Replace declaration with that of...
1082 (make_pass_cleanup_barriers): ...new function.
1083 (pass_delay_slots): Replace declaration with that of...
1084 (make_pass_delay_slots): ...new function.
1085 (pass_split_for_shorten_branches): Replace declaration with that of...
1086 (make_pass_split_for_shorten_branches): ...new function.
1087 (pass_split_before_regstack): Replace declaration with that of...
1088 (make_pass_split_before_regstack): ...new function.
1089 (pass_convert_to_eh_region_ranges): Replace declaration with that
1090 of...
1091 (make_pass_convert_to_eh_region_ranges): ...new function.
1092 (pass_shorten_branches): Replace declaration with that of...
1093 (make_pass_shorten_branches): ...new function.
1094 (pass_set_nothrow_function_flags): Replace declaration with that of...
1095 (make_pass_set_nothrow_function_flags): ...new function.
1096 (pass_dwarf2_frame): Replace declaration with that of...
1097 (make_pass_dwarf2_frame): ...new function.
1098 (pass_final): Replace declaration with that of...
1099 (make_pass_final): ...new function.
1100 (pass_rtl_seqabstr): Replace declaration with that of...
1101 (make_pass_rtl_seqabstr): ...new function.
1102 (pass_release_ssa_names): Replace declaration with that of...
1103 (make_pass_release_ssa_names): ...new function.
1104 (pass_early_inline): Replace declaration with that of...
1105 (make_pass_early_inline): ...new function.
1106 (pass_inline_parameters): Replace declaration with that of...
1107 (make_pass_inline_parameters): ...new function.
1108 (pass_update_address_taken): Replace declaration with that of...
1109 (make_pass_update_address_taken): ...new function.
1110 (pass_convert_switch): Replace declaration with that of...
1111 (make_pass_convert_switch): ...new function.
1112 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1113 to a subclass of simple_ipa_opt_pass along with...
1114 (pass_data_ipa_tree_profile): ...new pass_data instance and...
1115 (make_pass_ipa_tree_profile): ...new function.
1116 * tree-sra.c (pass_sra_early): Convert from a global struct to a
1117 subclass of gimple_opt_pass along with...
1118 (pass_data_sra_early): ...new pass_data instance and...
1119 (make_pass_sra_early): ...new function.
1120 (pass_sra): Convert from a global struct to a subclass of
1121 gimple_opt_pass along with...
1122 (pass_data_sra): ...new pass_data instance and...
1123 (make_pass_sra): ...new function.
1124 (pass_early_ipa_sra): Convert from a global struct to a subclass of
1125 gimple_opt_pass along with...
1126 (pass_data_early_ipa_sra): ...new pass_data instance and...
1127 (make_pass_early_ipa_sra): ...new function.
1128 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1129 subclass of gimple_opt_pass along with...
1130 (pass_data_ccp): ...new pass_data instance and...
1131 (make_pass_ccp): ...new function.
1132 (pass_fold_builtins): Convert from a global struct to a subclass of
1133 gimple_opt_pass along with...
1134 (pass_data_fold_builtins): ...new pass_data instance and...
1135 (make_pass_fold_builtins): ...new function.
1136 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1137 subclass of gimple_opt_pass along with...
1138 (pass_data_copy_prop): ...new pass_data instance and...
1139 (make_pass_copy_prop): ...new function.
1140 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1141 global struct to a subclass of gimple_opt_pass along with...
1142 (pass_data_rename_ssa_copies): ...new pass_data instance and...
1143 (make_pass_rename_ssa_copies): ...new function.
1144 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1145 subclass of gimple_opt_pass along with...
1146 (pass_data_dce): ...new pass_data instance and...
1147 (make_pass_dce): ...new function.
1148 (pass_dce_loop): Convert from a global struct to a subclass of
1149 gimple_opt_pass along with...
1150 (pass_data_dce_loop): ...new pass_data instance and...
1151 (make_pass_dce_loop): ...new function.
1152 (pass_cd_dce): Convert from a global struct to a subclass of
1153 gimple_opt_pass along with...
1154 (pass_data_cd_dce): ...new pass_data instance and...
1155 (make_pass_cd_dce): ...new function.
1156 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1157 subclass of gimple_opt_pass along with...
1158 (pass_data_dominator): ...new pass_data instance and...
1159 (make_pass_dominator): ...new function.
1160 (pass_phi_only_cprop): Convert from a global struct to a subclass of
1161 gimple_opt_pass along with...
1162 (pass_data_phi_only_cprop): ...new pass_data instance and...
1163 (make_pass_phi_only_cprop): ...new function.
1164 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1165 subclass of gimple_opt_pass along with...
1166 (pass_data_dse): ...new pass_data instance and...
1167 (make_pass_dse): ...new function.
1168 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1169 a subclass of gimple_opt_pass along with...
1170 (pass_data_forwprop): ...new pass_data instance and...
1171 (make_pass_forwprop): ...new function.
1172 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1173 struct to a subclass of gimple_opt_pass along with...
1174 (pass_data_tree_ifcombine): ...new pass_data instance and...
1175 (make_pass_tree_ifcombine): ...new function.
1176 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1177 subclass of gimple_opt_pass along with...
1178 (pass_data_ch): ...new pass_data instance and...
1179 (make_pass_ch): ...new function.
1180 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1181 subclass of gimple_opt_pass along with...
1182 (pass_data_tree_loop): ...new pass_data instance and...
1183 (make_pass_tree_loop): ...new function.
1184 (pass_tree_loop_init): Convert from a global struct to a subclass of
1185 gimple_opt_pass along with...
1186 (pass_data_tree_loop_init): ...new pass_data instance and...
1187 (make_pass_tree_loop_init): ...new function.
1188 (pass_lim): Convert from a global struct to a subclass of
1189 gimple_opt_pass along with...
1190 (pass_data_lim): ...new pass_data instance and...
1191 (make_pass_lim): ...new function.
1192 (pass_tree_unswitch): Convert from a global struct to a subclass of
1193 gimple_opt_pass along with...
1194 (pass_data_tree_unswitch): ...new pass_data instance and...
1195 (make_pass_tree_unswitch): ...new function.
1196 (pass_predcom): Convert from a global struct to a subclass of
1197 gimple_opt_pass along with...
1198 (pass_data_predcom): ...new pass_data instance and...
1199 (make_pass_predcom): ...new function.
1200 (pass_vectorize): Convert from a global struct to a subclass of
1201 gimple_opt_pass along with...
1202 (pass_data_vectorize): ...new pass_data instance and...
1203 (make_pass_vectorize): ...new function.
1204 (pass_graphite): Convert from a global struct to a subclass of
1205 gimple_opt_pass along with...
1206 (pass_data_graphite): ...new pass_data instance and...
1207 (make_pass_graphite): ...new function.
1208 (pass_graphite_transforms): Convert from a global struct to a subclass
1209 of gimple_opt_pass along with...
1210 (pass_data_graphite_transforms): ...new pass_data instance and...
1211 (make_pass_graphite_transforms): ...new function.
1212 (pass_check_data_deps): Convert from a global struct to a subclass of
1213 gimple_opt_pass along with...
1214 (pass_data_check_data_deps): ...new pass_data instance and...
1215 (make_pass_check_data_deps): ...new function.
1216 (pass_iv_canon): Convert from a global struct to a subclass of
1217 gimple_opt_pass along with...
1218 (pass_data_iv_canon): ...new pass_data instance and...
1219 (make_pass_iv_canon): ...new function.
1220 (pass_scev_cprop): Convert from a global struct to a subclass of
1221 gimple_opt_pass along with...
1222 (pass_data_scev_cprop): ...new pass_data instance and...
1223 (make_pass_scev_cprop): ...new function.
1224 (pass_record_bounds): Convert from a global struct to a subclass of
1225 gimple_opt_pass along with...
1226 (pass_data_record_bounds): ...new pass_data instance and...
1227 (make_pass_record_bounds): ...new function.
1228 (pass_complete_unroll): Convert from a global struct to a subclass of
1229 gimple_opt_pass along with...
1230 (pass_data_complete_unroll): ...new pass_data instance and...
1231 (make_pass_complete_unroll): ...new function.
1232 (pass_complete_unrolli): Convert from a global struct to a subclass of
1233 gimple_opt_pass along with...
1234 (pass_data_complete_unrolli): ...new pass_data instance and...
1235 (make_pass_complete_unrolli): ...new function.
1236 (pass_parallelize_loops): Convert from a global struct to a subclass
1237 of gimple_opt_pass along with...
1238 (pass_data_parallelize_loops): ...new pass_data instance and...
1239 (make_pass_parallelize_loops): ...new function.
1240 (pass_loop_prefetch): Convert from a global struct to a subclass of
1241 gimple_opt_pass along with...
1242 (pass_data_loop_prefetch): ...new pass_data instance and...
1243 (make_pass_loop_prefetch): ...new function.
1244 (pass_iv_optimize): Convert from a global struct to a subclass of
1245 gimple_opt_pass along with...
1246 (pass_data_iv_optimize): ...new pass_data instance and...
1247 (make_pass_iv_optimize): ...new function.
1248 (pass_tree_loop_done): Convert from a global struct to a subclass of
1249 gimple_opt_pass along with...
1250 (pass_data_tree_loop_done): ...new pass_data instance and...
1251 (make_pass_tree_loop_done): ...new function.
1252 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
1253 struct to a subclass of gimple_opt_pass along with...
1254 (pass_data_cse_reciprocals): ...new pass_data instance and...
1255 (make_pass_cse_reciprocals): ...new function.
1256 (pass_cse_sincos): Convert from a global struct to a subclass of
1257 gimple_opt_pass along with...
1258 (pass_data_cse_sincos): ...new pass_data instance and...
1259 (make_pass_cse_sincos): ...new function.
1260 (pass_optimize_bswap): Convert from a global struct to a subclass of
1261 gimple_opt_pass along with...
1262 (pass_data_optimize_bswap): ...new pass_data instance and...
1263 (make_pass_optimize_bswap): ...new function.
1264 (pass_optimize_widening_mul): Convert from a global struct to a
1265 subclass of gimple_opt_pass along with...
1266 (pass_data_optimize_widening_mul): ...new pass_data instance and...
1267 (make_pass_optimize_widening_mul): ...new function.
1268 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
1269 subclass of gimple_opt_pass along with...
1270 (pass_data_phiopt): ...new pass_data instance and...
1271 (make_pass_phiopt): ...new function.
1272 (pass_cselim): Convert from a global struct to a subclass of
1273 gimple_opt_pass along with...
1274 (pass_data_cselim): ...new pass_data instance and...
1275 (make_pass_cselim): ...new function.
1276 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
1277 subclass of gimple_opt_pass along with...
1278 (pass_data_phiprop): ...new pass_data instance and...
1279 (make_pass_phiprop): ...new function.
1280 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
1281 subclass of gimple_opt_pass along with...
1282 (pass_data_pre): ...new pass_data instance and...
1283 (make_pass_pre): ...new function.
1284 (pass_fre): Convert from a global struct to a subclass of
1285 gimple_opt_pass along with...
1286 (pass_data_fre): ...new pass_data instance and...
1287 (make_pass_fre): ...new function.
1288 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
1289 subclass of gimple_opt_pass along with...
1290 (pass_data_reassoc): ...new pass_data instance and...
1291 (make_pass_reassoc): ...new function.
1292 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
1293 subclass of gimple_opt_pass along with...
1294 (pass_data_sink_code): ...new pass_data instance and...
1295 (make_pass_sink_code): ...new function.
1296 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
1297 subclass of gimple_opt_pass along with...
1298 (pass_data_strlen): ...new pass_data instance and...
1299 (make_pass_strlen): ...new function.
1300 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
1301 struct to a subclass of gimple_opt_pass along with...
1302 (pass_data_build_alias): ...new pass_data instance and...
1303 (make_pass_build_alias): ...new function.
1304 (pass_build_ealias): Convert from a global struct to a subclass of
1305 gimple_opt_pass along with...
1306 (pass_data_build_ealias): ...new pass_data instance and...
1307 (make_pass_build_ealias): ...new function.
1308 (pass_ipa_pta): Convert from a global struct to a subclass of
1309 simple_ipa_opt_pass along with...
1310 (pass_data_ipa_pta): ...new pass_data instance and...
1311 (make_pass_ipa_pta): ...new function.
1312 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
1313 subclass of gimple_opt_pass along with...
1314 (pass_data_uncprop): ...new pass_data instance and...
1315 (make_pass_uncprop): ...new function.
1316 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
1317 global struct to a subclass of gimple_opt_pass along with...
1318 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
1319 (make_pass_late_warn_uninitialized): ...new function.
1320 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
1321 to a subclass of gimple_opt_pass along with...
1322 (pass_data_init_datastructures): ...new pass_data instance and...
1323 (make_pass_init_datastructures): ...new function.
1324 (pass_early_warn_uninitialized): Convert from a global struct to a
1325 subclass of gimple_opt_pass along with...
1326 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
1327 (make_pass_early_warn_uninitialized): ...new function.
1328 (pass_update_address_taken): Convert from a global struct to a
1329 subclass of gimple_opt_pass along with...
1330 (pass_data_update_address_taken): ...new pass_data instance and...
1331 (make_pass_update_address_taken): ...new function.
1332 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
1333 struct to a subclass of gimple_opt_pass along with...
1334 (pass_data_release_ssa_names): ...new pass_data instance and...
1335 (make_pass_release_ssa_names): ...new function.
1336 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
1337 subclass of gimple_opt_pass along with...
1338 (pass_data_stdarg): ...new pass_data instance and...
1339 (make_pass_stdarg): ...new function.
1340 * tree-switch-conversion.c (pass_convert_switch): Convert from a
1341 global struct to a subclass of gimple_opt_pass along with...
1342 (pass_data_convert_switch): ...new pass_data instance and...
1343 (make_pass_convert_switch): ...new function.
1344 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
1345 to a subclass of gimple_opt_pass along with...
1346 (pass_data_tail_recursion): ...new pass_data instance and...
1347 (make_pass_tail_recursion): ...new function.
1348 (pass_tail_calls): Convert from a global struct to a subclass of
1349 gimple_opt_pass along with...
1350 (pass_data_tail_calls): ...new pass_data instance and...
1351 (make_pass_tail_calls): ...new function.
1352 * tree-vect-generic.c (pass_lower_vector): Convert from a global
1353 struct to a subclass of gimple_opt_pass along with...
1354 (pass_data_lower_vector): ...new pass_data instance and...
1355 (make_pass_lower_vector): ...new function.
1356 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
1357 gimple_opt_pass along with...
1358 (pass_data_lower_vector_ssa): ...new pass_data instance and...
1359 (make_pass_lower_vector_ssa): ...new function.
1360 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
1361 to a subclass of gimple_opt_pass along with...
1362 (pass_data_slp_vectorize): ...new pass_data instance and...
1363 (make_pass_slp_vectorize): ...new function.
1364 (pass_ipa_increase_alignment): Convert from a global struct to a
1365 subclass of simple_ipa_opt_pass along with...
1366 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
1367 (make_pass_ipa_increase_alignment): ...new function.
1368 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
1369 gimple_opt_pass along with...
1370 (pass_data_vrp): ...new pass_data instance and...
1371 (make_pass_vrp): ...new function.
1372 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
1373 subclass of simple_ipa_opt_pass along with...
1374 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
1375 (make_pass_ipa_free_lang_data): ...new function.
1376 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
1377 gimple_opt_pass along with...
1378 (pass_data_tsan): ...new pass_data instance and...
1379 (make_pass_tsan): ...new function.
1380 (pass_tsan_O0): Convert from a global struct to a subclass of
1381 gimple_opt_pass along with...
1382 (pass_data_tsan_O0): ...new pass_data instance and...
1383 (make_pass_tsan_O0): ...new function.
1384 * var-tracking.c (pass_variable_tracking): Convert from a global
1385 struct to a subclass of rtl_opt_pass along with...
1386 (pass_data_variable_tracking): ...new pass_data instance and...
1387 (make_pass_variable_tracking): ...new function.
1388 * web.c (pass_web): Convert from a global struct to a subclass of
1389 rtl_opt_pass along with...
1390 (pass_data_web): ...new pass_data instance and...
1391 (make_pass_web): ...new function.
1392 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
1393 declaration with that of...
1394 (make_pass_mode_switch_use): ...new function.
1395 (pass_resolve_sw_modes): Replace declaration with that of...
1396 (make_pass_resolve_sw_modes): ...new function.
1397 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
1398 from a global struct to a subclass of rtl_opt_pass along with...
1399 (pass_data_mode_switch_use): ...new pass_data instance and...
1400 (make_pass_mode_switch_use): ...new function.
1401 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
1402 from a global struct to a subclass of rtl_opt_pass along with...
1403 (pass_data_resolve_sw_modes): ...new pass_data instance and...
1404 (make_pass_resolve_sw_modes): ...new function.
1405 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
1406 struct to a subclass of rtl_opt_pass along with...
1407 (pass_data_insert_vzeroupper): ...new pass_data instance and...
1408 (make_pass_insert_vzeroupper): ...new function.
1409 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
1410 global struct to a subclass of rtl_opt_pass along with...
1411 (pass_data_work_around_errata): ...new pass_data instance and...
1412 (make_pass_work_around_errata): ...new function.
1413 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
1414 struct to a subclass of rtl_opt_pass along with...
1415 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
1416 (make_pass_mips_machine_reorg2): ...new function.
1417
04b48e18 14182013-08-05 David Malcolm <dmalcolm@redhat.com>
1419
1420 * passes.c (pass_manager::operator new): New.
1421
bcfddb5b 14222013-08-05 David Malcolm <dmalcolm@redhat.com>
1423
1424 Handwritten part of conversion of passes to C++ classes.
1425
1426 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
1427 (toplev.o): Add dep on PASS_MANAGER_H.
1428 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
1429 of early local pases to reflect this moving from a global to a
1430 member of gcc::pass_manager.
1431 (cgraph_add_new_function): Likewise.
1432 * lto-cgraph.c (lto_output_node): Update for conversion of
1433 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
1434 * passes.c (opt_pass::clone): New.
1435 (opt_pass::gate): New.
1436 (opt_pass::execute): New.
1437 (opt_pass::opt_pass): New.
1438 (pass_manager::execute_early_local_passes): New.
1439 (pass_manager::execute_pass_mode_switching): new.
1440 (finish_optimization_passes): Convert to...
1441 (pass_manager::finish_optimization_passes): ...this.
1442 (finish_optimization_passes): Update for conversion of passes to
1443 C++ classes.
1444 (register_dump_files_1): Use has_gate since we cannot portably
1445 check a vtable entry against NULL.
1446 (dump_one_pass): Likewise.
1447 (ipa_write_summaries_2): Likewise.
1448 (ipa_write_optimization_summaries_1): Likewise.
1449 (ipa_read_summaries_1): Likewise.
1450 (ipa_read_optimization_summaries_1): Likewise.
1451 (execute_ipa_stmt_fixups): Likewise.
1452 (pass_manager::pass_manager): Rewrite pass-creation, invoking
1453 pass-creation functions rather than wiring up globals, and
1454 storing the results in fields of pass_manager generated using
1455 pass-instances.def.
1456 (pass_manager::dump_profile_report): Update for conversion of
1457 passes to C++ classes.
1458 (pass_manager::execute_ipa_summary_passes): Likewise.
1459 (execute_one_ipa_transform_pass): Likewise.
1460 (execute_one_pass): Use has_gate and has_execute since we cannot
1461 portably check a vtable entry against NULL.
1462 * pass_manager.h (pass_manager::finish_optimization_passes): New.
1463 (pass_manager): Use pass-instances.def to add fields for the
1464 various pass instances.
1465 * toplev.c (finalize): Update for move of
1466 finish_optimization_passes to a method of gcc::pass_manager.
1467 * toplev.h (finish_optimization_passes): Move to method of class
1468 pass_manager.
1469 * tree-pass.h (struct pass_data): New.
1470 (opt_pass): Convert to C++ class, make it a subclass of
1471 pass_data.
1472 (opt_pass::gate): Convert to virtual function.
1473 (opt_pass::~opt_pass): New.
1474 (opt_pass::clone): New.
1475 (opt_pass::execute): Convert to virtual function.
1476 (opt_pass::opt_pass): New.
1477 (opt_pass::ctxt_): new.
1478 (gimple_opt_pass): Convert to subclass of opt_pass.
1479 (gimple_opt_pass::gimple_opt_pass): New.
1480 (rtl_opt_pass): Convert to subclass of opt_pass.
1481 (rtl_opt_pass::rtl_opt_pass): New.
1482 (ipa_opt_pass_d): Convert to subclass of opt_pass.
1483 (ipa_opt_pass_d::ipa_opt_pass_d): New.
1484 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
1485 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
1486 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
1487 invocation of pass_mode_switching to reflect this moving from a
1488 global to a member of gcc::pass_manager.
1489 (ix86_option_override): Rework how pass_insert_vzeroupper is
1490 added to the pass_manager to reflect autogenerated changes.
1491 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and
1492 PASS_MANAGER_H.
1493
cdbf60c9 14942013-08-05 Richard Earnshaw <rearnsha@arm.com>
1495
1496 PR rtl-optimization/57708
1497 * recog.c (peep2_find_free_register): Validate all regs in a
1498 multi-reg mode.
1499
281dea26 15002013-08-05 Jan Hubicka <jh@suse.cz>
1501
1502 PR lto/57602
1503 * cgraph.c (verify_cgraph_node): Accept local flags from other partitions.
1504 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
1505 (function_and_variable_visibility): Likewise.
1506 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
1507
4e765d3c 15082013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
1509
1510 * graph.c (init_graph_slim_pretty_print): Remove.
1511 (print_graph_cfg): Do not call it. Use local pretty printer.
1512 (start_graph_dump): Likewise.
1513
ba0714c6 15142013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
1515
1516 * gimple-pretty-print.c (buffer): Remove.
1517 (initialized): Likewise.
1518 (maybe_init_pretty_print): Likewise.
1519 (print_gimple_stmt): Do not call it. Use non-static local
1520 pretty_printer variable.
1521 (print_gimple_expr): Likewise.
1522 (print_gimple_seq): Likewise.
1523 (gimple_dump_bb): Likewise.
1524
b75d2c15 15252013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
1526
1527 * asan.c (asan_pp): Remove.
1528 (asan_pp_initialized): Likewise.
1529 (asan_pp_initialize): Likewise.
1530 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
1531 (asan_emit_stack_protection): Tidy. Use local pretty printer.
1532 (asan_add_global): Likewise.
1533
a94db6b0 15342013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
1535
1536 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
1537 * diagnostic.h (diagnostic_flush_buffer): Adjust.
1538 * pretty-print.c (pp_formatted_text_data): Likewise.
1539 (pp_indent): Rename from pp_base_indent.
1540 (pp_format): Rename from pp_base_format.
1541 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
1542 (pp_format_verbatim): Rename from pp_base_format_verbatim.
1543 (pp_flush): Rename from pp_base_flush.
1544 (pp_set_line_maximum_length): Rename from
1545 pp_base_set_line_maximum_length.
1546 (pp_clear_output_area): Rename from pp_base_clear_output_area.
1547 (pp_set_prefix): Rename from pp_base_set_prefix.
1548 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
1549 (pp_emit_prefix): Rename from pp_base_emit_prefix.
1550 (pp_append_text): Rename from pp_base_append_text.
1551 (pp_formatted_text): Rename from pp_base_formatted_text.
1552 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
1553 (pp_remaining_character_count_for_line): Rename from
1554 pp_base_remaining_character_count_for_line.
1555 (pp_newline): Rename from pp_base_newline.
1556 (pp_character): Rename from pp_base_character.
1557 (pp_string): Rename from pp_base_string.
1558 (pp_maybe_space): Rename from pp_base_maybe_space.
1559 * asan.c (asan_pp_string): Adjust.
1560 (asan_emit_stack_protection): Likewise.
1561 (asan_add_global): Likewise.
1562 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
1563 * tree-mudflap.c (mf_varname_tree): Likewise.
1564 * tree-pretty-print.c (pp_tree_identifier): Rename from
1565 pp_base_tree_identifier.
1566 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
1567 Declare as function.
1568
70d60d1d 15692013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
1570
1571 * pretty-print.h (pp_bar_bar): New.
1572 (pp_ampersand_ampersand): Likewise.
1573 (pp_less_equal): Likewise.
1574 (pp_greater_equal): Likewise.
1575 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
1576 printer functions instead of pp_string or operators and punctuators.
1577 (dump_gimple_call): Likewise.
1578 (dump_gimple_omp_for): Likewise.
1579 (dump_gimple_transaction): Likewise.
1580 (dump_gimple_phi): Likewise.
1581 (pp_gimple_stmt_1): Likewise.
1582 * sched-vis.c (print_insn): Likewise.
1583 * tree-mudflap.c (mf_varname_tree): Likewise.
1584 * tree-pretty-print.c (dump_block_node): Likewise.
1585 (dump_generic_node): Likewise.
1586
4df870fd 15872013-08-02 Jan Hubicka <jh@suse.cz>
1588
1589 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
1590 boundaries.
1591 * lto-streamer-out.c (tree_is_indexable): Results decls and
1592 parm decls are not indexable.
1593 (DFS_write_tree_body): Do not follow args and results.
1594 (hash_tree): Likewise.
1595 (output_functions): Rearrange so struct function is needed
1596 only when real body is output; be able to also ouptut abstract
1597 functions; output DECL_ARGUMENTS and DECL_RESULT.
1598 (lto_output): When not in WPA, ale store abstract functions.
1599 (write_symbol): Do not care about RESULT_DECL.
1600 (output_symbol_p): Handle correctly sbtract decls.
1601 * lto-streamer-in.c (input_function): Rearrange so struct
1602 function can be NULL at entry; allow streaming of
1603 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
1604 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
1605 sanity check during LTO.
1606 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
1607 RESULT_DECl and DECL_ARGUMENTS.
1608 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
1609 Likewise.
1610
dda4f0ec 16112013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
1612
1613 * pretty-print.h (pp_underscore): New.
1614 (pp_comma): Tidy.
1615 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
1616 printer functions instead of pp_character.
1617 (dump_binary_rhs): Likewise.
1618 (dump_ternary_rhs): Likewise.
1619 (dump_gimple_call_args): Likewise.
1620 (pp_points_to_solution): Likewise.
1621 (dump_gimple_call): Likewise.
1622 (dump_gimple_switch): Likewise.
1623 (dump_gimple_cond): Likewise.
1624 (dump_gimple_bind): Likewise.
1625 (dump_gimple_try): Likewise.
1626 (dump_gimple_omp_for): Likewise.
1627 (dump_gimple_omp_continue): Likewise.
1628 (dump_gimple_omp_single): Likewise.
1629 (dump_gimple_omp_sections): Likewise.
1630 (dump_gimple_omp_block): Likewise.
1631 (dump_gimple_omp_critical): Likewise.
1632 (dump_gimple_transaction): Likewise.
1633 (dump_gimple_asm): Likewise.
1634 (dump_gimple_phi): Likewise.
1635 (dump_gimple_omp_parallel): Likewise.
1636 (dump_gimple_omp_task): Likewise.
1637 (dump_gimple_omp_atomic_load): Likewise.
1638 (dump_gimple_omp_atomic_store): Likewise.
1639 (dump_gimple_mem_ops): Likewise.
1640 (pp_gimple_stmt_1): Likewise.
1641 (pp_cfg_jump): Likewise.
1642 (dump_implicit_edges): Likewise.
1643 (gimple_dump_bb_for_graph): Likewise.
1644 * graph.c (draw_cfg_node): Likewise.
1645 * langhooks.c (lhd_print_error_function): Likewise.
1646 * sched-vis.c (print_exp): Likewise.
1647 (print_value): Likewise.
1648 (print_pattern): Likewise.
1649 (print_insn): Likewise.
1650 (rtl_dump_bb_for_graph): Likewise.
1651 * tree-pretty-print.c (dump_function_declaration): Likewise.
1652 (dump_array_domain): Likewise.
1653 (dump_omp_clause): Likewise.
1654 (dump_location): Likewise.
1655 (dump_generic_node): Likewise.
1656 (print_struct_decl): Likewise.
1657 * diagnostic.c (diagnostic_show_locus): Use pp_space.
1658
dc77dea1 16592013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1660
1661 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
1662 candidate table when replacing a candidate statement.
1663 (replace_rhs_if_not_dup): Likewise.
1664 (replace_one_candidate): Likewise.
1665
f2c6b33f 16662013-08-02 Jan Hubicka <jh@suse.cz>
741d4110 1667 Martin Liska <marxin.liska@gmail.com>
f2c6b33f 1668
1669 * cgraphunit.c (add_new_function): Fix logic when adding from
1670 late IPA pass.
1671 (assemble_thunk): Rename to ...
1672 (expand_thunk); .. this one; export; get it working with general functions;
1673 make produced gimple valid.
1674 * cgraph.h (expand_thunk): Declare.
1675
09ab6335 16762013-08-02 Jan Hubicka <jh@suse.cz>
1677
1678 * ipa-cp.c (gather_context_independent_values): Use ipa_get_param_move_cost.
1679 (get_replacement_map): Remove PARAM; move parameter folding into tree-inline.c
1680 (create_specialized_node): Update.
1681 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
1682 assert that we have gimple body; update move_cost.
1683 (count_formal_params): Assert that we have gimple body.
1684 (ipa_dump_param): New function.
1685 (ipa_alloc_node_params): Break out from ...
1686 (ipa_initialize_node_params): ... here.
1687 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
1688 (ipa_write_node_info): Stream move costs.
1689 (ipa_read_node_info): Read move costs.
1690 (ipa_update_after_lto_read): Do not recompute node params.
1691 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
1692 (ipa_get_param): Check we are not in WPA.
1693 (ipa_get_param_move_cost): New.
1694 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
1695 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
1696 parm numbers to be present.
1697
1524bcdc 16982013-08-02 Vladimir Makarov <vmakarov@redhat.com>
1699
1700 PR rtl-optimization/58048
1701 * lra-constraints.c (process_alt_operands): Don't check asm
1702 operand on register.
1703
49539027 17042013-08-02 Eric Botcazou <ebotcazou@adacore.com>
1705
1706 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
1707 the implied StoreLoad barrier for atomic operations if before.
1708
44e82502 17092013-08-02 Jan Hubicka <jh@suse.cz>
1710 Martin Liska <marxin.liska@gmail.com>
1711
1712 * cgraph.c (cgraph_function_body_availability): Do not check cgrpah flags.
1713 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
1714 symtab_node_availability): Declare.
1715 * ipa.c (can_replace_by_local_alias): New.
1716 (function_and_variable_visibility): Use it.
1717 * symtab.c (symtab_for_node_and_aliases, symtab_nonoverwritable_alias_1,
1718 symtab_nonoverwritable_alias): New.
1719
691cfda4 17202013-08-02 Vladimir Makarov <vmakarov@redhat.com>
1721
1722 PR rtl-optimization/57963
1723 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p):
1724 New.
1725 (lra_constraints): Use them.
1726
062447bb 17272013-08-02 Sofiane Naci <sofiane.naci@arm.com>
1728
1729 * config/arm/types.md (define_attr "type"): Add "load_acq" and "store_rel".
1730 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
1731 changes.
1732 (cortex_a53_store1): Likewise.
1733
f08369a7 17342013-08-01 Jan Hubicka <jh@suse.cz>
1735
1736 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other partitions are
1737 not needed.
1738
3f40d0f8 17392013-08-01 Uros Bizjak <ubizjak@gmail.com>
1740
1741 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
1742 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
1743 MAYBE_NON_Q_CLASS_P where appropriate.
1744
adf64732 17452013-08-01 Jan Hubicka <jh@suse.cz>
1746
1747 * cgraph.h (release_function_body): Declare.
1748 * tree.c (free_lang_data_in_decl): Free, parameters and return values
1749 of unused delcarations.
1750
36ee0cde 17512013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1752
1753 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical RTL form
1754 when subtracting a constant.
1755
31991287 17562013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1757
1758 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
1759 Generate canonical plus rtx with negated immediate instead of minus
1760 where appropriate.
1761 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
1762
abb1a237 17632013-08-01 Jan Hubicka <jh@suse.cz>
1764
1765 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
1766 (cgraph_release_function_body): Likewise.
1767 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
1768 * cgraph.h (cgrpah_node): Rename abstract_and_needed
1769 to used_as_abstract_origin.
1770 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
1771 symbols used as abstract origins.
1772 * cgraphunit.c (analyze_functions): Update.
3f40d0f8 1773 * ipa.c (symtab_remove_unreachable_nodes): Recompute
1774 used_as_abstract_origin.
abb1a237 1775 * tree-inline.c (tree_function_versioning): Update
3f40d0f8 1776 used_as_abstract_origin; be ready for DECL_RESULT and
1777 DECL_ARGUMENTS to be NULL.
abb1a237 1778
3f40d0f8 1779 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
1780 for abstract functions.
1781 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
1782 real symbols.
abb1a237 1783
d4974b91 17842013-08-01 Jan Hubicka <jh@suse.cz>
1785
1786 * profile.c (compute_value_histograms): Fix thinko.
1787
aa7f84c2 17882013-08-01 Sofiane Naci <sofiane.naci@arm.com>
1789
1790 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
1791 aarch-common-protos.h to extra_headers.
1792 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
1793 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
1794 * config/aarch64/t-aarch64 (aarch-common.o): Define.
1795
0c9bcd94 17962013-08-01 Sofiane Naci <sofiane.naci@arm.com>
1797
1798 * config/aarch64/aarch64.md (define_attr "type"): Delete.
1799 Include "../arm/types.md". Define "type" attribute for all patterns.
1800 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
1801 attribute changes.
1802
a081294b 18032013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1804
1805 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
1806 to support power8 load fusion.
1807 (fusion_gpr_mem_load): Likewise.
1808
1809 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
1810
1811 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
1812 declarations for power8 load fusion.
1813 (emit_fusion_gpr_load): Likewise.
1814
1815 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
1816 tuning for power8, turn on fusion mode by default. Turn on sign
1817 extending fusion mode if normal fusion mode is on, and we are at
1818 -O2 or -O3.
1819 (fusion_gpr_load_p): New function, return true if we can fuse an
1820 addis instruction with a dependent load to a GPR.
1821 (emit_fusion_gpr_load): Emit the instructions for power8 load
1822 fusion to GPRs.
1823
3f40d0f8 1824 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
a081294b 1825 (VSX load fusion peepholes): New peepholes to fuse together an
1826 addi instruction with a VSX load instruction.
1827
1828 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
1829 peepholes to fuse an addis instruction with a load to a GPR base
1830 register. If we are supporting sign extending fusions, convert
1831 sign extending loads to zero extending loads and add an explicit
1832 sign extension.
1833
015adf41 18342013-07-31 Sofiane Naci <sofiane.naci@arm.com>
1835
1836 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
1837 aarch-common-protos.h to extra_headers.
1838 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
1839 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
1840 (arm_early_store_addr_dep): Likewise.
3f40d0f8 1841 (arm_no_early_alu_shift_dep): Likewise.
1842 (arm_no_early_alu_shift_value_dep): Likewise.
1843 (arm_no_early_mul_dep): Likewise.
1844 (arm_no_early_store_addr_dep): Likewise.
1845 (arm_mac_accumulator_is_mul_result): Likewise.
1846 (arm_mac_accumulator_is_result): Likewise.
015adf41 1847 * config/arm/aarch-common.c: ... here. New file.
3f40d0f8 1848 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
1849 here to ...
015adf41 1850 (arm_early_store_addr_dep): Likewise.
3f40d0f8 1851 (arm_no_early_alu_shift_dep): Likewise.
1852 (arm_no_early_alu_shift_value_dep): Likewise.
1853 (arm_no_early_mul_dep): Likewise.
1854 (arm_no_early_store_addr_dep): Likewise.
1855 (arm_mac_accumulator_is_mul_result): Likewise.
1856 (arm_mac_accumulator_is_result): Likewise.
015adf41 1857 * config/arm/aarch-common-protos.h: ... here. New file.
1858 * config/arm/t-arm (aarch-common.o): Define.
1859
18602013-07-31 Sofiane Naci <sofiane.naci@arm.com>
1861
1862 * config/arm/arm.md: Include new file "types.md".
1863 (define_attr "type"): Move from here to ...
1864 (define_attr "mul32"): Likewise.
1865 (define_attr "mul64"): Likewise.
1866 * config/arm/types.md: ... here. New file.
1867
d69684b5 18682013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
1869
1870 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
1871 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
1872
dee86a54 18732013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
90bccf02 1874
1875 * gen-pass-instances.awk: Fix offset of substr().
1876
743d5ab7 18772013-07-31 David Malcolm <dmalcolm@redhat.com>
1878
1879 * Makefile.in (pass-instances.def): New.
1880 (passes.o): Replace dependency on passes.def with one on
1881 pass-instances.def
1882
1883 * gen-pass-instances.awk: New.
1884
1885 * passes.c (pass_manager::pass_manager): Use pass-instances.def
1886 rather than passes.def, updating local definition of NEXT_PASS
1887 macro to add an extra NUM parameter (currently unused).
1888
3ea50c01 18892013-07-30 David Malcolm <dmalcolm@redhat.com>
1890
1891 * Makefile.in (PASS_MANAGER_H): New.
1892 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
1893 (passes.o): Likewise.
1894 (statistics.o): Likewise.
1895 (cgraphunit.o): Likewise.
1896 (context.o): Depend on PASS_MANAGER_H.
1897
1898 * pass_manager.h: New.
1899
1900 * cgraphunit.c (cgraph_add_new_function): Update for moves
1901 of globals to fields of pass_manager.
1902 (analyze_function): Likewise.
1903 (expand_function): Likewise.
1904 (ipa_passes): Likewise.
1905 (compile): Likewise.
1906
1907 * context.c (context::context): New.
1908 * context.h (context::context): New.
1909 (context::get_passes): New.
1910 (context::passes_): New.
1911
1912 * lto-cgraph.c (input_node): Update for moves of globals to
1913 fields of pass_manager.
1914
1915 * passes.c (all_passes): Remove, in favor of a field of the
1916 same name within the new class pass_manager.
1917 (all_small_ipa_passes): Likewise.
1918 (all_lowering_passes): Likewise.
1919 (all_regular_ipa_passes): Likewise.
1920 (all_late_ipa_passes): Likewise.
1921 (all_lto_gen_passes): Likewise.
1922 (passes_by_id): Likewise.
1923 (passes_by_id_size): Likewise.
1924 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
1925 the new class pass_manager.
1926 (set_pass_for_id): Convert to...
1927 (pass_manager::set_pass_for_id): ...method.
1928 (get_pass_for_id): Convert to...
1929 (pass_manager::get_pass_for_id): ...method.
1930 (register_one_dump_file): Move body of implementation into...
1931 (pass_manager::register_one_dump_file): ...here.
1932 (register_dump_files_1): Convert to...
1933 (pass_manager::register_dump_files_1): ...method.
1934 (register_dump_files): Convert to...
1935 (pass_manager::register_dump_files): ...method.
1936 (create_pass_tab): Update for moves of globals to fields of
1937 pass_manager.
1938 (dump_passes): Move body of implementation into...
1939 (pass_manager::dump_passes): ...here.
1940 (register_pass): Move body of implementation into...
1941 (pass_manager::register_pass): ...here.
1942 (init_optimization_passes): Convert into...
1943 (pass_manager::pass_manager): ...constructor for new
1944 pass_manager class, and initialize the pass_lists array.
1945 (check_profile_consistency): Update for moves of globals to
1946 fields of pass_manager.
1947 (dump_profile_report): Move body of implementation into...
1948 (pass_manager::dump_profile_report): ...here.
1949 (ipa_write_summaries_1): Update for moves of pass lists from
1950 being globals to fields of pass_manager.
1951 (ipa_write_optimization_summaries): Likewise.
1952 (ipa_read_summaries): Likewise.
1953 (ipa_read_optimization_summaries): Likewise.
1954 (execute_all_ipa_stmt_fixups): Likewise.
1955
1956 * statistics.c (statistics_fini): Update for moves of globals to
1957 fields of pass_manager.
1958
1959 * toplev.c (general_init): Replace call to
1960 init_optimization_passes with construction of the pass_manager
1961 instance.
1962
1963 * tree-pass.h (all_passes): Remove, in favor of a field of the
1964 same name within the new class pass_manager.
1965 (all_small_ipa_passes): Likewise.
1966 (all_lowering_passes): Likewise.
1967 (all_regular_ipa_passes): Likewise.
1968 (all_lto_gen_passes): Likewise.
1969 (all_late_ipa_passes): Likewise.
1970 (passes_by_id): Likewise.
1971 (passes_by_id_size): Likewise.
1972 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
1973 the new class pass_manager.
1974 (get_pass_for_id): Remove.
1975
d9395640 19762013-07-30 Richard Earnshaw <rearnsha@arm.com>
1977
1978 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
1979 configs.
1980
4422d91f 19812013-07-30 Richard Earnshaw <rearnsha@arm.com>
1982
1983 * arm.md (mulhi3): New expand pattern.
1984
2084f25d 19852013-07-30 Jan Hubicka <jh@suse.cz>
1986 Martin Liska <marxin.liska@gmail.com>
1987
1988 * profile.c (compute_value_histograms): Do not ICE when
1989 there is mismatch only on some counters.
1990
48c8dd80 19912013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1992
1993 PR rtl-optimization/57637
1994 * function.c (move_insn_for_shrink_wrap): Also check the
1995 GEN set of the LIVE problem for the liveness analysis
1996 if it exists, otherwise give up.
1997
8f0676e4 19982013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
1999
2000 PR tree-optimization/57993
2001 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2002 replaced statement in the candidate table.
2003 (phi_add_costs): Return infinite cost when the hidden basis does
2004 not dominate all phis on which the candidate is dependent.
2005 (replace_one_candidate): Record replaced statement in the
2006 candidate table.
2007
37905fbf 20082013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
2009
2010 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2011 (ashlv2si3): New expander.
2012 (*ashlv2si3_i): New define_insn_and_split.
2013 * predicates.md (float_operation): Allow patterns with three
2014 basic sub-patterns.
2015
4da102b5 2016 PR rtl-optimization/58021
2017 * mode-switching.c (create_pre_exit): Always split off preceding
2018 insns if we are not at the basic block head.
2019
0bd32132 20202013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
2021
2022 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2023 (UCLIBC_DYNAMIC_LINKER): New macro.
2024 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2025 `-mnan=2008'.
2026 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2027 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
2028 `-mnan=2008'.
2029 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2030 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2031 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2032 for SF and DF modes. Use ieee_quad_format for TF mode.
2033 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2034 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2035 (mips_option_override): Handle `-mnan=legacy'.
2036 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2037 `-mabs=2008' and `-mnan=2008'.
2038 (OPTION_DEFAULT_SPECS): Add "nan" default.
2039 (ASM_SPEC): Handle `-mnan='.
2040 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2041 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2042 comment accordingly.
2043 (neg<mode>2): Likewise.
2044 * config/mips/mips.opt (mabs, mnan): New options.
2045 * doc/install.texi (Configuration): Document `--with-nan=' option.
2046 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2047 `-mnan=' options.
2048 (MIPS Options): Document them.
2049 * config.gcc <mips*-*-*>: Handle `--with-nan='.
2050 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2051 * configure: Regenerate.
2052 * config.in: Regenerate.
2053
9834debf 20542013-07-29 Uros Bizjak <ubizjak@gmail.com>
2055
2056 * config/i386/i386.md (float post-reload splitters): Do not check
2057 for subregs of SSE registers.
2058
20592013-07-29 Uros Bizjak <ubizjak@gmail.com>
2060 H.J. Lu <hongjiu.lu@intel.com>
2061
2062 PR target/57954
2063 PR target/57988
2064 * config/i386/i386.md (post-reload splitter
2065 to avoid partial SSE reg dependency stalls): New pattern.
2066
8a92cb4c 20672013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
2068
2069 * config/s390/s390.md ("movcc"): Swap load and store instructions.
2070
d7fb9af4 20712013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
2072
2073 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2074 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2075
09bda26a 20762013-07-26 Cary Coutant <ccoutant@google.com>
2077
2078 * dwarf2out.c (die_checksum_ordered): Don't include template
2079 instantiations in signature.
2080 (is_template_parameter): New function.
2081 (is_template_instantiation): New function.
2082 (generate_skeleton_bottom_up): Don't include template instantiations
2083 in type unit DIE.
2084 (generate_skeleton): Likewise.
2085 (break_out_comdat_types): Move recursive call to break out nested
2086 types earlier.
2087 (prune_unused_types_mark_generic_parms_dies): Call
2088 is_template_parameter.
2089
d0755ec6 20902013-07-26 Ian Bolton <ian.bolton@arm.com>
2091
2092 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2093 uses vector registers.
47c77d8b 2094 * config/aarch64/iterators.md: Add attributes rtn and vas.
d0755ec6 2095
90404b57 20962013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9834debf 2097 Richard Earnshaw <richard.earnshaw@arm.com>
90404b57 2098
2099 * combine.c (simplify_comparison): Re-canonicalize operands
9834debf 2100 where appropriate.
90404b57 2101 * config/arm/arm.md (movcond_addsi): New splitter.
2102
50b2b09e 21032013-07-25 Sterling Augustine <saugustine@google.com>
2104
2105 * dwarf2out.c (size_of_pubnames): Move code to...
2106 (include_pubname_in_output): ...here. New.
2107 (want_pubnames): Rearrange.
2108 (output_pubnames): Call include_pubname_in_output. Move assertion.
2109
542c9bab 21102013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
2111
2112 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2113
21142013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
2115
2116 PR target/38836
2117 * doc/extend.texi: Remove obsolete builtins. Fix
2118 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2119
79e830ee 21202013-07-25 Jan Hubicka <jh@suse.cz>
2121
2122 * cgraph.c (release_function_body): Break out from ...
2123 (cgraph_release_function_body): ... this one; also release DECL_RESULT
2124 and DECL_ARGUMENTS.
2125 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2126 old_tree in the map.
2127 (create_specialized_node): Update.
2128 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2129 into index.
542c9bab 2130 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2131 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
79e830ee 2132 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2133 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2134 DECL_RESULT.
2135
a0b3420a 21362013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2137
2138 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2139 addsi3_carryin_alt2_<optab>): Correct output template.
2140
a2c9e1be 21412013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2142
2143 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2144 Adjust for arm_restrict_it.
2145 Remove trailing whitespace.
2146
c6b1c11d 21472013-07-25  Mark Kettenis  <kettenis@openbsd.org>
2148
01dd1649 2149 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2150 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2151
c6b1c11d 2152 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2153
8177deec 21542013-07-25 Vladimir Makarov <vmakarov@redhat.com>
2155
2156 PR rtl-optimization/57960
2157 * lra-constraints.c (process_alt_operands): Use the right mode
2158 when checking strict_low.
2159
b9b49047 21602013-07-25 Jan Hubicka <jh@suse.cz>
2161
2162 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2163 * cgraph.c (cgraph_remove_node): Do not release function body
2164 when in cgraph streaming.
2165 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
542c9bab 2166 in other partitions are not considered reachable; fix handling of
2167 clones.
b9b49047 2168
947d113e 21692013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2170
2171 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2172
84ce8e5c 21732013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2174
2175 PR target/19599
2176 PR target/57731
931e9e20 2177 PR target/57837
947d113e 2178 * config/arm/arm.md ("*sibcall_insn): Replace use of
84ce8e5c 2179 Ss with US. Adjust output for v5 and v4t.
542c9bab 2180 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
947d113e 2181
84ce8e5c 2182 * config/arm/constraints.md ("Ss"): Rename to US.
2183
13a19995 21842013-07-25 Terry Guo <terry.guo@arm.com>
2185
2186 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2187 shift_add/shift_sub0/shift_sub1 RTXs.
2188
bdf9958f 21892013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
2190 Anton Blanchard <anton@au1.ibm.com>
2191
2192 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2193 (altivec_vpks<VI_char>ss): Likewise.
2194 (altivec_vpks<VI_char>us): Likewise.
2195 (altivec_vpku<VI_char>us): Likewise.
2196 (altivec_vpku<VI_char>um): Likewise.
2197
8d0badf1 21982013-07-24 David Malcolm <dmalcolm@redhat.com>
2199
2200 Introduce context class.
2201
2202 * Makefile.in (CONTEXT_H): New.
2203 (OBJS): Add context.o.
2204 (toplev.o): Add CONTEXT_H to dependencies.
2205 (context.o): New.
2206
542c9bab 2207 * toplev.c (general_init): Create the singleton gcc::context instance.
8d0badf1 2208
2209 * context.c: New.
2210
2211 * context.h: New.
2212
39be9d26 22132013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
2214
2215 PR rtl-optimization/57968
2216 * mode-switching.c (create_pre_exit): Allow instructions that
2217 don't set a return register to need a non-exit mode.
2218
c2dae78e 22192013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2220 Anton Blanchard <anton@au1.ibm.com>
2221
5ecb6d80 2222 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
2223 operands to vperm for little endian.
2224 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
2225 of lvsl to create the control mask for a vperm for little endian.
c2dae78e 2226
7b1292cc 22272013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2228 Anton Blanchard <anton@au1.ibm.com>
2229
2230 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2231 two operands for little-endian.
2232
1d5fba0b 22332013-07-23 Steve Ellcey <sellcey@mips.com>
2234
2235 * config/mips/mips.c (mips_case_values_threshold): New.
2236 (TARGET_CASE_VALUES_THRESHOLD): Define.
2237
712bc3b7 22382013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2239 Anton Blanchard <anton@au1.ibm.com>
2240
2241 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
2242 selection of field for vector splat in little endian mode.
2243
91bf8773 22442013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2245
2246 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
2247 expanders to rs6000.md.
2248 (ior<mode>3): Likewise.
2249 (and<mode>3): Likewise.
2250 (one_cmpl<mode>2): Likewise.
2251 (nor<mode>3): Likewise.
2252 (andc<mode>3): Likewise.
2253 (eqv<mode>3): Likewise.
2254 (nand<mode>3): Likewise.
2255 (orc<mode>3): Likewise.
2256
2257 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
2258 declaration.
2259
2260 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
2261 to split multi-word logical operations.
2262 (rs6000_split_logical_di): Likewise.
2263 (rs6000_split_logical): Likewise.
2264
2265 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
2266 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
2267 and allow TImode operations in 32-bit.
2268 (vsx_and<mode>3_64bit): Likewise.
2269 (vsx_ior<mode>3_32bit): Likewise.
2270 (vsx_ior<mode>3_64bit): Likewise.
2271 (vsx_xor<mode>3_32bit): Likewise.
2272 (vsx_xor<mode>3_64bit): Likewise.
2273 (vsx_one_cmpl<mode>2_32bit): Likewise.
2274 (vsx_one_cmpl<mode>2_64bit): Likewise.
2275 (vsx_nor<mode>3_32bit): Likewise.
2276 (vsx_nor<mode>3_64bit): Likewise.
2277 (vsx_andc<mode>3_32bit): Likewise.
2278 (vsx_andc<mode>3_64bit): Likewise.
2279 (vsx_eqv<mode>3_32bit): Likewise.
2280 (vsx_eqv<mode>3_64bit): Likewise.
2281 (vsx_nand<mode>3_32bit): Likewise.
2282 (vsx_nand<mode>3_64bit): Likewise.
2283 (vsx_orc<mode>3_32bit): Likewise.
2284 (vsx_orc<mode>3_64bit): Likewise.
2285
2286 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
2287 logical types in GPRs.
2288
2289 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
2290 logical insns to rs6000.md, and allow TImode operations in
2291 32-bit.
2292 (altivec_ior<mode>3): Likewise.
2293 (altivec_xor<mode>3): Likewise.
2294 (altivec_one_cmpl<mode>2): Likewise.
2295 (altivec_nor<mode>3): Likewise.
2296 (altivec_andc<mode>3): Likewise.
2297
2298 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
2299 attributes for moving the 128-bit logical operations into
2300 rs6000.md.
2301 (BOOL_REGS_OUTPUT): Likewise.
2302 (BOOL_REGS_OP1): Likewise.
2303 (BOOL_REGS_OP2): Likewise.
2304 (BOOL_REGS_UNARY): Likewise.
2305 (BOOL_REGS_AND_CR0): Likewise.
2306 (one_cmpl<mode>2): Add support for DI logical operations on
2307 32-bit, splitting the operations to 32-bit.
2308 (anddi3): Likewise.
2309 (iordi3): Likewise.
2310 (xordi3): Likewise.
2311 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
2312 changes to combine the 32/64-bit code, allow logical operations on
2313 TI mode in 32-bit, and to use similar match_operator patterns like
2314 scalar mode uses. Combine the Altivec and VSX code for logical
2315 operations, and move it here.
2316 (ior<mode>3, 128-bit types): Likewise.
2317 (xor<mode>3, 128-bit types): Likewise.
2318 (one_cmpl<mode>3, 128-bit types): Likewise.
2319 (nor<mode>3, 128-bit types): Likewise.
2320 (andc<mode>3, 128-bit types): Likewise.
2321 (eqv<mode>3, 128-bit types): Likewise.
2322 (nand<mode>3, 128-bit types): Likewise.
2323 (orc<mode>3, 128-bit types): Likewise.
2324 (and<mode>3_internal): Likewise.
2325 (bool<mode>3_internal): Likewise.
2326 (boolc<mode>3_internal1): Likewise.
2327 (boolc<mode>3_internal2): Likewise.
2328 (boolcc<mode>3_internal1): Likewise.
2329 (boolcc<mode>3_internal2): Likewise.
2330 (eqv<mode>3_internal1): Likewise.
2331 (eqv<mode>3_internal2): Likewise.
2332 (one_cmpl1<mode>3_internal): Likewise.
2333
a86bc377 23342013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
2335
2336 * config/microblaze/microblaze.c (microblaze_expand_prologue):
2337 Rename flag_stack_usage to flag_stack_usage_info.
2338
c4fcbf4c 23392013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
2340
bc3d75b4 2341 * config/microblaze/sync.md: New file.
2342 * config/microblaze/microblaze.md: Include sync.md
2343 * config/microblaze/microblaze.c: Add print_operand 'y'.
2344 * config/microblaze/constraints.md: Add memory_contraint
2345 'Q' which is a single register.
c4fcbf4c 2346
7207a928 23472013-07-23 Eric Botcazou <ebotcazou@adacore.com>
2348
2349 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
2350
b766ccbc 23512013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
2352
2353 * reload.c (find_reloads): Exit loop once we find this operand
2354 cannot be reloaded somehow for this alternative.
2355
2356 * reload.c (find_reloads): Exit loop once we find a hard register.
2357
2358 * rtlanal.c (computed_jump_p): Exit loop once we find label
2359 reference is used.
2360
2361 * i386.c (ix86_pad_returns): Exit loop after setting replace.
2362
2363 * cfgloopmanip.c (remove_path): Exit loop after setting
2364 irred_invalidated.
2365
2366 * gensupport.c (subst_dup): Avoid loop if code is not
2367 MATCH_DUP nor MATCH_OP_DUP.
2368
c2d280c9 23692013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
2370
2371 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
2372
ec8b8726 23732013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2374
2375 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
2376 true for SP_REGNUM if mode == ptr_mode.
2377 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
2378 with value R0_REGNUM + 31.
2379
1ae2a9c0 23802013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2381
2382 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
2383 pad pointer-typed argument downward.
2384
b01b37f5 23852013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2386
2387 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
2388 and __ILP32__ when the ILP32 model is in use.
2389
011aed18 23902013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2391
2392 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
2393 (aarch64_load_symref_appropriately): In the case of
2394 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
2395 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
2396 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
2397 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
2398 if 'mode' doesn't equal to 'ptr_mode'.
2399 (aarch64_output_mi_thunk): Add an assertion on the alignment of
2400 'vcall_offset'; change to call aarch64_emit_move differently depending
2401 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
2402 to calculate the upper bound of 'vcall_offset'.
2403 (aarch64_cannot_force_const_mem): Change to also return true if
2404 mode != ptr_mode.
2405 (aarch64_legitimize_reload_address): In the case of large
2406 displacements, add new local variable 'xmode' and an assertion
2407 based on it; change to use 'xmode' to generate the new rtx and
2408 reload.
2409 (aarch64_asm_trampoline_template): Change to generate the template
2410 differently depending on TARGET_ILP32 or not; change to use
2411 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
2412 (aarch64_trampoline_size): Removed.
2413 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
2414 and replace immediate literals with it. Change to use 'ptr_mode'
2415 instead of 'DImode' and call convert_memory_address if the mode
2416 of 'fnaddr' doesn't equal to 'ptr_mode'.
2417 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
2418 to output symbol.
2419 (aarch64_elf_asm_destructor): Likewise.
2420 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
2421 on TARGET_ILP32 instead of aarch64_trampoline_size.
2422 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
2423 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
2424 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
2425 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
2426 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
2427 (storewb_pair<GPI:mode>_<P:mode>): ... this.
2428 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
2429 depending on the value of 'mode'.
2430 (add_losym_<mode>): New.
2431 (ldr_got_small_<mode>): New, based on ldr_got_small.
2432 (ldr_got_small): Remove.
2433 (ldr_got_small_sidi): New.
2434 * config/aarch64/iterators.md (P): New.
2435 (PTR): Change to 'ptr_mode' in the condition.
2436
bb009361 24372013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2438
2439 * config.gcc (aarch64*-*-*): Support --with-abi.
2440 (aarch64*-*-elf): Support --with-multilib-list.
2441 (aarch64*-*-linux*): Likewise.
2442 (supported_defaults): Add abi to aarch64*-*-*.
2443 * configure.ac: Mention AArch64 for --with-multilib-list.
2444 * configure: Re-generated.
2445 * config/aarch64/biarchilp32.h: New file.
2446 * config/aarch64/biarchlp64.h: New file.
2447 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
2448 (ABI_SPEC): Ditto.
2449 (MULTILIB_DEFAULTS): Ditto.
2450 (DRIVER_SELF_SPECS): Ditto.
2451 (ASM_SPEC): Update to also substitute -mabi.
2452 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
2453 file whose name depends on -mabi= and -mbig-endian.
2454 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
2455 TARGET_ILP32.
2456 (POINTER_SIZE): New define.
2457 (POINTERS_EXTEND_UNSIGNED): Ditto.
2458 (enum aarch64_abi_type): New enumeration tag.
2459 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
2460 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
2461 (TARGET_ILP32): New define.
2462 * config/aarch64/aarch64.opt (mabi): New.
2463 (aarch64_abi): New.
2464 (ilp32, lp64): New values for -mabi.
2465 * config/aarch64/t-aarch64 (comma): New define.
2466 (MULTILIB_OPTIONS): Ditto.
2467 (MULTILIB_DIRNAMES): Ditto.
2468 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
2469 * doc/invoke.texi: Document -mabi for AArch64.
2470
4bafdf2c 24712013-07-23 Georg-Johann Lay <avr@gjlay.de>
2472
2473 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
2474
b21ad496 24752013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2476 Anton Blanchard <anton@au1.ibm.com>
2477
2478 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
2479 endianness when selecting field to splat.
2480
109b4732 24812013-07-22 Eric Christopher <echristo@gmail.com>
2482
2483 * dwarf2out.c (die_odr_checksum): New function to use
2484 CHECKSUM_ macros and ULEB128 for DIE tag.
2485 (generate_type_signature): Use.
542c9bab 2486
191a0bb7 24872013-07-22 Eric Botcazou <ebotcazou@adacore.com>
2488
2489 * config.gcc (sparc*-*-*): Accept leon3 processor.
2490 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
2491 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
2492 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
2493 * config/sparc/sparc.opt (enum processor_type): Add leon3.
2494 (mfix-ut699): Adjust comment.
2495 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
2496 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
2497 (CPP_CPU_SPEC): Likewise.
2498 (ASM_CPU_SPEC): Likewise.
2499 * config/sparc/sparc.c (leon3_cost): New constant.
2500 (sparc_option_override): Add leon3 support.
2501 (mem_ref): New function.
2502 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
2503 (sparc_do_work_around_errata): Look into the instruction in the delay
2504 slot and adjust accordingly. Add fix for the data cache nullify issues
2505 of the UT699. Change insertion position for the NOP.
2506 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
2507 (leon3_load): New reservation.
2508 (leon_store): Bump latency to 2.
2509 (grfpu): New automaton.
2510 (grfpu_alu): New unit.
2511 (grfpu_ds): Likewise.
2512 (leon_fp_alu): Adjust.
2513 (leon_fp_mult): Delete.
2514 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
2515 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
2516 * config/sparc/sparc.md (cpu): Add leon3.
2517 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
2518 (swapsi): Likewise.
2519 (atomic_test_and_set): Likewise.
2520 (ldstub): Likewise.
2521
394563b3 25222013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
2523
2524 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
2525 default for R5900 targets.
2526 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
2527 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
2528 * config/mips/mips.c (mips_option_override): Report an error for
2529 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
2530 for -march=r5900 -mhard-float.
2531
b766ccbc 25322013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
73bd65ab 2533
2534 * df-problems.c (can_move_insns_across): Exit loop once we
2535 find a non-fixed, non-global register.
2536
2537 * ipa-pure-const.c (propagate_nothrow): Exit loop after
2538 setting can_throw.
2539
2540 * omega.c (omega_eliminate_red): Break after setting red_found.
2541 (omega_problem_has_red_equations): Similarly after setting found.
2542 (omega_query_variable): Similarly after setting coupled.
2543
4d334484 25442013-07-22 Marek Polacek <polacek@redhat.com>
2545
2546 * gimplify.c: Don't include gimple.h twice.
2547
7c36fe71 25482013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2549
2550 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
2551 instead of TARGET_THUMB1.
2552 (Pz): New constraint.
2553 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
2554 encodings.
2555 (compare_negsi_si): Likewise.
2556 (compare_addsi2_op0): Likewise.
2557 (compare_addsi2_op1): Likewise.
2558 (addsi3_carryin_<optab>): Likewise.
2559 (addsi3_carryin_alt2_<optab>): Likewise.
2560 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
2561 for arm_restrict_it.
2562 (subsi3_carryin): Likewise.
2563 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
2564 (minmax_arithsi): Disable for arm_restrict_it.
2565 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
2566 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
2567 (satsi_<SAT:code>_shift): Likewise.
2568 (arm_shiftsi3): Add alternative for 16-bit encoding.
2569 (arm32_movhf): Disable for arm_restrict_it.
2570 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
2571 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
2572
1aed5204 25732013-07-22 Sofiane Naci <sofiane.naci@arm.com>
2574
2575 * config/arm/arm.md (attribute "insn"): Delete.
2576 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
2577 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
2578 (not_shiftsi): Update for attribute change.
2579 (not_shiftsi_compare0): Likewise.
2580 (not_shiftsi_compare0_scratch): Likewise.
2581 (arm_one_cmplsi2): Likewise.
2582 (thumb1_one_cmplsi2): Likewise.
2583 (notsi_compare0): Likewise.
2584 (notsi_compare0_scratch): Likewise.
2585 (thumb1_movdi_insn): Likewise.
2586 (arm_movsi_insn): Likewise.
2587 (movhi_insn_arch4): Likewise.
2588 (movhi_bytes): Likewise.
2589 (arm_movqi_insn): Likewise.
2590 (thumb1_movqi_insn): Likewise.
2591 (arm32_movhf): Likewise.
2592 (thumb1_movhf): Likewise.
2593 (arm_movsf_soft_insn): Likewise.
2594 (thumb1_movsf_insn): Likewise.
2595 (thumb_movdf_insn): Likewise.
2596 (movsicc_insn): Likewise.
2597 (movsfcc_soft_insn): Likewise.
2598 (and_scc): Likewise.
2599 (cond_move): Likewise.
2600 (if_move_not): Likewise.
2601 (if_not_move): Likewise.
2602 (if_shift_move): Likewise.
2603 (if_move_shift): Likewise.
2604 (if_shift_shift): Likewise.
2605 (if_not_arith): Likewise.
2606 (if_arith_not): Likewise.
2607 (cond_move_not): Likewise.
2608 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
2609 (neon_mov<mode>): Likewise.
2610 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
2611 (thumb2_movsi_vfp): Likewise.
2612 (movsf_vfp): Likewise.
2613 (thumb2_movsf_vfp): Likewise.
5f7f04d8 2614 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
2615 change.
1aed5204 2616 (cortexa7_older_only): Likewise.
2617 (cortexa7_younger): Likewise.
2618 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
2619 (1020alu_shift_op): Likewise.
2620 (1020alu_shift_reg_op): Likewise.
2621 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
2622 (alu_shift_op): Likewise.
2623 (alu_shift_reg_op): Likewise.
2624 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
2625 (11_alu_shift_op): Likewise.
2626 (11_alu_shift_reg_op): Likewise.
2627 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
2628 (9_alu_shift_reg_op): Likewise.
5f7f04d8 2629 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
2630 change.
1aed5204 2631 (cortex_a15_alu_shift): Likewise.
2632 (cortex_a15_alu_shift_reg): Likewise.
2633 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
2634 (cortex_a5_alu_shift): Likewise.
5f7f04d8 2635 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
2636 change.
1aed5204 2637 (cortex_a53_alu_shift): Likewise.
5f7f04d8 2638 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
2639 change.
1aed5204 2640 (cortex_a7_alu_reg): Likewise.
2641 (cortex_a7_alu_shift): Likewise.
2642 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
2643 (cortex_a8_alu_shift): Likewise.
2644 (cortex_a8_alu_shift_reg): Likewise.
2645 (cortex_a8_mov): Likewise.
2646 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
2647 (cortex_a9_dp_shift): Likewise.
2648 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
2649 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
2650 (cortex_r4_mov): Likewise.
2651 (cortex_r4_alu_shift): Likewise.
2652 (cortex_r4_alu_shift_reg): Likewise.
2653 * config/arm/fa526.md (526_alu_op): Update for attribute change.
2654 (526_alu_shift_op): Likewise.
2655 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
2656 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
2657 (626te_alu_shift_op): Likewise.
2658 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
2659 (726te_alu_op): Likewise.
2660 (726te_alu_shift_op): Likewise.
2661 (726te_alu_shift_reg_op): Likewise.
2662 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
2663 (mp626_alu_shift_op): Likewise.
2664 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
2665 (pj4_alu_e1_conds): Likewise.
2666 (pj4_alu): Likewise.
2667 (pj4_alu_conds): Likewise.
2668 (pj4_shift): Likewise.
2669 (pj4_shift_conds): Likewise.
2670 (pj4_alu_shift): Likewise.
2671 (pj4_alu_shift_conds): Likewise.
2672
4369e6ad 26732013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2674
2675 * config/arm/predicates.md (shiftable_operator_strict_it):
2676 New predicate.
2677 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
2678 Disable cond_exec version for arm_restrict_it.
2679 (thumb2_smaxsi3): Convert to generate cond_exec.
2680 (thumb2_sminsi3): Likewise.
2681 (thumb32_umaxsi3): Likewise.
2682 (thumb2_uminsi3): Likewise.
2683 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
2684 (thumb2_neg_abssi2): Likewise.
2685 (thumb2_mov_scc): Add alternative for 16-bit encoding.
2686 (thumb2_movsicc_insn): Adjust alternatives.
2687 (thumb2_mov_negscc): Disable for arm_restrict_it.
2688 (thumb2_mov_negscc_strict_it): New pattern.
2689 (thumb2_mov_notscc_strict_it): New pattern.
2690 (thumb2_mov_notscc): Disable for arm_restrict_it.
2691 (thumb2_ior_scc): Likewise.
2692 (thumb2_ior_scc_strict_it): New pattern.
2693 (thumb2_cond_move): Adjust for arm_restrict_it.
2694 (thumb2_cond_arith): Disable for arm_restrict_it.
2695 (thumb2_cond_arith_strict_it): New pattern.
2696 (thumb2_cond_sub): Adjust for arm_restrict_it.
2697 (thumb2_movcond): Likewise.
2698 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
2699 (thumb2_zero_extendhisi2_v6): Likewise.
2700 (thumb2_zero_extendqisi2_v6): Likewise.
2701 (orsi_notsi_si): Likewise.
2702 (orsi_not_shiftsi_si): Likewise.
2703
4f3c1595 27042013-07-22 Georg-Johann Lay <avr@gjlay.de>
2705
2706 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
2707 instruction sequence is 1 byte shorter.
2708
896bc7cd 27092013-07-22 Uros Bizjak <ubizjak@gmail.com>
2710
2711 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
2712 it is not needed after split.
2713
0605a3f7 27142013-07-20 Iain Sandoe <iain@codesourcery.com>
2715
2716 PR target/51784
2717 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
2718 second label for nonlocal goto receivers. Don't output pic base labels
2719 unless we're producing PIC; mark that action unreachable().
2720 (ix86_save_reg): If the function contains a nonlocal label, save the
2721 PIC base reg.
2722 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
5f7f04d8 2723 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
0605a3f7 2724 (update_pic_label_number_if_needed): New.
2725 (machopic_output_function_base_name): Adjust for nonlocal receiver
2726 case.
2727 (machopic_should_output_picbase_label): New.
2728 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
2729 (nonlocal_goto_receiver): New insn and split.
2730
b9a85db4 27312013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
2732
2733 * config/aarch64/aarch64-builtins.c
2734 (aarch64_fold_builtin): Fold abs in all modes.
2735 * config/aarch64/aarch64-simd-builtins.def
2736 (abs): Enable for all modes.
2737 * config/aarch64/arm_neon.h
2738 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
2739 (vabs_f64): Add missing intrinsic.
2740
9bf7136b 27412013-07-19 Ian Bolton <ian.bolton@arm.com>
2742
2743 * config/aarch64/arm_neon.h (vabs_s64): New function
2744
fdbf5b19 27452013-07-19 Georg-Johann Lay <avr@gjlay.de>
2746
2747 PR target/57516
2748 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
2749 * config/avr/avr.md (adjust_len): Add `round'.
2750 * config/avr/avr-protos.h (avr_out_round): New prototype.
2751 (avr_out_plus): Add `out_label' argument.
2752 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
2753 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
2754 Handle the case where `insn' is just a pattern.
2755 (avr_out_bitop): Handle the case where `insn' is just a pattern.
2756 (avr_out_round): New function.
2757 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
2758
db58342c 27592013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
2760
2761 * config/microblaze/microblaze.c (microblaze_expand_prologue):
2762 Add check for flag_stack_usage to handle -fstack-usage support
2763
890f32a6 27642013-07-18 Pat Haugen <pthaugen@us.ibm.com>
2765
2766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
2767 interaction for new Power8 flags and VSX.
2768
c440293e 27692013-07-18 Sriraman Tallam <tmsriram@google.com>
2770
2771 PR middle-end/57698
2772 * tree-inline.c (expand_call_inline): Emit errors during
2773 early_inlining only if optimization is not turned on.
2774
ca4be2a0 27752013-07-18 David Malcolm <dmalcolm@redhat.com>
2776
2777 * passes.def: New.
2778
2779 * passes.c (init_optimization_passes): Move the construction of
2780 the pass hierarchy into a new passes.def file.
2781
2782 * Makefile.in (passes.o): Add dependency on passes.def.
2783
95c45f89 27842013-07-18 David Malcolm <dmalcolm@redhat.com>
2785
2786 * passes.c (init_optimization_passes): Introduce macros for
2787 constructing the tree of passes (INSERT_PASSES_AFTER,
2788 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
2789 TERMINATE_PASS_LIST).
2790
8033d76f 27912013-07-18 Vladimir Makarov <vmakarov@redhat.com>
2792 Wei Mi <wmi@google.com>
2793
2794 PR rtl-optimization/57878
2795 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
2796 top.
2797 (reload_pseudo_compare_func): Check nregs first for reload
2798 pseudos.
2799
69c20f28 28002013-07-18 David Malcolm <dmalcolm@redhat.com>
2801
2802 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
2803
b766ccbc 28042013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
d13600bf 2805
2806 * read-rtl.c (validate_const_int): Once an invalid character is
2807 seen, quit the loop.
2808
2809 * gengtype.c (write_roots): Similarly once we find the "deletable"
2810 or "if_marked" option.
2811
bebe9bbb 28122013-07-18 Sofiane Naci <sofiane.naci@arm.com>
2813
2814 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
2815 "xtab" and "sat". Move value "clz" from here to ...
2816 (attriubte "type"): ... here.
2817 (satsi_<SAT:code>): Delete "insn" attribute.
2818 (satsi_<SAT:code>_shift): Likewise.
2819 (arm_zero_extendqisi2addsi): Likewise.
2820 (arm_extendqisi2addsi): Likewise.
2821 (clzsi2): Update for attribute changes.
2822 (rbitsi2): Likewise.
5f7f04d8 2823 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
2824 attribute.
bebe9bbb 2825 (arm_usatsihi): Likewise.
2826 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
2827
6b6abc9c 28282013-07-18 Sofiane Naci <sofiane.naci@arm.com>
2829
2830 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
5f7f04d8 2831 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
2832 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
6b6abc9c 2833 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
2834 in alphabetical order.
2835 (attribute "core_cycles"): Update for attribute changes.
2836 (arm_addsi3): Likewise.
2837 (addsi3_compare0): Likewise.
2838 (addsi3_compare0_scratch): Likewise.
2839 (addsi3_compare_op1): Likewise.
2840 (addsi3_compare_op2): Likewise.
2841 (compare_addsi2_op0): Likewise.
2842 (compare_addsi2_op1): Likewise.
2843 (addsi3_carryin_shift_<optab>): Likewise.
2844 (subsi3_carryin_shift): Likewise.
2845 (rsbsi3_carryin_shift): Likewise.
2846 (arm_subsi3_insn): Likewise.
2847 (subsi3_compare0): Likewise.
2848 (subsi3_compare): Likewise.
2849 (arm_andsi3_insn): Likewise.
2850 (thumb1_andsi3_insn): Likewise.
2851 (andsi3_compare0): Likewise.
2852 (andsi3_compare0_scratch): Likewise.
2853 (zeroextractsi_compare0_scratch
2854 (andsi_not_shiftsi_si): Likewise.
2855 (iorsi3_insn): Likewise.
2856 (iorsi3_compare0): Likewise.
2857 (iorsi3_compare0_scratch): Likewise.
2858 (arm_xorsi3): Likewise.
2859 (thumb1_xorsi3_insn): Likewise.
2860 (xorsi3_compare0): Likewise.
2861 (xorsi3_compare0_scratch): Likewise.
2862 (satsi_<SAT:code>_shift): Likewise.
2863 (rrx): Likewise.
2864 (arm_shiftsi3): Likewise.
2865 (shiftsi3_compare0): Likewise.
2866 (not_shiftsi): Likewise.
2867 (not_shiftsi_compare0): Likewise.
2868 (not_shiftsi_compare0_scratch): Likewise.
2869 (arm_one_cmplsi2): Likewise.
2870 (thumb_one_complsi2): Likewise.
2871 (notsi_compare0): Likewise.
2872 (notsi_compare0_scratch): Likewise.
2873 (thumb1_zero_extendhisi2): Likewise.
2874 (arm_zero_extendhisi2): Likewise.
2875 (arm_zero_extendhisi2_v6): Likewise.
2876 (arm_zero_extendhisi2addsi): Likewise.
2877 (thumb1_zero_extendqisi2): Likewise.
2878 (thumb1_zero_extendqisi2_v6): Likewise.
2879 (arm_zero_extendqisi2): Likewise.
2880 (arm_zero_extendqisi2_v6): Likewise.
2881 (arm_zero_extendqisi2addsi): Likewise.
2882 (thumb1_extendhisi2): Likewise.
2883 (arm_extendhisi2): Likewise.
2884 (arm_extendhisi2_v6): Likewise.
2885 (arm_extendqisi): Likewise.
2886 (arm_extendqisi_v6): Likewise.
2887 (arm_extendqisi2addsi): Likewise.
2888 (thumb1_extendqisi2): Likewise.
2889 (thumb1_movdi_insn): Likewise.
2890 (arm_movsi_insn): Likewise.
2891 (movsi_compare0): Likewise.
2892 (movhi_insn_arch4): Likewise.
2893 (movhi_bytes): Likewise.
2894 (arm_movqi_insn): Likewise.
2895 (thumb1_movqi_insn): Likewise.
2896 (arm32_movhf): Likewise.
2897 (thumb1_movhf): Likewise.
2898 (arm_movsf_soft_insn): Likewise.
2899 (thumb1_movsf_insn): Likewise.
2900 (movdf_soft_insn): Likewise.
2901 (thumb_movdf_insn): Likewise.
2902 (arm_cmpsi_insn): Likewise.
2903 (cmpsi_shiftsi): Likewise.
2904 (cmpsi_shiftsi_swp): Likewise.
2905 (arm_cmpsi_negshiftsi_si): Likewise.
2906 (movsicc_insn): Likewise.
2907 (movsfcc_soft_insn): Likewise.
2908 (arith_shiftsi): Likewise.
2909 (arith_shiftsi_compare0
2910 (arith_shiftsi_compare0_scratch
2911 (sub_shiftsi): Likewise.
2912 (sub_shiftsi_compare0
2913 (sub_shiftsi_compare0_scratch
2914 (and_scc): Likewise.
2915 (cond_move): Likewise.
2916 (if_plus_move): Likewise.
2917 (if_move_plus): Likewise.
2918 (if_move_not): Likewise.
2919 (if_not_move): Likewise.
2920 (if_shift_move): Likewise.
2921 (if_move_shift): Likewise.
2922 (if_shift_shift): Likewise.
2923 (if_not_arith): Likewise.
2924 (if_arith_not): Likewise.
2925 (cond_move_not): Likewise.
2926 (thumb1_ashlsi3): Set type attribute.
2927 (thumb1_ashrsi3): Likewise.
2928 (thumb1_lshrsi3): Likewise.
2929 (thumb1_rotrsi3): Likewise.
2930 (shiftsi3_compare0_scratch): Likewise.
2931 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
2932 (neon_mov<mode>): Likewise.
5f7f04d8 2933 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
2934 attribute changes.
6b6abc9c 2935 (thumb2_movsi_insn): Likewise.
2936 (thumb2_cmpsi_neg_shiftsi): Likewise.
2937 (thumb2_extendqisi_v6): Likewise.
2938 (thumb2_zero_extendhisi2_v6): Likewise.
2939 (thumb2_zero_extendqisi2_v6): Likewise.
2940 (thumb2_shiftsi3_short): Likewise.
2941 (thumb2_addsi3_compare0_scratch): Likewise.
2942 (orsi_not_shiftsi_si): Likewise.
2943 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
2944 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
2945 changes.
2946 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
2947 (1020alu_shift_op): Likewise.
2948 (1020alu_shift_reg_op): Likewise.
2949 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
2950 (alu_shift_op): Likewise.
2951 (alu_shift_reg_op): Likewise.
2952 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
2953 (11_alu_shift_op): Likewise.
2954 (11_alu_shift_reg_op): Likewise.
2955 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
2956 (9_alu_shift_reg_op): Likewise.
5f7f04d8 2957 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
2958 changes.
6b6abc9c 2959 (cortex_a15_alu_shift): Likewise.
2960 (cortex_a15_alu_shift_reg): Likewise.
5f7f04d8 2961 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
2962 changes.
6b6abc9c 2963 (cortex_a5_alu_shift): Likewise.
2964 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
2965 changes.
2966 (cortex_a53_alu_shift): Likewise.
2967 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
2968 changes.
2969 (cortex_a7_alu_reg): Likewise.
2970 (cortex_a7_alu_shift): Likewise.
5f7f04d8 2971 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
2972 changes.
6b6abc9c 2973 (cortex_a8_alu_shift): Likewise.
2974 (cortex_a8_alu_shift_reg): Likewise.
2975 (cortex_a8_mov): Likewise.
2976 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
2977 (cortex_a9_dp_shift): Likewise.
5f7f04d8 2978 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
2979 changes.
2980 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
2981 changes.
6b6abc9c 2982 (cortex_r4_mov): Likewise.
2983 (cortex_r4_alu_shift): Likewise.
2984 (cortex_r4_alu_shift_reg): Likewise.
2985 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
2986 (526_alu_shift_op): Likewise.
2987 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
2988 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
2989 (626te_alu_shift_op): Likewise.
2990 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
2991 (726te_alu_op): Likewise.
2992 (726te_alu_shift_op): Likewise.
2993 (726te_alu_shift_reg_op): Likewise.
2994 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
2995 (mp626_alu_shift_op): Likewise.
2996 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
2997 (pj4_alu_e1_conds): Likewise.
2998 (pj4_alu): Likewise.
2999 (pj4_alu_conds): Likewise.
3000 (pj4_shift): Likewise.
3001 (pj4_shift_conds): Likewise.
3002 (pj4_alu_shift): Likewise.
3003 (pj4_alu_shift_conds): Likewise.
5f7f04d8 3004 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3005 changes.
6b6abc9c 3006 (cortexa7_older_only): Likewise.
3007 (cortexa7_younger): Likewise.
3008
80880eb6 30092013-07-18 David Malcolm <dmalcolm@redhat.com>
3010
3011 * ipa-pure-const.c (generate_summary): Rename to...
3012 (pure_const_generate_summary): ... this.
3013
79d68a42 30142013-07-17 Iain Sandoe <iain@codesourcery.com>
3015
3016 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3017
3eafc9ff 30182013-07-17 Yvan Roux <yvan.roux@linaro.org>
3019
3020 PR target/57909
3021 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3022 usage in HI mode.
3023
ee2dbc39 30242013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3025
3026 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3027 enabled without -march=zEC12.
3028 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3029 flags to be set.
3030
18af479d 30312013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3032
3033 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3034 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3035 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3036 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3037 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3038 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3039 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3040 ISA_HAS_FP4.
3041 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3042 and ISA_HAS_NMADD3_NMSUB3.
3043 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3044 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3045 (nmsub4<mode>, nmsub3<mode>): Likewise.
3046 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3047
57f12f14 30482013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3049
3050 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3051 TARGET_MIPS5400 checking.
3052
8e5b1909 30532013-07-16 Jakub Jelinek <jakub@redhat.com>
3054 Peter Bergner <bergner@vnet.ibm.com>
3055
3056 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3057 registers in the comment.
3058 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3059 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3060 rather than FIRST_PSEUDO_REGISTERS.
3061
30622013-07-16 Peter Bergner <bergner@vnet.ibm.com>
6e33a792 3063
3064 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3065 enable extra ISA flags with TARGET_HTM.
3066
512f12d6 30672013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3068
3069 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3070 Fix comment typos.
3071
9b44c85a 30722013-07-15 Cong Hou <congh@google.com>
3073
3074 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3075 in compare function for sorting.
3076
8e5b1909 30772013-07-15 Peter Bergner <bergner@vnet.ibm.com>
5088e479 3078
3079 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3080 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3081 * config/rs6000/rs6000.opt: Add -mhtm option.
3082 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3083 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3084 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3085 __HTM__ if the HTM instructions are available.
3086 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3087 htm_spr_reg_operand): New define_predicates.
3088 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3089 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3090 Include htm.md.
3091 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3092 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3093 HTM builtin functions.
3094 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3095 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3096 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3097 (rs6000_builtin_mask_calculate): Likewise.
3098 (rs6000_option_override_internal): Likewise.
3099 (bdesc_htm): Add new HTM builtin support.
3100 (htm_spr_num): New function.
3101 (htm_spr_regno): Likewise.
3102 (rs6000_htm_spr_icode): Likewise.
3103 (htm_expand_builtin): Likewise.
3104 (htm_init_builtins): Likewise.
3105 (rs6000_expand_builtin): Add support for HTM builtin functions.
3106 (rs6000_init_builtins): Likewise.
5f7f04d8 3107 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3108 option.
5088e479 3109 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3110 (TARGET_HTM, MASK_HTM): Define macros.
3111 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3112 (FIXED_REGISTERS): Likewise.
3113 (CALL_USED_REGISTERS): Likewise.
3114 (CALL_REALLY_USED_REGISTERS): Likewise.
3115 (REG_ALLOC_ORDER): Likewise.
3116 (enum reg_class): Likewise.
3117 (REG_CLASS_NAMES): Likewise.
3118 (REG_CLASS_CONTENTS): Likewise.
3119 (REGISTER_NAMES): Likewise.
3120 (ADDITIONAL_REGISTER_NAMES): Likewise.
3121 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3122 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3123 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3124 * config/rs6000/htm.md: New file.
3125 * config/rs6000/htmintrin.h: New file.
3126 * config/rs6000/htmxlintrin.h: New file.
3127
2c97ec73 31282013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
3129
3130 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3131 Define SYMBOL_TINY_GOT, update comment.
3132 * config/aarch64/aarch64.c
3133 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3134 (aarch64_expand_mov_immediate): Likewise.
3135 (aarch64_print_operand): Likewise.
3136 (aarch64_classify_symbol): Likewise.
3137 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3138 (ldr_got_tiny): Define.
3139
bffcae34 31402013-07-13 Tobias Grosser <tobias@grosser.es>
3141
3142 PR tree-optimization/54094
3143 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3144 scheduling dimension for the parallelism check from the polyhedral
3145 information in the AST.
3146 * graphite-dependences.c (carries_deps): Do not assume the schedule is
3147 in 2D + 1 form.
3148
84fcb947 31492013-07-13 Jason Merrill <jason@redhat.com>
3150
3151 * print-tree.c (debug_vec_tree): Use debug_raw.
3152 (debug_raw (vec<tree, va_gc> &)): New.
3153 (debug_raw (vec<tree, va_gc> *)): New.
3154 * tree.h: Declare them.
3155
abd03efe 31562013-07-13 Bin Cheng <bin.cheng@arm.com>
3157
3158 * ifcvt.c (ifcvt_after_combine): New static variable.
3159 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3160 for size.
3161 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
3162 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3163 rest_of_handle_if_after_reload): Pass new argument for if_convert.
3164
83773cef 31652013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
3166
3167 * config/mips/mips.c (mips_expand_call): Remove empty statement.
3168
0dc99c85 31692013-07-12 Michael Matz <matz@suse.de>
3170
3171 PR middle-end/55771
3172 * convert.c (convert_to_real): Reject non-float inner types.
3173
8458c9e9 31742013-07-12 Tejas Belagod <tejas.belagod@arm.com>
3175
3176 * config/aarch64/aarch64-protos.h
3177 (aarch64_simd_immediate_valid_for_move): Remove.
3178 * config/aarch64/aarch64.c (simd_immediate_info): New member.
3179 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3180 cases.
3181 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3182
68a45531 31832013-07-11 Steve Ellcey <sellcey@mips.com>
3184
3185 * config/mips/mips.c (mips_conditional_register_usage): Do not
3186 use t[0-7] registers in MIPS16 mode when optimizing for size.
3187
db56f021 31882013-07-11 Sriraman Tallam <tmsriram@google.com>
3189
3190 * config/i386/i386.c (dispatch_function_versions): Fix array
3191 indexing of function_version_info to match actual_versions.
3192
182e7ecd 31932013-07-11 Teresa Johnson <tejohnson@google.com>
3194
3195 * vec.h (struct va_gc): Move release out-of-line.
3196 (va_gc::release): Call ggc_free on released vec.
3197
fe14b54e 31982013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3199
3200 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3201 Require GOT register as additional operand in UNSPEC.
3202 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3203 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3204 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3205 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3206 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3207 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3208 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3209
248c10bf 32102013-07-11 Georg-Johann Lay <avr@gjlay.de>
3211
3212 PR target/57631
3213 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3214 name seen by assembler/linker rather if available.
3215
945153c1 32162013-07-11 Andreas Schwab <schwab@suse.de>
3217
3218 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
3219
a519b04a 32202013-07-10 Vladimir Makarov <vmakarov@redhat.com>
3221
5f7f04d8 3222 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
a519b04a 3223
f8e3aa87 32242013-07-10 Joseph Myers <joseph@codesourcery.com>
3225
d4173f0f 3226 * doc/tm.texi.in: Move hook documentation to ....
3227 * target.def: ... here.
3228
f8e3aa87 3229 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
3230 text on @hook line.
3231 * doc/tm.texi: Regenerate.
3232
839f2f70 32332013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
3234
3235 PR c++/57869
3236 * doc/invoke.texi: Document -Wconditionally-supported.
3237
cde06e71 32382013-07-10 Georg-Johann Lay <avr@gjlay.de>
3239
3240 PR target/57844
3241 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
3242 of my_fp.
3243
d0c1e059 32442013-07-10 Georg-Johann Lay <avr@gjlay.de>
3245
3246 PR target/57506
3247 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
3248 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
3249 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
3250 Remove duplicate devices.
3251 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
3252 * config/avr/t-multilib: Regenerate.
3253 * config/avr/avr-tables.opt: Regenerate.
3254 * doc/avr-mmcu.texi: Regenerate.
3255
636ded30 32562013-07-10 Georg-Johann Lay <avr@gjlay.de>
3257
3258 PR target/56987
3259 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
3260
0060eac4 32612013-07-10 Graham Stott <graham.stott@btinternet.com>
73f353d0 3262
0060eac4 3263 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
3264 the cost of MULT when optimizing for size.
3265
42dda6b1 32662013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3267
3268 * config/cr16/cr16-protos.h: Don't include target.h.
3269
d738e1ee 32702013-07-09 Joseph Myers <joseph@codesourcery.com>
3271
3272 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
3273 adjust register size for TDmode and TFmode for VSX registers.
3274
15c57ef1 32752013-07-08 Kai Tietz <ktietz@redhat.com>
3276
3277 PR target/56892
3278 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
3279 hook_bool_const_tree_true.
3280
6a2469fe 32812013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3282
3283 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
3284 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
3285 * config/s390/s390.md: Define FPR*_REGNUM constants.
3286 Fix FPR2_REGNUM constant (18 -> 17).
3287 ("*trunc<BFP:mode><DFP_ALL:mode>2")
3288 ("*trunc<DFP_ALL:mode><BFP:mode>2")
3289 ("trunc<BFP:mode><DFP_ALL:mode>2")
3290 ("trunc<DFP_ALL:mode><BFP:mode>2")
3291 ("*extend<BFP:mode><DFP_ALL:mode>2")
3292 ("*extend<DFP_ALL:mode><BFP:mode>2")
3293 ("extend<BFP:mode><DFP_ALL:mode>2")
3294 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
3295 FPR4_REGNUM.
3296
6b77556b 32972013-07-08 Graham Stott <graham.stott@btinternet.com>
6b77556b 3298
73f353d0 3299 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
562fcf12 3300
29439367 33012013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3302
3303 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
3304 and cfun_fpr_bit_p to cfun_fpr_save_p.
3305 (s390_frame_area, s390_register_info, s390_frame_info)
3306 (s390_emit_prologue, s390_emit_epilogue)
3307 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
3308 register numbers.
3309 * config/s390/s390.h: Define *_REGNUM macros for floating point
3310 register numbers.
3311
9c9668f3 33122013-07-08 Eric Botcazou <ebotcazou@adacore.com>
3313
3314 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
3315
ed3a8d75 33162013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
3317
3318 PR rtl-optimization/57786
3319 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
3320 and break out of the loop when it is set to false.
3321
ec7f9058 33222013-07-08 Jakub Jelinek <jakub@redhat.com>
3323
1bf020d9 3324 PR target/57819
3325 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
3326 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
3327 (const_int 63)) 0)).
3328 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
3329 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
3330 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
3331
ec7f9058 3332 PR rtl-optimization/57829
3333 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
3334 mask bits outside of mode are just sign-extension from mode to HWI.
3335
7645814e 33362013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
3337
3338 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
3339 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
3340 adjust_address instead of change_address to keep info about alignment.
3341 (emit_strmov): Remove.
3342 (emit_memmov): New function.
3343 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
3344 (expand_movmem_epilogue): Likewise and return updated rtx for
3345 destination.
3346 (expand_constant_movmem_prologue): Likewise and return updated rtx for
3347 destination and source.
3348 (decide_alignment): Refactor, handle vector_loop.
3349 (ix86_expand_movmem): Likewise.
3350 (ix86_expand_setmem): Likewise.
3351 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
3352
9bfede72 33532013-07-07 Uros Bizjak <ubizjak@gmail.com>
3354
3355 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
3356 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
3357
c358a059 33582013-07-06 Uros Bizjak <ubizjak@gmail.com>
3359
3360 * config/i386/sse.md (sse_movlhps): Change alternative 3
3361 of operand 2 to "m".
3362
33632013-07-06 Uros Bizjak <ubizjak@gmail.com>
3364
3365 PR target/57807
3366 * config/i386/sse.md (iptr): New mode attribute.
3367 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
3368 (<sse>_vm<plusminus_insn><mode>3): Ditto.
3369 (<sse>_vmmul<mode>3): Ditto.
3370 (<sse>_vmdiv<mode>3): Ditto.
3371 (sse_vmrcpv4sf2): Ditto.
3372 (<sse>_vmsqrt<mode>2): Ditto.
3373 (sse_vmrsqrtv4sf2): Ditto.
3374 (<sse>_vm<code><mode>3): Ditto.
3375 (avx_vmcmp<mode>3): Ditto.
3376 (<sse>_vmmaskcmp<mode>3): Ditto.
3377 (<sse>_comi): Ditto.
3378 (<sse>_ucomi): Ditto.
3379 (*xop_vmfrcz_<mode>): Ditto.
3380 (*fmai_fmadd_<mode>): Ditto.
3381 (*fmai_fmsub_<mode>): Ditto.
3382 (*fmai_fnmadd_<mode>): Ditto.
3383 (*fmai_fnmsub_<mode>): Ditto.
3384 (*fma4i_vmfmadd_<mode>): Ditto.
3385 (*fma4i_vmfmsub_<mode>): Ditto.
3386 (*fma4i_vmfnmadd_<mode>): Ditto.
3387 (*fma4i_vmfnmsub_<mode>): Ditto.
3388 (*xop_vmfrcz_<mode>): Ditto.
3389 (sse_cvtps2pi): Ditto.
3390 (sse_cvttps2pi): Ditto.
3391 (sse_cvtss2si): Ditto.
3392 (sse_cvtss2si_2): Ditto.
3393 (sse_cvtss2siq_2): Ditto.
3394 (sse_cvttss2si): Ditto.
3395 (sse_cvttss2siq): Ditto.
3396 (sse_cvtsd2si): Ditto.
3397 (sse_cvtsd2si_2): Ditto.
3398 (sse_cvtsd2siq_2): Ditto.
3399 (sse_cvttsd2si): Ditto.
3400 (sse_cvttsd2siq): Ditto.
3401 (sse_cvtsd2ss): Ditto.
3402 (sse_cvtss2sd): Ditto.
3403 (avx2_pbroadcast<mode>): Ditto.
3404 (avx2_pbroadcast<mode>_1): Ditto.
3405 (*avx_vperm_broadcast_v4sf): Ditto.
3406
3407 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
3408 (sse_movlhps): Ditto.
3409 (sse_storehps): Ditto.
3410 (sse_loadhps): Ditto.
3411 (sse_storelps): Ditto.
3412 (sse_loadlps): Ditto.
3413 (*vec_concatv4sf): Ditto.
3414 (*vec_interleave_highv2df): Ditto.
3415 (*vec_interleave_lowv2df): Ditto.
3416 (*vec_extractv2df_1_sse): Ditto.
3417 (*vec_extractv2df_0_sse): Ditto.
3418 (sse2_storelpd): Ditto.
3419 (sse2_loadlpd): Ditto.
3420 (sse2_movsd): Ditto.
3421 (*vec_concatv4si): Ditto.
3422 (vec_concatv2di): Ditto.
3423
3424 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
3425 for Intel asm dialect.
3426 (mmx_punpcklwd): Ditto.
3427 (mmx_punpckldq): Ditto.
3428
3429 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
3430 for intel assembler dialect.
3431
64b5be08 34322013-07-06 Jakub Jelinek <jakub@redhat.com>
3433
3434 PR target/29776
3435 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
3436 for BUILT_IN_C{LZ,LRSB}*.
3437 * tree.h (CASE_INT_FN): Add FN##IMAX case.
3438 * tree-vrp.c (extract_range_basic): Handle
3439 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
3440 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
3441 fall thru to code calling set_value*.
3442 * builtins.c (expand_builtin): Remove *IMAX cases.
3443 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
3444 if width is bigger than 2*HWI.
3445
1f3a048a 34462013-07-05 Vladimir Makarov <vmakarov@redhat.com>
3447
3448 PR rtl-optimization/55342
3449 * lra-int.h (lra_subreg_reload_pseudos): New.
3450 * lra.c: Add undoing optional reloads to the block diagram.
3451 (lra_subreg_reload_pseudos): New.
3452 (lra_optional_reload_pseudos): Change comments.
3453 (lra): Init and clear lra_subreg_reload_pseudos. Clear
3454 lra_optional_reload_pseudos after undo transformations.
3455 * lra-assigns.c (pseudo_prefix_title): New.
3456 (lra_setup_reg_renumber): Use it.
3457 (spill_for): Ditto. Check subreg reload pseudos too.
3458 (assign_by_spills): Consider subreg reload pseudos too.
3459 * lra-constraints.c (simplify_operand_subreg): Use
3460 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
3461 (curr_insn_transform): Recognize and do optional reloads.
3462 (undo_optional_reloads): New.
3463 (lra_undo_inheritance): Call undo_optional_reloads.
3464
4d6b2b7e 34652013-07-05 Thomas Quinot <quinot@adacore.com>
3466
3467 * tree-complex.c (expand_complex_operations_1): Fix typo.
3468
61d9499e 34692013-07-04 Tejas Belagod <tejas.belagod@arm.com>
3470
3471 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
3472 (tune_params): New member 'const vec_costs'.
3473 * config/aarch64/aarch64.c (generic_vector_cost): New.
3474 (generic_tunings): New member 'generic_vector_cost'.
3475 (aarch64_builtin_vectorization_cost): New.
3476 (aarch64_add_stmt_cost): New.
3477 (TARGET_VECTORIZE_ADD_STMT_COST): New.
3478 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
3479
f4d378a8 34802013-07-03 Jakub Jelinek <jakub@redhat.com>
3481
3482 PR target/57777
3483 * config/i386/predicates.md (vsib_address_operand): Disallow
3484 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
3485
4598ade9 34862013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
3487
3488 PR middle-end/55030
3489 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
3490 expand_builtin_setjmp_receiver.
3491 (expand_label): Adjust, call expand_builtin_setjmp_receiver
3492 with NULL for the label parameter.
3493 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
3494 the frame-pointer. Adjust comments.
3495 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
3496 only if LABEL is non-NULL.
3497
04a3900d 34982013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
3499
3500 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
3501 (ARM_ABI_AAPCS64): Ditto.
3502 (arm_abi): Ditto.
3503 (ARM_DEFAULT_ABI): Ditto.
3504
88c9a347 35052013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3506
3507 * config/aarch64/aarch64-builtins.c
3508 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
3509 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
3510 (st1): Likewise.
3511 * config/aarch64/aarch64-simd.md
3512 (aarch64_ld1<VALL:mode>): New.
3513 (aarch64_st1<VALL:mode>): Likewise.
3514 * config/aarch64/arm_neon.h
3515 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
3516
2ce8ff38 35172013-07-02 Sriraman Tallam <tmsriram@google.com>
3518
3af61095 3519 * config/i386/i386.c (gate_insert_vzeroupper): Check if
3520 target ISA is AVX.
2ce8ff38 3521 (ix86_option_override_internal):Turn on all -mavx target flags by
3522 default as they are dependent on AVX anyway.
3523
1bfbf223 35242013-07-02 Cary Coutant <ccoutant@google.com>
3525
3526 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
3527 deterministic hash.
3528 (loc_checksum_ordered): Likewise.
3529 (hash_loc_operands): Remove inline keyword.
3530
1d62df1c 35312013-07-02 Jakub Jelinek <jakub@redhat.com>
3532
3533 PR tree-optimization/57741
3534 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
3535 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
3536 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
3537 Allow REAL_CST step_exprs if flag_associative_math.
3538 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
3539
1648ecaa 35402013-07-02 Ian Bolton <ian.bolton@arm.com>
3541
3af61095 3542 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
1648ecaa 3543
0e14bc84 35442013-07-02 Ian Bolton <ian.bolton@arm.com>
3545
3546 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
3547
29e234a3 35482013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3549
3550 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
3551 encoding.
3552 (iorsi3_insn): Likewise.
3553 (arm_xorsi3): Likewise.
3554
7cb6c048 35552013-07-01 Sofiane Naci <sofiane.naci@arm.com>
3556
3557 * arm.md (attribute "wtype"): Delete. Move attribute values from here
3558 to ...
3559 (attribute "type"): ... here, and prefix with "wmmx_".
3560 (attribute "core_cycles"): Update for attribute changes.
3561 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
3562 (tbcstv4hi): Likewise.
3563 (tbcstv2si): Likewise.
3564 (iwmmxt_iordi3): Likewise.
3565 (iwmmxt_xordi3): Likewise.
3566 (iwmmxt_anddi3): Likewise.
3567 (iwmmxt_nanddi3): Likewise.
3568 (iwmmxt_arm_movdi): Likewise.
3569 (iwmmxt_movsi_insn): Likewise.
3570 (mov<mode>_internal): Likewise.
3571 (and<mode>3_iwmmxt): Likewise.
3572 (ior<mode>3_iwmmxt): Likewise.
3573 (xor<mode>3_iwmmxt): Likewise.
3574 (add<mode>3_iwmmxt): Likewise.
3575 (ssaddv8qi3): Likewise.
3576 (ssaddv4hi3): Likewise.
3577 (ssaddv2si3): Likewise.
3578 (usaddv8qi3): Likewise.
3579 (usaddv4hi3): Likewise.
3580 (usaddv2si3): Likewise.
3581 (sub<mode>3_iwmmxt): Likewise.
3582 (sssubv8qi3): Likewise.
3583 (sssubv4hi3): Likewise.
3584 (sssubv2si3): Likewise.
3585 (ussubv8qi3): Likewise.
3586 (ussubv4hi3): Likewise.
3587 (ussubv2si3): Likewise.
3588 (mulv4hi3_iwmmxt): Likewise.
3589 (smulv4hi3_highpart): Likewise.
3590 (umulv4hi3_highpart): Likewise.
3591 (iwmmxt_wmacs): Likewise.
3592 (iwmmxt_wmacsz): Likewise.
3593 (iwmmxt_wmacu): Likewise.
3594 (iwmmxt_wmacuz): Likewise.
3595 (iwmmxt_clrdi): Likewise.
3596 (iwmmxt_clrv8qi): Likewise.
3597 (iwmmxt_clr4hi): Likewise.
3598 (iwmmxt_clr2si): Likewise.
3599 (iwmmxt_uavgrndv8qi3): Likewise.
3600 (iwmmxt_uavgrndv4hi3): Likewise.
3601 (iwmmxt_uavgv8qi3): Likewise.
3602 (iwmmxt_uavgv4hi3): Likewise.
3603 (iwmmxt_tinsrb): Likewise.
3604 (iwmmxt_tinsrh): Likewise.
3605 (iwmmxt_tinsrw): Likewise.
3606 (iwmmxt_textrmub): Likewise.
3607 (iwmmxt_textrmsb): Likewise.
3608 (iwmmxt_textrmuh): Likewise.
3609 (iwmmxt_textrmsh): Likewise.
3610 (iwmmxt_textrmw): Likewise.
3611 (iwmxxt_wshufh): Likewise.
3612 (eqv8qi3): Likewise.
3613 (eqv4hi3): Likewise.
3614 (eqv2si3): Likewise.
3615 (gtuv8qi3): Likewise.
3616 (gtuv4hi3): Likewise.
3617 (gtuv2si3): Likewise.
3618 (gtv8qi3): Likewise.
3619 (gtv4hi3): Likewise.
3620 (gtv2si3): Likewise.
3621 (smax<mode>3_iwmmxt): Likewise.
3622 (umax<mode>3_iwmmxt): Likewise.
3623 (smin<mode>3_iwmmxt): Likewise.
3624 (umin<mode>3_iwmmxt): Likewise.
3625 (iwmmxt_wpackhss): Likewise.
3626 (iwmmxt_wpackwss): Likewise.
3627 (iwmmxt_wpackdss): Likewise.
3628 (iwmmxt_wpackhus): Likewise.
3629 (iwmmxt_wpackwus): Likewise.
3630 (iwmmxt_wpackdus): Likewise.
3631 (iwmmxt_wunpckihb): Likewise.
3632 (iwmmxt_wunpckihh): Likewise.
3633 (iwmmxt_wunpckihw): Likewise.
3634 (iwmmxt_wunpckilb): Likewise.
3635 (iwmmxt_wunpckilh): Likewise.
3636 (iwmmxt_wunpckilw): Likewise.
3637 (iwmmxt_wunpckehub): Likewise.
3638 (iwmmxt_wunpckehuh): Likewise.
3639 (iwmmxt_wunpckehuw): Likewise.
3640 (iwmmxt_wunpckehsb): Likewise.
3641 (iwmmxt_wunpckehsh): Likewise.
3642 (iwmmxt_wunpckehsw): Likewise.
3643 (iwmmxt_wunpckelub): Likewise.
3644 (iwmmxt_wunpckeluh): Likewise.
3645 (iwmmxt_wunpckeluw): Likewise.
3646 (iwmmxt_wunpckelsb): Likewise.
3647 (iwmmxt_wunpckelsh): Likewise.
3648 (iwmmxt_wunpckelsw): Likewise.
3649 (ror<mode>3): Likewise.
3650 (ashr<mode>3_iwmmxt): Likewise.
3651 (lshr<mode>3_iwmmxt): Likewise.
3652 (ashl<mode>3_iwmmxt): Likewise.
3653 (ror<mode>3_di): Likewise.
3654 (ashr<mode>3_di): Likewise.
3655 (lshr<mode>3_di): Likewise.
3656 (ashl<mode>3_di): Likewise.
3657 (iwmmxt_wmadds): Likewise.
3658 (iwmmxt_wmaddu): Likewise.
3659 (iwmmxt_tmia): Likewise.
3660 (iwmmxt_tmiaph): Likewise.
3661 (iwmmxt_tmiabb): Likewise.
3662 (iwmmxt_tmiatb): Likewise.
3663 (iwmmxt_tmiabt): Likewise.
3664 (iwmmxt_tmiatt): Likewise.
3665 (iwmmxt_tmovmskb): Likewise.
3666 (iwmmxt_tmovmskh): Likewise.
3667 (iwmmxt_tmovmskw): Likewise.
3668 (iwmmxt_waccb): Likewise.
3669 (iwmmxt_wacch): Likewise.
3670 (iwmmxt_waccw): Likewise.
3671 (iwmmxt_waligni): Likewise.
3672 (iwmmxt_walignr): Likewise.
3673 (iwmmxt_walignr0): Likewise.
3674 (iwmmxt_walignr1): Likewise.
3675 (iwmmxt_walignr2): Likewise.
3676 (iwmmxt_walignr3): Likewise.
3677 (iwmmxt_wsadb): Likewise.
3678 (iwmmxt_wsadh): Likewise.
3679 (iwmmxt_wsadbz): Likewise.
3680 (iwmmxt_wsadhz): Likewise.
3681 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
3682 (iwmmxt_wabsdiffb): Likewise.
3683 (iwmmxt_wabsdiffh): Likewise.
3684 (iwmmxt_wabsdiffw): Likewise.
3685 (iwmmxt_waddsubhx): Likewise
3686 (iwmmxt_wsubaddhx): Likewise.
3687 (addc<mode>3): Likewise.
3688 (iwmmxt_avg4): Likewise.
3689 (iwmmxt_avg4r): Likewise.
3690 (iwmmxt_wmaddsx): Likewise.
3691 (iwmmxt_wmaddux): Likewise.
3692 (iwmmxt_wmaddsn): Likewise.
3693 (iwmmxt_wmaddun): Likewise.
3694 (iwmmxt_wmulwsm): Likewise.
3695 (iwmmxt_wmulwum): Likewise.
3696 (iwmmxt_wmulsmr): Likewise.
3697 (iwmmxt_wmulumr): Likewise.
3698 (iwmmxt_wmulwsmr): Likewise.
3699 (iwmmxt_wmulwumr): Likewise.
3700 (iwmmxt_wmulwl): Likewise.
3701 (iwmmxt_wqmulm): Likewise.
3702 (iwmmxt_wqmulwm): Likewise.
3703 (iwmmxt_wqmulmr): Likewise.
3704 (iwmmxt_wqmulwmr): Likewise.
3705 (iwmmxt_waddbhusm): Likewise.
3706 (iwmmxt_waddbhusl): Likewise.
3707 (iwmmxt_wqmiabb): Likewise.
3708 (iwmmxt_wqmiabt): Likewise.
3709 (iwmmxt_wqmiatb): Likewise.
3710 (iwmmxt_wqmiatt): Likewise.
3711 (iwmmxt_wqmiabbn): Likewise.
3712 (iwmmxt_wqmiabtn): Likewise.
3713 (iwmmxt_wqmiatbn): Likewise.
3714 (iwmmxt_wqmiattn): Likewise.
3715 (iwmmxt_wmiabb): Likewise.
3716 (iwmmxt_wmiabt): Likewise.
3717 (iwmmxt_wmiatb): Likewise.
3718 (iwmmxt_wmiatt): Likewise.
3719 (iwmmxt_wmiabbn): Likewise.
3720 (iwmmxt_wmiabtn): Likewise.
3721 (iwmmxt_wmiatbn): Likewise.
3722 (iwmmxt_wmiattn): Likewise.
3723 (iwmmxt_wmiawbb): Likewise.
3724 (iwmmxt_wmiawbt): Likewise.
3725 (iwmmxt_wmiawtb): Likewise.
3726 (iwmmxt_wmiawtt): Likewise.
3727 (iwmmxt_wmiawbbn): Likewise.
3728 (iwmmxt_wmiawbtn): Likewise.
3729 (iwmmxt_wmiawtbn): Likewise.
3730 (iwmmxt_wmiawttn): Likewise.
3731 (iwmmxt_wmerge): Likewise.
3732 (iwmmxt_tandc<mode>3): Likewise.
3733 (iwmmxt_torc<mode>3): Likewise.
3734 (iwmmxt_torvsc<mode>3): Likewise.
3735 (iwmmxt_textrc<mode>3): Likewise.
3736 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
3737 (wmmxt_pack): Likewise.
3738 (wmmxt_mult_c1): Likewise.
3739 (wmmxt_mult_c2): Likewise.
3740 (wmmxt_alu_c1): Likewise.
3741 (wmmxt_alu_c2): Likewise.
3742 (wmmxt_alu_c3): Likewise.
3743 (wmmxt_transfer_c1): Likewise.
3744 (wmmxt_transfer_c2): Likewise.
3745 (wmmxt_transfer_c3): Likewise.
3746 (marvell_f_iwmmxt_wstr): Likewise.
3747 (marvell_f_iwmmxt_wldr): Likewise.
3748
801070f5 37492013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
3750
3af61095 3751 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
801070f5 3752
8f67e177 37532013-06-28 Vladimir Makarov <vmakarov@redhat.com>
3754
3755 Revert:
3756 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
3757 * lra-constraints.c (need_for_split_p): Check call used hard regs
3758 living through calls.
3759
3760 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
3761 call used regs for call insn.
3762
df92075b 37632013-06-28 Jakub Jelinek <jakub@redhat.com>
3764
3765 PR target/57736
3af61095 3766 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
3767 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
df92075b 3768
936c3081 37692013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
3770
3771 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
8f67e177 3772
0157439c 37732013-06-28 Vladimir Makarov <vmakarov@redhat.com>
3774
3775 * lra-constraints.c (need_for_split_p): Check call used hard regs
3776 living through calls.
3777
53ee13c3 37782013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
3779
3780 PR target/57744
3781 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
3782 to tie with any other modes. Eliminate Altivec vector mode tests,
3783 since these are a subset of ALTIVEC or VSX vector modes. Simplify
3784 code, to return 0 if testing MODE2 for a condition, if we've
3785 already tested MODE1 for the same condition.
3786
beac24f9 37872013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
3788
3789 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
3790 layout.
3791
586bb065 37922013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
3793
3794 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3795 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
3796
8fd61aba 37972013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
3798
3af61095 3799 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
3800 Define.
8fd61aba 3801 (aarch64_symbolic_constant_p): Remove.
3802 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
3803 static. Fix line length and white space.
3804 (aarch64_symbolic_constant_p): Remove.
3805 * config/aarch64/predicates.md (aarch64_valid_symref):
3806 Use aarch64_classify_symbol_expression.
3807
fd711051 38082013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3809
3810 * config/arm/constraints.md (Ts): New constraint.
3811 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
3812 16-bit encodings.
3813 (compare_scc): Use "Ts" constraint for operand 0.
3814 (ior_scc_scc): Likewise.
3815 (and_scc_scc): Likewise.
3816 (and_scc_scc_nodom): Likewise.
3817 (ior_scc_scc_cmp): Likewise for operand 7.
3818 (and_scc_scc_cmp): Likewise.
3819 * config/arm/thumb2.md (thumb2_movsi_insn):
3820 Add alternatives for 16-bit encodings.
3821 (thumb2_movhi_insn): Likewise.
3822 (thumb2_movsicc_insn): Likewise.
3823 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
3824 (thumb2_negscc): Use "Ts" constraint.
3825 Move mvn instruction outside cond_exec block.
3826 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
3827 for 16-bit encodings.
3828
d952d547 38292013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3830
3831 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
3832 encoding.
3833 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
3834 (mulsi3subsi): Likewise.
3835 (mulsidi3adddi): Likewise.
3836 (mulsidi3_v6): Likewise.
3837 (umulsidi3_v6): Likewise.
3838 (umulsidi3adddi_v6): Likewise.
3839 (smulsi3_highpart_v6): Likewise.
3840 (umulsi3_highpart_v6): Likewise.
3841 (mulhisi3tb): Likewise.
3842 (mulhisi3bt): Likewise.
3843 (mulhisi3tt): Likewise.
3844 (maddhisi4): Likewise.
3845 (maddhisi4tb): Likewise.
3846 (maddhisi4tt): Likewise.
3847 (maddhidi4): Likewise.
3848 (maddhidi4tb): Likewise.
3849 (maddhidi4tt): Likewise.
3850 (zeroextractsi_compare0_scratch): Likewise.
3851 (insv_zero): Likewise.
3852 (insv_t2): Likewise.
3853 (anddi_notzesidi_di): Likewise.
3854 (anddi_notsesidi_di): Likewise.
3855 (andsi_notsi_si): Likewise.
3856 (iordi_zesidi_di): Likewise.
3857 (xordi_zesidi_di): Likewise.
3858 (andsi_iorsi3_notsi): Likewise.
3859 (smax_0): Likewise.
3860 (smax_m1): Likewise.
3861 (smin_0): Likewise.
3862 (not_shiftsi): Likewise.
3863 (unaligned_loadsi): Likewise.
3864 (unaligned_loadhis): Likewise.
3865 (unaligned_loadhiu): Likewise.
3866 (unaligned_storesi): Likewise.
3867 (unaligned_storehi): Likewise.
3868 (extv_reg): Likewise.
3869 (extzv_t2): Likewise.
3870 (divsi3): Likewise.
3871 (udivsi3): Likewise.
3872 (arm_zero_extendhisi2addsi): Likewise.
3873 (arm_zero_extendqisi2addsi): Likewise.
3874 (compareqi_eq0): Likewise.
3875 (arm_extendhisi2_v6): Likewise.
3876 (arm_extendqisi2addsi): Likewise.
3877 (arm_movt): Likewise.
3878 (thumb2_ldrd): Likewise.
3879 (thumb2_ldrd_base): Likewise.
3880 (thumb2_ldrd_base_neg): Likewise.
3881 (thumb2_strd): Likewise.
3882 (thumb2_strd_base): Likewise.
3883 (thumb2_strd_base_neg): Likewise.
3884 (arm_negsi2): Add alternative for 16-bit encoding.
3885 (arm_one_cmplsi2): Likewise.
3886
4df97ffc 38872013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3888
3889 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
3890 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
3891 (movdfcc): Likewise.
3892 * config/arm/vfp.md (*thumb2_movsf_vfp):
3893 Disable predication for arm_restrict_it.
3894 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
3895 (*thumb2_movdfcc_vfp): Likewise.
3896 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
3897 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
3898 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
3899 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
3900 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
3901 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
3902 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
3903 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
3904 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
3905 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
3906 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
3907 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
3908 Disable predication for arm_restrict_it.
3909
556f149b 39102013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
3911
3912 * config/i386/bmiintrin.h (_bextr_u32): New.
3913 (_bextr_u64): Ditto.
3914
c789b852 39152013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
3916
3917 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
3918 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
3919 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
3920 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
3921 * config/mips/n32-elf.h: ...this new file.
3922
3b76b707 39232013-06-27 Marc Glisse <marc.glisse@inria.fr>
3924
3925 PR target/57224
3926 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
3927 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
3928
351a64c5 39292013-06-27 Catherine Moore <clm@codesourcery.com>
3930
3931 * config/mips/mips-tables.opt: Regenerate.
3932 * config/mips/mips-cpus.def: Add m14ke and m14kec.
3af61095 3933 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
351a64c5 3934 * doc/invoke.texi: Add -m14kc.
556f149b 3935
b8f56e39 39362013-06-27 Jakub Jelinek <jakub@redhat.com>
3937
fdddc582 3938 PR target/57623
3939 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
3940 constraints of operand 1 and 2.
3941
b8f56e39 3942 PR target/57623
3943 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
3944 to match RTL canonicalization. Swap predicates and
3945 constraints of operand 1 and 2.
3946
bf63c98f 39472013-06-27 Vladimir Makarov <vmakarov@redhat.com>
3948
3af61095 3949 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
3950 Process OP_INOUT regs for splitting too.
bf63c98f 3951
5c90e7b6 39522013-06-27 Jakub Jelinek <jakub@redhat.com>
3953
3954 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
3955 decl before the loop, initialize to NULL.
3956 (vectorizable_load): Initialize ptr_incr to NULL.
3957
3c0fe71b 39582013-06-27 Martin Jambor <mjambor@suse.cz>
3959
3960 PR lto/57208
3961 * ipa-ref.h (ipa_maybe_record_reference): Declare.
3962 * ipa-ref.c (ipa_maybe_record_reference): New function.
3963 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
3964 * ipa-cp.c (create_specialized_node): Record potential references from
3965 aggvals.
3966 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
3967
6b765f96 39682013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
3969
3970 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
3971 parameter 'mode' of type 'enum machine_mode mode'; change to pass
3972 'mode' to force_reg.
3973 (aarch64_add_offset): Update calls to aarch64_force_temporary.
3974 (aarch64_expand_mov_immediate): Likewise.
3975
7f2c00e6 39762013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
3977
3978 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
3979 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
3980
5ada7a14 39812013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3982
3983 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
3984 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
3985 (struct machine_function): Add tbegin_p.
3986 (s390_canonicalize_comparison): Fold CC mode compares to
3987 conditional jump if possible.
3988 (s390_emit_jump): Return the emitted jump.
3989 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
3990 Handle CCRAWmode compares.
3991 (s390_option_override): Default to -mhtm if available.
3992 (s390_reg_clobbered_rtx): Handle floating point regs as well.
3993 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
3994 FPRs instead of df_regs_ever_live_p.
3995 (s390_optimize_nonescaping_tx): New function.
3996 (s390_init_frame_layout): Extend clobbered_regs array to cover
3997 FPRs as well.
3998 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
3999 (s390_expand_tbegin): New function.
4000 (enum s390_builtin): New enum definition.
4001 (code_for_builtin): New array definition.
4002 (s390_init_builtins): New function.
4003 (s390_expand_builtin): New function.
4004 (TARGET_INIT_BUILTINS): Define.
4005 (TARGET_EXPAND_BUILTIN): Define.
3af61095 4006 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5ada7a14 4007 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4008 (s390_alc_comparison): Likewise.
4009 * config/s390/s390-modes.def: Add CCRAWmode.
4010 * config/s390/s390.h (processor_flags): Add PF_TX.
4011 (TARGET_CPU_HTM): Define macro.
4012 (TARGET_HTM): Define macro.
4013 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4014 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4015 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4016 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
3af61095 4017 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5ada7a14 4018 (TBEGIN_MASK, TBEGINC_MASK): New constants.
4019 ("*cc_to_int"): Move up.
4020 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4021 constants other than 0.
4022 ("*ccraw_to_int"): New insn and splitter definition.
4023 ("tbegin", "tbegin_nofloat", "tbegin_retry")
4024 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4025 ("tx_assist"): New expander.
4026 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4027 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4028 * config/s390/s390.opt: Add -mhtm option.
4029 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4030 * config/s390/htmxlintrin.h: New file.
4031 * config/s390/htmintrin.h: New file.
4032 * config/s390/s390intrin.h: New file.
4033 * doc/extend.texi: Document htm builtins.
4034 * config.gcc: Add the new header files to extra_headers.
4035
b1d3018a 40362013-06-26 Thomas Schwinge <thomas@codesourcery.com>
4037
4038 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
3af61095 4039 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
b1d3018a 4040
ce67dec2 40412013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4042 Pat Haugen <pthaugen@us.ibm.com>
4043 Peter Bergner <bergner@vnet.ibm.com>
4044
4045 * config/rs6000/power8.md: New.
4046 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4047 setting for power8 entry.
4048 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4049 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4050 test for Power4/Power5 only.
4051 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4052 support.
4053 (force_new_group): Adjust comment.
4054 * config/rs6000/rs6000.md: Include power8.md.
4055
1e939b24 40562013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
4057
4058 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4059 * config/arm/arm-protos.h (arm_max_conditional_execute): New
4060 declaration.
4061 (tune_params): Update comment.
4062 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4063 (arm_max_conditional_execute): New function.
4064 (thumb2_final_prescan_insn): Use max_insn_skipped and
4065 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4066
bb0d2509 40672013-06-25 Jakub Jelinek <jakub@redhat.com>
4068
4069 PR tree-optimization/57705
4070 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4071 SSA_NAME step, provided that it is not defined inside the loop.
3af61095 4072 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
bb0d2509 4073 (get_initial_def_for_induction): Handle SSA_NAME IV step.
4074
2f6c1cf4 40752013-06-25 Martin Jambor <mjambor@suse.cz>
4076
4077 PR middle-end/57670
4078 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4079 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4080 calls in the dump.
4081 (ipa_note_param_call): Initialize member_ptr flag.
4082 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4083 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4084 (ipa_write_indirect_edge_info): Stream member_ptr flag.
4085 (ipa_read_indirect_edge_info): Likewise.
4086
37704d29 40872013-06-25 Richard Biener <rguenther@suse.de>
4088
4089 PR middle-end/56977
4090 * passes.c (init_optimization_passes): Move pass_fold_builtins
4091 and pass_dce earlier with -Og.
4092
f42d23fc 40932013-06-25 Eric Botcazou <ebotcazou@adacore.com>
4094
4095 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4096 <BIT_FIELD_REF>: Remove trailing TAB.
4097 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4098 remove blank line.
4099
6c0a4a25 41002013-06-24 Martin Jambor <mjambor@suse.cz>
4101
4102 PR tree-optimization/57358
4103 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4104 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4105 (ipa_analyze_params_uses): Generate pessimistic info when true.
4106
48f42a9a 41072013-06-24 Martin Jambor <mjambor@suse.cz>
4108
4109 PR tree-optimization/57539
4110 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4111 global.inlined_to of the new node to it. All callers changed.
4112 * ipa-inline-transform.c (clone_inlined_nodes): New variable
4113 inlining_into, pass it to cgraph_clone_node.
4114 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4115 ipa_free_edge_args_substructures.
4116 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4117 rdesc linked list. Do not assert rdesc edges have inlined caller.
4118 Assert we have found an rdesc in the rdesc list.
4119
ea1a929f 41202013-06-24 Richard Biener <rguenther@suse.de>
4121
3af61095 4122 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
ea1a929f 4123 (pointer_set_lookup): Declare.
4124 (class pointer_map): New template class implementing a
4125 generic pointer to T map.
4126 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4127 pointer_map<T>::contains, pointer_map<T>::insert,
4128 pointer_map<T>::traverse): New functions.
4129 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4130 (pointer_set_lookup): New function.
4131 (pointer_set_contains): Use pointer_set_lookup.
4132 (pointer_set_insert): Likewise.
4133 (insert_aux): Remove.
4134 (struct pointer_map_t): Embed a pointer_set_t.
4135 (pointer_map_create): Adjust.
4136 (pointer_map_destroy): Likewise.
4137 (pointer_map_contains): Likewise.
4138 (pointer_map_insert): Likewise.
4139 (pointer_map_traverse): Likewise.
4140 * tree-streamer.h (struct streamer_tree_cache_d): Use a
4141 pointer_map<unsigned> instead of a pointer_map_t.
4142 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4143 (streamer_tree_cache_lookup): Likewise.
4144 (streamer_tree_cache_create): Likewise.
4145 (streamer_tree_cache_delete): Likewise.
4146 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4147 pointer_map<unsigned> instead of a pointer_map_t.
4148 (lto_init_tree_ref_encoder): Adjust.
4149 (lto_destroy_tree_ref_encoder): Likewise.
4150 * lto-section-out.c (lto_output_decl_index): Likewise.
4151 (lto_record_function_out_decl_state): Likewise.
4152 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4153
5f751b20 41542013-06-24 Richard Biener <rguenther@suse.de>
4155
4156 PR tree-optimization/57488
4157 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4158
6611aa31 41592013-06-24 Alan Modra <amodra@gmail.com>
4160
4161 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4162 (gen_easy_altivec_constant): Likewise.
4163 * config/rs6000/predicates.md (easy_vector_constant_add_self,
4164 easy_vector_constant_msb): Likewise.
4165
15f8efea 41662013-06-23 Jakub Jelinek <jakub@redhat.com>
4167
4168 PR target/57688
4169 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4170 add missing return true.
4171
640c5f1f 41722013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
4173
4174 PR target/52483
4175 * config/sh/predicates.md (general_extend_operand): Invoke
4176 general_movsrc_operand for memory operands.
4177 (general_movsrc_operand): Allow reg+reg addressing, do not use
4178 general_operand for memory operands.
4179
ef21d40e 41802013-06-23 Sriraman Tallam <tmsriram@google.com>
4181
4182 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
4183 when current target options does not apply.
4184 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4185 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
3af61095 4186 * config/i386/bmiintrin.h: Pass appropriate target
4187 attributes to header.
ef21d40e 4188 * config/i386/mmintrin.h: Ditto.
4189 * config/i386/nmmintrin.h: Ditto.
4190 * config/i386/avx2intrin.h: Ditto.
4191 * config/i386/fxsrintrin.h: Ditto.
4192 * config/i386/tbmintrin.h: Ditto.
4193 * config/i386/xsaveintrin.h: Ditto.
4194 * config/i386/f16cintrin.h: Ditto.
4195 * config/i386/xtestintrin.h: Ditto.
4196 * config/i386/xsaveoptintrin.h: Ditto.
4197 * config/i386/bmi2intrin.h: Ditto.
4198 * config/i386/lzcntintrin.h: Ditto.
4199 * config/i386/smmintrin.h: Ditto.
4200 * config/i386/wmmintrin.h: Ditto.
4201 * config/i386/x86intrin.h: Remove all header include guards.
4202 * config/i386/prfchwintrin.h: Ditto.
4203 * config/i386/pmmintrin.h: Ditto.
4204 * config/i386/tmmintrin.h: Ditto.
4205 * config/i386/xmmintrin.h: Ditto.
4206 * config/i386/popcntintrin.h: Ditto.
4207 * config/i386/rdseedintrin.h: Ditto.
4208 * config/i386/ammintrin.h: Ditto.
4209 * config/i386/emmintrin.h: Ditto.
4210 * config/i386/immintrin.h: Remove all header include guards.
4211 * config/i386/fma4intrin.h: Ditto.
4212 * config/i386/lwpintrin.h: Ditto.
4213 * config/i386/xopintrin.h: Ditto.
4214 * config/i386/ia32intrin.h: Ditto.
4215 * config/i386/avxintrin.h: Ditto.
4216 * config/i386/rtmintrin.h: Ditto.
4217 * config/i386/fmaintrin.h: Ditto.
4218 * config/i386/mm3dnow.h: Ditto.
4219
cab840cb 42202013-06-22 Sriraman Tallam <tmsriram@google.com>
4221
4222 * common/config/i386/i386-common.c: Handle LZCNT.
4223
0604bcfb 42242013-06-22 Andi Kleen <ak@linux.intel.com>
4225
4226 * doc/extend.texi: Use __atomic_store_n instead of
4227 __atomic_store in HLE example.
4228
9553eedb 42292013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
4230
4231 * config/sh/sh.c: Remove <cstdlib> workaround.
4232
4aa48cf8 42332013-06-21 Andi Kleen <ak@linux.intel.com>
4234
3af61095 4235 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
4aa48cf8 4236
8e02d80e 42372013-06-21 Andi Kleen <ak@linux.intel.com>
4238
4239 * doc/extend.texi: Document that __atomic_clear and
4240 __atomic_test_and_set should only be used with bool.
4241
7371d299 42422013-06-20 Jan Hubicka <jh@suse.cz>
4243
4244 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
4245 types_same_for_odr.
4246 * tree.c (decls_same_for_odr): New function.
4247 (same_for_edr): New function.
4248 (types_same_for_odr): New function.
4249 (get_binfo_at_offset): Use it.
4250 * tree.h (types_same_for_odr): Declare.
4251
9553eedb 42522013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
d6601303 4253 Jason Merrill <jason@redhat.com>
4254
4255 * system.h: Include <cstdlib> as well as <stdlib.h>.
4256
43844d9d 42572013-06-20 Uros Bizjak <ubizjak@gmail.com>
4258
4259 PR target/57655
4260 * config/i386/i386.c (construct_container): Report error if
4261 long double is used with disabled x87 float returns.
4262
fdbe58cf 42632013-06-20 Jan Hubicka <jh@suse.cz>
4264
4265 * lto-cgraph.c (input_symtab): Do not set cgraph state.
4266
cb456db5 42672013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
4268
4269 PR rtl-optimization/57425
4270 PR rtl-optimization/57569
4271 * alias.c (write_dependence_p): Remove parameters mem_mode and
4272 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
4273 Changed all callers.
4274 (canon_anti_dependence): Get comments and semantics in sync.
4275 Add parameter mem_canonicalized. Changed all callers.
4276 * rtl.h (canon_anti_dependence): Update prototype.
4277
ab4542eb 42782013-06-20 Richard Biener <rguenther@suse.de>
4279
4280 * data-streamer-in.c (streamer_read_uhwi): Optimize single
4281 byte case, inline streamer_read_uchar and defer section
4282 overrun check.
4283
d51283e3 42842013-06-20 Richard Biener <rguenther@suse.de>
4285
4286 PR tree-optimization/57584
4287 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
4288 SSA names into the expanded expression that take part in
4289 abnormal coalescing.
4290
5f7f04d8 42912013-06-19 Sharad Singhai <singhai@google.com>
fafee80b 4292
4293 * gcov.c (print_usage): Handle new option.
4294 (process_args): Ditto.
4295 (get_gcov_intermediate_filename): New function.
4296 (output_intermediate_file): New function.
4297 (output_gcov_file): New function
4298 (generate_results): Handle new option.
4299 (release_function): Relase demangled name.
4300 (read_graph_file): Handle demangled name.
4301 (output_lines): Ditto.
4302 * doc/gcov.texi: Document gcov intermediate format.
4303
6c397456 43042013-06-19 Vladimir Makarov <vmakarov@redhat.com>
4305
4306 PR bootstrap/57604
4307 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
4308 (lra_emit_add): Use the functions. Add comment about Y as an
4309 address segment.
4310
908e0747 43112013-06-19 David Edelsohn <dje.gcc@gmail.com>
4312
4313 PR driver/57652
4314 * collect2.c (collect_atexit): New.
4315 (collect_exit): Delete.
4316 (main): Register collect_atexit with atexit.
4317 (collect_wait): Change collect_exit to exit.
4318 (do_wait): Same.
4319 * collect2.h (collect_exit): Delete.
4320 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
4321
febd1624 43222013-06-19 Wei Mi <wmi@google.com>
4323
4324 PR rtl-optimization/57518
4325 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
4326 if regno is used in paradoxical subreg.
4327 (update_equiv_regs): Check pdx_subregs[regno] before
4328 set a reg to be equivalent with a mem.
4329
0dc1ebca 43302013-06-19 Matthias Klose <doko@ubuntu.com>
4331
4332 PR driver/57651
4333 * file-find.h (find_a_file): Add a mode parameter.
4334 * file-find.c (find_a_file): Likewise.
4335 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
4336 with X_OK for the executables.
4337 * collect2.c (main): Call find_a_file with X_OK.
4338
b6c75192 43392013-06-19 Steve Ellcey <sellcey@mips.com>
4340
4341 PR target/56942
4342 * config/mips/mips.md (casesi_internal_mips16_<mode>):
4343 Use NEXT_INSN instead of next_real_insn.
4344
df8d3e89 43452013-06-19 Jan Hubicka <jh@suse.cz>
4346
4347 * cgraph.h (const_value_known_p): Replace by ...
4348 (ctor_for_folding): .. this one.
4349 * cgraphunit.c (process_function_and_variable_attributes): Use it.
4350 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
4351 * expr.c (expand_expr_real_1): Likewise.
4352 (string_constant): Likewise.
4353 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
4354 * ipa.c (process_references): Likewise.
4355 (symtab_remove_unreachable_nodes): Likewise.
4356 * ipa-inline-analysis.c (param_change_prob): Likewise.
4357 * gimple-fold.c (canonicalize_constructor_val): Likewise.
4358 (get_base_constructor): Likwise.
4359 * varpool.c (varpool_remove_node): Likewise.
4360 (varpool_remove_initializer): LIkewise.
4361 (dump_varpool_node): LIkwise.
4362 (const_value_known_p): Rewrite to ...
4363 (ctor_for_folding): ... this one.
4364
a90c63ff 43652013-06-19 Jakub Jelinek <jakub@redhat.com>
4366
4367 PR driver/57651
4368 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
4369 PERSONALITY in $PATH derived prefixes.
4370
16bc66ec 43712013-06-19 Jeff Law <law@redhat.com>
4372
040f64e5 4373 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
4374 in comment.
4375
16bc66ec 4376 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
4377 (simplify_bitwise_binary): Use it to simpify certain binary ops on
4378 booleans.
4379
70e6ce19 43802013-06-19 Sofiane Naci <sofiane.naci@arm.com>
4381
4382 * config/arm/vfp.md: Move VFP instruction classification documentation
4383 to ...
4384 * config/arm/arm.md: ... here. Update instruction classification
4385 documentation.
4386
98562479 43872013-06-19 Richard Earnshaw <rearnsha@arm.com>
4388
4389 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
4390 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
4391 pattern. Use more efficient sequences on ARMv5 and Thumb2.
4392
0cd9a9a9 43932013-06-19 Steven Bosscher <steven@gcc.gnu.org>
4394
4395 PR target/57609
4396 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4397 with NEXT_INSN. Use tablejump_p to check for jump table data
4398 insns.
4399
3390dcc4 44002013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
4401
4402 PR c++/56544
4403 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
4404 that now in C++ the value is correct per the C++ standards.
4405
7871778b 44062013-06-19 Richard Biener <rguenther@suse.de>
4407
4408 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
4409 for global context.
4410
c8de5323 44112013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4412
4413 Revert:
4414 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4415
4416 PR target/57609
4417 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4418 with next_active_insn.
4419
3bc4161a 44202013-06-18 Sriraman Tallam <tmsriram@google.com>
4421
4422 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
4423 functions are inlined during failures to flag an error.
4424 * tree-inline.c (expand_call_inline): Allow the error to be flagged
4425 in early inline pass.
3b9449a5 4426
e8df52ad 44272013-06-18 H.J. Lu <hongjiu.lu@intel.com>
4428
4429 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
4430 in comments.
4431
1c2054e4 44322013-06-18 Julian Brown <julian@codesourcery.com>
4433
4434 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
4435 Permit virtual register pre-reload if !strict.
4436 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
4437 change.
4438 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
4439 prototype.
4440 * config/arm/neon.md (movmisalign<mode>): Use
4441 neon_perm_struct_or_reg_operand instead of
4442 neon_struct_or_register_operand.
4443 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
4444 neon_permissive_struct_operand instead of neon_struct_operand.
4445 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
4446 neon_vector_mem_operand.
4447 * config/arm/predicates.md (neon_struct_operand): Adjust call to
4448 neon_vector_mem_operand.
4449 (neon_permissive_struct_operand): New.
4450 (neon_struct_or_register_operand): Rename to...
4451 (neon_perm_struct_or_reg_operand): This. Adjust call to
4452 neon_vector_mem_operand.
4453
32d76803 44542013-06-18 Richard Biener <rguenther@suse.de>
4455
4456 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
4457 * lto-streamer.h: Include pointer-set.h.
4458 (struct lto_decl_slot): Remove.
4459 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
4460 Remove next_index entry.
4461 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4462 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4463 (lto_init_tree_ref_encoder): Adjust.
4464 (lto_destroy_tree_ref_encoder): Likewise.
4465 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4466 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4467 (lto_output_decl_index): Adjust.
4468 (lto_new_out_decl_state): Likewise.
4469 (lto_record_function_out_decl_state): Likewise.
4470 * lto-streamer-out.c (copy_function): Likewise.
4471
7eacb0dd 44722013-06-18 Richard Biener <rguenther@suse.de>
4473
4474 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
4475 * cgraphunit.c: Include cfgloop.h.
4476 (init_lowered_empty_function): Initialize the loop tree.
4477 (assemble_thunk): Insert new BBs into loops.
4478
be1cd111 44792013-06-18 Richard Biener <rguenther@suse.de>
4480
4481 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
4482 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
4483 the map from cache entry to cache index optional.
4484 (streamer_tree_cache_replace_tree): Adjust accordingly.
4485 (streamer_tree_cache_append): Likewise.
4486 (streamer_tree_cache_delete): Likewise.
4487 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
4488 streamer cache map from cache entry to cache index.
4489 * lto-streamer-out.c (create_output_block): Adjust.
4490
9da0ec36 44912013-06-18 Sofiane Naci <sofiane.naci@arm.com>
4492
3b9449a5 4493 * config/arm/arm.md (attribute "insn"): Move multiplication and
4494 division attributes to...
9da0ec36 4495 (attribute "type"): ... here. Remove mult.
4496 (attribute "mul32"): New attribute.
4497 (attribute "mul64"): Add umaal.
4498 (*arm_mulsi3): Update attributes.
4499 (*arm_mulsi3_v6): Likewise.
4500 (*thumb_mulsi3): Likewise.
4501 (*thumb_mulsi3_v6): Likewise.
4502 (*mulsi3_compare0): Likewise.
4503 (*mulsi3_compare0_v6): Likewise.
4504 (*mulsi_compare0_scratch): Likewise.
4505 (*mulsi_compare0_scratch_v6): Likewise.
4506 (*mulsi3addsi): Likewise.
4507 (*mulsi3addsi_v6): Likewise.
4508 (*mulsi3addsi_compare0): Likewise.
4509 (*mulsi3addsi_compare0_v6): Likewise.
4510 (*mulsi3addsi_compare0_scratch): Likewise.
4511 (*mulsi3addsi_compare0_scratch_v6): Likewise.
4512 (*mulsi3subsi): Likewise.
4513 (*mulsidi3adddi): Likewise.
4514 (*mulsi3addsi_v6): Likewise.
4515 (*mulsidi3adddi_v6): Likewise.
4516 (*mulsidi3_nov6): Likewise.
4517 (*mulsidi3_v6): Likewise.
4518 (*umulsidi3_nov6): Likewise.
4519 (*umulsidi3_v6): Likewise.
4520 (*umulsidi3adddi): Likewise.
4521 (*umulsidi3adddi_v6): Likewise.
4522 (*smulsi3_highpart_nov6): Likewise.
4523 (*smulsi3_highpart_v6): Likewise.
4524 (*umulsi3_highpart_nov6): Likewise.
4525 (*umulsi3_highpart_v6): Likewise.
4526 (mulhisi3): Likewise.
4527 (*mulhisi3tb): Likewise.
4528 (*mulhisi3bt): Likewise.
4529 (*mulhisi3tt): Likewise.
4530 (maddhisi4): Likewise.
4531 (*maddhisi4tb): Likewise.
4532 (*maddhisi4tt): Likewise.
4533 (maddhidi4): Likewise.
4534 (*maddhidi4tb): Likewise.
4535 (*maddhidi4tt): Likewise.
4536 (divsi3): Likewise.
4537 (udivsi3): Likewise.
4538 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
4539 (thumb2_mulsi_short_compare0): Likewise.
4540 (thumb2_mulsi_short_compare0_scratch): Likewise.
4541 * config/arm/arm1020e.md (1020mult1): Update attribute change.
4542 (1020mult2): Likewise.
4543 (1020mult3): Likewise.
4544 (1020mult4): Likewise.
4545 (1020mult5): Likewise.
4546 (1020mult6): Likewise.
3b9449a5 4547 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
4548 change.
9da0ec36 4549 (cortex_a15_mult64): Likewise.
4550 (cortex_a15_sdiv): Likewise.
4551 (cortex_a15_udiv): Likewise.
4552 * config/arm/arm1026ejs.md (mult1): Update attribute change.
4553 (mult2): Likewise.
4554 (mult3): Likewise.
4555 (mult4): Likewise.
4556 (mult5): Likewise.
4557 (mult6): Likewise.
4558 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
4559 (pj4_ir_div): Likewise.
4560 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
4561 (11_mult2): Likewise.
4562 (11_mult3): Likewise.
4563 (11_mult4): Likewise.
4564 (11_mult5): Likewise.
4565 (11_mult6): Likewise.
4566 (11_mult7): Likewise.
4567 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
4568 (cortex_a8_mla): Likewise.
4569 (cortex_a8_mull): Likewise.
4570 (cortex_a8_smulwy): Likewise.
4571 (cortex_a8_smlald): Likewise.
4572 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
4573 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
4574 (cortex_r4_mul_3): Likewise.
4575 (cortex_r4_mla_4): Likewise.
4576 (cortex_r4_mla_3): Likewise.
4577 (cortex_r4_smlald): Likewise.
4578 (cortex_r4_mull): Likewise.
4579 (cortex_r4_sdiv): Likewise.
4580 (cortex_r4_udiv): Likewise.
4581 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
4582 (cortex_a7_idiv): Likewise.
4583 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
4584 (9_mult2): Likewise.
4585 (9_mult3): Likewise.
4586 (9_mult4): Likewise.
4587 (9_mult5): Likewise.
4588 (9_mult6): Likewise.
4589 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
4590 (cortex_a53_sdiv): Likewise.
4591 (cortex_a53_udiv): Likewise.
4592 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
4593 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
4594 (mp626_mult2): Likewise.
4595 (mp626_mult3): Likewise.
4596 (mp626_mult4): Likewise.
4597 * config/arm/fa526.md (526_mult1): Update attribute change.
4598 (526_mult2): Likewise.
4599 * config/arm/arm-generic.md (mult): Update attribute change.
4600 (mult_ldsched_strongarm): Likewise.
4601 (mult_ldsched): Likewise.
4602 (multi_cycle): Likewise.
4603 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
4604 * config/arm/fa606te.md (606te_mult1): Update attribute change.
4605 (606te_mult2): Likewise.
4606 (606te_mult3): Likewise.
4607 (606te_mult4): Likewise.
4608 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
4609 (cortex_a9_mac16): Likewise.
4610 (cortex_a9_multiply): Likewise.
4611 (cortex_a9_mac): Likewise.
4612 (cortex_a9_multiply_long): Likewise.
4613 * config/arm/fa626te.md (626te_mult1): Update attribute change.
4614 (626te_mult2): Likewise.
4615 (626te_mult3): Likewise.
4616 (626te_mult4): Likewise.
4617
97e13fca 46182013-06-18 Richard Biener <rguenther@suse.de>
4619
4620 PR lto/57334
4621 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
4622
f5cf1225 46232013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4624
4625 PR target/57609
4626 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4627 with next_active_insn.
4628
efb66c8d 46292013-06-18 Alan Modra <amodra@gmail.com>
4630
4631 * config/rs6000/rs6000.h (enum data_align): New.
4632 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
4633 (DATA_ABI_ALIGNMENT): Define.
4634 (CONSTANT_ALIGNMENT): Correct comment.
4635 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
4636 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
4637
98044fac 46382013-06-17 David Malcolm <dmalcolm@redhat.com>
4639
4640 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
4641 ATTRIBUTE_UNUSED marking.
4642
fcd3260c 46432013-06-17 Sofiane Naci <sofiane.naci@arm.com>
4644
4645 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
4646 alternative and update.
4647 (aarch64_dup_lanedi): Delete.
4648 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
4649 * config/aarch64/aarch64-simd-builtins.def: Update.
4650
8ceff600 46512013-06-17 Richard Biener <rguenther@suse.de>
4652
4653 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
4654 (lto_input_scc): Declare.
4655 (lto_input_tree_1): Likewise.
4656 (struct lto_stats_d): Add num_tree_bodies_output and
4657 num_pickle_refs_output.
4658 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
4659 (lto_read_tree_1): Split out from ...
4660 (lto_read_tree): ... this.
4661 (lto_input_scc): New function.
4662 (lto_input_tree_1): Split out from ...
4663 (lto_input_tree): ... this. Handle LTO_tree_scc.
4664 (lto_data_in_create): Create the streamer cache without hashes.
4665 * lto-streamer-out.c (create_output_block): Create the streamer
4666 cache with hashes when not doing WPA.
4667 (lto_write_tree_1): Split out from ...
4668 (lto_write_tree): ... this.
4669 (get_symbol_initial_value): New function.
4670 (lto_output_tree_1): Split out from ...
4671 (lto_output_tree): ... this. Write trees as series of SCCs
4672 using a DFS walk via DFS_write_tree.
4673 (struct sccs, struct scc_entry): New types.
4674 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
4675 (DFS_write_tree_body): New function.
4676 (DFS_write_tree): Likewise.
4677 (hash_tree): Likewise.
4678 (scc_entry_compare): Likewise.
4679 (hash_scc): Likewise.
4680 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
4681 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
4682 TREE_CHAIN as regular reference.
4683 (streamer_read_integer_cst): Remove.
4684 (streamer_get_pickled_tree): Adjust.
4685 * tree-streamer-out.c (streamer_write_chain): Disable streaming
4686 of DECL_EXTERNALs in BLOCK_VARS for now.
4687 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
4688 reference.
4689 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
4690 Add hash value argument and record that if hashes are recorded
4691 in the cache.
4692 (streamer_tree_cache_insert_1): Adjust.
4693 (streamer_tree_cache_insert): Likewise.
4694 (streamer_tree_cache_insert_at): Rename to ...
4695 (streamer_tree_cache_replace_tree): ... this and adjust.
4696 (streamer_tree_cache_append): Adjust.
4697 (record_common_node): Likewise.
4698 (streamer_tree_cache_create): Add argument whether to
4699 record hash values together with trees.
4700 (streamer_tree_cache_delete): Adjust.
4701 * tree-streamer.h (struct streamer_tree_cache_d): Add
4702 vector of hashes.
4703 (streamer_read_integer_cst): Remove.
4704 (streamer_tree_cache_insert): Adjust.
4705 (streamer_tree_cache_append): Likewise.
4706 (streamer_tree_cache_insert_at): Rename to ...
4707 (streamer_tree_cache_replace_tree): ... this and adjust.
4708 (streamer_tree_cache_create): Add argument whether to record hashes.
4709 (streamer_tree_cache_get): Rename to ...
4710 (streamer_tree_cache_get_tree): ... this.
4711 (streamer_tree_cache_get_hash): New function.
4712 * tree.c (cache_integer_cst): New function.
4713 * tree.h (cache_integer_cst): Declare.
4714 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
4715 * lto-symtab.c (lto_varpool_replace_node): Only release
4716 DECL_INITIAL of non-prevailing decls.
4717 * varpool.c (varpool_remove_initializer): Do not release
4718 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
4719
0b4e9fcd 47202013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
4721
4722 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
4723 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
4724 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
4725 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
4726 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
4727 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
4728 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
4729 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
4730 instead of TARGET_64BIT.
4731 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
4732 Require ISA_HAS_<D>DIV.
4733
61ad2260 47342013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
4735
4736 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
4737 (mips*-*-linux*): Move default with_llsc setting to where other
4738 defaults are set.
4739 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
4740 with_arch block.
4741 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
4742 Likewise. Remove default with_tune setting. Move default float
4743 setting to its own block. Handle with_llsc in the same block as above.
4744
ddba76b8 47452013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
4746
4747 PR rtl-optimization/57425
4748 PR rtl-optimization/57569
f28ea912 4749 * alias.c (write_dependence_p): Add new parameters mem_mode,
ddba76b8 4750 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
4751 Changed all callers.
4752 (canon_anti_dependence): New function.
4753 * cse.c (check_dependence): Use canon_anti_dependence.
4754 * cselib.c (cselib_invalidate_mem): Likewise.
4755 * rtl.h (canon_anti_dependence): Declare.
4756
5e524dea 47572013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
4758
4759 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
4760 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
4761 ".set mips3" for 64-bit targets.
4762
341de017 47632013-06-15 Dehao Chen <dehao@google.com>
4764
4765 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
4766 * gimple-low.c (gimple_check_call_matching_types): Likewise.
4767 (gimple_check_call_args): Likewise.
4768 * value-prof.c (check_ic_target): Likewise.
4769 * ipa-inline.c (early_inliner): Likewise.
4770 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
4771 * cgraph.c (cgraph_create_edge_1): Likewise.
4772 (cgraph_make_edge_direct): Likewise.
4773
7a57f216 47742013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4775
4776 PR target/57615
4777 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
4778 rs6000_output_move_128bit to handle emitting quad memory
4779 operations. Set attribute length to 8 bytes.
4780
74375e70 47812013-06-14 Vidya Praveen <vidyapraveen@arm.com>
4782
4783 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
4784 New pattern.
4785 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
4786 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
4787 (aarch64_<su>mlsl<mode>): Likewise.
4788
e7a88601 47892013-06-14 Mike Stump <mikestump@comcast.net>
4790
4791 * Makefile.in (TARGET_H): Add insn-codes.h.
4792
780871fb 47932013-06-14 Alan Modra <amodra@gmail.com>
4794
4795 PR middle-end/57134
4796 PR middle-end/57586
4797 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
4798 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
4799 bitfield expansion when EXPAND_MEMORY.
4800 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
4801
8a4a56e3 48022013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4803
4804 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
4805 test for clearing quad memory on 32-bit later.
4806
4895a1c6 48072013-06-13 Marc Glisse <marc.glisse@inria.fr>
4808
4809 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
4810 (fold_negate_expr): Likewise.
4811 (fold_real_zero_addition_p): Handle vectors.
4812 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
4813
9db6f2e0 48142013-06-14 Alan Modra <amodra@gmail.com>
4815
4816 * varasm.c (force_const_mem): Revert 2013-06-07 change.
4817
a1d135c3 48182013-06-13 Jan Hubicka <jh@suse.cz>
4819
4820 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
4821 Local comdats are not externally visible.
4822 * symtab.c (dump_symtab_base): Dump externally visible.
4823 (verify_symtab_base): Verify back links in the symtab hash.
4824
fb4441fa 48252013-06-13 Bin Cheng <bin.cheng@arm.com>
4826
4827 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
4828 CONVERT_EXPR as equal nodes.
4829
3f0def8e 48302013-06-13 Bin Cheng <bin.cheng@arm.com>
4831
3b9449a5 4832 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
3f0def8e 4833
2c83a45e 48342013-06-13 Marc Glisse <marc.glisse@inria.fr>
4835
4836 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
4837 Generalize to complex and vector.
4838 * tree.c (build_all_ones_cst): New function.
4839 * tree.h (build_all_ones_cst): Declare it.
4840
74a1167a 48412013-06-13 Alan Modra <amodra@gmail.com>
4842
4843 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
4844 * config/rs6000/rs6000.md (signbittf2): New insn.
4845 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
4846 (abstf2_internal, cmptf_internal2): Likewise.
4847 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
4848
488befe3 48492013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4850 Pat Haugen <pthaugen@us.ibm.com>
4851 Peter Bergner <bergner@vnet.ibm.com>
4852
4853 * config/rs6000/rs6000.c (emit_load_locked): Add support for
4854 power8 byte, half-word, and quad-word atomic instructions.
4855 (emit_store_conditional): Likewise.
4856 (rs6000_expand_atomic_compare_and_swap): Likewise.
4857 (rs6000_expand_atomic_op): Likewise.
4858
4859 * config/rs6000/sync.md (larx): Add new modes for power8.
4860 (stcx): Likewise.
4861 (AINT): New mode iterator to include TImode as well as normal
4862 integer modes on power8.
4863 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
4864 that VSX registers are not considered. Use AINT mode iterator
4865 instead of INT1 to allow inclusion of quad word atomic operations
4866 on power8.
4867 (load_locked<mode>): Likewise.
4868 (store_conditional<mode>): Likewise.
4869 (atomic_compare_and_swap<mode>): Likewise.
4870 (atomic_exchange<mode>): Likewise.
4871 (atomic_nand<mode>): Likewise.
4872 (atomic_fetch_<fetchop_name><mode>): Likewise.
4873 (atomic_nand_fetch<mode>): Likewise.
4874 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
4875 each type.
4876 (ATOMIC): On power8, add QImode, HImode modes.
4877 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
4878 modes that promote to SImode.
4879 (load_lockedti): Convert TImode arguments to PTImode, so that we
4880 get a guaranteed even/odd register pair.
4881 (load_lockedpti): Likewise.
4882 (store_conditionalti): Likewise.
4883 (store_conditionalpti): Likewise.
4884
4885 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
4886 atomic load/store instructions.
4887 (HSI): Likewise.
4888
885edc78 48892013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
4890
4891 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
4892 loads.
4893 (insn_count): New attribute, with most cases extracted from...
4894 (length): ...here. Redefine most cases in terms of insn_count.
4895 (single_insn): Delete.
4896 (can_delay): Use insn_count to check for single instructions.
4897 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
4898 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
4899 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
4900 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
4901 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
4902 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
4903 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
4904 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
4905 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
4906 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
4907 rather than "length".
4908 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
4909 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
4910 Use "insn_count" rather than "length".
4911 * config/mips/mips-dsp.md
4912 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
4913 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
4914 length attributes.
4915
7855598c 49162013-06-12 Marc Glisse <marc.glisse@inria.fr>
4917
4918 PR tree-optimization/57361
4919 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
4920
d820433c 49212013-06-12 Sofiane Naci <sofiane.naci@arm.com>
4922
3b9449a5 4923 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
4924 to split.
d820433c 4925 (aarch64_simd_combine<mode>): New instruction expansion.
4926 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
4927 function prototype.
4928 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
4929 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
4930
da4b8721 49312013-06-12 Jan Hubicka <jh@suse.cz>
4932
4933 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
4934 decl has when in streaming stage.
4935 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
4936 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
4937
e7346145 49382013-06-12 Roland Stigge <stigge@antcom.de>
4939
4940 PR target/57578
4941 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
4942
d82cd73d 49432013-06-12 Jakub Jelinek <jakub@redhat.com>
4944
4945 PR tree-optimization/57537
4946 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
4947 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
4948
2f5bed2a 49492013-06-12 Richard Biener <rguenther@suse.de>
4950
4951 * data-streamer.h (streamer_write_char_stream): CSE
4952 obs->current_pointer.
4953 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
4954 streamer_write_char_stream manually and optimize the resulting loop.
4955 (streamer_write_hwi_stream): Likewise.
4956
9e9c3e92 49572013-06-12 Jan Hubicka <jh@suse.cz>
4958
4959 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
4960 * cgraph.h (varpool_create_empty_node): Declare.
4961 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
4962 duplicated nodes.
4963 * symtab.c (symtab_unregister_node): Be lax about missin entries
4964 in node hash.
4965 (symtab_get_node): Update comment.
4966 * varpool.c (varpool_create_empty_node): Break out from ...
4967 (varpool_node_for_decl): ... here.
4968 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
4969
e9b15297 49702013-06-12 Eric Botcazou <ebotcazou@adacore.com>
4971
4972 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
4973 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
4974 part. Use straight-line flow at the end.
4975 <COMPONENT_REF>: Remove superfluous else.
4976 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
4977
5658fec0 49782013-06-12 Jakub Jelinek <jakub@redhat.com>
4979
4980 PR target/56564
4981 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
4982 target hook even for !TREE_PUBLIC decls. If no resolution info
4983 is available, return false for common and external decls.
4984
bb56c23a 49852013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
4986
4987 * config/rl78/constraints.md (U): New constraint.
4988 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
4989 valloc attribute.
4990
8f88661d 49912013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4992
4993 PR target/57589
4994 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
4995 to allow returning address to AT_PLATFORM name.
4996
f2526cce 49972013-06-11 Jan Hubicka <jh@suse.cz>
4998
4999 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5000 * cgraph.h (symtab_node_base): Add weakref flag.
5001 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5002 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5003 (output_weakrefs): Use weakref flag.
5004 * fold-const.c (simple_operand_p): Handle WEAK.
5005 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5006 * ipa.c (varpool_externally_visible_p): Drop weakref.
5007 (function_and_variable_visibility): Update comment; fix weakref
5008 sanity checks; do not clear DECL_WEAK on them.
5009 * lto-cgraph.c (lto_output_node): update.
5010 (lto_output_varpool_node): Update.
5011 (input_overwrite_node): Update.
5012 (input_node): Update.
5013 (input_varpool_node): Update.
5014 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5015 (lto_symtab_merge_symbols): Add sanity check.
5016 (lto_symtab_prevailing_decl): Do not special case weakrefs.
5017 * passes.c (rest_of_decl_compilation): Set static flag, too.
5018 * symtab.c (dump_symtab_base): Dump weakref.
5019 (verify_symtab_base): Sanity check weakrefs.
5020 (symtab_make_decl_local): Remove duplicated code.
5021 (symtab_alias_ultimate_target): Simplify.
5022 * varpool.c (varpool_create_variable_alias): Set weakref flag.
8f88661d 5023
99bdcd9d 50242013-06-11 DJ Delorie <dj@redhat.com>
5025
5026 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5027 (rl78_unwind_word_mode): New.
5028
94a2945a 50292013-06-11 David Malcolm <dmalcolm@redhat.com>
5030
5031 * final.c (debug_prefix_maps): Make static.
5032
5ef9352d 50332013-06-11 David Malcolm <dmalcolm@redhat.com>
5034
5035 * function.c (initial_trampoline): Remove stray copy.
5036
20431bc8 50372013-06-11 Sofiane Naci <sofiane.naci@arm.com>
5038
5039 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5040
fd8b458b 50412013-06-11 Martin Jambor <mjambor@suse.cz>
5042
5043 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5044 within bounds at the beginning of the function.
5045
71079e07 50462013-06-11 Alan Modra <amodra@gmail.com>
5047
5048 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
5049 reporting.
5050 (get_named_section): Don't NULL !DECL_P decl.
5051
80f0bcca 50522013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
5053
5054 * doc/invoke.texi (core-avx2): Document.
5055 (slm): Likewise.
5056 (atom): Updated with MOVBE.
5057
9a65bef9 50582013-06-11 Richard Biener <rguenther@suse.de>
5059
3b9449a5 5060 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9a65bef9 5061
a6b212e1 50622013-06-11 Anton Blanchard <anton@samba.org>
5063
5064 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5065 correct shift value in little-endian mode.
5066
8ef2f3bf 50672013-06-11 Jakub Jelinek <jakub@redhat.com>
5068
5069 PR target/56564
5070 * varasm.c (get_variable_align): Move #endif to the right place.
5071
daccd6b3 50722013-06-10 Cary Coutant <ccoutant@google.com>
5073
5074 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5075 for hash so that hash table traversal order is deterministic.
5076
f88fbcb9 50772013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5078 Pat Haugen <pthaugen@us.ibm.com>
5079 Peter Bergner <bergner@vnet.ibm.com>
5080
5081 * config/rs6000/vector.md (GPR move splitter): Do not split moves
5082 of vectors in GPRS if they are direct moves or quad word load or
5083 store moves.
5084
5085 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5086 declaration.
5087 (direct_move_p): Likewise.
5088 (quad_load_store_p): Likewise.
5089
5090 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5091 classes into bins based on the physical register type.
5092 (reg_class_to_reg_type): Likewise.
5093 (IS_STD_REG_TYPE): Likewise.
5094 (IS_FP_VECT_REG_TYPE): Likewise.
5095 (reload_fpr_gpr): Arrays to determine what insn to use if we can
5096 use direct move instructions.
5097 (reload_gpr_vsx): Likewise.
5098 (reload_vsx_gpr): Likewise.
5099 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5100 information that is a simplification of register classes. Also
5101 precalculate direct move reload helpers.
5102 (direct_move_p): New function to return true if the operation can
5103 be done as a direct move instruciton.
5104 (quad_load_store_p): New function to return true if the operation
5105 is a quad memory operation.
5106 (rs6000_legitimize_address): If quad memory, only allow register
5107 indirect for TImode addresses.
5108 (rs6000_legitimate_address_p): Likewise.
5109 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5110 (rs6000_reload_register_type): Likewise.
5111 (register_to_reg_type): Return register type.
5112 (rs6000_secondary_reload_simple_move): New helper function for
5113 secondary reload and secondary memory needed to identify anything
5114 that is a simple move, and does not need reloading.
5115 (rs6000_secondary_reload_direct_move): New helper function for
5116 secondary reload to identify cases that can be done with several
5117 instructions via the direct move instructions.
5118 (rs6000_secondary_reload_move): New helper function for secondary
5119 reload to identify moves between register types that can be done.
5120 (rs6000_secondary_reload): Add support for quad memory operations
5121 and for direct move.
5122 (rs6000_secondary_memory_needed): Likewise.
5123 (rs6000_debug_secondary_memory_needed): Change argument names.
5124 (rs6000_output_move_128bit): New function to return the move to
5125 use for 128-bit moves, including knowing about the various
5126 limitations of quad memory operations.
5127
5128 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5129 memory operations. call rs6000_output_move_128bit for the actual
5130 instruciton(s) to generate.
5131 (vsx_movti_64bit): Likewise.
5132
5133 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5134 (UNSPEC_P8V_MTVSRWZ): Likewise.
5135 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5136 (UNSPEC_P8V_MTVSRD): Likewise.
5137 (UNSPEC_P8V_XXPERMDI): Likewise.
5138 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5139 (UNSPEC_FUSION_GPR): Likewise.
5140 (FMOVE128_GPR): New iterator for direct move.
3b9449a5 5141 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
f88fbcb9 5142 (f32_sv): Likewise.
5143 (f32_dm): Likewise.
5144 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5145 loads and direct move instructions.
5146 (zero_extendsidi2_lfiwzx): Likewise.
5147 (extendsidi2_lfiwax): Likewise.
5148 (extendsidi2_nocell): Likewise.
5149 (floatsi<mode>2_lfiwax): Likewise.
5150 (lfiwax): Likewise.
5151 (floatunssi<mode>2_lfiwzx): Likewise.
5152 (lfiwzx): Likewise.
5153 (fix_trunc<mode>_stfiwx): Likewise.
5154 (fixuns_trunc<mode>_stfiwx): Likewise.
5155 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5156 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5157 (parity<mode>2_cmpb): Set length/type attr.
5158 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5159 for 'mr.' to fast_compare.
5160 (bpermd_<mode>): Change type attr to popcnt.
5161 (p8_fmrgow_<mode>): New insns for power8 direct move support.
5162 (p8_mtvsrwz_1): Likewise.
5163 (p8_mtvsrwz_2): Likewise.
5164 (reload_fpr_from_gpr<mode>): Likewise.
5165 (p8_mtvsrd_1): Likewise.
5166 (p8_mtvsrd_2): Likewise.
5167 (p8_xxpermdi_<mode>): Likewise.
5168 (reload_vsx_from_gpr<mode>): Likewise.
5169 (reload_vsx_from_gprsf): Likewise.
5170 (p8_mfvsrd_3_<mode>): LIkewise.
5171 (reload_gpr_from_vsx<mode>): Likewise.
5172 (reload_gpr_from_vsxsf): Likewise.
5173 (p8_mfvsrd_4_disf): Likewise.
5174 (multi-word GPR splits): Do not split direct moves or quad memory
5175 operations.
5176
47c1894f 51772013-06-10 David Malcolm <dmalcolm@redhat.com>
5178
5179 * tree-into-ssa.c (interesting_blocks): Make static.
5180
a16734cd 51812013-06-10 Jakub Jelinek <jakub@redhat.com>
5182
5183 PR target/56564
5184 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5185 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5186 Use DATA_ABI_ALIGNMENT for that case instead if defined.
5187 (get_variable_align): New function.
5188 (get_variable_section, emit_bss, emit_common,
5189 assemble_variable_contents, place_block_symbol): Use
5190 get_variable_align instead of DECL_ALIGN.
5191 (assemble_noswitch_variable): Add align argument, use it
5192 instead of DECL_ALIGN.
5193 (assemble_variable): Adjust caller. Use get_variable_align
5194 instead of DECL_ALIGN.
5195 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5196 caller.
5197 (DATA_ABI_ALIGNMENT): Define.
5198 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5199 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
5200 opt is false, only return the psABI mandated alignment increase.
5201 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5202 (DATA_ABI_ALIGNMENT): ... this.
5203 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5204 (DATA_ABI_ALIGNMENT): ... this.
5205 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5206 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5207 (DATA_ABI_ALIGNMENT): ... this.
5208 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5209 * doc/tm.texi: Regenerated.
5210
ad94304e 52112013-06-10 Uros Bizjak <ubizjak@gmail.com>
5212
5213 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5214 cmp_code to construct REG_EQUAL note.
5215
1bb66e41 52162013-06-09 Jakub Jelinek <jakub@redhat.com>
5217
5218 PR target/57568
5219 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
5220 that operands[2] doesn't overlap with operands[0].
5221
85897772 52222013-06-09 David Edelsohn <dje.gcc@gmail.com>
5223 Jan Hubicka <jh@suse.cz>
5224
5225 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
5226 hack to mark symbols as used.
5227
0178c26e 52282013-06-08 Vladimir Makarov <vmakarov@redhat.com>
5229
5230 PR rtl-optimization/57559
5231 * lra-constraints.c (process_alt_operands): Don't discourage
5232 memory with known offset for offsetable memory constraint.
5233 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
5234
b0e0d151 52352013-06-08 Eric Botcazou <ebotcazou@adacore.com>
5236
5237 * varasm.c (struct oc_local_state): Reorder fields.
5238 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
5239 and adjust accordingly.
5240 (output_constructor): Reorder initialization code and adjust call to
5241 output_constructor_bitfield.
5242
55fd7c31 52432013-06-07 Jan Hubicka <jh@suse.cz>
5244
5245 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
5246
45fe2c0f 52472013-06-07 David Malcolm <dmalcolm@redhat.com>
5248
5249 * tree-object-size.c (unknown): Make const.
5250
b1032ad8 52512013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5252
5253 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
5254 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
5255 for last alternative in the cpu_facility attribute.
5256
8ee7dc6f 52572013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5258
5259 PR target/56315
5260 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
5261 (xordi3): Change operand 2 constraint to arm_xordi_operand.
5262 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
5263 * config/arm/constraints.md (Dg): New constraint.
5264 * config/arm/neon.md (xordi3_neon): Remove.
5265 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
5266 * config/arm/predicates.md (arm_xordi_operand): New predicate.
5267
0a314dcd 52682013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5269
5270 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
5271 Clean up alternatives.
5272
ce1273ae 52732013-06-07 Alan Modra <amodra@gmail.com>
5274
5275 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
5276 va_list_gpr_size.
5277
0eb8eaa7 52782013-06-07 Alan Modra <amodra@gmail.com>
5279
5280 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
5281
e0fe6977 52822013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5283
5284 * config/arm/constraints.md (Df): New constraint.
5285 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
5286 Correct length attribute for last two alternatives.
5287
c7f18d78 52882013-06-07 Alan Modra <amodra@gmail.com>
5289
5290 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5291 override user -mfp-in-toc.
5292 (offsettable_ok_by_alignment): Consider just the current access
5293 rather than the whole object, unless BLKmode. Handle
5294 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
5295 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
5296 for -mcmodel=medium.
5297 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
5298 override user -mfp-in-toc or -msum-in-toc. Default to
5299 -mno-fp-in-toc for -mcmodel=medium.
5300
556726e1 53012013-06-06 DJ Delorie <dj@redhat.com>
5302
5303 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
5304 TARGET_VALID_POINTER_MODE.
5305
7028141d 53062013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5307 Pat Haugen <pthaugen@us.ibm.com>
5308 Peter Bergner <bergner@vnet.ibm.com>
5309
5310 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5311 Document new power8 builtins.
5312
5313 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
5314 condition code register, to allow 128-bit logical operations to be
5315 done in the VSX or GPR registers.
5316 (nor<mode>3): Use the canonical form for nor.
5317 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
5318 vclz*, and vpopcnt* vector instructions.
5319 (nand<mode>3): Likewise.
5320 (orc<mode>3): Likewise.
5321 (clz<mode>2): LIkewise.
5322 (popcount<mode>2): Likewise.
5323
5324 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
5325 that only the GPRs are recognized.
5326
5327 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5328 support for new power8 builtins.
5329
5330 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
5331 builtin functions.
5332 (xscvdpspn): Likewise.
5333 (vclz): Likewise.
5334 (vclzb): Likewise.
5335 (vclzh): Likewise.
5336 (vclzw): Likewise.
5337 (vclzd): Likewise.
5338 (vpopcnt): Likewise.
5339 (vpopcntb): Likewise.
5340 (vpopcnth): Likewise.
5341 (vpopcntw): Likewise.
5342 (vpopcntd): Likewise.
5343 (vgbbd): Likewise.
5344 (vmrgew): Likewise.
5345 (vmrgow): Likewise.
5346 (eqv): Likewise.
5347 (eqv_v16qi3): Likewise.
5348 (eqv_v8hi3): Likewise.
5349 (eqv_v4si3): Likewise.
5350 (eqv_v2di3): Likewise.
5351 (eqv_v4sf3): Likewise.
5352 (eqv_v2df3): Likewise.
5353 (nand): Likewise.
5354 (nand_v16qi3): Likewise.
5355 (nand_v8hi3): Likewise.
5356 (nand_v4si3): Likewise.
5357 (nand_v2di3): Likewise.
5358 (nand_v4sf3): Likewise.
5359 (nand_v2df3): Likewise.
5360 (orc): Likewise.
5361 (orc_v16qi3): Likewise.
5362 (orc_v8hi3): Likewise.
5363 (orc_v4si3): Likewise.
5364 (orc_v2di3): Likewise.
5365 (orc_v4sf3): Likewise.
5366 (orc_v2df3): Likewise.
5367
5368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
5369 allow power8 quad mode in 64-bit.
5370 (rs6000_builtin_vectorized_function): Add support to vectorize
5371 ISA 2.07 count leading zeros, population count builtins.
5372 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
5373 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
5374 (builtin_function_type): Add vgbbd builtin function which takes an
5375 unsigned argument.
5376 (altivec_expand_vec_perm_const): Add support for new power8 merge
5377 instructions.
5378
5379 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
5380 that does not include TImdoe for use with 32-bit.
5381 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
5382 instructions.
5383 (UNSPEC_VSX_CVDPSPN): Likewise.
5384 (vsx_xscvdpspn): Likewise.
5385 (vsx_xscvspdpn): Likewise.
5386 (vsx_xscvdpspn_scalar): Likewise.
5387 (vsx_xscvspdpn_directmove): Likewise.
5388 (vsx_and<mode>3): Split logical operations into 32-bit and
5389 64-bit. Add support to do logical operations on TImode as well as
5390 VSX vector types. Allow logical operations to be done in either
5391 VSX registers or in general purpose registers in 64-bit mode. Add
5392 splitters if GPRs were used. For AND, add clobber of CCmode to
5393 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
5394 encoding.
5395 (vsx_and<mode>3_32bit): Likewise.
5396 (vsx_and<mode>3_64bit): Likewise.
5397 (vsx_ior<mode>3): Likewise.
5398 (vsx_ior<mode>3_32bit): Likewise.
5399 (vsx_ior<mode>3_64bit): Likewise.
5400 (vsx_xor<mode>3): Likewise.
5401 (vsx_xor<mode>3_32bit): Likewise.
5402 (vsx_xor<mode>3_64bit): Likewise.
5403 (vsx_one_cmpl<mode>2): Likewise.
5404 (vsx_one_cmpl<mode>2_32bit): Likewise.
5405 (vsx_one_cmpl<mode>2_64bit): Likewise.
5406 (vsx_nor<mode>3): Likewise.
5407 (vsx_nor<mode>3_32bit): Likewise.
5408 (vsx_nor<mode>3_64bit): Likewise.
5409 (vsx_andc<mode>3): Likewise.
5410 (vsx_andc<mode>3_32bit): Likewise.
5411 (vsx_andc<mode>3_64bit): Likewise.
5412 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
5413 and xxlorc instructions.
5414 (vsx_eqv<mode>3_64bit): Likewise.
5415 (vsx_nand<mode>3_32bit): Likewise.
5416 (vsx_nand<mode>3_64bit): Likewise.
5417 (vsx_orc<mode>3_32bit): Likewise.
5418 (vsx_orc<mode>3_64bit): Likewise.
5419
5420 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
5421
5422 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
5423 instruction.
5424 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
5425 (p8_vmrgow): Likewise.
5426 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
5427 GPRs to be split under VSX.
5428 (p8v_clz<mode>2): Add power8 count leading zero support.
5429 (p8v_popcount<mode>2): Add power8 population count support.
5430 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
5431 support.
5432
5433 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
5434 instruction.
5435
5436 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
5437 builtin functions.
5438 (vec_nand): Likewise.
5439 (vec_vclz): Likewise.
5440 (vec_vclzb): Likewise.
5441 (vec_vclzd): Likewise.
5442 (vec_vclzh): Likewise.
5443 (vec_vclzw): Likewise.
5444 (vec_vgbbd): Likewise.
5445 (vec_vmrgew): Likewise.
5446 (vec_vmrgow): Likewise.
5447 (vec_vpopcnt): Likewise.
5448 (vec_vpopcntb): Likewise.
5449 (vec_vpopcntd): Likewise.
5450 (vec_vpopcnth): Likewise.
5451 (vec_vpopcntw): Likewise.
5452
3620f4de 54532013-06-06 Vladimir Makarov <vmakarov@redhat.com>
5454
5455 PR rtl-optimization/57468
5456 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
5457 spilled pseudos.
5458
53d78539 54592013-06-06 Vladimir Makarov <vmakarov@redhat.com>
5460
5461 PR rtl-optimization/57459
5462 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
5463 type when setting live regs.
5464
7b1bda1c 54652013-06-06 Vladimir Makarov <vmakarov@redhat.com>
5466
5467 * config/s390/s390.opt (mlra): New option.
5468 * config/s390/s390.c (s390_decompose_address): Check displacement
5469 for all registers for LRA.
5470 (s390_secondary_reload): Don't used secondary reloads for LRA.
5471 (s390_lra_p): New function.
5472 (TARGET_LRA_P): Define.
5473 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
5474 of attribute cpu_facility to zarch for the last alternative.
5475 (*cmpmem_short): Ditto.
5476
eec6d059 54772013-06-06 Eric Botcazou <ebotcazou@adacore.com>
5478
5479 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
5480 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
5481 (arm_expand_prologue): Likewise.
5482
aa78dca5 54832013-06-06 Teresa Johnson <tejohnson@google.com>
5484
5485 PR c++/53743
5486 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
5487 as this is now done by redirect_edge_and_branch_force.
5488 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
5489 barriers, and fix interaction with splitting.
5490 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
5491 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
5492 reflect changes made in the routine.
5493 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
5494 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
5495 since this is called in cfglayout mode, and replace partition fixup
5496 with assert as that is now done by force_nonfallthru_and_redirect.
5497 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
5498 already be marked with region crossing note.
5499 (insert_section_boundary_note): Make non-static, gate on flag
5500 has_bb_partition, rewrite to also check for multiple partitions.
5501 (rest_of_handle_reorder_blocks): Remove call to
5502 insert_section_boundary_note, now done later during free_cfg.
5503 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
5504 * bb-reorder.h (insert_section_boundary_note): Declare.
5505 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
5506 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
5507 invoke insert_section_boundary_note.
5508 (try_redirect_by_replacing_jump): Remove unnecessary
5509 check for region crossing note.
5510 (fixup_partition_crossing): New function.
5511 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
5512 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
5513 in non-cfglayout mode.
5514 (force_nonfallthru_and_redirect): Fixup partition boundaries,
5515 remove old code that tried to do this. Emit barrier correctly
5516 when we are in cfglayout mode.
5517 (last_bb_in_partition): New function.
5518 (rtl_split_edge): Correctly fixup partition boundaries.
5519 (commit_one_edge_insertion): Remove old code that tried to
5520 fixup region crossing edge since this is now handled in
5521 split_block, and set up insertion point correctly since
5522 block may now end in a jump.
5523 (verify_hot_cold_block_grouping): Guard against checking when not in
5524 linearized RTL mode.
5525 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
5526 notes.
5527 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
5528 rtl_verify_flow_info, so not called in cfglayout mode.
5529 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
5530 (fixup_reorder_chain): Remove old code that attempted to fixup region
5531 crossing note as this is now handled in force_nonfallthru_and_redirect.
5532 (duplicate_insn_chain): Don't duplicate switch section notes.
5533 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
5534 note.
5535 * basic-block.h (emit_barrier_after_bb): Declare.
5536
940beb92 55372013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5538
5539 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
5540 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
5541 arm_usatsihi): Adjust alternatives for arm_restrict_it.
5542
27153331 55432013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5544
5545 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
5546 where appropriate.
5547 * config/arm/ldmstm.md: Regenerate.
5548
5a109f6c 55492013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5550
5551 * config/arm/sync.md (atomic_loaddi_1):
5552 Disable predication for arm_restrict_it.
5553 (arm_load_exclusive<mode>): Likewise.
5554 (arm_load_exclusivesi): Likewise.
5555 (arm_load_exclusivedi): Likewise.
5556 (arm_load_acquire_exclusive<mode>): Likewise.
5557 (arm_load_acquire_exclusivesi): Likewise.
5558 (arm_load_acquire_exclusivedi): Likewise.
5559 (arm_store_exclusive<mode>): Likewise.
5560 (arm_store_exclusive<mode>): Likewise.
5561 (arm_store_release_exclusivedi): Likewise.
5562 (arm_store_release_exclusive<mode>): Likewise.
5563
e09e0df2 55642013-06-06 Richard Biener <rguenther@suse.de>
5565
5566 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
5567 after LTO_null.
5568 (lto_tag_is_tree_code_p): Adjust.
5569 (lto_tag_is_gimple_code_p): Likewise.
5570 (lto_gimple_code_to_tag): Likewise.
5571 (lto_tag_to_gimple_code): Likewise.
5572 (lto_tree_code_to_tag): Likewise.
5573 (lto_tag_to_tree_code): Likewise.
5574 * data-streamer.h (streamer_write_hwi_in_range): Use
5575 uhwi streaming to stream the normalized range.
5576 (streamer_read_hwi_in_range): Likewise.
5577
c7a58118 55782013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5579
5580 * config/arm/arm.md (enabled_for_depr_it): New attribute.
5581 (predicable_short_it): Likewise.
5582 (predicated): Likewise.
5583 (enabled): Handle above.
5584 (define_cond_exec): Set predicated attribute to yes.
5585
910d871e 55862013-06-05 Mike Stump <mikestump@comcast.net>
5587
5588 * gdbinit.in (__FUNCTION__): Add.
5589
cf1f9f8b 55902013-06-05 Uros Bizjak <ubizjak@gmail.com>
5591
5592 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
5593 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
5594
709f8747 55952013-06-05 Jan Hubicka <jh@suse.cz>
5596
5597 * varasm.c (mark_decl_referenced): Revert the removal until targets
5598 are fixed.
5599
bc5f662c 56002013-06-05 David Edelsohn <dje.gcc@gmail.com>
5601
5602 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
5603 instead of mark_decl_referenced.
5604
6a1c0403 56052013-06-05 Jan Hubicka <jh@suse.cz>
5606
5607 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
5608 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
5609 and symtab_used_from_object_file_p.
5610 (cgraph_make_node_local_1): Clear forced_by_abi.
5611 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
5612 * cgraph.h (symtab_node_base): Add forced_by_abi.
5613 (decide_is_variable_needed): Remove.
5614 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
5615 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
5616 (decide_is_symbol_needed): ... this one; handle symbols in general;
5617 always analyze virtuals; honnor forced_by_abi.
5618 (cgraph_finalize_function): Update.
5619 (varpool_finalize_decl): Update.
5620 (symbol_defined_and_needed): Remove.
5621 (analyze_functions): Update.
5622 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
5623 output_refs, input_overwrite_node): Handle forced_by_abi.
5624 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
5625 (address_taken_from_non_vtable_p): ... this one.
5626 (comdat_can_be_unshared_p_1): New function.
5627 (cgraph_comdat_can_be_unshared_p): Rename to ...
5628 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
5629 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
5630 (function_and_variable_visibility): Clear forced_by_abi as needed.
5631 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
5632 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6a1c0403 5633 * symtab.c (dump_symtab_base): Dump forced_by_abi.
5634 * varpool.c (decide_is_variable_needed): Remove.
5635
889fe57c 56362013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5637
5638 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
5639 (arm_option_override): Override arm_restrict_it where appropriate.
5640 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
5641 * config/arm/arm.opt (mrestrict-it): New command-line option.
5642 * doc/invoke.texi: Document -mrestrict-it.
5643
9d34f097 56442013-06-05 David Malcolm <dmalcolm@redhat.com>
5645
5646 * tsan.c (tsan_atomic_table): Make const.
5647
4b927f54 56482013-06-05 Richard Biener <rguenther@suse.de>
5649
5650 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
5651 index associated with the tree we are supposed to replace.
5652 * tree-streamer-out.c (pack_ts_base_value_fields): Output
5653 TREE_ASM_WRITTEN as zero for everything but SSA names.
5654
94771494 56552013-06-05 David Malcolm <dmalcolm@redhat.com>
5656
5657 * tree-ssa-structalias.c (call_stmt_vars): Make static.
5658
f1a7feee 56592013-06-04 Jan Hubicka <jh@suse.cz>
5660
5661 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
5662 (input_node, input_varpool_node): Handle correctly external same
5663 body aliases.
5664 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
5665 nodes at ltrans stage.
5666
9817f2cd 56672013-06-04 Jan Hubicka <jh@suse.cz>
5668
5669 * ipa-inline.c (update_caller_keys): Fix availability test.
5670 (update_callee_keys): Likewise.
5671 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
5672 to follow ELF standard.
5673
60b34f30 56742013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
5675
5676 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
5677 (mips64r5900el-*-elf*): New configurations.
5678 * config/mips/mips-cpus.def (r5900): New processor.
5679 * config/mips/mips-tables.opt: Regenerate.
5680 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
5681 (mips_issue_rate): Handle PROCESSOR_R5900.
5682 (mips_reorg_process_insns): Force reorder mode for the R5900.
5683 * config/mips/mips.h (TARGET_MIPS5900): Define.
5684 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
5685 TARGET_MIPS5900.
5686 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
5687 TARGET_MIPS5900.
5688 * config/mips/mips.md (processor): Add r5900.
5689 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
5690
dae88020 56912013-06-04 Ian Bolton <ian.bolton@arm.com>
5692
5693 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
5694 into function to generate MOVI instruction.
cf1f9f8b 5695 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
dae88020 5696 (aarch64_preferred_simd_mode): Turn into wrapper.
5697 (aarch64_output_scalar_simd_mov_immediate): New function.
5698 * config/aarch64/aarch64-protos.h: Add prototype for above.
5699
04b042b2 57002013-06-04 Ian Bolton <ian.bolton@arm.com>
5701
5702 * config/aarch64/aarch64.c (simd_immediate_info): Remove
5703 element_char member.
5704 (sizetochar): Return signed char.
5705 (aarch64_simd_valid_immediate): Remove elchar and other
5706 unnecessary variables.
5707 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
5708 Calculate element_char as required.
5709 * config/aarch64/aarch64-protos.h: Update and move prototype
5710 for aarch64_output_simd_mov_immediate.
5711 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5712 Update arguments.
5713
fc3eb658 57142013-06-04 Ian Bolton <ian.bolton@arm.com>
5715
5716 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
5717 information completed by aarch64_simd_valid_immediate.
5718 (aarch64_legitimate_constant_p): Update arguments.
5719 (aarch64_simd_valid_immediate): Work with struct rather than many
5720 pointers.
5721 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
5722 (aarch64_simd_make_constant): Update arguments.
5723 (aarch64_output_simd_mov_immediate): Work with struct rather than
5724 many pointers. Output immediate directly rather than as operand.
5725 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
5726 Update prototype.
5727 * config/aarch64/constraints.md (Dn): Update arguments.
5728
74c40875 57292013-06-04 Ian Bolton <ian.bolton@arm.com>
5730
5731 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
5732 longer static.
5733 (aarch64_simd_immediate_valid_for_move): Remove.
5734 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
5735 (aarch64_simd_make_constant): Update call.
5736 (aarch64_output_simd_mov_immediate): Update call.
5737 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
5738 Add prototype.
5739 * config/aarch64/constraints.md (Dn): Update call.
5740
4eb118f1 57412013-06-04 Ian Bolton <ian.bolton@arm.com>
5742
5743 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
5744 return type to bool for prototype.
5745 (aarch64_legitimate_constant_p): Check for true instead of not -1.
5746 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
5747 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
5748
3721f3a0 57492013-06-04 Catherine Moore <clm@codesourcery.com>
5750
5751 * config/mips/mips.opt (meva): New.
cf1f9f8b 5752 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
3721f3a0 5753 (ASM_SPEC): Handle -meva.
5754 * doc/invoke.texi (meva): Document.
5755
6c8e6c47 57562013-06-04 Alan Modra <amodra@gmail.com>
5757
5758 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
5759 constant output.
5760
8c45824d 57612013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5762
5763 * rtl.def: Add extra fourth optional field to define_cond_exec.
5764 * gensupport.c (process_one_cond_exec): Process attributes from
5765 define_cond_exec.
5766 * doc/md.texi: Document fourth field in define_cond_exec.
5767
8744b4c5 57682013-06-04 Eric Botcazou <ebotcazou@adacore.com>
5769
5770 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
5771 out the processing order as in store_bit_field_1.
5772
a55453ec 57732013-06-04 Jan Hubicka <jh@suse.cz>
5774
5775 PR middle-end/57500
5776 * cgraphunit.c (cgraph_process_same_body_aliases): Create
5777 non-VAR_DECL node if it does not exist yet.
5778
53c5ddd2 57792013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
5780
5781 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
5782 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
5783 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
5784 target_cpu_default setting.
5785
a21425b5 57862013-06-03 Teresa Johnson <tejohnson@google.com>
5787
5788 * dumpfile.c (opt_info_switch_p): Change -fopt-info
5789 default to -fopt-info=optimized instead of all.
5790 * doc/invoke.texi: Ditto.
5791 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
5792 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
5793 (execute_vect_slp): Emit BB vectorization success under
5794 MSG_OPTIMIZED_LOCATIONS.
5795 * tree-vect-slp.c (vect_slp_transform_bb): Change
5796 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
5797 * tree-vect-loop.c (vect_transform_loop): Ditto.
5798
79f0505e 57992013-06-03 Jason Merrill <jason@redhat.com>
5800
5801 PR c++/57415
5802 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5803 Use TARGET_EXPR for C++.
5804
c2066d40 58052013-06-03 Jakub Jelinek <jakub@redhat.com>
5806
5807 PR rtl-optimization/57268
8744b4c5 5808 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
c2066d40 5809 if DEBUG_INSN_P (insn).
5810
5811 Reapply
5812 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
5813
5814 PR rtl-optimization/57268
5815 * sched-deps.c (sched_analyze_2): Flush dependence lists if
5816 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
5817
fbcee6d1 58182013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
5819
5820 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
5821 (ix86_avoid_lea_for_addr): Likewise.
5822 (exact_dependency_1): Likewise.
5823 (ix86_adjust_cost): Likewise.
5824 (swap_top_of_ready_list): Fix formatting and !reload_completed check
5825 removed.
5826 (do_reorder_for_imul): Fix typo, formatting and
5827 !reload_completed check removed.
5828 (ix86_sched_reorder): Fix typo and formatting.
5829 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
5830 list.
5831
e018d84c 58322013-06-03 Sofiane Naci <sofiane.naci@arm.com>
5833
5834 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
5835
9322e825 58362013-06-03 Eric Botcazou <ebotcazou@adacore.com>
5837
5838 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
5839 <STRING_CST>: Likewise.
5840 <VECTOR_CST>: Likewise.
5841
f3b719b3 58422013-06-01 Janus Weil <janus@gcc.gnu.org>
5843 Mikael Morin <mikael@gcc.gnu.org>
5844
5845 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
5846 * config.in: Regenerated.
5847 * configure: Regenerated.
5848
4cd8ae4b 58492013-06-01 Jan Hubicka <jh@suse.cz>
5850
5851 PR middle-end/57366
5852 * cgraphunit.c (compile): When weakref is not supported,
5853 set up transparent aliases before final output pass.
5854 * varasm.c (assemble_alias): Do not try to do it here.
5855
e4f633f7 58562013-06-01 Jan Hubicka <jh@suse.cz>
5857
5858 PR middle-end/57467
5859 * passes.c (for_per_function): Skip unanalyzed functions.
5860
48669653 58612013-06-01 Jan Hubicka <jh@suse.cz>
5862
5863 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
5864 (lto_symtab_merge_symbols_1): ... this one.
5865 (lto_symtab_merge_cgraph_nodes): Rename to ...
5866 (lto_symtab_merge_symbols): ... this one; simplify.
5867 * cgraph.c (same_body_aliases_done): Rename to ...
5868 (cpp_implicit_aliases_done): ... this one.
5869 (cgraph_create_function_alias): Update.
5870 (cgraph_same_body_alias): Update.
cf1f9f8b 5871 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
48669653 5872 (verify_edge_corresponds_to_fndecl): Simplify.
cf1f9f8b 5873 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
48669653 5874 (cgraph_node): Remove same_body_alias.
5875 (varpool_node): Remove alias_of and extra_name_alias.
5876 (same_body_aliases_done): Rename to ..
5877 (cpp_implicit_aliases_done): ... this one.
5878 (symtab_alias_ultimate_target): Add default parameter.
5879 (symtab_resolve_alias): New function.
5880 (fixup_same_cpp_alias_visibility): Declare.
5881 (cgraph_function_node): Add default parameter.
5882 (cgraph_node_asm_name): Likewise.
5883 (cgraph_function_or_thunk_node): Add default parameter; do
5884 not ICE when it is NULL.
5885 (varpool_variable_node): Likewise.
5886 * tree-emutls.c (create_emultls_var): Update.
5887 (ipa_lower_emutls): Update.
5888 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
5889 (cgraph_reset_node): Reset alias info.
5890 (cgraph_finalize_function): Update.
5891 (fixup_same_cpp_alias_visibility): Move to symtab.c.
5892 (analyze_function): Simplify.
5893 (cgraph_process_same_body_aliases): Simplify.
5894 (analyze_functions): Fixup same body aliases.
5895 (handle_alias_pairs): Simplify.
5896 (assemble_thunk): Update.
5897 (assemble_thunks_and_aliases): Update.
5898 (output_weakrefs): Rewrite.
5899 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
5900 (lto_output_varpool_node): Likewise.
5901 (compute_ltrans_boundary): Remve assert.
5902 (get_alias_symbol): New functoin.
5903 (input_node): Rewrite alias handling.
5904 (input_varpool_node): Likewise.
5905 * ipa-pure-const.c (propagate_pure_const): Fix formating.
5906 * ipa.c (process_references): Handle weakrefs correctly.
5907 (symtab_remove_unreachable_nodes): Likewise.
5908 * trans-mem.c (get_cg_data): Update.
5909 (ipa_tm_create_version_alias): Update.
5910 (ipa_tm_execute): Update.
5911 * symtab.c (dump_symtab_base): Dump aliases.
5912 (verify_symtab_base): Verify aliases.
5913 (symtab_node_availability): New function.
5914 (symtab_alias_ultimate_target): Simplify.
5915 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
5916 handle all the fixup cases.
5917 (symtab_resolve_alias): New function.
5918 * passes.c (ipa_write_summaries): Handle weakrefs.
5919 * varpool.c (varpool_analyze_node): Simplify.
5920 (assemble_aliases): Update.
5921 (varpool_create_variable_alias): Simplify.
5922 (varpool_extra_name_alias): Simplify.
5923 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
5924 (lto_symtab_merge_symbols): ... this one.
5925
a47321b7 59262013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
5927
5928 Revert
5929 PR rtl-optimization/57268
5930 * sched-deps.c (sched_analyze_2): Flush dependence lists if
5931 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
5932
d48be958 59332013-06-01 Tobias Burnus <burnus@net-b.de>
5934
5935 Partially reverted:
5936 2013-05-31 Tobias Burnus <burnus@net-b.de>
5937
5938 PR middle-end/57073
5939 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
5940 further up.
5941
62c26259 59422013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
5943
5944 PR rtl-optimization/57268
5945 * sched-deps.c (sched_analyze_2): Flush dependence lists if
5946 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
5947
fb80316e 59482013-05-31 Eric Botcazou <ebotcazou@adacore.com>
5949
5950 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
5951 unordered comparison operators when -fno-trapping-math is in effect
5952 on the e500.
5953 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
5954 and implement unordered comparison operators properly on the e500.
5955
efb076ad 59562013-05-31 Eric Botcazou <ebotcazou@adacore.com>
5957
5958 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
5959 for constant scalar integers.
5960 (simplify_relational_operation_1): Likewise.
5961
317121db 59622013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
5963
5964 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
5965 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
5966 Fix comment.
5967
f3de38f1 59682013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
5969 Igor Zamyatin <igor.zamyatin@intel.com>
5970
5971 Silvermont (SLM) architecture performance tuning.
5972 * config/i386/i386.h (enum ix86_tune_indices): Add
5973 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
5974 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
5975
5976 * config/i386/i386.c (initial_ix86_tune_features)
5977 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
5978 (ix86_lea_outperforms): Handle Silvermont tuning.
5979 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
5980 call.
5981 (ix86_use_lea_for_mov): Likewise.
5982 (ix86_avoid_lea_for_addr): Likewise.
5983 (ix86_lea_for_add_ok): Likewise.
5984 (exact_dependency_1): New function.
5985 (exact_store_load_dependency): Likewise.
5986 (ix86_adjust_cost): Handle Silvermont tuning.
5987 (do_reoder_for_imul): Likewise.
5988 (swap_top_of_ready_list): New function.
5989 (ix86_sched_reorder): Changed to handle Silvermont tuning.
5990
5991 * config/i386/i386.md (peepholes that split memory operand in fp
5992 converts): New.
5993
78d82aec 59942013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
5995
5996 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5997 Remove un-necessary braces.
5998
12a60807 59992013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
6000
6001 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6002 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6003
6dfe7d53 60042013-05-31 Tobias Burnus <burnus@net-b.de>
6005
6006 PR middle-end/57073
cf1f9f8b 6007 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6dfe7d53 6008
74d6113f 60092013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6010
6011 PR target/56315
6012 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6013 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6014 * config/arm/neon.md (iordi3_neon): Remove.
6015 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6016 * config/arm/predicates.md (imm_for_neon_logic_operand):
6017 Move to earlier in the file.
6018 (neon_logic_op2): Likewise.
6019 (arm_iordi_operand_neon): New predicate.
6020
5cfd0eb5 60212013-05-31 Richard Biener <rguenther@suse.de>
6022
6023 PR tree-optimization/57478
6024 PR tree-optimization/57453
6025 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6026 are life as well.
6027
789a3e1c 60282013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
6029
6030 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6031 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6032
377db285 60332013-05-30 Tobias Burnus <burnus@net-b.de>
6034 Thomas Koenig <tkoenig@gcc.gnu.org>
6035
6036 PR middle-end/57073
6037 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6038 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6039
8a139381 60402013-05-30 Steven Bosscher <steven@gcc.gnu.org>
6041
377db285 6042 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8a139381 6043
4a2ca8f3 60442013-05-30 Vladimir Makarov <vmakarov@redhat.com>
6045
6046 * target.def (register_usage_leveling_p): New hook.
6047 * targhooks.c (default_register_usage_leveling_p): New.
6048 * targhooks.h (default_register_usage_leveling_p): New prototype.
6049 * lra-assigns.c (register_usage_leveling_p): Use the hook.
6050 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6051 * doc/tm.texi: Update.
6052 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6053
fb08a7c6 60542013-05-30 Ian Bolton <ian.bolton@arm.com>
6055
6056 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6057 (*insv_reg<mode>): New define_insn.
6058
5bea3269 60592013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
6060
6061 PR rtl-optimization/57439
6062 * postreload.c (move2add_valid_value_p): Check that we have
6063 a zero subreg_regno_offset when accessing the register in
6064 the requested mode.
6065
564c5b22 60662013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6067 Igor Zamyatin <igor.zamyatin@intel.com>
6068
6069 Silvermont (SLM) architecture pipeline model, tuning and
6070 insn selection.
6071 * config.gcc: Add slm config options and target.
6072
6073 * config/i386/slm.md: New.
6074
6075 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6076
6077 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
6078 PROCESSOR_SLM.
6079 (ix86_target_macros_internal): Likewise.
6080
6081 * gcc/config/i386/i386.c (slm_cost): New cost.
6082 (m_SLM): New macro flag.
6083 (initial_ix86_tune_features): Set m_SLM.
6084 (x86_accumulate_outgoing_args): Likewise.
6085 (x86_arch_always_fancy_math_387): Likewise.
6086 (processor_target_table): Add slm cost.
6087 (cpu_names): Add slm cpu name.
6088 (x86_option_override_internal): Set SLM ISA.
6089 (ix86_issue_rate): New case PROCESSOR_SLM.
6090 (ia32_multipass_dfa_lookahead): Likewise.
6091 (fold_builtin_cpu): Add slm.
6092
6093 * config/i386/i386.h (TARGET_SLM): New target macro.
6094 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6095 (processor_type): Add PROCESSOR_SLM.
6096
6097 * config/i386/i386.md (cpu): Add new value "slm".
6098 (slm.md): Include slm.md.
6099
0686440e 61002013-05-30 Bernd Schmidt <bernds@codesourcery.com>
6101 Zhenqiang Chen <zhenqiang.chen@linaro.org>
6102
6103 * config/arm/arm-protos.h: Add and update function protos.
6104 * config/arm/arm.c (use_simple_return_p): New added.
6105 (thumb2_expand_return): Check simple_return flag.
6106 * config/arm/arm.md: Add simple_return and conditional simple_return.
6107 * config/arm/iterators.md: Add iterator for return and simple_return.
6108
cf94bbe6 61092013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
6110
6111 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6112 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6113 (arm_emit_vfp_multi_reg_pop): Likewise.
6114 (thumb2_emit_ldrd_pop): Likewise.
6115 (arm_expand_epilogue): Add misc REG_CFA notes.
6116 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6117
7b7a707b 61182013-05-29 Lawrence Crowl <crowl@google.com>
6119
6120 * config/arm/t-arm: Update for below.
6121
6122 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6123 Change type to hash_table. Update dependent calls and types.
6124
6125 * config/i386/t-cygming: Update for below.
6126
6127 * config/i386/t-interix: Update for below.
6128
6129 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6130 Change type to hash_table. Update dependent calls and types.
6131 (i386_find_on_wrapper_list::wrappers): Likewise.
6132
6133 * config/ia64/t-ia64: Update for below.
6134
6135 * config/ia64/ia64.c (bundle_state_table):
6136 Change type to hash_table. Update dependent calls and types.
6137
6138 * config/mips/mips.c (mips_reorg_process_insns::htab):
6139 Change type to hash_table. Update dependent calls and types.
6140
6141 * config/sol2.c (solaris_comdat_htab):
6142 Change type to hash_table. Update dependent calls and types.
6143
6144 * config/t-sol2: Update for above.
6145
e6c8824e 61462013-05-29 Teresa Johnson <tejohnson@google.com>
6147
6148 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6149 functions are not yet marked as defined.
6150
ae52d0de 61512013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6152 Pat Haugen <pthaugen@us.ibm.com>
6153 Peter Bergner <bergner@vnet.ibm.com>
6154
6155 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6156 instructions.
6157 (VEC_A): Likewise.
6158 (VEC_C): Likewise.
6159 (vrotl<mode>3): Likewise.
6160 (vashl<mode>3): Likewise.
6161 (vlshr<mode>3): Likewise.
6162 (vashr<mode>3): Likewise.
6163
6164 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6165 support for power8 V2DI builtins.
6166
6167 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6168 power8 V2DI builtins.
6169 (vupkhsw): Likewise.
6170 (vupklsw): Likewise.
6171 (vaddudm): Likewise.
6172 (vminsd): Likewise.
6173 (vmaxsd): Likewise.
6174 (vminud): Likewise.
6175 (vmaxud): Likewise.
6176 (vpkudum): Likewise.
6177 (vpksdss): Likewise.
6178 (vpkudus): Likewise.
6179 (vpksdus): Likewise.
6180 (vrld): Likewise.
6181 (vsld): Likewise.
6182 (vsrd): Likewise.
6183 (vsrad): Likewise.
6184 (vsubudm): Likewise.
6185 (vcmpequd): Likewise.
6186 (vcmpgtsd): Likewise.
6187 (vcmpgtud): Likewise.
6188 (vcmpequd_p): Likewise.
6189 (vcmpgtsd_p): Likewise.
6190 (vcmpgtud_p): Likewise.
6191 (vupkhsw): Likewise.
6192 (vupklsw): Likewise.
6193 (vaddudm): Likewise.
6194 (vmaxsd): Likewise.
6195 (vmaxud): Likewise.
6196 (vminsd): Likewise.
6197 (vminud): Likewise.
6198 (vpksdss): Likewise.
6199 (vpksdus): Likewise.
6200 (vpkudum): Likewise.
6201 (vpkudus): Likewise.
6202 (vrld): Likewise.
6203 (vsld): Likewise.
6204 (vsrad): Likewise.
6205 (vsrd): Likewise.
6206 (vsubudm): Likewise.
6207
6208 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6209 support for power8 V2DI instructions.
6210
6211 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6212 power8 V2DI instructions. Combine pack and unpack insns to use an
6213 iterator for each mode. Check whether a particular mode supports
6214 Altivec instructions instead of just checking TARGET_ALTIVEC.
6215 (UNSPEC_VPKUWUM): Likewise.
6216 (UNSPEC_VPKSHSS): Likewise.
6217 (UNSPEC_VPKSWSS): Likewise.
6218 (UNSPEC_VPKUHUS): Likewise.
6219 (UNSPEC_VPKSHUS): Likewise.
6220 (UNSPEC_VPKUWUS): Likewise.
6221 (UNSPEC_VPKSWUS): Likewise.
6222 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
6223 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
6224 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
6225 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
6226 (UNSPEC_VUPKHSB): Likewise.
6227 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
6228 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
6229 (UNSPEC_VUPKHSH): Likewise.
6230 (UNSPEC_VUPKLSB): Likewise.
6231 (UNSPEC_VUPKLSH): Likewise.
6232 (VI2): Likewise.
6233 (VI_char): Likewise.
6234 (VI_scalar): Likewise.
6235 (VI_unit): Likewise.
6236 (VP): Likewise.
6237 (VP_small): Likewise.
6238 (VP_small_lc): Likewise.
6239 (VU_char): Likewise.
6240 (add<mode>3): Likewise.
6241 (altivec_vaddcuw): Likewise.
6242 (altivec_vaddu<VI_char>s): Likewise.
6243 (altivec_vadds<VI_char>s): Likewise.
6244 (sub<mode>3): Likewise.
6245 (altivec_vsubcuw): Likewise.
6246 (altivec_vsubu<VI_char>s): Likewise.
6247 (altivec_vsubs<VI_char>s): Likewise.
6248 (altivec_vavgs<VI_char>): Likewise.
6249 (altivec_vcmpbfp): Likewise.
6250 (altivec_eq<mode>): Likewise.
6251 (altivec_gt<mode>): Likewise.
6252 (altivec_gtu<mode>): Likewise.
6253 (umax<mode>3): Likewise.
6254 (smax<mode>3): Likewise.
6255 (umin<mode>3): Likewise.
6256 (smin<mode>3): Likewise.
6257 (altivec_vpkuhum): Likewise.
6258 (altivec_vpkuwum): Likewise.
6259 (altivec_vpkshss): Likewise.
6260 (altivec_vpkswss): Likewise.
6261 (altivec_vpkuhus): Likewise.
6262 (altivec_vpkshus): Likewise.
6263 (altivec_vpkuwus): Likewise.
6264 (altivec_vpkswus): Likewise.
6265 (altivec_vpks<VI_char>ss): Likewise.
6266 (altivec_vpks<VI_char>us): Likewise.
6267 (altivec_vpku<VI_char>us): Likewise.
6268 (altivec_vpku<VI_char>um): Likewise.
6269 (altivec_vrl<VI_char>): Likewise.
6270 (altivec_vsl<VI_char>): Likewise.
6271 (altivec_vsr<VI_char>): Likewise.
6272 (altivec_vsra<VI_char>): Likewise.
6273 (altivec_vsldoi_<mode>): Likewise.
6274 (altivec_vupkhsb): Likewise.
6275 (altivec_vupkhs<VU_char>): Likewise.
6276 (altivec_vupkls<VU_char>): Likewise.
6277 (altivec_vupkhsh): Likewise.
6278 (altivec_vupklsb): Likewise.
6279 (altivec_vupklsh): Likewise.
6280 (altivec_vcmpequ<VI_char>_p): Likewise.
6281 (altivec_vcmpgts<VI_char>_p): Likewise.
6282 (altivec_vcmpgtu<VI_char>_p): Likewise.
6283 (abs<mode>2): Likewise.
6284 (vec_unpacks_hi_v16qi): Likewise.
6285 (vec_unpacks_hi_v8hi): Likewise.
6286 (vec_unpacks_lo_v16qi): Likewise.
6287 (vec_unpacks_hi_<VP_small_lc>): Likewise.
6288 (vec_unpacks_lo_v8hi): Likewise.
6289 (vec_unpacks_lo_<VP_small_lc>): Likewise.
6290 (vec_pack_trunc_v8h): Likewise.
6291 (vec_pack_trunc_v4si): Likewise.
6292 (vec_pack_trunc_<mode>): Likewise.
6293
6294 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
6295 V2DI builtins.
6296 (vec_vmaxsd): Likewise.
6297 (vec_vmaxud): Likewise.
6298 (vec_vminsd): Likewise.
6299 (vec_vminud): Likewise.
6300 (vec_vpksdss): Likewise.
6301 (vec_vpksdus): Likewise.
6302 (vec_vpkudum): Likewise.
6303 (vec_vpkudus): Likewise.
6304 (vec_vrld): Likewise.
6305 (vec_vsld): Likewise.
6306 (vec_vsrad): Likewise.
6307 (vec_vsrd): Likewise.
6308 (vec_vsubudm): Likewise.
6309 (vec_vupkhsw): Likewise.
6310 (vec_vupklsw): Likewise.
6311
15ca8f90 63122013-05-29 Jan Hubicka <jh@suse.cz>
6313
6314 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
6315 flags; reorder rest of fields in more consistent way.
6316 (varpool_node): Remove analyzed, finalized and alias.
6317 (cgraph_ndoe): Likewise.
6318 (symtab_alias_ultimate_target): New function.
6319 (cgraph_function_node): Move offline.
6320 (cgraph_reset_node): Declare.
6321 (cgraph_comdat_can_be_unshared_p): Remove.
6322 (varpool_remove_initializer): Declare.
6323 (varpool_first_defined_variable, varpool_next_defined_variable
6324 cgraph_first_defined_function, cgraph_next_defined_function): Update.
6325 (cgraph_function_with_gimple_body_p): Update.
6326 (varpool_all_refs_explicit_p): Update.
6327 (symtab_alias_target): New function.
6328 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
6329 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
cf1f9f8b 6330 (cgraph_function_or_thunk_node): Simplify using
6331 symtab_alias_ultimate_target.
15ca8f90 6332 (varpool_variable_node): Likewise.
6333 * cgraph.c (cgraph_create_function_alias): Update.
6334 (cgraph_add_thunk): Update.
6335 (cgraph_remove_node): Update.
6336 (dump_cgraph_node): Do not dump removed flags.
6337 (cgraph_function_body_availability): Update.
6338 (cgraph_propagate_frequency): Update.
6339 (verify_cgraph_node): Check sanity of local flag.
6340 (cgraph_function_node): Move here from cgraph.h; revamp for
6341 cgraph_function_or_thunk_node.
6342 * lto-symtab.c (lto_varpool_replace_node): Update.
6343 (lto_symtab_resolve_can_prevail_p): Update.
6344 (lto_symtab_merge_cgraph_nodes): Update.
6345 * ipa-cp.c (determine_versionability, initialize_node_lattices,
6346 propagate_constants_accross_call, devirtualization_time_bonus,
6347 ipcp_propagate_stage): Update.
6348 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
cf1f9f8b 6349 * ipa-inline-transform.c (clone_inlined_nodes,
6350 preserve_function_body_p): Update.
15ca8f90 6351 * ipa-reference.c (propagate): Update.
6352 (write_node_summary_p): Update.
6353 * toplev.c (wrapup_global_declaration_2): Update.
6354 * cgraphunit.c (cgraph_analyze_function): Rename to ...
6355 (analyze_function) ... this one.
6356 (cgraph_process_new_functions): Update.
6357 (cgraph_reset_node): Export.
6358 (cgraph_finalize_function): Update.
6359 (cgraph_add_new_function): Update.
6360 (process_function_and_variable_attributes): Update.
6361 (varpool_finalize_decl): Update.
6362 (symbol_finalized): Remove.
6363 (symbol_finalized_and_needed): Rename to ...
6364 (symbol_defined_and_needed): ... update.
6365 (cgraph_analyze_functions): Update.
6366 (handle_alias_pairs): Update.
6367 (mark_functions_to_output): Update.
6368 (assemble_thunk): Update.
6369 (output_in_order): Update.
6370 (output_weakrefs): Update.
6371 (finalize_compilation_unit): Update.
6372 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
6373 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
6374 input_node, input_varpool_node): Update.
6375 * dbxout.c (dbxout_expand_expr): Update.
6376 * cgraphclones.c (cgraph_clone_node): Update.
6377 (cgraph_copy_node_for_versioning): Update.
6378 (cgraph_materialize_clone): Update.
6379 (cgraph_materialize_all_clones): Update.
6380 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
6381 propagate_pure_const, propagate_nothrow): Update.
6382 * lto-streamer-out.c (lto_output, write_symbol): Update.
6383 * ipa-utils.c (ipa_reverse_postorder): Update.
6384 * ipa-inline.c (can_inline_edge_p): Update.
6385 (update_caller_keys, ipa_inline): Update.
6386 * dwarf2out.c (reference_to_unused,
6387 premark_types_used_by_global_vars_helper): Update.
6388 * tree-eh.c (tree_could_trap_p): Update.
6389 * ipa-split.c (consider_split, execute_split_functions): Update.
6390 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
cf1f9f8b 6391 has_addr_references_p): Update; move ahead in file for better
6392 readability.
15ca8f90 6393 (process_references): Simplify.
6394 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
6395 bodies are removed.
6396 (cgraph_comdat_can_be_unshared_p): Make static.
6397 (cgraph_externally_visible_p): Update.
6398 (varpool_externally_visible_p): Update.
6399 (function_and_variable_visibility): Update.
6400 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
6401 ipa_tm_mark_force_output_node): Update.
6402 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
6403 estimate_edge_devirt_benefit, inline_generate_summary,
6404 inline_write_summary): Update.
6405 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
6406 * ipa-prop.c (ipa_compute_jump_functions): Update.
cf1f9f8b 6407 (ipa_print_node_params, ipa_prop_read_section,
6408 ipa_update_after_lto_read, read_replacements_section): Update.
15ca8f90 6409 * varasm.c (mark_decl_referenced): Update.
6410 (assemble_alias, dump_tm_clone_pairs): Update.
6411 * tree-inline.c (copy_bb): Update.
6412 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
6413 Update.
6414 * symtab.c (dump_symtab_base): Print new flags.
6415 (verify_symtab_base): Verify new flags.
6416 (symtab_alias_ultimate_target): New function.
6417 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
6418 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
6419 Update.
cf1f9f8b 6420 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
6421 Update.
15ca8f90 6422 * i386.c (ix86_get_function_versions_dispatcher,
6423 ix86_generate_version_dispatcher_body): Update.
6424 (fold_builtin_cpu): Use varpool_add_new_variable.
6425 * varpool.c (varpool_remove_initializer): Break out from ...
6426 (varpool_remove_node): ... this one.
6427 (dump_varpool_node, varpool_node_for_asm,
6428 cgraph_variable_initializer_availability, varpool_analyze_node,
6429 varpool_assemble_decl, varpool_remove_unreferenced_decls,
cf1f9f8b 6430 varpool_finalize_named_section_flags, varpool_create_variable_alias):
6431 Update.
15ca8f90 6432
993fd1eb 64332013-05-29 Jan Hubicka <jh@suse.cz>
6434
6435 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
6436
b3c1a1cd 64372013-05-29 Easwaran Raman <eraman@google.com>
6438
6439 PR tree-optimization/57442
6440 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
6441 when control exits the main loop.
6442
f0964309 64432013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
6444
6445 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
6446 and RX600.
cf1f9f8b 6447 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
f0964309 6448 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
6449 * rx/t-rx: Add rx100 under multi library matches option for nofpu
6450 option.
6451
259c15e6 64522013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6453
6454 PR tree-optimization/57441
6455 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
6456 Don't limit size of incr_vec to number of candidates.
6457
0985b4a2 64582013-05-29 Steve Ellcey <sellcey@imgtec.com>
6459
6460 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
6461 and mips16 directories.
cf1f9f8b 6462 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
0985b4a2 6463 (MULTILIB_DIRNAMES): Ditto.
6464 (MULTILIB_EXCEPTIONS): Add new exceptions.
6465 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
6466 (MULTILIB_DIRNAMES): Ditto.
6467 (MULTILIB_EXCEPTIONS): Add new exceptions.
6468
5137d3cb 64692012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6470 Marcus Shawcroft <marcus.shawcroft@arm.com>
6471
6472 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
6473 SYMBOL_TINY_ABSOLUTE.
6474 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
6475 SYMBOL_TINY_ABSOLUTE.
6476 (aarch64_expand_mov_immediate): Likewise.
6477 (aarch64_classify_symbol): Likewise.
6478 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
6479 Permit SYMBOL_TINY_ABSOLUTE.
6480 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
6481
e03531ec 64822013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6483 Marcus Shawcroft <marcus.shawcroft@arm.com>
6484
6485 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
6486 Refactor if/switch. Replace gcc_assert with if.
6487
46f4ca2e 64882013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
6489
6490 * config/i386/i386.c (initial_ix86_tune_features): Enable
6491 FP Reassociation for AMD bdver1 and bdver2.
6492
9b3c8ca4 64932013-05-29 Martin Jambor <mjambor@suse.cz>
6494
6495 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
6496 and IMAGPART_EXPR do not occur within other handled_components.
6497
65526f7f 64982013-05-29 Richard Biener <rguenther@suse.de>
6499
6500 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
6501 access on whether the use is in the BB we currently try to
6502 vectorize.
6503 (vect_bb_vectorization_profitable_p): Pass the BB we currently
6504 vectorize to vect_bb_slp_scalar_cost.
6505
32afb748 65062013-05-29 Richard Biener <rguenther@suse.de>
6507
6508 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
6509 computing scalar cost offsetted by stmts that are kept live
6510 by scalar uses.
6511 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
6512 for computation of scalar cost.
6513
d36d2527 65142013-05-28 Steve Ellcey <sellcey@mips.com>
6515
6516 * config/mips/mips-cpus.def (mips32r2): Change processor type.
6517
3c6d4197 65182013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
6519
6520 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
6521 array notation built-in reduction functions.
6522 * doc/passes.texi (Passes): Added documentation about changes done
6523 for Cilk Plus.
6524 * doc/invoke.texi (C Dialect Options): Added documentation about
6525 the -fcilkplus flag.
6526 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
6527 (BUILTINS_DEF): Depend on cilkplus.def.
6528 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
6529 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
6530 * cilkplus.def: New file.
6531
17ce39e3 65322013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
6533
6534 PR rtl-optimization/57439
6535 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
6536
bb88a6c7 65372013-05-28 Easwaran Raman <eraman@google.com>
6538
6539 PR tree-optimization/57337
6540 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
6541 (find_insert_point): Correctly identify the insertion point
6542 when two statements with the same UID is compared.
6543
8911f4de 65442013-05-28 Richard Biener <rguenther@suse.de>
6545
6546 PR tree-optimization/56787
6547 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
6548 from the list of data references.
6549 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
6550 clobbers.
6551 (vect_analyze_loop_operations): Likewise.
6552 (vect_transform_loop): Remove clobbers.
6553
49c71be6 65542013-05-28 Martin Jambor <mjambor@suse.cz>
6555
6556 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
6557 and REALPART_EXPRs have scalar type.
6558
a3c29ebc 65592013-05-28 Richard Biener <rguenther@suse.de>
6560
6561 PR tree-optimization/57411
6562 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
6563 virtual operands.
6564 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
6565 virtual operand propagation.
6566
65c294e9 65672013-05-28 Eric Botcazou <ebotcazou@adacore.com>
6568
6569 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
6570 destination register for bmasksi_vis.
6571 (vector_init_bshuffle): Likewise.
6572 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
6573
d6a0a4b0 65742013-05-28 Eric Botcazou <ebotcazou@adacore.com>
6575
6576 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
6577 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
6578 mode if the instruction isn't available in the original mode.
6579 * config/sparc/sparc.opt (mfix-ut699): New option.
6580 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
6581 (divdf3): Turn into expander.
6582 (divdf3_nofix): New insn.
6583 (divdf3_fix): Likewise.
6584 (divsf3): Disable if -mfix-ut699.
6585 (sqrtdf2): Turn into expander.
6586 (sqrtdf2_nofix): New insn.
6587 (sqrtdf2_fix): Likewise.
6588 (sqrtsf2): Disable if -mfix-ut699.
6589
5f037457 65902013-05-27 Richard Biener <rguenther@suse.de>
6591
6592 PR middle-end/57412
6593 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
6594 block for the new loop.
6595
e4800344 65962013-05-27 Richard Biener <rguenther@suse.de>
6597
6598 PR tree-optimization/57343
6599 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
6600 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
6601 (number_of_iterations_cond): Do not build the folded tree.
6602
d68e9408 66032013-05-27 Richard Biener <rguenther@suse.de>
6604
6605 Revert
6606 PR middle-end/57381
6607 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
6608 OEP_CONSTANT_ADDRESS_OF retained.
6609
6610 PR tree-optimization/57417
6611 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
6612 for unchanged base.
6613 (set_ssa_val_to): Compare addresses using
6614 get_addr_base_and_unit_offset.
6615
b6b86e87 66162013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
6617
6618 PR rtl-optimization/56833
6619 * postreload.c (move2add_record_mode): New function.
6620 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
6621 (move2add_use_add2_insn): Use move2add_record_sym_value.
6622 (move2add_use_add3_insn): Likewise.
6623 (reload_cse_move2add): Use move2add_valid_value_p and
6624 move2add_record_mode. Invalidate call-clobbered and REG_INC
6625 affected regs by setting reg_mode to VOIDmode.
6626 (move2add_note_store): Don't pretend the inside of a SUBREG is
6627 the actual destination. Invalidate single/leading registers by
6628 setting reg_mode to VOIDmode.
6629 Use move2add_record_sym_value, move2add_valid_value_p and
6630 move2add_record_mode.
6631
575a3e39 66322013-05-27 Richard Biener <rguenther@suse.de>
6633
6634 PR tree-optimization/57396
6635 * tree-affine.c (double_int_constant_multiple_p): Properly
6636 return false for val == 0 and div != 0.
6637
e0462563 66382013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
6639
6640 * config/mips/mips.h: Use #elif in preprocessor conditions.
6641
188bbdec 66422013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
6643
6644 PR target/53916
6645 * config/mips/constraints.md (kl): New constraint.
6646 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
6647 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
6648 constraint for operand 0. Split after CSE for MIPS16. Emit a move
6649 from LO for MIPS16.
6650 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
6651
073ac520 66522013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
6653
6654 PR target/55777
6655 * config/mips/mips.c (mips_can_inline_p): New function.
6656 (TARGET_CAN_INLINE_P): Define.
6657
242d1ee6 66582013-05-25 Steven Bosscher <steven@gcc.gnu.org>
6659
6660 * sched-int.h (ds_t, dw_t): Make unsigned int.
6661 Fix documentation that describes how all the ds_t bits are used.
6662 Reserve the last bit for delayed-branch scheduling.
6663 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
6664 (BITS_PER_DEP_WEAK): Fix definition and documentation.
6665 (gen_dep_weak_1): Remove prototype.
6666 * sched-deps.c (get_dep_weak_1): Make static.
6667 * target.def (speculate_insn, needs_block_p, gen_spec_check,
6668 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
6669 * doc/tm.texi: Regenerate.
6670 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
6671
2fdeadc1 66722013-05-24 Steven Bosscher <steven@gcc.gnu.org>
6673
6674 PR debug/56950
6675 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
6676
473d3d50 66772013-05-24 Nathan Sidwell <nathan@codesourcery.com>
6678 Sandra Loosemore <sandra@codesourcery.com>
6679
6680 * config.gcc (powerpc-*): Allow native for with-cpu.
6681
65a8f1a1 66822013-05-24 Jeff Law <law@redhat.com>
6683
6684 PR tree-optimization/57124
6685 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
6686 conversion feeding a condition if the range has an overflow
6687 if -fstrict-overflow. Add warnings for when we do make the
6688 transformation.
6689
0dc2f395 66902013-05-24 Dehao Chen <dehao@google.com>
6691
89321355 6692 * tree-cfg.c (locus_discrim_map): Fix the typo.
0dc2f395 6693 (locus_discrim_hasher): Likewise.
6694 (locus_discrim_hasher::hash): Likewise.
6695 (locus_discrim_hasher::equal): Likewise.
6696
7d9f258f 66972013-05-24 Martin Jambor <mjambor@suse.cz>
6698
6699 PR tree-optimization/57294
6700 * cgraph.h (ipa_record_stmt_references): Declare.
6701 * cgraphbuild.c (ipa_record_stmt_references): New function.
6702 (build_cgraph_edges): Use ipa_record_stmt_references.
6703 (rebuild_cgraph_edges): Likewise.
6704 (cgraph_rebuild_references): Likewise.
6705 * ipa-prop.c (ipa_modify_call_arguments): Discard references
6706 associated with the old statement and build references from the
6707 newly built statements.
6708 * ipa-ref.c (ipa_remove_stmt_references): New function.
6709 * ipa-ref.h (ipa_remove_stmt_references): Declare.
6710
cc0dc61b 67112013-05-24 Vladimir Makarov <vmakarov@redhat.com>
6712
f3de38f1 6713 * lra-constraints.c (emit_spill_move): Use smaller mode for
cc0dc61b 6714 mem-mem moves.
f3de38f1 6715 (check_and_process_move): Consider mem-reg moves for secondary
cc0dc61b 6716 too.
f3de38f1 6717 (curr_insn_transform): Don't lose insns emitted before for
cc0dc61b 6718 secondary memory moves.
6719 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
6720 reg set up in the current insn.
6721
601ee7b3 67222013-05-24 Dehao Chen <dehao@google.com>
6723
89321355 6724 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
601ee7b3 6725 hash function.
6726 (locus_descrim_hasher::equal): Likewise.
89321355 6727 (build_gimple_cfg): New discriminator assignment algorithm.
601ee7b3 6728 (make_edges): Likewise.
6729 (next_discriminator_for_locus): Likewise.
6730 (same_line_p): Likewise.
6731 (assign_discriminators): Likewise.
6732 (make_cond_expr_edges): Likewise.
6733 (make_gimple_switch_edges): Likewise.
6734 (make_goto_expr_edges): Likewise.
6735 (make_gimple_asm_edges): Likewise.
6736
2f42b453 67372013-05-24 Ian Bolton <ian.bolton@arm.com>
6738
6739 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
6740 X format specifier to only display bottom 16 bits.
6741 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
6742 immediate to match for operand 2, since it will be masked.
6743
5f4dd0d0 67442013-05-24 Richard Biener <rguenther@suse.de>
6745
6746 PR tree-optimization/57287
6747 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
6748 all SSA names that occur in abnormal PHIs.
6749
f982ec99 67502013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
6751
6752 PR tree-ssa/57385
6753 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
6754 that index is not negative.
6755
fe081106 67562013-05-24 Eric Botcazou <ebotcazou@adacore.com>
6757
6758 PR rtl-optimization/55177
6759 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
6760 (simplify_byte_swapping_operation): New.
6761 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
6762 (simplify_relational_operation_1): Deal with BSWAP.
6763
3111042a 67642013-05-23 Richard Henderson <rth@redhat.com>
6765
6766 PR target/56742
6767 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
6768 (ix86_reorg): Call it.
6769
8ebff8a4 67702013-05-23 Uros Bizjak <ubizjak@gmail.com>
6771
6772 PR target/57379
6773 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
6774 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
6775 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
6776
56cff37b 67772013-05-23 Christian Bruel <christian.bruel@st.com>
6778
6779 PR debug/57351
6780 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
6781
a5ebd4c4 67822013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6783 Marcus Shawcroft <marcus.shawcroft@arm.com>
6784
6785 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
6786 * config/aarch64/constraints.md (Usa): Remove.
6787 * doc/md.texi (AArch64 Usa): Remove.
6788
fda698d7 67892013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6790 Marcus Shawcroft <marcus.shawcroft@arm.com>
6791
6792 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
6793 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
6794 * config/aarch64/predicates.md (aarch64_const_address): Remove.
6795 (aarch64_mov_operand): Use aarch64_mov_operand_p.
6796
9cacc1ad 67972013-05-23 Vidya Praveen <vidyapraveen@arm.com>
6798
6799 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
6800 instruction (AdvSIMD).
6801 * config/aarch64/aarch64-builtins.c
6802 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
6803 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
6804
74d5ea33 68052013-05-23 Martin Jambor <mjambor@suse.cz>
fb8b391e 6806
6807 PR middle-end/57347
6808 * tree.h (contains_bitfld_component_ref_p): Declare.
6809 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8ebff8a4 6810 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
6811 caller.
fb8b391e 6812 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
6813 not access a bit-field. Assert all final offsets are byte-aligned.
6814
6979886b 68152013-05-23 Richard Biener <rguenther@suse.de>
6816
6817 PR tree-optimization/57380
6818 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
6819 least one invariant or re-used load.
6820 * passes.c (init_optimization_passes): Move pass_phiprop before
6821 pass_forwprop.
6822
b776525d 68232013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
6824
6825 * config/aarch64/aarch64-simd.md
6826 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
6827
324d4d99 68282013-05-23 Richard Biener <rguenther@suse.de>
6829
6830 PR middle-end/57381
6831 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
6832 OEP_CONSTANT_ADDRESS_OF retained.
6833
98908b23 68342013-05-23 Jakub Jelinek <jakub@redhat.com>
6835
6836 PR middle-end/57344
8ebff8a4 6837 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
6838 don't lower unit. Handle unit not being always BITS_PER_WORD.
98908b23 6839
81c3220e 68402013-05-23 Richard Biener <rguenther@suse.de>
6841
6842 PR rtl-optimization/57341
6843 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
6844 instead of true_dependence.
6845
7b38ffff 68462013-05-22 David Malcolm <dmalcolm@redhat.com>
6847
72524e91 6848 * bb-reorder.c (branch_threshold): Make const.
6849 (exec_threshold): Ditto.
7b38ffff 6850
81f0e7d0 68512013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
6852 Pat Haugen <pthaugen@us.ibm.com>
6853 Peter Bergner <bergner@vnet.ibm.com>
6854
6855 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
6856 documentation for the power8 crypto builtins.
6857
6858 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
6859
6860 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
6861 macros for defining power8 builtin functions.
6862 (BU_P8V_AV_2): Likewise.
6863 (BU_P8V_AV_P): Likewise.
6864 (BU_P8V_VSX_1): Likewise.
6865 (BU_P8V_OVERLOAD_1): Likewise.
6866 (BU_P8V_OVERLOAD_2): Likewise.
6867 (BU_CRYPTO_1): Likewise.
6868 (BU_CRYPTO_2): Likewise.
6869 (BU_CRYPTO_3): Likewise.
6870 (BU_CRYPTO_OVERLOAD_1): Likewise.
6871 (BU_CRYPTO_OVERLOAD_2): Likewise.
6872 (XSCVSPDP): Fix typo, point to the correct instruction.
6873 (VCIPHER): Add power8 crypto builtins.
6874 (VCIPHERLAST): Likewise.
6875 (VNCIPHER): Likewise.
6876 (VNCIPHERLAST): Likewise.
6877 (VPMSUMB): Likewise.
6878 (VPMSUMH): Likewise.
6879 (VPMSUMW): Likewise.
6880 (VPERMXOR_V2DI): Likewise.
6881 (VPERMXOR_V4SI: Likewise.
6882 (VPERMXOR_V8HI: Likewise.
6883 (VPERMXOR_V16QI: Likewise.
6884 (VSHASIGMAW): Likewise.
6885 (VSHASIGMAD): Likewise.
6886 (VPMSUM): Likewise.
6887 (VPERMXOR): Likewise.
6888 (VSHASIGMA): Likewise.
6889
6890 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
6891 __CRYPTO__ if the crypto instructions are available.
6892 (altivec_overloaded_builtins): Add support for overloaded power8
6893 builtins.
6894
6895 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
6896 support for power8 crypto builtins.
6897 (builtin_function_type): Likewise.
6898 (altivec_init_builtins): Add support for builtins that take vector
6899 long long (V2DI) arguments.
6900
6901 * config/rs6000/crypto.md: New file, define power8 crypto
6902 instructions.
6903
69042013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
6905 Pat Haugen <pthaugen@us.ibm.com>
6906 Peter Bergner <bergner@vnet.ibm.com>
6907
6908 * doc/invoke.texi (Option Summary): Add power8 options.
6909 (RS/6000 and PowerPC Options): Likewise.
6910
6911 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
6912 constraints.md instead of rs6000.h. Reorder w* constraints. Add
6913 wm, wn, wr documentation.
6914
6915 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
6916 registers if direct move instructions are enabled.
6917 (wn): New constraint for no registers.
6918 (wq): New constraint for quad word even GPR registers.
6919 (wr): New constraint if 64-bit instructions are enabled.
6920 (wv): New constraint if power8 vector instructions are enabled.
6921 (wQ): New constraint for quad word memory locations.
6922
6923 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
6924 constraint for 0..15 for crypto instructions.
6925 (gpc_reg_operand): If VSX allow registers in VSX registers as well
6926 as GPR and floating point registers.
6927 (int_reg_operand): New predicate to match only GPR registers.
6928 (base_reg_operand): New predicate to match base registers.
6929 (quad_int_reg_operand): New predicate to match even GPR registers
6930 for quad memory operations.
6931 (vsx_reg_or_cint_operand): New predicate to allow vector logical
6932 operations in both GPR and VSX registers.
6933 (quad_memory_operand): New predicate for quad memory operations.
6934 (reg_or_indexed_operand): New predicate for direct move support.
6935
6936 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
6937 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
6938 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
6939 (POWERPC_MASKS): Add power8 options.
6940 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
6941 various options.
6942
6943 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
6944 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
6945
6946 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
6947 (-mpower8-fusion): New power8 options.
6948 (-mpower8-fusion-sign): Likewise.
6949 (-mpower8-vector): Likewise.
6950 (-mcrypto): Likewise.
6951 (-mdirect-move): Likewise.
6952 (-mquad-memory): Likewise.
6953
6954 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
6955 power8.
6956 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
6957 registers.
8ebff8a4 6958 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
81f0e7d0 6959 (rs6000_debug_vector_unit): Add p8_vector.
6960 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
6961 definitions. Also print fusion state.
6962 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
6963 (rs6000_builtin_mask_calculate): Add power8 builtin support.
6964 (rs6000_option_override_internal): Add support for power8.
6965 (rs6000_common_init_builtins): Add debugging for skipped builtins
6966 if -mdebug=builtin.
6967 (rs6000_adjust_cost): Add power8 support.
6968 (rs6000_issue_rate): Likewise.
6969 (insn_must_be_first_in_group): Likewise.
6970 (insn_must_be_last_in_group): Likewise.
6971 (force_new_group): Likewise.
6972 (rs6000_register_move_cost): Likewise.
6973 (rs6000_opt_masks): Likewise.
6974
6975 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
6976 power8 capable assembler, default to power7 options.
6977 (TARGET_DIRECT_MOVE): Likewise.
6978 (TARGET_CRYPTO): Likewise.
6979 (TARGET_P8_VECTOR): Likewise.
6980 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
6981 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
6982 (VECTOR_MEM_P8_VECTOR_P): Likewise.
6983 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
6984 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
6985 (TARGET_XSCVDPSPN): Likewise.
6986 (TARGET_XSCVSPDPN): Likewsie.
6987 (TARGET_SYNC_HI_QI): Likewise.
6988 (TARGET_SYNC_TI): Likewise.
6989 (MASK_CRYPTO): Likewise.
6990 (MASK_DIRECT_MOVE): Likewise.
6991 (MASK_P8_FUSION): Likewise.
6992 (MASK_P8_VECTOR): Likewise.
8ebff8a4 6993 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
6994 temporary used by some of the direct move instructions to get two FP
6995 temporary registers does not force creation of a stack frame.
81f0e7d0 6996 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
6997 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
6998 that any VSX registers are tieable, even if they are also an
6999 Altivec vector mode.
7000 (r6000_reg_class_enum): Add wm, wr, wv constraints.
7001 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7002 (RS6000_BTM_CRYPTO): Likewise.
7003 (RS6000_BTM_COMMON): Likewise.
7004
7005 * config/rs6000/rs6000.md (cpu attribute): Add power8.
7006 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7007 (enum rs6000_vector): Add power8 vector support.
7008
7fa9818b 70092013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7010
7011 PR target/19599
7012 PR target/57340
7013 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7014 (any_sibcall_could_use_r3): this and handle indirect calls.
7015 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7016
ed11924e 70172013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7018
7019 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7020
bb268663 70212013-05-22 Richard Biener <rguenther@suse.de>
7022
7023 PR middle-end/57349
7024 * profile.c (branch_prob): Do not split blocks that are
7025 abnormally receiving from ECF_RETURNS_TWICE functions.
7026
e7dd8581 70272013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
7028
7029 * recog.c (offsettable_address_addr_space_p): Fix calculation of
7030 address mode. Move pointer mode initialization to the same place.
7031
873bf3e1 70322013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
7033
7034 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7035 while it has any effect.
7036
64e4dbea 70372013-05-21 Easwaran Raman <eraman@google.com>
7038
7039 PR tree-optimization/57322
bb88a6c7 7040 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7041 UID of the statement added to the BB to be 1.
64e4dbea 7042
1c872eee 70432013-05-21 Jakub Jelinek <jakub@redhat.com>
7044
7045 PR tree-optimization/57331
8ebff8a4 7046 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7047 of conversion from pointer type to integral type with integer.
1c872eee 7048
f965232f 70492013-05-21 Martin Jambor <mjambor@suse.cz>
7050
7051 PR lto/57289
7052 * ipa-prop.c (ipa_read_node_info): Process param_used and
7053 controlled_uses in the same order as when writing.
7054
86642a49 70552013-05-21 Magnus Granberg <baldrick@free.fr>
7056
7057 PR plugins/56754
515b2f23 7058 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
86642a49 7059
ae8a8b85 70602013-05-21 Richard Biener <rguenther@suse.de>
7061
7062 PR tree-optimization/57318
7063 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7064 estimate stmts with side-effects as likely eliminated.
7065
35d3304d 70662013-05-21 Richard Biener <rguenther@suse.de>
7067
7068 PR tree-optimization/57330
7069 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7070 preserve the call stmts fntype.
7071
03dccd68 70722013-05-21 Richard Biener <rguenther@suse.de>
7073
7074 PR tree-optimization/57303
7075 * tree-ssa-sink.c (statement_sink_location): Improve killing
7076 stmt detection and properly handle self-assignments.
7077
3983036a 70782013-05-21 Christian Bruel <christian.bruel@st.com>
7079
f3de38f1 7080 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7081 spanning registers. LEAF_REG_REMAP is supported only for contiguous
3983036a 7082 registers. Set register size out of the PARALLEL loop.
7083
dbc31e6d 70842013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
7085
7086 PR target/56547
7087 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7088 (*fmasf4, *fmasf4_media): New insns.
7089
6bb79546 70902013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7091
7092 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7093 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7094 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7095 (mips_idiv_insns): Update the comments to say that the returned
7096 instruction counts are in units of BASE_INSN_LENGTH.
7097 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7098 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
7099 using 2 rather than 4 as the length of indirect MIPS16 and
7100 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
7101 length of a NOP. Don't divide MIPS16 lengths by 2.
7102 (mips16_split_long_branches): Assume a branch is long if the
7103 length is greater than 4 rather than 8.
7104 * config/mips/mips.md (length): Give MIPS16 lengths directly,
7105 rather than multiplying them by 2. Multiply instruction counts
7106 by BASE_INSN_LENGTH rather than 4.
7107 (*jump_mips16, tls_get_tp_mips16_<mode>)
7108 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7109
18ea5fe9 71102013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7111
7112 * config/mips/mips.md (extended_mips16): Remove branch case.
7113 (length): Remove duplicated extended_mips16 test.
7114
28763a15 71152013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7116
7117 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7118
a31fe3d0 71192013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
7120
7121 * recog.h (Recog_data): Rename to...
7122 (recog_data_d): ...this.
7123 (recog_data): Update accordingly.
7124 * recog.c (recog_data): Likewise.
7125 * reload.c (save_recog_data): Likewise.
7126 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7127 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7128
f473eb72 71292013-05-17 Julian Brown <julian@codesourcery.com>
7130
7131 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7132 found in a REG_EQUAL note, invalidate it.
7133
5f7f04d8 71342013-05-17 Easwaran Raman <eraman@google.com>
a2bd0c99 7135
7136 * tree-ssa-reassoc.c (find_insert_point): New function.
7137 (insert_stmt_after): Likewise.
7138 (get_def_stmt): Likewise.
7139 (ensure_ops_are_available): Likewise.
7140 (not_dominated_by): Likewise.
7141 (rewrite_expr_tree): Do not move statements beyond what is
7142 necessary. Remove call to swap_ops_for_binary_stmt...
7143 (reassociate_bb): ... and move it here.
7144 (build_and_add_sum): Assign UIDs for new statements.
7145 (linearize_expr): Likewise.
7146 (do_reassoc): Renumber gimple statement UIDs.
7147
fc8456b4 71482013-05-17 Jan Hubicka <jh@suse.cz>
7149
7150 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7151 weakrefs.
7152 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
ae837da3 7153 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7154 weakrefs.
fc8456b4 7155 (output_weakrefs): Update.
7156
9ea71c42 71572013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
7158 Martin Jambor <mjambor@suse.cz>
7159
7160 PR middle-end/57276
7161 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7162 value that corresponds to the given aggval is found in values vector.
7163
fb0f74a7 71642013-05-17 Uros Bizjak <ubizjak@gmail.com>
7165
7166 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7167 sse, sse2, sse3, ssse3 and sse4a flags to options.
7168
a50dd382 71692013-05-17 David Malcolm <dmalcolm@redhat.com>
7170
7171 * gengtype-state.c: (s_expr_writer): New class, to handle
7172 prettifying of output layout of s-expressions.
7173 (state_writer): New class, to write out gtype.state.
7174 (state_written_type_count): Move this variable into member data of
7175 state_writer.
7176 (s_expr_writer::s_expr_writer): New code: constructor for new class
7177 (state_writer::state_writer(): ditto
7178 (s_expr_writer::write_new_line): New function
7179 (s_expr_writer::write_any_indent): ditto
7180 (s_expr_writer::begin_s_expr): ditto
7181 (s_expr_writer::end_s_expr): ditto
7182 (write_state_fileloc): convert to method of state_writer...
7183 (state_writer:: write_state_fileloc): ...and use methods of
7184 s_expr_writer to write indentation into the gtype.state output file
7185 to visually represent the hierarchical structure of the list
7186 structures
7187 (write_state_fields): ditto, renaming to...
7188 (state_writer::write_state_fields)
7189 (write_state_a_string): ditto, renaming to...
7190 (state_writer::write_state_a_string)
7191 (write_state_string_option): ditto, renaming to...
7192 (state_writer::write_state_string_option)
7193 (write_state_type_option): ditto, renaming to...
7194 (state_writer::write_state_type_option)
7195 (write_state_nested_option): ditto, renaming to...
7196 (state_writer::write_state_nested_option)
7197 (write_state_option): ditto, renaming to...
7198 (state_writer::write_state_option)
7199 (write_state_options): ditto, renaming to...
7200 (state_writer::write_state_options)
7201 (write_state_lang_bitmap): ditto, renaming to...
7202 (state_writer::write_state_lang_bitmap)
7203 (write_state_version): ditto, renaming to...
7204 (state_writer::write_state_version)
7205 (write_state_scalar_type): ditto, renaming to...
7206 (state_writer::write_state_scalar_type)
7207 (write_state_string_type): ditto, renaming to...
7208 (state_writer::write_state_string_type)
7209 (write_state_undefined_type): ditto, renaming to...
7210 (state_writer::write_state_undefined_type)
7211 (write_state_struct_union_type): ditto, renaming to...
7212 (state_writer::write_state_struct_union_type)
7213 (write_state_struct_type): ditto, renaming to...
7214 (state_writer::write_state_struct_type)
7215 (write_state_user_struct_type): ditto, renaming to...
7216 (state_writer::write_state_user_struct_type)
7217 (write_state_lang_struct_type): ditto, renaming to...
7218 (state_writer::write_state_lang_struct_type)
7219 (write_state_param_struct_type): ditto, renaming to...
7220 (state_writer::write_state_param_struct_type)
7221 (write_state_pointer_type): ditto, renaming to...
7222 (state_writer::write_state_pointer_type)
7223 (write_state_array_type): ditto, renaming to...
7224 (state_writer::write_state_array_type)
7225 (write_state_gc_used): ditto, renaming to...
7226 (state_writer::write_state_gc_used)
7227 (write_state_common_type_content): ditto, renaming to...
7228 (state_writer::write_state_common_type_content)
7229 (write_state_type): ditto, renaming to...
7230 (state_writer::write_state_type)
7231 (write_state_pair_list): ditto, renaming to...
7232 (state_writer::write_state_pair_list)
7233 (write_state_pair): ditto, renaming to...
7234 (state_writer::write_state_pair)
7235 (write_state_typedefs): ditto, renaming to...
7236 (state_writer::write_state_typedefs)
7237 (write_state_structures): ditto, renaming to...
7238 (state_writer::write_state_structures)
7239 (write_state_param_structs): ditto, renaming to...
7240 (state_writer::write_state_param_structs)
7241 (write_state_variables): ditto, renaming to...
7242 (state_writer::write_state_variables)
7243 (write_state_srcdir): ditto, renaming to...
7244 (state_writer::write_state_srcdir)
7245 (write_state_files_list): ditto, renaming to...
7246 (state_writer::write_state_files_list)
7247 (write_state_languages): ditto, renaming to...
7248 (state_writer::write_state_languages)
7249 (write_state): create a state_writer instance and use it when
7250 writing out the state file
7251
639d27bb 72522013-05-17 Mike Stump <mikestump@comcast.net>
7253
65e47909 7254 PR rtl-optimization/57304
639d27bb 7255 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
7256 accessing DF_REF_REAL_LOC.
7257
6150dc35 72582013-05-17 Jakub Jelinek <jakub@redhat.com>
7259
7260 PR rtl-optimization/57281
7261 PR rtl-optimization/57300
7262 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
7263 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
7264 what the other splitter did if the registers are dead.
7265
83d18d6d 72662013-05-17 Richard Biener <rguenther@suse.de>
7267
7268 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
7269 MEM_REF offsets.
7270
86d5f657 72712013-05-17 Jakub Jelinek <jakub@redhat.com>
7272
7273 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
7274 linking.
7275
1390d90a 72762013-05-17 Marek Polacek <polacek@redhat.com>
7277
fb0f74a7 7278 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
7279 length when doing non-zero store of storing '\0' to '\0'.
1390d90a 7280
d38deed0 72812013-05-17 Jakub Jelinek <jakub@redhat.com>
7282
0a015fc7 7283 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
7284 vect_external_def oprnd1 with loop_vinfo, try to emit
7285 optional cast, negation and and stmts on the loop preheader
7286 edge instead of into the pattern def seq.
7287
d38deed0 7288 PR tree-optimization/57051
7289 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
7290 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
7291
fbac2992 72922013-05-16 Nick Clifton <nickc@redhat.com>
7293
7294 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
7295 (rl78_is_naked_func): New function.
47dd4311 7296 (rl78_expand_prologue): Skip prologue generation for naked functions.
7297 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
fbac2992 7298 * doc/extend.texi (naked): Add RL78 to the list of processors
7299 that supports this attribute.
7300
1ba55558 73012013-05-16 Jeff Law <law@redhat.com>
7302
7303 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
7304
e35017c0 73052013-05-16 Uros Bizjak <ubizjak@gmail.com>
7306
7307 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
7308 cache parameters using detect_caches_amd also for CYRIX,
7309 NSC and TM2 signatures.
7310
6d17d827 73112013-05-16 Uros Bizjak <ubizjak@gmail.com>
7312 Dzianis Kahanovich <mahatma@eu.by>
7313
7314 PR target/45359
7315 PR target/46396
7316 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7317 VIA/Centaur processors and determine their cache parameters
7318 using detect_caches_amd.
7319
4bf32a77 73202013-05-16 Teresa Johnson <tejohnson@google.com>
7321
7322 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
7323 (rtl_verify_edges): New function.
7324 (rtl_verify_bb_insns): Ditto.
7325 (rtl_verify_bb_pointers): Ditto.
7326 (rtl_verify_bb_insn_chain): Ditto.
7327 (rtl_verify_fallthru): Ditto.
7328 (rtl_verify_bb_layout): Ditto.
7329 (rtl_verify_flow_info_1): Outline checks into new functions.
7330 (rtl_verify_flow_info): Ditto.
7331
d99f53b2 73322013-05-16 Steve Ellcey <sellcey@imgtec.com>
7333
7334 * cfghooks.c (copy_bbs): Add update_dominance argument.
7335 * cfghooks.h (copy_bbs): Update prototype.
7336 * tree-cfg.c (gimple_duplicate_sese_region):
7337 Add update_dominance argument.
7338 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
7339 * tree-ssa-loop-ch.c (copy_loop_headers): Update
7340 gimple_duplicate_sese_region call.
7341 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
7342 Update copy_bbs call.
7343 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
7344 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
7345
cb4792d2 73462013-05-16 Jakub Jelinek <jakub@redhat.com>
7347
7348 * tree-vectorizer.h (NUM_PATTERNS): Increment.
7349 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
7350 vect_recog_rotate_pattern.
7351 (vect_recog_rotate_pattern): New function.
7352
be5f9e46 73532013-05-16 Jason Merrill <jason@redhat.com>
7354
7355 * Makefile.in (LLINKER): New variable.
7356 (mostlyclean): Remove link mutex.
7357 * configure.ac: Handle --enable-link-mutex.
7358 * lock-and-run.sh: New script.
7359
9dbfd743 73602013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7361
7362 PR target/19599
7363 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
7364 for NULL decl.
7365
a98345a3 73662013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7367
7368 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
7369
ae51a965 73702013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
7371
7372 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
7373 * config/arm/arm.c (next_consecutive_mem): New function.
7374 (gen_movmem_ldrd_strd): Likewise.
7375 * config/arm/arm.md (movmemqi): Update condition and code.
7376 (unaligned_loaddi, unaligned_storedi): New patterns.
7377
062ed845 73782013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7379
7380 * config.gcc: Obsolete *-*-solaris2.9*.
7381 * doc/install.texi (Specific, *-*-solaris2*): Document it.
7382
9c7f0e62 73832013-05-16 Richard Biener <rguenther@suse.de>
7384
7385 * passes.c (init_optimization_passes): Move pass_parallelize_loops
7386 earlier, after GRAPHITE transforms and IV canonicalization.
7387
8e6b4515 73882013-05-16 Jakub Jelinek <jakub@redhat.com>
7389
7390 * omp-low.c (extract_omp_for_data): For collapsed loops,
7391 if at least one of the loops is known at compile time to
7392 iterate zero times, set count to 0.
7393 (expand_omp_regimplify_p): New function.
7394 (expand_omp_for_generic): For collapsed loops, if at least
7395 one of the loops isn't known to iterate at least once,
7396 add runtime check with setting count to 0.
7397 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
7398 For unsigned types if it isn't known at compile time that
7399 the loop will iterate at least once, add runtime check to bypass
7400 the whole loop if initial condition isn't true.
7401
5c62c52a 74022013-05-16 Nathan Sidwell <nathan@codesourcery.com>
7403
7404 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
7405
ea1b629e 74062013-05-16 Marc Glisse <marc.glisse@inria.fr>
7407
7408 PR middle-end/57286
7409 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
7410 transformations to avoid an infinite loop.
7411
d70b7140 74122013-05-16 Marek Polacek <polacek@redhat.com>
7413
7414 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
7415
93d015a0 74162013-05-15 Leif Ekblad <leif@rdos.net>
7417
7418 * config/i386/i386.c (ix86_decompose_address): Use
7419 DEFAULT_TLS_SEG_REG to access TLS segment register.
7420 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
7421 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
7422 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
7423
c0de0233 74242013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
7425
7426 PR target/57260
7427 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
7428 sibling calls to functions that would normally be lazily bound,
7429 unless $gp is call-clobbered.
7430
c27c2857 74312013-05-15 Uros Bizjak <ubizjak@gmail.com>
7432
6d17d827 7433 * config/i386/i386.c (ix86_option_override_internal): Update
c27c2857 7434 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
55472891 7435 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
7436 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
c27c2857 7437 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
7438 of TARGET_3DNOW.
7439 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
7440
e2be515a 74412013-05-15 Andreas Schwab <schwab@suse.de>
7442
7443 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
7444 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
7445 third operand.
7446
830b796a 74472013-05-15 Teresa Johnson <tejohnson@google.com>
7448
7449 * loop-unroll.c (report_unroll_peel): Check decision before
7450 emitting unroll/peel message.
7451
812ca88e 74522013-05-15 Teresa Johnson <tejohnson@google.com>
7453
7454 * function.h (has_bb_partition): New rtl_data flag.
7455 (bb_reorder_complete): Ditto.
7456 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
7457 instead of flag_reorder_blocks_and_partition.
7458 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
7459 with some enhancements.
7460 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
7461 * bb-reorder.c (connect_traces): Check for has_bb_partition
7462 instead of flag_reorder_blocks_and_partition.
7463 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
7464 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
7465 verify_hot_cold_block_grouping.
7466 (partition_hot_cold_basic_blocks): Set has_bb_partition.
7467
33ae7c4b 74682013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7469
7470 PR target/19599
7471 * config/arm/predicates.md (call_insn_operand): New predicate.
7472 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
7473 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
7474 if insn is not a tail call.
7475 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
7476 registers.
7477 * config/arm/arm.h (enum reg_class): New caller save register class.
7478 (REG_CLASS_NAMES): Likewise.
7479 (REG_CLASS_CONTENTS): Likewise.
7480 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
7481 without decls.
7482
b055bc88 74832013-05-15 Richard Biener <rguenther@suse.de>
7484
7485 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
7486 of MSG_OPTIMIZED_LOCATIONS.
7487 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
7488 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
7489 message.
7490 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
7491 of MSG_OPTIMIZED_LOCATIONS.
7492 (execute_vect_slp): Likewise.
7493 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
7494 (vect_create_cond_for_alias_checks): Likewise.
7495 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
7496 (vect_recog_widen_mult_pattern): Likewise.
7497 (vect_recog_widen_sum_pattern): Likewise.
7498 (vect_recog_over_widening_pattern): Likewise.
7499 (vect_recog_widen_shift_pattern): Likewise.
7500 (vect_recog_vector_vector_shift_pattern): Likewise.
7501 (vect_recog_divmod_pattern): Likewise.
7502 (vect_recog_mixed_size_cond_pattern): Likewise.
7503 (vect_recog_bool_pattern): Likewise.
7504 (vect_pattern_recog_1): Likewise.
7505
95fb3203 75062013-05-15 Martin Jambor <mjambor@suse.cz>
7507
7508 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
7509 non-functions to builtin_unreachable.
7510 * ipa-inline-transform.c (inline_call): Do not assert estimates were
7511 correct when new direct edges were discovered.
7512
15c999e3 75132013-05-15 Martin Jambor <mjambor@suse.cz>
7514
7515 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
7516 header, print symbol order instead of node uid, print more information
7517 about indirect edge targets.
7518 (ipa_make_edge_direct_to_target): Print symbol order instead of node
7519 uids.
7520 (ipa_make_edge_direct_to_target): Likewise.
7521 (remove_described_reference): Likewise.
7522 (propagate_controlled_uses): Likewise.
7523 (ipa_print_node_params): Also print symbol order.
7524 (ipcp_transform_function): Print symbol order instead of node uids.
7525 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
7526 (cgraph_get_create_real_symbol_node): Likewise.
7527 * ipa-cp.c (print_lattice): Likewise.
7528 (print_all_lattices): Likewise.
7529 (determine_versionability): Likewise.
7530 (initialize_node_lattices): Likewise.
7531 (estimate_local_effects): Likewise.
7532 (update_profiling_info): Likewise.
7533 (create_specialized_node): Likewise.
7534 (perhaps_add_new_callers): Likewise.
7535 (decide_about_value): Likewise.
7536 (decide_whether_version_node): Likewise.
7537 (identify_dead_nodes): Likewise.
7538 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
7539 (dump_inline_summary): Likewise.
7540 (estimate_node_size_and_time): Likewise.
7541 (inline_analyze_function): Likewise.
7542 * ipa-inline.c (report_inline_failed_reason): Likewise.
7543 (want_early_inline_function_p): Likewise.
7544 (edge_badness): Likewise.
7545 (update_edge_key): Likewise.
7546 (inline_small_functions): Likewise. Add dumping of order to two other
7547 dumps.
7548 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
7549 instead of node uids.
7550 (propagate_pure_const): Likewise.
7551 (propagate_pure_const): Likewise.
7552 * ipa-utils.c (dump_cgraph_node_set): Likewise.
7553 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
7554 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
7555 of node uids.
7556 * tree-pretty-print.c (dump_function_header): Likewise.
7557 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
7558 Print symbol order instead of node uids.
7559
7af74943 75602013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7561
7562 * config/s390/s390.c (s390_register_move_cost): Don't impose the
7563 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
7564
4c2623ad 75652013-05-15 Richard Biener <rguenther@suse.de>
7566
7567 PR tree-optimization/57275
7568 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
7569 return value for fail to do runtime alias checks for gather loads.
7570
33864132 75712013-05-15 Jan Hubicka <jh@suse.cz>
7572
7573 PR lto/57038
7574 PR lto/47375
c27c2857 7575 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
7576 weakrefs are not external.
7577 (lto_symtab_merge_decls): Fix thinko when dealing with
7578 non-lto_symtab decls.
33864132 7579 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
7580 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
7581 * varpool.c (dump_varpool_node): Dump more flags.
7582
9b0c154a 75832013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
7584
7585 * config/i386/i386.c (processor_alias_table): Add instruction
7586 FSGSBASE for AMD bdver3 architecture.
7587
721cebe6 75882013-05-14 Jakub Jelinek <jakub@redhat.com>
7589
7590 * tree.c (warn_deprecated_use): Print file:line using locus color.
7591 * diagnostic.c (diagnostic_report_current_module): Print file:line
7592 and file:line:column using locus color.
7593
832f4182 75942013-05-14 Mike Stump <mikestump@comcast.net>
7595
7596 * gdbinit.in: Add __null.
7597
31beb4cd 75982013-05-14 Mike Stump <mikestump@comcast.net>
7599
7600 * recog.h: Rename struct recog_data to Recog_data.
7601 * recog.c: Likewise.
7602 * reload.c (can_reload_into): Likewise.
7603 * config/picochip/picochip.c: Likewise.
7604
14d27f18 76052013-05-14 Mike Stump <mikestump@comcast.net>
7606
7607 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
7608
9d853199 76092013-05-14 Steven Bosscher <steven@gcc.gnu.org>
7610
e1ed78c8 7611 * resource.h (struct resources): Remove unch_memory member.
7612 (CLEAR_RESOURCE): Don't clear unch_memory.
7613 * resource.c (mark_referenced_resources): Don't set it.
7614 (mark_set_resources): Likewise.
7615 (mark_target_live_regs): Don't clear it.
7616 (init_resource_info): Likewise.
7617 * reorg.c (resource_conflicts_p): Don't compare it.
7618 (redundant_insn): Don't set it.
7619
9d853199 7620 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
7621 Remove prototypes.
7622 * emit-rtl.c (next_label): Remove unused function.
7623 (skip_consecutive_labels, link_cc0_insns): Move to ...
7624 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
7625 only place where these functions are used, and make them static.
7626
9be60878 76272013-05-14 Marc Glisse <marc.glisse@inria.fr>
7628
7629 * fold-const.c (fold_negate_expr): Handle vectors.
7630 (fold_truth_not_expr): Make it static.
7631 (fold_invert_truthvalue): New static function.
7632 (invert_truthvalue_loc): Handle vectors. Do not call
7633 fold_truth_not_expr directly.
7634 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
7635 <TRUTH_NOT_EXPR>: Do not cast to boolean.
7636 (fold_comparison): Handle vector constants.
7637 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
7638 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
7639 * tree.h (fold_truth_not_expr): Remove declaration.
7640
2fcc3fb4 76412013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
7642
7643 * config/aarch64/aarch64-simd.md
7644 (aarch64_vcond_internal<mode>): Rename to...
7645 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
7646 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
7647 float modes. Clarify all iterator modes.
7648 (vcond<mode><mode>): Use new name for vcond expanders.
7649 (vcond<v_cmp_result><mode>): Likewise.
7650 (vcondu<mode><mode>: Likewise.
7651 * config/aarch64/iterators.md (VDQF_COND): New.
7652
dbdcf3c6 76532013-05-14 Marc Glisse <marc.glisse@inria.fr>
7654
7655 PR bootstrap/57266
7656 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
7657 variable for the shift amount. Check that we shift by non-negative
7658 amounts.
7659
8d42bec1 76602013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
7661
7662 PR target/42017
7663 * config/arm/arm.h (EPILOGUE_USES): Only return true
7664 for LR_REGNUM after epilogue_completed.
7665
b8f9423d 76662013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
7667
7668 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
c27c2857 7669 is error_mark_node.
b8f9423d 7670
a861c00f 76712013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7672
7673 PR target/57261
7674 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
7675 and Solaris 11+/x86 with gld.
7676 * configure: Regenerate.
7677
b96ca9ca 76782013-05-14 Jakub Jelinek <jakub@redhat.com>
7679
7680 * expmed.c (expand_shift_1): Canonicalize rotates by
7681 constant bitsize / 2 to bitsize - 1.
b3ffac0b 7682 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
b96ca9ca 7683 case ROTATERT>: Likewise.
7684
7685 Revert:
7686 2013-05-10 Jakub Jelinek <jakub@redhat.com>
7687
7688 * config/i386/i386.md (rotateinv): New code attr.
7689 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
7690 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
7691 roll $31, %eax, etc.
7692
8aacb2c5 76932013-05-14 Richard Biener <rguenther@suse.de>
7694
7695 PR middle-end/57235
7696 * tree-eh.c (sink_clobbers): Give up for successors with
7697 multiple predecessors and no virtual uses.
7698
3726209b 76992013-05-14 Eric Botcazou <ebotcazou@adacore.com>
7700
7701 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
7702 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
7703
4620c2de 77042013-05-14 Jakub Jelinek <jakub@redhat.com>
7705
7706 PR middle-end/57251
7707 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
7708 the case when both op0 and op1 have VOIDmode.
7709
57518f85 77102013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
7711
7712 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
7713 in multiply-accumulate mode.
7714
2fd796fd 77152013-05-13 Guozhi Wei <carrot@google.com>
7716
7717 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
7718
cb2d23dd 77192013-05-13 Kai Tietz <ktietz@redhat.com>
7720
7721 PR target/56975
7722 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
c27c2857 7723 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
cb2d23dd 7724 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
c27c2857 7725 * config/i386/i386.c (ix86_option_override_internal): Likewise.
cb2d23dd 7726 (ix86_expand_prologue): Likewise.
7727 (ix86_expand_split_stack_prologue): Likewise.
7728 (legitimate_pic_address_disp_p): Likewise.
7729 (legitimize_pic_address): Likewise.
7730 (legitimize_tls_address): Likewise.
7731 (legitimize_pe_coff_symbol): Likewise.
7732 (output_pic_addr_const): Likewise.
7733 (construct_plt_address): Likewise.
7734 (ix86_expand_call): Likewise.
7735 (x86_output_mi_thunk): Likewise.
7736 (x86_function_profiler): Likewise.
7737
e30d3162 77382013-05-13 Sofiane Naci <sofiane.naci@arm.com>
7739
7740 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
7741 similar switch cases.
7742 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
7743 (aarch64_simd_mov_to_<mode>low): Delete.
7744 (aarch64_simd_mov_to_<mode>high): Delete.
7745 (move_lo_quad_<mode>): Add w<-r alternative.
7746 (aarch64_simd_move_hi_quad_<mode>): Likewise.
7747 (aarch64_simd_mov_from_*): Update type attribute.
7748 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
7749 statement.
7750
54ef9b16 77512013-05-13 Jan Hubicka <jh@suse.cz>
7752
7753 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
7754 * config/i386/i386.c (ix86_compute_frame_layout,
c27c2857 7755 ix86_expand_epilogue, emit_i387_cw_initialization,
7756 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
7757 ix86_local_alignment): Fix use of size/speed predicates.
54ef9b16 7758
043ce677 77592013-05-13 Jakub Jelinek <jakub@redhat.com>
7760
7761 PR tree-optimization/45216
7762 PR tree-optimization/57157
7763 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
7764 the (-Y) & (B - 1) variant if OP is |.
7765 * expmed.c (expand_shift_1): For rotations by const0_rtx just
7766 return shifted. Use (-op1) & (prec - 1) as other_amount
7767 instead of prec - op1.
7768
096295f6 77692013-05-13 Martin Jambor <mjambor@suse.cz>
7770
7771 PR middle-end/42371
7772 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
7773 (ipa_constant_data): New type.
7774 (ipa_jump_func): Use ipa_constant_data to hold information about
7775 constant jump functions.
7776 (ipa_get_jf_constant): Adjust to jump function type changes.
7777 (ipa_get_jf_constant_rdesc): New function.
7778 (ipa_param_descriptor): New field controlled_uses.
7779 (ipa_get_controlled_uses): New function.
7780 (ipa_set_controlled_uses): Likewise.
7781 * ipa-ref.h (ipa_find_reference): Declare.
7782 * ipa-prop.c (ipa_cst_ref_desc): New type.
7783 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
7784 changes.
7785 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
7786 New parameter cs. Adjust all callers.
7787 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
7788 (remove_described_reference): New function.
7789 (jfunc_rdesc_usable): Likewise.
7790 (try_make_edge_direct_simple_call): Decrement controlled use count,
7791 attempt to remove reference if it hits zero.
7792 (combine_controlled_uses_counters): New function.
7793 (propagate_controlled_uses): Likewise.
7794 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
7795 (ipa_edge_duplication_hook): Duplicate reference descriptions.
7796 (ipa_print_node_params): Print described use counter.
7797 (ipa_write_jump_function): Adjust to jump function type changes.
7798 (ipa_read_jump_function): New parameter CS, pass it to
7799 ipa_set_jf_constant. Adjust caller.
7800 (ipa_write_node_info): Stream controlled use count
7801 (ipa_read_node_info): Likewise.
7802 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
7803 asserting.
7804 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
7805 count. Remove cloning-added reference if it reaches zero.
7806 * ipa-ref.c (ipa_find_reference): New function.
7807
3a407061 78082013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
7809
c27c2857 7810 * config/i386/i386.c (processor_target_table): Modified default
3a407061 7811 alignment values for AMD BD and BT architectures.
7812
67f16f2b 78132013-05-13 Marc Glisse <marc.glisse@inria.fr>
7814
7815 * tree-vect-generic.c (uniform_vector_p): Move ...
7816 * tree.c (uniform_vector_p): ... here.
7817 * tree.h (uniform_vector_p): Declare it.
7818 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
7819 into a scalar.
7820
22b4b13d 78212013-05-13 Jakub Jelinek <jakub@redhat.com>
7822
2047c70c 7823 PR tree-optimization/57230
7824 * tree-ssa-strlen.c (handle_char_store): Record length for
7825 array store from STRING_CST.
7826
22b4b13d 7827 PR tree-optimization/57230
7828 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
7829 check.
7830
c85716aa 78312013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
7832
7833 * config/epiphany/epiphany.c (epiphany_init): Check size of
7834 NUM_MODES_FOR_MODE_SWITCHING.
7835 (epiphany_expand_prologue):
7836 Remove CONFIG_REGNUM initial value handling code.
7837 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
7838 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
4af825f3 7839 (emit_set_fp_mode, epiphany_mode_after): Likewise.
c85716aa 7840 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
7841 Don't return 1 for FP_MODE_NONE.
7842 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
7843 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
7844 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
7845 * config/epiphany/epiphany.md (save_config): New pattern.
7846
5afd8bcb 78472013-05-12 Uros Bizjak <ubizjak@gmail.com>
7848
7849 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
7850
f66a9337 78512013-05-10 Uros Bizjak <ubizjak@gmail.com>
7852
7853 * config/i386/i386.md (memory): Handle sseishft1.
7854 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
7855 (*vec_extractv2di_1): Ditto.
7856
904ea1db 78572013-05-10 Vladimir Makarov <vmakarov@redhat.com>
7858
7859 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
7860 saved registers.
7861
82b24465 78622013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
e70cd596 7863
7864 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
7865 Add mthumb/march=armv7-a multilib.
7866 Add mthumb/march=armv7-r multilib.
7867 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
7868
1bc74e6e 78692013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
7870
7871 * config/v850/t-rtems: Add more multilibs.
7872
fca2aa67 78732013-05-10 Richard Biener <rguenther@suse.de>
7874
7875 PR tree-optimization/57214
7876 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
7877 not propagate from SSA names that occur in abnormal PHI nodes.
7878
432dd330 78792013-05-10 Marc Glisse <marc.glisse@inria.fr>
7880
7881 * stor-layout.c (element_precision): New function.
7882 * machmode.h (element_precision): Declare it.
7883 * tree.c (build_minus_one_cst): New function.
7884 (element_precision): Likewise.
7885 * tree.h (build_minus_one_cst): Declare new function.
7886 (element_precision): Likewise.
7887 * fold-const.c (operand_equal_p): Use element_precision.
7888 (fold_binary_loc): Handle vector types.
7889 * convert.c (convert_to_integer): Use element_precision.
7890 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
7891 separately.
7892
287e3614 78932013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
7894
7895 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
7896 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
7897 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
7898 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
7899 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
7900 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
7901 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
7902 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
7903 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
7904 (Uuw8): New constraints.
7905 (Usb4): Move into alphabetical order.
7906 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
7907 (sd8_operand, ub8_operand, uw8_operand): New predicates.
7908 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
7909 previously unnamed patterns.
7910 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
7911 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
7912 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
7913 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
7914 of set_attr_alternative/if_then_else. Use extended_mips16 instead
7915 of specific lengths.
7916
3b8827a2 79172013-05-10 Jakub Jelinek <jakub@redhat.com>
7918
bf7ec243 7919 * config/i386/i386.md (rotateinv): New code attr.
7920 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
7921 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
7922 roll $31, %eax, etc.
7923
3b8827a2 7924 PR tree-optimization/45216
7925 PR tree-optimization/57157
7926 * tree-ssa-forwprop.c (simplify_rotate): New function.
7927 (ssa_forward_propagate_and_combine): Call it.
7928
2cd0995e 79292013-05-10 Richard Biener <rguenther@suse.de>
7930
7931 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
7932 disable peeling when we version for aliasing.
7933 (vector_alignment_reachable_p): Honor explicit user alignment.
7934 (vect_supportable_dr_alignment): Likewise.
7935 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
7936 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
7937 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
7938 then peeling to arrange for the cost-model check to come first.
7939
15ba1c5f 79402013-05-10 Alan Modra <amodra@gmail.com>
7941
7942 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
7943 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
7944 * configure: Regenerate.
7945
916b643b 79462013-05-10 Alan Modra <amodra@gmail.com>
7947
7948 PR target/55033
7949 * varasm.c (default_elf_select_section): Move !DECL_P check..
7950 (get_named_section): ..to here before calling get_section_name.
7951 Adjust assertion.
7952 (default_section_type_flags): Add DECL_P check.
7953 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
7954 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
7955
3f5a5361 79562013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
7957
7958 * config/epiphany/epiphany.c (epiphany_expand_prologue):
7959 When using gen_stack_adjust_str with a register offset, add a
7960 REG_FRAME_RELATED_EXPR note.
7961
b8403b2e 79622013-05-09 Uros Bizjak <ubizjak@gmail.com>
7963
7964 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
7965 (*vec_extractv4si_zext_mem): Ditto.
7966 (*vec_extractv2di): Add 0->x and x->x alternatives.
7967 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
7968 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
7969
4fa680c3 79702013-05-09 Jason Merrill <jason@redhat.com>
7971
b749017f 7972 N3639 C++1y VLA support
7973 * gimplify.c (gimplify_vla_decl): Don't touch an existing
7974 DECL_VALUE_EXPR.
7975
4fa680c3 7976 * tree.c (build_constructor_va): New.
7977 * tree.h: Declare it.
7978
0a2d5b65 79792013-05-09 Martin Jambor <mjambor@suse.cz>
7980
7981 PR lto/57084
7982 * gimple-fold.c (canonicalize_constructor_val): Call
7983 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
7984
85b7ebb2 79852013-05-09 Jan Hubicka <jh@suse.cz>
7986 Richard Biener <rguenther@suse.de>
7987
7988 PR lto/54095
7989 * symtab.c (symtab_make_decl_local): Do not add private names.
7990
6f4cfc6d 79912013-05-09 Jan Hubicka <jh@suse.cz>
7992
7993 PR lto/54095
7994 * symtab.c (insert_to_assembler_name_hash): Handle clones.
7995 (unlink_from_assembler_name_hash): Likewise.
7996 (symtab_prevail_in_asm_name_hash, symtab_register_node,
7997 symtab_unregister_node, symtab_initialize_asm_name_hash,
7998 change_decl_assembler_name): Update.
7999
eff87958 80002013-05-09 Sofiane Naci <sofiane.naci@arm.com>
8001
8002 * config/aarch64/aarch64.md: New movtf split.
8003 (*movtf_aarch64): Update.
8004 (aarch64_movdi_tilow): Handle TF modes and rename to
8005 aarch64_movdi_<mode>low.
8006 (aarch64_movdi_tihigh): Handle TF modes and rename to
8007 aarch64_movdi_<mode>high
8008 (aarch64_movtihigh_di): Handle TF modes and rename to
8009 aarch64_mov<mode>high_di
8010 (aarch64_movtilow_di): Handle TF modes and rename to
8011 aarch64_mov<mode>low_di
8012 (aarch64_movtilow_tilow): Remove spurious whitespace.
8013 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8014 splits.
8015 (aarch64_print_operand): Update.
8016
e0b071f6 80172013-05-09 Alan Modra <amodra@gmail.com>
8018
8019 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8020 powerpc64le.
8021 * configure: Regenerate.
8022
823a2ddd 80232013-05-08 Uros Bizjak <ubizjak@gmail.com>
8024
8025 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8026 splitter preparation statements.
8027 * config/i386/sse.md (*vec_extract* splitters): Ditto.
8028 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8029 adjust_address_nv.
8030
d873e825 80312013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8032
8033 * gimple-ssa-strength-reduction.c (count_candidates): Change
8034 return value to int.
8035 (analyze_candidates_and_replace): Change type of length to int.
8036
27fc86e0 80372013-05-08 Uros Bizjak <ubizjak@gmail.com>
8038
8039 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8040 (*vec_extract<mode>): Use VI12_128 mode iterator.
8041 (*vec_extract<mode>_mem): Ditto.
8042 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8043 attribute.
8044
45cde32d 80452013-05-08 Diego Novillo <dnovillo@google.com>
8046
8047 PR bootstrap/54659
8048
8049 Revert:
45cde32d 8050 2012-08-17 Diego Novillo <dnovillo@google.com>
8051
27fc86e0 8052 PR bootstrap/54281
8053 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8054 * config.in: Regenerate.
8055 * configure: Regenerate.
8056 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
45cde32d 8057
5c368d8a 80582013-05-08 Jan Hubicka <jh@suse.cz>
8059
8060 PR lto/54095
8061 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8062 * cgraph.h (symtab_node_base): Add unique_name.
8063 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8064 input_overwrite_node, input_varpool_node): Stream unique_name.
8065 * cgraphclones.c (cgraph_create_virtual_clone,
8066 cgraph_function_versioning): Set unique_name.
8067 * ipa.c (function_and_variable_visibility): Set unique_name.
8068
8858a76d 80692013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8070
8071 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8072 (alloc_cand_and_find_basis): Restrict conditional candidate
8073 processing to CAND_MULTs.
8074
48bbff00 80752013-05-08 Jan Hubicka <jh@suse.cz>
8076
8077 PR lto/54095
8078 lto-symtab.c (lto_symtab_symbol_p): New function.
8079 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8080 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8081 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8082 Skip static symbols.
8083
531a0a15 80842013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
8085
8086 PR tree-optimization/57200
8087 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8088 Only call inform if the preceding warning_at returns true.
8089
b156ec37 80902013-05-07 Han Shen <shenhan@google.com>
8091
8092 * cfgexpand.c (record_or_union_type_has_array_p): New function.
8093 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8094 * common.opt (fstack-protector-strong): New option.
8095 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8096 * doc/invoke.texi (Optimization Options): Document
8097 "-fstack-protector-strong".
8098 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8099
5c2d214d 81002013-05-06 Steven Bosscher <steven@gcc.gnu.org>
8101
8102 * config/mips/mips.c (mips_machine_reorg2): Return 0.
8103
c625778b 81042013-05-07 Vladimir Makarov <vmakarov@redhat.com>
8105
8106 * ira.c (update_equiv_regs): Add insn having equiv memory even if
8107 it is not lhs of the insn.
8108 (setup_reg_equiv): Remove insn having equiv memory which it is not
8109 lhs of the insn.
8110 * lra-constraints.c (process_address): Try to improve generation
8111 code for address base + disp.
8112 (lra_constraints): Make correct the code for checking insn setting
8113 up backward equivalence. Remove insn only if it is in the init
8114 insn list.
8115 * lra-eliminations.c (update_reg_eliminate): Change return value.
8116 (lra_eliminate): Use the result.
8117
1087c60b 81182013-05-07 Uros Bizjak <ubizjak@gmail.com>
8119
8120 * config/i386/sse.md (ssescalarnummask): New mode attribute.
8121 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8122 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8123 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8124 register target operands.
8125 (*vec_extractv8hi_sse2): New pattern.
8126 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8127 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8128 (*vec_extract<mode>_mem): New insn and split pattern.
8129
3e8585ab 81302013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
8131
8132 * config/arm/arm.c (arm_asan_shadow_offset): New function.
8133 (TARGET_ASAN_SHADOW_OFFSET): Define.
8134 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8135 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8136
d62a8480 81372013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8138
8139 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8140 (incr_vec_index): Return -1 if increment not found.
8141 (create_add_on_incoming_edge): Assert if increment not found.
8142 (record_increment): Limit number of increments recorded.
8143 (all_phi_incrs_profitable): Return false if an increment not found.
8144 (replace_profitable_candidates): Don't process increments that were
8145 not recorded.
8146 (analyze_candidates_and_replace): Limit size of incr_vec.
8147
3621f68e 81482013-05-07 Richard Biener <rguenther@suse.de>
8149
8150 * calls.c (special_function_p): setjmp-like functions are leaf.
8151 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8152 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8153
e0e03aa1 81542013-05-07 Sofiane Naci <sofiane.naci@arm.com>
8155
8156 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8157 (aarch64_simd_mov<mode>): New expander.
8158 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8159 (aarch64_simd_mov_to_<mode>high): Likewise.
8160 (aarch64_simd_mov_from_<mode>low): Likewise.
8161 (aarch64_simd_mov_from_<mode>high): Likewise.
8162 (aarch64_dup_lane<mode>): Update.
8163 (aarch64_dup_lanedi): New instruction pattern.
8164 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8165 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8166
d54ab372 81672013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8168
8169 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8170 (replace_mult_candidate): Remove unnecessary argument; remove
8171 unnecessary parameter from call to introduce_cast_before_cand.
8172 (replace_unconditional_candidate): Remove unnecessary parameter
8173 from call to replace_mult_candidate.
8174 (replace_conditional_candidate): Likewise.
8175 (insert_initializers): Use make_temp_ssa_name.
8176 (introduce_cast_before_cand): Remove unnecessary argument; use
8177 make_temp_ssa_name.
8178 (replace_one_candidate): Remove unnecessary argument; remove
8179 unnecessary parameter from calls to introduce_cast_before_cand.
8180 (replace_profitable_candidates): Remove unnecessary parameters
8181 from calls to replace_one_candidate.
8182
cebae8f7 81832013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8184
8185 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8186 phi def as possibly hiding a basis for a CAND_ADD whose operands
8187 have been commuted in the analysis.
8188 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8189
dcb79d5f 81902013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
8191
8192 * config/aarch64/aarch64.md
8193 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8194 shift value between 0-4.
8195
75aefb7b 81962013-05-07 Richard Biener <rguenther@suse.de>
8197
8198 * double-int.h (rshift): New overload.
8199 * double-int.c (rshift): New function.
8200 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8201 (create_reference_ops_from_ref): Remove.
8202 (vn_reference_insert): Use shared ops for constructing the
8203 reference and copy it.
8204
587a19f1 82052013-05-07 Richard Biener <rguenther@suse.de>
8206
8207 PR middle-end/57190
8208 * tree-eh.c (sink_clobbers): Properly propagate
8209 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8210
03404fe6 82112013-05-07 Jakub Jelinek <jakub@redhat.com>
8212
182debc1 8213 PR tree-optimization/57149
8214 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
8215 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
8216 collect_phi_def_edges, execute_late_warn_uninitialized): Use
8217 uninit_undefined_value_p instead of ssa_undefined_value_p.
8218
03404fe6 8219 PR debug/57184
8220 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
8221 for modifier == EXPAND_INITIALIZER.
8222
7c291732 82232013-05-07 Anton Blanchard <anton@samba.org>
8224
8225 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
8226 for powerpc64 little endian.
8227 * configure: Regenerate.
8228
6a219c68 82292013-05-06 Graham Stott <grahams@btinternet.com>
8230
8231 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
8232 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
8233 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
8234 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
8235
82362013-05-06 Graham Stott <grahams@btinternet.com>
7e91113e 8237
8238 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
8239 codes which allow non-lvalues.
8240
5a4f3016 82412013-05-06 Marc Glisse <marc.glisse@inria.fr>
8242
8243 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
8244 components are all 1s.
8245 (integer_minus_onep): New function.
8246 * tree.h (integer_minus_onep): Declare it.
8247 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
8248 integer_minus_onep instead of integer_all_onesp.
8249
8de9db6e 82502013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
8251
8252 PR target/52933
8253 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
8254 variations of these patterns.
8255
2071a184 82562013-05-06 Uros Bizjak <ubizjak@gmail.com>
8257
8258 * config/i386/i386.md (isa): Add x64_sse4 member.
8259 (enabled): Handle x64_sse4.
8260 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
8261 instruction for 64bit SSE4_1 targets. Update insn attributes.
8262 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
8263 instruction for SSE4_1 targets. Update insn attributes.
8264 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8265 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
8266 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
8267 const_1 selector.
8268 (*vec_extractv4si): Rename from *sse4_1_pextrd.
8269 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
8270 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
8271
e3c1cc51 82722013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
8273
8274 PR target/57108
8275 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
8276
82353beb 82772013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
8278
8279 * final.c (do_assembler_dialects): Don't handle curly braces and
8280 vertical bar escaped by % as dialect delimiters.
8281 (output_asm_insn): Print curly braces and vertical bar if escaped
8282 by % and ASSEMBLER_DIALECT defined.
8283 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
8284 * doc/tm.texi: Regenerated.
8285
5a2fb01f 82862013-05-06 Steven Bosscher <steven@gcc.gnu.org>
8287
5a2fb01f 8288 * config/mips/mips.c: Include tree-pass.h.
8289 (mips_reorg): Split in pre- and post-dbr_schedule parts.
8290 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
8291 (pass_mips_machine_reorg2): New machine specific pass.
8292 (insert_pass_mips_machine_reorg2): New pass plugin definition.
8293 (mips_option_override): Register the new pass.
8294 * rtl.h (cleanup_barriers): Remove prototype.
8295 (dbr_schedule): Likewise.
8296 * jump.c (cleanup_barriers): Make static.
8297 * reorg.c (dbr_schedule): Likewise.
8298
59c0ed80 82992013-05-06 Richard Biener <rguenther@suse.de>
8300
8301 PR tree-optimization/57185
8302 * tree-parloops.c (add_field_for_reduction): Handle anonymous
8303 SSA names properly.
8304
0b9c4816 83052013-05-06 Uros Bizjak <ubizjak@gmail.com>
8306
8307 PR target/57106
8308 * config/i386/i386.c (add_parameter_dependencies): Add dependence
8309 between "first_arg" and "insn", not "last" and "insn".
8310
7090b697 83112013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8312
8313 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
8314 (find_candidates_in_block): Re-enable slsr_process_phi.
8315 (create_phi_basis): Fix double counting of candidate adjustment.
8316
e2a6b9da 83172013-05-06 Richard Biener <rguenther@suse.de>
8318
8319 PR middle-end/57147
8320 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
8321 the edge is also fallthru, preserve it and just clear the
8322 abnormal flag.
8323 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
8324 also complex, preserve that and just clear the fallthru flag.
8325 * tree-inline.c (update_ssa_across_abnormal_edges): Also
8326 update virtual operands.
8327
a6245fde 83282013-05-06 Alan Modra <amodra@gmail.com>
8329
8330 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
8331 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
8332 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
8333 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
8334 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8335 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
8336
23c6c8a2 83372013-05-06 Alan Modra <amodra@gmail.com>
8338
8339 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
8340 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
8341 (DEFAULT_ASM_ENDIAN): Define.
8342 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8343 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
8344 Update -K PIC clause from sysv4.h.
8345 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
8346 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
8347
75e370bf 83482013-05-06 Alan Modra <amodra@gmail.com>
8349
8350 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
8351 twice for little-endian.
8352 (ashrdi3_no_power, ashrdi3): Support little-endian.
8353
a61eea5a 83542013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
8355
8356 PR target/55303
8357 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
8358 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
8359 related expanders.
8360 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
8361 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
8362 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
8363 New predicates.
8364
b8f55b74 83652013-05-05 Steven Bosscher <steven@gcc.gnu.org>
8366 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
8367
8368 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
8369 * config/pa/pa.opt: Make mbig-switch a no-op.
8370 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
8371 (CASE_VECTOR_MODE): Always return SImode.
8372 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
8373 for the !TARGET_BIG_SWITCH case.
8374 * config/pa/pa-linux.h: Likewise.
8375 * config/pa/pa-openbsd.h: Likewise.
8376 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
8377 * config/pa/pa.md (short_jump): Remove define_insn.
8378 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
8379 (casesi0): Remove define_insn.
8380 (type): Remove btable_branch.
8381 (pa_combine_type): Likewise.
8382 (in_nullified_branch_delay): Likewise.
8383 (in_call_delay): Likewise.
8384 (define_delay): Likewise.
8385 (define_insn_reservation "Z3"): Likewise.
8386 (define_insn_reservation "Z4"): Likewise.
8387 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
8388 (pa_adjust_insn_length): Remove adjustment for btable branches.
8389 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
8390 and mno-big-switch
8391
fe4df2ce 83922013-05-05 Uros Bizjak <ubizjak@gmail.com>
8393
8394 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8395 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
8396 Add m->r,x alternatives.
8397 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
8398 splitters using SWI48x mode iterator.
8399 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
8400 TARGET_64BIT. Add m->x alternative.
8401 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
8402 Add o->x alternative. Enable for TARGET_SSE.
8403 (sse_storeq): Remove expander.
8404 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
8405 with memory input operand.
8406 (*vec_extractv2di_1 splitter): New.
8407 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
8408 * config/i386/i386.md (ssevecmodelower): New mode attribute.
8409
97f92bff 84102013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
8411
8412 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
8413 (INT_LOWPART): Delete.
8414 (extract_MB): Adjust.
8415 (extract_ME): Adjust.
8416 (print_operand): Adjust.
8417
42408413 84182013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
8419
8420 * config/rs6000/predicates.md (reg_or_add_cint_operand,
8421 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8422 (reg_or_logical_cint_operand, easy_fp_constant,
8423 logical_const_operand): Delete "CONST_DOUBLE" case.
8424 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
8425 "HOST_BITS_PER_WIDE_INT == 64" test.
fe4df2ce 8426 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
42408413 8427 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
8428 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
8429 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
8430 test.
8431 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
8432 CONST_DOUBLE DImode/VOIDmode case.
8433 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
8434 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
8435 CONST_DOUBLE VOIDmode case.
8436 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8437 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
8438 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
8439 Delete CONST_DOUBLE case.
8440 (splitters for mov FMOVE64 const_double): Delete
8441 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
8442 "HOST_BITS_PER_WIDE_INT >= 64" test.
8443 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
8444 case.
8445 (mov DI const_double): Delete.
8446
fd304a93 84472013-05-04 Jakub Jelinek <jakub@redhat.com>
8448
8449 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
8450 on op shows all bits zero in mode of a lowpart subreg, return zero.
8451
21e5dc01 84522013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8453
8454 PR target/57150
8455 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
8456 to save TFmode registers and DImode to save TImode registers for
8457 caller save operations.
8458 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
8459 mark being partially clobbered since they only use the first
8460 double word.
8461
8462 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
8463 and TDmode only use the upper 64-bits of each VSX register.
8464
e64c90eb 84652013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8466
8467 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
8468 (find_candidates_in_block): Disable slsr_process_phi.
8469
bd13f5e8 84702013-05-03 Guozhi Wei <carrot@google.com>
8471
8472 * coverage.c (coverage_obj_init): Move the construction of gcov
fe4df2ce 8473 constructor to ...
8474 (build_init_ctor): ... here.
bd13f5e8 8475
1168caca 84762013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8477
8478 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
8479 (slsr_cand_d): Redefine def_phi.
8480 (stride_status, phi_adjust_status, count_phis_status): New enums.
8481 (find_phi_def): New.
8482 (find_basis_for_base_expr): New.
8483 (find_basis_for_candidate): Handle hidden bases.
8484 (alloc_cand_and_find_basis): Handle phi candidates.
8485 (slsr_process_phi): New.
8486 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
8487 (create_mul_imm_cand): Likewise.
8488 (create_add_ssa_cand): Exclude phi base candidates.
8489 (create_add_imm_cand): Likewise.
8490 (slsr_process_cast): Likewise.
8491 (slsr_process_copy): Likewise.
8492 (find_candidates_in_block): Handle phi candidates.
8493 (dump_candidate): Likewise.
8494 (unconditional_cands): Delete.
8495 (unconditional_cands_with_known_stride_p): Delete.
8496 (phi_dependent_cand_p): New.
8497 (cand_increment): Handle phi-dependent candidates.
8498 (replace_dependent): Delete.
8499 (replace_mult_candidate): New.
8500 (replace_unconditional_candidate): New.
8501 (incr_vec_index): Move to avoid forward reference.
8502 (create_add_on_incoming_edge): New.
8503 (create_phi_basis): New.
8504 (replace_dependents): Delete.
8505 (replace_conditional_candidate): New.
8506 (phi_add_costs): New.
8507 (replace_uncond_cands_and_profitable_phis): New.
8508 (record_increment): Handle phi adjustments.
8509 (record_phi_increments): New.
8510 (record_increments): Handle phi adjustments.
8511 (phi_incr_cost): New.
8512 (lowest_cost_path): Handle phis.
8513 (total_savings): Likewise.
8514 (analyze_increments): Likewise.
8515 (ncd_with_phi): New.
8516 (ncd_of_cand_and_phis): New.
8517 (nearest_common_dominator_for_cands): Handle phi increments.
8518 (all_phi_incrs_profitable): New.
8519 (replace_profitable_candidates): Handle phi-dependent candidates.
8520 (analyze_candidates_and_replace): Likewise.
8521
df21e330 85222013-05-03 Teresa Johnson <tejohnson@google.com>
8523
8524 PR bootstrap/57154
8525 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
8526 do not exceed REG_BR_PROB_BASE.
8527
d06dae50 85282013-05-03 Jeff Law <law@redhat.com>
8529
66c357ba 8530 PR tree-optimization/57144
d06dae50 8531 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
8532 operand of the condition will bit into the new type when eliminating
8533 a cast feeding a condition.
8534
cea05f44 85352013-05-03 Jakub Jelinek <jakub@redhat.com>
8536
8537 PR rtl-optimization/57130
fe4df2ce 8538 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
8539 of COMPARE as in_code to the recursive call if needed.
cea05f44 8540
65c52515 85412013-05-03 Uros Bizjak <ubizjak@gmail.com>
8542
8543 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
8544 (enabled): Handle new members.
8545 * config/i386/sse.md (*vec_concatv2si): Merge from
8546 *vec_concatv2si_sse2 and vec_concatv2si_sse.
8547 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
8548
058e9571 85492013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
8550
8551 PR tree-optimization/57027
8552 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
8553 for fnms opportunity, check we got the prerequisite kind
8554 of tree / gimple before using accessor functions.
8555
9a8b9e26 85562013-05-03 Richard Biener <rguenther@suse.de>
8557
8558 * double-int.h (lshift): New overload without precision
8559 and arith argument.
8560 (operator *=, operator +=, operator -=): Move ...
8561 * double-int.c (operator *=, operator +=, operator -=): ... here
8562 and implement more efficiently.
8563 (mul_double_with_sign): Remove.
8564 (lshift_double): Adjust to take unsinged shift argument, push
8565 dispatching code to callers.
8566 (mul_double_wide_with_sign): Add early out for callers that
8567 are not interested in high parts or overflow.
8568 (lshift): New function.
8569 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
8570 dispatch code here.
8571 (lrotate, rrotate): Use logical shifts.
8572 * expr.c (get_inner_reference): Use lshift.
8573 * fixed-value.c (do_fixed_divide): Likewise.
8574 * tree-dfa.c (get_ref_base_and_extent): Likewise.
8575 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
8576 (indirect_refs_may_alias_p): Likewise.
8577 (stmt_kills_ref_p_1): Likewise.
8578
91b3c1af 85792013-05-03 Vidya Praveen <vidyapraveen@arm.com>
8580
8581 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
8582
4d9d91e8 85832013-05-03 Vidya Praveen <vidyapraveen@arm.com>
8584
8585 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
8586 scalar form of FABD instruction.
8587
04dda2a2 85882013-05-02 Vladimir Makarov <vmakarov@redhat.com>
8589
8590 * lra-constraints.c (process_alt_operands): Add checking alt
8591 number to choose the best alternative.
8592
077b5b8a 85932013-05-02 Richard Biener <rguenther@suse.de>
8594
8595 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
8596 bitmap and its handling.
8597 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
8598
92b43d63 85992013-05-02 Richard Biener <rguenther@suse.de>
8600
8601 PR middle-end/57140
8602 * tree-inline.c (copy_loops): Properly handle removed loops.
8603 (copy_cfg_body): Mark destination loops for fixup if source
8604 loops needed fixup.
8605
d93eb6ad 86062013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
8607
8608 PR target/56732
8609 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
8610 generating simple_return for naked functions.
8611
c42e4f2e 86122013-05-02 Martin Jambor <mjambor@suse.cz>
8613
8614 PR middle-end/56988
8615 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
8616 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
8617 flags match.
8618 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
8619 ipa_agg_replacement_value structures.
8620 (known_aggs_to_agg_replacement_list): Likewise.
8621 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
8622 (read_agg_replacement_chain): Likewise.
8623 (ipcp_transform_function): Also check that by_ref flags match.
8624
73c6d54e 86252013-05-02 Richard Biener <rguenther@suse.de>
8626
8627 * graphds.h (struct graph): Add obstack member.
8628 * graphds.c (new_graph): Initialize obstack and allocate
8629 vertices from it.
8630 (add_edge): Allocate edge from the obstack.
65c52515 8631 (free_graph): Free the obstack instead of all edges and vertices.
73c6d54e 8632
70074000 86332013-05-02 Teresa Johnson <tejohnson@google.com>
8634
8635 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
8636 divides.
8637 * cfg.c (update_bb_profile_for_threading): Ditto.
8638 * tree-inline.c (copy_bb): Ditto.
8639 (copy_edges_for_bb): Ditto.
8640 (initialize_cfun): Ditto.
8641 (copy_cfg_body): Ditto.
8642 (expand_call_inline): Ditto.
8643 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
8644 (estimate_node_size_and_time): Ditto.
8645 (inline_merge_summary): Ditto.
8646 * cgraphclones.c (cgraph_clone_edge): Ditto.
8647 (cgraph_clone_node): Ditto.
8648 * sched-rgn.c (compute_dom_prob_ps): Ditto.
8649 (compute_trg_info): Ditto.
8650
4817649a 86512013-05-02 Ian Bolton <ian.bolton@arm.com>
8652
8653 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
8654 S reg when fp attribute set.
8655 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
8656
623ae2ee 86572013-05-02 Ian Bolton <ian.bolton@arm.com>
8658
8659 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
8660 New pattern.
8661 (*and_one_cmplsi3_compare0_uxtw): Likewise.
8662 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
8663 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
8664
e23d3784 86652013-05-02 Richard Biener <rguenther@suse.de>
8666
8667 * tree-scalar-evolution.c (scev_info_hasher): Remove.
8668 (struct instantiate_cache_entry): New type.
8669 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
8670 (struct instantiate_cache_type): New type.
8671 (set_instantiated_value, get_instantiated_value): Remove.
8672 (get_instantiated_value_entry): New function.
8673 (instantiate_scev_name): Use the new cache and adjust.
8674 (instantiate_scev_poly): Adjust.
8675 (instantiate_scev_binary): Likewise.
8676 (instantiate_array_ref): Likewise.
8677 (instantiate_scev_convert): Likewise.
8678 (instantiate_scev_not): Likewise.
8679 (instantiate_scev_3): Likewise.
8680 (instantiate_scev_2): Likewise.
8681 (instantiate_scev_r): Likewise.
8682 (instantiate_scev): Likewise.
8683 (resolve_mixers): Likewise.
8684
273c330a 86852013-05-01 Vladimir Makarov <vmakarov@redhat.com>
8686
8687 PR target/57091
8688 * lra-constraints.c (best_small_class_operands_num): Remove.
8689 (process_alt_operands): Remove small_class_operands_num. Take
8690 small classes operands into losers and only if the operand is not
8691 matched. Modify debugging output.
8692 (curr_insn_transform): Remove best_small_class_operands_num.
8693 Print insn name.
8694
4491e251 86952013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8696
8697 * config/aarch64/aarch64-builtins.c
8698 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
8699 * config/aarch64/aarch64-simd-builtins.def
8700 (reduc_splus_): Add new modes.
8701 (reduc_uplus_): New.
8702 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
8703 (reduc_uplus_v4sf): Likewise.
8704 (reduc_splus_v4sf): Likewise.
8705 (aarch64_addv<mode>): Likewise.
8706 (reduc_uplus_<mode>): Likewise.
8707 (reduc_splus_<mode>): Likewise.
8708 (aarch64_addvv2di): Likewise.
8709 (reduc_uplus_v2di): Likewise.
8710 (reduc_splus_v2di): Likewise.
8711 (aarch64_addvv2si): Likewise.
8712 (reduc_uplus_v2si): Likewise.
8713 (reduc_splus_v2si): Likewise.
8714 (reduc_<sur>plus_<mode>): New.
8715 (reduc_<sur>plus_v2di): Likewise.
8716 (reduc_<sur>plus_v2si): Likewise.
8717 (reduc_<sur>plus_v4sf): Likewise.
8718 (aarch64_addpv4sf): Likewise.
8719 * config/aarch64/arm_neon.h
8720 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
8721 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
8722 add UNSPEC_SADDV, UNSPEC_UADDV.
8723 (SUADDV): New.
8724 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
8725
3bbdb270 87262013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8727
8728 * config/aarch64/arm_neon.h
8729 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
8730
867d4a60 87312013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8732
8733 * config/aarch64/aarch64-builtins
8734 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
8735
2520d4b7 87362013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8737
8738 * config/aarch64/aarch64-simd-builtins.def
8739 (reduc_smax_): New.
8740 (reduc_smin_): Likewise.
8741 (reduc_umax_): Likewise.
8742 (reduc_umin_): Likewise.
8743 (reduc_smax_nan_): Likewise.
8744 (reduc_smin_nan_): Likewise.
8745 (fmax): Remove.
8746 (fmin): Likewise.
8747 (smax): Update for V2SF, V4SF and V2DF modes.
8748 (smin): Likewise.
8749 (smax_nan): New.
8750 (smin_nan): Likewise.
8751 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
8752 (<su><maxmin><mode>3): ...This, refactor.
8753 (s<maxmin><mode>3): New.
8754 (<maxmin_uns><mode>3): Likewise.
8755 (reduc_<maxmin_uns>_<mode>): Refactor.
8756 (reduc_<maxmin_uns>_v4sf): Likewise.
8757 (reduc_<maxmin_uns>_v2si): Likewise.
8758 (aarch64_<fmaxmin><mode>: Remove.
8759 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
8760 new builtin names.
8761 (vmin<q>_f<32,64>): Likewise.
8762 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
8763 (FMAXMIN): New.
8764 (su): Add mappings for smax, smin, umax, umin.
8765 (maxmin): New.
8766 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
8767 (FMAXMIN): Rename as...
8768 (FMAXMIN_UNS): ...This.
8769 (maxminv): Remove.
8770 (fmaxminv): Likewise.
8771 (fmaxmin): Likewise.
8772 (maxmin_uns): New.
8773 (maxmin_uns_op): Likewise.
8774
2fe1d3d6 87752013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8776
8777 * config/aarch64/arm_neon.h
8778 (vac<ge, gt><sd>_f<32, 64>): Rename to...
8779 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
8780 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
8781
3310ab7f 87822013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8783
8784 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
8785 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
8786
a793ba7d 87872013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8788
8789 * config/aarch64/aarch64-simd.md
8790 (vcond<mode>_internal): Handle special cases for constant masks.
8791 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
8792 (vcondu<mode><mode>): Likewise.
8793 (vcond<v_cmp_result><mode>): New.
8794
c4759747 87952013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8796
8797 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
8798 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
8799 * config/aarch64/aarch64-simd-builtins.def
8800 (cmeq): Update to BUILTIN_VALLDI.
8801 (cmgt): Likewise.
8802 (cmge): Likewise.
8803 (cmle): Likewise.
8804 (cmlt): Likewise.
8805 * config/aarch64/arm_neon.h
8806 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
8807 to builtins or C as appropriate.
8808
6ba1316b 88092013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
8810
8811 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
8812 (cmgeu): ...This.
8813 (cmhi): Rename to...
8814 (cmgtu): ...This.
8815 * config/aarch64/aarch64-simd.md
8816 (simd_mode): Add SF.
8817 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
8818 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
8819 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
8820 (cstore<mode>_neg): ...This.
8821 * config/aarch64/iterators.md
8822 (VALLF): new.
8823 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
8824 (COMPARISONS): New.
8825 (UCOMPARISONS): Likewise.
8826 (optab): Add missing comparisons.
8827 (n_optab): New.
8828 (cmp_1): Likewise.
8829 (cmp_2): Likewise.
8830 (CMP): Likewise.
8831 (cmp): Remove.
8832 (VCMP_S): Likewise.
8833 (VCMP_U): Likewise.
8834 (V_cmp_result): Add DF, SF modes.
8835 (v_cmp_result): Likewise.
8836 (v): Likewise.
8837 (vmtype): Likewise.
8838 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
8839
50148b9a 88402013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
8841
8842 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
8843 define_insn to define_insn_and_split.
8844 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
8845 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
8846 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
8847 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
8848 (thumb2_negscc): Likewise.
8849
e57907f3 88502013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
8851
8852 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
8853
d9eaa291 88542013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
8855
8856 * config/arm/thumb2.md: Remove trailing whitespaces.
8857
e3fc9ffb 88582013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
8859
8860 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
8861 Use gen_int_mode rather than GEN_INT.
8862
c350d42f 88632013-04-30 H.J. Lu <hongjiu.lu@intel.com>
8864
9baae48f 8865 * value-prof.c (stream_in_histogram_value): Remove the strayed
c350d42f 8866 debug_gimple_stmt.
8867
65a02176 88682013-04-30 Richard Biener <rguenther@suse.de>
8869
8870 PR middle-end/57122
65c52515 8871 * cfghooks.c (split_edge): Properly check for the loop latch edge.
65a02176 8872
0ba38440 88732013-04-30 Richard Biener <rguenther@suse.de>
8874
8875 PR middle-end/57107
8876 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
8877
d5897457 88782013-04-30 Andrey Belevantsev <abel@ispras.ru>
8879
8880 PR rtl-optimization/56957
8881 PR rtl-optimization/57105
d5897457 8882 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
8883 variable. Use just INSN_UID for determining whether an insn
8884 should be only disconnected from the insn stream.
8885 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
8886
060d11f4 88872013-04-30 Jakub Jelinek <jakub@redhat.com>
8888
8889 PR tree-optimization/57104
8890 * tsan.c (instrument_expr): Don't instrument accesses to
8891 DECL_HARD_REGISTER VAR_DECLs.
8892
41f75a99 88932013-04-30 Richard Biener <rguenther@suse.de>
8894
8895 * function.h (loops_for_fn): New inline function.
8896 (set_loops_for_fn): Likewise.
8897 * cfgloop.h (place_new_loop): Add struct function parameter.
8898 (get_loop): Likewise.
8899 (get_loops): Likewise.
8900 (number_of_loops): Likewise.
8901 (fel_next): Adjust.
8902 (fel_init): Likewise.
8903 * cfg.c (get_loop_copy): Adjust.
8904 * cfgloop.c (flow_loops_dump): Likewise.
8905 (record_loop_exits): Likewise.
8906 (verify_loop_structure): Likewise.
8907 * cfgloopanal.c (mark_irreducible_loops): Likewise.
8908 (estimate_reg_pressure_cost): Likewise.
8909 (mark_loop_exit_edges): Likewise.
8910 * cfgloopmanip.c (place_new_loop): Likewise.
8911 (add_loop): Likewise.
8912 (duplicate_loop): Likewise.
8913 * graph.c (draw_cfg_nodes): Likewise.
8914 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
8915 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
8916 (extract_affine_chrec): Likewise.
8917 (build_scop_iteration_domain): Likewise.
8918 * graphite.c (graphite_initialize): Likewise.
8919 * ira-build.c (create_loop_tree_nodes): Likewise.
8920 (more_one_region_p): Likewise.
8921 (rebuild_regno_allocno_maps): Likewise.
8922 (mark_loops_for_removal): Likewise.
8923 (mark_all_loops_for_removal): Likewise.
8924 (remove_unnecessary_regions): Likewise.
8925 (ira_build): Likewise.
8926 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
8927 * loop-init.c (fix_loop_structure): Likewise.
8928 (gate_rtl_move_loop_invariants): Likewise.
8929 (gate_rtl_unswitch): Likewise.
8930 (gate_rtl_unroll_and_peel_loops): Likewise.
8931 (rtl_doloop): Likewise.
8932 * lto-streamer-in.c (input_cfg): Likewise.
8933 * lto-streamer-out.c (output_cfg): Likewise.
8934 * modulo-sched.c (sms_schedule): Likewise.
8935 * predict.c (tree_estimate_probability): Likewise.
8936 (tree_estimate_probability_driver): Likewise.
8937 (estimate_loops): Likewise.
8938 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
8939 (move_sese_region_to_fn): Likewise.
8940 (debug_loop_num): Likewise.
8941 * tree-chrec.c (chrec_evaluate): Likewise.
8942 (hide_evolution_in_other_loops_than_loop): Likewise.
8943 (chrec_component_in_loop_num): Likewise.
8944 (reset_evolution_in_loop): Likewise.
8945 (evolution_function_is_invariant_rec_p): Likewise.
8946 * tree-if-conv.c (main_tree_if_conversion): Likewise.
8947 * tree-inline.c (copy_loops): Likewise.
8948 (copy_cfg_body): Likewise.
8949 (tree_function_versioning): Likewise.
8950 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
8951 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
8952 Likewise.
8953 (add_to_evolution_1): Likewise.
8954 (scev_const_prop): Likewise.
8955 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
8956 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
8957 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
8958 (tree_ssa_lim_initialize): Likewise.
8959 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
8960 (verify_loop_closed_ssa): Likewise.
8961 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
8962 (tree_ssa_loop_im): Likewise.
8963 (tree_ssa_loop_unswitch): Likewise.
8964 (tree_vectorize): Likewise.
8965 (check_data_deps): Likewise.
8966 (tree_ssa_loop_ivcanon): Likewise.
8967 (tree_ssa_loop_bounds): Likewise.
8968 (tree_complete_unroll): Likewise.
8969 (tree_complete_unroll_inner): Likewise.
8970 (tree_parallelize_loops): Likewise.
8971 (tree_ssa_loop_prefetch): Likewise.
8972 (tree_ssa_loop_ivopts): Likewise.
8973 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
8974 * tree-vectorizer.c (vectorize_loops): Likewise.
8975
2090f4f0 89762013-04-29 Mike Frysinger <vapier@gentoo.org>
8977
8978 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
8979 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
8980 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
8981 with EABI_LINK_SPEC.
8982
34e882e1 89832013-04-29 Uros Bizjak <ubizjak@gmail.com>
8984
8985 PR target/44578
8986 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
8987 alternative.
8988
f7c98bb1 89892013-04-29 Vladimir Makarov <vmakarov@redhat.com>
8990
8991 PR target/57097
2090f4f0 8992 * lra-constraints.c (process_alt_operands): Discourage a bit more
f7c98bb1 8993 using memory for pseudos. Print cost dump for alternatives.
8994 Modify cost values for conflicts with early clobbers.
8995 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
8996
4d9f9757 89972013-04-29 Uros Bizjak <ubizjak@gmail.com>
8998
8999 PR target/57098
9000 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9001
35a93fee 90022013-04-29 Ian Bolton <ian.bolton@arm.com>
9003
9004 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9005 from/to S register.
9006 (movdi_aarch64): Support LDR/STR from/to D register.
9007
65c4a832 90082013-04-29 Ian Bolton <ian.bolton@arm.com>
9009
9010 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9011 or higher by default.
9012
caf1e632 90132013-04-29 Richard Biener <rguenther@suse.de>
9014
9015 PR middle-end/57075
9016 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9017 even if not adding abnormal edges for calls that can make
9018 abnormal gotos.
9019
00ef57e3 90202013-04-29 Richard Biener <rguenther@suse.de>
9021
9022 PR middle-end/57103
9023 * tree-cfg.c (move_stmt_op): Fix condition under which to update
9024 TREE_BLOCK.
9025 (move_stmt_r): Remove redundant checking.
9026
e2bc4ec8 90272013-04-29 Teresa Johnson <tejohnson@google.com>
9028
9029 PR bootstrap/57077
9030 * basic-block.h (apply_scale): New function.
9031 (apply_probability): Use apply_scale.
9032 * gimple-streamer-in.c (input_bb): Ditto.
9033 * lto-streamer-in.c (input_cfg): Ditto.
9034 * lto-cgraph.c (merge_profile_summaries): Ditto.
9035 * tree-optimize.c (execute_fixup_cfg): Ditto.
4d9f9757 9036 * tree-inline.c (copy_bb): Update comment to use apply_scale.
e2bc4ec8 9037 (copy_edges_for_bb): Ditto.
9038 (copy_cfg_body): Ditto.
9039
c48f77d5 90402013-04-29 Tom de Vries <tom@codesourcery.com>
9041
9042 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9043 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9044 (tail_merge_optimize): Handle current_loops == NULL.
9045
3172284a 90462013-04-26 Jeff Law <law@redhat.com>
9047
9048 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9049 (simplify_cond_using_ranges): Generalize code to simplify
9050 COND_EXPRs where one argument is a constant and the other
9051 is an SSA_NAME created by an integral type conversion.
9052
f15d4d96 90532013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9054
9055 * config/arm/arm.md (store_minmaxsi): Use only when
9056 optimize_insn_for_size_p.
9057
7d523484 90582013-04-29 Christian Bruel <christian.bruel@st.com>
9059
9060 PR target/57108
9061 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9062
04c2922b 90632013-04-29 Richard Biener <rguenther@suse.de>
9064
9065 PR middle-end/57089
4d9f9757 9066 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9067 loop tree make sure to schedule a fixup for the child as well.
04c2922b 9068 (expand_omp_for_generic): Properly add loops.
9069 (expand_omp_for_static_nochunk): Likewise.
9070 (expand_omp_for_static_chunk): Likewise.
9071 (expand_omp_for): For the degenerate case fixup loops.
9072 (expand_omp_sections): Fix default bb placement in loops.
9073 (expand_omp_atomic_pipeline): Properly add loops.
9074
3349fabd 90752013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9076
9077 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9078
a441795e 90792013-04-29 Tom de Vries <tom@codesourcery.com>
9080
9081 * tree-ssa-tail-merge.c: Update header comment.
9082
8111d37b 90832013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9084
9085 * config/aarch64/arm_neon.h
9086 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9087 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9088 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9089 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9090 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9091 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9092 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9093 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9094
c14c74cf 90952013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9096
9097 * config/aarch64/aarch64-simd.md
9098 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9099 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9100 fix_trunc, fixuns_trunc.
9101 (ftrunc<VDQF:mode>2): New.
9102 * config/aarch64/iterators.md (optab): Add fix, fixuns.
9103 (fix_trunc_optab): New.
9104
2ddf5043 91052013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9106
9107 * config/aarch64/aarch64-builtins.c
9108 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9109 iceilf, lround, iroundf.
9110
2a631a93 91112013-04-29 Uros Bizjak <ubizjak@gmail.com>
9112
9113 PR target/54349
9114 * config/i386/i386.h (enum ix86_tune_indices)
9115 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9116 New, split from X86_TUNE_INTER_UNIT_MOVES.
9117 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9118 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9119 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9120 (TARGET_INTER_UNIT_MOVES): Remove.
9121 * config/i386/i386.c (initial_ix86_tune_features): Update.
9122 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9123 (ix86_expand_convert_uns_didf_sse): Use
9124 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9125 (ix86_expand_vector_init_one_nonzero): Ditto.
9126 (ix86_expand_vector_init_interleave): Ditto.
9127 (inline_secondary_memory_needed): Return true for moves from SSE class
9128 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9129 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9130 * config/i386/constraints.md (Yi, Ym): Depend on
9131 TARGET_INTER_UNIT_MOVES_TO_VEC.
9132 (Yj, Yn): New constraints.
9133 * config/i386/i386.md (*movdi_internal): Change constraints of
9134 operand 1 from Yi to Yj and from Ym to Yn.
9135 (*movsi_internal): Ditto.
9136 (*movdf_internal): Ditto.
9137 (*movsf_internal): Ditto.
9138 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9139 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9140 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9141 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9142 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9143 * config/i386/sse.md (movdi_to_sse): Ditto.
9144 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9145 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9146 TARGET_INTER_UNIT_MOVES.
9147 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9148 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9149 instead of TARGET_INTER_UNIT_MOVES.
9150 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9151 operand 1 from Yi to Yj and from Ym to Yn.
9152
9f224c46 91532013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9154
9155 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9156 (float_truncate_hi_): Likewise.
9157 (float_extend_lo_): Likewise.
9158 (float_truncate_lo_): Likewise.
9159 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9160 (aarch64_float_extend_lo_v2df): Likewise.
9161 (vec_unpacks_hi_v4sf): Likewise.
9162 (aarch64_float_truncate_lo_v2sf): Likewise.
9163 (aarch64_float_truncate_hi_v4sf): Likewise.
9164 (vec_pack_trunc_v2df): Likewise.
9165 (vec_pack_trunc_df): Likewise.
9166
865dd64d 91672013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9168
9169 * config/aarch64/aarch64-builtins.c
9170 (aarch64_fold_builtin): Fold float conversions.
9171 * config/aarch64/aarch64-simd-builtins.def
9172 (floatv2si, floatv4si, floatv2di): New.
9173 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9174 * config/aarch64/aarch64-simd.md
9175 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9176 * config/aarch64/iterators.md (FLOATUORS): New.
9177 (optab): Add float, floatuns.
9178 (su_optab): Likewise.
9179
2fc9b3fc 91802013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9181
9182 * config/aarch64/aarch64-builtins.c
9183 (aarch64_builtin_vectorized_function): Use new names for
9184 fcvt builtins.
9185 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9186 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9187 (fcvtzu): Split as...
9188 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9189 (fcvtas): Split as...
9190 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9191 (fcvtau): Split as...
9192 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9193 (fcvtps): Split as...
9194 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9195 (fcvtpu): Split as...
9196 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9197 (fcvtms): Split as...
9198 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9199 (fcvtmu): Split as...
9200 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9201 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9202 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9203 (lfrintnusf, lfrintnudf): Likewise.
9204 * config/aarch64/aarch64-simd.md
9205 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9206 define_insn.
9207 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9208 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9209 (fcvt_pattern): Likewise.
9210
38c0c992 92112013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9212
9213 * config/aarch64/aarch64-simd.md
9214 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
9215 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
9216
0c882306 92172013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9218
9219 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
9220 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
9221 (vrnd<a,m,n,p>_f32): Implement using builtins.
9222 (vrnd<i,x><q>_f<32, 64>): New.
9223
abb2e961 92242013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9225
9226 * config/aarch64/aarch64-builtins.c
9227 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
9228 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
9229 (frintz): Rename to...
9230 (btrunc): ...this.
9231 (frintp): Rename to...
9232 (ceil): ...this.
9233 (frintm): Rename to...
9234 (floor): ...this.
9235 (frinti): Rename to...
9236 (nearbyint): ...this.
9237 (frintx): Rename to...
9238 (rint): ...this.
9239 (frinta): Rename to...
9240 (round): ...this.
9241 * config/aarch64/aarch64-simd.md
9242 (aarch64_frint<frint_suffix><mode>): Delete.
9243 (<frint_pattern><mode>2): Convert to insn.
9244 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
9245 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
9246 (frint_pattern): Likewise.
9247 (frint_suffix): Likewise.
9248
9584aa9d 92492013-04-29 Richard Biener <rguenther@suse.de>
9250
9251 PR tree-optimization/57081
9252 * loop-init.c: Include tree-flow.h.
9253 (loop_optimizer_finalize): Free number of iteration estimates.
9254 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
9255
ebdfd365 92562013-04-29 Jakub Jelinek <jakub@redhat.com>
9257
5b566583 9258 PR tree-optimization/57083
9259 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
9260 non-singleton shift count range, zero extend low_bound for uns case.
9261
ebdfd365 9262 * config/i386/predicates.md (general_vector_operand): New predicate.
9263 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
9264 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
9265 if they aren't nonimmediate operands. If their original values
9266 satisfy const_vector_equal_evenodd_p, don't shift them.
9267 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
9268 predicates. For the SSE4.1 case force operands[{1,2}] into registers
9269 if not nonimmediate_operand.
9270 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
9271 instead of register_operand.
9272 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
9273
992a05af 92742013-04-28 Eric Botcazou <ebotcazou@adacore.com>
c6cfb282 9275
9276 * stor-layout.c (finalize_size_functions): Allocate a structure and
9277 reset cfun before dumping the functions.
9278
a31cce64 92792013-04-27 Jakub Jelinek <jakub@redhat.com>
9280
95d51e38 9281 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
9282
a31cce64 9283 PR target/56866
9284 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
9285 use xop_pmacsdqh if uns_p.
9286 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
9287 the immediate rotate count.
9288
ea99c7a1 92892013-04-26 Vladimir Makarov <vmakarov@redhat.com>
9290
9291 * rtl.h (struct rtx_def): Add comment for field jump.
9292 (LRA_SUBREG_P): New macro.
9293 * recog.c (register_operand): Check LRA_SUBREG_P.
9294 * lra.c (lra): Add note at the end of RTL code. Align non-empty
9295 stack frame.
9296 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
9297 (lra_final_code_change): Skip subreg change for operators.
9298 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
9299 if there are no operand changes.
9300 * lra-constraints.c (curr_insn_set): New.
9301 (match_reload): Set LRA_SUBREG_P.
9302 (emit_spill_move): Ditto.
9303 (check_and_process_move): Use curr_insn_set. Process only single
9304 set insns. Don't initialize sec_mem_p and change_p.
9305 (simplify_operand_subreg): Use LRA_SUBREG_P.
9306 (reg_in_class_p): New function.
9307 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
9308 of #ifdef. Add code to remove cycling.
9309 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
9310 non-null disp. Reload inner instead of disp when base and index
9311 are null. Try to put lo_sum into register.
9312 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
9313 (check_and_process_move): Move code for move cost check to
9314 simple_move_p. Remove equiv_substitution.
9315 (simple_move_p): New function.
9316 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
9317 curr_insn_set. Call check_and_process_move only for single set
9318 insns. Use the new function. Move call of check_and_process_move
9319 after operand equiv substitution and address process.
9320
f585c295 93212013-04-26 Jakub Jelinek <jakub@redhat.com>
9322
9323 PR go/57045
9324 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
9325 with nonlocal goto receivers or returns twice calls, ignore
9326 unininitialized values from abnormal edges to nl goto receiver
9327 or returns twice call.
9328
b7219196 93292013-04-26 Jakub Jelinek <jakub@redhat.com>
9330
9331 PR tree-optimization/57051
9332 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
9333 and VEC_RSHIFT_EXPR if shift count is a multiple of element
9334 bitsize.
9335
82b40354 93362013-04-26 Richard Biener <rguenther@suse.de>
9337
9338 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
9339 (expand_omp_taskreg): Likewise. Mark loops for fixup.
9340 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
9341 (fixup_loop_arrays_after_move): New function.
9342 (move_sese_region_to_fn): Properly outline the loop tree parts
9343 of the SESE region.
9344
baad29fe 93452013-04-26 Uros Bizjak <ubizjak@gmail.com>
9346
9347 * config/i386/i386.md (type, unit): Fix long lines.
9348
7a569539 93492013-04-26 Richard Biener <rguenther@suse.de>
9350
9351 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
9352 (lto-streamer-out.o): Likewise.
9353 * cfgloop.c (init_loops_structure): Export, add struct function
9354 argument and adjust.
9355 (flow_loops_find): Adjust.
9356 * cfgloop.h (enum loop_estimation): Add EST_LAST.
9357 (init_loops_structure): Declare.
9358 * lto-streamer-in.c: Include cfgloop.h.
9359 (input_cfg): Input the loop tree.
9360 * lto-streamer-out.c: Include cfgloop.h.
9361 (output_cfg): Output the loop tree.
9362 (output_struct_function_base): Do not drop PROP_loops.
9363
f6568ea4 93642013-03-26 Richard Biener <rguenther@suse.de>
9365
9366 * tree-cfg.c (execute_build_cfg): Build the loop tree.
9367 (pass_build_cfg): Provide PROP_loops.
9368 (move_sese_region_to_fn): Remove loops that are outlined into fn
9369 for now.
9370 * tree-inline.c: Include cfgloop.h.
9371 (initialize_cfun): Do not drop PROP_loops.
9372 (copy_loops): New function.
9373 (copy_cfg_body): Copy loop structure.
9374 (tree_function_versioning): Initialize destination loop tree.
9375 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
9376 (pass_parallelize_loops): Do IL verification.
9377 * loop-init.c (loop_optimizer_init): Fixup loops if required.
9378 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
9379 the CFG make sure we fixup loops as well.
9380 * tree-ssa-tail-merge.c: Include cfgloop.h.
9381 (replace_block_by): When merging loop latches mark loops for fixup.
9382 * lto-streamer-out.c (output_struct_function_base): Drop
9383 PROP_loops for now.
9384 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
9385 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
9386 * ipa-split.c: Include cfgloop.h.
9387 (split_function): Add the new return block to the loop tree root.
9388 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
9389 whether we have removed the forwarder block.
9390 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
9391 * cfgloop.h (place_new_loop): Declare.
9392 * cfgloopmanip.c (place_new_loop): Export.
9393 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
9394 (tree-switch-conversion.o): Likewise.
9395 (tree-complex.o): Likewise.
9396 (tree-inline.o): Likewise.
9397 (tree-ssa-tailmerge.o): Likewise.
9398 (ipa-split.o): Likewise.
9399 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
9400 (tree-ssa-copy.o): Likewise.
9401 * tree-switch-conversion.c: Include cfgloop.h
9402 (process_switch): If we emit a bit-test cascade, schedule loops
9403 for fixup.
9404 * tree-complex.c: Include cfgloop.h.
9405 (expand_complex_div_wide): Properly add new basic-blocks to loops.
9406 * asan.c: Include cfgloop.h.
9407 (create_cond_insert_point): Properly add new basic-blocks to
9408 loops, schedule loop fixup.
9409 * cfgloop.c (verify_loop_structure): Check that looks are not
9410 marked for fixup.
9411 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
9412 to loops.
9413 (expand_omp_for_generic): Likewise.
9414 (expand_omp_sections): Likewise.
9415 (expand_omp_atomic_pipeline): Schedule loops for fixup.
9416 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
9417 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
9418 is initialized, not when loops are present.
9419 * tree-parloops.c (parallelize_loops): Remove checking here.
9420 * passes.c (init_optimization_passes): Schedule a copy-propagation
9421 pass before complete unrolling of inner loops.
9422
2cafe211 94232013-04-26 Jakub Jelinek <jakub@redhat.com>
9424
71eeb8d3 9425 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
9426 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
9427 (colorize_init): Add argument to _WIN32 version.
9428 * toplev.c: Include diagnostic-color.h.
9429 (process_options): Default to -fdiagnostics-color=auto if
9430 GCC_COLORS env var is in the environment.
9431 * common.opt (fdiagnostics-color=): Add Var and Init.
baad29fe 9432 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
9433 env var is in the environment, the default is auto rather than never.
71eeb8d3 9434
2cafe211 9435 * diagnostic.h (file_name_as_prefix): Add context argument.
9436 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
9437 the string as locus.
9438 * langhooks.c (lhd_print_error_function): Adjust caller.
9439
33df9a15 94402013-04-25 Lawrence Crowl <crowl@google.com>
9441
9442 * var-tracking.c (shared_hash_def::htab):
9443 Change type to hash_table. Update dependent calls and types.
9444
d9dd21a8 94452013-04-25 Lawrence Crowl <crowl@google.com>
9446
9447 * Makefile.in: Update as needed below.
9448
9449 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
9450 Move declaration to after the type's method definitons.
9451
9452 * attribs.c (htab_t scoped_attributes::attribute_hash):
9453 Change type to hash_table. Update dependent calls and types.
9454
9455 * bitmap.c (htab_t bitmap_desc_hash):
9456 Change type to hash_table. Update dependent calls and types.
9457
9458 * cselib.c (htab_t cselib_hash_table):
9459 Change type to hash_table. Update dependent calls and types.
9460
9461 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
9462 (hash_string_slot_node): Move implementation into lto-streamer.h
9463 struct string_slot_hasher.
9464 (eq_string_slot_node): Likewise.
9465
9466 * data-streamer-out.c: Update output_block::string_hash_table
9467 dependent calls and types.
9468
9469 * dwarf2cfi.c (htab_t trace_index):
9470 Change type to hash_table. Update dependent calls and types.
9471
9472 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
9473 Change type to hash_table. Update dependent calls and types.
9474 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
9475 (htab_t optimize_external_refs::map): Likewise.
9476 (htab_t output_comp_unit::extern_map): Likewise.
9477 (htab_t output_comdat_type_unit::extern_map): Likewise.
9478 (htab_t output_macinfo::macinfo_htab): Likewise.
9479 (htab_t optimize_location_lists::htab): Likewise.
9480 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
9481
9482 * except.c (htab_t ehspec_hash_type):
9483 Change type to hash_table. Update dependent calls and types.
9484 (assign_filter_values::ttypes): Likewise.
9485 (assign_filter_values::ehspec): Likewise.
9486 (sjlj_assign_call_site_values::ar_hash): Likewise.
9487 (convert_to_eh_region_ranges::ar_hash): Likewise.
9488
9489 * gcse.c (htab_t pre_ldst_table):
9490 Change type to hash_table. Update dependent calls and types.
9491
9492 * ggc-common.c (htab_t saving_htab):
9493 Change type to hash_table. Update dependent calls and types.
9494 (htab_t loc_hash): Likewise.
9495 (htab_t ptr_hash): Likewise.
9496 (call_count): Rename ggc_call_count.
9497 (call_alloc): Rename ggc_call_alloc.
9498 (loc_descriptor): Rename make_loc_descriptor.
9499 (add_statistics): Rename ggc_add_statistics.
9500
9501 * ggc-common.c (saving_htab):
9502 Change type to hash_table. Update dependent calls and types.
9503
9504 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
9505 (push_gimplify_context): Likewise.
9506 (pop_gimplify_context): Likewise.
9507 (struct gimple_temp_hash_elt): Added.
9508 (struct gimplify_hasher): Likewise.
9509 (struct gimplify_ctx.temp_htab):
9510 Change type to hash_table. Update dependent calls and types.
9511
9512 * gimple-fold.c: Include gimplify-ctx.h.
9513
9514 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
9515 Change type to hash_table. Update dependent calls and types.
9516 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
9517 avoid potential global name collision.
9518
9519 * gimplify.c: Include gimplify-ctx.h.
9520 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
9521 (htab_t gimplify_ctx::temp_htab):
9522 Update dependent calls and types for new type hash_table.
9523 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
9524 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
9525
9526 * gimplify-ctx.h: New.
9527 (struct gimple_temp_hash_elt): Move from gimplify.c.
9528 (class gimplify_hasher): New.
9529 (struct gimplify_ctx): Move from gimple.h.
9530 (htab_t gimplify_ctx::temp_htab):
9531 Change type to hash_table. Update dependent calls and types.
9532
9533 * graphite-clast-to-gimple.c: Include graphite-htab.h.
9534 (htab_t ivs_params::newivs_index):
9535 Change type to hash_table. Update dependent calls and types.
9536 (htab_t ivs_params::params_index): Likewise.
9537 (htab_t print_generated_program::params_index): Likewise.
9538 (htab_t gloog::newivs_index): Likewise.
9539 (htab_t gloog::params_index): Likewise.
9540
9541 * graphite.c: Include graphite-htab.h.
9542 4htab_t graphite_transform_loops::bb_pbb_mapping):
9543 Change type to hash_table. Update dependent calls and types.
9544
9545 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
9546 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
9547 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
9548
9549 * graphite-dependences.c: Include graphite-htab.h.
9550 (loop_is_parallel_p): Change hash table type of parameter.
9551
9552 * graphite-htab.h: New.
9553 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
9554 (extern find_pbb_via_hash): Move from graphite-poly.h.
9555 (extern loop_is_parallel_p): Move from graphite-poly.h.
9556 (extern get_loop_body_pbbs): Move from graphite-poly.h.
9557
9558 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
9559 (extern loop_is_parallel_p): Move to graphite-htab.h.
9560 (extern get_loop_body_pbbs): Move to graphite-htab.h.
9561
9562 * haifa-sched.c (htab_t delay_htab):
9563 Change type to hash_table. Update dependent calls and types.
9564 (htab_t delay_htab_i2): Likewise.
9565
9566 * ira-color.c (htab_t allocno_hard_regs_htab):
9567 Change type to hash_table. Update dependent calls and types.
9568
9569 * ira-costs.c (htab_t cost_classes_htab):
9570 Change type to hash_table. Update dependent calls and types.
9571
9572 * loop-invariant.c (htab_t merge_identical_invariants::eq):
9573 Change type to hash_table. Update dependent calls and types.
9574
9575 * loop-iv.c (htab_t bivs):
9576 Change type to hash_table. Update dependent calls and types.
9577
9578 * loop-unroll.c (htab_t opt_info::insns_to_split):
9579 Change type to hash_table. Update dependent calls and types.
9580 (htab_t opt_info::insns_with_var_to_expand): Likewise.
9581
9582 * lto-streamer.h (struct string_slot): Move from data-streamer.h
9583 (struct string_slot_hasher): New.
9584 (htab_t output_block::string_hash_table):
9585 Change type to hash_table. Update dependent calls and types.
9586
9587 * lto-streamer-in.c (freeing_string_slot_hasher): New.
9588 (htab_t file_name_hash_table):
9589 Change type to hash_table. Update dependent calls and types.
9590
9591 * lto-streamer-out.c: Update output_block::string_hash_table dependent
9592 calls and types.
9593
9594 * lto-streamer.c (htab_t tree_htab):
9595 Change type to hash_table. Update dependent calls and types.
9596
9597 * omp-low.c: Include gimplify-ctx.h.
9598
9599 * passes.c (htab_t name_to_pass_map):
9600 Change type to hash_table. Update dependent calls and types.
9601 (pass_traverse): Rename to passes_pass_traverse.
9602
9603 * plugin.c (htab_t event_tab):
9604 Change type to hash_table. Update dependent calls and types.
9605
9606 * postreload-gcse.c (htab_t expr_table):
9607 Change type to hash_table. Update dependent calls and types.
9608 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
9609
9610 * sese.c (debug_rename_map_1): Make extern.
9611 (htab_t copy_bb_and_scalar_dependences::rename_map):
9612 Change type to hash_table. Update dependent calls and types.
9613
9614 * sese.h (extern debug_rename_map): Move to .c file.
9615
9616 * store-motion.c (htab_t store_motion_mems_table):
9617 Change type to hash_table. Update dependent calls and types.
9618
9619 * trans-mem.c (htab_t tm_new_mem_hash):
9620 Change type to hash_table. Update dependent calls and types.
9621
9622 * tree-browser.c (htab_t TB_up_ht):
9623 Change type to hash_table. Update dependent calls and types.
9624
9625 * tree-cfg.c (htab_t discriminator_per_locus):
9626 Change type to hash_table. Update dependent calls and types.
9627
9628 * tree-complex.c: Include tree-hasher.h
9629 (htab_t complex_variable_components):
9630 Change type to hash_table. Update dependent calls and types.
9631
9632 * tree-eh.c (htab_t finally_tree):
9633 Change type to hash_table. Update dependent calls and types.
9634
9635 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
9636 struct int_tree_hasher.
9637 (extern int_tree_map_eq): Likewise.
9638 (uid_decl_map_hash): Removed.
9639 (extern decl_tree_map_eq): Likewise.
9640
9641 * tree-hasher.h: New.
9642 (struct int_tree_hasher): New.
9643 (typedef int_tree_htab_type): New.
9644
9645 * tree-inline.c: Include gimplify-ctx.h.
9646
9647 * tree-mudflap.c: Include gimplify-ctx.h.
9648
9649 * tree-parloops.c: Include tree-hasher.h.
9650 (htab_t eliminate_local_variables_stmt::decl_address):
9651 Change type to hash_table. Update dependent calls and types.
9652 (htab_t separate_decls_in_region::decl_copies): Likewise.
9653
9654 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
9655 Change type to hash_table. Update dependent calls and types.
9656
9657 * tree-sra.c (candidates):
9658 Change type to hash_table. Update dependent calls and types.
9659
9660 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
9661 in tree-flow.h.
9662 (int_tree_map_hash): Likewise.
9663
9664 * tree-ssa-dom.c (htab_t avail_exprs):
9665 Change type to hash_table. Update dependent calls and types.
9666
9667 * tree-ssa-live.c (var_map_base_init::tree_to_index):
9668 Change type to hash_table. Update dependent calls and types.
9669
9670 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
9671 Change type to hash_table. Update dependent calls and types.
9672
9673 * tree-ssa-phiopt.c (seen_ssa_names):
9674 Change type to hash_table. Update dependent calls and types.
9675
9676 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
9677 Change type to hash_table. Update dependent calls and types.
9678
9679 * tree-ssa-uncprop.c (equiv):
9680 Change type to hash_table. Update dependent calls and types.
9681
ca719585 96822013-04-25 Jakub Jelinek <jakub@redhat.com>
9683
9684 PR rtl-optimization/57003
9685 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
9686 call note_stores with kill_clobbered_value callback again after
9687 killing regs_invalidated_by_call.
9688
0a627f67 96892013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
9690
9691 * config/aarch64/aarch64-simd.md
9692 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
9693 (aarch64_simd_bsl<mode>): Likewise.
9694 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
9695
7695fea9 96962013-04-25 Marek Polacek <polacek@redhat.com>
9697
9698 PR tree-optimization/57066
65c52515 9699 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
7695fea9 9700
535e3387 97012013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
9702
9703 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
9704
f164e9a5 97052013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
9706
9707 * config/aarch64/aarch64-builtins.c
9708 (aarch64_fold_builtin): New.
9709 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
9710 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
9711 * config/aarch64/aarch64-simd-builtins.def (abs): New.
9712 * config/aarch64/arm_neon.h
9713 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
9714
58aab7ce 97152013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
9716 Tejas Belagod <tejas.belagod@arm.com>
9717
9718 * config/aarch64/aarch64-builtins.c
9719 (aarch64_gimple_fold_builtin): New.
9720 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
9721 * config/aarch64/aarch64-simd-builtins.def (addv): New.
9722 * config/aarch64/aarch64-simd.md (addpv4sf): New.
9723 (addvv4sf): Update.
9724 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
9725
c7262ba2 97262013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
9727
baad29fe 9728 * config/aarch64/aarch64.md
c7262ba2 9729 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
9730
993fed27 97312013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
9732
9733 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
9734 (*ngcsi_uxtw): New pattern.
9735
8518cada 97362013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
65c52515 9737 Julian Brown <julian@codesourcery.com>
8518cada 9738
9739 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
9740 (TB_DREG): Add T_V4HF.
9741 (v4hf_UP): New macro.
9742 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
baad29fe 9743 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
8518cada 9744 Handle initialisation of V4HF. Adjust initialisation of reinterpret
9745 built-ins.
baad29fe 9746 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
8518cada 9747 (arm_vector_mode_supported_p): Handle V4HF.
9748 (arm_mangle_map): Handle V4HFmode.
9749 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
9750 * config/arm/arm_neon_builtins.def: Add entries for
9751 vcvtv4hfv4sf, vcvtv4sfv4hf.
9752 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
9753 (neon_vcvtv4hfv4sf): Likewise.
9754 * config/arm/neon-gen.ml: Handle half-precision floating point
9755 features.
9756 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
9757 * config/arm/arm_neon.h: Regenerate.
9758 * config/arm/neon.ml (type elts): Add F16.
9759 (type vectype): Add T_float16x4, T_floatHF.
9760 (type vecmode): Add V4HF.
9761 (type features): Add Requires_FP_bit feature.
9762 (elt_width): Handle F16.
9763 (elt_class): Likewise.
9764 (elt_of_class_width): Likewise.
9765 (mode_of_elt): Refactor.
9766 (type_for_elt): Handle F16, fix error messages.
9767 (vectype_size): Handle T_float16x4.
9768 (vcvt_sh): New function.
9769 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
9770 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
9771 (string_of_mode): Handle V4HF.
9772 * doc/arm-neon-intrinsics.texi: Regenerate.
9773
66a71b3d 97742013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
9775
9776 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
9777 format specifier in 'X' case.
9778
0ee54873 97792013-04-25 Alan Modra <amodra@gmail.com>
9780
9781 PR target/57052
9782 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
9783 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
9784 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
9785 Repeat for many other rotate/shift and mask patterns using subregs.
9786 Name lshiftrt insns.
9787 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
9788 on WORDS_BIG_ENDIAN.
9789
d80dc920 97902013-04-25 Alan Modra <amodra@gmail.com>
9791
9792 * config.gcc: Support little-endian powerpc-linux targets.
9793 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
9794 (LINK_OS_LINUX_SPEC): Define.
9795 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
9796 Preserve MASK_LITTLE_ENDIAN.
9797 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
9798 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
9799 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
9800 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
9801 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
9802 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
9803 Correct fp word order for little-endian. Don't shift toc entries
9804 smaller than a word for little-endian.
9805 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
9806 (bswapdi2 splits): Correct low-part subreg for little-endian.
9807 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
9808 low/high where such is correct only for be.
9809 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
9810 little-endian for -mcall-aixdesc.
9811
d04c855d 98122013-04-25 Alan Modra <amodra@gmail.com>
9813
9814 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
9815 replace_equiv_address_nv.
9816
1efbef15 98172013-04-25 Alan Modra <amodra@gmail.com>
9818
9819 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
9820
50580188 98212013-04-24 Vladimir Makarov <vmakarov@redhat.com>
9822
9823 Revert:
9824 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
9825 * rtl.h (struct rtx_def): ...
1efbef15 9826
74855d08 98272013-04-24 Vladimir Makarov <vmakarov@redhat.com>
9828
9829 PR rtl-optimizations/57046
9830 * lra-constraints (split_reg): Set up lra_risky_transformations_p
9831 for multi-reg splits.
9832
6c4ccb7b 98332013-04-24 H.J. Lu <hongjiu.lu@intel.com>
9834
9835 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
9836
1c74176c 98372013-04-24 Sterling Augustine <saugustine@google.com>
9838
9839 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
9840 (comp_dir_string, debug_str_dwo_section): New.
9841 (DEBUG_STR_DWO_SECTION): Rename to ...
9842 (DEBUG_DWO_STR_SECTION): ... this.
9843 (DEBUG_NORM_STR_SECTION): Delete.
9844 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
9845 (DEBUG_STR_DWO_SECTION_FLAGS): New.
9846 (find_AT_string): Move most logic to ...
9847 (find_AT_string_in_table): ... here. New.
9848 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
9849 add_skeleton_AT_string. Delete logic.
9850 (output_skeleton_debug_sections): Remove call to
9851 add_top_level_skeleton_die_attrs.
9852 (add_comp_dir_attribute): Move logic to comp_dir_string.
9853 (dwarf2out_init): Initialize debug_str_dwo_section.
9854 (output_indirect_string): Call find_string_form.
9855 (output_indirect_strings): Rewrite.
9856 (prune_unused_types): Empty skeleton_debug_str_hash.
9857 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
9858 (dwarf2out_finish): Call output_indirect_strings.
9859
949dbf93 98602013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
9861
9862 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
9863
bac08f26 98642013-04-24 Vladimir Makarov <vmakarov@redhat.com>
9865
1efbef15 9866 * rtl.h (struct rtx_def): Add comment for field jump.
bac08f26 9867 (LRA_SUBREG_P): New macro.
9868 * recog.c (register_operand): Check LRA_SUBREG_P.
9869 * lra.c (lra): Add note at the end of RTL code. Align non-empty
9870 stack frame.
9871 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
9872 (lra_final_code_change): Skip subreg change for operators.
9873 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
9874 if there are no operand changes.
9875 * lra-constraints.c (curr_insn_set): New.
9876 (match_reload): Set LRA_SUBREG_P.
9877 (emit_spill_move): Ditto.
9878 (check_and_process_move): Use curr_insn_set. Process only single
9879 set insns. Don't initialize sec_mem_p and change_p.
9880 (simplify_operand_subreg): Use LRA_SUBREG_P.
9881 (reg_in_class_p): New function.
9882 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
9883 of #ifdef. Add code to remove cycling.
9884 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
9885 non-null disp. Reload inner instead of disp when base and index
9886 are null. Try to put lo_sum into register.
9887 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
1efbef15 9888 (check_and_process_move): Move code for move cost check to
bac08f26 9889 simple_move_p. Remove equiv_substitution.
1efbef15 9890 (simple_move_p): New function.
bac08f26 9891 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
9892 curr_insn_set. Call check_and_process_move only for single set
9893 insns. Use the new function. Move call of check_and_process_move
9894 after operand equiv substitution and address process.
9895
2d62f55c 98962013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
9897
9898 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
9899 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
9900 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
9901
05d0bce1 99022013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
9903
9904 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
9905
0d009ac2 99062013-04-24 Marek Polacek <polacek@redhat.com>
9907
9908 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
9909 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
9910 (select_loops_exit_conditions): Likewise.
9911 (number_of_iterations_for_all_loops): Likewise.
9912 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
9913 (scev_analysis): Likewise.
9914
21e02cfa 99152013-04-02 Catherine Moore <clm@codesourcery.com>
9916 Chao-ying Fu <fu@mips.com>
9917
1efbef15 9918 * config/mips/micromips.md (jraddiusp): New pattern.
9919 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
9920 instruction if possible.
21e02cfa 9921
99646b03 99222013-04-24 Alan Modra <amodra@gmail.com>
9923
9924 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
9925
47b5e9f1 99262013-04-24 Julian Brown <julian@codesourcery.com>
9927 Chung-Lin Tang <cltang@codesourcery.com>
9928
9929 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
baad29fe 9930 dependency behavior in enumeration type DIE generation. Add TODO note
9931 to comments about future DW_FORM_sdata/udata re-work of related code.
47b5e9f1 9932
3e871d4d 99332013-04-23 Lawrence Crowl <crowl@google.com>
9934
9935 * Makefile.in: Update as needed below.
9936
9937 * hash-table.h (class hash_table):
9938 Correct many methods with parameter types compare_type to the correct
9939 value_type. (Correct code was unlikely to notice the change.)
9940 (hash_table::elements_with_deleted) New.
9941 (class hashtable::iterator): New.
9942 (hashtable::begin()): New.
9943 (hashtable::end()): New.
9944 (FOR_EACH_HASH_TABLE_ELEMENT): New.
9945
9946 * statistics.c (statistics_hashes):
9947 Change type to hash_table. Update dependent calls and types.
9948
9949 * tree-into-ssa.c (var_infos):
9950 Change type to hash_table. Update dependent calls and types.
9951
9952 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
9953 Change type to hash_table. Update dependent calls and types.
9954
9955 * tree-ssa-loop-im.c (struct mem_ref.refs):
9956 Change type to hash_table. Update dependent calls and types.
9957
9958 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
9959 Change type to hash_table. Update dependent calls and types.
9960
9961 * tree-ssa-sccvn.c (vn_tables_s::nary):
9962 Change type to hash_table. Update dependent calls and types.
9963 (vn_tables_s::phis): Likewise.
9964 (vn_tables_s::references): Likewise.
9965
9966 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
9967 (vn_reference_eq): Update parameter and return types.
9968
9969 * tree-ssa-structalias.c (pointer_equiv_class_table):
9970 Change type to hash_table. Update dependent calls and types.
9971 (location_equiv_class_table): Likewise.
9972
9973 * tree-vect-data-refs.c: Consequential changes for making
9974 peeling a hash_table.
9975
9976 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
9977 (destroy_loop_vec_info): Dependent hash_table update.
9978
9979 * tree-vectorizer.h (peeling_htab):
9980 Change type to hash_table. Update dependent calls and types.
9981
a1064490 99822013-04-23 Shiva Chen <shiva0217@gmail.com>
9983
1efbef15 9984 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
9985 to check the register content is equal or not.
9986 * lra-constraints.c (match_reload): Use lra_assign_reg_val
9987 to assign register content record.
9988 * lra-eliminations.c (update_reg_eliminate): Use
a1064490 9989 lra_update_reg_val_offset to update register content offset.
1efbef15 9990 * lra-int.h (struct lra_reg): Add offset member.
9991 (lra_reg_val_equal_p): New static inline function.
9992 (lra_update_reg_val_offset): New static inline function.
9993 (lra_assign_reg_val): New static inline function.
9994 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
9995 to assign register content record.
9996 (initialize_lra_reg_info_element): Initial offset to zero.
a1064490 9997
5e1781b6 99982013-04-23 Catherine Moore <clm@codesourcery.com>
9999
10000 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10001 operands. Record compression.
10002
6f2b83c8 100032013-04-23 Xinliang David Li <davidxl@google.com>
10004
10005 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10006
5f0af0d8 100072013-04-23 Richard Biener <rguenther@suse.de>
10008
10009 PR middle-end/57036
10010 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10011 parameter, only add abnormal goto edges from the copied body
10012 if the call could perform abnormal gotos.
10013 (copy_cfg_body): Adjust.
10014
562e5728 100152013-04-23 Sofiane Naci <sofiane.naci@arm.com>
10016
10017 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10018
d3c57e54 100192013-04-23 Andreas Schwab <schwab@linux-m68k.org>
10020
10021 * coretypes.h (gimple_stmt_iterator): Add struct to make
10022 compatible with C.
10023
3a4a5c9c 100242013-04-23 Richard Biener <rguenther@suse.de>
10025
10026 PR tree-optimization/57026
10027 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10028 from SSA names occuring in abnormal PHI nodes.
10029
8af0a4fc 100302013-04-22 Andi Kleen <ak@linux.intel.com>
10031
10032 * lto/lto.c (print_lto_report_1): Fix LTO report names.
10033
f4c8156d 100342013-04-22 Andi Kleen <ak@linux.intel.com>
10035
10036 * lto/lto.c (print_lto_report_1): Declare early.
10037 (read_cgraph_and_symbols): Call print_lto_report_1 early.
10038
b63a7a3e 100392013-04-22 Andi Kleen <ak@linux.intel.com>
10040
10041 * common.opt (-flto-report-wpa): Add.
10042 * doc/invoke.texi (-flto-report-wpa): Add.
10043 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10044 (lto_main): dito.
10045
5d24e2d1 100462013-04-22 Xinliang David Li <davidxl@google.com>
10047
10048 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10049 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10050 * Makefile.in: New dependency
10051
1efbef15 10052 David Daney <ddaney.cavm@gmail.com>
bcba791d 10053
1efbef15 10054 * configure.ac (gcc_cv_as_micromips_support): Use the
10055 --fatal-warnings option.
10056 * configure: Regenerate.
bcba791d 10057
888b31c9 100582013-04-22 Marek Polacek <polacek@redhat.com>
10059
10060 PR sanitizer/56990
10061 * tsan.c (instrument_expr): Don't instrument expression
10062 in case its size is zero.
10063
31ba2027 100642013-04-22 Uros Bizjak <ubizjak@gmail.com>
10065
10066 PR target/57032
10067 Revert:
10068 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
10069
10070 * config/alpha/alpha.c (TARGET_LRA_P): New define.
10071
29cad3b6 100722013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10073
10074 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10075 (gimple_stmt_iterator): New typedef.
10076 * gimple.h (gimple_stmt_iterator): Rename to...
10077 (gimple_stmt_iterator_d): ... This.
10078 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10079 trees be valid for GIMPLE and GENERIC.
10080 (TARGET_GIMPLE_FOLD_BUILTIN): New.
10081 * gimple-fold.c (gimple_fold_call): Call target hook
10082 gimple_fold_builtin.
10083 * hooks.c (hook_bool_gsiptr_false): New.
10084 * hooks.h (hook_bool_gsiptr_false): New.
10085 * target.def (fold_stmt): New.
10086 * doc/tm.texi: Regenerate.
10087
fbb45ff1 100882013-04-22 Vladimir Makarov <vmakarov@redhat.com>
10089
10090 PR target/57018
10091 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10092 a set sp if no stack realignment.
10093
ea4d38ee 100942013-04-22 Nick Clifton <nickc@redhat.com>
10095
10096 * config.gcc (tilegx-linux): Extend extra_objs rather than
10097 overwriting it.
10098 (tilepro-linux): Likewise.
10099
5a8d54a9 101002013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10101
10102 * config/aarch64/aarch64-builtins.c
10103 (CF): Remove.
10104 (CF0, CF1, CF2, CF3, CF4, CF10): New.
10105 (VAR<1-12>): Add MAP parameter.
10106 (BUILTIN_*): Likewise.
10107 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10108 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10109 (aarch64_ushl_n<mode>): Likewise.
10110 (aarch64_sshr_n<mode>): Likewise.
10111 (aarch64_ushr_n<mode>): Likewise.
10112 (aarch64_<maxmin><mode>): Likewise.
10113 (aarch64_sqrt<mode>): Likewise.
10114 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10115 (vshr<q>_n_*): Likewise.
10116
7b990298 101172013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10118
10119 * config/aarch64/aarch64-builtins.c
10120 (aarch64_simd_builtin_type_mode): Handle SF types.
10121 (sf_UP): Define.
10122 (BUILTIN_GPF): Define.
10123 (aarch64_init_simd_builtins): Handle SF types.
10124 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10125 (frecps): Likewise.
10126 (frecpx): Likewise.
10127 * config/aarch64/aarch64-simd.md
10128 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10129 (aarch64_frecpe<mode>): New.
10130 (aarch64_frecps<mode>): Likewise.
10131 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10132 (v8type): Add frecp<esx>.
10133 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10134 (aarch64_frecps<mode>): Likewise.
10135 * config/aarch64/iterators.md (FRECP): New.
10136 (frecp_suffix): Likewise.
10137 * config/aarch64/arm_neon.h
10138 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10139
c7a79d83 101402013-04-22 Christian Bruel <christian.bruel@st.com>
ec3c739c 10141
10142 PR target/56995
10143 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10144 (REG_CLASS_NAMES): Idem.
10145 (REG_CLASS_CONTENTS): Idem.
10146 (REGCLASS_HAS_FP_REG): Idem.
10147 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10148 (sh_conditional_register_usage): Idem.
10149
d23e1965 101502013-04-21 Jeff Law <law@redhat.com>
10151
10152 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10153 (ssa_forward_propagate_and_combine): Use it.
10154
b28ae2d4 101552013-04-19 Vladimir Makarov <vmakarov@redhat.com>
10156
10157 * lra.c: Update the flow chart diagram.
10158
a89f8bac 101592013-04-19 Vladimir Makarov <vmakarov@redhat.com>
10160
10161 PR rtl-optimization/56847
10162 * lra-constraints.c (process_alt_operands): Discourage alternative
10163 with non-matche doffsettable memory constraint fro memory with
10164 known offset.
10165
a5ef9e4d 101662013-04-19 Richard Biener <rguenther@suse.de>
10167
10168 PR tree-optimization/56982
10169 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10170 function.
10171 * gimplify.c (gimplify_call_expr): Notice special calls.
10172 (gimplify_modify_expr): Likewise.
10173 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10174 abnormal control flow receivers.
10175 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10176 in the same way as cfun->has_nonlocal_labels.
10177 (gimple_purge_dead_abnormal_call_edges): Likewise.
10178 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10179 receivers start a basic-block.
10180
678e3d6e 101812013-04-19 Richard Biener <rguenther@suse.de>
10182
10183 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10184 member ...
10185 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
10186 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10187 (SLP_TREE_LOAD_PERMUTATION): Add.
10188 (vect_transform_slp_perm_load): Adjust prototype.
10189 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10190 (vect_free_slp_instance): Likewise.
10191 (vect_create_new_slp_node): Likewise.
10192 (vect_supported_slp_permutation_p): Remove.
10193 (vect_slp_rearrange_stmts): Adjust.
10194 (vect_supported_load_permutation_p): Likewise. Inline
10195 vect_supported_slp_permutation_p here.
10196 (vect_analyze_slp_instance): Compute load permutations per
10197 slp node instead of per instance.
10198 (vect_get_slp_defs): Adjust.
10199 (vect_transform_slp_perm_load): Likewise.
10200 (vect_schedule_slp_instance): Remove redundant code.
10201 (vect_schedule_slp): Remove hack for PR56270, add it ...
10202 * tree-vect-stmts.c (vectorizable_load): ... here, do not
10203 CSE loads for SLP. Adjust.
10204
c210e3da 102052013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
10206
10207 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10208 spelling in two comments.
10209
a5c5144f 102102013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
10211
10212 PR target/56797
10213 * config/arm/arm.c (load_multiple_sequence): Require SP
10214 as base register for loads if SP is in the register list.
10215
bee52153 102162013-04-19 Martin Jambor <mjambor@suse.cz>
10217
10218 PR tree-optimization/56718
10219 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
10220 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
10221 and made public. Adjusted all callers.
10222 (ipa_intraprocedural_devirtualization): New function.
10223 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
10224 (ipa_intraprocedural_devirtualization): Likewise.
10225 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
10226
204c295f 102272013-04-19 Richard Biener <rguenther@suse.de>
10228
10229 PR tree-optimization/57000
10230 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
10231
3153fc2e 102322013-04-19 Terry Guo <terry.guo@arm.com>
10233
10234 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
10235 Replace with ...
10236 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
10237 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
10238 (cortex_m4_fmacs): Use new reservations.
10239 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
10240
638e746e 102412013-04-18 Vladimir Makarov <vmakarov@redhat.com>
10242
c0d5925a 10243 PR rtl-optimization/56999
638e746e 10244 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
10245 related code.
10246 (lra_coalesce): Remove split_origin_bitmap and related code.
10247 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
10248 ranges if necessary.
10249
17e854e6 102502013-04-18 Uros Bizjak <ubizjak@gmail.com>
10251
10252 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
10253 New array.
10254 (ix86_expand_call): Remove clobbered_registers array and use
10255 x86_64_ms_sysv_extra_clobbered_registers instead.
10256 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
10257 Declare here.
10258 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
10259 predicate.
10260 * config/i386/i386.md (*call_rex64_ms_sysv): Use
10261 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
10262 (*call_value_rex64_ms_sysv): Ditto.
10263
faeb0946 102642013-04-18 Cary Coutant <ccoutant@google.com>
10265
10266 * dwarf2out.c (output_pubnames): Check die_perennial_p of
10267 parent instead of die_mark.
10268
32bd7708 102692013-04-18 Diego Novillo <dnovillo@google.com>
10270
10271 * gimple.c (create_gimple_tmp): New.
10272 (get_expr_type): New.
10273 (build_assign): New.
10274 (build_type_cast): New.
10275 * gimple.h (enum ssa_mode): Define.
10276 (gimple_seq_set_location): New.
10277 * asan.c (build_check_stmt): Change some gimple_build_* calls
17e854e6 10278 to use build_assign and build_type_cast.
32bd7708 10279
8bbe6b75 102802013-04-18 Richard Biener <rguenther@suse.de>
10281
10282 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
10283 handle negative step. Remove redundant checks.
10284 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
10285 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
10286 for negative step and grouped loads fail to vectorize.
10287
1e9af25c 102882013-04-18 Steven Bosscher <steven@gcc.gnu.org>
10289
10290 * emit-rtl.c (reset_insn_used_flags): New function.
10291 (reset_all_used_flags): Use it.
10292 (verify_insn_sharing): New function.
10293 (verify_rtl_sharing): Fix verification for SEQUENCEs.
10294
7c4be923 102952013-04-18 Jakub Jelinek <jakub@redhat.com>
10296
10297 PR tree-optimization/56984
10298 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
10299 and (x >> M) >= N don't register any assertion if N << M is the
10300 minimum value.
10301
c7ee1482 103022013-04-18 Steven Bosscher <steven@gcc.gnu.org>
10303
10304 * lower-subreg.c (resolve_simple_move): If called self-recursive,
10305 do not delete_insn insns that have not yet been emitted, only
10306 unlink them with remove_insn.
10307 * df-scan.c (df_insn_delete): Revert r197492.
10308
34f5b9ac 103092013-04-17 Steven Bosscher <steven@gcc.gnu.org>
10310
10311 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
10312 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
10313
103142013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
190efb17 10315
10316 * config/arm/arm.md (movsicc_insn): Convert define_insn into
10317 define_insn_and_split.
10318 (and_scc,ior_scc,negscc): Likewise.
10319 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
10320
34f5b9ac 103212013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
35c10aa8 10322
10323 * config/arm/arm.c (use_return_insn): Return 0 for targets that
10324 can benefit from using a sequence of LDRD instructions in epilogue
10325 instead of a single LDM instruction.
10326
d21e37af 103272013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
10328
10329 PR 45688
10330 * doc/extend.texi: Fix typo.
10331
16513aa4 103322013-04-17 Richard Biener <rguenther@suse.de>
10333
10334 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
10335 (vect_build_slp_tree): ... here.
10336 (vect_build_slp_tree_1): Compute which stmts of the SLP group
10337 match. Remove special-casing of mismatched complex loads.
10338 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
10339 re-try the match with swapped commutative operands.
10340 (vect_supported_load_permutation_p): Remove special-casing of
10341 mismatched complex loads.
10342 (vect_analyze_slp_instance): Adjust.
10343
427286b3 103442013-04-17 Richard Biener <rguenther@suse.de>
10345
10346 PR rtl-optimization/56921
10347 * cfgloop.h (struct loop): Add simple_loop_desc member.
10348 (struct niter_desc): Mark with GTY(()).
10349 (simple_loop_desc): Do not use aux field but simple_loop_desc.
10350 * loop-iv.c (get_simple_loop_desc): Likewise.
10351 (free_simple_loop_desc): Likewise.
10352
10353 Revert
10354 2013-04-16 Richard Biener <rguenther@suse.de>
10355
10356 PR rtl-optimization/56921
10357 * loop-init.c (pass_rtl_move_loop_invariants): Add
10358 TODO_do_not_ggc_collect to todo_flags_finish.
10359 (pass_rtl_unswitch): Same.
10360 (pass_rtl_unroll_and_peel_loops): Same.
10361 (pass_rtl_doloop): Same.
10362
740a2cb6 103632013-04-17 Eric Botcazou <ebotcazou@adacore.com>
10364
10365 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
10366 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
10367 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
10368 references.
10369 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
10370 * tree-streamer.c (record_common_node): Adjust reference in comment.
10371
c20f53ce 103722013-04-17 Terry Guo <terry.guo@arm.com>
10373
10374 * config/arm/cortex-m4.md: Add a new bypass.
10375
31ba2027 103762013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
7425af27 10377
10378 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
10379 New pattern.
10380 (*subs_<optab><mode>_multp2): New pattern.
10381 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10382 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10383
31ba2027 103842013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
16b547de 10385
10386 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
10387 (*subs_mul_imm_<mode>): New pattern.
10388
bd59dd08 103892013-04-16 David Edelsohn <dje.gcc@gmail.com>
10390
10391 PR target/56948
10392 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
10393 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
10394 (vsx_movti_32bit): Change j->wa to O->wa.
10395
507c20f1 103962013-04-16 Richard Biener <rguenther@suse.de>
10397
10398 PR rtl-optimization/56921
10399 * loop-init.c (pass_rtl_move_loop_invariants): Add
10400 TODO_do_not_ggc_collect to todo_flags_finish.
10401 (pass_rtl_unswitch): Same.
10402 (pass_rtl_unroll_and_peel_loops): Same.
10403 (pass_rtl_doloop): Same.
10404
1e9af25c 104052013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
9e0accd9 10406
10407 * config/arm/arm.c (emit_multi_reg_push): New declaration
10408 for an existing function.
10409 (arm_emit_strd_push): New function.
10410 (arm_expand_prologue): Used here.
10411 (arm_emit_ldrd_pop): New function.
10412 (arm_expand_epilogue): Used here.
10413 (arm_get_frame_offsets): Update condition.
10414 (arm_emit_multi_reg_pop): Add a special case for load of a single
10415 register with writeback.
10416
69ac6d73 104172013-04-16 Uros Bizjak <ubizjak@gmail.com>
10418
10419 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
10420 description.
10421
bc9af2dd 104222013-04-16 Richard Biener <rguenther@suse.de>
10423
10424 PR tree-optimization/56756
10425 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
10426 (first_mem_ref_loc): New.
10427 (execute_sm): Place the load temporarily before a previous
10428 access instead of in the latch edge to ensure its SSA dependencies
10429 are defined at points dominating the load.
10430
35f3420b 104312013-04-16 Steven Bosscher <steven@gcc.gnu.org>
10432
eda57d74 10433 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
10434 correct fix by moving header and footer insn to the footer of
10435 the merged basic block. Clear BB_END of the merged-away block.
10436
35f3420b 10437 PR middle-end/43631
10438 * emit-rtl.c (make_note_raw): New function.
10439 (link_insn_into_chain): New static inline function.
10440 (add_insn): Use it.
10441 (add_insn_before, add_insn_after): Factor insn chain linking code...
10442 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
10443 using link_insn_into_chain.
10444 (note_outside_basic_block_p): New helper function for emit_note_after
10445 and emit_note_before.
10446 (emit_note_after): Use nobb variant of add_insn_after if the note
10447 should not be contained in a basic block.
10448 (emit_note_before): Use nobb variant of add_insn_before if the note
10449 should not be contained in a basic block.
10450 (emit_note_copy): Use make_note_raw.
10451 (emit_note): Likewise.
10452 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
10453 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
10454 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
10455 the moved barrier the tail of the basic block it follows.
10456 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
10457
9de38bc3 104582013-04-15 Jakub Jelinek <jakub@redhat.com>
10459
10460 PR tree-optimization/56962
10461 * gimple-ssa-strength-reduction.c (record_increment): Only set
10462 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
10463 either rhs1 or rhs2 is equal to c->base_expr.
10464
4d525783 104652013-04-15 Richard Biener <rguenther@suse.de>
10466
10467 PR tree-optimization/56933
10468 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
10469 member.
10470 (GROUP_READ_WRITE_DEPENDENCE): Remove.
10471 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
10472 * tree-vect-data-refs.c (vect_analyze_group_access): Move
10473 dependence check ...
10474 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
10475 ... here.
10476 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
10477 GROUP_READ_WRITE_DEPENDENCE.
10478
7cdd84a2 104792013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
10480
10481 * emit-rtl.c (reset_all_used_flags): New function.
10482 (verify_rtl_sharing): Call reset_all_used_flags before and after
10483 performing the checks.
10484
f6bbdcf6 104852013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10486
10487 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
10488 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
10489 * config/arm/constraints.md (De): New constraint.
10490 * config/arm/neon.md (anddi3_neon): Delete.
10491 (neon_vand<mode>): Expand to standard anddi3 pattern.
10492 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
10493 Move earlier in the file.
10494 (neon_inv_logic_op2): Likewise.
10495 (arm_anddi_operand_neon): New predicate.
10496
8aaed91d 104972013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10498
10499 * configure.ac (gcc_cv_ld_as_needed): Set
10500 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
10501 Use -z ignore, -z record on *-*-solaris2*.
10502 (HAVE_LD_AS_NEEDED): Update comment.
10503 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
10504 * configure: Regenerate.
10505 * config.in: Regenerate.
10506 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
10507 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
10508 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
10509 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
10510 equivalents. Fix markup.
10511 * doc/tm.texi: Regenerate.
10512
98c0d657 105132013-04-15 Andrew Hsieh <andrewhsieh.google.com>
10514
10515 * config/i386/i386.opt: New option mstack-protector-guard=.
10516 * config/i386/i386-opts.h: Add enum stack_protector_guard.
10517 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
10518 TARGET_SSP_TLS_GUARD.
10519 * config/i386/i386.c (ix86_option_override_internal): Set
10520 ix86_stack_protector_guard.
10521 * config/i386/i386.md (stack_protect_set): Enable for
10522 TARGET_SSP_TLS_GUARD only.
10523 (stack_protect_set_<mode>): Ditto.
10524 (stack_protect_test): Ditto.
10525 (stack_protect_test_<mode>): Ditto.
10526 * doc/invoke.texi (i386 Option): Document.
10527
3eb1f083 105282013-04-15 Eric Botcazou <ebotcazou@adacore.com>
10529
10530 PR target/56890
10531 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
10532 (S_MODES): Set H_MODE bit.
10533 (SF_MODES): Set only S_MODE and SF_MODE bits.
10534 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
10535 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
10536 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
10537 <MODE_FLOAT>: Likewise.
10538
756b7d6a 105392013-04-15 Joey Ye <joey.ye@arm.com>
10540
10541 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
10542
8d04d59e 105432013-04-15 Joey Ye <joey.ye@arm.com>
10544
10545 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
10546 for real far jump.
10547 (thumb_far_jump_used_p): Count instruction size and set
10548 far_jump_used.
10549
fdb35bef 105502013-04-14 Eric Botcazou <ebotcazou@adacore.com>
10551
10552 * reorg.c (fill_simple_delay_slots): Reindent block of code.
10553 * resource.c (mark_target_live_regs): Reformat conditional block.
10554
8ea0810a 105552013-04-13 Steven Bosscher <steven@gcc.gnu.org>
10556
10557 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
10558 notes, they are emitted only just before final.
10559 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
10560
93ff53d3 105612013-04-13 Steven Bosscher <steven@gcc.gnu.org>
10562
10563 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
10564 * cfgrtl.c (delete_insn): Call it here instead.
10565 * lra-spills.c (lra_final_code_change): Use delete_insn.
10566 * haifa-sched.c (sched_remove_insn): Likewise.
10567 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
10568 returning to the nop pool.
10569 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
10570 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
10571
720b223d 105722013-04-12 Steven Bosscher <steven@gcc.gnu.org>
10573
10574 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
10575 * doc/tm.texi: Regenerated.
10576
8f028904 105772013-04-12 Uros Bizjak <ubizjak@gmail.com>
10578
10579 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
10580 QImode checks.
10581
09f44684 105822013-04-12 Steven Bosscher <steven@gcc.gnu.org>
10583
10584 * df-core.c (df_find_def): Compare register numbers.
10585 (df_find_use): Likewise.
10586
45237259 105872013-04-12 Vladimir Makarov <vmakarov@redhat.com>
10588
10589 PR target/56903
10590 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
10591 lra_in_progress for return.
10592
a3b84066 105932013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
10594
10595 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
10596 define_insn into define_insn_and_split and emit movsicc patterns.
10597
b3e65331 105982013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
10599
10600 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
10601
533c15bc 106022013-04-12 Richard Biener <rguenther@suse.de>
10603
10604 * tree-pass.h (TODO_do_not_ggc_collect): New.
10605 * passes.c (execute_one_ipa_transform_pass): Honor
10606 TODO_do_not_ggc_collect.
10607 (execute_one_pass): Likewise.
10608
10609 Revert
10610 2013-04-10 Richard Biener <rguenther@suse.de>
10611
10612 * passes.c (init_optimization_passes): Remove reload pass.
10613 * ira.c (do_reload): Merge into ...
10614 (ira): ... this.
10615 (rest_of_handle_reload): Remove.
10616 (pass_reload): Likewise.
10617 * config/i386/i386.c (ix86_option_override): Refer to ira instead
10618 of reload for vzeroupper pass placement.
10619
577e2925 106202013-04-12 Jakub Jelinek <jakub@redhat.com>
10621
10622 PR tree-optimization/56918
10623 PR tree-optimization/56920
10624 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
10625 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
10626 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
10627 use wide_mul_with_sign method.
10628
5b787773 106292013-04-12 Richard Biener <rguenther@suse.de>
10630
10631 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
10632 not be considered a gimple constant.
10633
ef4a904b 106342013-04-12 Marc Glisse <marc.glisse@inria.fr>
10635
10636 * fold-const.c (const_binop): Handle vector shifts by a scalar.
10637 (fold_binary_loc): Call const_binop also for mixed vector-scalar
10638 operations.
10639
41609f8b 106402013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
10641 Jakub Jelinek <jakub@redhat.com>
10642
10643 * opts.c: Include diagnostic-color.h.
10644 (common_handle_option): Handle OPT_fdiagnostics_color_.
10645 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
10646 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
10647 (diagnostic-color.o): New.
10648 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
10649 (diagnostic_color_rule): New enum.
10650 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
10651 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
10652 the location string.
10653 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
10654 either NULL, or color kind.
10655 * diagnostic-color.c: New file.
10656 * diagnostic-color.h: New file.
10657 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
10658 arguments.
10659 * doc/invoke.texi (-fdiagnostics-color): Document.
10660 * pretty-print.h (pp_show_color): Define.
10661 (struct pretty_print_info): Add show_color field.
10662 * diagnostic.c: Include diagnostic-color.h.
10663 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
10664 macros. Colorize error:, warning: etc. strings and also the location
10665 string.
10666 (diagnostic_show_locus): Colorize the caret line.
10667 * pretty-print.c: Include diagnostic-color.h.
10668 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
10669 inside of %< %> quotes or quoted through q format modifier.
10670
79f85ed5 106712013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
10672
8f028904 10673 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
79f85ed5 10674
8f028904 106752013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
62bad789 10676
10677 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
10678 code in CC_NZ mode.
10679 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
10680 pattern.
10681
abf3fde4 106822013-04-11 Marek Polacek <polacek@redhat.com>
10683
10684 PR tree-optimization/48184
8f028904 10685 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
abf3fde4 10686
3303f1bb 106872013-04-11 Eric Botcazou <ebotcazou@adacore.com>
10688
10689 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
10690 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
10691 (skip_simple_arithmetic): Tidy up.
10692 * tree.h (skip_simple_constant_arithmetic): Declare.
10693
8f028904 106942013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
d449a7bb 10695
10696 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
10697
6c2c88c7 106982013-04-11 Richard Biener <rguenther@suse.de>
10699
10700 * tree-vect-loop.c (get_initial_def_for_induction): Properly
10701 generate vector constants.
10702
6046367e 107032013-04-11 Richard Biener <rguenther@suse.de>
10704
10705 PR tree-optimization/56878
10706 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
10707 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
10708 New function.
10709 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10710 Prefer to align the DR with the most invariant base address.
10711
6c1f6b49 107122013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
10713
10714 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
10715 comment.
10716
3b11c427 107172013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
10718
10719 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
10720 floating-point vector comparisons against 0.
10721
4debb326 107222013-04-11 Jakub Jelinek <jakub@redhat.com>
10723
10724 PR tree-optimization/56899
10725 * fold-const.c (extract_muldiv_1): Apply distributive law
10726 only if TYPE_OVERFLOW_WRAPS (ctype).
10727
a0ca7f3d 107282013-04-11 Bin Cheng <bin.cheng@arm.com>
10729
10730 PR target/56124
10731 * ira-costs.c (scan_one_insn): Check whether the source rtx of
10732 loading has side effect.
10733
a1adea34 107342013-04-10 Steven Bosscher <steven@gcc.gnu.org>
10735
10736 * config/sparc/sparc.c: Include tree-pass.h.
10737 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
10738 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
10739 head of file. Change return type. Split off gate function.
10740 (sparc_gate_work_around_errata): New function.
10741 (pass_work_around_errata): New pass definition.
10742 (insert_pass_work_around_errata) New pass insert definition to
10743 insert pass_work_around_errata just after delayed-branch scheduling.
10744 (sparc_option_override): Insert the pass.
10745 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
10746
17f446a0 107472013-04-10 David S. Miller <davem@davemloft.net>
10748
c18356f5 10749 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
10750 or -mcpu=hypersparc.
10751
17f446a0 10752 * target.def (cstore_mode): New hook.
10753 * target.h: Include insn-codes.h
10754 * targhooks.c: Likewise.
10755 (default_cstore_mode): New function.
10756 * targhooks.h: Declare it.
10757 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
10758 * doc/tm.texi: Rebuild.
10759 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
10760 target hook, rather than inspecting the insn_data.
10761 * config/sparc/sparc.c (sparc_cstore_mode): New function.
10762 (TARGET_CSTORE_MODE): Redefine.
10763 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
10764 result patterns.
10765 * config/sparc/predicates.md (cstore_result_operand): New special
10766 predicate.
10767 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
10768 Use it for operand 0.
10769 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
10770 (*snesi_special): Likewise.
10771 (*snesi_zero): Likewise.
10772 (*seqsi_zero): Likewise.
10773 (*sltu_insn): Likewise.
10774 (*sgeu_insn): Likewise.
10775 (*seqdi_special): Make operand 0 and comparison operation be of
10776 DImode.
10777 (*snedi_special): Likewise.
10778 (*snedi_special_vis3): Likewise.
10779 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
10780 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
10781 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
10782 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
10783 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
10784 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
10785 (*sltu_extend_sp64): Likewise.
10786 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
10787 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
10788 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
10789 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
10790 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
10791
69b8a0da 107922013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
10793
10794 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
10795 (aarch64_start_file): Use the new function.
10796
8786b72c 107972013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
8f028904 10798 Jason Merrill <jason@redhat.com>
8786b72c 10799
10800 * common.opt: Add -gdwarf.
10801 * opts.c (common_handle_option): Handle it.
10802 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
10803
ef3baff5 108042013-04-10 Richard Biener <rguenther@suse.de>
10805
10806 * passes.c (execute_todo): Do not call ggc_collect conditional here.
10807 (execute_one_ipa_transform_pass): But unconditionally here.
10808 (execute_one_pass): And here.
10809 (init_optimization_passes): Remove reload pass.
10810 * tree-pass.h (TODO_ggc_collect): Remove.
10811 (pass_reload): Likewise.
10812 * ira.c (do_reload): Merge into ...
10813 (ira): ... this.
10814 (rest_of_handle_reload): Remove.
10815 (pass_reload): Likewise.
10816 * config/i386/i386.c (ix86_option_override): Refer to ira instead
10817 of reload for vzeroupper pass placement.
10818 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
10819 and todo_flags_finish of all passes.
10820
3533e58d 108212013-04-10 Richard Biener <rguenther@suse.de>
10822
10823 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
10824 first_const_oprnd field, rename first_def_type to first_op_type.
10825 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
10826 (vect_get_and_check_slp_defs): Always use the type of the
10827 operand. Allow mixed vect_external_def, vect_constant_def types.
10828 (vect_get_constant_vectors): Handle mixed vect_external_def,
10829 vect_constant_def types.
10830
b095bd6a 108312013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
10832
10833 PR tree-optimization/55524
10834 * tree-ssa-math-opts.c
10835 (convert_mult_to_fma): Don't use an fms construct
10836 when we don't have an fms operation, but fnma, and it looks
10837 likely that we'll be able to use the latter.
10838
479b4ace 108392013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
10840
10841 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
10842 function.
10843 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
10844 inline fail caused by overwritable functions.
10845
241f5399 108462013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
10847
10848 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
10849 unnecessary bits in the constant power of two case.
10850
0eba4748 108512013-04-10 Richard Biener <rguenther@suse.de>
10852
10853 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
10854 broken code swapping operands.
10855 (vect_build_slp_tree): Do not compute load permutations here.
10856 (vect_analyze_slp_instance): Compute load permutations here,
10857 after building the SLP tree.
10858
60ac2a59 108592013-04-09 Christian Bruel <christian.bruel@st.com>
10860
10861 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
10862 of next/prev_real_insn.
10863
7dfe7ce5 108642013-04-09 Jan Hubicka <jh@suse.cz>
10865
0eba4748 10866 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
10867 Drop aliased parameter.
7dfe7ce5 10868 (function_and_variable_visibility): Do not handle alias pairs.
10869 * cgraph.c (varpool_externally_visible_p): Update prototype.
10870 * varpool.c (varpool_add_new_variable): Update.
10871
4164bca1 108722013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10873
10874 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
10875
6ef7d725 108762013-04-09 Steven Bosscher <steven@gcc.gnu.org>
10877
adb4cb31 10878 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
10879
6ef7d725 10880 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
10881
efbfe80e 108822013-04-09 Marek Polacek <polacek@redhat.com>
10883
10884 PR tree-optimization/48762
8f028904 10885 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
efbfe80e 10886
ee6e9ba5 108872013-04-09 Richard Biener <rguenther@suse.de>
10888
10889 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
10890 dealing with cost.
10891 (vect_build_slp_tree): Likewise.
10892 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
10893 calculating the cost of a SLP instance.
10894 (vect_analyze_slp_instance): Use it from here, after building
10895 the SLP tree.
10896
4abecb72 108972013-04-09 Jakub Jelinek <jakub@redhat.com>
10898
10899 PR middle-end/56883
10900 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
10901 expand_omp_for_static_chunk): Use simple_p = true in
10902 force_gimple_operand_gsi calls when assigning to addressable decls.
10903
813adf93 109042013-04-09 Jeff Law <law@redhat.com>
10905
10906 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
10907 when the boolean was created by converting a wider object which
10908 had a boolean range.
10909
40bcc7c2 109102013-04-09 Richard Biener <rguenther@suse.de>
10911
10912 * tree-vectorizer.h (slp_void_p): Remove.
10913 (slp_tree): Typedef before _slp_tree declaration.
10914 (struct _slp_tree): Use a vector of slp_tree as children.
10915 (vect_get_place_in_interleaving_chain): Remove.
10916 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
10917 Move ...
10918 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
10919 and make static.
10920 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
10921 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
10922 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
10923 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
10924 Use slp_node instead of slp_void_p and adjust.
10925
c8498e04 109262013-04-09 Richard Biener <rguenther@suse.de>
10927
10928 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
10929 work that is not necessary.
10930
f6e2e4ff 109312013-04-09 Jakub Jelinek <jakub@redhat.com>
10932
10933 PR tree-optimization/56854
10934 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
10935 forward into clobber stmts if it would change MEM_REF lhs into
10936 non-MEM_REF.
10937
45e5c263 109382013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
10939
10940 * tree.c (type_hash_lookup, type_hash_add): Make static.
10941 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
10942
26a87c78 109432013-04-09 Richard Biener <rguenther@suse.de>
10944
10945 * tree.h (unsave_expr_now): Remove.
10946 * tree-inline.c (mark_local_for_remap_r): Remove.
10947 (unsave_expr_1): Likewise.
10948 (unsave_r): Likewise.
10949 (unsave_expr_now): Likewise.
10950 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
10951 (propagate_tree_value): Likewise.
10952
3ff7f7e7 109532013-04-08 Steven Bosscher <steven@gcc.gnu.org>
10954
10955 * doc/rtl.texi (sequence): Rewrite documentation to match the
10956 current use of SEQUENCE rtl objects.
10957 * rtl.def (SEQUENCE): Likewise.
10958
10959 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
10960 Update documentation.
10961 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
10962 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
10963
10964 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
10965
f9d4b7f4 109662013-04-08 Teresa Johnson <tejohnson@google.com>
10967
10968 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
10969 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
8f028904 10970 methods.
f9d4b7f4 10971 (estimate_edge_size_and_time): Add comment to suggest using rounding
10972 methods.
10973 (estimate_node_size_and_time): Ditto.
10974 (remap_edge_change_prob): Use helper rounding divide methods.
10975 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
10976 (gimple_mod_pow2_value_transform): Ditto.
10977 (gimple_mod_subtract_transform): Ditto.
10978 (gimple_ic_transform): Ditto.
10979 (gimple_stringops_transform): Ditto.
10980 * stmt.c (conditional_probability): Ditto.
10981 (emit_case_dispatch_table): Ditto.
10982 * lto-cgraph.c (merge_profile_summaries): Ditto.
10983 * tree-optimize.c (execute_fixup_cfg): Ditto.
10984 * cfgcleanup.c (try_forward_edges): Ditto.
10985 * cfgloopmanip.c (scale_loop_profile): Ditto.
10986 (loopify): Ditto.
10987 (duplicate_loop_to_header_edge): Ditto.
10988 (lv_adjust_loop_entry_edge): Ditto.
10989 * tree-vect-loop.c (vect_transform_loop): Ditto.
10990 * profile.c (compute_branch_probabilities): Ditto.
10991 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
10992 * lto-streamer-in.c (input_cfg): Ditto.
10993 * gimple-streamer-in.c (input_bb): Ditto.
10994 * ipa-cp.c (update_profiling_info): Ditto.
10995 (update_specialized_profile): Ditto.
10996 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
10997 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
8f028904 10998 rounding methods.
f9d4b7f4 10999 * sched-rgn.c (compute_dom_prob_ps): Ditto.
11000 (compute_trg_info): Ditto.
11001 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11002 (purge_dead_edges): Ditto.
11003 * loop-unswitch.c (unswitch_loop): Ditto.
11004 * cgraphclones.c (cgraph_clone_edge): Ditto.
11005 (cgraph_clone_node): Ditto.
11006 * tree-inline.c (copy_bb): Ditto.
11007 (copy_edges_for_bb): Ditto.
11008 (initialize_cfun): Ditto.
11009 (copy_cfg_body): Ditto.
11010 (expand_call_inline): Ditto.
11011
5a405481 110122013-04-08 Kai Tietz <ktietz@redhat.com>
11013
11014 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11015 TARGET_CYGWIN64 by TARGET_64BIT.
11016
88b2f163 110172013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
11018
11019 * config/epiphany/epiphany.md (GPR_1): New constant.
11020 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11021 * config/epiphany/epiphany.c (gen_compare_reg):
11022 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11023 is already in place.
11024 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11025 Don't require being called during rtl expansion; If y operlaps r0,
11026 return 0.
11027 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11028 (epiphany_expand_epilogue): Likewise.
11029
3d5b5f5d 11030 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11031 Don't use CC_FPmode for ORDERED / UNORDERED.
8f028904 11032 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
3d5b5f5d 11033
9864d85d 11034 * config/epiphany/constraints.md (CnL): New constraint.
11035 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11036 * config/epiphany/predicates.md (add_operand): Allow 1024.
11037
309ff476 11038 * config/epiphany/epiphany.md (logical_op): New code iterator.
11039 (op_mnc): New code attribute.
11040 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11041 (mov_f+1, mov_f+2): New peephole2 patterns.
11042
b434a9d7 11043 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11044 (cstoresi4): Also allow re-use of zero result when doing a NE
11045 comparison to a non-zero operand.
2467b384 11046 Use (clobber (scratch)) for first insn if the gpr output is not needed.
b434a9d7 11047
9d8aebe9 11048 * config/epiphany/epiphany.md (<insn_opname>v2si3):
11049 Use gen_addsi3_i / gen_subsi3_i.
11050
896a0c42 110512013-04-08 Jakub Jelinek <jakub@redhat.com>
11052
11053 PR c++/34949
11054 PR c++/50243
11055 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11056 contain anything but clobbers, at most one __builtin_stack_restore,
11057 optionally debug stmts and final resx, and if it has at least one
11058 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
11059 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11060 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11061 which isn't defaut definition, remove them.
11062 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11063 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11064 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11065 with MEM_REF LHS with SSA_NAME address.
11066
d701b871 110672013-04-08 Jeff Law <law@redhat.com>
11068
11069 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
813adf93 11070
0bf5f81b 110712013-04-08 Richard Biener <rguenther@suse.de>
11072
11073 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11074 extra newline.
11075 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11076 determined vector type.
11077 (vect_analyze_data_refs): Likewise.
11078 (vect_get_new_vect_var): Adjust.
11079 (vect_create_destination_var): Preserve SSA name versions.
11080 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11081 not dump anything here.
11082
78d11f8a 110832013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
11084
11085 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11086 Add member lr_slot_known.
11087 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11088 if necessary.
11089 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11090 Remove code that sets lr_slot_offset according to what a previous
11091 version of epiphany_emit_save_restore used to do.
11092 (epiphany_emit_save_restore): When doing an lr save or restore,
11093 set/verify lr_slot_known and lr_slot_offset.
11094
0a36a221 110952013-04-08 Xinyu Qi <xyqi@marvell.com>
11096
8f028904 11097 PR target/54338
11098 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
0a36a221 11099 in ALL_REGS.
11100
ce0deccc 111012013-04-08 Richard Biener <rguenther@suse.de>
11102
11103 * alias.c (find_base_term): Fix thinko in previous change.
11104
0dc30b0c 111052013-04-08 Jakub Jelinek <jakub@redhat.com>
11106
11107 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11108 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11109 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
11110 if possible to compute val.
11111 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11112 For QImode integers don't require anything about precision. Use
11113 const_with_all_bytes_same to find out if the constant doesn't have
11114 repeated bytes in it.
11115
31efd1ec 111162013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11117
11118 * config/s390/s390.c (s390_expand_insv): Only accept insertions
11119 within mode size.
11120
ce4f2aee 111212013-04-08 Marek Polacek <polacek@redhat.com>
11122
11123 PR rtl-optimization/48182
11124 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11125 value to 1.
11126
4aa618ec 111272013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
11128
11129 PR target/55487
11130 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11131 nuses, make sure we have a label.
11132
c68f8616 111332013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11134
11135 PR target/56843
11136 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11137 (rs6000_emit_swdiv_low_precision): Remove.
11138 (rs6000_emit_swdiv): Rewrite to handle between one and four
11139 iterations of Newton-Raphson generally; modify required number of
11140 iterations for some cases.
11141 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11142
b257b412 111432013-04-05 Steven Bosscher <steven@gcc.gnu.org>
11144
11145 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11146 set-but-unused variable.
11147
11148 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11149 basic blocks of released function bodies garbage-collectable.
11150
11151 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11152 (struct rtl_opt_pass): Add TODO_df_finish.
11153
11154 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11155
353cf59a 111562013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11157
11158 * config/arm/constraints.md (q): New constraint.
11159 * config/arm/ldrdstrd.md: New file.
11160 * config/arm/arm.md (ldrdstrd.md) New include.
11161 (arm_movdi): Use "q" instead of "r" constraint
11162 for double-word memory access.
11163 (movdf_soft_insn): Likewise.
11164 * config/arm/vfp.md (movdi_vfp): Likewise.
11165 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
8f028904 11166 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
353cf59a 11167 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11168 (mem_ok_for_ldrd_strd): Likewise.
11169 (output_move_double): Update assertion.
11170
cb860649 111712013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11172
11173 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11174
3dc953f2 111752013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11176
11177 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11178 define_insn_and_split.
11179 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11180
a5b9aa37 111812013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11182
11183 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11184 define_insn_and_split.
8f028904 11185 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
a5b9aa37 11186 (shiftsi3_compare): New pattern.
11187 (rrx): New pattern.
11188 * config/arm/unspecs.md (UNSPEC_RRX): New.
11189
83e2b922 111902013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11191
11192 * config/arm/arm.md (negdi_extendsidi): New pattern.
11193 (negdi_zero_extendsidi): Likewise.
11194
ba6a3b2f 111952013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11196
11197 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11198 define_insn_and_split.
11199 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11200 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11201
2f9b23e3 112022013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11203
11204 * config/arm/arm.md (arm_subdi3): Convert define_insn into
11205 define_insn_and_split.
11206 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11207 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11208
9154bd82 112092013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11210
11211 * config/arm/arm.md (subsi3_carryin): New pattern.
11212 (subsi3_carryin_const): Likewise.
11213 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11214 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
11215
560d151e 112162013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11217
11218 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
11219
a7de272d 112202013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11221
11222 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
9154bd82 11223 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
a7de272d 11224
30441c92 112252013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11226
11227 * config/arm/arm.c (arm_expand_builtin): Change fcode
11228 type to unsigned int.
11229
f2ba16e7 112302013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11231
11232 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
11233
224959cb 112342013-04-04 Ian Lance Taylor <iant@google.com>
11235
8f028904 11236 * doc/standards.texi (Standards): The Go frontend supports the Go 1
11237 language standard.
224959cb 11238
581d1e0f 112392013-04-04 Steven Bosscher <steven@gcc.gnu.org>
11240
11241 PR middle-end/56729
11242 * df-scan.c (df_insn_delete): Disable failing assert.
11243
23877cbc 112442013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11245
11246 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
11247 New function prototype.
11248 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
11249 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
11250 (arm_builtin_vectorized_function): New function.
11251
4bc74efc 112522013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11253
11254 * config/arm/arm_neon_builtins.def: New file.
11255 * config/arm/arm.c (neon_builtin_data): Move contents to
11256 arm_neon_builtins.def.
11257 (enum arm_builtins): Include neon builtin definitions.
11258 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
8f028904 11259 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
4bc74efc 11260
6040d650 112612013-04-04 Marek Polacek <polacek@redhat.com>
11262
11263 PR tree-optimization/48186
11264 * predict.c (maybe_hot_frequency_p): Return false if
11265 HOT_BB_FREQUENCY_FRACTION is 0.
11266 (cgraph_maybe_hot_edge_p): Likewise.
11267
f77a1eec 112682013-04-04 Richard Biener <rguenther@suse.de>
11269
11270 PR tree-optimization/56826
11271 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
11272 more accurately.
11273
2dca30a9 112742013-04-04 Richard Biener <rguenther@suse.de>
11275
11276 PR tree-optimization/56213
11277 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
8f028904 11278 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
2dca30a9 11279
616d052d 112802013-04-04 Richard Biener <rguenther@suse.de>
11281
11282 PR tree-optimization/56837
11283 * tree-loop-distribution.c (classify_partition): For non-zero
11284 values require that the value has the same precision as its
11285 mode to be useful as memset value.
11286
34bcbb16 112872013-04-03 Nick Clifton <nickc@redhat.com>
11288
8f028904 11289 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
34bcbb16 11290 (fmssf4): Use fmsf.s on E3V5 architectures.
11291 (fnmasf4): Use fnmaf.s on E3V5 architectures.
11292 (fnmssf4): Use fnmsf.s on E3V5 architectures.
11293
2aebdec3 112942013-04-03 Jeff Law <law@redhat.com>
11295
11296 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
11297 (lra-eliminations.o): Likewise.
11298
8515a84d 112992013-04-03 Teresa Johnson <tejohnson@google.com>
11300
11301 * gcov-io.c (compute_working_sets): Moved most of body of old
8f028904 11302 compute_working_sets here from profile.c.
8515a84d 11303 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
8f028904 11304 (gcov_working_set_t): Moved typedef here from basic-block.h
11305 (compute_working_set): Declare.
8515a84d 11306 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
11307 (get_working_sets): Renamed from compute_working_set,
8f028904 11308 replace most of body with call to new compute_working_sets.
8515a84d 11309 (get_exec_counts): Replace call to compute_working_sets
8f028904 11310 to get_working_sets.
11311 * profile.h (get_working_sets): Renamed from compute_working_set.
8515a84d 11312 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
8f028904 11313 to get_working_sets.
8515a84d 11314 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
11315 * gcov-dump.c (dump_working_sets): New function.
11316
687a1ea1 113172013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
11318
11319 * hwint.c (sext_hwi, zext_hwi): New functions.
11320 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
11321 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
11322 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
11323 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
11324 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
11325 (sext_hwi, zext_hwi): New functions.
34bcbb16 11326
99db618c 113272013-04-03 Jeff Law <law@redhat.com>
11328
11329 PR tree-optimization/56799
8f028904 11330 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
11331 back test for widening conversion erroneously dropped in prior change.
99db618c 11332
9743e73f 113332013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11334
11335 PR target/56809
11336 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
11337 instead of next_real_insn.
11338
ef951254 113392013-04-03 Marek Polacek <polacek@redhat.com>
11340
11341 PR sanitizer/55702
8f028904 11342 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
ef951254 11343
795e7242 113442013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11345
11346 PR target/56809
11347 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
11348 next_real_insn.
11349 (thumb1_output_casesi): Likewise.
11350 (thumb2_output_casesi): Likewise.
11351
042301ef 113522013-04-03 Richard Biener <rguenther@suse.de>
11353
11354 PR tree-optimization/56817
11355 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
11356 Split out ...
11357 (tree_unroll_loops_completely_1): ... new function to manually
11358 walk the loop tree, properly defering outer loops of unrolled
11359 loops to later iterations.
11360
19bacd59 113612013-04-03 Marc Glisse <marc.glisse@inria.fr>
11362
11363 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
11364 (vectorizable_load): Likewise.
11365 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
11366 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
11367
872e3ff8 113682013-04-03 Marc Glisse <marc.glisse@inria.fr>
11369
11370 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
11371 BIT_FIELD_REF.
11372
b1135d9a 113732013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
11374
11375 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
11376
bb760dac 113772013-04-03 Bin Cheng <bin.cheng@arm.com>
11378
11379 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
11380
07681e3a 113812013-04-03 Marc Glisse <marc.glisse@inria.fr>
11382
11383 PR tree-optimization/56790
8f028904 11384 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
11385 folding.
07681e3a 11386
f0a05bd5 113872013-04-03 Marc Glisse <marc.glisse@inria.fr>
11388
11389 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
11390 Handle VEC_MERGE.
11391 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
11392 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
11393 equal arguments.
11394
0b049e15 113952013-04-03 Jakub Jelinek <jakub@redhat.com>
11396
11397 PR c/19449
11398 * tree.h (force_folding_builtin_constant_p): New decl.
11399 * builtins.c (force_folding_builtin_constant_p): New variable.
11400 (fold_builtin_constant_p): Fold immediately also if
11401 force_folding_builtin_constant_p.
11402
1fa434e3 114032013-04-03 Richard Biener <rguenther@suse.de>
11404
11405 PR tree-optimization/56812
11406 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
11407 DRs of the same interleaving chain are independent.
11408
0a56fff4 114092013-04-02 Jason Merrill <jason@redhat.com>
11410
11411 * gdbinit.in (pbb): Use debug fn.
11412
f8477cfc 114132013-04-02 Lawrence Crowl <crowl@google.com>
11414
11415 * sese.h (struct ivtype_map_elt_s): Remove unused.
11416 (extern debug_ivtype_map): Remove unused.
11417 (extern eq_ivtype_map_elts): Remove unused.
11418 * sese.c (debug_ivtype_map): Removed unused.
11419 (debug_ivtype_map_1): Removed unused.
11420 (debug_ivtype_elt): Remove unused.
11421 (eq_ivtype_map_elts): Remove unused.
11422
11423
5496fac8 114242013-04-02 Kai Tietz <ktietz@redhat.com>
11425
11426 PR target/52790
11427 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
11428 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
8f028904 11429 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
11430 function.
5496fac8 11431 (legitimize_pe_coff_symbol): Likewise.
11432 (is_imported_p): New helper-function.
11433 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
11434 for Windows x64 targets.
11435 (ix86_expand_prologue): Optimize for pe-coff targets.
11436 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
11437 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
11438 medium/large code-model.
11439 (legitimize_pic_address): Likewise.
11440 (legitimize_tls_address): Likewise.
11441 (ix86_expand_call): Likewise.
11442 (x86_output_mi_thunk): Likewise.
11443 (get_dllimport_decl): Add new beimport argument.
11444 (construct_plt_address): Don't assert for x64 pe-coff targets.
11445 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
11446 targets.
11447 (SYMBOL_FLAG_STUBVAR): New macro.
11448 (SYMBOL_REF_STUBVAR_P): Likewise.
11449 * config/i386/winnt.c (stub_list): New structure.
11450 (stub_head): New local variable.
11451 (i386_pe_record_stub): New function.
11452 (i386_pe_file_end): Emit refptr-stubs.
11453
9f559b20 114542013-04-02 Jakub Jelinek <jakub@redhat.com>
11455
50212417 11456 PR rtl-optimization/56745
11457 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
11458 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
11459
137eddca 11460 PR c++/34949
11461 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
11462 and both of them are MEM_REFs, just compare first argument for
11463 equality and attempt to deal even with differing offsets.
11464
9f559b20 11465 PR c++/34949
11466 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
11467 of gimple_clobber_p to be MEM_REF.
11468 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
11469 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
11470 after gimplification.
11471 * asan.c (get_mem_ref_of_assignment): Don't instrument
11472 gimple_clobber_p stmts.
11473 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
11474 gimple_clobber_p stmt if they have MEM_REF lhs and
11475 are dead because of another gimple_clobber_p stmt.
11476 * tree-ssa-live.c (clear_unused_block_pointer): Treat
11477 gimple_clobber_p stmts like debug stmts.
11478 (remove_unused_locals): Remove clobbers with MEM_REF lhs
11479 that refer to unused VAR_DECLs or uninitialized values.
11480 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
11481 gimple_clobber_p stmts if they refer to removed parameters.
11482 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
11483 formatting.
11484
7d71feb3 114852013-04-02 Uros Bizjak <ubizjak@gmail.com>
11486
11487 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
11488 using SWI48 mode attribute.
11489
10ef9b48 114902013-04-02 Wei Mi <wmi@google.com>
11491
11492 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
11493 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
11494 *<rotate_insn><mode>3_mask in i386.md.
11495
ced500f0 114962013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
11497
11498 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
11499
95e19962 115002013-04-02 Richard Biener <rguenther@suse.de>
11501
11502 PR tree-optimization/56778
11503 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11504 Runtime alias tests are not supported for gather loads.
11505 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
11506 stmts referenced from SSA operands before updating SSA form.
11507
da10bc87 115082013-04-02 Ian Caulfield <ian.caulfield@arm.com>
11509 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11510
11511 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
7d71feb3 11512 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
11513 * config/arm/cortex-a53.md: New file.
11514 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
11515 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
11516 * config/arm/arm.c (arm_issue_rate): Likewise.
11517 * config/arm/arm-tune.md: Regenerate
11518 * config/arm/arm-tables.opt: Regenerate.
11519 * config/arm/arm-cores.def: Add cortex-a53.
da10bc87 11520
d62eaedc 115212013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
11522
11523 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
11524 non-static link.
11525
fb436d5f 115262013-04-02 Sofiane Naci <sofiane.naci@arm.com>
11527
da10bc87 11528 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
11529 scalar load/store operations using B/H registers.
fb436d5f 11530 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
11531
43bd8f5c 115322013-04-02 Sofiane Naci <sofiane.naci@arm.com>
11533
11534 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
11535 scalar move.
11536 * config/aarch64/aarch64.c
11537 (aarch64_simd_scalar_immediate_valid_for_move): New.
11538 * config/aarch64/aarch64-protos.h
11539 (aarch64_simd_scalar_immediate_valid_for_move): New.
11540 * config/aarch64/constraints.md (Dh, Dq): New.
11541 * config/aarch64/iterators.md (hq): New.
11542
965df672 115432013-04-02 Eric Botcazou <ebotcazou@adacore.com>
11544
11545 * reorg.c (get_branch_condition): Deal with conditional returns.
11546 (fill_simple_delay_slots): Remove dead code dealing with jumps.
11547
8089c750 115482013-04-01 Wei Mi <wmi@google.com>
11549
11550 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
11551 Truncate operand 2 using %b asm operand modifier.
11552 (*<shift_insn><mode>3_mask): Ditto.
11553 (*<rotate_insn><mode>3_mask): Ditto.
11554
53859f1c 115552013-04-01 Steven Bosscher <steven@gcc.gnu.org>
11556
11557 PR middle-end/56798
11558 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
11559
262e5908 115602013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
11561
11562 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
11563 of next_real_insn.
11564 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
11565
e85eaec5 115662013-03-30 Lawrence Crowl <crowl@google.com>
11567
11568 * dse.c (clear_alias_sets): Remove never set.
11569 (disqualified_clear_alias_sets): Remove never set.
11570 (clear_alias_mode_pool): Remove never set.
11571 (dse_step0): Remove condition that is never true.
11572 (canon_address): Remove condition that is never true.
11573 (dse_step7): Remove condition that is never true.
11574 (rest_of_handle_dse): Remove condition that is never true.
11575 (rest_of_handle_dse::did_global): Remove never read from above.
11576 (dse_step2_spill): Remove never called from above.
11577 (dse_step5_spill): Remove never called from above.
11578
91f71fa3 115792013-03-30 Steven Bosscher <steven@gcc.gnu.org>
11580
9ddbb404 11581 * doc/md.texi (Standard Names) <casesi>: Update documentation for
11582 JUMP_TABLE_DATA changes.
11583 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
11584 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
11585 (Insns) <jump_table_data>: New entry.
11586 * doc/tm.texi: Regenerate.
11587
91f71fa3 11588 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
11589
11590 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
11591 for table jump at the end of a basic block using tablejump_p.
11592 * targhooks.c (default_invalid_within_doloop): Likewise.
11593 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
11594 target hook implementation that is identical to the default hook.
11595 (rs6000_invalid_within_doloop): Remove.
11596
11597 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
11598 unused variable from tablejump_p call.
11599
11600 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
11601 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
11602 (INSN_DELETED_P): Likewise.
11603 (emit_jump_table_data): New prototype.
11604 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
11605 after 4th as unused.
11606 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
11607 * sched-vis.c (print_insn): Likewise.
11608 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
11609 insn for compatibility with back ends that use next_active_insn to
11610 identify jump table data.
11611 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
11612 (remove_insn): Likewise.
11613 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
11614 to be emitted.
11615 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
11616 (emit_jump_table_data): New function.
11617
11618 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
11619 basic block, a JUMP_TABLE_DATA never is.
11620 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
11621 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
11622 off from code handling real insns.
11623 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
11624 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
11625 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
11626 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
11627 is not a NONDEBUG_INSN_P.
11628 * ira-costs.c (scan_one_insn): Likewise.
11629 * jump.c (mark_all_labels): Likewise.
11630 (mark_jump_label_1): Likewise.
11631 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
11632 * lra.c (get_insn_freq): Expect all insns reaching here to be in
11633 a basic block.
11634 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
11635 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
11636 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
11637 JUMP_TABLE_DATA_P insns.
11638 (calculate_elim_costs_all_insns): Likewise.
11639 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
11640 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
11641 (delete_output_reload): Code style fixups.
11642 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
11643 insn flags on this non-insn.
11644 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
11645 as scheduling barriers, for pre-change compatibility.
11646 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
11647 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
11648
11649 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
11650 redundant JUMP_TABLE_DATA_P test.
11651 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
11652 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
11653 (frv_for_each_packet): Likewise.
11654 * config/i386/i386.c (min_insn_size): Likewise.
11655 (ix86_avoid_jump_mispredicts): Likewise.
11656 * config/m32r/m32r.c (m32r_is_insn): Likewise.
11657 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
11658 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
11659 (mips16_insn_length): Robustify.
11660 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
11661 (mips16_split_long_branches): Likewise.
11662 * config/pa/pa.c (pa_combine_instructions): Likewise.
11663 * config/rs6000/rs6000.c (get_next_active_insn): Treat
11664 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
11665 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
11666 as contributing to pool range lengths.
11667 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
11668 Remove redundant JUMP_TABLE_DATA_P test.
11669 (sh_loop_align): Likewise.
11670 (split_branches): Likewise.
11671 (sh_insn_length_adjustment): Likewise.
11672 * config/spu/spu.c (get_branch_target): Likewise.
11673
9e179a64 116742013-03-29 Jan Hubicka <jh@suse.cz>
11675
11676 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
11677 gcov streaming; stream hot bb threshold to ltrans.
11678 * predict.c (get_hot_bb_threshold): Break out from ....
11679 (maybe_hot_count_p): ... here.
11680 (set_hot_bb_threshold): New function.
11681 * lto-section-in.c (lto_section_name): Add profile.
11682 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
11683 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
11684 and data-streamer.h
11685 (histogram_entry): New structure.
11686 (histogram, histogram_pool): New global vars.
11687 (histogram_hash): New structure.
11688 (histogram_hash::hash): New method.
11689 (histogram_hash::equal): Likewise.
11690 (account_time_size): New function.
11691 (cmp_counts): New function.
11692 (dump_histogram): New function.
11693 (ipa_profile_generate_summary): New function.
11694 (ipa_profile_write_summary): New function.
11695 (ipa_profile_read_summary): New function.
11696 (ipa_profile): Decide on threshold.
7d71feb3 11697 (pass_ipa_profile): Add ipa_profile_write_summary and
11698 ipa_profile_read_summary.
9e179a64 11699 * Makefile.in (ipa.o): Update dependencies.
11700 * lto-streamer.h (LTO_section_ipa_profile): New section.
11701
80a58eb0 117022013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
11703
11704 * tree.h (VAR_P): New.
11705
51bad497 117062013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
11707
11708 PR lto/56777
11709 * doc/invoke.texi ([-fwhole-program]): Fix typo.
11710
77985f1a 117112013-03-29 Steven Bosscher <steven@gcc.gnu.org>
11712
11713 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
11714 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
11715 (control_flow_insn_p): Likewise.
11716 * cfgrtl.c (duplicate_insn_chain): Likewise.
11717 * final.c (get_attr_length_1): Likewise.
11718 (shorten_branches): Likewise.
11719 (final_scan_insn): Likewise.
11720 * function.c (instantiate_virtual_regs): Likewise.
11721 * gcse.c (insert_insn_end_basic_block): Likewise.
11722 * ira-costs.c (scan_one_insn): Likewise.
11723 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
11724 * lra.c (check_rtl): Likewise.
11725 * reload1.c (elimination_costs_in_insn): Likewise.
11726 * reorg.c (follow_jumps): Likewise.
11727
11728 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
11729 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
11730 (thumb_far_jump_used_p): Likewise.
11731 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
11732 (workaround_speculation): Likewise.
11733 (add_sched_insns_for_speculation): Likewise.
11734 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
11735 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
11736 (frv_for_each_packet): Likewise.
11737 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
11738 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
11739 (final_emit_insn_group_barriers): Likewise.
11740 * config/m32r/m32r.c (m32r_is_insn): Likewise.
11741 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
11742 (mips16_insn_length): Likewise.
11743 * config/pa/pa.c (pa_reorg): Likewise.
11744 (pa_combine_instructions): Likewise.
11745 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
11746 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
11747 (sh_reorg): Likewise.
11748 (split_branches): Likewise.
11749 * config/spu/spu.c (get_branch_target): Likewise.
11750
11751 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
11752 JUMP_TABLE_DATA_P.
11753
78e8d5ff 117542013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
11755
51bad497 11756 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
78e8d5ff 11757 Fix declaration name.
11758
31d49894 117592013-03-28 Lawrence Crowl <crowl@google.com>
11760
11761 * graphds.h (struct graph.indicies): Remove unused.
11762 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
11763 (SCOP_ORIGINAL_PDDRS): Remove unused.
11764 * sese.h (extern insert_loop_close_phis): Removed unused.
11765 (extern insert_guard_phis): Removed unused.
11766 (extern ivtype_map_elt_info): Removed unused.
11767 (new_ivtype_map_elt): Removed unused.
11768 * sese.c (ivtype_map_elt_info): Removed unused.
11769
c7d89805 117702013-03-28 Lawrence Crowl <crowl@google.com>
11771
11772 * Makefile.in: Add several missing include dependences.
11773 (DUMPFILE_H): New.
11774 (test-dump.o): New. This object is not added to any executable,
11775 but is present for ad-hoc testing.
11776 * bitmap.c
11777 (debug (const bitmap_head_def &)): New.
11778 (debug (const bitmap_head_def *)): New.
11779 * bitmap.h
11780 (extern debug (const bitmap_head_def &)): New.
11781 (extern debug (const bitmap_head_def *)): New.
11782 * cfg.c
11783 (debug (edge_def &)): New.
11784 (debug (edge_def *)): New.
11785 * cfghooks.c
11786 (debug (basic_block_def &)): New.
11787 (debug (basic_block_def *)): New.
11788 * dumpfile.h
11789 (dump_node (const_tree, int, FILE *)): Correct source file.
11790 * dwarf2out.c
11791 (debug (die_struct &)): New.
11792 (debug (die_struct *)): New.
11793 * dwarf2out.h
11794 (extern debug (die_struct &)): New.
11795 (extern debug (die_struct *)): New.
11796 * gimple-pretty-print.c
11797 (debug (gimple_statement_d &)): New.
11798 (debug (gimple_statement_d *)): New.
11799 * gimple-pretty-print.h
11800 (extern debug (gimple_statement_d &)): New.
11801 (extern debug (gimple_statement_d *)): New.
11802 * ira-build.c
11803 (debug (ira_allocno_copy &)): New.
11804 (debug (ira_allocno_copy *)): New.
11805 (debug (ira_allocno &)): New.
11806 (debug (ira_allocno *)): New.
11807 * ira-int.h
11808 (extern debug (ira_allocno_copy &)): New.
11809 (extern debug (ira_allocno_copy *)): New.
11810 (extern debug (ira_allocno &)): New.
11811 (extern debug (ira_allocno *)): New.
11812 * ira-lives.c
11813 (debug (live_range &)): New.
11814 (debug (live_range *)): New.
11815 * lra-int.h
11816 (debug (lra_live_range &)): New.
11817 (debug (lra_live_range *)): New.
11818 * lra-lives.c
11819 (debug (lra_live_range &)): New.
11820 (debug (lra_live_range *)): New.
11821 * omega.c
11822 (debug (omega_pb_d &)): New.
11823 (debug (omega_pb_d *)): New.
11824 * omega.h
11825 (extern debug (omega_pb_d &)): New.
11826 (extern debug (omega_pb_d *)): New.
11827 * print-rtl.c
11828 (debug (const rtx_def &)): New.
11829 (debug (const rtx_def *)): New.
11830 * print-tree.c
11831 (debug_tree (tree): Move within file.
11832 (debug_raw (const tree_node &)): New.
11833 (debug_raw (const tree_node *)): New.
11834 (dump_tree_via_hooks (const tree_node *, int)): New.
11835 (debug (const tree_node &)): New.
11836 (debug (const tree_node *)): New.
11837 (debug_verbose (const tree_node &)): New.
11838 (debug_verbose (const tree_node *)): New.
11839 (debug_head (const tree_node &)): New.
11840 (debug_head (const tree_node *)): New.
11841 (debug_body (const tree_node &)): New.
11842 (debug_body (const tree_node *)): New.
11843 (debug_vec_tree (tree): Move and reimplement in terms of dump.
11844 (debug (vec<tree, va_gc> &)): New.
11845 (debug (vec<tree, va_gc> *)): New.
11846 * rtl.h
11847 (extern debug (const rtx_def &)): New.
11848 (extern debug (const rtx_def *)): New.
11849 * sbitmap.c
11850 (debug_raw (simple_bitmap_def &)): New.
11851 (debug_raw (simple_bitmap_def *)): New.
11852 (debug (simple_bitmap_def &)): New.
11853 (debug (simple_bitmap_def *)): New.
11854 * sbitmap.h
11855 (extern debug (simple_bitmap_def &)): New.
11856 (extern debug (simple_bitmap_def *)): New.
11857 (extern debug_raw (simple_bitmap_def &)): New.
11858 (extern debug_raw (simple_bitmap_def *)): New.
11859 * sel-sched-dump.c
11860 (debug (vinsn_def &)): New.
11861 (debug (vinsn_def *)): New.
11862 (debug_verbose (vinsn_def &)): New.
11863 (debug_verbose (vinsn_def *)): New.
11864 (debug (expr_def &)): New.
11865 (debug (expr_def *)): New.
11866 (debug_verbose (expr_def &)): New.
11867 (debug_verbose (expr_def *)): New.
11868 (debug (vec<rtx> &)): New.
11869 (debug (vec<rtx> *)): New.
11870 * sel-sched-dump.h
11871 (extern debug (vinsn_def &)): New.
11872 (extern debug (vinsn_def *)): New.
11873 (extern debug_verbose (vinsn_def &)): New.
11874 (extern debug_verbose (vinsn_def *)): New.
11875 (extern debug (expr_def &)): New.
11876 (extern debug (expr_def *)): New.
11877 (extern debug_verbose (expr_def &)): New.
11878 (extern debug_verbose (expr_def *)): New.
11879 (extern debug (vec<rtx> &)): New.
11880 (extern debug (vec<rtx> *)): New.
11881 * sel-sched-ir.h
11882 (_list_iter_cond_expr): Make inline instead of static.
11883 * sreal.c
11884 (debug (sreal &)): New.
11885 (debug (sreal *)): New.
11886 * sreal.h
11887 (extern debug (sreal &)): New.
11888 (extern debug (sreal *)): New.
11889 * tree.h
11890 (extern debug_raw (const tree_node &)): New.
11891 (extern debug_raw (const tree_node *)): New.
11892 (extern debug (const tree_node &)): New.
11893 (extern debug (const tree_node *)): New.
11894 (extern debug_verbose (const tree_node &)): New.
11895 (extern debug_verbose (const tree_node *)): New.
11896 (extern debug_head (const tree_node &)): New.
11897 (extern debug_head (const tree_node *)): New.
11898 (extern debug_body (const tree_node &)): New.
11899 (extern debug_body (const tree_node *)): New.
11900 (extern debug (vec<tree, va_gc> &)): New.
11901 (extern debug (vec<tree, va_gc> *)): New.
11902 * tree-cfg.c
11903 (debug (struct loop &)): New.
11904 (debug (struct loop *)): New.
11905 (debug_verbose (struct loop &)): New.
11906 (debug_verbose (struct loop *)): New.
11907 * tree-dump.c: Add header dependence.
11908 * tree-flow.h
11909 (extern debug (struct loop &)): New.
11910 (extern debug (struct loop *)): New.
11911 (extern debug_verbose (struct loop &)): New.
11912 (extern debug_verbose (struct loop *)): New.
11913 * tree-data-ref.c
11914 (debug (data_reference &)): New.
11915 (debug (data_reference *)): New.
11916 (debug (vec<data_reference_p> &)): New.
11917 (debug (vec<data_reference_p> *)): New.
11918 (debug (vec<ddr_p> &)): New.
11919 (debug (vec<ddr_p> *)): New.
11920 * tree-data-ref.h
11921 (extern debug (data_reference &)): New.
11922 (extern debug (data_reference *)): New.
11923 (extern debug (vec<data_reference_p> &)): New.
11924 (extern debug (vec<data_reference_p> *)): New.
11925 (extern debug (vec<ddr_p> &)): New.
11926 (extern debug (vec<ddr_p> *)): New.
11927 * tree-ssa-alias.c
11928 (debug (pt_solution &)): New.
11929 (debug (pt_solution *)): New.
11930 * tree-ssa-alias.h
11931 (extern debug (pt_solution &)): New.
11932 (extern debug (pt_solution *)): New.
11933 * tree-ssa-alias.c
11934 (debug (_var_map &)): New.
11935 (debug (_var_map *)): New.
11936 (debug (tree_live_info_d &)): New.
11937 (debug (tree_live_info_d *)): New.
11938 * tree-ssa-alias.h
11939 (extern debug (_var_map &)): New.
11940 (extern debug (_var_map *)): New.
11941 (extern debug (tree_live_info_d &)): New.
11942 (extern debug (tree_live_info_d *)): New.
11943
ed8b1c1f 119442013-03-28 Jan Hubicka <jh@suse.cz>
11945
11946 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
11947
0d62a093 119482013-03-28 Ian Bolton <ian.bolton@arm.com>
11949
11950 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
11951 record only when desired or required.
11952
cea39eb6 119532013-03-28 Uros Bizjak <ubizjak@gmail.com>
11954
11955 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
11956 *vec_extractv2di_1_rex64. Use x64 isa attribute.
11957
8f028904 119582013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
7f4f914a 11959
11960 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
11961 (*andsi3_compare0_uxtw): New pattern.
11962 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
11963 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
11964
fc44a215 119652013-03-28 Jan Hubicka <jh@suse.cz>
11966
11967 * data-streamer-in.c (streamer_read_gcov_count): New function.
11968 * gimple-streamer-out.c: Include value-prof.h.
11969 (output_gimple_stmt): Output histogram.
11970 (output_bb): Use streamer_write_gcov_count.
11971 * value-prof.c: Include data-streamer.h
11972 (dump_histogram_value): Add HIST_TYPE_MAX.
11973 (stream_out_histogram_value): New function.
11974 (stream_in_histogram_value): New function.
11975 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
11976 (stream_out_histogram_value, stream_in_histogram_value): Declare.
11977 * data-streamer-out.c (streamer_write_gcov_count): New function.
11978 (streamer_write_gcov_count_stream): New function.
11979 * lto-cgraph.c (lto_output_edge): Update counter streaming.
11980 (lto_output_node): Likewise.
11981 (input_node, input_edge): Likewise.
11982 * lto-streamer-out.c (output_cfg): Update streaming.
11983 * lto-streamer-in.c (input_cfg): Likewise.
11984 * data-streamer.h (streamer_write_gcov_count,
11985 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
11986 * gimple-streamer-in.c: Include value-prof.h
11987 (input_gimple_stmt): Input histograms.
11988 (input_bb): Update profile streaming.
11989
7d71feb3 119902013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
64a8b204 11991
8f028904 11992 * genmodes.c (emit_max_int): New function.
f9cb58ae 11993 (emit_insn_modes_h): Added call to emit_max_function.
51bad497 11994 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
11995 Added doc.
f9cb58ae 11996 * machmode.def: Fixed comment.
fc44a215 11997
7d71feb3 119982013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
f9cb58ae 11999
12000 * combine.c (try_combine): Removed useless assert.
64a8b204 12001 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12002
43d7caca 120032013-03-28 Marek Polacek <polacek@redhat.com>
12004 Richard Biener <rguenther@suse.de>
12005
12006 PR tree-optimization/56695
12007 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12008 build signed result of a vector comparison.
12009 * tree-cfg.c (verify_gimple_comparison): Check that a result
12010 of a vector comparison has signed type.
12011
8f3e4987 120122013-03-28 Richard Biener <rguenther@suse.de>
12013
12014 PR tree-optimization/37021
12015 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12016 do not restrict gaps between groups.
12017 * tree-vect-stmts.c (vectorizable_load): Properly account for
12018 a gap between groups.
12019
5b0d71ce 120202013-03-28 Eric Botcazou <ebotcazou@adacore.com>
12021
12022 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12023 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12024 is not enabled.
12025
886c5e35 120262013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
12027
12028 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12029 * doc/extend.texi (Named Address Spaces): Ditto.
12030 (Variable Attributes): Ditto.
12031
8f3115a2 120322013-03-27 Kai Tietz <ktietz@redhat.com>
12033
10b0ba24 12034 * config.build: Add support for cygwin x64 target.
12035 * config.gcc: Likewise.
12036 * config.host: Likewise.
12037 * configure.ac: Likewise
12038 * configure: Regenerated.
8f3115a2 12039
5f7f04d8 120402013-03-27 Kai Tietz <ktietz@redhat.com>
cb8160ad 12041
12042 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12043 * config/i386/t-cygwin-w64: New file.
12044 * config/i386/cygwin-w64.h: New file.
12045 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12046 and add support for x64-cygwin target.
12047 (CPP_SPEC): Likewise.
12048 (CXX_WRAP_SPEC_LIST): Undefine before define.
12049 (LIBGCJ_SONAME): Use 15 as version.
12050
6c9b05aa 120512013-03-27 Richard Biener <rguenther@suse.de>
12052
12053 PR tree-optimization/56716
12054 * tree-ssa-structalias.c (perform_var_substitution): Adjust
12055 dumping for ref nodes.
12056
4f0c9cf2 120572013-03-27 Martin Jambor <mjambor@suse.cz>
12058
12059 PR tree-optimization/55334
12060 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12061 restricted pointers to arrays.
12062
e79e977d 120632013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
12064
12065 * Makefile.in (.SUFFIXES): Add .cc.
12066 (.c.o): Apply same recipe for implicit rule .cc.o.
12067
f837b1f7 120682013-03-27 Richard Biener <rguenther@suse.de>
12069
12070 PR tree-optimization/37021
12071 * tree-vect-data-refs.c (vect_check_strided_load): Allow
12072 REALPART/IMAGPART_EXPRs around the supported refs.
12073 * tree-ssa-structalias.c (find_func_aliases): Assume that
12074 floating-point values are not used to transfer pointers.
12075
df9f2e40 120762013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
12077
51bad497 12078 * target.def (TARGET_HAS_IFUNC_P): New target hook.
12079 * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12080 * doc/tm.texi: Regenerate.
12081 * targhooks.h (default_has_ifunc_p): New.
12082 * targhooks.c (default_has_ifunc_p): Ditto.
12083 * config/linux-protos.h: New file.
12084 * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
df9f2e40 12085 this hook for linux which disables support of indirect functions in
12086 android.
51bad497 12087 * config/linux-android.c: New file.
12088 * config/t-linux-android.c: Ditto.
12089 * config.gcc: Added new object file linux-android.o.
12090 * config/i386/i386.c (ix86_get_function_versions_dispatcher):
df9f2e40 12091 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
51bad497 12092 * varasm.c (do_assemble_alias): Likewise.
df9f2e40 12093 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12094 doesn't support indirect functions.
12095 * configure: Regenerate.
12096
d18af62e 120972013-03-27 Bin Cheng <bin.cheng@arm.com>
12098
12099 PR target/56102
12100 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12101 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12102 mult-word mode.
12103
2e9f00a3 121042013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12105
12106 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12107
6ee84d16 121082013-03-27 Terry Guo <terry.guo@arm.com>
12109
12110 * config/arm/arm-cores.def: Added core cortex-r7.
12111 * config/arm/arm-tune.md: Regenerated.
12112 * config/arm/arm-tables.opt: Regenerated.
12113 * doc/invoke.texi: Added entry for core cortex-r7.
12114
9f1b21e1 121152013-03-27 Walter Lee <walt@tilera.com>
12116
12117 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12118 double-decrement of next_scratch_regno.
12119
21743daa 121202013-03-27 Walter Lee <walt@tilera.com>
12121
7946a1e8 12122 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21743daa 12123 input operands.
12124 (insn_v1mulus): Ditto.
12125 (insn_v2muls): Ditto.
12126
d7c3902f 121272013-03-27 Walter Lee <walt@tilera.com>
12128
7d71feb3 12129 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
d7c3902f 12130 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12131
cd002190 121322013-03-27 Walter Lee <walt@tilera.com>
12133
7d71feb3 12134 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
cd002190 12135 (*sibcall_value): Ditto.
12136
ddff9023 121372013-03-27 Walter Lee <walt@tilera.com>
12138
12139 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12140 (insn_mnz_v8qi): ... this ...
12141 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
12142 vector equivalent.
12143 (insn_v<n>mnz): Replaced by ...
12144 (insn_v1mnz): ... this ...
12145 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
12146 equivalent.
12147 (insn_mz_<mode>): Replaced by ...
12148 (insn_mz_v8qi): ... this ...
12149 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
12150 vector equivalent.
12151 (insn_v<n>mz): Replaced by ...
12152 (insn_v1mz): ... this ...
12153 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
12154 equivalent.
12155
79c96309 121562013-03-26 Eric Botcazou <ebotcazou@adacore.com>
12157
12158 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12159
b48eb844 121602013-03-26 Roland McGrath <mcgrathr@google.com>
12161
12162 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12163 than fprintf with a non-constant, non-format string.
12164
ee9e7384 121652013-03-26 Uros Bizjak <ubizjak@gmail.com>
12166
12167 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12168 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
12169 operand 0 predicate.
12170 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12171 attribute. Use general_x64nomem_operand as operand 1 predicate.
12172 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12173 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
12174 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12175 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
12176 (mov<mode>_insv_1): Remove expander. Merge insn with
12177 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12178 Use general_x64nomem_operand as operand 1 predicate.
12179 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12180 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12181 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
12182 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12183 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
12184 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12185 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
12186 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12187 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
12188 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12189 (general_x64nomem_operand): Ditto.
12190
57c8b8f0 121912013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
2c4fdad4 12192
12193 * config/rtems.opt: Add -pthread option.
12194
5abc99e4 121952013-03-26 Richard Biener <rguenther@suse.de>
12196
ee9e7384 12197 * alias.c (find_base_term): Avoid redundant and not used recursion.
12198 (base_alias_check): Get the initial base term from the caller.
5abc99e4 12199 (true_dependence_1): Compute and pass base terms to base_alias_check.
12200 (write_dependence_p): Likewise.
12201 (may_alias_p): Likewise.
12202
df67a123 122032013-03-26 Sofiane Naci <sofiane.naci@arm.com>
12204
12205 * config/aarch64/aarch64.c (aarch64_classify_address): Support
12206 PC-relative load in SI modes and above only.
12207
9d4b6482 122082013-03-26 Xinyu Qi <xyqi@marvell.com>
12209
12210 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
8f028904 12211 * config/arm/iwmmxt.md (WCGR0): Update.
12212 (WCGR1, WCGR2, WCGR3): Likewise.
9d4b6482 12213
c8244d17 122142013-03-26 Uros Bizjak <ubizjak@gmail.com>
12215
12216 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
12217 Use x64 and nox64 isa attributes.
12218
9eec20bf 122192013-03-26 Richard Biener <rguenther@suse.de>
12220
12221 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
12222 alignment computations and rely on get_object_alignment_1
12223 for the !TYPE_P case.
12224 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
12225
684c0f06 122262013-03-26 Walter Lee <walt@tilera.com>
12227
12228 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
12229 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
12230
96a1fa3a 122312013-03-25 Jeff Law <law@redhat.com>
99db618c 12232
8f028904 12233 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
12234 check for INTEGRAL_TYPE_P that was missing due to checking in
96a1fa3a 12235 wrong version of prior patch.
12236
60cae6e1 122372013-03-25 Walter Lee <walt@tilera.com>
12238
12239 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
12240 TILEGX_INSN_SHUFFLEBYTES1.
12241 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
12242 shufflebytes1.
12243 (tilegx_builtins): Ditto.
12244 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
12245
a116add9 122462013-03-25 Walter Lee <walt@tilera.com>
12247
12248 * config/tilegx/tilegx.md (floatsisf2): New pattern.
60cae6e1 12249 (floatunssisf2): New pattern.
12250 (floatsidf2): New pattern.
12251 (floatunssidf2): New pattern.
a116add9 12252
19f27a0c 122532013-03-25 Walter Lee <walt@tilera.com>
12254
12255 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
12256 tests for constraint J, K, N, P.
12257
b429d641 122582013-03-25 Walter Lee <walt@tilera.com>
12259
12260 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
12261 Use indirect/pcrel encoding.
12262 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
12263 Ditto.
12264
14844884 122652013-03-25 Steve Ellcey <sellcey@mips.com>
12266
12267 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
12268 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
12269 * config/mips/mips.c (mips_option_override): Set IMADD default.
12270 * config/mips/mips.h (PTF_AVOID_IMADD): New.
12271 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
12272 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
12273 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
12274 * doc/invoke.texi (-mimadd/-mno-imadd): New.
12275
5aa24950 122762013-03-25 Jeff Law <law@redhat.com>
12277
12278 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
12279 slightly to avoid creating and folding useless trees. Simplify
12280 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
12281
c7222732 122822013-03-25 Uros Bizjak <ubizjak@gmail.com>
12283
12284 * config/i386/i386.md (*zero_extendsidi2): Merge with
12285 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
12286 * config/i386/predicates.md (x86_64_zext_operand): Rename from
12287 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
12288 targets. Clarify comment.
12289
30a240df 122902013-03-25 Martin Jambor <mjambor@suse.cz>
12291
12292 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
12293 pass-through jump functions differently.
12294 (ipa_read_jump_function): Likewise. Also use setter functions to set
12295 up jump functions.
12296
265c4eb2 122972013-03-25 Martin Jambor <mjambor@suse.cz>
12298
12299 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
12300 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
12301 process it.
12302 (ipa_get_indirect_edge_target): New function.
12303 (devirtualization_time_bonus): New parameter known_aggs, pass it to
12304 ipa_get_indirect_edge_target. Update all callers.
12305 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
12306 ipa_get_indirect_edge_target_1 instead of calling
12307 ipa_get_indirect_edge_target.
12308 (create_specialized_node): Pass aggvlas to
12309 ipcp_discover_new_direct_edges.
12310
91cb50d2 123112013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12312
12313 * config/arm/arm.md (f_sels, f_seld): New types.
12314 (*cmov<mode>): New pattern.
12315 * config/arm/predicates.md (arm_vsel_comparison_operator): New
12316 predicate.
12317
1ff5fe3b 123182013-03-25 Kai Tietz <ktietz@redhat.com>
12319
12320 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
12321 POSIX-printf for mingw-hosted builds.
12322
0565e0b5 123232013-03-25 Richard Biener <rguenther@suse.de>
12324
12325 PR middle-end/56694
12326 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
12327 must-not-throw stmt location.
12328
47d66364 123292013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12330
12331 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
12332 Emit load-acquire versions when acq is true.
12333 (arm_emit_store_exclusive): Add rel parameter.
12334 Emit store-release versions when rel is true.
12335 (arm_split_compare_and_swap): Use acquire-release instructions
12336 instead.
12337 of barriers when appropriate.
12338 (arm_split_atomic_op): Likewise.
12339 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
12340 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
12341 (VUNSPEC_SLX): Likewise.
12342 (VUNSPEC_LDA): Likewise.
12343 (VUNSPEC_STL): Likewise.
12344 * config/arm/sync.md (atomic_load<mode>): New pattern.
12345 (atomic_store<mode>): Likewise.
12346 (arm_load_acquire_exclusive<mode>): Likewise.
12347 (arm_load_acquire_exclusivesi): Likewise.
12348 (arm_load_acquire_exclusivedi): Likewise.
12349 (arm_store_release_exclusive<mode>): Likewise.
12350
087a99ba 123512013-03-25 Catherine Moore <clm@codesourcery.com>
12352
12353 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
12354 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
12355 * config/mip/predicates.md (lwsp_swsp_operand,
12356 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
12357 sb16_operand, db4_operand, db7_operand, ib3_operand,
12358 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
12359 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
12360 andi16_operand): New predicates.
12361 * config/mips/mips.md (compression): New attribute.
12362 (enabled): New attribute.
12363 (length): Consider compression in computing length.
12364 (shift_compression): New code attribute.
12365 (*add<mode>3): New operands. Record compression.
12366 (sub<mode>3): Likewise.
12367 (one_cmpl<mode>2): Likewise.
12368 (*and<mode>3): Likewise.
12369 (*ior<mode>3): Likewise.
12370 (unnamed pattern for xor): Likewise.
12371 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
12372 (*<optab><mode>3): Likewise.
12373 (*mov<mode>_internal: Likewise.
12374 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
12375 (mips_unsigned_immediate_p): New.
12376 (umips_lwsp_swsp_address_p): New.
12377 (m16_based_address_p): New.
12378 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
12379 (mips_unsigned_immediate_p): New prototype.
12380 (lwsp_swsp_address_p): New prototype.
12381 (m16_based_address_p): New prototype.
12382 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
12383 (mips_signed_immediate_p): New function.
12384 (m16_based_address_p): New function.
12385 (lwsp_swsp_address_p): New function.
12386 (mips_print_operand_punctuation): Recognize short delay slot insns
12387 for microMIPS.add<mode>3"
12388
2f17f336 123892013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9793b57b 12390
2f17f336 12391 PR target/56720
12392 * config/arm/iterators.md (v_cmp_result): New mode attribute.
12393 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
12394
d5459a15 123952013-03-25 Richard Biener <rguenther@suse.de>
12396
12397 PR tree-optimization/56689
12398 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
12399 any edge.
12400
feba4360 124012013-03-25 Richard Biener <rguenther@suse.de>
12402
12403 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
12404 of bitmap.
12405 (memory_references): Likewise.
12406 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
12407 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
12408 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
12409 (gather_mem_refs_in_loops): Fold into ...
12410 (analyze_memory_references): ... this. Move initialization
12411 to tree_ssa_lim_initialize.
12412 (fill_always_executed_in): Rename to ...
12413 (fill_always_executed_in_1): ... this.
12414 (fill_always_executed_in): Move contains_call computation to
12415 this new function from ...
12416 (tree_ssa_lim_initialize): ... here.
12417 (tree_ssa_lim): Call fill_always_executed_in.
12418
7777a939 124192013-03-25 Eric Botcazou <ebotcazou@adacore.com>
12420
12421 * postreload.c (reload_combine): Fix code detecting returns.
12422
96129913 124232013-03-25 Eric Botcazou <ebotcazou@adacore.com>
12424
12425 * function.c (emit_use_return_register_into_block): On cc0 targets,
12426 do not emit the sequence between cc0 setter and user.
12427
ed5a7307 124282013-03-25 Kai Tietz <ktietz@redhat.com>
12429
96129913 12430 * config/i386/predicates.md (local_symbolic_operand): Interpret
12431 dll-imported symbols as none-local.
ed5a7307 12432
a2c22c0d 124332013-03-25 Richard Biener <rguenther@suse.de>
12434
12435 * tree-ssa-loop-im.c (struct depend): Remove.
12436 (struct lim_aux_data): Make depends a vec of gimples.
12437 (free_lim_aux_data): Adjust.
12438 (add_dependency): Likewise.
12439 (set_level): Likewise.
12440
10836fcc 124412013-03-25 Richard Biener <rguenther@suse.de>
12442
12443 PR middle-end/56434
12444 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
12445 the pointer returned by calls with ECF_MALLOC set.
12446
1d61c5cb 124472013-03-24 Uros Bizjak <ubizjak@gmail.com>
12448
83b62e99 12449 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
12450
124512013-03-24 Uros Bizjak <ubizjak@gmail.com>
12452
12453 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
1d61c5cb 12454 using MMXMODE mode iterator.
12455 (*move<mode>_internal): Merge with *movv2sf_internal and
12456 *movv2sf_internal_rex64 using MMXMODE mode iterator.
12457
690973a8 124582013-03-23 Steven Bosscher <steven@gcc.gnu.org>
12459
37495d69 12460 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
12461 (record_last_mem_set_info): Likewise.
12462
690973a8 12463 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
12464 of XNEWVEC followed by memset.
12465 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
12466
aa90bb35 124672013-03-23 Steven Bosscher <steven@gcc.gnu.org>
12468
12469 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
12470 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
12471 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
12472 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
12473 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
12474 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
12475 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
12476 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
12477 BARRIER_P instead of GET_CODE.
12478
ada2d491 124792013-03-23 Eric Botcazou <ebotcazou@adacore.com>
12480
12481 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
12482 inaccuracy in the probing code.
12483
12484 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
12485 (ctrapdi4): Likewise.
12486
d8ef55fc 124872013-03-23 Eric Botcazou <ebotcazou@adacore.com>
12488
12489 * calls.c (expand_call): Add missing guard to code handling return
12490 of non-BLKmode structures in MSB.
12491 * function.c (expand_function_end): Likewise.
12492
32cf0798 124932013-03-23 Eric Botcazou <ebotcazou@adacore.com>
12494
12495 * combine.c (try_combine): Adjust comment. Do not add the set of
12496 insn #0 if the destination indirectly is set or dies in insn #2.
12497 Tidy up code to distribute a new note.
12498
dc624382 124992013-03-22 Uros Bizjak <ubizjak@gmail.com>
12500
12501 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
12502 also for alternatives 16 and 17.
12503
76cfb9fa 125042013-03-22 Uros Bizjak <ubizjak@gmail.com>
12505
12506 * config/i386/sse.md (*mov<mode>_internal): Merge with
12507 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
12508 Emit insn template depending on type attribute. Use
12509 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
12510 movd instead of movq mnemonic for interunit moves. Rewrite mode
12511 attribute calculation. Remove unit attribute calculation.
12512 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
12513 Set prefix_data16 attribute for DImode ssemov types.
12514 Use Ym instead of y for SSE-MMX conversion alternatives.
12515 Reorder operand constraints.
12516
f09c66e0 125172013-03-22 Steven Bosscher <steven@gcc.gnu.org>
12518
12519 * df.h (df_insn_delete): Adjust prototype.
12520 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
12521 and let it decide whether mark the basic block dirty.
12522 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
12523 * df-scan.c (df_insn_info_delete): New helper function, split
12524 off from df_insn_delete.
12525 (df_scan_free_bb_info): Use it.
12526 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
12527 Likewise.
12528 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
12529 that the insn is actually an insn and it has a non-NULL basic block.
12530 Do not mark basic block dirty if only deleting a DEBUG_INSN.
12531
c202575b 125322013-03-22 Richard Biener <rguenther@suse.de>
12533
12534 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
12535 dep_ref members.
12536 (mem_ref_alloc): Do not allocate them.
12537 (refs_independent_p): Do not query or maintain a cache.
12538
111767fb 125392013-03-22 Richard Biener <rguenther@suse.de>
12540
12541 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
12542 (gather_mem_refs_in_loops): Do not compute it.
12543 (analyze_memory_references): Do not allocate it.
12544 (tree_ssa_lim_finalize): Do not free it.
12545 (for_all_locs_in_loop): Do not query all_refs_in_loop.
12546
cd828e36 125472013-03-22 Richard Biener <rguenther@suse.de>
12548
12549 * is-a.h (as_a): Use gcc_checking_assert.
76cfb9fa 12550
c6ed7a1a 125512013-03-22 Ian Bolton <ian.bolton@arm.com>
12552
12553 * config/aarch64/aarch64.c (aarch64_print_operand): New
12554 format specifier for printing a constant in hex.
12555 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
12556 format specifier for printing second operand.
12557
8f9829e8 125582013-03-22 Richard Biener <rguenther@suse.de>
12559
12560 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
12561 bitmaps.
12562 (gather_mem_refs_in_loops): Perform store accumulation here.
12563 (create_vop_ref_mapping_loop): Remove.
12564 (create_vop_ref_mapping): Likewise.
12565 (analyze_memory_references): Initialize refs_stored_in_loop.
76cfb9fa 12566 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
8f9829e8 12567 (record_indep_loop): Remove.
12568 (record_dep_loop): New function.
12569 (ref_indep_loop_p_1): Adjust to only walk over references
12570 in the loop, not its subloops.
12571 (ref_indep_loop_p): Rename to ...
12572 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
12573 maintaining a more fine-grained cache.
12574 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
12575 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
12576
5c8a8be3 125772013-03-22 Richard Biener <rguenther@suse.de>
12578
12579 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
12580 (struct mem_ref): Make accesses_in_loop a vec of a vec of
12581 aggregate mem_ref_loc.
12582 (free_mem_ref_locs): Inline into ...
12583 (memref_free): ... this and adjust.
12584 (mem_ref_alloc): Adjust.
12585 (mem_ref_locs_alloc): Remove.
12586 (record_mem_ref_loc): Adjust.
12587 (get_all_locs_in_loop): Rewrite into ...
12588 (for_all_locs_in_loop): ... this iterator.
12589 (rewrite_mem_ref_loc): New functor.
12590 (rewrite_mem_refs): Use for_all_locs_in_loop.
12591 (sm_set_flag_if_changed): New functor.
12592 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
12593 (ref_always_accessed): New functor.
12594 (ref_always_accessed_p): Use for_all_locs_in_loop.
12595
7c3b431d 125962013-03-21 Marc Glisse <marc.glisse@inria.fr>
12597
12598 * tree-pass.h (PROP_gimple_lvec): New.
12599 * passes.c (dump_properties): Handle PROP_gimple_lvec.
12600 (init_optimization_passes): Move pass_lower_vector.
12601 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
12602 PROP_gimple_lvec.
12603 (pass_lower_vector): Provide PROP_gimple_lvec.
12604 (pass_lower_vector_ssa): Likewise.
12605 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
12606
0b436012 126072013-03-21 Mark Wielaard <mjw@redhat.com>
12608
12609 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
12610
5c38eac8 126112013-03-21 Uros Bizjak <ubizjak@gmail.com>
12612
12613 * config/i386/i386.md (*movdi_internal): Disparage slightly
12614 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
12615 conversion alternatives.
12616
2c8b23a2 126172013-03-21 Jakub Jelinek <jakub@redhat.com>
12618
d9d1b0a4 12619 PR middle-end/48087
12620 * diagnostic.def (DK_WERROR): New kind.
12621 * diagnostic.h (werrorcount): Define.
12622 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
12623 promoted to DK_ERROR, increment DK_WERROR counter instead of
12624 DK_ERROR counter.
12625 * toplev.c (toplev_main): Call print_ignored_options even if
12626 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
12627 even if just werrorcount is non-zero.
12628
2c8b23a2 12629 PR debug/55608
12630 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
12631 on failure.
12632 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
12633 (string_cst_pool_decl): New function.
12634 (optimize_one_addr_into_implicit_ptr): New function.
12635 (resolve_addr_in_expr): Optimize DWARF location expression
12636 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
12637 which doesn't live in memory, but has DW_AT_location or
12638 DW_AT_const_value, or refers to a string literal, into
12639 DW_OP_GNU_implicit_pointer.
12640 (optimize_location_into_implicit_ptr): New function.
12641 (resolve_addr): If removing DW_AT_location of a variable because
12642 it was DW_OP_addr of address of the variable, but the variable doesn't
12643 live in memory, try to emit const value attribute for the initializer.
12644
8ee286c8 126452013-03-21 Marc Glisse <marc.glisse@inria.fr>
12646
12647 * tree.h (VECTOR_TYPE_P): New macro.
12648 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
12649 TYPE_MODE): Use it.
12650 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
12651 VEC_COND_EXPR cannot be lvalues.
12652 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
12653
d09e949c 126542013-03-21 Marc Glisse <marc.glisse@inria.fr>
12655
12656 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
12657 Restrict the transformation to equal modes.
12658
04911505 126592013-03-21 Richard Biener <rguenther@suse.de>
12660
12661 PR tree-optimization/39326
12662 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
12663 (MEM_ANALYZABLE): Adjust.
12664 (record_mem_ref_loc): Move bitmap ops ...
12665 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
12666 unanalyzable refs, do not record locations for it.
12667 (analyze_memory_references): Allocate ref zero as shared
12668 unanalyzable ref.
12669 (refs_independent_p): Do not test for unanalyzed mems here.
12670 (ref_indep_loop_p_1): Special-case disambiguation against
12671 the unanalyzed ref.
5c38eac8 12672 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
04911505 12673
b6779ddc 126742013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
12675
12676 * config/arm/arm-protos.h (tune_params): Add
12677 prefer_neon_for_64bits field.
12678 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
12679 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
12680 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
12681 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
12682 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
12683 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
12684 (arm_option_override): Handle -mneon-for-64bits new option.
12685 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
12686 (prefer_neon_for_64bits): Declare new variable.
12687 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
5c38eac8 12688 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
b6779ddc 12689 (arch_enabled): Handle new arch types. Remove support for onlya8
12690 and nota8.
12691 (one_cmpldi2): Use new arch names.
a1bd99a8 12692 (zero_extend<mode>di2, extend<mode>di2): Ditto.
b6779ddc 12693 * config/arm/arm.opt (mneon-for-64bits): Add option.
12694 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
12695 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
12696 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
12697 of onlya8.
12698 * doc/invoke.texi (-mneon-for-64bits): Document.
12699
7894a3d9 127002013-03-21 Richard Biener <rguenther@suse.de>
12701
12702 PR tree-optimization/39326
12703 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
12704 (sort_bbs_in_loop_postorder_cmp): New function.
5c38eac8 12705 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
7894a3d9 12706
68f15e9d 127072013-03-21 Richard Biener <rguenther@suse.de>
12708
12709 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
12710 (vect_insert_into_interleaving_chain): Likewise.
12711 (vect_drs_dependent_in_basic_block): Inline ...
12712 (vect_slp_analyze_data_ref_dependence): ... here. New function,
12713 split out from ...
12714 (vect_analyze_data_ref_dependence): ... here. Simplify.
12715 (vect_check_interleaving): Simplify.
12716 (vect_analyze_data_ref_dependences): Likewise. Split out ...
12717 (vect_slp_analyze_data_ref_dependences): ... this new function.
12718 (dr_group_sort_cmp): New function.
12719 (vect_analyze_data_ref_accesses): Compute data-reference groups
12720 here instead of in vect_analyze_data_ref_dependence. Use
12721 a more efficient algorithm.
12722 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
12723 vect_slp_analyze_data_ref_dependences. Call
12724 vect_analyze_data_ref_accesses earlier.
12725 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12726 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
12727 (vect_slp_analyze_data_ref_dependences): New prototype.
12728
39f8afdf 127292013-03-21 Richard Biener <rguenther@suse.de>
12730
12731 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
12732 ref is stored in the loop.
12733 (find_refs_for_sm): Walk only over all stores.
12734 (store_motion_loop): Allocate from lim_bitmap_obstack.
12735 (store_motion): Likewise.
12736
f68f7ce8 127372013-03-21 Richard Biener <rguenther@suse.de>
12738
12739 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
12740 Update virtual SSA form.
12741
75d28287 127422013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12743
12744 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
12745 * configure: Regenerate.
12746 * config.in: Regenerate.
12747 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
12748 if !HAVE_LD_EH_FRAME_CIEV3.
12749
2428e53a 127502013-03-21 Richard Biener <rguenther@suse.de>
12751
12752 * tree-cfg.c (verify_expr_no_block): New function.
12753 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
12754 nor DECL_VALUE_EXPR have locations with associated blocks.
12755 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
12756 (clear_unused_block_pointer): Remove code dealing with
12757 blocks in DECL_DEBUG_EXPR locations.
12758
8e966116 127592013-03-21 Richard Biener <rguenther@suse.de>
12760
12761 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
12762 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
12763 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
12764 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
12765 instead of DECL_DEBUG_EXPR_IS_FROM.
12766 * gimplify.c (gimplify_modify_expr): Likewise.
12767 * tree-cfg.c (verify_expr_location_1): Likewise.
12768 * tree-complex.c (create_one_component_var): Likewise.
12769 * tree-sra.c (create_access_replacement): Likewise.
12770 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
12771 (clear_unused_block_pointer): Likewise.
12772 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
12773 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
12774 * var-tracking.c (var_debug_decl): Likewise.
12775 (track_expr_p): Likewise.
12776 * tree-inline.c (add_local_variables): Likewise. Set
12777 DECL_HAS_DEBUG_EXPR_P after copying it.
12778 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
12779 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
12780
9e2820ab 127812013-03-21 Uros Bizjak <ubizjak@gmail.com>
12782
12783 PR bootstrap/56656
12784 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
12785 * configure: Regenerate.
12786 * config.in: Regenerate.
12787 * config/i386/i386.md (*movdf_internal): Use
12788 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
12789 movd instead of movq mnemonic for interunit moves.
12790 (*movdi_internal): Ditto.
12791
d0360773 127922013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
02ab8c72 12793
12794 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
12795 (abd<mode>_3): New pattern.
12796 (aba<mode>_3): New pattern.
12797 (fabd<mode>_3): New pattern.
12798
d0360773 127992013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
1b457b76 12800
12801 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
12802 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
12803 occurrence of REGISTER_PREFIX as its empty string.
12804
799b7423 128052013-03-20 Jeff Law <law@redhat.com>
12806
12807 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
12808 addititional equivalences for equality comparisons between an SSA_NAME
12809 and a constant where the SSA_NAME was set from a widening conversion.
12810
4e38a95d 128112013-03-20 Walter Lee <walt@tilera.com>
12812
12813 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
12814
59ff41d8 128152013-03-20 Uros Bizjak <ubizjak@gmail.com>
12816
12817 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
12818 depending on type attribute.
12819 (*movti_internal): Ditto.
12820 (*movtf_internal): Ditto.
12821 (*movxf_internal): Ditto.
12822 (*movdf_internal): Ditto.
12823 (*movsf_internal): Ditto.
12824
7b536847 128252013-03-20 Uros Bizjak <ubizjak@gmail.com>
12826
12827 * config/i386/i386.md (*movti_internal): Set prefix attribute to
12828 maybe_vex for sselog1 and ssemov types.
12829 (*movdi_internal): Reorder operand constraints.
12830 (*movsi_internal): Ditto. Set prefix attribute to
12831 maybe_vex for sselog1 and ssemov types.
12832 (*movtf_internal): Set prefix attribute to maybe_vex
12833 for sselog1 and ssemov types.
12834 (*movdf_internal): Ditto. Set prefix_data16 attribute for
12835 DImode ssemov types. Reorder operand constraints.
12836 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
12837 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
12838 attribute for SImode ssemov types. Reorder operand constraints.
12839
3a1cb879 128402013-03-20 Martin Jambor <mjambor@suse.cz>
12841
12842 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
12843 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
12844
33a2d887 128452013-03-20 Pat Haugen <pthaugen@us.ibm.com>
12846
12847 * config/rs6000/predicates.md (indexed_address, update_address_mem
12848 update_indexed_address_mem): New predicates.
12849 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
12850 attribute for load/store instructions.
12851 * config/rs6000/dfp.md (movsd_store): Likewise.
12852 (movsd_load): Likewise.
12853 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
12854 (unnamed HI->DI extend define_insn): Likewise.
12855 (unnamed SI->DI extend define_insn): Likewise.
12856 (unnamed QI->SI extend define_insn): Likewise.
12857 (unnamed QI->HI extend define_insn): Likewise.
12858 (unnamed HI->SI extend define_insn): Likewise.
12859 (unnamed HI->SI extend define_insn): Likewise.
12860 (extendsfdf2_fpr): Likewise.
12861 (movsi_internal1): Likewise.
12862 (movsi_internal1_single): Likewise.
12863 (movhi_internal): Likewise.
12864 (movqi_internal): Likewise.
12865 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
12866 attribute for load/store instructions.
12867 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
12868 instructions.
12869 (mov<mode>_softfloat): Likewise.
12870 (mov<mode>_hardfloat32): Likewise.
12871 (mov<mode>_hardfloat64): Likewise.
12872 (mov<mode>_softfloat64): Likewise.
12873 (movdi_internal32): Likewise.
12874 (movdi_internal64): Likewise.
12875 (probe_stack_<mode>): Likewise.
12876
128772013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12878
12879 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
12880 floating point, and decimal floating point to reload iterator.
12881
12882 * config/rs6000/constraints.md (wl constraint): New constraints to
12883 return FLOAT_REGS if certain options are used to reduce the number
12884 of separate patterns that exist in the file.
12885 (wx constraint): Likewise.
12886 (wz constraint): Likewise.
12887
12888 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
12889 -mdebug=reg, print wg, wl, wx, and wz constraints.
12890 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
12891 Initialize the reload functions for 64-bit binary/decimal floating
12892 point types.
12893 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
12894 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
12895 create the buffer on the stack to overcome not having a 32-bit
12896 load and store.
12897 (rs6000_emit_move): Likewise.
12898 (rs6000_secondary_memory_needed_rtx): Likewise.
12899 (rs6000_alloc_sdmode_stack_slot): Likewise.
12900 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
12901 via xxlxor, just like DFmode 0.0.
12902
12903 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
12904 define as 1 if we are running on a power7 or newer.
12905 (enum r6000_reg_class_enum): Add new constraints.
12906
12907 * config/rs6000/dfp.md (movsd): Delete, combine with binary
12908 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
12909 with other moves by using conditional constraits (wg). Use LFIWZX
d0360773 12910 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
33a2d887 12911 (movsd splitter): Likewise.
12912 (movsd_hardfloat): Likewise.
12913 (movsd_softfloat): Likewise.
12914
12915 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
12916 binary and decimal floating point moves.
12917 (fmove_ok): New attributes to combine binary and decimal floating
12918 point moves, and to combine power6x (mfpgpr) moves along normal
12919 floating moves.
12920 (real_value_to_target): Likewise.
12921 (f32_lr): Likewise.
12922 (f32_lm): Likewise.
12923 (f32_li): Likewise.
12924 (f32_sr): Likewise.
12925 (f32_sm): Likewise.
12926 (f32_si): Likewise.
12927 (movsf): Combine binary and decimal floating point moves. Combine
12928 power6x (mfpgpr) moves with other moves by using conditional
d0360773 12929 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
33a2d887 12930 (mov<mode> for SFmode/SDmode); Likewise.
12931 (SFmode/SDmode splitters): Likewise.
12932 (movsf_hardfloat): Likewise.
12933 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
12934 (movsf_softfloat): Likewise.
12935 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
12936
12937 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
12938 wx and wz constraints.
12939
12940 * config/rs6000/constraints.md (wg constraint): New constraint to
12941 return FLOAT_REGS if -mmfpgpr (power6x) was used.
12942
12943 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
12944 constraint.
12945
12946 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
12947 -mdebug=reg, print wg, wl, wx, and wz constraints.
12948 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
12949 Initialize the reload functions for 64-bit binary/decimal floating
12950 point types.
12951 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
12952 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
12953 create the buffer on the stack to overcome not having a 32-bit
12954 load and store.
12955 (rs6000_emit_move): Likewise.
12956 (rs6000_secondary_memory_needed_rtx): Likewise.
12957 (rs6000_alloc_sdmode_stack_slot): Likewise.
12958 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
12959 via xxlxor, just like DFmode 0.0.
12960
33a2d887 12961 * config/rs6000/dfp.md (movdd): Delete, combine with binary
12962 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
12963 with other moves by using conditional constraits (wg). Use LFIWZX
12964 and STFIWX for loading SDmode on power7.
12965 (movdd splitters): Likewise.
12966 (movdd_hardfloat32): Likewise.
12967 (movdd_softfloat32): Likewise.
12968 (movdd_hardfloat64_mfpgpr): Likewise.
12969 (movdd_hardfloat64): Likewise.
12970 (movdd_softfloat64): Likewise.
12971
12972 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
12973 64-bit binary and decimal floating point moves.
12974 (FMOVE64X): Likewise.
12975 (movdf): Combine 64-bit binary and decimal floating point moves.
12976 Combine power6x (mfpgpr) moves with other moves by using
12977 conditional constraits (wg).
12978 (mov<mode> for DFmode/DDmode): Likewise.
12979 (DFmode/DDmode splitters): Likewise.
12980 (movdf_hardfloat32): Likewise.
12981 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
12982 (movdf_softfloat32): Likewise.
12983 (movdf_hardfloat64_mfpgpr): Likewise.
12984 (movdf_hardfloat64): Likewise.
12985 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
12986 (movdf_softfloat64): Likewise.
12987 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
12988 (reload_<mode>_load): Move to later in the file so they aren't in
12989 the middle of the floating point move insns.
12990 (reload_<mode>_store): Likewise.
12991
12992 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
12993 constraint.
12994
12995 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
12996 constraint if -mdebug=reg.
d0360773 12997 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
12998 Enable using dd reload support if needed.
33a2d887 12999
13000 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13001 binary and decimal floating point moves in rs6000.md.
13002 (movtd_internal): Likewise.
13003
13004 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13005 decimal floating point moves.
13006 (movtf): Likewise.
13007 (movtf_internal): Likewise.
13008 (mov<mode>_internal, TDmode/TFmode): Likewise.
13009 (movtf_softfloat): Likewise.
13010 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13011
13012 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13013 movdi_internal64, using wg constraint for move direct operations.
13014 (movdi_internal64): Likewise.
13015
13016 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13017 MODES_TIEABLE_P for selected modes. Print the numerical value of
13018 the various virtual registers. Use GPR/FPR first/last values,
13019 instead of hard coding the register numbers. Print which modes
13020 have reload functions registered.
d0360773 13021 (rs6000_option_override_internal): If -mdebug=reg, trace the options
13022 settings before/after setting cpu, target and subtarget settings.
13023 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13024 and for secondary reload failures in rs6000_secondary_reload_inner.
33a2d887 13025 (rs6000_secondary_reload_fail): Likewise.
13026 (rs6000_secondary_reload_inner): Likewise.
13027
13028 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13029 macros for first/last GPR and FPR registers.
13030 (LAST_GPR_REGNO): Likewise.
13031 (FIRST_FPR_REGNO): Likewise.
13032 (LAST_FPR_REGNO): Likewise.
13033
13034 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13035 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13036 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13037 (vcond<mode><mode>): Likewise.
13038 (vcondu<mode><mode>): Likewise.
13039 (vector_gtu<mode>): Likewise.
13040 (vector_gte<mode>): Likewise.
13041 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
d0360773 13042 to prevent the compiler from converting DImode operations to TImode.
33a2d887 13043 (ior<mode>3): Likewise.
13044 (and<mode>3): Likewise.
13045 (one_cmpl<mode>2): Likewise.
13046 (nor<mode>3): Likewise.
13047 (andc<mode>3): Likewise.
13048
13049 * config/rs6000/constraints.md (wt constraint): New constraint
13050 that returns VSX_REGS if TImode is allowed in VSX registers.
13051
13052 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13053 constant under VSX.
13054
13055 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13056 similar to TImode, but it is restricted to being in the GPRs.
13057
13058 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13059 TImode to occupy a single VSX register.
13060
13061 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13062 -mvsx-timode for power7/power8.
13063 (power7 cpu): Likewise.
13064 (power8 cpu): Likewise.
13065
13066 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13067 sure that TFmode/TDmode take up two registers if they are ever
13068 allowed in the upper VSX registers.
13069 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13070 registers.
13071 (rs6000_init_hard_regno_mode_ok): Likewise.
13072 (rs6000_debug_reg_global): Add debugging for PTImode and wt
13073 constraint. Print if LRA is turned on.
13074 (rs6000_option_override_internal): Give an error if -mvsx-timode
13075 and VSX is not enabled.
13076 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
13077 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13078 to reg+offset addressing. Use PTImode when checking offset
13079 addresses for validity.
13080 (reg_offset_addressing_ok_p): Likewise.
13081 (rs6000_legitimate_offset_address_p): Likewise.
13082 (rs6000_legitimize_address): Likewise.
13083 (rs6000_legitimize_reload_address): Likewise.
13084 (rs6000_legitimate_address_p): Likewise.
13085 (rs6000_eliminate_indexed_memrefs): Likewise.
13086 (rs6000_emit_move): Likewise.
13087 (rs6000_secondary_reload): Likewise.
13088 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
13089 reloads to fpr registers to continue to use reg+offset addressing,
13090 but 64-bit reloads to altivec registers need reg+reg addressing.
13091 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13092 it. Treat LO_SUM like a PLUS operation.
13093 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
d0360773 13094 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
33a2d887 13095 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13096 registers to share a register with a smaller sized type, since VSX
13097 puts scalars in the upper 64-bits.
13098 (print_operand): Add support for PTImode.
13099 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13100 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13101 registers, but don't have arithmetic support.
13102 (rs6000_memory_move_cost): Add test for VSX.
13103 (rs6000_opt_masks): Add -mvsx-timode.
13104
13105 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13106 for TImode.
13107 (VSs): Likewise.
13108 (VSr): Use wt constraint for TImode.
13109 (VSv): Drop TImode support.
13110 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13111 (vsx_movti_64bit): Likewise.
13112 (vsx_movti_32bit): Likewise.
13113 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13114 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13115 one '?' on the appropriate output constraint. Do not allow TImode
13116 logical operations on 32-bit systems.
13117 (vsx_ior<mode>3): Likewise.
13118 (vsx_xor<mode>3): Likewise.
13119 (vsx_one_cmpl<mode>2): Likewise.
13120 (vsx_nor<mode>3): Likewise.
13121 (vsx_andc<mode>3): Likewise.
13122 (vsx_concat_<mode>): Likewise.
13123 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13124
13125 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13126 OPTION_MASK_VSX_TIMODE.
13127 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13128 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13129
13130 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13131 (TI2 iterator): New iterator for TImode, PTImode.
13132 (wd mode attribute): Add values for vector types.
d0360773 13133 (movti_string): Replace TI move operations with operations for TImode
13134 and PTImode. Add support for TImode being allowed in VSX registers.
33a2d887 13135 (mov<mode>_string, TImode/PTImode): Likewise.
13136 (movti_ppc64): Likewise.
13137 (mov<mode>_ppc64, TImode/PTImode): Likewise.
13138 (TI mode splitters): Likewise.
13139
13140 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13141 constraint.
13142
a9073c2f 131432013-03-20 Marc Glisse <marc.glisse@inria.fr>
13144
13145 PR tree-optimization/56355
13146 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13147 Also handle integers with undefined overflow.
13148
ff9c1bc1 131492013-03-20 Catherine Moore <clm@codesourcery.com>
13150 Maciej W. Rozycki <macro@codesourcery.com>
13151 Tom de Vries <tom@codesourcery.com>
13152 Nathan Sidwell <nathan@codesourcery.com>
13153 Iain Sandoe <iain@codesourcery.com>
13154 Nathan Froyd <froydnj@codesourcery.com>
13155 Chao-ying Fu <fu@mips.com>
13156
13157 * doc/extend.texi: (micromips, nomicromips, nocompression):
7b536847 13158 Document new function attributes.
ff9c1bc1 13159 * doc/invoke.texi (minterlink-compressed, mmicromips,
13160 m14k, m14ke, m14kec): Document new options.
13161 (minterlink-mips16): Update documentation.
13162 * doc/md.texi (ZC, ZD): Document new constraints.
13163 * configure.ac (gcc_cv_as_micromips): Check if linker
13164 supports the .set micromips directive.
13165 * configure: Regenerate.
13166 * config.in: Regenerate.
13167 * config/mips/mips-tables.opt: Regenerate.
13168 * config/mips/micromips.md: New file.
13169 * constraints.md (ZC, ZD): New constraints.
13170 * config/mips/predicates.md (movep_src_register): New predicate.
13171 (movep_src_operand): New predicate.
13172 (non_volatile_mem_operand): New predicate.
13173 * config/mips/mips.md (multimem): New type.
13174 (length): Differentiate between 17-bit and 18-bit branch offsets.
13175 (MOVEP1, MOVEP2): New mode iterator.
8f028904 13176 (mov_<load>l): Use ZC constraint.
ff9c1bc1 13177 (mov_<load>r): Likewise.
13178 (mov_<store>l): Likewise.
13179 (mov_<store>r): Likewise.
13180 (*branch_equality<mode>_inverted): Add microMIPS support.
13181 (*branch_equality<mode>): Likewise.
13182 (*jump_absolute): Likewise.
13183 (indirect_jump_<mode>): Likewise.
13184 (tablejump_<mode>): Likewise.
13185 (<optab>_internal): Likewise.
13186 (sibcall_internal): Likewise.
13187 (sibcall_value_internal): Likewise.
13188 (prefetch): Use constraint ZD.
13189 * config/mips/mips.opt (minterlink-compressed): New option.
13190 (minterlink-mips16): Now an alias for minterlink-compressed.
13191 (mmicromips): New option.
13192 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13193 (compare_and_swap_12): Likewise.
13194 (sync_add<mode>): Likewise.
13195 (sync_<optab>_12): Likewise.
13196 (sync_old_<optab>_12): Likewise.
13197 (sync_new_<optab>_12): Likewise.
13198 (sync_nand_12): Likewise.
13199 (sync_old_nand_12): Likewise.
13200 (sync_new_nand_12): Likewise.
13201 (sync_sub<mode>): Likewise.
13202 (sync_old_add<mode>): Likewise.
13203 (sync_old_sub<mode>): Likewise.
13204 (sync_new_add<mode>): Likewise.
13205 (sync_new_sub<mode>): Likewise.
13206 (sync_<optab><mode>): Likewise.
13207 (sync_old_<optab><mode>): Likewise.
13208 (sync_new_<optab><mode>): Likewise.
13209 (sync_nand<mode>): Likewise.
13210 (sync_old_nand<mode>): Likewise.
13211 (sync_new_nand<mode>): Likewise.
13212 (sync_lock_test_and_set<mode>): Likewise.
13213 (test_and_set_12): Likewise.
13214 (atomic_compare_and_swap<mode>): Likewise.
13215 (atomic_exchange<mode>_llsc): Likewise.
13216 (atomic_fetch_add<mode>_llsc): Likewise.
13217 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
13218 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
13219 (umips_save_restore_pattern_p): Likewise.
13220 (umips_load_store_pair_p): Likewise.
13221 (umips_output_load_store_pair): Likewise.
13222 (umips_movep_target_p): Likewise.
13223 (umips_12bit_offset_address_p): Likewise.
13224 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
13225 (mips_base_mips16): Rename this...
13226 (mips_base_compression_flags): ...to this. Update all uses.
13227 (mips_attribute_table): Add micromips, nomicromips and nocompression.
13228 (mips_mips16_decl_p): Delete.
13229 (mips_nomips16_decl_p): Delete.
8f028904 13230 (mips_get_compress_on_flags): New function.
ff9c1bc1 13231 (mips_get_compress_off_flags): New function.
13232 (mips_get_compress_mode): New function.
13233 (mips_get_compress_on_name): New function.
13234 (mips_get_compress_off_name): New function.
13235 (mips_insert_attributes): Support multiple compression types.
13236 (mips_merge_decl_attributes): Likewise.
13237 (umips_12bit_offset_address_p): New function.
13238 (mips_start_function_definition): Emit .set micromips directive.
13239 (mips_call_may_need_jalx_p): New function.
13240 (mips_function_ok_for_sibcall): Add microMIPS support.
13241 (mips_print_operand_punctuation): Support short delay slots and
13242 compact jumps.
13243 (umips_swm_mask, umips_swm_encoding): New.
13244 (umips_build_save_restore): New function.
13245 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
13246 (was_mips16_p): Remove.
13247 (old_compression_mode): New.
13248 (mips_set_compression_mode): New function.
13249 (mips_set_current_function): Add microMIPS support.
13250 (mips_option_override): Likewise.
13251 (umips_save_restore_pattern_p): New function.
13252 (umips_output_save_restore): New function.
13253 (umips_load_store_pair_p_1): New function.
13254 (umips_load_store_pair_p): New function.
13255 (umips_output_load_store_pair_1): New function.
13256 (umips_output_load_store_pair): New function.
13257 (umips_movep_target_p) New function.
13258 (mips_prepare_pch_save): Add microMIPS support.
13259 * config/mips/mips.h (TARGET_COMPRESSION): New.
13260 (TARGET_CPU_CPP_BUILTINS): Update macro
13261 to use new compression flags and to support microMIPS.
13262 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
13263 (MIPS_ARCH_FLOAT_SPEC): Likewise.
13264 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
13265 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
13266 (ASM_SPEC): Support mmicromips and mno-micromips.
13267 (M16STORE_REG_P): New macro.
13268 (MIPS_CALL): Support TARGET_MICROMIPS.
13269 (MICROMIPS_J): New macro.
13270 (mips_base_mips16): Rename this...
13271 (mips_base_compression_flags): ...to this.
13272 (UMIPS_12BIT_OFFSET_P): New macro.
13273 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
13274 (MULTILIB_DIRNAMES): Likewise.
affbda71 132752013-03-20 Richard Biener <rguenther@suse.de>
13276
13277 PR tree-optimization/56661
13278 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
13279 the result does not have to be distinct.
13280
b94b4cf4 132812013-03-20 Richard Biener <rguenther@suse.de>
13282
13283 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
13284 remap_gimple_op_r.
13285
a6d46489 132862013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13287 Steven Bosscher <steven@gcc.gnu.org>
13288
13289 PR rtl-optimization/56605
13290 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
13291
76468b73 132922013-03-20 Uros Bizjak <ubizjak@gmail.com>
13293
13294 PR bootstrap/56656
13295 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
13296 that require movd instead of movq.
13297
5a950977 132982013-03-20 Richard Biener <rguenther@suse.de>
13299
13300 * tree-ssa-structalias.c (struct variable_info): Add pointer
13301 to the first field of an aggregate with sub-vars. Make
13302 this and the pointer to the next subfield its ID.
13303 (vi_next): New function.
13304 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
13305 storedanything_id, integer_id): Increment by one.
13306 (new_var_info, get_call_vi, lookup_call_clobber_vi,
13307 get_call_clobber_vi): Adjust.
13308 (solution_set_expand): Simplify and speedup.
13309 (solution_set_add): Inline into ...
13310 (set_union_with_increment): ... this. Adjust accordingly.
13311 (do_sd_constraint): Likewise.
13312 (do_ds_constraint): Likewise.
13313 (do_complex_constraint): Simplify.
13314 (build_pred_graph): Adjust.
13315 (solve_graph): Likewise. Simplify and speedup.
13316 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
13317 get_constraint_for_component_ref, get_constraint_for_1,
13318 first_vi_for_offset, first_or_preceding_vi_for_offset,
13319 create_function_info_for, create_variable_info_for_1,
13320 create_variable_info_for, intra_create_variable_infos): Adjust.
13321 (init_base_vars): Push NULL for ID zero.
13322 (compute_points_to_sets): Adjust.
13323
cd9916a9 133242013-03-20 Richard Biener <rguenther@suse.de>
13325
13326 * cfgloop.c (verify_loop_structure): Streamline and avoid
13327 ICEing on corrupt loop tree.
13328 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
13329 loop tree.
13330
3767bd95 133312013-03-20 Richard Biener <rguenther@suse.de>
13332
13333 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
13334 check whether an SSA update is needed.
13335
ac721588 133362013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
13337
13338 * config/mips/constraints.md (T): Rename to...
13339 (Yf): ...this.
13340 (U): Rename to...
13341 (Yd): ...this.
13342 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
13343 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
13344
00f5db85 133452013-03-19 Ian Bolton <ian.bolton@arm.com>
13346
13347 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
13348 (*subsi3_carryin_uxtw): Likewise.
13349
4d83c223 133502013-03-19 Ian Bolton <ian.bolton@arm.com>
13351
13352 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
13353 (*rorsi3_insn_uxtw): Likewise.
13354
45244f9b 133552013-03-19 Ian Bolton <ian.bolton@arm.com>
13356
13357 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
13358 (*extrsi5_insn_uxtw): Likewise.
13359
0df18bd2 133602013-03-19 Richard Biener <rguenther@suse.de>
13361
13362 PR tree-optimization/56273
13363 * passes.c (init_optimization_passes): Move second VRP after DOM.
13364
4c5ef8d5 133652013-03-19 Uros Bizjak <ubizjak@gmail.com>
13366
13367 * config/i386/i386.md (*movti_internal): Merge from
13368 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
13369 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
13370 nox64 isa attributes.
13371
6a5ca1fc 133722013-03-18 Richard Biener <rguenther@suse.de>
13373
13374 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
13375 (unite): Likewise.
13376 (merge_node_constraints): Likewise.
13377 (build_succ_graph): Likewise.
13378 (valid_graph_edge): Inline into single caller.
13379 (unify_nodes): Likewise. Use bitmap_set_bit return value
13380 and cache varinfo.
13381 (scc_visit): Fix formatting and variable use.
13382 (do_sd_constraint): Use gcc_checking_assert.
13383 (do_ds_constraint): Likewise.
13384 (do_complex_constraint): Likewise.
13385 (condense_visit): Likewise. Cleanup.
13386 (dump_pred_graph): New function.
13387 (perform_var_substitution): Dump the pred-graph before
13388 variable substitution.
13389 (find_equivalent_node): Use gcc_checking_assert.
13390 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
13391
90d4c4af 133922013-03-18 Richard Biener <rguenther@suse.de>
13393
13394 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
13395 Remove cond_expr_stmt_list argument and do not gimplify the
13396 built expression.
13397 (vect_loop_versioning): Adjust.
13398 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
13399 Cleanup to use less temporaries.
13400 (vect_create_data_ref_ptr): Cleanup.
13401
70347500 134022013-03-18 Jakub Jelinek <jakub@redhat.com>
13403
13404 PR tree-optimization/56635
13405 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
13406 require types_compatible_p types.
13407
23ecf105 134082013-03-18 Nick Clifton <nickc@redhat.com>
13409
c694a5cd 13410 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
13411 spurious backslash.
13412
23ecf105 13413 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
13414 Add missing line to comment describing function.
13415
4953672f 134162013-03-18 Richard Biener <rguenther@suse.de>
13417
13418 PR tree-optimization/56210
13419 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13420 Handle string / character search functions.
13421 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
13422
9532a315 134232013-03-18 Richard Biener <rguenther@suse.de>
13424
13425 PR middle-end/56483
13426 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
13427 and implement properly.
13428 * gimple.h (gimple_cond_single_var_p): Remove.
13429
07e3bcbf 134302013-03-18 Richard Biener <rguenther@suse.de>
13431
13432 * tree-data-ref.h (find_data_references_in_loop): Declare.
13433 * tree-data-ref.c (get_references_in_stmt): Use a stack
13434 vector pre-allocated in the callers.
13435 (find_data_references_in_stmt): Adjust.
13436 (graphite_find_data_references_in_stmt): Likewise.
13437 (create_rdg_vertices): Likewise.
13438 (find_data_references_in_loop): Export.
13439 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13440 Compute dependences here...
13441 (vect_analyze_data_refs): ...not here. When we encounter
13442 a non-vectorizable data reference in basic-block vectorization
13443 truncate the data reference vector. Do not bother to
13444 fixup data-dependence information for gather loads.
13445 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
13446 of data references, as reported.
13447
cc99c1e1 134482013-03-18 Richard Biener <rguenther@suse.de>
13449
13450 PR tree-optimization/3713
13451 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
13452 has_constants and expr.
13453 (stmt_has_constants): Properly valueize SSA names when deciding
13454 whether the stmt has constants.
13455
f55f91f5 134562013-03-18 Richard Biener <rguenther@suse.de>
13457
13458 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
13459 whole function when there is nothing to do.
13460 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
13461 * tree-vectorizer.c (vectorize_loops): Update virtual and
13462 loop-closed SSA once.
13463 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
13464
0da53361 134652013-03-18 Richard Biener <rguenther@suse.de>
13466
13467 PR middle-end/56113
13468 * domwalk.c (bb_postorder): New global static.
13469 (cmp_bb_postorder): New function.
13470 (walk_dominator_tree): Replace scheme imposing an order for
13471 visiting dominator sons by one sorting them at the time they
13472 are pushed on the stack.
13473
09f4cf62 134742013-03-18 Richard Biener <rguenther@suse.de>
13475
13476 PR tree-optimization/39326
13477 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
13478 (struct mem_ref): Replace mem member with ao_ref typed member.
13479 (MEM_ANALYZABLE): Adjust.
13480 (memref_eq): Likewise.
13481 (mem_ref_alloc): Likewise.
13482 (gather_mem_refs_stmt): Likewise.
13483 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
13484 (execute_sm_if_changed_flag_set): Adjust.
13485 (execute_sm): Likewise.
13486 (ref_always_accessed_p): Likewise.
13487 (refs_independent_p): Likewise.
13488 (can_sm_ref_p): Likewise.
13489
800469aa 134902013-03-18 Jakub Jelinek <jakub@redhat.com>
13491
13492 PR c/56566
13493 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
13494 return 1 even for !unsignedp.
13495
099188cb 134962013-03-17 Uros Bizjak <ubizjak@gmail.com>
13497
13498 * config/i386/i386.md (isa): Add x64 and nox64.
13499 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
13500 (*pushtf): Enable *roF alternative for x64 isa only.
13501 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
13502 mode attribute of integer alternatives to DImode for TARGET_64BIT.
13503 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
13504 (*movtf_internal): Merge from *movtf_internal_rex64 and
13505 *movtf_internal_sse. Use x64 and nox64 isa attributes.
13506 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
13507 nox64 isa attributes.
13508 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
13509 nox64 isa attributes.
13510 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
13511
862c63eb 135122013-03-17 Uros Bizjak <ubizjak@gmail.com>
13513
13514 * config/alpha/alpha.c (TARGET_LRA_P): New define.
13515
595169d8 135162013-03-17 Jakub Jelinek <jakub@redhat.com>
13517
13518 PR target/56640
13519 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
13520 class names. Remove trailing comma after "ALL_REGS".
13521
7d52f1a3 135222013-03-16 Jan Hubicka <jh@suse.cz>
13523
13524 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
13525 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
13526 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
13527 of cgraph_get_create_node.
13528 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13529
48adf3f3 135302013-03-16 Jason Merrill <jason@redhat.com>
13531
13532 PR debug/49090
13533 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
13534 with DW_AT_default_value.
13535
67f9d821 135362013-03-16 Jakub Jelinek <jakub@redhat.com>
13537
13538 * BASE-VER: Set to 4.9.0.
13539
34e2916a 135402013-03-14 Andi Kleen <ak@linux.intel.com>
13541
13542 PR target/56619
13543 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
13544 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
13545 Document _x* TSX intrinsics.
13546
5e5f9379 135472013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
13548 David Holsgrove <david.holsgrove@xilinx.com>
4eae25aa 13549
13550 * configure.ac: Add MicroBlaze TLS support detection.
13551 * configure: Regenerate.
862c63eb 13552 * config/microblaze/microblaze-protos.h
13553 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
13554 symbol_mentioned_p, label_mentioned_p): Add prototypes.
13555 * config/microblaze/microblaze.c (microblaze_address_type): Add
13556 ADDRESS_TLS and tls_reloc address types.
4eae25aa 13557 (microblaze_address_info): Add tls_reloc.
13558 (TARGET_HAVE_TLS): Define.
13559 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
13560 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
862c63eb 13561 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
13562 load_tls_operand, microblaze_call_tls_get_addr,
13563 microblaze_legitimize_tls_address): New functions.
4eae25aa 13564 (microblaze_classify_unspec): Handle UNSPEC_TLS.
13565 (get_base_reg): Use microblaze_tls_symbol_p.
13566 (microblaze_classify_address): Handle TLS.
862c63eb 13567 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
13568 label_mentioned_p and microblaze_tls_referenced_p.
4eae25aa 13569 (microblaze_legitimize_address): Handle TLS.
13570 (microblaze_address_insns): Handle ADDRESS_TLS.
13571 (pic_address_needs_scratch): Handle TLS.
13572 (print_operand_address): Handle TLS.
13573 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
13574 (microblaze_expand_move): Handle TLS.
862c63eb 13575 (microblaze_legitimate_constant_p): Check
13576 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
4eae25aa 13577 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
13578 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
13579 (PIC_OFFSET_TABLE_REGNUM): Set.
13580 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
13581 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
13582 (addsi3, movsi_internal2, movdf_internal): Update constraints
13583 * config/microblaze/predicates.md (arith_plus_operand): Define
862c63eb 13584 (move_operand): Redefine as move_src_operand,
13585 check microblaze_tls_referenced_p.
4eae25aa 13586
341c5ffb 135872013-03-14 Ian Bolton <ian.bolton@arm.com>
13588
13589 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
13590 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
13591
34cf9362 135922013-03-14 Ian Bolton <ian.bolton@arm.com>
13593
13594 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
13595 CC mode for AND.
13596
fc162120 135972013-03-14 Jakub Jelinek <jakub@redhat.com>
13598
228bf2b8 13599 PR tree-optimization/53265
13600 * common.opt (Waggressive-loop-optimizations): New option.
13601 * tree-ssa-loop-niter.c: Include tree-pass.h.
13602 (do_warn_aggressive_loop_optimizations): New function.
13603 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
13604 if number_of_latch_executions returned constant.
13605 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
13606 early. If number_of_latch_executions returned constant, set
13607 nb_iterations_upper_bound back to it.
13608 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
13609 field.
13610 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
13611 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
13612
fc162120 13613 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
13614 (MULTILIB_OSDIRNAMES): Set.
13615 * genmultilib: If defaultosdirname doesn't start with :: , set
13616 defaultosdirname2 instead, clear it and emit two . multilib_raw
13617 entries instead of just one.
13618
ae979f32 136192013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
13620
13621 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
13622 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
13623 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
13624 (SUBTARGET_OVERRIDE_OPTIONS): New.
13625
bb057878 136262013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
13627
13628 PR target/49880
13629 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
13630 (musermode): Convert to Var(TARGET_USERMODE).
13631 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
13632 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
13633 * config/sh/sh.c (sh_option_override): Use
13634 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
13635 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
13636 condition.
13637 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
13638 TARGET_SH4.
13639 (udivsi3_i4_single, divsi3_i4_single): Use
13640 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
13641
cfdddc4b 136422013-03-13 Dave Korn <dave.korn.cygwin@....>
13643
13644 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
13645 default setting.
13646
814c4f66 136472013-03-13 Richard Biener <rguenther@suse.de>
13648
13649 PR tree-optimization/56608
13650 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
13651 calls when vectorizing basic-blocks.
13652
043cabe3 136532013-03-13 Jakub Jelinek <jakub@redhat.com>
13654
13655 PR plugins/45078
13656 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
13657 tm_file.
13658
1e93bc00 136592013-03-12 Jakub Jelinek <jakub@redhat.com>
13660
13661 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
13662
344cbf0a 136632013-03-11 Jan Hubicka <jh@suse.cz>
13664
13665 PR lto/56557
13666 * lto-streamer-out.c (output_symbol_p): Skip references from
13667 constructors of external variables.
13668
b291008a 136692013-03-11 Jan Hubicka <jh@suse.cz>
13670
13671 PR middle-end/56571
13672 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
13673 from pseudos.
13674 * emit-rtl.c (verify_rtx_sharing): Likewise.
13675 (copy_insn_1): Likewise.
13676 * rtl.c (copy_rtx): Likewise.
13677
803e7ca1 136782013-03-11 Georg-Johann Lay <avr@gjlay.de>
13679
13680 PR target/56591
13681 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
13682 output_operand_lossage message.
862c63eb 13683
f8825f5d 136842013-03-11 Richard Earnshaw <rearnsha@arm.com>
13685
13686 PR target/56470
13687 * arm.c (shift_op): Validate RTL pattern on the fly.
13688 (arm_print_operand, case 'S'): Don't use shift_operator to validate
13689 the RTL.
13690
a1fcab4c 136912013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
13692
13693 PR target/56347
13694 * config/pa/pa.md (call_value): Check for calls to powf and direct to
13695 new call patterns that clobber %fr12.
13696 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
13697 split and postreload patterns.
13698 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
13699 registers %fr12 and %fr12R as call used.
13700
1ca59310 137012013-03-09 Steven Bosscher <steven@gcc.gnu.org>
13702
13703 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
13704 (canon_address, record_store, replace_read, check_mem_read_rtx,
13705 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
13706 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
13707 rest_of_handle_dse): Likewise.
13708
9d3fa937 137092013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
13710
13711 PR middle-end/56524
13712 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
13713 Add base_optabs.
13714 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
13715 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
13716 (save_optabs_if_changed): Replace with...
13717 (init_tree_optimization_optabs): ...this.
13718 * optabs.c (save_optabs_if_changed): Rename to...
13719 (init_tree_optimization_optabs): ...this. Take the optimization node
13720 as argument. Do nothing if the base optabs are already correct.
13721 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
13722 to recompute optabs.
13723 * function.h (function): Remove optabs field.
13724 * function.c (invoke_set_current_function_hook): Call
13725 init_tree_optimization_optabs. Use the result to initialize
13726 this_fn_optabs.
13727
1910089e 137282013-02-27 Aldy Hernandez <aldyh@redhat.com>
13729
13730 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
13731 if GTMA_HAS_NO_INSTRUMENTATION.
13732 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
13733 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
13734 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
13735 * gimple-pretty-print.c (dump_gimple_transaction): Handle
13736 GTMA_HAS_NO_INSTRUMENTATION.
13737
62efcea5 137382013-03-08 Jakub Jelinek <jakub@redhat.com>
13739
13740 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
13741 libasan_preinit.o.
13742
b3269f54 137432013-03-08 Marek Polacek <polacek@redhat.com>
13744 Jakub Jelinek <jakub@redhat.com>
13745
13746 PR tree-optimization/56478
13747 * predict.c (is_comparison_with_loop_invariant_p): Change the
13748 type of loop_step to tree.
13749 (predict_loops): Adjust.
13750 (predict_iv_comparison): Perform the computations on double_ints.
13751
4c04afc7 137522013-03-08 Richard Biener <rguenther@suse.de>
13753
13754 PR tree-optimization/56570
13755 * tree-cfg.c (verify_expr_location_1): Verify locations for
13756 DECL_DEBUG_EXPR.
13757 * tree-sra.c (create_access_replacement): Strip locations
13758 from DECL_DEBUG_EXPRs.
13759
dc426203 137602013-03-08 Richard Biener <rguenther@suse.de>
13761
13762 * tree-inline.c (expand_call_inline): Do not associate
13763 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
13764 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
13765
58fd1738 137662013-03-08 Richard Biener <rguenther@suse.de>
13767
13768 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
13769 or block changes with -Og. Fix for location / block encoding
13770 changes and PHI arguments with locations.
13771
28d3effc 137722013-03-07 Steven Bosscher <steven@gcc.gnu.org>
13773
13774 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
13775 for all counters.
13776 (struct output_info): Likewise.
13777 (register_overhead): Remove bad gcc_assert.
13778 (bitmap_find_bit): If there is only a single bitmap element, do not
13779 count a miss as a search.
13780 (print_statistics): Update for counter type changes.
13781 (dump_bitmap_statistics): Likewise. Print headers such that they
13782 are properly lined up with the printed counters.
13783
bdb5a35a 137842013-03-07 Jakub Jelinek <jakub@redhat.com>
13785
13786 PR tree-optimization/56559
13787 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
13788 check that it has only a single use.
13789
f7961441 137902013-03-07 Richard Biener <rguenther@suse.de>
13791
13792 * doc/invoke.texi (fwhole-program): Discourage use in combination
13793 with -flto.
13794
31c66114 137952013-03-06 Jakub Jelinek <jakub@redhat.com>
13796
23910ff9 13797 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
13798
f5c4bbcd 13799 PR tree-optimization/56539
13800 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
13801 instead of GSI_CONTINUE_LINKING as last argument to
13802 force_gimple_operand_gsi. Adjust function comment.
13803
48ff81f3 13804 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
13805 aarch64-cores.def.
13806
31c66114 13807 PR middle-end/56548
13808 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
13809 promoted mode, convert the result back to the original mode.
13810
487c79c3 138112013-03-06 Richard Biener <rguenther@suse.de>
13812
13813 PR middle-end/56294
13814 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
13815 (insert_updated_phi_nodes_compare_uids): New function.
13816 (update_ssa): Sort symbols_to_rename after UID before
13817 traversing it to insert PHI nodes.
13818
90a674eb 138192013-03-06 Richard Biener <rguenther@suse.de>
13820
13821 PR middle-end/50494
13822 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
13823 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
13824
13825 Revert
13826 2013-02-13 Richard Biener <rguenther@suse.de>
13827
13828 PR lto/50494
13829 * varasm.c (output_constant_def_1): Get the decl representing
13830 the constant as argument.
13831 (output_constant_def): Wrap output_constant_def_1.
13832 (make_decl_rtl): Use output_constant_def_1 with the decl
13833 representing the constant.
13834 (build_constant_desc): Optionally re-use a decl already
13835 representing the constant.
13836 (tree_output_constant_def): Adjust.
13837
40c059b1 138382013-03-06 Joey Ye <joey.ye@arm.com>
13839
13840 PR lto/50293
13841 * gcc.c (convert_white_space): New function.
13842 (main): Handles white space in function name.
13843
9fe603c3 138442013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
13845
13846 PR target/56529
13847 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
13848 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
13849 to SH_DIV_CALL_TABLE for TARGET_SH2.
13850 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
13851 list.
13852 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
13853 call-table options.
13854
24a5ae15 138552013-03-05 Sterling Augustine <saugustine@google.com>
13856 Cary Coutant <ccoutant@google.com>
13857
13858 PR debug/55364
13859 * dwarf2out.c (resolve_addr): Don't call
13860 remove_loc_list_addr_table_entries a second time for the same
13861 expression.
13862
848d0536 138632013-03-05 Jakub Jelinek <jakub@redhat.com>
d9c1b601 13864
848d0536 13865 PR debug/56510
13866 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
13867 (avoid_complex_debug_insns): New function.
13868 (expand_debug_locations): Call it.
13869
13870 PR rtl-optimization/56484
13871 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
13872 lifetimes of hard registers on small register class machines.
13873
138742013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
13875
13876 * config/microblaze/microblaze-protos.h: Rename
d9c1b601 13877 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
848d0536 13878 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
d9c1b601 13879 fast_interrupt.
13880 (microblaze_fast_interrupt_function_p): New function.
13881 (microblaze_is_interrupt_handler): Rename to
13882 microblaze_is_interrupt_variant and add fast_interrupt check.
13883 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
13884 (save_restore_insns): Likewise.
13885 (compute_frame_size): Likewise.
13886 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
13887 (microblaze_globalize_label): Likewise.
848d0536 13888 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
13889 * config/microblaze/microblaze.md: Use wrapper
d9c1b601 13890 microblaze_is_interrupt_variant.
13891
d641a4ab 138922013-03-05 Kai Tietz <ktietz@redhat.com>
13893
13894 * sdbout.c (sdbout_one_type): Switch to current function's section
13895 supporting cold/hot.
13896
31c66114 138972013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
76cc869a 13898
13899 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
13900 -mxl-reorder.
13901
a776325f 139022013-03-05 Jakub Jelinek <jakub@redhat.com>
13903
5ceebb21 13904 PR middle-end/56461
13905 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
13906 if VALGRIND_GET_VBITS is defined, temporarily make object
13907 memory all defined, and restore previous valgrind addressability
13908 and definability afterwards. Free this_object at the end.
13909
d3d0b390 13910 PR middle-end/56461
13911 * lra.c (lra): Call lra_clear_live_ranges if live_p,
13912 right before calling lra_create_live_ranges, also call it
13913 when clearing live_p. Only call lra_clear_live_ranges
13914 at the end if live_p.
13915
a776325f 13916 PR middle-end/56461
13917 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
13918
63628665 139192013-03-05 Richard Biener <rguenther@suse.de>
13920
13921 PR tree-optimization/56521
13922 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
13923 value-id.
13924
390f4a4b 139252013-03-05 Steven Bosscher <steven@gcc.gnu.org>
13926
13927 PR c++/55135
13928 * except.h (remove_unreachable_eh_regions): New prototype.
13929 * except.c (remove_eh_handler_splicer): New function, split out
13930 of remove_eh_handler.
13931 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
13932 warning about running it on many EH regions one at a time.
13933 (remove_unreachable_eh_regions_worker): New function, walk the
13934 EH tree in depth-first order and remove non-marked regions.
13935 (remove_unreachable_eh_regions): New function.
13936 * tree-eh.c (mark_reachable_handlers): New function, split out
13937 from remove_unreachable_handlers.
13938 (remove_unreachable_handlers): Use mark_reachable_handlers and
13939 remove_unreachable_eh_regions.
13940 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
13941 and remove_unreachable_eh_regions.
13942
84087109 139432013-03-05 Richard Biener <rguenther@suse.de>
13944
13945 PR middle-end/56525
13946 * loop-init.c (fix_loop_structure): Remove loops in two stages,
13947 not freeing them until the end.
13948
a22cd2a7 139492013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
13950
13951 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
13952
6de71799 139532013-03-05 Richard Biener <rguenther@suse.de>
13954
13955 PR tree-optimization/56270
13956 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
13957 of loads after scheduling an SLP instance.
13958
6ae8a044 139592013-03-05 Jakub Jelinek <jakub@redhat.com>
13960
94ce8195 13961 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
13962 tic6x.exp.
13963 (check_gcc_parallelize): Run guality.exp as a separate job from
13964 vect.exp with unsorted.exp and $(dg_target_exps) separately from
13965 struct-layout-1.exp with stackalign.exp.
13966
7a2f311c 13967 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
13968
cae603e9 13969 PR middle-end/56461
13970 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
13971 load_index sbitmap even if some bit in it isn't set.
13972
b71850a1 13973 PR middle-end/56461
13974 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
13975 (discover_iteration_bound_by_body_walk): Change queues to
13976 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
13977 spelling in comment. Call safe_push on queues[bound_index] directly.
13978 Release queues[queue_index] in every iteration unconditionally.
13979 Release bounds vector.
13980
6ae8a044 13981 PR middle-end/56461
13982 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
13983 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
13984 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
13985 inner_phis vector.
13986
c251ba7f 139872013-03-05 Richard Biener <rguenther@suse.de>
13988
13989 PR lto/56515
13990 * tree-inline.c (remap_blocks_to_null): New function.
13991 (expand_call_inline): When expanding a call stmt without
13992 an associated block inline remap all callee blocks to NULL.
13993
70639a9d 139942013-03-05 Jakub Jelinek <jakub@redhat.com>
13995
bdcf717b 13996 PR rtl-optimization/56494
13997 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
13998 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
13999 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14000
862c1934 14001 PR middle-end/56461
14002 * sel-sched-ir.c (free_sched_pools): Release
14003 succs_info_pool.stack[succs_info_pool.max_top] vectors too
14004 if succs_info_pool.max_top isn't -1.
14005
70639a9d 14006 PR bootstrap/56509
14007 * opts.c (opts_obstack, opts_concat): Moved to...
14008 * opts-common.c (opts_obstack, opts_concat): ... here.
14009
29d52322 140102013-03-04 Jakub Jelinek <jakub@redhat.com>
14011
14012 PR middle-end/56461
14013 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14014
a1a00916 140152013-03-04 Martin Jambor <mjambor@suse.cz>
14016
14017 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14018 all appropriate places.
14019
8ad1eb69 140202013-01-04 Eric Botcazou <ebotcazou@adacore.com>
14021
14022 PR tree-optimization/56424
14023 * ipa-split.c (split_function): Do not set the RSO flag if result is
14024 not by reference and its type is a register type.
14025
31c66114 140262013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
4f12cc20 14027
862c63eb 14028 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
4f12cc20 14029 (microblaze_legitimate_pic_operand): Likewise
862c63eb 14030 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
4f12cc20 14031 new function microblaze_legitimate_pic_operand
14032 * config/microblaze/microblaze-protos.h
14033 (microblaze_legitimate_pic_operand): Declare.
14034
31c66114 140352013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
ff86513b 14036
31c66114 14037 * config/microblaze/predicates.md (call_insn_simple_operand):
ff86513b 14038 New predicate for supported rtx code types.
31c66114 14039 * config/microblaze/microblaze.md (call_internal1): Use
ff86513b 14040 call_insn_simple_operand predicate.
14041
033ee56d 140422013-03-04 Jakub Jelinek <jakub@redhat.com>
14043
b3df3de1 14044 PR middle-end/56461
14045 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14046 partitions.ordered_remove.
14047
1d4dcacf 14048 PR middle-end/56461
14049 * tree-vect-stmts.c (vectorizable_conversion): Don't call
14050 vec_oprnds0.create (1) for modifier == NONE.
14051
87c0beb4 14052 PR middle-end/56461
14053 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14054 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14055 vec_oprnds1 right before pushing anything to it for
14056 scalar_shift_arg.
14057
033ee56d 14058 PR middle-end/56461
14059 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14060 set nbbs to 0 instead of having separate code path.
14061 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14062 instead of false as last argument if returning NULL.
14063
090eadfb 140642013-03-03 Sandra Loosemore <sandra@codesourcery.com>
14065
14066 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14067 the attribute is now called "target" instead of "option".
14068 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14069 * doc/tm.texi.in (Target Attributes): Likewise document the correct
14070 attribute/pragma name for TARGET_OPTION_VALID_P and
14071 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
14072 * doc/tm.texi: Regenerated.
14073
87c0beb4 140742013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
db38a061 14075
87c0beb4 14076 * config/microblaze/microblaze.c:
db38a061 14077 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
87c0beb4 14078 * config/microblaze/microblaze.h: Add -mxl-reorder to
14079 DRIVER_SELF_SPECS.
14080 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14081 instructions emitted if TARGET_REORDER.
862c63eb 14082 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14083 or 0 for -m/-mno case, but initialises as 2 to detect default use case
87c0beb4 14084 separately.
db38a061 14085
ae814dd3 140862013-03-01 Xinliang David Li <davidxl@google.com>
14087
14088 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14089 walk length.
14090
57fad2bf 140912013-03-01 Jakub Jelinek <jakub@redhat.com>
14092
04437ab6 14093 PR middle-end/56461
14094 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14095 vector even when returning true. Fix up function comment formatting.
14096
f9feeefd 14097 PR middle-end/56461
14098 * ira-build.c (ira_loop_nodes_count): New variable.
14099 (create_loop_tree_nodes): Initialize it.
14100 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14101
f40aaf2d 14102 PR middle-end/56461
14103 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14104 method on dr_chain and result_chain.
14105 * tree-vect-stmts.c (vectorizable_store): Only call
14106 result_chain.create if j == 0.
14107
57fad2bf 14108 PR middle-end/56461
14109 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14110 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14111 before overwriting it.
14112
32ddc0d1 141132013-03-01 Tobias Burnus <burnus@net-b.de>
14114
14115 * doc/extended.texi (C Extensions): Change order in @menu
14116 to match @node.
14117 (Other MIPS Built-in Functions): Move last MIPS entry before
14118 "picoChip Built-in Functions".
14119 (SH Built-in Functions): Move after RX Built-in Functions.
14120 * doc/gcc.texi (Introduction): Change order in @menu
14121 to match @node.
14122 * doc/md.texi (Constraints): Ditto.
14123 * gty.texi (Type Information): Ditto.
14124 (User-provided marking routines for template types): Make
14125 subsection.
14126 * doc/invoke.texi (AArch64 Options): Move before
14127 "Adapteva Epiphany Options".
14128
a9196da9 141292013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
14130 Jakub Jelinek <jakub@redhat.com>
14131
14132 PR sanitizer/56454
14133 * asan.c (gate_asan): Lookup no_sanitize_address instead of
14134 no_address_safety_analysis attribute.
14135 * doc/extend.texi (no_address_safety_attribute): Rename to
14136 no_sanitize_address attribute, mention no_address_safety_analysis
14137 attribute as deprecated alias.
14138
7f7695a7 141392013-02-28 Jakub Jelinek <jakub@redhat.com>
14140
14141 PR middle-end/56461
14142 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14143 type to vec<vec<tree> > *.
14144 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
14145 to be vec<tree> instead of vec<tree> *, set vec_defs
14146 to vNULL and call vec_defs.create (number_of_vects), adjust other
14147 uses of vec_defs.
14148 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14149 vectorizable_condition): Adjust vect_get_slp_defs callers.
14150
9439bdfe 141512013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14152
14153 * config/aarch64/aarch64.c
14154 (aarch64_float_const_representable): Remove unused variable.
14155
147f0a69 141562013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14157
14158 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14159
4f164a9f 141602013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14161
14162 * config/aarch64/aarch64-builtins.c
14163 (aarch64_init_simd_builtins): Make static.
14164
6d57cecd 141652013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14166
14167 * config/aarch64/aarch64.c
14168 (aarch64_simd_make_constant): Make static.
14169
e3fa6175 141702013-02-28 Martin Jambor <mjambor@suse.cz>
14171
14172 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14173 with no initialization to the RHS of debug statements.
14174
2e1d16b0 141752013-02-28 Martin Jambor <mjambor@suse.cz>
14176
14177 PR tree-optimization/56294
14178 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14179 Adjust dumping.
14180 (get_access_replacement): Do not call create_access_replacement.
14181 Assert a replacement exists.
14182 (get_repl_default_def_ssa_name): Create the replacement declaration
14183 itself.
14184
2c8489a0 141852013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14186
14187 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14188 final_end_function.
14189
f3721e91 141902013-02-28 Marek Polacek <polacek@redhat.com>
14191
14192 PR rtl-optimization/56466
14193 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14194 if we're changing a loop.
14195 (peel_loops_completely): Likewise.
14196
a8877909 141972013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
14198
14199 PR c++/55813
14200 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14201
3e88585b 142022013-02-28 Georg-Johann Lay <avr@gjlay.de>
14203
14204 PR target/56445
14205 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14206 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14207 INTX_FTYPE_FX, FX_FTYPE_INTX.
14208 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14209
724fcbc3 142102013-02-28 Georg-Johann Lay <avr@gjlay.de>
14211
14212 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14213 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14214 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
14215 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
14216 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
14217 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
14218 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
14219 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
14220 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
14221 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
14222 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
14223 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
14224 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
14225 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
14226 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
14227 (avrxmega6): Increase max flash segments from 5 to 6.
14228 * config/avr/t-multilib: Regenerate.
14229 * config/avr/avr-tables.opt: Regenerate.
14230 * doc/avr-mmcu.texi: Regenerate.
14231
28cd3cde 142322013-02-28 Georg-Johann Lay <avr@gjlay.de>
14233
14234 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
14235 (avr_device_to_arch): Rename to avr_device_to_ld.
14236 (avr_device_to_as): New prototype.
14237 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
14238 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
14239 * config/avr/driver-avr.c (avr_device_to_as): New.
14240 (avr_device_to_arch): Rename to avr_device_to_ld.
14241
3bda6402 142422013-02-27 Jakub Jelinek <jakub@redhat.com>
14243
1648f21f 14244 PR middle-end/56461
14245 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
14246 method on dr_chain and result_chain.
14247
6a191849 14248 PR middle-end/56461
14249 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
14250 pointer_set_destroy on not_executed_last_iteration.
14251
1224d221 14252 PR middle-end/56461
862c63eb 14253 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
1224d221 14254
c6d207aa 14255 PR middle-end/56461
14256 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
14257 FOR_EACH_DEFINED_FUNCTION when freeing state.
14258
73dcdb84 14259 PR middle-end/56461
14260 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
14261 pool_free.
14262 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
14263 overwriting it.
14264
e4898110 14265 PR middle-end/56461
14266 * ipa-cp.c (decide_whether_version_node): Call vec_free on
14267 known_aggs[i].items and release known_aggs vector.
14268
3bda6402 14269 PR middle-end/56461
14270 * ipa-reference.c (propagate): Free node_info even for alias nodes.
14271
142722013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
e249ec67 14273
3bda6402 14274 * config/microblaze/microblaze.c (microblaze_emit_compare):
14275 Use xor for EQ/NE comparisions.
14276 * config/microblaze/microblaze.md (cstoresf4): Add constraints
14277 (cbranchsf4): Adjust operator to comparison_operator.
e249ec67 14278
dcd56d11 142792013-02-27 Jakub Jelinek <jakub@redhat.com>
14280
14281 PR middle-end/56461
14282 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
14283 vector.
14284 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
14285 vec_safe_push, always update *slot.
14286 (redirect_edge_var_map_clear): Use vec_free.
14287 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
14288 (free_var_map_entry): Use vec_free.
14289 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
14290 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
14291
936ab1d9 142922013-02-27 Andrey Belevantsev <abel@ispras.ru>
14293
14294 PR middle-end/45472
936ab1d9 14295 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
14296 when the may_trap_p bit of the exprs being merged differs.
936ab1d9 14297 Reorder tests for speculativeness in the logical and operator.
14298
d5417a49 142992013-02-27 Jakub Jelinek <jakub@redhat.com>
14300
632c3914 14301 * incpath.c (add_standard_paths): Use reconcat instead of concat
14302 where appropriate and avoid leaking memory.
14303
ba30d337 14304 * opts.h: Include obstack.h.
14305 (opts_concat): New prototype.
14306 (opts_obstack): New declaration.
14307 * opts.c (opts_concat): New function.
14308 (opts_obstack): New variable.
14309 (init_options_struct): Call gcc_init_obstack on opts_obstack.
14310 (finish_options): Use opts_concat instead of concat
14311 and XOBNEWVEC instead of XNEWVEC.
14312 * opts-common.c (generate_canonical_option, decode_cmdline_option,
14313 generate_option): Likewise.
14314 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
14315 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
14316
d5417a49 14317 PR target/56455
14318 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
14319 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
14320
ed8fbc55 143212013-02-26 Jakub Jelinek <jakub@redhat.com>
14322
14323 PR middle-end/56461
14324 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
14325
931645bc 143262013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
14327
14328 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
14329 (arm_block_move_unaligned_straight): Likewise.
14330 (arm_adjust_block_mem): Likewise.
14331
7b04a0cb 143322013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
14333
14334 PR target/48901
14335 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
14336 temp, cond and label.
14337 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
14338
5d77e25b 14339 PR target/52500
14340 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
14341 * config/c6x/c6x.h (dbx_register_map): Update declaration.
14342
e8b0bf40 14343 PR target/52501
14344 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
14345 of prologue/epilogue functions.
14346
7ae884c9 14347 PR target/52550
14348 * config/tilegx/tilegx.c (tilegx_expand_prologue):
14349 Remove unused variable cfa_offset.
14350 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
14351
df70f973 14352 PR target/54639
14353 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
14354 type promotion to unsigned.
14355
5e10bd08 14356 PR target/54640
14357 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
14358 for HOST_WIDE_INT of 32 bit / same size as int.
14359 (arm_block_move_unaligned_straight): Likewise.
14360 (arm_adjust_block_mem): Likewise.
14361
46645ed9 14362 PR target/54662
14363 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
14364 ALL_CFLAGS.
14365
9df082d2 143662013-02-26 Marek Polacek <polacek@redhat.com>
14367
14368 PR tree-optimization/56426
862c63eb 14369 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
9df082d2 14370
258f96b3 143712013-02-26 Richard Biener <rguenther@suse.de>
14372
14373 PR target/56444
14374 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
14375 unused variable loops.
14376
50c96f17 143772013-02-26 Jakub Jelinek <jakub@redhat.com>
14378
db08f1c6 14379 PR tree-optimization/56448
14380 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
14381 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
14382 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
14383 later operands of the references, or even first operand for
14384 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
14385
50c96f17 14386 PR tree-optimization/56443
14387 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
14388 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
14389 to type_for_mode langhook.
14390
28c83d49 143912013-02-25 Matt Turner <mattst88@gmail.com>
14392
14393 * doc/invoke.texi: Document r4700.
14394
750e47f5 143952013-02-25 Richard Biener <rguenther@suse.de>
14396
14397 PR tree-optimization/56175
14398 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
14399 split out from ...
14400 (simplify_bitwise_binary): ... here. Also guard the conversion
14401 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
14402
ca7a473f 144032013-02-25 Catherine Moore <clm@codesourcery.com>
14404
14405 Revert:
14406 2013-02-24 Catherine Moore <clm@codesourcery.com>
14407 Maciej W. Rozycki <macro@codesourcery.com>
14408 Tom de Vries <tom@codesourcery.com>
31c66114 14409 Nathan Sidwell <nathan@codesourcery.com>
ca7a473f 14410 Iain Sandoe <iain@codesourcery.com>
14411 Nathan Froyd <froydnj@codesourcery.com>
31c66114 14412 Chao-ying Fu <fu@mips.com>
862c63eb 14413
ca7a473f 14414 * doc/extend.texi: (micromips, nomicromips, nocompression):
862c63eb 14415 Document new function attributes.
ca7a473f 14416 * doc/invoke.texi (minterlink-compressed, mmicromips,
14417 m14k, m14ke, m14kec): Document new options.
14418 (minterlink-mips16): Update documentation.
14419 * doc/md.texi (ZC, ZD): Document new constraints.
14420 * configure.ac (gcc_cv_as_micromips): Check if linker
14421 supports the .set micromips directive.
14422 * configure: Regenerate.
14423 * config.in: Regenerate.
14424 * config/mips/mips-tables.opt: Regenerate.
14425 * config/mips/micromips.md: New file.
14426 * constraints.md (ZC, AD): New constraints.
14427 * config/mips/predicates.md (movep_src_register): New predicate.
14428 (movep_src_operand): New predicate.
14429 (non_volatile_mem_operand): New predicate.
14430 * config/mips/mips.md (multimem): New type.
14431 (length): Differentiate between 17-bit and 18-bit branch offsets.
14432 (MOVEP1, MOVEP2): New mode iterator.
14433 (mov_<load>l): Use ZC constraint.
14434 (mov_<load>r): Likewise.
14435 (mov_<store>l): Likewise.
14436 (mov_<store>r): Likewise.
14437 (*branch_equality<mode>_inverted): Add microMIPS support.
14438 (*branch_equality<mode>): Likewise.
14439 (*jump_absolute): Likewise.
14440 (indirect_jump_<mode>): Likewise.
14441 (tablejump_<mode>): Likewise.
14442 (<optab>_internal): Likewise.
14443 (sibcall_internal): Likewise.
14444 (sibcall_value_internal): Likewise.
14445 (prefetch): Use constraint ZD.
14446 * config/mips/mips.opt (minterlink-compressed): New option.
14447 (minterlink-mips16): Now an alias for minterlink-compressed.
14448 (mmicromips): New option.
14449 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14450 (compare_and_swap_12): Likewise.
14451 (sync_add<mode>): Likewise.
14452 (sync_<optab>_12): Likewise.
14453 (sync_old_<optab>_12): Likewise.
14454 (sync_new_<optab>_12): Likewise.
14455 (sync_nand_12): Likewise.
14456 (sync_old_nand_12): Likewise.
14457 (sync_new_nand_12): Likewise.
14458 (sync_sub<mode>): Likewise.
14459 (sync_old_add<mode>): Likewise.
14460 (sync_old_sub<mode>): Likewise.
14461 (sync_new_add<mode>): Likewise.
14462 (sync_new_sub<mode>): Likewise.
14463 (sync_<optab><mode>): Likewise.
14464 (sync_old_<optab><mode>): Likewise.
14465 (sync_new_<optab><mode>): Likewise.
14466 (sync_nand<mode>): Likewise.
14467 (sync_old_nand<mode>): Likewise.
14468 (sync_new_nand<mode>): Likewise.
14469 (sync_lock_test_and_set<mode>): Likewise.
14470 (test_and_set_12): Likewise.
14471 (atomic_compare_and_swap<mode>): Likewise.
14472 (atomic_exchange<mode>_llsc): Likewise.
14473 (atomic_fetch_add<mode>_llsc): Likewise.
14474 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14475 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14476 (umips_save_restore_pattern_p): Likewise.
14477 (umips_load_store_pair_p): Likewise.
14478 (umips_output_load_store_pair): Likewise.
14479 (umips_movep_target_p): Likewise.
14480 (umips_12bit_offset_address_p): Likewise.
14481 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14482 (mips_base_mips16): Rename this...
14483 (mips_base_compression_flags): ...to this. Update all uses.
14484 (mips_attribute_table): Add micromips, nomicromips and nocompression.
14485 (mips_mips16_decl_p): Delete.
14486 (mips_nomips16_decl_p): Delete.
14487 (mips_get_compress_on_flags): New function.
14488 (mips_get_compress_off_flags): New function.
14489 (mips_get_compress_mode): New function.
14490 (mips_get_compress_on_name): New function.
14491 (mips_get_compress_off_name): New function.
14492 (mips_insert_attributes): Support multiple compression types.
14493 (mips_merge_decl_attributes): Likewise.
14494 (umips_12bit_offset_address_p): New function.
14495 (mips_start_function_definition): Emit .set micromips directive.
14496 (mips_call_may_need_jalx_p): New function.
14497 (mips_function_ok_for_sibcall): Add microMIPS support.
14498 (mips_print_operand_punctuation): Support short delay slots and
14499 compact jumps.
14500 (umips_swm_mask, umips_swm_encoding): New.
14501 (umips_build_save_restore): New function.
14502 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14503 (was_mips16_p): Remove.
14504 (old_compression_mode): New.
14505 (mips_set_compression_mode): New function.
14506 (mips_set_current_function): Add microMIPS support.
14507 (mips_option_override): Likewise.
14508 (umips_save_restore_pattern_p): New function.
14509 (umips_output_save_restore): New function.
14510 (umips_load_store_pair_p_1): New function.
14511 (umips_load_store_pair_p): New function.
14512 (umips_output_load_store_pair_1): New function.
14513 (umips_output_load_store_pair): New function.
14514 (umips_movep_target_p) New function.
14515 (mips_prepare_pch_save): Add microMIPS support.
14516 * config/mips/mips.h (TARGET_COMPRESSION): New.
14517 (TARGET_CPU_CPP_BUILTINS): Update macro
14518 to use new compression flags and to support microMIPS.
14519 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14520 (MIPS_ARCH_FLOAT_SPEC): Likewise.
14521 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14522 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14523 (ASM_SPEC): Support mmicromips and mno-micromips.
14524 (M16STORE_REG_P): New macro.
14525 (MIPS_CALL): Support TARGET_MICROMIPS.
14526 (MICROMIPS_J): New macro.
14527 (mips_base_mips16): Rename this...
14528 (mips_base_compression_flags): ...to this.
14529 (UMIPS_12BIT_OFFSET_P): New macro.
14530 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14531 (MULTILIB_DIRNAMES): Likewise.
14532
6de7acd5 145332013-02-25 Tom de Vries <tom@codesourcery.com>
14534
14535 PR rtl-optimization/56131
14536 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
14537 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
14538 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
14539
2d0b2b60 145402013-02-25 Tobias Burnus <burnus@net-b.de>
14541
14542 * doc/invoke.texi (-fsanitize=): Move from optimization
14543 to debugging options.
14544
f22426d2 145452013-02-25 Andrey Belevantsev <abel@ispras.ru>
14546
14547 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
14548
e965825b 145492013-02-25 Andrey Belevantsev <abel@ispras.ru>
14550 Alexander Monakov <amonakov@ispras.ru>
14551
14552 PR middle-end/56077
14553 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
14554 flush pending lists also on non-jumps. Adjust comment.
14555
7fd93f9c 145562013-02-24 Catherine Moore <clm@codesourcery.com>
14557 Maciej W. Rozycki <macro@codesourcery.com>
14558 Tom de Vries <tom@codesourcery.com>
31c66114 14559 Nathan Sidwell <nathan@codesourcery.com>
7fd93f9c 14560 Iain Sandoe <iain@codesourcery.com>
14561 Nathan Froyd <froydnj@codesourcery.com>
31c66114 14562 Chao-ying Fu <fu@mips.com>
7fd93f9c 14563
14564 * doc/extend.texi: (micromips, nomicromips, nocompression):
862c63eb 14565 Document new function attributes.
7fd93f9c 14566 * doc/invoke.texi (minterlink-compressed, mmicromips,
14567 m14k, m14ke, m14kec): Document new options.
14568 (minterlink-mips16): Update documentation.
14569 * doc/md.texi (ZC, ZD): Document new constraints.
14570 * configure.ac (gcc_cv_as_micromips): Check if linker
14571 supports the .set micromips directive.
14572 * configure: Regenerate.
14573 * config.in: Regenerate.
14574 * config/mips/mips-tables.opt: Regenerate.
14575 * config/mips/micromips.md: New file.
14576 * constraints.md (ZC, AD): New constraints.
14577 * config/mips/predicates.md (movep_src_register): New predicate.
14578 (movep_src_operand): New predicate.
14579 (non_volatile_mem_operand): New predicate.
14580 * config/mips/mips.md (multimem): New type.
14581 (length): Differentiate between 17-bit and 18-bit branch offsets.
14582 (MOVEP1, MOVEP2): New mode iterator.
2d0b2b60 14583 (mov_<load>l): Use ZC constraint.
7fd93f9c 14584 (mov_<load>r): Likewise.
14585 (mov_<store>l): Likewise.
14586 (mov_<store>r): Likewise.
14587 (*branch_equality<mode>_inverted): Add microMIPS support.
14588 (*branch_equality<mode>): Likewise.
14589 (*jump_absolute): Likewise.
14590 (indirect_jump_<mode>): Likewise.
14591 (tablejump_<mode>): Likewise.
14592 (<optab>_internal): Likewise.
14593 (sibcall_internal): Likewise.
14594 (sibcall_value_internal): Likewise.
14595 (prefetch): Use constraint ZD.
14596 * config/mips/mips.opt (minterlink-compressed): New option.
14597 (minterlink-mips16): Now an alias for minterlink-compressed.
14598 (mmicromips): New option.
14599 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14600 (compare_and_swap_12): Likewise.
14601 (sync_add<mode>): Likewise.
14602 (sync_<optab>_12): Likewise.
14603 (sync_old_<optab>_12): Likewise.
14604 (sync_new_<optab>_12): Likewise.
14605 (sync_nand_12): Likewise.
14606 (sync_old_nand_12): Likewise.
14607 (sync_new_nand_12): Likewise.
14608 (sync_sub<mode>): Likewise.
14609 (sync_old_add<mode>): Likewise.
14610 (sync_old_sub<mode>): Likewise.
14611 (sync_new_add<mode>): Likewise.
14612 (sync_new_sub<mode>): Likewise.
14613 (sync_<optab><mode>): Likewise.
14614 (sync_old_<optab><mode>): Likewise.
14615 (sync_new_<optab><mode>): Likewise.
14616 (sync_nand<mode>): Likewise.
14617 (sync_old_nand<mode>): Likewise.
14618 (sync_new_nand<mode>): Likewise.
14619 (sync_lock_test_and_set<mode>): Likewise.
14620 (test_and_set_12): Likewise.
14621 (atomic_compare_and_swap<mode>): Likewise.
14622 (atomic_exchange<mode>_llsc): Likewise.
14623 (atomic_fetch_add<mode>_llsc): Likewise.
14624 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14625 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14626 (umips_save_restore_pattern_p): Likewise.
14627 (umips_load_store_pair_p): Likewise.
14628 (umips_output_load_store_pair): Likewise.
14629 (umips_movep_target_p): Likewise.
14630 (umips_12bit_offset_address_p): Likewise.
14631 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14632 (mips_base_mips16): Rename this...
14633 (mips_base_compression_flags): ...to this. Update all uses.
14634 (mips_attribute_table): Add micromips, nomicromips and nocompression.
14635 (mips_mips16_decl_p): Delete.
14636 (mips_nomips16_decl_p): Delete.
2d0b2b60 14637 (mips_get_compress_on_flags): New function.
7fd93f9c 14638 (mips_get_compress_off_flags): New function.
14639 (mips_get_compress_mode): New function.
14640 (mips_get_compress_on_name): New function.
14641 (mips_get_compress_off_name): New function.
14642 (mips_insert_attributes): Support multiple compression types.
14643 (mips_merge_decl_attributes): Likewise.
14644 (umips_12bit_offset_address_p): New function.
14645 (mips_start_function_definition): Emit .set micromips directive.
14646 (mips_call_may_need_jalx_p): New function.
14647 (mips_function_ok_for_sibcall): Add microMIPS support.
14648 (mips_print_operand_punctuation): Support short delay slots and
14649 compact jumps.
14650 (umips_swm_mask, umips_swm_encoding): New.
14651 (umips_build_save_restore): New function.
14652 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14653 (was_mips16_p): Remove.
14654 (old_compression_mode): New.
14655 (mips_set_compression_mode): New function.
14656 (mips_set_current_function): Add microMIPS support.
14657 (mips_option_override): Likewise.
14658 (umips_save_restore_pattern_p): New function.
14659 (umips_output_save_restore): New function.
14660 (umips_load_store_pair_p_1): New function.
14661 (umips_load_store_pair_p): New function.
14662 (umips_output_load_store_pair_1): New function.
14663 (umips_output_load_store_pair): New function.
14664 (umips_movep_target_p) New function.
14665 (mips_prepare_pch_save): Add microMIPS support.
14666 * config/mips/mips.h (TARGET_COMPRESSION): New.
14667 (TARGET_CPU_CPP_BUILTINS): Update macro
14668 to use new compression flags and to support microMIPS.
14669 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14670 (MIPS_ARCH_FLOAT_SPEC): Likewise.
14671 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14672 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14673 (ASM_SPEC): Support mmicromips and mno-micromips.
14674 (M16STORE_REG_P): New macro.
14675 (MIPS_CALL): Support TARGET_MICROMIPS.
14676 (MICROMIPS_J): New macro.
14677 (mips_base_mips16): Rename this...
14678 (mips_base_compression_flags): ...to this.
14679 (UMIPS_12BIT_OFFSET_P): New macro.
14680 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14681 (MULTILIB_DIRNAMES): Likewise.
14682
9a9a97e2 146832013-02-24 Jakub Jelinek <jakub@redhat.com>
14684
14685 PR target/52555
14686 * target-globals.c (save_target_globals): For init_reg_sets and
14687 target_reinit remporarily set this_fn_optabs to this_target_optabs.
14688
752358b2 146892013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
14690
14691 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
14692 * config/aarch64/t-aarch64
14693 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
14694
8c3a9b39 146952013-02-22 Vladimir Makarov <vmakarov@redhat.com>
14696
14697 PR inline-asm/56148
14698 * lra-constraints.c (process_alt_operands): Reload operand
14699 conflicting with earlier clobber only if no more other conflicting
14700 operands.
14701
a16e0d2d 147022013-02-22 Jakub Jelinek <jakub@redhat.com>
14703
14704 PR sanitizer/56393
14705 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
14706 if not linking a shared library.
14707
0e0256f2 147082013-02-22 Seth LaForge <sethml@google.com>
14709
14710 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
14711
6aa689e8 147122013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
14713
2d0b2b60 14714 * config/arm/arm.md (split for extendsidi): Update condition.
14715 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
14716 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
14717 (qhs_zextenddi_cstr): Likewise.
6aa689e8 14718
4d5cf08a 147192013-02-21 Jakub Jelinek <jakub@redhat.com>
14720
af0907da 14721 PR middle-end/56420
14722 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
14723 avoid signed wrapping.
14724 (expand_mult): Handle properly multiplication by
14725 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
14726 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
14727 in the compiler if coeff is HOST_WIDE_INT_MIN.
14728 (expand_divmod): Don't make ext_op1 static, change it's type to
14729 uhwi. Avoid undefined behavior in -INTVAL (op1).
14730
4d5cf08a 14731 PR rtl-optimization/50339
14732 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
14733 field.
14734 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
14735 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
14736 into splitting_ashiftrt field.
14737 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
14738 ASHIFTRT.
14739 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
14740 choices.
14741
ded1a556 147422013-02-20 Aldy Hernandez <aldyh@redhat.com>
14743
14744 PR middle-end/56108
14745 * trans-mem.c (execute_tm_mark): Do not expand transactions that
14746 are sure to go irrevocable.
14747
16a3d053 147482013-02-21 Hans-Peter Nilsson <hp@axis.com>
14749
14750 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
14751 scalars are valid operands.
14752
9bd3a517 147532013-02-21 Martin Jambor <mjambor@suse.cz>
14754
14755 PR tree-optimization/56310
14756 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
14757 only matching indices and non-negative final offsets.
14758 (intersect_aggregates_with_edge): Pass src_idx to
14759 agg_replacements_to_vector. Pass src_idx insstead of index to
14760 intersect_with_agg_replacements.
14761
cfb368a3 147622013-02-21 Martin Jambor <mjambor@suse.cz>
14763
14764 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
14765 instead of hard-wired defaults.
14766
c856a95f 147672013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
14768
14769 * doc/invoke.texi (MIPS Options): Update documentation of the
14770 floating-point multiply-accumulate instruction restrictions.
14771
9dff5bbc 147722013-02-21 Kostya Serebryany <kcc@google.com>
5a59af67 14773
14774 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
14775 asan_shadow_offset on x86_64 linux.
14776
7662ff7e 147772013-02-21 Richard Biener <rguenther@suse.de>
14778
14779 PR tree-optimization/56415
14780 Revert
14781 2013-02-11 Richard Biener <rguenther@suse.de>
14782
14783 PR tree-optimization/56273
14784 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
14785 first VRP run.
14786
28a9e8c4 147872013-02-21 Jakub Jelinek <jakub@redhat.com>
14788
551e6483 14789 PR bootstrap/56258
14790 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
14791 instead of @itemx.
14792
28a9e8c4 14793 PR inline-asm/56405
14794 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
14795 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
14796
16473e06 147972013-02-20 Jan Hubicka <jh@suse.cz>
14798
14799 PR tree-optimization/56265
862c63eb 14800 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
14801 when target is referenced for first time.
16473e06 14802
560965e9 148032013-02-20 Richard Biener <rguenther@suse.de>
14804
14805 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
14806 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
14807 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
14808 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
14809 not return anything.
14810 (rename_ssa_copies): Do not remove unused locals.
14811 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
862c63eb 14812 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
560965e9 14813 * passes.c (execute_function_todo): Do not schedule unused locals
14814 removal if cleanup_tree_cfg did something.
14815 * tree-ssa-live.c (remove_unused_locals): Dump statistics
14816 about the number of removed locals.
14817
2510e5cd 148182013-02-20 Richard Biener <rguenther@suse.de>
14819
14820 PR tree-optimization/56398
862c63eb 14821 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
2510e5cd 14822
170754bc 148232013-02-20 Martin Jambor <mjambor@suse.cz>
14824
14825 PR tree-optimization/55334
14826 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
14827 restricted pointers to arrays.
14828
285df01b 148292013-02-20 Richard Biener <rguenther@suse.de>
862c63eb 14830 Jakub Jelinek <jakub@redhat.com>
285df01b 14831
14832 PR tree-optimization/56396
14833 * tree-ssa-ccp.c (n_const_val): New static variable.
14834 (get_value): Return NULL for SSA names we don't have a lattice
14835 entry for.
14836 (ccp_initialize): Initialize n_const_val.
14837 * tree-ssa-copy.c (n_copy_of): New static variable.
14838 (init_copy_prop): Initialize n_copy_of.
14839 (get_value): Return NULL_TREE for SSA names we don't have a
14840 lattice entry for.
14841
1cf0acc3 148422013-02-20 Martin Jambor <mjambor@suse.cz>
14843
14844 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
14845
0b588b3d 148462013-02-20 Richard Biener <rguenther@suse.de>
14847
14848 * genpreds.c (write_lookup_constraint): Do not compare first
14849 letter of the constraint again.
14850
d2ec9d3b 148512013-02-20 Richard Biener <rguenther@suse.de>
14852
14853 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
14854 and ceil_log2.
14855 (get_use_iv_cost): Terminate hashtable walk when coming across
14856 an empty entry.
14857
18daa717 148582013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
14859
14860 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
14861 reassociation for avx2 targets.
14862
31c66114 148632012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
d2ec9d3b 14864
f8a5cd80 14865 * config/microblaze/microblaze.c: microblaze_has_clz = 0
14866 Add version check for v8.10.a to enable microblaze_has_clz
862c63eb 14867 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
f8a5cd80 14868 version and TARGET_PATTERN_COMPARE check
14869 * config/microblaze/microblaze.md: New clzsi2 instruction
2d0b2b60 14870
31c66114 148712012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
dd21a8f3 14872
31c66114 14873 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
dd21a8f3 14874 function before branching.
14875
09a6d85b 148762012-02-19 Andrey Belevantsev <abel@ispras.ru>
14877
14878 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
14879 DUMP_INSN_RTX_UID.
14880 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
14881
3e1f1f1a 148822012-02-19 Andrey Belevantsev <abel@ispras.ru>
14883
14884 PR middle-end/55889
3e1f1f1a 14885 * sel-sched.c: Include ira.h.
14886 (implicit_clobber_conflict_p): New function.
14887 (moveup_expr): Use it.
862c63eb 14888 * Makefile.in (sel-sched.o): Depend on ira.h.
3e1f1f1a 14889
82a7a70c 148902013-02-19 Richard Biener <rguenther@suse.de>
14891
14892 PR tree-optimization/56384
14893 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
14894 (vn_hash_type): Split out from ...
14895 (vn_hash_constant_with_type): ... here.
14896 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
14897 (vn_phi_eq): Compare types from vn_phi_s structure.
14898 (vn_phi_lookup): Populate vn_phi_s type.
14899 (vn_phi_insert): Likewise.
14900
15c27dda 149012013-02-19 Jakub Jelinek <jakub@redhat.com>
14902
a82fc9c6 14903 PR tree-optimization/56350
14904 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
14905 if haven't found reduction or nested cycle operand, rather than
14906 asserting we must find it.
14907
15c27dda 14908 PR tree-optimization/56381
14909 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
14910 to fold_build3.
14911
149122013-02-18 Aldy Hernandez <aldyh@redhat.com>
08c7d04b 14913 Jakub Jelinek <jakub@redhat.com>
14914
14915 PR target/52555
14916 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
14917 (swap_optab_enable): Same.
14918 (init_all_optabs): Use argument instead of global.
862c63eb 14919 * tree.h (struct tree_optimization_option): New field target_optabs.
08c7d04b 14920 * expr.h (init_all_optabs): Add argument to prototype.
14921 (TREE_OPTIMIZATION_OPTABS): New.
14922 (save_optabs_if_changed): Protoize.
14923 * optabs.h: Declare this_fn_optabs.
14924 * optabs.c (save_optabs_if_changed): New.
14925 Declare this_fn_optabs.
14926 (init_optabs): Add argument to init_all_optabs() call.
14927 * function.c (invoke_set_current_function_hook): Handle per
14928 function optabs.
14929 * function.h (struct function): New field optabs.
14930 * config/mips/mips.c (mips_set_mips16_mode): Handle when
14931 optimization_current_node has changed.
14932 * target-globals.h (save_target_globals_default_opts): Protoize.
14933 * target-globals.c (save_target_globals_default_opts): New.
14934
4883d82f 149352013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
14936
14937 PR target/56347
14938 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
14939 registers %fr12 and %fr12R as call used.
172258a4 14940
14941 PR target/56214
14942 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
862c63eb 14943 and HImode, require all displacements to be an integer multiple of
14944 their mode size.
172258a4 14945 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
14946 only allow QImode and HImode when reload is in progress and strict is
14947 true. Likewise for symbolic addresses. Use base14_operand to check
14948 displacements in REG+BASE addresses.
14949
abad9af1 149502013-02-18 Richard Biener <rguenther@suse.de>
14951
14952 PR tree-optimization/56366
14953 * tree-vect-loop.c (get_initial_def_for_induction): Properly
14954 handle sign-conversion of outer-loop initial induction value.
14955
b375c775 149562013-02-18 Richard Biener <rguenther@suse.de>
14957
1450b4dd 14958 PR middle-end/56349
b375c775 14959 * cfghooks.c (merge_blocks): If we merge a latch into another
14960 block adjust references to it.
14961 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
14962 (verify_loop_structure): Verify that a recorded latch is in fact
14963 a latch.
14964
e6b37e57 149652013-02-18 Richard Biener <rguenther@suse.de>
14966
14967 PR tree-optimization/56321
14968 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
14969 order SSA name release and virtual operand unlinking.
14970
71cfadef 149712013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
14972
14973 * config/microblaze/microblaze.md (save_stack_block): Define.
14974 (restore_stack_block): Likewise.
14975
f9ca9e46 149762013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
14977
14978 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
14979 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
14980 * config/microblaze/microblaze.c (microblaze_option_override):
14981 Bail out early for PIC modes when target does not support PIC.
14982
10baa3f3 149832013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
14984
862c63eb 14985 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
10baa3f3 14986 Replace with a microblaze version.
14987 (microblaze_trampoline_init): Adapt for microblaze.
9a26e430 14988 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
10baa3f3 14989 microblaze.
14990
d9dc05a1 149912013-02-16 Jakub Jelinek <jakub@redhat.com>
14992 Dodji Seketeli <dodji@redhat.com>
14993
14994 PR asan/56330
862c63eb 14995 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
d9dc05a1 14996 (instrument_mem_region_access): Do not forget to always put
14997 instrumentation of the of 'base' and 'base + len' in a "if (len !=
14998 0) statement, even for cases where either 'base' or 'base + len'
14999 are not instrumented -- because they have been previously
15000 instrumented. Simplify the logic by putting all the statements
15001 instrument 'base + len' inside a sequence, and then insert that
15002 sequence right before the current insertion point. Then, to
15003 instrument 'base + len', just get an iterator on that statement.
15004 And do not forget to update the pointer to iterator the function
15005 received as argument.
15006
a27d5d8c 150072013-02-15 Vladimir Makarov <vmakarov@redhat.com>
15008
15009 PR rtl-optimization/56348
15010 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15011
69df7536 150122013-02-15 Steven Bosscher <steven@gcc.gnu.org>
15013
15014 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15015 (clean_graph_dump_file): Pass base to start_graph_dump.
15016
36c98bd9 150172013-02-14 Richard Henderson <rth@redhat.com>
15018
15019 PR target/55941
15020 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15021
6975c724 150222013-02-14 Steven Bosscher <steven@gcc.gnu.org>
15023
15024 * collect2-aix.h: Define F_LOADONLY.
15025
ca797a74 150262013-02-14 Richard Biener <rguenther@suse.de>
15027
15028 PR lto/50494
15029 * varasm.c (output_constant_def_1): Get the decl representing
15030 the constant as argument.
15031 (output_constant_def): Wrap output_constant_def_1.
15032 (make_decl_rtl): Use output_constant_def_1 with the decl
15033 representing the constant.
15034 (build_constant_desc): Optionally re-use a decl already
15035 representing the constant.
15036 (tree_output_constant_def): Adjust.
15037
0f2b58ae 150382013-02-14 Dodji Seketeli <dodji@redhat.com>
15039
15040 Fix an asan crash
15041 * asan.c (instrument_builtin_call): Really put the length of the
15042 second source argument into src1_len.
15043
e8d4d8a9 150442013-02-13 Jakub Jelinek <jakub@redhat.com>
15045
15046 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15047 argument. If it is false, don't create edge from then_bb to
15048 fallthru_bb.
15049 (insert_if_then_before_iter): Pass true to it.
15050 (build_check_stmt): Pass false to it.
15051 (transform_statements): Flush hash table only on extended basic
15052 block boundaries, rather than at the beginning of every bb.
15053 Don't flush hash table on nonfreeing_call_p calls.
15054 * tree-flow.h (nonfreeing_call_p): New prototype.
15055 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15056
944c7922 150572013-02-13 David S. Miller <davem@davemloft.net>
15058
15059 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15060
f03b2d9d 150612013-02-13 Vladimir Makarov <vmakarov@redhat.com>
15062
15063 PR target/56184
15064 * ira.c (max_regno_before_ira): Move from ...
15065 (ira): ... here.
15066 (fix_reg_equiv_init): Use max_regno_before_ira instead of
15067 vec_safe_length.
15068
77c5abed 150692013-02-13 Jakub Jelinek <jakub@redhat.com>
15070
15071 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15072
ecfa1cec 150732013-02-13 Richard Biener <rguenther@suse.de>
15074
15075 PR lto/56295
15076 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15077 globals in MEM_REFs.
15078
51f707ce 150792013-02-13 Richard Biener <rguenther@suse.de>
15080
15081 * loop-init.c (loop_optimizer_init): Clear loop state when
15082 re-initializing preserved loops.
15083 * loop-unswitch.c (unswitch_single_loop): Return whether
15084 we unswitched the loop. Do not verify loop state here.
862c63eb 15085 (unswitch_loops): When we unswitched a loop discover new loops.
51f707ce 15086
7da89e80 150872013-02-13 Kostya Serebryany <kcc@google.com>
15088
862c63eb 15089 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15090 on x86_64 linux.
7da89e80 15091 * sanitizer.def: Rename __asan_init to __asan_init_v1.
15092
c31c80df 150932013-02-12 Dodji Seketeli <dodji@redhat.com>
15094
15095 Avoid instrumenting duplicated memory access in the same basic block
15096 * Makefile.in (asan.o): Add new dependency on hash-table.h
15097 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15098 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15099 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15100 (free_mem_ref_resources, has_mem_ref_been_instrumented)
15101 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15102 (get_mem_ref_of_assignment): New functions.
15103 (get_mem_refs_of_builtin_call): Extract from
15104 instrument_builtin_call and tweak a little bit to make it fit with
15105 the new signature.
15106 (instrument_builtin_call): Use the new
15107 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
15108 of is_gimple_builtin_call.
15109 (instrument_derefs, instrument_mem_region_access): Insert the
15110 instrumented memory reference into the hash table.
15111 (maybe_instrument_assignment): Renamed instrument_assignment into
15112 this, and change it to advance the iterator when instrumentation
15113 actually happened and return true in that case. This makes it
15114 homogeneous with maybe_instrument_assignment, and thus give a
15115 chance to callers to be more 'regular'.
15116 (transform_statements): Clear the memory reference hash table
15117 whenever we enter a new BB, when we cross a function call, or when
15118 we are done transforming statements. Use
15119 maybe_instrument_assignment instead of instrumentation. No more
15120 need to special case maybe_instrument_assignment and advance the
15121 iterator after calling it; it's now handled just like
15122 maybe_instrument_call. Update comment.
15123
319f4d7d 151242013-02-13 Richard Biener <rguenther@suse.de>
15125
15126 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15127 Fix loop discovery code.
15128
89c2edcf 151292013-02-12 Vladimir Makarov <vmakarov@redhat.com>
15130
15131 PR inline-asm/56148
15132 * lra-constraints.c (process_alt_operands): Match early clobber
f03b2d9d 15133 operand with itself. Check conflicts with earlyclobber only if
89c2edcf 15134 the operand is not reloaded. Prefer to reload conflicting operand
15135 if earlyclobber and matching operands are the same.
15136
d648b2ee 151372013-02-12 Richard Biener <rguenther@suse.de>
15138
15139 PR lto/56297
15140 * lto-streamer-out.c (write_symbol): Do not output symbols
15141 for hard register variables.
15142
cfbcb7d5 151432013-02-12 Georg-Johann Lay <avr@gjlay.de>
15144
15145 PR target/54222
15146 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15147 (umulsidi3_insn, mulsidi3_insn): New insns.
15148
31c66114 151492013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
7ad0a5e6 15150
15151 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15152 (struct tune_params): Add vec_costs field.
15153 * config/arm/arm.c (arm_builtin_vectorization_cost)
15154 (arm_add_stmt_cost): New functions.
15155 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15156 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15157 (arm_default_vec_cost): New struct of type cpu_vec_costs.
15158 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15159 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15160 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15161 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15162
972031ce 151632013-02-12 Richard Biener <rguenther@suse.de>
15164
15165 PR lto/56295
15166 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15167 decls again if possible.
15168
bcef4e8e 151692013-02-12 Richard Biener <rguenther@suse.de>
15170
15171 PR middle-end/56288
15172 * tree-ssa.c (verify_ssa_name): Fix check, move
15173 SSA_NAME_IN_FREE_LIST check up.
15174
193baafd 151752013-02-12 Jakub Jelinek <jakub@redhat.com>
15176 Steven Bosscher <steven@gcc.gnu.org>
15177
15178 PR rtl-optimization/56151
15179 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15180 equal to op0 or op1, and last_insn pattern is CODE operation
15181 with MEM dest and one of the operands matches that MEM.
15182
fd457cef 151832013-02-11 Sriraman Tallam <tmsriramgoogle.com>
15184
15185 * doc/extend.texi: Document Function Multiversioning and "default"
15186 parameter string to target attribute.
15187 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15188 target attribute parameter is "default".
15189 (ix86_compare_version_priority): Remove checks for target attribute.
15190 (ix86_mangle_function_version_assembler_name): Change error to sorry.
15191 Remove check for target attribute equal to NULL. Add assert.
15192 (ix86_generate_version_dispatcher_body): Change error to sorry.
15193
48b8e70a 151942013-02-11 Iain Sandoe <iain@codesourcery.com>
15195 Jack Howarth <howarth@bromo.med.uc.edu>
15196 Patrick Marlier <patrick.marlier@gmail.com>
15197
15198 PR libitm/55693
15199 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15200 define ENDFILE_SPEC as TM_DESTRUCTOR.
15201 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15202
31c66114 152032013-02-11 Alexander Potapenko <glider@google.com>
ed161dab 15204 Jack Howarth <howarth@bromo.med.uc.edu>
15205 Jakub Jelinek <jakub@redhat.com>
15206
15207 PR sanitizer/55617
15208 * config/darwin.c (cdtor_record): Rename ctor_record.
15209 (sort_cdtor_records): Rename sort_ctor_records.
15210 (finalize_dtors): New routine to sort destructors by
15211 priority before use in assemble_integer.
15212 (machopic_asm_out_destructor): Use finalize_dtors if needed.
15213
ffc1977a 152142013-02-11 Uros Bizjak <ubizjak@gmail.com>
15215
15216 PR rtl-optimization/56275
15217 * simplify-rtx.c (avoid_constant_pool_reference): Check that
15218 offset is non-negative and less than cmode size before
15219 calling simplify_subreg.
15220
b6f3c6f1 152212013-02-11 Richard Biener <rguenther@suse.de>
15222
15223 PR tree-optimization/56264
15224 * cfgloop.h (fix_loop_structure): Adjust prototype.
15225 * loop-init.c (fix_loop_structure): Return the number of
15226 newly discovered loops.
15227 * tree-cfgcleanup.c (repair_loop_structures): When new loops
15228 are discovered, do a full loop-closed SSA rewrite.
15229
4d739378 152302013-02-11 Richard Biener <rguenther@suse.de>
15231
15232 PR tree-optimization/56273
15233 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15234 first VRP run.
15235 (check_array_ref): Fix missing newline in dumps.
15236 (search_for_addr_array): Likewise.
15237
31a1ce3e 152382013-02-09 David Edelsohn <dje.gcc@gmail.com>
15239
15240 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
15241
18163dd8 152422013-02-09 Jakub Jelinek <jakub@redhat.com>
15243
15244 PR target/56256
15245 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
15246
4aa54340 152472013-02-08 Vladimir Makarov <vmakarov@redhat.com>
15248
15249 PR rtl-optimization/56246
31a1ce3e 15250 * lra-constraints.c (simplify_operand_subreg): Try to reuse
4aa54340 15251 reload pseudo.
15252 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
15253 constraints are satisfied.
15254
ea239197 152552013-02-08 Jeff Law <law@redhat.com>
15256
15257 PR debug/53948
15258 * emit-rtl.c (reg_is_parm_p): New function.
15259 * regs.h (reg_is_parm_p): New prototype.
15260 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
15261 callee-clobbered registers.
15262
b85af273 152632013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
15264
15265 PR target/56043
15266 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
15267 If there is no implicit builtin declaration, just return NULL.
15268
21a6219e 152692013-02-08 Uros Bizjak <ubizjak@gmail.com>
15270
15271 * config/i386/sse.md (FMAMODEM): New mode iterator.
15272 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
15273 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
15274
a3c139e5 152752013-02-08 Uros Bizjak <ubizjak@gmail.com>
e975e345 15276
a3c139e5 15277 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
15278 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
15279 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
15280
152812013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15282
15283 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
15284 (microblaze*-*-elf): Likewise.
15285 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
15286 LINK_SPEC.
15287 * config/microblaze/microblaze-c.c: Add builtin defines for
15288 _LITTLE_ENDIAN and _BIG_ENDIAN.
15289 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
15290 add to TARGET_DEFAULT flags.
e975e345 15291 Expand ASM_SPEC and LINK_SPEC.
a3c139e5 15292 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
15293 * config/microblaze/microblaze.md: Update extendsidi2 and
15294 movdi_internal instructions to use low-order / high-order reg
15295 print_operands.
15296 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
15297 options and inversemask / mask of LITTLE_ENDIAN.
15298 * config/microblaze/t-microblaze: Expand multilib options to
15299 include mlittle-endian (le) and update exceptions patterns.
15300
1a8b7013 153012013-02-08 Jakub Jelinek <jakub@redhat.com>
15302
5ef30aea 15303 PR rtl-optimization/56195
15304 * lra-constraints.c (get_reload_reg): Don't reuse regs
15305 if they have smaller mode than requested, if they have
15306 wider mode than requested, try to return a SUBREG.
15307
1a8b7013 15308 PR tree-optimization/56250
15309 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
15310 if type is unsigned and code isn't MULT_EXPR.
15311
992b7387 153122013-02-08 Georg-Johann Lay <avr@gjlay.de>
15313
15314 PR tree-optimization/56064
15315 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
15316 bits according to mode.
15317 * fixed-value.h (fixed_from_double_int)
15318 (const_fixed_from_double_int): Adjust comments.
15319
ab3c0986 153202013-02-08 Richard Biener <rguenther@suse.de>
15321
15322 PR lto/56231
15323 * lto-streamer.h (struct data_in): Remove current_file, current_line
15324 and current_col members.
15325 * lto-streamer-out.c (lto_output_location): Stream changed bits
15326 en-block for efficiency.
15327 * lto-streamer-in.c (clear_line_info): Remove.
15328 (lto_input_location): Cache current file, line and column
15329 globally via local statics. Read changed bits en-block.
15330 (input_function): Do not call clear_line_info.
15331 (lto_read_body): Likewise.
15332 (lto_input_toplevel_asms): Likewise.
15333
42540642 153342013-02-08 Michael Matz <matz@suse.de>
15335
15336 PR tree-optimization/52448
15337 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
15338 (nt_call_phase): New static.
15339 (add_or_mark_expr): Only mark accesses with newer phase than any
15340 call seen.
15341 (nonfreeing_call_p): New.
15342 (nt_init_block): Update nt_call_phase, mark blocks as visited.
15343 (nt_fini_block): Keep blocks marked as visited.
15344 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
15345
1a8dfee2 153462013-02-08 Richard Biener <rguenther@suse.de>
15347
15348 * ira.c (ira): Free broken dominator information.
15349
eaa0ee9f 153502013-02-08 Uros Bizjak <ubizjak@gmail.com>
15351
15352 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
15353
14a0d8d7 153542013-02-08 Marek Polacek <polacek@redhat.com>
15355
eaa0ee9f 15356 * cfgloop.c (verify_loop_structure): Add more checking of headers.
14a0d8d7 15357
ff829efa 153582013-02-08 Richard Biener <rguenther@suse.de>
15359
15360 PR middle-end/56181
15361 * cfgloop.h (flow_loops_find): Adjust.
15362 (bb_loop_header_p): Declare.
15363 * cfgloop.c (bb_loop_header_p): New function split out from ...
15364 (flow_loops_find): ... here. Adjust function signature,
15365 support incremental loop structure update.
15366 (verify_loop_structure): Cleanup. Verify a loop is a loop.
15367 * cfgloopmanip.c (fix_loop_structure): Move ...
15368 * loop-init.c (fix_loop_structure): ... here.
15369 (apply_loop_flags): Split out from ...
15370 (loop_optimizer_init): ... here.
15371 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
15372 in incremental mode, only remove dead loops here.
15373
eafea251 153742013-02-08 Georg-Johann Lay <avr@gjlay.de>
15375
15376 PR target/54222
15377 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
15378 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
15379 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
15380 (*round<mode>3.libgcc): New insns for fixed-modes.
15381 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
15382 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
15383 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
15384 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
15385 implementations. Define to __builtin_avr_absFX,
15386 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
15387 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
15388 __builtin_avr_countlsFX, respectively.
15389 * config/avr/avr-c.c (target.h): Include it.
15390 (enum avr_builtin_id): New enum.
15391 (avr_resolve_overloaded_builtin): New static function.
15392 (avr_register_target_pragmas): Use it to set
15393 targetm.resolve_overloaded_builtin.
15394 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
15395 tree nodes used by DEF_BUILTIN.
15396 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
15397 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
15398 <AVR_BUILTIN_xxBITS>: Same.
15399
1032e48d 154002013-02-08 Richard Biener <rguenther@suse.de>
15401
15402 * cfgloop.c (verify_loop_structure): Properly handle
15403 a loop exiting to another loop header.
15404 * ira-int.h (ira_loops): Remove.
15405 * ira.c (ira_loops): Remove.
15406 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
15407 (do_reload): Use loop_optimizer_finalize.
15408 * ira-build.c (create_loop_tree_nodes): Use get_loops and
15409 number_of_loops to access the loop tree.
15410 (more_one_region_p): Likewise.
15411 (finish_loop_tree_nodes): Likewise.
15412 (rebuild_regno_allocno_maps): Likewise.
15413 (mark_loops_for_removal): Likewise.
15414 (mark_all_loops_for_removal): Likewise.
15415 (remove_unnecessary_regions): Likewise.
15416 (ira_build): Likewise.
15417 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
15418
8ff30f9a 154192013-02-08 Richard Biener <rguenther@suse.de>
15420
15421 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
15422 * ipa-pure-const.c (analyze_function): Avoid calling
15423 mark_irreducible_loops twice.
eaa0ee9f 15424 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
8ff30f9a 15425
690bfd6a 154262013-02-07 David S. Miller <davem@davemloft.net>
15427
15428 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
15429 on 'reg'.
15430 * var-tracking.c (vt_add_function_parameter): Test the presence of
15431 HAVE_window_save properly and do not remap argument registers when
15432 we have a leaf function.
15433
7a87c8e1 154342013-02-07 Uros Bizjak <ubizjak@gmail.com>
15435
15436 PR bootstrap/56227
15437 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
15438 instead of "ll".
15439 * config/i386/i386.c (ix86_print_operand): Ditto.
15440
92dfb77d 154412013-02-07 Vladimir Makarov <vmakarov@redhat.com>
15442
7a87c8e1 15443 * lra-constraints.c (process_alt_operands): Fix recently added comment.
92dfb77d 15444
2e620dc7 154452013-02-07 Vladimir Makarov <vmakarov@redhat.com>
15446
15447 PR rtl-optimization/56225
15448 * lra-constraints.c (process_alt_operands): Check that reload hard
15449 reg can hold value for strict_low_part.
15450
67a34241 154512013-02-07 Jakub Jelinek <jakub@redhat.com>
15452
15453 PR debug/56154
15454 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
15455 dwarf2out_end_function.
15456 (in_first_function_p, maybe_at_text_label_p,
15457 first_loclabel_num_not_at_text_label): New variables.
15458 (dwarf2out_var_location): In the first function find out
7a87c8e1 15459 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
67a34241 15460 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
15461 functions.
15462
a24ec999 154632013-02-07 Eric Botcazou <ebotcazou@adacore.com>
15464
15465 PR rtl-optimization/56178
15466 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
15467 SUBREG of a register. Tidy up related block of code.
15468 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
15469 note if the source is a register or a SUBREG of a register.
15470
bef70214 154712013-02-07 Jakub Jelinek <jakub@redhat.com>
15472
15473 PR target/56228
15474 * config/rs6000/rs6000.md (ptrm): New mode attr.
15475 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
15476 call_value_indirect_aix<pttrsize>,
15477 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
15478 m in constraints.
15479
f30f08a4 154802013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
15481
15482 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
15483 if -bnortl. Convert to strcmp and strncmp.
15484
9c74eaeb 154852013-02-07 Alan Modra <amodra@gmail.com>
15486
15487 PR target/54009
15488 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
15489 addresses won't wrap when offsetting.
15490 (rs6000_secondary_reload): Provide secondary reloads needed for
15491 wrapping LO_SUM addresses.
15492
e7306dad 154932013-02-06 Thomas Schwinge <thomas@codesourcery.com>
15494
15495 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
15496 MACH, just __MACH__.
15497
caa1ac55 154982013-02-06 Richard Biener <rguenther@suse.de>
15499
15500 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
15501 instead of calling fix_loop_structure.
15502
b8214689 155032013-02-06 Jakub Jelinek <jakub@redhat.com>
15504
15505 PR middle-end/56217
15506 * omp-low.c (use_pointer_for_field): Return false if
15507 lower_send_shared_vars doesn't generate any copy-out code.
15508
fb848efb 155092013-02-06 Tom de Vries <tom@codesourcery.com>
15510
15511 PR rtl-optimization/56131
15512 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
15513 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
15514 of the label is NULL. Add comment.
15515
2a111350 155162013-02-05 Jakub Jelinek <jakub@redhat.com>
15517
4f364235 15518 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
15519
5f13745b 15520 PR sanitizer/55374
15521 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
15522 (STATIC_LIBTSAN_LIBS): Likewise.
15523 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
15524 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
15525 is defined, don't add anything else beyond that.
15526 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
15527 (LINK_COMMAND_SPEC): Use them.
15528
2a111350 15529 PR tree-optimization/56205
15530 * tree-stdarg.c (check_all_va_list_escapes): Return true if
15531 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
15532 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
15533
bc17236a 155342013-02-05 Richard Biener <rguenther@suse.de>
15535
15536 PR tree-optimization/53342
15537 PR tree-optimization/53185
15538 * tree-vectorizer.h (vect_check_strided_load): Remove.
15539 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
15540 not disallow peeling for vectorized strided loads.
15541 (vect_check_strided_load): Make static and simplify.
15542 (vect_analyze_data_refs): Adjust.
15543 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
15544 correctly when vectorizing strided loads.
15545
b6a9ee95 155462013-02-05 Richard Biener <rguenther@suse.de>
15547
15548 * doc/install.texi: Refer to ISL, not PPL.
15549
8afc926b 155502013-02-05 Jan Hubicka <jh@suse.cz>
15551
b9b931d9 15552 PR tree-optimization/55789
15553 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
15554
155552013-02-05 Jan Hubicka <jh@suse.cz>
15556
15557 PR tree-optimization/55789
8afc926b 15558 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
15559 the dead call anyway.
15560
b514e580 155612013-02-05 Eric Botcazou <ebotcazou@adacore.com>
15562
15563 PR sanitizer/55374
15564 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
15565
155662013-02-04 Alexander Potapenko <glider@google.com>
15567 Jack Howarth <howarth@bromo.med.uc.edu>
4b313269 15568 Jakub Jelinek <jakub@redhat.com>
15569
15570 PR sanitizer/55617
15571 * config/darwin.c (sort_ctor_records): Stabilized qsort
15572 on constructor priority by using original position.
15573 (finalize_ctors): New routine to sort constructors by
15574 priority before use in assemble_integer.
15575 (machopic_asm_out_constructor): Use finalize_ctors if needed.
15576
4475ab54 155772013-02-04 Jakub Jelinek <jakub@redhat.com>
15578
15579 PR libstdc++/54314
15580 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
15581 about visibility on artificial decls.
15582 * config/sol2.c (solaris_assemble_visibility): Likewise.
15583
f04dc164 155842013-02-04 Kai Tietz <ktietz@redhat.com>
15585
15586 PR target/56186
15587 * config/i386/i386.c (function_value_ms_64): Add additional valtype
15588 argument and improve checking of return-argument types for 16-byte
15589 modes.
15590 (ix86_function_value_1): Add additional valtype argument on call
15591 of function_value_64.
15592 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
15593 handling infunction_value_64 function.
15594
f3567af6 155952013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
15596
4475ab54 15597 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
f3567af6 15598
5342937d 155992013-02-04 Richard Biener <rguenther@suse.de>
15600
15601 PR tree-optimization/56188
15602 * tree-ssa-structalias.c (label_visit): Consider case with
15603 initially non-empty points-to set.
15604 (perform_var_substitution): Dump node mapping and clean up.
15605
871609a8 156062013-02-04 Richard Guenther <rguenther@suse.de>
15607
15608 PR lto/56168
15609 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
15610 node prevail as last resort.
15611 (lto_symtab_merge_decls): Remove guard on LTRANS here.
7a87c8e1 15612 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
871609a8 15613
98102711 156142013-02-04 Richard Biener <rguenther@suse.de>
15615
15616 PR tree-optimization/56113
15617 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
15618 Merge into ...
15619 (equiv_class_lookup_or_add): ... this.
15620 (label_visit): Adjust and fix error in previous patch.
15621 (perform_var_substitution): Adjust.
15622
6c049e03 156232013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
15624
15625 * config/sh/divtab.c: Fix formatting and comments throughout the file.
15626 * config/sh/sh4-300.md: Likewise.
15627 * config/sh/sh4a.md: Likewise.
15628 * config/sh/constraints.md: Likewise.
15629 * config/sh/sh.md: Likewise.
15630 * config/sh/netbsd-elf.h: Likewise.
15631 * config/sh/predicates.md: Likewise.
15632 * config/sh/sh-protos.h: Likewise.
15633 * config/sh/ushmedia.h: Likewise.
15634 * config/sh/linux.h: Likewise.
15635 * config/sh/sh.c: Likewise.
15636 * config/sh/superh.h: Likewise.
15637 * config/sh/elf.h: Likewise.
15638 * config/sh/sh4.md: Likewise.
15639 * config/sh/sh.h: Likewise.
15640
e8248b41 156412013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15642
15643 * config/pa/constraints.md: Adjust unused letters. Change "T"
15644 constraint to match_test floating_point_store_memory_operand().
15645 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
15646 (base14_operand): New.
15647 (floating_point_store_memory_operand): New.
15648 (integer_store_memory_operand): Revise to use base14_operand and
15649 reg_plus_base_memory_operand.
15650 (move_dest_operand): Allow symbolic_memory_operands.
15651 (symbolic_memory_operand): Check for LO_SOM.
15652 (symbolic_operand): Change default case to break.
15653 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
15654 CONST_DOUBLE values to be reloaded by putting them into memory when
15655 the destination is a floating point register.
15656 (movdf): Remove code to handle CONST_DOUBLE.
15657 (movsf): Likewise.
15658 (reload_indf_r1): New.
15659 (reload_insf_r1): New.
15660 Consistently use "Q" and "T" constraints with integer and floating
15661 point move instructions, respectively.
15662 (movdi): Remove FAIL.
15663 Change predicate for source operand unamed DImode move from
15664 general_operand to move_src_operand.
15665 (umulsidi3): Change predicate for destination operand to
15666 register_operand.
15667 Likewise for similar unamed patterns.
15668 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
15669 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
15670 (hppa_legitimize_address): Simplify mask calculation.
15671 (pa_emit_move_sequence): Revised handling of secondary reloads from
15672 REG+D addresses for floating point loads and stores. Directly handle
15673 loading CONST0_RTX (mode) to a floating point register.
15674 (pa_secondary_reload): Handle reloading DF and SFmode constant values
15675 to floating point registers. Don't restrict secondary reloads to
15676 floating point registers to integer modes. Revise some comments and
15677 cleanup some code.
15678 (TARGET_LEGITIMATE_ADDRESS_P): Define.
15679 (pa_legitimate_address_p): New.
15680 (pa_legitimize_reload_address): New.
15681 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
15682 (STRICT_REG_OK_FOR_BASE_P): New.
15683 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
15684 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
15685
8bd7002d 156862013-02-03 David Edelsohn <dje.gcc@gmail.com>
15687 Andrew Dixie <andrewd@gentrack.com>
15688
15689 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
15690 flag set.
15691
bd23a499 156922013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
15693
15694 * expmed.c (extract_bit_field_1): Pass the full width of the
15695 structure to get_best_reg_extraction_insn.
15696
a8af2dd5 156972013-02-01 David Edelsohn <dje.gcc@gmail.com>
15698
15699 PR target/54601
15700 * configure.ac (use_cxa_atexit): Add AIX.
15701 * configure: Regenerate.
15702
15703 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
15704
2436e54f 157052013-02-01 Jakub Jelinek <jakub@redhat.com>
15706
15707 PR debug/54793
15708 * final.c (need_profile_function): New variable.
15709 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
15710 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
15711 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
15712 notes, targetm.asm_out.function_prologue doesn't emit anything,
15713 HAVE_prologue and profiler should be emitted before prologue,
15714 set need_profile_function instead of emitting it.
15715 (final_scan_insn): If need_profile_function, emit
15716 profile_function on the first NOTE_INSN_BASIC_BLOCK or
15717 NOTE_INSN_FUNCTION_BEG note.
15718
157192013-02-01 Richard Henderson <rth@redhat.com>
5b0b7360 15720
15721 * config/rs6000/rs6000.md (smulditi3): New.
15722 (umulditi3): New.
15723
67336b74 15724 * config/alpha/alpha.md (umulditi3): New.
15725
29f92520 157262013-02-01 David Edelsohn <dje.gcc@gmail.com>
15727
15728 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
15729 (ASM_OUTPUT_ALIGNED_LOCAL): New.
15730
c74cfdaf 157312013-02-01 Richard Biener <rguenther@suse.de>
15732
15733 PR tree-optimization/56113
15734 * tree-ssa-structalias.c (label_visit): Reduce work for
15735 single-predecessor nodes.
15736
6b40a1a5 157372013-02-01 Eric Botcazou <ebotcazou@adacore.com>
15738
15739 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
15740 range isn't testing for zero.
15741
9add6570 157422013-01-31 Steven Bosscher <steven@gcc.gnu.org>
15743
15744 PR middle-end/56113
15745 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
15746
9a5788ea 157472013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
15748 Nick Clifton <nickc@redhat.com>
15749
15750 * config/v850/constraints.md (Q): Define as a memory constraint.
15751 * config/v850/predicates.md (label_ref_operand): New predicate.
15752 (e3v5_shift_operand): New predicate.
15753 (ior_operator): New predicate.
15754 * config/v850/t-v850: Add e3v5 multilib.
15755 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
15756 (v850_gen_movdi): Prototype.
15757 * config/v850/v850.c: Add support for e3v5 architecture.
15758 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
15759 TARGET_V850E_UP.
15760 (construct_save_jarl): Add e3v5 long JARL support.
15761 (v850_adjust_insn_length): New function. Adjust length of call
15762 insns when using e3v5 instructions.
15763 (v850_gen_movdi): New function: Generate instructions to move a
15764 DImode value.
15765 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
15766 (CPP_SPEC): Define __v850e3v5__ as appropriate.
15767 (TARGET_USE_FPU): Enable for e3v5.
15768 (CONST_OK_FOR_W): New macro.
15769 (ADJUST_INSN_LENGTH): Define.
15770 * config/v850/v850.md (UNSPEC_LOOP): Define.
15771 (attr cpu): Add v850e3v5.
15772 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
15773 (movdi): New pattern.
15774 (movdi_internal): New pattern.
15775 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
15776 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
15777 (cstoresf4): Likewise.
15778 (cstoredf4): Likewise.
15779 (insv): New pattern.
15780 (rotlso3_a): New pattern.
15781 (rotlsi3_b): New pattern
15782 (rotlsi3_v850e3v5): New pattern.
15783 (doloop_begin): New pattern.
15784 (fix_loop_counter): New pattern.
15785 (doloop_end): New pattern.
15786 (branch_normal): Add e3v5 long branch support.
15787 (branch_invert): Likewise.
15788 (branch_z_normal): Likewise.
15789 (branch_z_invert): Likewise.
15790 (branch_nz_normal): Likewise.
15791 (branch_nz_invert): Likewise.
15792 (call_internal_short): Add e3v5 register-indirect JARL support.
15793 (call_internal_long): Likewise.
15794 (call_value_internal_short): Likewise.
15795 (call_value_internal_long): Likewise.
15796 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
15797 (mloop): New option.
15798 * config.gcc: Add support for configuring v840e3v5 target.
7a87c8e1 15799 * doc/invoke.texi: Document new v850 specific command line options.
9a5788ea 15800
7a85a361 158012013-01-31 Paul Koning <ni1d@arrl.net>
15802
15803 PR debug/55059
15804 PR debug/54508
15805 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
15806 children if parent is a class.
15807 (prune_unused_types_prune): Don't add DW_AT_declaration.
15808
a8bcfda2 158092013-01-31 Richard Biener <rguenther@suse.de>
15810
15811 PR tree-optimization/56157
15812 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
15813 match up operand with SLP child.
15814
6b71bf7b 158152013-01-31 Jason Merrill <jason@redhat.com>
15816
ebb59397 15817 PR debug/54410
6b71bf7b 15818 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
15819 parameters the first time.
15820 (gen_scheduled_generic_parms_dies): Check completeness here.
15821
56dfa44a 158222013-01-31 Richard Biener <rguenther@suse.de>
15823
15824 PR middle-end/53073
15825 * common.opt (faggressive-loop-optimizations): New flag,
15826 enabled by default.
15827 * doc/invoke.texi (faggressive-loop-optimizations): Document.
15828 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
15829 infer_loop_bounds_from_undefined by it.
15830
5789f1f8 158312013-01-31 Richard Biener <rguenther@suse.de>
15832
15833 PR tree-optimization/56150
15834 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
15835 visit virtual operands.
15836 (find_uses_to_rename_bb): Likewise.
15837
158382013-01-31 Richard Biener <rguenther@suse.de>
fad68b9f 15839
15840 PR tree-optimization/56150
15841 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
15842 mixed store non-store stmts.
15843
ae0c3984 158442013-01-30 Jakub Jelinek <jakub@redhat.com>
15845
53486a83 15846 PR sanitizer/55374
15847 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
15848 LIBASAN_EARLY_SPEC is defined.
15849 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
15850 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
15851 before %o.
15852 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
15853
ae0c3984 15854 PR c++/55742
15855 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
15856 invalid args instead of ICEing on it.
15857 (ix86_valid_target_attribute_tree): Return error_mark_node if
15858 ix86_valid_target_attribute_inner_p failed.
15859 (ix86_valid_target_attribute_p): Return false only if
15860 ix86_valid_target_attribute_tree returned error_mark_node. Allow
15861 target("default") attribute.
15862 (sorted_attr_string): Change argument from const char * to tree,
15863 merge in all target attribute arguments rather than just one.
15864 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
15865 instead of free. Avoid using strcat.
15866 (ix86_mangle_function_version_assembler_name): Mangle
15867 target("default") as if no target attribute is present. Adjust
15868 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
15869 instead of xmalloc and XDELETEVEC instead of free.
15870 (ix86_function_versions): Don't return true if one of the decls
15871 doesn't have target attribute. If they don't and one of the decls
15872 is DECL_FUNCTION_VERSIONED, report an error. Adjust
15873 sorted_attr_string caller. Use XDELETEVEC instead of free.
15874 (ix86_supports_function_versions): Remove.
15875 (make_name): Fix up formatting.
15876 (make_dispatcher_decl): Remove resolver_name and its initialization.
15877 Avoid leaking memory.
15878 (is_function_default_version): Return true if there is
15879 target("default") attribute rather than no target attribute at all.
15880 (make_resolver_func): Avoid leaking memory.
15881 (ix86_generate_version_dispatcher_body): Likewise.
15882 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
15883 * target.def (supports_function_versions): Remove.
15884 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
15885 * doc/tm.texi: Regenerated.
15886
85276115 158872013-01-30 Vladimir Makarov <vmakarov@redhat.com>
15888
15889 PR rtl-optimization/56144
15890 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
15891 for values with side effects.
15892
545cf51c 158932013-01-30 Richard Biener <rguenther@suse.de>
15894
15895 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
15896 (sparseset_pop): Likewise.
15897 * cfganal.c (compute_idf): Likewise. Increase work-stack size
15898 to be able to use quick_push in the worker loop.
15899
022d4ccc 159002013-01-30 Marek Polacek <polacek@redhat.com>
15901
15902 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
15903
224a26c5 159042013-01-30 Richard Biener <rguenther@suse.de>
15905
15906 PR lto/56147
7a87c8e1 15907 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
224a26c5 15908
52cd005d 159092013-01-30 Georg-Johann Lay <avr@gjlay.de>
15910
15911 PR tree-optimization/56064
15912 * fixed-value.c (fixed_from_double_int): New function.
15913 * fixed-value.h (fixed_from_double_int): New prototype.
15914 (const_fixed_from_double_int): New static inline function.
15915 * fold-const.c (native_interpret_fixed): New static function.
15916 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
15917 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
15918 (native_encode_fixed): New static function.
15919 (native_encode_expr) <FIXED_CST>: Use it.
15920 (native_interpret_int): Move double_int worker code to...
15921 * double-int.c (double_int::from_buffer): ...this new static method.
15922 * double-int.h (double_int::from_buffer): Prototype it.
15923
dd6f8b2c 159242013-01-30 Richard Biener <rguenther@suse.de>
15925
15926 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
15927 New pointer-map and obstack.
15928 (init_alias_vars): Allocate pointer-map and obstack.
15929 (delete_points_to_sets): Free them.
15930 (find_what_var_points_to): Cache result.
15931 (find_what_p_points_to): Adjust for changed interface of
15932 find_what_var_points_to.
15933 (compute_points_to_sets): Likewise.
15934 (ipa_pta_execute): Likewise.
15935
a602740a 159362013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15937
15938 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
15939 * configure: Regenerate.
15940 * config.in: Regenerate.
15941 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
15942 #nobits/#progbits if supported.
15943
9081e4b3 159442013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
15945
15946 PR target/56121
15947 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
15948 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
15949 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
15950
a3826176 159512013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
15952
6effa4dc 15953 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
15954 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
a3826176 15955
f9e7462d 159562013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
15957
6effa4dc 15958 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
15959 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
f9e7462d 15960
eca9ccc8 159612013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
15962
6effa4dc 15963 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
15964 declaration.
15965 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
15966 * config/arm/cortex-a7.md: New bypasses using
15967 arm_mac_accumulator_is_result.
eca9ccc8 15968
fd923cfd 159692013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
15970
6effa4dc 15971 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
fd923cfd 15972 (cortex_a7_neon_mla): Likewise.
6effa4dc 15973 (cortex_a7_fpfmad): New reservation.
15974 (cortex_a7_fpmacs): Use ffmas and update required units.
15975 (cortex_a7_fpmuld): Update required units and latency.
15976 (cortex_a7_fpmacd): Likewise.
15977 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
15978 (cortex_a7_neon). Likewise.
15979 (bypass) Update participating units.
fd923cfd 15980
56df0c05 159812013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
15982
6effa4dc 15983 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
15984 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
15985 from fmac to ffma.
15986 * config/arm/vfp11.md (vfp_farith): Use ffmas.
15987 (vfp_fmul): Use ffmad.
15988 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
15989 (cortex_r4_fmacd): Use ffmad.
15990 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
15991 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
15992 (cortex_a9_fmacd): Use ffmad.
15993 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
15994 (cortex_a8_vfp_macd): Use ffmad.
15995 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
15996 (cortex_a5_fpmacd): Use ffmad.
15997 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
15998 (cortex_a15_vfp_macd): Use ffmad.
15999 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
56df0c05 16000
f88705ba 160012013-01-29 Jason Merrill <jason@redhat.com>
16002
16003 PR libstdc++/54314
16004 * varasm.c (default_assemble_visibility): Don't warn about
16005 visibility on artificial decls.
16006
783d1e89 160072013-01-29 Richard Biener <rguenther@suse.de>
16008
16009 PR tree-optimization/56113
16010 * tree-ssa-structalias.c (equiv_class_lookup): Also return
16011 the bitmap leader.
16012 (label_visit): Free duplicate bitmaps and record the leader instead.
16013 (perform_var_substitution): Adjust.
16014
5daac166 160152013-01-29 Richard Biener <rguenther@suse.de>
16016
16017 PR tree-optimization/55270
16018 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16019 the CFG, schedule loops for fixup.
16020
f716766f 160212013-01-29 Nick Clifton <nickc@redhat.com>
16022
16023 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16024 SP_REG.
16025
b433697b 160262013-01-28 Leif Ekblad <leif@rdos.net>
16027
16028 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16029 * config/i386/i386.h (TARGET_RDOS): New macro.
16030 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16031 * config/i386/i386.c (ix86_option_override_internal): For 64bit
16032 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16033 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16034 DEFAULT_LARGE_SECTION_THRESHOLD.
16035 * config/i386/i386.md (R14_REG, R15_REG): New constants.
16036 * config/i386/rdos.h: New file.
16037 * config/i386/rdos64.h: New file.
16038
25399338 160392013-01-28 Bernd Schmidt <bernds@codesourcery.com>
16040
16041 PR other/54814
16042 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16043 TEST_HARD_REG_BIT.
16044
12e51a1d 160452013-01-28 Jakub Jelinek <jakub@redhat.com>
16046
16047 PR rtl-optimization/56117
16048 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16049 call cselib_lookup_from_insn on the MEM before calling
16050 add_insn_mem_dependence.
16051
688b0b5b 160522013-01-28 Richard Biener <rguenther@suse.de>
16053
16054 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16055 to a stmt that didn't have one.
16056 (copy_phis_for_bb): Likewise for PHI arguments.
16057 (copy_debug_stmt): Likewise for debug stmts.
16058
ac7a1007 160592013-01-28 Richard Biener <rguenther@suse.de>
16060
16061 PR tree-optimization/56034
7a87c8e1 16062 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
ac7a1007 16063 (partition_builtin_p): Adjust.
16064 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
16065 it is the last partition.
16066 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16067 up the vertex for the definition.
16068 (classify_partition): Classify whether a partition is a
16069 PKIND_REDUCTION, thus has uses outside of the loop.
16070 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16071 Merge all PKIND_REDUCTION partitions into the last partition.
16072 (tree_loop_distribution): Seed partitions from reductions as well.
16073
04ff97ce 160742013-01-28 Jakub Jelinek <jakub@redhat.com>
16075
0190fe95 16076 PR tree-optimization/56125
16077 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16078 pow(x,c) into sqrt(x) * powi(x, n/2) or
16079 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16080 optimizing for size.
16081 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16082 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16083 integer.
16084
04ff97ce 16085 PR tree-optimization/56094
16086 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16087 to UNKNOWN_LOCATION while gimplifying expr.
16088
f2897e01 160892013-01-27 Uros Bizjak <ubizjak@gmail.com>
16090
16091 PR target/56114
16092 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16093 operand 0 in movabs insn template for -masm=intel asm alternative.
16094 (*movabs<mode>_2): Ditto for operand 1.
16095
0190fe95 160962013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
c48df545 16097
16098 PR target/54663
16099 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16100 of microblaze-c.o
16101
0190fe95 161022013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
98f2ac05 16103
16104 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16105 tm_file.
16106
0190fe95 161072013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
9fc8b0bf 16108
16109 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16110 Undef to avoid warning.
16111
82163b89 161122013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
16113
16114 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16115 * configure: Regenerate.
16116
6cc085b6 161172013-01-25 Jakub Jelinek <jakub@redhat.com>
16118
16119 PR tree-optimization/56098
16120 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16121 for stmts with volatile ops.
16122 (cond_store_replacement): Don't optimize if assign has volatile ops.
16123 (cond_if_else_store_replacement_1): Don't optimize if either
16124 then_assign or else_assign have volatile ops.
16125 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16126 volatile ops.
16127
d2ac69cc 161282013-01-25 Georg-Johann Lay <avr@gjlay.de>
16129
16130 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16131
63b6db20 161322013-01-25 Georg-Johann Lay <avr@gjlay.de>
16133
16134 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16135 missing ':' in asm example.
16136
431fc079 161372013-01-25 Tejas Belagod <tejas.belagod@arm.com>
16138
16139 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16140 entries into lane and laneq entries.
f2897e01 16141 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16142 Remove AdvSIMD scalar modes.
431fc079 16143 (aarch64_sq<r>dmulh_laneq<mode>): New.
16144 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16145 modes.
16146 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16147 builtin implementations to relfect changes in RTL in aarch64-simd.md.
16148 * config/aarch64/iterators.md (VCOND): New.
16149 (VCONQ): New.
16150
f3297245 161512013-01-25 Georg-Johann Lay <avr@gjlay.de>
16152
16153 PR target/54222
16154 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16155 Add NULL LIBNAME argument to existing definitions.
16156 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16157 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16158 * config/avr/avr.c (DEF_BUILTIN): Same.
16159 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16160 (avr_expand_builtin): Expand to a vanilla call if a libgcc
16161 implementation is available (DECL_ASSEMBLER_NAME is set).
16162 (avr_fold_absfx): New static function.
16163 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16164 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16165 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16166 AVR_BUILTIN_ABSLLK.
16167 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16168 (abshk, absk, abslk, absllk): Provide as static inline functions.
16169
8e3bada1 161702013-01-25 Marek Polacek <polacek@redhat.com>
16171
16172 PR tree-optimization/56035
16173 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16174
cd0e2161 161752012-01-24 Uros Bizjak <ubizjak@gmail.com>
16176
16177 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16178 (*movtf_internal_rex64): Add (!o,C) alternative
16179 (*movxf_internal_rex64): Ditto.
16180 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16181
2b139ca6 161822013-01-24 Shenghou Ma <minux.ma@gmail.com>
16183
16184 * doc/invoke.texi: fix typo.
16185 * doc/objc.texi: fix typo.
16186
1283f262 161872013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
16188
16189 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16190 for the first two alternatives.
16191
5cc13354 161922013-01-24 Diego Novillo <dnovillo@google.com>
16193
f2897e01 16194 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
5cc13354 16195 (ggc-zone.o): Remove.
16196 * configure.ac: Remove option --with-gc.
16197 * configure: Re-generate.
16198 * doc/install.texi: Remove documentation for --with-gc.
16199 * gengtype.c (write_enum_defn): Remove. Update all users.
16200 (write_Types_process_field): Remove generation of gt_e_* argument.
16201 (output_type_enum): Remove. Update all users.
16202 (write_enum_defn): Remove. Update all users.
16203 (enum alloc_zone): Remove. Update all users.
f2897e01 16204 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
5cc13354 16205 * ggc-common.c (ggc_splay_alloc): Remove first argument.
16206 Update all callers.
16207 (struct ptr_data): Remove field TYPE. Update all users.
f2897e01 16208 (gt_pch_note_object): Remove argument TYPE. Update all users.
5cc13354 16209 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16210 Update all users.
96e241e8 16211 * ggc-none.c (ggc_alloc_typed_stat): Remove.
5cc13354 16212 (struct alloc_zone): Remove.
16213 (ggc_internal_alloc_zone_stat): Remove.
16214 (ggc_internal_cleared_alloc_zone_stat): Remove.
16215 * ggc-page.c (ggc_alloc_typed_stat): Remove.
f2897e01 16216 (ggc_pch_count_object): Remove last argument. Update all users.
16217 (ggc_pch_alloc_object): Remove last argument. Update all users.
5cc13354 16218 (struct alloc_zone): Remove.
16219 * ggc-zone.c: Remove.
f2897e01 16220 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
5cc13354 16221 (struct alloc_zone): Remove.
16222 (ggc_alloc_typed_stat): Remove.
16223 (ggc_alloc_typed): Remove.
16224 (ggc_splay_alloc): Remove first argument.
16225 (rtl_zone): Remove. Update all users.
16226 (tree_zone): Remove. Update all users.
16227 (tree_id_zone): Remove. Update all users.
16228 (ggc_internal_zone_alloc_stat): Remove. Update all users.
f2897e01 16229 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
5cc13354 16230 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
f2897e01 16231 * tree-ssanames.c: Remove references to zone allocator in comments.
5cc13354 16232
2b5c3dd1 162332013-01-24 Georg-Johann Lay <avr@gjlay.de>
16234
16235 * config/avr/avr.c (avr_out_fract): Make register numbers that
16236 might be outside of source operand signed.
16237
80946863 162382013-01-24 Uros Bizjak <ubizjak@gmail.com>
16239
16240 * config/i386/constraints.md (Yf): New constraint.
16241 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
16242 of f constraint to conditionaly disable x87 register preferences.
16243 (*movdf_internal): Ditto.
16244 (*movsf_internal): Ditto.
16245
aa46b107 162462013-01-24 Steven Bosscher <steven@gcc.gnu.org>
16247
16248 PR inline-asm/55934
16249 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
16250 that have operands with impossible constraints.
16251 Add a FIXME for a speed-up opportunity.
16252 * lra-constraints.c (process_alt_operands): Verify that a class
16253 selected from constraints on asms is valid for the operand mode.
16254 (curr_insn_transform): Remove incorrect comment.
16255
657dd719 162562013-01-23 David Edelsohn <dje.gcc@gmail.com>
16257
16258 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
16259 TOC operand is a valid symbol ref in the constant pool.
16260
162612013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
d6fe7425 16262
96e241e8 16263 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
d6fe7425 16264
c9b315f4 162652013-01-23 Georg-Johann Lay <avr@gjlay.de>
16266
16267 PR target/54222
16268 * config/avr/stdfix.h: New file.
16269 * t-avr (stdfix-gcc.h): New rule to build it.
16270 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
16271
4a2c1ffc 162722013-01-23 Kostya Serebryany <kcc@google.com>
16273
f2897e01 16274 * config/darwin.h: remove dependency on
16275 CoreFoundation (asan on Mac OS).
4a2c1ffc 16276
b8eae306 162772013-01-23 Jakub Jelinek <jakub@redhat.com>
16278
16279 PR target/49069
16280 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
16281 instead of cmpdi_operand for first comparison operand.
16282 Don't assert that comparison operands aren't both constants.
16283
feadd43d 162842013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
16285
16286 * doc/install.texi (Downloading the Source): Update references to
16287 downloading separate components.
16288
162892013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
32b80c6a 16290
16291 * doc/extend.texi (__int128): Improve grammar.
16292
feadd43d 162932013-01-22 Uros Bizjak <ubizjak@gmail.com>
38a03d63 16294
16295 PR target/56028
16296 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
16297 alternative to (o,r).
16298 (*movdi_internal_rex64): Remove (!o,n) alternative.
16299 (DImode immediate->memory splitter): Remove.
16300 (DImode immediate->memory peephole2): Remove.
16301 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
16302 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
16303 alternative to (!o,*r).
16304 (*movtf_internal_sse): New pattern.
16305 (*movxf_internal_rex64): New pattern.
16306 (*movxf_internal): Disable for TARGET_64BIT.
16307 (*movdf_internal_rex64): Remove (!o,F) alternative.
16308
e6d8288f 163092013-01-22 Jakub Jelinek <jakub@redhat.com>
16310
b2d978a6 16311 PR middle-end/56074
16312 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
16313 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
16314 * tree-vect-loop-manip.c (find_loop_location): Also ignore
16315 stmt locations where LOCATION_LOCUS of the stmt location is
16316 UNKNOWN_LOCATION or BUILTINS_LOCATION.
16317
e6d8288f 16318 PR target/55686
16319 * config/i386/i386.md (UNSPEC_STOS): New.
16320 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
16321 *strsetqi_1): Add UNSPEC_STOS.
16322
81b38777 163232013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
16324
16325 PR c++/56067
16326 * doc/invoke.texi: Remove left over -Wsynth example.
16327
3e97846e 163282013-01-21 Jakub Jelinek <jakub@redhat.com>
16329
16330 PR tree-optimization/56051
16331 * fold-const.c (fold_binary_loc): Don't fold
16332 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
16333 a narrowing conversion, or widening conversion from signed
16334 to unsigned.
16335
feadd43d 163362013-01-21 Uros Bizjak <ubizjak@gmail.com>
ae3eaa0f 16337
16338 PR rtl-optimization/56023
16339 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
16340 dependent on debug instruction.
16341
a956a7a6 163422013-01-21 Martin Jambor <mjambor@suse.cz>
16343
16344 PR middle-end/56022
16345 * function.c (allocate_struct_function): Call
16346 invoke_set_current_function_hook earlier.
16347
cd5c8367 163482013-01-21 Jakub Jelinek <jakub@redhat.com>
16349
16350 * reload1.c (init_reload): Only initialize reload_obstack
16351 during the first call.
16352
dcb14b61 163532013-01-21 Marek Polacek <polacek@redhat.com>
16354
16355 * cfgloop.c (verify_loop_structure): Fix up grammar.
16356
1a262a91 163572013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
16358
16359 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
16360 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
16361
43115c41 163622013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16363
16364 PR target/56058
16365 * config/arm/marvell-pj4.md: Update copyright year.
16366 Fix up use of alu to alu_reg and simple_alu_imm.
16367
feadd43d 163682013-01-21 Uros Bizjak <ubizjak@gmail.com>
914b013a 16369
16370 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
16371
e0420317 163722013-01-20 Vladimir Makarov <vmakarov@redhat.com>
16373
16374 PR target/55433
16375 * lra-constraints.c (curr_insn_transform): Don't reuse original
914b013a 16376 insn for secondary memory move when memory mode should be different.
e0420317 16377
e7bca8d9 163782013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16379
16380 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
16381 atomic_storedi_1): New patterns.
16382
6470d004 163832013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
16384
16385 btver2 pipeline descriptions.
16386 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
16387 descriptions.
16388 * config/i386/i386.md (btver2_decode): New type attributes.
914b013a 16389 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
6470d004 16390 type attributes.
16391 * config/i386/btver2.md: New file describing btver2 pipelines.
16392
3a73dcc2 163932013-01-19 Andrew Pinski <apinski@cavium.com>
16394
16395 PR tree-optimization/52631
16396 * tree-ssa-sccvn (visit_use): Before looking up the original
16397 statement, try looking up the simplified expression.
16398
a3076649 163992013-01-19 Anthony Green <green@moxielogic.com>
16400
16401 * config/moxie/moxie.c (moxie_expand_prologue): Set
16402 current_function_static_stack_size.
16403
16f02f09 164042013-01-18 Jakub Jelinek <jakub@redhat.com>
16405
16406 PR tree-optimization/56029
16407 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
16408 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
16409
c4c46233 164102013-01-18 Sharad Singhai <singhai@google.com>
16411
16412 PR tree-optimization/55995
16413 * dumpfile.c (dump_loc): Print location only if available.
16414 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
16415
c47331e3 164162013-01-18 Vladimir Makarov <vmakarov@redhat.com>
16417
16418 PR target/55433
16419 * lra-constraints.c (curr_insn_transform): Reuse original insn for
16420 secondary memory move.
16421 (inherit_reload_reg): Use rclass instead of cl for
16422 check_secondary_memory_needed_p.
16423
83014b20 164242013-01-18 Jakub Jelinek <jakub@redhat.com>
16425
16426 PR middle-end/56015
16427 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
914b013a 16428 the case where writing real complex part of target modifies op1.
83014b20 16429
15953bb4 164302013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
16431
16432 * config/aarch64/aarch64-simd.md
16433 (aarch64_vcond_internal<mode>): Handle unordered cases.
16434 * config/aarch64/iterators.md (v_cmp_result): New.
16435
ea7d210b 164362013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
16437 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16438
16439 * config/arm/marvell-pj4.md: New file.
16440 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
16441 * config/arm/arm.md (generic_sched): Add marvell_pj4.
16442 (generic_vfp): Likewise.
16443 * config/arm/arm-cores.def: Add marvell-pj4.
16444 * config/arm/arm-tune.md: Regenerate.
16445 * config/arm/arm-tables.opt: Regenerate.
16446 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
16447 * doc/invoke.texi: Document marvell-pj4.
16448
7a28aaef 164492013-01-18 Tejas Belagod <tejas.belagod@arm.com>
16450
16451 * config/aarch64/arm_neon.h: Map scalar types to standard types.
16452
0190fe95 164532013-01-18 Alexandre Oliva <aoliva@redhat.com>
861a6de4 16454
16455 PR debug/54114
16456 PR debug/54402
16457 PR debug/49888
16458 * var-tracking.c (negative_power_of_two_p): New.
16459 (global_get_addr_cache, local_get_addr_cache): New.
16460 (get_addr_from_global_cache, get_addr_from_local_cache): New.
16461 (vt_canonicalize_addr): Rewrite using the above. Adjust the
16462 heading comment.
16463 (vt_stack_offset_p): Remove.
16464 (vt_canon_true_dep): Always canonicalize loc's address.
16465 (clobber_overlapping_mems): Make sure we have a MEM.
16466 (local_get_addr_clear_given_value): New.
16467 (val_reset): Clear local cached entries.
16468 (compute_bb_dataflow): Create and release the local cache.
16469 Disable duplicate MEMs clobbering.
16470 (emit_notes_in_bb): Clobber MEMs likewise.
16471 (vt_emit_notes): Create and release the local cache.
16472 (vt_initialize, vt_finalize): Create and release the global
16473 cache, respectively.
914b013a 16474 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
861a6de4 16475
0190fe95 164762013-01-18 Alexandre Oliva <aoliva@redhat.com>
a9490210 16477
16478 PR libmudflap/53359
16479 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
16480 not found in the symtab.
16481
0190fe95 164822013-01-18 Alexandre Oliva <aoliva@redhat.com>
340ad617 16483
1ff3b42b 16484 PR debug/56006
340ad617 16485 PR rtl-optimization/55547
16486 PR rtl-optimization/53827
16487 PR debug/53671
16488 PR debug/49888
16489 * alias.c (offset_overlap_p): New, factored out of...
16490 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
16491 the conservative special case for symbolic constants. Don't
16492 adjust zero sizes on alignment.
16493
1bcf869e 164942013-01-18 Bernd Schmidt <bernds@codesourcery.com>
16495
16496 PR rtl-optimization/52573
16497 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
16498 REG_UNUSED for the same register.
16499
c1f96968 165002013-01-17 Richard Biener <rguenther@suse.de>
16501 Marek Polacek <polacek@redhat.com>
16502
16503 PR rtl-optimization/55833
16504 * loop-unswitch.c (unswitch_loops): Move loop verification...
16505 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
16506 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
16507 Set it to true when we're removing a loop from hierarchy tree in
16508 an irreducible region.
16509 (fix_bb_placements): Adjust caller.
16510 (fix_loop_placements): Likewise.
16511
73263209 165122013-01-17 Georg-Johann Lay <avr@gjlay.de>
16513
16514 * config/avr/builtins.def (DEF_BUILTIN): Factor out
16515 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
16516 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
16517 Remove ID. Adjust comments.
16518 * config/avr/avr-c.c (avr_builtin_name): Remove.
16519 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
16520 * config/avr/avr.c (avr_tolower): New static function.
16521 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
16522 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
16523 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
16524 default expansion.
16525
9ce07db4 165262013-01-17 Jan Hubicka <jh@suse.cz>
16527
cb7b3a2f 16528 PR tree-optimization/55273
9ce07db4 16529 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
16530
feadd43d 165312013-01-17 Uros Bizjak <ubizjak@gmail.com>
a5b5a55a 16532
16533 PR target/55981
16534 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
16535 store through atomic_store<mode>_1.
16536 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
16537
3f442959 165382013-01-17 Martin Jambor <mjambor@suse.cz>
16539
16540 PR tree-optimizations/55264
16541 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
16542 for virtual methods.
16543 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
16544 virtual methods before inlining is over.
16545 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
16546 virtual functions.
16547 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
16548 non-virtual.
16549
a2e20fd2 165502013-01-16 Vladimir Makarov <vmakarov@redhat.com>
16551
16552 PR rtl-optimization/56005
16553 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
16554 pending reads for prefetch.
16555
e7e5bbc5 165562013-01-16 Ian Bolton <ian.bolton@arm.com>
16557
96e241e8 16558 * config/aarch64/aarch64.md
e7e5bbc5 16559 (*cstoresi_neg_uxtw): New pattern.
16560 (*cmovsi_insn_uxtw): New pattern.
16561 (*<optab>si3_uxtw): New pattern.
16562 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
16563 (*<optab>si3_insn_uxtw): New pattern.
16564 (*bswapsi2_uxtw): New pattern.
16565
a74056fd 165662013-01-16 Richard Biener <rguenther@suse.de>
16567
16568 * tree-inline.c (tree_function_versioning): Remove set but
16569 never used variable.
16570
c9b2c569 165712013-01-16 Richard Biener <rguenther@suse.de>
16572
16573 PR tree-optimization/55964
16574 * tree-flow.h (rename_variables_in_loop): Remove.
16575 (rename_variables_in_bb): Likewise.
16576 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
16577 (copy_loop_before): Adjust and delete update-ssa status.
16578 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
a5b5a55a 16579 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
c9b2c569 16580 (rename_variables_in_loop): Remove.
16581 (slpeel_update_phis_for_duplicate_loop): Likewise.
16582 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
16583 use available cfg machinery instead of duplicating it.
16584 Update PHI nodes and perform poor-mans SSA update here.
16585 (slpeel_tree_peel_loop_to_edge): Adjust.
16586
d590d541 165872013-01-16 Richard Biener <rguenther@suse.de>
16588
16589 PR tree-optimization/54767
16590 PR tree-optimization/53465
16591 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
16592 (vrp_visit_phi_node): For PHI arguments coming via backedges
16593 drop all symbolical range information.
16594 (execute_vrp): Compute backedges.
16595
c7cd65bf 165962013-01-16 Richard Biener <rguenther@suse.de>
16597
16598 * doc/install.texi: Update CLooG and ISL requirements to
16599 0.18.0 and 0.11.1.
16600
31fd081d 166012013-01-16 Christian Bruel <christian.bruel@st.com>
16602
16603 PR target/55301
16604 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
16605 (broken_move): Handle UNSPECV_SP_SWITCH_B.
16606 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
16607
166082013-01-16 DJ Delorie <dj@redhat.com>
16609
16610 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
16611 (UNSPECV_SP_SWITCH_E): New.
16612 (sp_switch_1): Change to an unspec.
16613 (sp_switch_2): Change to an unspec. Don't use post-inc when we
16614 replace $r15.
16615
166162013-01-16 Uros Bizjak <ubizjak@gmail.com>
1a9fa1dd 16617
16618 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
16619 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
16620 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
16621 (expand_mem_thread_fence): Ditto.
16622 (expand_mem_signal_fence): Ditto.
16623 (expand_atomic_load): Ditto.
16624 (expand_atomic_store): Ditto.
16625
0190fe95 166262013-01-16 Alexandre Oliva <aoliva@redhat.com>
bb2bc4f3 16627
16628 PR rtl-optimization/55547
16629 PR rtl-optimization/53827
16630 PR debug/53671
16631 PR debug/49888
16632 * alias.c (memrefs_conflict_p): Set sizes to negative after
16633 AND adjustments.
16634
392a7dd5 166352013-01-15 Jakub Jelinek <jakub@redhat.com>
16636
16637 PR target/55940
16638 * function.c (thread_prologue_and_epilogue_insns): Always
16639 add crtl->drap_reg to set_up_by_prologue.set, even if
16640 stack_realign_drap is false.
16641
8aaf4223 166422013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
16643
16644 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
16645 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
16646 *call): Fix indention.
16647
c35e21be 166482013-01-15 Tom de Vries <tom@codesourcery.com>
16649
16650 PR target/55876
16651 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
16652 Update comment.
16653
42fb263f 166542013-01-15 Vladimir Makarov <vmakarov@redhat.com>
16655
392a7dd5 16656 PR rtl-optimization/55153
42fb263f 16657 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
16658
1789bced 166592013-01-15 Martin Jambor <mjambor@suse.cz>
16660
16661 PR tree-optimization/55920
16662 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
16663 accesses as grp_to_be_debug_replaced.
16664
ddce22b8 166652013-01-15 Jakub Jelinek <jakub@redhat.com>
16666
16667 PR tree-optimization/55920
16668 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
16669 there is non-useless type conversion needed from debug rhs to lhs,
16670 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
16671
1969f96d 166722013-01-15 Joseph Myers <joseph@codesourcery.com>
16673 Mikael Pettersson <mikpe@it.uu.se>
16674
16675 PR target/43961
16676 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
16677 Thumb.
16678 (ASM_OUTPUT_CASE_LABEL): Remove.
16679 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
16680 * final.c (shorten_branches): Update alignment of labels before
16681 jump tables if CASE_VECTOR_SHORTEN_MODE.
16682
21be23eb 166832013-01-15 Richard Biener <rguenther@suse.de>
16684
16685 PR bootstrap/55961
16686 * system.h: Do not include gmp.h for building host tools.
16687
323cf38f 166882013-01-15 Richard Biener <rguenther@suse.de>
16689
16690 PR middle-end/55882
16691 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
16692 account for bitpos when computing alignment.
16693
56de744c 166942013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
16695
16696 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
16697 (ix86_target_macros_internal): Likewise.
16698
16699 * config/i386/i386.c (m_CORE2I7): Removed.
16700 (m_CORE_HASWELL): New macro.
16701 (m_CORE_ALL): Likewise.
16702 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
16703 (initial_ix86_arch_features): Likewise.
16704 (processor_target_table): Initializations for Core avx2.
16705 (cpu_names): New names "core-avx2".
16706 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
16707 PROCESSOR_CORE_HASWELL.
16708 (ix86_issue_rate): New case.
16709 (ia32_multipass_dfa_lookahead): Likewise.
16710 (ix86_sched_init_global): Likewise.
16711
16712 * config/i386/i386.h (TARGET_HASWELL): New macro.
16713 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
16714 (processor_type): New PROCESSOR_HASWELL.
16715
8365f6c8 167162013-01-15 Jakub Jelinek <jakub@redhat.com>
16717
2d788f29 16718 PR tree-optimization/55955
16719 * tree-vect-loop.c (vectorizable_reduction): Give up early on
16720 *SHIFT_EXPR and *ROTATE_EXPR codes.
16721
8365f6c8 16722 PR tree-optimization/48766
16723 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
16724 -ftrapv disable -fwrapv.
16725
02edff68 167262013-01-14 Georg-Johann Lay <avr@gjlay.de>
16727
16728 PR target/55974
16729 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
16730 etc. to 1 and not to __flash.
16731 Use LL suffix for __INT24_MAX__ with -mint8.
16732 Use ULL suffix for __UINT24_MAX__ with -mint8.
16733
ce544647 167342013-01-14 Georg-Johann Lay <avr@gjlay.de>
16735
16736 * config/avr/avr-arch.h
16737 (struct base_arch_s): Use typedef avr_arch_t instead.
16738 (struct arch_info_s): Use typedef avr_arch_info_t instead.
16739 (struct mcu_type_s): Use typedef avr_mcu_t instead.
16740 * config/avr/avr.c: Same.
16741 * config/avr/avr-devices.c: Same.
16742 * config/avr/driver-avr.c: Same.
16743 * config/avr/gen-avr-mmcu-texi.c: Same.
16744 * config/avr/avr-mcus.def: Adjust comment.
16745
df83fa4d 167462013-01-14 Tejas Belagod <tejas.belagod@arm.com>
16747
1a9fa1dd 16748 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
16749 * config/aarch64/iterators.md (VALLDI): New.
df83fa4d 16750
feadd43d 167512013-01-14 Uros Bizjak <ubizjak@gmail.com>
5fc510c2 16752 Andi Kleen <ak@linux.intel.com>
16753
16754 PR target/55948
16755 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
16756 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
16757 memmodel flag.
16758
0dff9558 167592013-01-14 Georg-Johann Lay <avr@gjlay.de>
16760
16761 * config/avr/avr-stdint.h: Remove trailing blanks.
16762 * config/avr/avr-log.h: Same.
16763 * config/avr/avr-arch.h: Same.
16764 * config/avr/avr-devices.c: Same.
16765 * config/avr/avr-dimode.md: Same.
16766 * config/avr/predicates.md: Same.
16767 * config/avr/avr-c.c: Same. And fix typo.
5fc510c2 16768
0dff9558 16769 * config/avr/avr-protos.h: Same. And:
16770 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
16771 (init_cumulative_args): Rename to avr_init_cumulative_args.
16772 (expand_prologue): Rename to avr_expand_prologue.
16773 (expand_epilogue): Rename to avr_expand_epilogue.
16774 (adjust_insn_length): Rename to avr_adjust_insn_length.
16775 (notice_update_cc): Rename to avr_notice_update_cc.
16776 (final_prescan_insn): Rename to avr_final_prescan_insn.
16777 * config/avr/avr.c: Same.
16778 * config/avr/avr.h: Same.
16779 * config/avr/avr.md: Remove trailing blanks.
16780 (prologue): Use avr_expand_prologue.
16781 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
16782
66366b96 167832013-01-14 Richard Biener <rguenther@suse.de>
16784
16785 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
16786 verify_location, collect_subblocks): New functions.
16787 (verify_gimple_in_cfg): Verify that locations only reference
16788 BLOCKs in the functions BLOCK tree.
16789
11db727d 167902013-01-14 Richard Biener <rguenther@suse.de>
16791
16792 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
16793 PHI argument.
16794 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
16795 unshare reference.
16796 (insert_out_of_ssa_copy_on_edge): Likewise.
16797 (rewrite_close_phi_out_of_ssa): Likewise.
16798 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
16799 debug expressions.
16800 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
16801 propagated constants.
16802 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
16803 can not be shared.
16804
1e6735ea 168052013-01-14 Georg-Johann Lay <avr@gjlay.de>
16806
16807 * config/avr/avr-modes.def: Add GPL copyright notice.
16808
86eac679 168092013-01-13 Uros Bizjak <ubizjak@gmail.com>
16810
16811 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
16812 MEMMODEL_MASK to determine memory model.
16813 (atomic_store<mode>): Ditto from operands[2].
16814 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
16815
8f266cd9 168162013-01-13 Jakub Jelinek <jakub@redhat.com>
16817
16818 PR fortran/55935
86eac679 16819 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
8f266cd9 16820 (fold_gimple_assign): Don't call unshare_expr here.
16821 (fold_ctor_reference): Call unshare_expr.
16822
7137cdac 168232013-01-13 Terry Guo <terry.guo@arm.com>
16824
96e241e8 16825 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
16826 * doc/fragments.texi: Document MULTILIB_REUSE.
16827 * gcc.c (multilib_reuse): New internal spec.
7137cdac 16828 (set_multilib_dir): Also search multilib from multilib_reuse.
96e241e8 16829 * genmultilib (tmpmultilib3): Refactor code.
7137cdac 16830 (tmpmultilib4): Ditto.
16831 (multilib_reuse): New multilib argument.
16832
9231cb67 168332013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
16834
16835 * Makefile.in: Update copyright.
16836
3d38d682 168372013-01-12 Tom de Vries <tom@codesourcery.com>
16838
16839 PR middle-end/55890
16840 * calls.c (expand_call): Check if arg_nr is valid.
16841
5c598e22 168422013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16843
16844 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
16845 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
21b14971 16846 documentation. Add missing '__' in front of
16847 __builtin_ia32_packssdw256.
5c598e22 16848
de47476b 168492013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
16850
16851 PR target/55719
16852 * config/s390/s390.c (s390_preferred_reload_class): Do not return
16853 NO_REGS for larl operands.
16854 (s390_reload_larl_operand): Use s390_load_address instead of
16855 emit_move_insn.
16856
bf27bcdd 168572013-01-11 Richard Biener <rguenther@suse.de>
16858
16859 * tree-cfg.c (verify_node_sharing_1): Split out from ...
16860 (verify_node_sharing): ... here.
16861 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
16862
1410e798 168632013-01-11 Eric Botcazou <ebotcazou@adacore.com>
16864
16865 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
16866 Substitute TREECHECKING.
16867 * configure: Regenerate.
16868 * Makefile.in (TREECHECKING): New.
16869
feadd43d 168702013-01-11 Richard Guenther <rguenther@suse.de>
742ca2be 16871
16872 PR tree-optimization/44061
16873 * tree-vrp.c (extract_range_basic): Compute zero as
16874 value-range for __builtin_constant_p of function parameters.
16875
711789cc 168762013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
16877
86eac679 16878 Update copyright years.
711789cc 16879
9441d07f 168802013-01-10 Vladimir Makarov <vmakarov@redhat.com>
16881
8f266cd9 16882 PR rtl-optimization/55672
8b599a09 16883 * lra-eliminations.c (mark_not_eliminable): Permit addition with
16884 const to be eliminable.
9441d07f 16885
e69effa4 168862013-01-10 David Edelsohn <dje.gcc@gmail.com>
16887
16888 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
16889 * configure: Regenerate.
16890
f4146cb8 168912013-01-10 Richard Biener <rguenther@suse.de>
16892
91f28fc1 16893 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
f4146cb8 16894
e60062ea 168952013-01-10 Richard Biener <rguenther@suse.de>
16896
16897 PR bootstrap/55792
16898 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
16899 locations for virtual PHI arguments.
16900 (rewrite_update_phi_arguments): Likewise.
16901
efc3a86d 169022013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
16903
16904 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
16905 on to assembler.
16906
349ff920 169072013-01-10 Jakub Jelinek <jakub@redhat.com>
16908
16909 PR tree-optimization/55921
16910 * tree-complex.c (expand_complex_asm): New function.
16911 (expand_complex_operations_1): Call it for GIMPLE_ASM.
16912
2a672556 169132013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
16914
16915 PR target/55718
16916 * config/s390/s390.c (s390_symref_operand_p)
16917 (s390_loadrelative_operand_p): Merge the two functions.
16918 (s390_check_qrst_address, print_operand_address): Add parameters
16919 to s390_loadrelative_operand_p invokation.
16920 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
16921 (s390_reload_larl_operand, s390_secondary_reload): Use
16922 s390_loadrelative_operand_p instead of s390_symref_operand_p.
16923 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
16924
e4209874 169252013-01-09 Mike Stump <mikestump@comcast.net>
16926
16927 * dse.c (record_store): Remove unnecessary assert.
16928
dde4834c 169292013-01-09 Jan Hubicka <jh@suse.cz>
16930
16931 PR tree-optimization/55569
16932 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
16933 * cfgloop.h (scale_loop_profile): Likewise.
16934
c7fbaa62 169352013-01-09 Jan Hubicka <jh@suse.cz>
16936
16937 PR lto/45375
91f28fc1 16938 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
16939 functions.
c7fbaa62 16940 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
16941
933eb13a 169422013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
16943
16944 PR middle-end/55114
16945 * expr.h (maybe_emit_group_store): Declare.
16946 * expr.c (maybe_emit_group_store): New function.
16947 * builtins.c (expand_builtin_int_roundingfn): Call it.
16948 (expand_builtin_int_roundingfn_2): Likewise.
16949
aa3ce8ba 169502013-01-09 Vladimir Makarov <vmakarov@redhat.com>
16951
701835de 16952 PR rtl-optimization/55829
aa3ce8ba 16953 * lra-constraints.c (match_reload): Add code for absent output.
16954 (curr_insn_transform): Add code for reloads of matched inputs
16955 without output.
16956
889d67a8 169572013-01-09 Uros Bizjak <ubizjak@gmail.com>
16958
16959 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
16960 attribute of movddup insn to DF.
16961 (*vec_interleave_lowv2df): Ditto.
16962 (vec_dupv2df): Ditto.
16963
64552007 169642013-01-09 Jan Hubicka <jh@suse.cz>
16965
16966 PR tree-optimiation/55875
16967 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
16968 EVERY_ITERATION parameter.
889d67a8 16969 (number_of_iterations_exit): Check if exit is executed every iteration.
64552007 16970 (idx_infer_loop_bounds): Similarly here.
16971 (n_of_executions_at_most): Simplify
16972 to only test for cases where statement is dominated by the
889d67a8 16973 particular bound; handle correctly the "postdominance" test.
64552007 16974 (scev_probably_wraps_p): Use max loop iterations info
16975 as a global bound first.
16976
e1181777 169772013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
bd38291a 16978 Nick Clifton <nickc@redhat.com>
16979
16980 * config/v850/v850.md (cbranchsf4): New pattern.
16981 (cstoresf4): New pattern.
16982 (cbranchdf4): New pattern.
16983 (cstoredf4): New pattern.
16984 (movsicc): Disallow floating point comparisons.
16985 (cmpsf_le_insn): Fix order of operators.
16986 (cmpsf_lt_insn): Likewise.
16987 (cmpsf_eq_insn): Likewise.
16988 (cmpdf_le_insn): Likewise.
16989 (cmpdf_lt_insn): Likewise.
16990 (cmpdf_eq_insn): Likewise.
16991 (cmpsf_ge_insn): Use LE comparison.
16992 (cmpdf_ge_insn): Likewise.
16993 (cmpsf_gt_insn): Use LT comparison.
16994 (cmpdf_gt_insn): Likewise.
16995 (cmpsf_ne_insn): Delete pattern.
16996 (cmpdf_ne_insn): Delete pattern.
16997 * config/v850/v850.c (v850_gen_float_compare): Use
16998 gen_cmpdf_eq_insn for NE comparison.
16999 (v850_float_z_comparison_operator)
17000 (v850_float_nz_comparison_operator): Move from here ...
17001 * config/v850/predicates.md: ... to here. Move GT and GE
17002 comparisons into v850_float_z_comparison_operator.
17003 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17004 Delete prototype.
17005 (v850_float_nz_comparison_operator): Likewise.
17006
81d5e322 170072013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17008
17009 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17010 with calls to gen_insvsi/gen_insvdi.
17011
3a33f4ce 170122013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
17013
17014 * config/i386/i386.c (initial_ix86_tune_features): Set up
17015 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17016
ed60f27f 170172013-01-09 Steven Bosscher <steven@gcc.gnu.org>
17018 Jakub Jelinek <jakub@redhat.com>
17019
17020 PR tree-optimization/48189
17021 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17022 If nitercst is 0, don't predict the exit edge.
17023
7a87c8e1 170242013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
b9ccfbdf 17025
17026 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17027 in asm_fprintf with reg_names.
17028 (aarch64_print_operand_address): Likewise.
17029 (aarch64_return_addr): Likewise.
17030 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17031
21522827 170322013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17033
17034 * config/pa/pa.h (VAL_U6_BITS_P): Define.
17035 (INT_U6_BITS): Likewise.
17036 * config/pa/predicates.md (uint6_operand): New predicate.
17037 (shift5_operand, shift6_operand): Likewise.
17038 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17039 arith32_operand.
17040 (lshrdi3): Use shift6_operand.
17041 (shrpsi4, shrpdi4): New insn patterns.
17042 (extzv): Delete expander.
17043 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
17044 predicates in unamed zero extract patterns. Tighten common constraint.
17045 (extv): Delete expander.
17046 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
17047 predicates in unamed sign extract patterns. Tighten common constraint.
17048 (insv): Delete expander.
17049 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
17050 predicates in unamed insert patterns. Tighten common constraint.
17051 Change uint32_operand predicate to uint6_operand predicate in unamed
17052 DImode pattern to insert constant values of type 1...1xxxx.
17053
a240d038 170542013-01-04 Jan Hubicka <jh@suse.cz>
17055
17056 PR tree-optimization/55823
889d67a8 17057 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17058 issue.
a240d038 17059
feadd43d 170602013-01-08 Jakub Jelinek <jakub@redhat.com>
bbfdbc02 17061 Uros Bizjak <ubizjak@gmail.com>
17062
17063 PR rtl-optimization/55845
17064 * df-problems.c (can_move_insns_across): Stop scanning at
17065 volatile_insn_p source instruction or give up if
17066 across_from .. across_to range contains any volatile_insn_p
17067 instructions.
17068
501336f7 170692013-01-08 Tejas Belagod <tejas.belagod@arm.com>
17070
889d67a8 17071 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17072 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17073 Declare.
501336f7 17074 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
889d67a8 17075 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
501336f7 17076
cc72d6e9 170772013-01-08 Jakub Jelinek <jakub@redhat.com>
17078
17079 PR fortran/55341
17080 * asan.c (asan_clear_shadow): New function.
17081 (asan_emit_stack_protection): Use it.
17082
b3f1c89d 170832013-01-08 Tejas Belagod <tejas.belagod@arm.com>
17084
17085 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17086 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17087 with tab instead of space.
17088
8d5f09ff 170892013-01-08 Nick Clifton <nickc@redhat.com>
17090
17091 * config/rl78/rl78.c (rl78_expand_prologue): Always select
17092 register bank 0 at the start of an interrupt handler.
50008a8e 17093 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17094 MDBH registers.
8d5f09ff 17095
1eb525c3 170962013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
17097
17098 * config/aarch64/aarch64-simd.md
17099 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17100 (aarch64_simd_bsl): Likewise.
17101 (aarch64_vcond_internal<mode>): Likewise.
17102 (vcond<mode><mode>): Likewise.
17103 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17104 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17105
5b4374a1 171062013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
17107
17108 * config/aarch64/aarch64-builtins.c
17109 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17110
5721b378 171112013-01-08 Martin Jambor <mjambor@suse.cz>
17112
17113 PR debug/55579
17114 * tree-sra.c (analyze_access_subtree): Return true also after
17115 potentially creating a debug-only replacement.
17116
4b9630f0 171172013-01-08 Jakub Jelinek <jakub@redhat.com>
17118
0b4f0116 17119 PR middle-end/55890
17120 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17121
4b9630f0 17122 PR tree-optimization/54120
17123 * tree-vrp.c (range_fits_type_p): Don't allow
17124 src_precision < precision from signed vr to unsigned_p
17125 if vr->min or vr->max is negative.
17126 (simplify_float_conversion_using_ranges): Test can_float_p
17127 against CODE_FOR_nothing.
17128
100d7996 171292013-01-08 Jakub Jelinek <jakub@redhat.com>
17130 Richard Biener <rguenther@suse.de>
17131
17132 PR middle-end/55851
17133 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17134 types instead of just INTEGER_TYPE types.
17135
6e029611 171362013-01-07 Mark Kettenis <kettenis@openbsd.org>
17137
17138 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17139 TF_SIZE): Define.
8d5f09ff 17140
8a5bc191 171412013-01-07 Steve Ellcey <sellcey@mips.com>
17142
17143 PR target/42661
17144 * config/mips/mips.opt: Change mad to mmad to match documentation.
17145
975036ab 171462013-01-07 Georg-Johann Lay <avr@gjlay.de>
17147
17148 PR target/55897
17149 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17150 .progmemx.data now.
17151
0558f5da 171522013-01-07 Georg-Johann Lay <avr@gjlay.de>
17153
17154 PR target/55897
17155 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17156 (avr_addrspace_t): Add .section_name field.
17157 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17158 array size.
17159 (avr_addrspace): Same. Initialize .section_name. Remove last
17160 NULL entry. Put __memx into .progmemx.data.
17161 (progmem_section_prefix): Remove.
17162 (avr_asm_init_sections): No need to initialize progmem_section.
17163 (avr_asm_named_section): Use avr_addrspace[].section_name to get
17164 section name prefix.
17165 (avr_asm_select_section): Ditto. And use get_unnamed_section to
17166 retrieve the progmem section.
17167 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17168 boundary to run over avr_addrspace[].
17169 (avr_register_target_pragmas): Ditto.
17170
046a838d 171712013-01-06 Jakub Jelinek <jakub@redhat.com>
17172
17173 * varasm.c (output_constant_def_contents): For asan_protect_global
17174 protected strings, adjust DECL_ALIGN if needed, before testing for
17175 anchored symbols.
17176 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17177 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
17178 normal decls.
17179 (output_object_block): For asan protected decls, emit asan padding
17180 after their contents.
17181 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17182 (asan_finish_file): Test it here instead.
17183
7a87c8e1 171842013-01-07 Nick Clifton <nickc@redhat.com>
17185 Matthias Klose <doko@debian.org>
17186 Doug Kwan <dougkwan@google.com>
17187 H.J. Lu <hongjiu.lu@intel.com>
99d5fe2d 17188
17189 PR driver/55470
17190 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17191
17192 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17193
17194 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17195
889d67a8 17196 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
99d5fe2d 17197
17198 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17199
67cec000 172002013-01-07 Georg-Johann Lay <avr@gjlay.de>
17201
5b804f86 17202 PR target/54461
67cec000 17203 * doc/install.texi (Cross-Compiler-Specific Options): Document
17204 --with-avrlibc.
17205
9de19dd2 172062013-01-07 Tejas Belagod <tejas.belagod@arm.com>
17207
17208 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17209 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17210 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17211 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17212 vqmovun_high_s64): Fix source operand number and update copyright.
17213
789a8d72 172142013-01-07 Richard Biener <rguenther@suse.de>
17215
17216 PR middle-end/55890
17217 * gimple.h (gimple_call_builtin_p): New overload.
17218 * gimple.c (validate_call): New function.
17219 (gimple_call_builtin_p): Likewise.
17220 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17221 Use gimple_call_builtin_p.
17222 (find_func_clobbers): Likewise.
17223 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
17224 (strlen_optimize_stmt): Likewise.
17225
71e72cc2 172262013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
17227
17228 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
17229 (vld1q_dup_*): Likewise.
17230 (vld1_*): Likewise.
17231 (vld1q_*): Likewise.
17232 (vld1_lane_*): Likewise.
17233 (vld1q_lane_*): Likewise.
17234
85fd8f96 172352013-01-07 Richard Biener <rguenther@suse.de>
17236
17237 * lto-streamer.h (LTO_minor_version): Bump to 2.
17238
72841352 172392013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
17240
17241 * config/aarch64/aarch64-protos.h
17242 (aarch64_const_double_zero_rtx_p): Rename to...
17243 (aarch64_float_const_zero_rtx_p): ...this.
17244 (aarch64_float_const_representable_p): New.
17245 (aarch64_output_simd_mov_immediate): Likewise.
17246 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
17247 move immediate case.
17248 * config/aarch64/aarch64.c
17249 (aarch64_const_double_zero_rtx_p): Rename to...
17250 (aarch64_float_const_zero_rtx_p): ...this.
17251 (aarch64_print_operand): Allow printing of new constants.
17252 (aarch64_valid_floating_const): New.
17253 (aarch64_legitimate_constant_p): Check for valid floating-point
17254 constants.
17255 (aarch64_simd_valid_immediate): Likewise.
17256 (aarch64_vect_float_const_representable_p): New.
17257 (aarch64_float_const_representable_p): Likewise.
17258 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
17259 (aarch64_output_simd_mov_immediate): New.
17260 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
17261 (*movdf_aarch64): Likewise.
17262 * config/aarch64/constraints.md (Ufc): New.
17263 (Y): call aarch64_float_const_zero_rtx.
17264 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
17265
4a101a28 172662013-01-07 Richard Biener <rguenther@suse.de>
17267
17268 PR tree-optimization/55888
17269 PR tree-optimization/55862
17270 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
17271 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
17272 not if it is contained therein.
17273
3b808880 172742013-01-07 Georg-Johann Lay <avr@gjlay.de>
17275
17276 * config/avr/t-avr: Typo.
17277
8d0c391e 172782013-01-07 Georg-Johann Lay <avr@gjlay.de>
17279
17280 PR55243
17281 * config/avr/t-avr: Don't automatically rebuild
17282 $(srcdir)/config/avr/t-multilib
17283 $(srcdir)/config/avr/avr-tables.opt
17284 $(srcdir)/doc/avr-mmcu.texi
17285 (avr-mcus): New phony target to build them on request.
17286 (s-avr-mlib, s-avr-mmcu-texi): Remove.
17287 * avr/avr-mcus.def: Adjust comments.
17288
e655d10e 172892013-01-07 Uros Bizjak <ubizjak@gmail.com>
17290
17291 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
17292
e2585731 172932013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
17294
17295 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
17296
e19a420f 172972013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
17298
17299 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
17300
eab80725 173012013-01-05 David Edelsohn <dje.gcc@gmail.com>
17302
17303 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
17304 to generate profiling.
17305 * config/rs6000/aix64.h (LIB_SPEC): Same.
17306
b49f35d1 173072013-01-04 Andrew Pinski <apinski@cavium.com>
17308
17309 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
17310 New function.
17311 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
17312
370a62b6 173132013-01-04 Uros Bizjak <ubizjak@gmail.com>
17314
17315 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
17316 unconditionally.
17317 (ix86_expand_move): Ditto.
17318 (ix86_zero_extend_to_Pmode): Ditto.
17319 (ix86_expand_call): Ditto.
17320 (ix86_expand_special_args_builtin): Ditto.
17321 (ix86_expand_builtin): Ditto.
17322
15b93bb9 173232013-01-04 Richard Biener <rguenther@suse.de>
17324
17325 PR tree-optimization/55862
17326 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
17327 translating them through PHI nodes.
17328
d210dffc 173292013-01-04 Martin Jambor <mjambor@suse.cz>
17330
17331 PR tree-optimization/55755
17332 * tree-sra.c (sra_modify_assign): Do not check that an access has no
17333 children when trying to avoid producing a VIEW_CONVERT_EXPR.
17334
9448cf4a 173352013-01-04 Marek Polacek <polacek@redhat.com>
17336
17337 PR middle-end/55859
17338 * opts.c (default_options_optimization): Clarify error message.
17339
65afafea 173402013-01-04 Richard Biener <rguenther@suse.de>
17341
17342 PR middle-end/55863
17343 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
17344 reassociation.
17345
dfb5e2bf 173462013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17347
17348 PR target/53789
17349 * config/pa/pa.md (movsi): Revert previous change.
a97fe562 17350 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
17351 references.
dfb5e2bf 17352
8c7f6993 173532013-01-03 Richard Henderson <rth@redhat.com>
17354
17355 * config/i386/i386.c (ix86_expand_move): Always assign to op1
17356 after eliminating TLS symbols.
17357
de2e6b8a 173582013-01-03 Marc Glisse <marc.glisse@inria.fr>
17359
17360 PR bootstrap/50167
17361 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
17362 * graphite-poly.c (debug_gmp_value): Likewise.
17363
a678c584 173642013-01-03 Uros Bizjak <ubizjak@gmail.com>
17365
17366 PR target/55712
17367 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
17368 selected code model, define __code_mode_small__, __code_model_medium__,
17369 __code_model_large__, __code_model_32__ or __code_model_kernel__.
17370 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
17371 xchg temporary register with %k. Declare temporary register as
17372 early clobbered.
17373 [__x86_64__]: For medium and large code models, preserve %rbx register.
17374
984cebc1 173752013-01-03 Richard Biener <rguenther@suse.de>
17376
98e7c1f5 17377 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
984cebc1 17378 (dump_subscript): Adjust.
17379 (finalize_ddr_dependent): Do not dump redundant info.
17380 (analyze_siv_subscript): Adjust.
17381 (subscript_dependence_tester): Likewise.
17382 (compute_affine_dependence): Likewise.
17383
6f75c845 173842013-01-03 Richard Biener <rguenther@suse.de>
17385
17386 Revert
17387 2013-01-03 Richard Biener <rguenther@suse.de>
17388
17389 PR tree-optimization/55857
17390 * tree-vect-stmts.c (vectorizable_load): Do not setup
17391 re-alignment for invariant loads.
17392
17393 2013-01-02 Richard Biener <rguenther@suse.de>
17394
17395 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
98e7c1f5 17396 invariant load do not generate a vector load from the scalar location.
6f75c845 17397
3c18ea71 173982013-01-03 Richard Biener <rguenther@suse.de>
17399
17400 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
17401 for not vectorizing.
17402 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
17403 not build INDIRECT_REFs, call get_name once only.
17404 (vect_create_data_ref_ptr): Likewise. Dump base object kind
17405 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
17406
59ff1f4a 174072013-01-03 Richard Biener <rguenther@suse.de>
17408
17409 PR tree-optimization/55857
17410 * tree-vect-stmts.c (vectorizable_load): Do not setup
17411 re-alignment for invariant loads.
17412
5b777cea 174132013-01-03 Richard Biener <rguenther@suse.de>
17414
17415 PR lto/55848
17416 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
17417 prefer a built-in decl.
17418
e4bef271 174192013-01-03 Jakub Jelinek <jakub@redhat.com>
17420
1f3894c9 17421 * gcc.c (process_command): Update copyright notice dates.
17422 * gcov.c (print_version): Likewise.
17423 * gcov-dump.c (print_version): Likewise.
17424
e4bef271 17425 PR rtl-optimization/55838
17426 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
17427 iv0.step, iv1.step and step.
17428
d871c5dc 174292013-01-03 Jakub Jelinek <jakub@redhat.com>
17430 Marc Glisse <marc.glisse@inria.fr>
17431
17432 PR tree-optimization/55832
17433 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
17434 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
17435 integer_{one,zero}_node.
17436
78ea005d 174372013-01-03 Jakub Jelinek <jakub@redhat.com>
17438
17439 PR debug/54402
17440 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
17441 * var-tracking.c (reverse_op): Don't add reverse ops to
17442 VALUEs that have already
98e7c1f5 17443 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
78ea005d 17444
df8f46da 174452013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
17446
17447 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
17448
f55775aa 174492013-01-02 Teresa Johnson <tejohnson@google.com>
17450
17451 * dumpfile.c (dump_loc): Print filename with location.
17452 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
17453 new location_t parameter to emit complete unroll message with
17454 new dump framework.
17455 (canonicalize_loop_induction_variables): Compute loops location
17456 and pass to try_unroll_loop_completely.
17457 * loop-unroll.c (report_unroll_peel): New function.
17458 (peel_loops_completely): Use new dump format with location
17459 for main dumpfile message, and invoke report_unroll_peel on success.
17460 (decide_unrolling_and_peeling): Ditto.
17461 (decide_peel_once_rolling): Remove old dumpfile message subsumed
17462 by report_unroll_peel.
17463 (decide_peel_completely): Ditto.
17464 (decide_unroll_constant_iterations): Ditto.
17465 (decide_unroll_runtime_iterations): Ditto.
17466 (decide_peel_simple): Ditto.
17467 (decide_unroll_stupid): Ditto.
17468 * cfgloop.c (get_loop_location): New function.
17469 * cfgloop.h (get_loop_location): Declare.
17470
13367f52 174712013-01-02 Sriraman Tallam <tmsriram@google.com>
17472
17473 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
17474 NULL.
17475
7b642468 174762013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17477
17478 PR middle-end/55198
17479 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
17480 BLKmode objects when EXPAND_MEMORY is specified.
17481
bad821ff 174822013-01-02 Sriraman Tallam <tmsriram@google.com>
17483
17484 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
17485 in loop predicate.
17486 (fold_builtin_cpu): Do not share cpu model decls across statements.
17487
c5cfcf40 174882013-01-02 Jason Merrill <jason@redhat.com>
17489
17490 PR c++/55804
17491 * tree.c (build_array_type_1): Revert earlier change.
17492
47509174 174932013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
17494
17495 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
17496 "cortex-a57".
17497 * config/aarch64/aarch64-tune.md: Re-generate.
17498
bca2a61e 174992013-01-02 Richard Biener <rguenther@suse.de>
17500
17501 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
98e7c1f5 17502 invariant load do not generate a vector load from the scalar location.
bca2a61e 17503
f67e1abf 175042013-01-02 Richard Biener <rguenther@suse.de>
17505
17506 PR bootstrap/55784
17507 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
17508 * configure: Regenerate.
17509
de2e453e 175102013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
17511
17512 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
17513 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
17514 (expand_builtin_int_roundingfn_2): Keep the original target around
17515 for the fallback case.
17516
ab4cb266 175172013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
17518
17519 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
17520 to be clear for sign changes.
17521
e876d531 175222013-01-01 Jan Hubicka <jh@suse.cz>
17523
17524 * ipa-inline-analysis.c: Fix formatting.
17525
c3c33891 175262013-01-01 Jakub Jelinek <jakub@redhat.com>
17527
17528 PR tree-optimization/55831
17529 * tree-vect-loop.c (get_initial_def_for_induction): Use
17530 gsi_after_labels instead of gsi_start_bb.
7dfbd804 17531\f
b3c5d881 17532Copyright (C) 2013 Free Software Foundation, Inc.
7dfbd804 17533
17534Copying and distribution of this file, with or without modification,
17535are permitted in any medium without royalty provided the copyright
17536notice and this notice are preserved.