1 2014-04-22 Sandra Loosemore <sandra@codesourcery.com>
3 * config/nios2/nios2.md (UNSPEC_ROUND): New.
5 * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
6 * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
7 * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
8 (nios2_fpu_insn): Add entry for round.
9 (N2FPU_NO_ERRNO_P): Define.
10 (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
12 * doc/invoke.texi (Nios II Options): Document -mcustom-round.
14 2014-04-22 Richard Henderson <rth@redhat.com>
16 * config/aarch64/aarch64 (addti3, subti3): New expanders.
17 (add<GPI>3_compare0): Remove leading * from name.
18 (add<GPI>3_carryin): Likewise.
19 (sub<GPI>3_compare0): Likewise.
20 (sub<GPI>3_carryin): Likewise.
21 (<su_optab>mulditi3): New expander.
22 (multi3): New expander.
23 (madd<GPI>): Remove leading * from name.
25 2014-04-22 Martin Jambor <mjambor@suse.cz>
27 * cgraphclones.c (cgraph_function_versioning): Copy
28 ipa_transforms_to_apply instead of asserting it is empty.
30 2014-04-22 H.J. Lu <hongjiu.lu@intel.com>
33 * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
34 on count_exp to get mode.
36 2014-04-22 Andrew Pinski <apinski@cavium.com>
38 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
40 * config/aarch64/aarch64.md (tlsie_small): Rename to ...
41 (tlsie_small_<mode>): this and handle PTR.
42 (tlsie_small_sidi): New pattern.
43 (tlsle_small): Change to an expand to handle ILP32.
44 (tlsle_small_<mode>): New pattern.
45 (tlsdesc_small): Rename to ...
46 (tlsdesc_small_<mode>): this and handle PTR.
48 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
50 * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
52 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
54 * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
55 (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
56 (aarch64_types_signed_poly_qualifiers): Likewise.
57 (aarch64_types_unsigned_signed_qualifiers): Likewise.
58 (aarch64_types_poly_signed_qualifiers): Likewise.
59 (TYPES_REINTERP_SS): Type macro added.
60 (TYPES_REINTERP_SU): Likewise.
61 (TYPES_REINTERP_SP): Likewise.
62 (TYPES_REINTERP_US): Likewise.
63 (TYPES_REINTERP_PS): Likewise.
64 (aarch64_fold_builtin): New expression folding added.
65 * config/aarch64/aarch64-simd-builtins.def (REINTERP):
67 (REINTERP_SS): Declarations added.
68 (REINTERP_US): Likewise.
69 (REINTERP_PS): Likewise.
70 (REINTERP_SU): Likewise.
71 (REINTERP_SP): Likewise.
72 * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
73 (vreinterpretq_p8_f64): Likewise.
74 (vreinterpret_p16_f64): Likewise.
75 (vreinterpretq_p16_f64): Likewise.
76 (vreinterpret_f32_f64): Likewise.
77 (vreinterpretq_f32_f64): Likewise.
78 (vreinterpret_f64_f32): Likewise.
79 (vreinterpret_f64_p8): Likewise.
80 (vreinterpret_f64_p16): Likewise.
81 (vreinterpret_f64_s8): Likewise.
82 (vreinterpret_f64_s16): Likewise.
83 (vreinterpret_f64_s32): Likewise.
84 (vreinterpret_f64_s64): Likewise.
85 (vreinterpret_f64_u8): Likewise.
86 (vreinterpret_f64_u16): Likewise.
87 (vreinterpret_f64_u32): Likewise.
88 (vreinterpret_f64_u64): Likewise.
89 (vreinterpretq_f64_f32): Likewise.
90 (vreinterpretq_f64_p8): Likewise.
91 (vreinterpretq_f64_p16): Likewise.
92 (vreinterpretq_f64_s8): Likewise.
93 (vreinterpretq_f64_s16): Likewise.
94 (vreinterpretq_f64_s32): Likewise.
95 (vreinterpretq_f64_s64): Likewise.
96 (vreinterpretq_f64_u8): Likewise.
97 (vreinterpretq_f64_u16): Likewise.
98 (vreinterpretq_f64_u32): Likewise.
99 (vreinterpretq_f64_u64): Likewise.
100 (vreinterpret_s64_f64): Likewise.
101 (vreinterpretq_s64_f64): Likewise.
102 (vreinterpret_u64_f64): Likewise.
103 (vreinterpretq_u64_f64): Likewise.
104 (vreinterpret_s8_f64): Likewise.
105 (vreinterpretq_s8_f64): Likewise.
106 (vreinterpret_s16_f64): Likewise.
107 (vreinterpretq_s16_f64): Likewise.
108 (vreinterpret_s32_f64): Likewise.
109 (vreinterpretq_s32_f64): Likewise.
110 (vreinterpret_u8_f64): Likewise.
111 (vreinterpretq_u8_f64): Likewise.
112 (vreinterpret_u16_f64): Likewise.
113 (vreinterpretq_u16_f64): Likewise.
114 (vreinterpret_u32_f64): Likewise.
115 (vreinterpretq_u32_f64): Likewise.
117 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
119 * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
120 * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
121 (vreinterpret_p8_s8): Likewise.
122 * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
123 (vreinterpret_p8_s16): Likewise.
124 (vreinterpret_p8_s32): Likewise.
125 (vreinterpret_p8_s64): Likewise.
126 (vreinterpret_p8_f32): Likewise.
127 (vreinterpret_p8_u8): Likewise.
128 (vreinterpret_p8_u16): Likewise.
129 (vreinterpret_p8_u32): Likewise.
130 (vreinterpret_p8_u64): Likewise.
131 (vreinterpret_p8_p16): Likewise.
132 (vreinterpretq_p8_s8): Likewise.
133 (vreinterpretq_p8_s16): Likewise.
134 (vreinterpretq_p8_s32): Likewise.
135 (vreinterpretq_p8_s64): Likewise.
136 (vreinterpretq_p8_f32): Likewise.
137 (vreinterpretq_p8_u8): Likewise.
138 (vreinterpretq_p8_u16): Likewise.
139 (vreinterpretq_p8_u32): Likewise.
140 (vreinterpretq_p8_u64): Likewise.
141 (vreinterpretq_p8_p16): Likewise.
142 (vreinterpret_p16_s8): Likewise.
143 (vreinterpret_p16_s16): Likewise.
144 (vreinterpret_p16_s32): Likewise.
145 (vreinterpret_p16_s64): Likewise.
146 (vreinterpret_p16_f32): Likewise.
147 (vreinterpret_p16_u8): Likewise.
148 (vreinterpret_p16_u16): Likewise.
149 (vreinterpret_p16_u32): Likewise.
150 (vreinterpret_p16_u64): Likewise.
151 (vreinterpret_p16_p8): Likewise.
152 (vreinterpretq_p16_s8): Likewise.
153 (vreinterpretq_p16_s16): Likewise.
154 (vreinterpretq_p16_s32): Likewise.
155 (vreinterpretq_p16_s64): Likewise.
156 (vreinterpretq_p16_f32): Likewise.
157 (vreinterpretq_p16_u8): Likewise.
158 (vreinterpretq_p16_u16): Likewise.
159 (vreinterpretq_p16_u32): Likewise.
160 (vreinterpretq_p16_u64): Likewise.
161 (vreinterpretq_p16_p8): Likewise.
162 (vreinterpret_f32_s8): Likewise.
163 (vreinterpret_f32_s16): Likewise.
164 (vreinterpret_f32_s32): Likewise.
165 (vreinterpret_f32_s64): Likewise.
166 (vreinterpret_f32_u8): Likewise.
167 (vreinterpret_f32_u16): Likewise.
168 (vreinterpret_f32_u32): Likewise.
169 (vreinterpret_f32_u64): Likewise.
170 (vreinterpret_f32_p8): Likewise.
171 (vreinterpret_f32_p16): Likewise.
172 (vreinterpretq_f32_s8): Likewise.
173 (vreinterpretq_f32_s16): Likewise.
174 (vreinterpretq_f32_s32): Likewise.
175 (vreinterpretq_f32_s64): Likewise.
176 (vreinterpretq_f32_u8): Likewise.
177 (vreinterpretq_f32_u16): Likewise.
178 (vreinterpretq_f32_u32): Likewise.
179 (vreinterpretq_f32_u64): Likewise.
180 (vreinterpretq_f32_p8): Likewise.
181 (vreinterpretq_f32_p16): Likewise.
182 (vreinterpret_s64_s8): Likewise.
183 (vreinterpret_s64_s16): Likewise.
184 (vreinterpret_s64_s32): Likewise.
185 (vreinterpret_s64_f32): Likewise.
186 (vreinterpret_s64_u8): Likewise.
187 (vreinterpret_s64_u16): Likewise.
188 (vreinterpret_s64_u32): Likewise.
189 (vreinterpret_s64_u64): Likewise.
190 (vreinterpret_s64_p8): Likewise.
191 (vreinterpret_s64_p16): Likewise.
192 (vreinterpretq_s64_s8): Likewise.
193 (vreinterpretq_s64_s16): Likewise.
194 (vreinterpretq_s64_s32): Likewise.
195 (vreinterpretq_s64_f32): Likewise.
196 (vreinterpretq_s64_u8): Likewise.
197 (vreinterpretq_s64_u16): Likewise.
198 (vreinterpretq_s64_u32): Likewise.
199 (vreinterpretq_s64_u64): Likewise.
200 (vreinterpretq_s64_p8): Likewise.
201 (vreinterpretq_s64_p16): Likewise.
202 (vreinterpret_u64_s8): Likewise.
203 (vreinterpret_u64_s16): Likewise.
204 (vreinterpret_u64_s32): Likewise.
205 (vreinterpret_u64_s64): Likewise.
206 (vreinterpret_u64_f32): Likewise.
207 (vreinterpret_u64_u8): Likewise.
208 (vreinterpret_u64_u16): Likewise.
209 (vreinterpret_u64_u32): Likewise.
210 (vreinterpret_u64_p8): Likewise.
211 (vreinterpret_u64_p16): Likewise.
212 (vreinterpretq_u64_s8): Likewise.
213 (vreinterpretq_u64_s16): Likewise.
214 (vreinterpretq_u64_s32): Likewise.
215 (vreinterpretq_u64_s64): Likewise.
216 (vreinterpretq_u64_f32): Likewise.
217 (vreinterpretq_u64_u8): Likewise.
218 (vreinterpretq_u64_u16): Likewise.
219 (vreinterpretq_u64_u32): Likewise.
220 (vreinterpretq_u64_p8): Likewise.
221 (vreinterpretq_u64_p16): Likewise.
222 (vreinterpret_s8_s16): Likewise.
223 (vreinterpret_s8_s32): Likewise.
224 (vreinterpret_s8_s64): Likewise.
225 (vreinterpret_s8_f32): Likewise.
226 (vreinterpret_s8_u8): Likewise.
227 (vreinterpret_s8_u16): Likewise.
228 (vreinterpret_s8_u32): Likewise.
229 (vreinterpret_s8_u64): Likewise.
230 (vreinterpret_s8_p8): Likewise.
231 (vreinterpret_s8_p16): Likewise.
232 (vreinterpretq_s8_s16): Likewise.
233 (vreinterpretq_s8_s32): Likewise.
234 (vreinterpretq_s8_s64): Likewise.
235 (vreinterpretq_s8_f32): Likewise.
236 (vreinterpretq_s8_u8): Likewise.
237 (vreinterpretq_s8_u16): Likewise.
238 (vreinterpretq_s8_u32): Likewise.
239 (vreinterpretq_s8_u64): Likewise.
240 (vreinterpretq_s8_p8): Likewise.
241 (vreinterpretq_s8_p16): Likewise.
242 (vreinterpret_s16_s8): Likewise.
243 (vreinterpret_s16_s32): Likewise.
244 (vreinterpret_s16_s64): Likewise.
245 (vreinterpret_s16_f32): Likewise.
246 (vreinterpret_s16_u8): Likewise.
247 (vreinterpret_s16_u16): Likewise.
248 (vreinterpret_s16_u32): Likewise.
249 (vreinterpret_s16_u64): Likewise.
250 (vreinterpret_s16_p8): Likewise.
251 (vreinterpret_s16_p16): Likewise.
252 (vreinterpretq_s16_s8): Likewise.
253 (vreinterpretq_s16_s32): Likewise.
254 (vreinterpretq_s16_s64): Likewise.
255 (vreinterpretq_s16_f32): Likewise.
256 (vreinterpretq_s16_u8): Likewise.
257 (vreinterpretq_s16_u16): Likewise.
258 (vreinterpretq_s16_u32): Likewise.
259 (vreinterpretq_s16_u64): Likewise.
260 (vreinterpretq_s16_p8): Likewise.
261 (vreinterpretq_s16_p16): Likewise.
262 (vreinterpret_s32_s8): Likewise.
263 (vreinterpret_s32_s16): Likewise.
264 (vreinterpret_s32_s64): Likewise.
265 (vreinterpret_s32_f32): Likewise.
266 (vreinterpret_s32_u8): Likewise.
267 (vreinterpret_s32_u16): Likewise.
268 (vreinterpret_s32_u32): Likewise.
269 (vreinterpret_s32_u64): Likewise.
270 (vreinterpret_s32_p8): Likewise.
271 (vreinterpret_s32_p16): Likewise.
272 (vreinterpretq_s32_s8): Likewise.
273 (vreinterpretq_s32_s16): Likewise.
274 (vreinterpretq_s32_s64): Likewise.
275 (vreinterpretq_s32_f32): Likewise.
276 (vreinterpretq_s32_u8): Likewise.
277 (vreinterpretq_s32_u16): Likewise.
278 (vreinterpretq_s32_u32): Likewise.
279 (vreinterpretq_s32_u64): Likewise.
280 (vreinterpretq_s32_p8): Likewise.
281 (vreinterpretq_s32_p16): Likewise.
282 (vreinterpret_u8_s8): Likewise.
283 (vreinterpret_u8_s16): Likewise.
284 (vreinterpret_u8_s32): Likewise.
285 (vreinterpret_u8_s64): Likewise.
286 (vreinterpret_u8_f32): Likewise.
287 (vreinterpret_u8_u16): Likewise.
288 (vreinterpret_u8_u32): Likewise.
289 (vreinterpret_u8_u64): Likewise.
290 (vreinterpret_u8_p8): Likewise.
291 (vreinterpret_u8_p16): Likewise.
292 (vreinterpretq_u8_s8): Likewise.
293 (vreinterpretq_u8_s16): Likewise.
294 (vreinterpretq_u8_s32): Likewise.
295 (vreinterpretq_u8_s64): Likewise.
296 (vreinterpretq_u8_f32): Likewise.
297 (vreinterpretq_u8_u16): Likewise.
298 (vreinterpretq_u8_u32): Likewise.
299 (vreinterpretq_u8_u64): Likewise.
300 (vreinterpretq_u8_p8): Likewise.
301 (vreinterpretq_u8_p16): Likewise.
302 (vreinterpret_u16_s8): Likewise.
303 (vreinterpret_u16_s16): Likewise.
304 (vreinterpret_u16_s32): Likewise.
305 (vreinterpret_u16_s64): Likewise.
306 (vreinterpret_u16_f32): Likewise.
307 (vreinterpret_u16_u8): Likewise.
308 (vreinterpret_u16_u32): Likewise.
309 (vreinterpret_u16_u64): Likewise.
310 (vreinterpret_u16_p8): Likewise.
311 (vreinterpret_u16_p16): Likewise.
312 (vreinterpretq_u16_s8): Likewise.
313 (vreinterpretq_u16_s16): Likewise.
314 (vreinterpretq_u16_s32): Likewise.
315 (vreinterpretq_u16_s64): Likewise.
316 (vreinterpretq_u16_f32): Likewise.
317 (vreinterpretq_u16_u8): Likewise.
318 (vreinterpretq_u16_u32): Likewise.
319 (vreinterpretq_u16_u64): Likewise.
320 (vreinterpretq_u16_p8): Likewise.
321 (vreinterpretq_u16_p16): Likewise.
322 (vreinterpret_u32_s8): Likewise.
323 (vreinterpret_u32_s16): Likewise.
324 (vreinterpret_u32_s32): Likewise.
325 (vreinterpret_u32_s64): Likewise.
326 (vreinterpret_u32_f32): Likewise.
327 (vreinterpret_u32_u8): Likewise.
328 (vreinterpret_u32_u16): Likewise.
329 (vreinterpret_u32_u64): Likewise.
330 (vreinterpret_u32_p8): Likewise.
331 (vreinterpret_u32_p16): Likewise.
332 (vreinterpretq_u32_s8): Likewise.
333 (vreinterpretq_u32_s16): Likewise.
334 (vreinterpretq_u32_s32): Likewise.
335 (vreinterpretq_u32_s64): Likewise.
336 (vreinterpretq_u32_f32): Likewise.
337 (vreinterpretq_u32_u8): Likewise.
338 (vreinterpretq_u32_u16): Likewise.
339 (vreinterpretq_u32_u64): Likewise.
340 (vreinterpretq_u32_p8): Likewise.
341 (vreinterpretq_u32_p16): Likewise.
343 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
345 * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
347 * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
350 * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
351 (vqnegd_s64): Likewise.
352 (vqabs_s64): Likewise.
353 (vqabsd_s64): Likewise.
355 2014-04-22 Richard Henderson <rth@redhat.com>
357 * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
358 computation to the top of the loop.
360 2014-04-22 Renlin <renlin.li@arm.com>
361 Jiong Wang <jiong.wang@arm.com>
363 * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
364 * config/aarch64/aarch64.c (aarch64_layout_frame)
365 (aarch64_initial_elimination_offset): Likewise.
367 2014-04-22 Marcus Shawcroft <marcus.shawcroft@arm.com>
369 * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
372 2014-04-22 Richard Sandiford <rdsandiford@googlemail.com>
374 * machmode.h (bitwise_mode_for_mode): Declare.
375 * stor-layout.h (bitwise_type_for_mode): Likewise.
376 * stor-layout.c (bitwise_mode_for_mode): New function.
377 (bitwise_type_for_mode): Likewise.
378 * builtins.c (fold_builtin_memory_op): Use it instead of
379 int_mode_for_mode and build_nonstandard_integer_type.
381 2014-04-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
383 * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
384 (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
385 (*-*-solaris2*): Simplify.
386 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
387 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
388 *-*-solaris2.9* handling.
390 * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
392 (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
393 (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
395 (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
396 * configure: Regenerate.
398 * config/i386/sol2-9.h: Remove.
400 * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
401 (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
402 Remove Solaris 9 references.
404 2014-04-22 Vidya Praveen <vidyapraveen@arm.com>
406 * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
407 (floatuns<GPI:mode><GPF:mode>2): Remove.
408 (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
409 and floatuns conversions.
410 (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
411 and floatuns conversions.
412 * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
413 (w1,w2): New mode attributes for inequal width conversions.
415 2014-04-22 Renlin Li <Renlin.Li@arm.com>
417 * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
418 the output asm format.
420 2014-04-22 James Greenhalgh <james.greenhalgh@arm.com>
422 * config/aarch64/aarch64-simd.md
423 (aarch64_cm<optab>di): Always split.
424 (*aarch64_cm<optab>di): New.
425 (aarch64_cmtstdi): Always split.
426 (*aarch64_cmtstdi): New.
428 2014-04-22 Jakub Jelinek <jakub@redhat.com>
430 PR tree-optimization/60823
431 * omp-low.c (ipa_simd_modify_function_body): Go through
432 all SSA_NAMEs and for those refering to vector arguments
433 which are going to be replaced adjust SSA_NAME_VAR and,
434 if it is a default definition, change it into a non-default
435 definition assigned at the beginning of function from new_decl.
436 (ipa_simd_modify_stmt_ops): Rewritten.
437 * tree-dfa.c (set_ssa_default_def): When removing default def,
438 check for NULL loc instead of NULL *loc.
440 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
442 * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
443 restrictions on core registers for DImode values in Thumb2.
445 2014-04-22 Ian Bolton <ian.bolton@arm.com>
447 * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
448 * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
450 2014-04-22 Ian Bolton <ian.bolton@arm.com>
452 * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
453 (*iordi_notzesidi_di): Likewise.
454 (*iordi_notsesidi_di): Likewise.
456 2014-04-22 Ian Bolton <ian.bolton@arm.com>
458 * config/arm/arm-protos.h (tune_params): New struct members.
459 * config/arm/arm.c: Initialise tune_params per processor.
460 (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
461 for speed, based on new tune_params.
463 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
465 * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
467 * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
469 * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
471 * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
472 * config/aarch64/arm_neon.h (vrnd_f64): Added.
473 (vrnda_f64): Likewise.
474 (vrndi_f64): Likewise.
475 (vrndm_f64): Likewise.
476 (vrndn_f64): Likewise.
477 (vrndp_f64): Likewise.
478 (vrndx_f64): Likewise.
480 2014-04-22 Zhenqiang Chen <zhenqiang.chen@linaro.org>
482 * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
483 GET_MODE_SIZE argument is enum machine_mode.
485 2014-04-22 Jakub Jelinek <jakub@redhat.com>
488 * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
489 value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
491 2014-04-22 Lin Zuojian <manjian2006@gmail.com>
494 * asan.c (asan_emit_stack_protection): Force the base to align to
495 appropriate bits if STRICT_ALIGNMENT. Set shadow_mem align to
496 appropriate bits if STRICT_ALIGNMENT.
497 * cfgexpand.c (expand_stack_vars): Set base_align appropriately
499 (expand_used_vars): Leave a space in the stack frame for alignment
502 2014-04-21 David Malcolm <dmalcolm@redhat.com>
504 * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
506 (gimple_store_p): Likewise.
507 (gimple_assign_load_p): Likewise.
508 (gimple_assign_cast_p): Likewise.
509 (gimple_clobber_p): Likewise.
511 * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
512 rather than a gimple.
513 (gimple_assign_cast_p): Likewise.
515 2014-04-21 Michael Meissner <meissner@linux.vnet.ibm.com>
518 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
519 If mode is DDmode and TARGET_E500_DOUBLE allow move.
521 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
522 more debug information for E500 if -mdebug=reg.
524 2014-04-21 Uros Bizjak <ubizjak@gmail.com>
527 * config/i386/i386.c (ix86_expand_builtin)
528 <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
529 register for target RTX.
530 <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
532 2014-04-18 Cong Hou <congh@google.com>
534 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
535 the widen-mult pattern by handling two operands with different sizes,
536 and operands whose size is smaller than half of the result type.
538 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
540 * ipa-inline.h (INLINE_HINT_known_hot): New hint.
541 * ipa-inline-analysis.c (dump_inline_hints): Dump it.
542 (do_estimate_edge_time): Compute it.
543 * ipa-inline.c (want_inline_small_function_p): Bypass
544 INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
546 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
548 * ipa-inline.c (spec_rem): New static variable.
549 (dump_overall_stats): New function.
550 (dump_inline_stats): New function.
552 2014-04-18 Richard Henderson <rth@redhat.com>
554 * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
555 to GET_MODE_SIZE, not a reg_class_t.
557 2014-04-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
559 * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
560 (vsx_xxmrglw_<mode>): Likewise.
562 2014-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
565 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
566 GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
567 (rs6000_init_hard_regno_mode_ok): Likewise.
569 2014-04-17 Jan Hubicka <hubicka@ucw.cz>
571 * ipa-inline.c (inline_small_functions): Account only non-cold
573 * doc/invoke.texi (inline-unit-growth): Update documentation.
575 2014-04-17 Pat Haugen <pthaugen@us.ibm.com>
577 * config/rs6000/rs6000.md (addti3, subti3): New.
579 2014-04-17 H.J. Lu <hongjiu.lu@intel.com>
582 * config/i386/i386.c (ix86_expand_clear): Remove outdated
583 comment. Check optimize_insn_for_size_p instead of
584 optimize_insn_for_speed_p.
586 2014-04-17 Martin Jambor <mjambor@suse.cz>
588 * gimple-iterator.c (gsi_start_edge): New function.
589 * gimple-iterator.h (gsi_start_edge): Declare.
590 * tree-sra.c (single_non_eh_succ): New function.
591 (disqualify_ops_if_throwing_stmt): Renamed to
592 disqualify_if_bad_bb_terminating_stmt. Allow throwing statements
593 having one non-EH successor BB.
594 (sra_modify_expr): If stmt ends bb, use single non-EH successor to
595 generate loads into replacements.
596 (sra_modify_assign): Likewise and and also use the simple path for
598 (sra_modify_function_body): Commit statements on edges.
600 2014-04-17 Richard Biener <rguenther@suse.de>
603 * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
604 comparison results and add clarifying comment.
606 2014-04-17 Jakub Jelinek <jakub@redhat.com>
608 * genmodes.c (struct mode_data): Add need_bytesize_adj field.
609 (blank_mode): Initialize it.
610 (emit_mode_size_inline, emit_mode_nunits_inline,
611 emit_mode_inner_inline): New functions.
612 (emit_insn_modes_h): Call them and surround their output with
613 #if GCC_VERSION >= 4001 ... #endif.
614 * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
615 For GCC_VERSION >= 4001 use mode_*_inline routines instead of
616 mode_* arrays if the argument is __builtin_constant_p.
617 * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
618 is enum machine_mode.
620 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
622 * passes.c (opt_pass::execute): Adjust.
623 (pass_manager::execute_pass_mode_switching): Likewise.
624 (early_local_passes::execute): Likewise.
625 (execute_one_pass): Pass cfun to the pass's execute method.
626 * tree-pass.h (opt_pass::execute): Add function * argument.
627 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
628 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
629 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
630 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
631 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
632 config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
633 except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
634 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
635 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
636 ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
637 ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
638 lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
639 postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
640 reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
641 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
642 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
643 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
644 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
645 tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
646 tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
647 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
648 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
649 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
650 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
651 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
652 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
653 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
654 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
655 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
658 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
660 * passes.c (opt_pass::gate): Take function * argument.
661 (gate_all_early_local_passes): Merge into
662 (early_local_passes::gate): this.
663 (gate_all_early_optimizations): Merge into
664 (all_early_optimizations::gate): this.
665 (gate_all_optimizations): Mege into
666 (all_optimizations::gate): this.
667 (gate_all_optimizations_g): Merge into
668 (all_optimizations_g::gate): this.
669 (gate_rest_of_compilation): Mege into
670 (rest_of_compilation::gate): this.
671 (gate_postreload): Merge into
672 (postreload::gate): this.
673 (dump_one_pass): Pass cfun to the pass's gate method.
674 (execute_ipa_summary_passes): Likewise.
675 (execute_one_pass): Likewise.
676 (ipa_write_summaries_2): Likewise.
677 (ipa_write_optimization_summaries_1): Likewise.
678 (ipa_read_summaries_1): Likewise.
679 (ipa_read_optimization_summaries_1): Likewise.
680 (execute_ipa_stmt_fixups): Likewise.
681 * tree-pass.h (opt_pass::gate): Add function * argument.
682 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
683 combine-stack-adj.c, combine.c, compare-elim.c,
684 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
685 config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
686 config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
687 dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c, fwprop.c, gcse.c,
688 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
689 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
690 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
691 loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
692 omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
693 reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
694 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
695 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
696 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
697 tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
698 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
699 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
700 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
701 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
702 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
703 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
704 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
705 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
706 tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
707 tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
708 var-tracking.c, vtable-verify.c, web.c: Adjust.
710 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
712 * configure.ac: Check for -Woverloaded-virtual and enable it if found.
713 * configure: Regenerate.
715 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
717 * passes.c (dump_one_pass): don't check pass->has_gate.
718 (execute_ipa_summary_passes): Likewise.
719 (execute_one_pass): Likewise.
720 (ipa_write_summaries_2): Likewise.
721 (ipa_write_optimization_summaries_1): Likewise.
722 (ipa_read_optimization_summaries_1): Likewise.
723 (execute_ipa_stmt_fixups): Likewise.
724 * tree-pass.h (pass_data::has_gate): Remove.
725 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
726 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
727 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
728 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
729 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
730 config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
731 config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
732 dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
733 gimple-low.c, gimple-ssa-isolate-paths.c,
734 gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
735 ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
736 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
737 ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
738 modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
739 recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
740 reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
741 trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
742 tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
743 tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
744 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
745 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
746 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
747 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
748 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
749 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
750 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
751 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
752 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
753 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
754 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
755 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
758 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
760 * pass_manager.h (pass_manager::register_dump_files_1): Remove
762 * passes.c (pass_manager::register_dump_files_1): Merge into
763 (pass_manager::register_dump_files): this, and remove its handling of
764 properties since the pass always has the properties anyway.
765 (pass_manager::pass_manager): Adjust.
767 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
769 * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
770 * passes.c (pass_manager::register_dump_files_1): Remove dead code
771 dealing with properties.
772 (pass_manager::register_dump_files): Adjust.
774 2014-03-20 Mark Wielaard <mjw@redhat.com>
776 * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
777 then represent the bound as normal constant value.
779 2014-04-17 Jakub Jelinek <jakub@redhat.com>
782 Forward port from 4.8 branch
783 2013-07-19 Kirill Yukhin <kirill.yukhin@intel.com>
785 * config/i386/bmiintrin.h (_blsi_u32): New.
789 (_blsmsk_u32): Ditto.
790 (_blsmsk_u64): Ditto.
794 2014-04-17 Kito Cheng <kito@0xlab.org>
796 * gcc.c (used_arg): Prevent out of bound access for multilib_options.
798 2014-04-17 Richard Biener <rguenther@suse.de>
801 * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
802 boolean results for comparisons.
804 2014-04-17 Richard Biener <rguenther@suse.de>
806 PR tree-optimization/60836
807 * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
808 initial PHI args to be gimple values.
810 2014-04-17 Richard Biener <rguenther@suse.de>
812 PR tree-optimization/60841
813 * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
814 * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
815 of stmts to SLP build.
816 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
817 (vect_analyze_slp): Likewise.
818 (vect_analyze_slp_instance): Likewise.
819 (vect_build_slp_tree): Limit overall SLP tree growth.
820 * tree-vectorizer.h (vect_analyze_data_refs,
821 vect_analyze_slp): Adjust prototypes.
823 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
825 * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
828 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
830 * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
831 * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
832 * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
833 for TARGET_SLOW_PSHUFB
835 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
837 * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
838 * config/i386/i386.c (intel_cost): Ditto.
840 2014-04-17 Joey Ye <joey.ye@arm.com>
842 * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
844 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
846 * opts.c (common_handle_option): Disable -fipa-reference coorectly
849 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
851 * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
852 (type_all_derivations_known_p): New predicate.
853 (type_all_ctors_visible_p): New predicate.
854 (type_possibly_instantiated_p): New predicate.
855 (get_odr_type): Compute all_derivations_known.
856 (dump_odr_type): Dump the flag.
857 (maybe_record_type): Cleanup.
858 (record_target_from_binfo): Add bases_to_consider array;
859 record bases for types w/o instances and skip CXX destructor.
860 (possible_polymorphic_call_targets_1): Add bases_to_consider
861 and consider_construction parameters; check if type may have instance.
862 (get_polymorphic_call_info): Set maybe_in_construction to true
863 when we know nothing.
864 (record_targets_from_bases): Skip CXX destructors; they are
865 never called for types in construction.
866 (possible_polymorphic_call_targets): Do not record target when
867 type may not have instance.
869 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
872 * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
873 external aliases alive, too.
875 2014-04-16 Andrew Pinski <apinski@cavium.com>
877 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
880 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
882 * final.c (compute_alignments): Do not apply loop alignment to a block
883 falling through to the exit.
885 2014-04-16 Catherine Moore <clm@codesourcery.com>
887 * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
888 Adjust constraints for microMIPS store patterns.
890 2014-04-16 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
892 * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
894 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
896 * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
897 (append_use): Run at -O0.
898 (append_vdef): Likewise.
899 * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
900 * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
902 2014-04-16 Jakub Jelinek <jakub@redhat.com>
904 PR tree-optimization/60844
905 * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
906 (propagate_op_to_single_use, remove_visited_stmt_chain,
907 linearize_expr, repropagate_negates, reassociate_bb): Use it
908 instead of gsi_remove.
910 2014-04-16 Martin Jambor <mjambor@suse.cz>
912 * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
913 ipa_transforms_to_apply.
914 (cgraph_function_versioning): Assert that old_node has empty
915 ipa_transforms_to_apply.
916 * trans-mem.c (ipa_tm_create_version): Likewise.
917 * tree-inline.c (tree_function_versioning): Do not duplicate
918 ipa_transforms_to_apply.
920 2014-04-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
923 * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
925 Pass necessary as flags on 64-bit Solaris/x86.
926 Use lowercase relocs for x86_64-*-*.
927 * configure: Regenerate.
929 2014-04-15 Jan Hubicka <jh@suse.cz>
931 * ipa-devirt.c (referenced_from_vtable_p): New predicate.
932 (maybe_record_node, likely_target_p): Use it.
934 2014-04-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
937 Revert following patch
939 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
942 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
943 software floating point or no floating point registers, do not
944 allow any type in the FPRs. Eliminate a test for SPE SIMD types
945 in GPRs that occurs after we tested for GPRs that would never be
948 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
949 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
950 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
951 specifically allow DDmode, since that does not use the SPE SIMD
954 2014-03-21 Mark Wielaard <mjw@redhat.com>
956 * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
957 as unsigned or int depending on type and value used.
959 2014-04-15 Richard Biener <rguenther@suse.de>
961 PR rtl-optimization/56965
962 * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
963 * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
965 * alias.c (true_dependence_1): Do not call
966 nonoverlapping_component_refs_p.
967 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
968 nonoverlapping_component_refs_p.
969 (indirect_refs_may_alias_p): Likewise.
971 2014-04-15 Teresa Johnson <tejohnson@google.com>
973 * cfg.c (dump_bb_info): Fix flags check.
974 * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
976 2014-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
978 PR rtl-optimization/60663
979 * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
982 2014-04-15 Richard Biener <rguenther@suse.de>
984 * lto-streamer.h (LTO_major_version): Bump to 4.
986 2014-04-15 Richard Biener <rguenther@suse.de>
988 * common.opt (lto_partition_model): New enum.
989 (flto-partition=): Merge separate options with a single with argument,
990 add -flto-partition=one support.
991 * flag-types.h (enum lto_partition_model): Declare.
992 * opts.c (finish_options): Remove duplicate -flto-partition=
994 * lto-wrapper.c (run_gcc): Adjust.
996 2014-04-15 Richard Biener <rguenther@suse.de>
998 * alias.c (ncr_compar): New function.
999 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
1001 2014-04-15 Richard Biener <rguenther@suse.de>
1003 * alias.c (record_component_aliases): Do not walk BINFOs.
1005 2014-04-15 Richard Biener <rguenther@suse.de>
1007 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1008 Add struct function argument and adjust.
1009 (find_func_aliases_for_call): Likewise.
1010 (find_func_aliases): Likewise.
1011 (find_func_clobbers): Likewise.
1012 (intra_create_variable_infos): Likewise.
1013 (compute_points_to_sets): Likewise.
1014 (ipa_pta_execute): Adjust. Do not push/pop cfun.
1016 2014-04-15 Richard Biener <rguenther@suse.de>
1018 * tree.c (iterative_hash_expr): Use enum tree_code_class
1019 to store TREE_CODE_CLASS.
1020 (tree_block): Likewise.
1021 (tree_set_block): Likewise.
1022 * tree.h (fold_build_pointer_plus_loc): Use
1023 convert_to_ptrofftype_loc.
1025 2014-04-15 Jakub Jelinek <jakub@redhat.com>
1028 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
1031 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
1033 * cfgloop.h (struct loop): Move force_vectorize down.
1034 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
1035 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
1036 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
1037 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
1038 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
1039 * tree-core.h (enum annot_expr_kind): Add new kind values.
1040 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
1041 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
1043 * tree.def (ANNOTATE_EXPR): Tweak comment.
1045 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1047 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
1050 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
1052 * tree.h (TYPE_IDENTIFIER): Declare.
1053 * tree.c (subrange_type_for_debug_p): Use it.
1054 * godump.c (go_format_type): Likewise.
1055 * dwarf2out.c (is_cxx_auto, modified_type_die,
1056 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
1057 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
1059 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1062 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
1064 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
1066 * config/i386/i386.c (examine_argument): Return bool. Return true if
1067 parameter should be passed in memory.
1068 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
1069 (construct_container): Update calls to examine_argument.
1070 (function_arg_advance_64): Ditto.
1071 (return_in_memory_32): Merge with ix86_return_in_memory.
1072 (return_in_memory_64): Ditto.
1073 (return_in_memory_ms_64): Ditto.
1075 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1077 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
1078 * coverage.c (coverage_compute_profile_id): Handle externally visible
1081 2014-04-14 Martin Jambor <mjambor@suse.cz>
1083 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
1084 DECL_DISREGARD_INLINE_LIMITS functions.
1086 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
1089 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
1091 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
1094 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
1095 optimize_insn_for_speed_p instead of
1096 optimize_function_for_speed_p.
1098 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
1100 * doc/invoke.texi (free): Document AArch64.
1102 2014-04-14 Richard Biener <rguenther@suse.de>
1104 PR tree-optimization/60042
1105 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
1106 (insert_into_preds_of_block): Do not prevent PHI insertion
1107 for REFERENCE exprs here ...
1108 (eliminate_dom_walker::before_dom_children): ... but prevent
1109 their use here under similar conditions when applied to the
1110 IL after PRE optimizations.
1112 2014-04-14 Richard Biener <rguenther@suse.de>
1114 * passes.def: Move early points-to after early SRA.
1116 2014-04-14 Richard Biener <rguenther@suse.de>
1118 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
1119 check for which sign-changes we allow when forwarding
1120 a converted value into a switch.
1122 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
1124 * stor-layout.c (place_field): Finalize non-constant offset for the
1127 2014-04-14 Richard Biener <rguenther@suse.de>
1129 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
1131 (expand_switch_using_bit_tests_p): Likewise.
1132 (process_switch): Compute and pass on speed_p based on the
1134 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
1135 optimize_bb_for_speed_p.
1137 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
1139 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
1140 * function.h (struct function): Rename has_force_vect_loops into
1141 has_force_vectorize_loops.
1142 * lto-streamer-in.c (input_cfg): Adjust for renaming.
1143 (input_struct_function_base): Likewise.
1144 * lto-streamer-out.c (output_cfg): Likewise.
1145 (output_struct_function_base): Likewise.
1146 * omp-low.c (expand_omp_simd): Likewise.
1147 * tree-cfg.c (move_sese_region_to_fn): Likewise.
1148 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1149 (version_loop_for_if_conversion): Likewise.
1150 (tree_if_conversion): Likewise.
1151 (main_tree_if_conversion): Likewise.
1152 (gate_tree_if_conversion): Likewise.
1153 * tree-inline.c (copy_loops): Likewise.
1154 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
1155 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
1156 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
1157 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
1158 * tree-vectorizer.c (vectorize_loops): Likewise.
1159 * tree-vectorizer.h (unlimited_cost_model): Likewise.
1161 2014-04-14 Richard Biener <rguenther@suse.de>
1164 * lto-streamer-out.c (wrap_refs): New function.
1165 (lto_output): Wrap symbol references in global initializes in
1166 type-preserving MEM_REFs.
1168 2014-04-14 Christian Bruel <christian.bruel@st.com>
1170 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
1172 2014-04-14 Christian Bruel <christian.bruel@st.com>
1174 * config/sh/sh.md (setmemqi): New expand pattern.
1175 * config/sh/sh.h (CLEAR_RATIO): Define.
1176 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
1177 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
1179 2014-04-14 Richard Biener <rguenther@suse.de>
1182 * fold-const.c (negate_expr_p): Don't negate directional rounding
1184 (fold_negate_expr): Likewise.
1186 2014-04-14 Richard Biener <rguenther@suse.de>
1188 PR tree-optimization/59817
1189 PR tree-optimization/60453
1190 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
1191 recursion to catch all CHRECs in the scalar evolution and restrict
1192 the predicate for the remains appropriately.
1194 2014-04-12 Catherine Moore <clm@codesourcery.com>
1196 * config/mips/constraints.md: Add new register constraint "kb".
1197 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
1198 (*movhi_internal): Likewise.
1199 (*movqi_internal): Likewise.
1200 * config/mips/mips.h (M16_STORE_REGS): New register class.
1201 (REG_CLASS_NAMES): Add M16_STORE_REGS.
1202 (REG_CLASS_CONTENTS): Likewise.
1203 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
1205 2014-04-11 Tobias Burnus <burnus@net-b.de>
1208 * doc/invoke.texi (-Wformat-signedness): Document it.
1209 (Wformat=2): Mention that this enables -Wformat-signedness.
1211 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1213 * common/config/epiphany/epiphany-common.c
1214 (epiphany_option_optimization_table): Enable section anchors by
1215 default at -O1 or higher.
1216 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
1217 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
1218 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
1219 carries no extra cost.
1220 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
1221 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
1222 * config/epiphany/predicates.md (memclob_operand): New predicate.
1223 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
1224 Use memclob_operand predicate and X constraint for operand 3.
1226 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1228 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
1229 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
1232 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1234 PR rtl-optimization/60651
1235 * mode-switching.c (optimize_mode_switching): Make sure to emit
1236 sets of a lower numbered entity before sets of a higher numbered
1237 entity to a mode of the same or lower priority.
1238 When creating a seginfo for a basic block that starts with a code
1239 label, move the insertion point past the code label.
1240 (new_seginfo): Document and enforce requirement that
1241 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
1242 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
1243 * doc/tm.texi: Regenerate.
1245 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
1248 * config/arc/arc.c (arc_save_restore): Fix assert typo.
1250 2013-04-11 Jakub Jelinek <jakub@redhat.com>
1252 * BASE-VER: Set to 4.10.0.
1254 2014-04-11 Tobias Burnus <burnus@net-b.de>
1257 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
1258 * doc/gcc.texi (Service): Update description in the @menu
1259 * doc/invoke.texi (Option Summary): Remove misplaced and
1262 2014-04-11 Steve Ellcey <sellcey@mips.com>
1263 Jakub Jelinek <jakub@redhat.com>
1266 * expr.c (convert_move): Use emit_store_flag_force instead of
1267 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
1270 2014-04-11 Richard Biener <rguenther@suse.de>
1273 * varasm.c (assemble_alias): Avoid endless error reporting
1274 recursion by setting TREE_ASM_WRITTEN.
1276 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1278 * config/s390/s390.md: Add a splitter for NOT rtx.
1280 2014-04-11 Jakub Jelinek <jakub@redhat.com>
1282 PR rtl-optimization/60663
1283 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
1285 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
1286 Jakub Jelinek <jakub@redhat.com>
1289 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
1290 flag from decl_node to node.
1292 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1295 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
1296 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
1297 ameliorating the cases where it can be.
1299 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
1302 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
1304 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1305 (loadsync_<mode>): Change mode.
1306 (load_quadpti, store_quadpti): New.
1307 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1308 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1309 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
1311 2014-04-09 Cong Hou <congh@google.com>
1314 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
1317 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1319 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
1320 instead of vnor to exploit possible fusion opportunity in the
1322 (altivec_expand_vec_perm_const_le): Likewise.
1324 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
1326 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1327 (loadsync_<mode>): Change mode.
1328 (load_quadpti, store_quadpti): New.
1329 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1330 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1332 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
1335 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
1336 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
1337 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
1339 2014-04-08 Richard Biener <rguenther@suse.de>
1342 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
1343 a 64bit widest int print double-int similar to on HWI64 hosts.
1345 2014-04-08 Richard Biener <rguenther@suse.de>
1347 PR tree-optimization/60785
1348 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
1349 default defs properly.
1351 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
1353 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
1354 (Weffc++): Likewise.
1356 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
1358 * ipa-devirt.c (maybe_record_node): When node is not recorded,
1359 set completep to false rather than true.
1361 2014-04-07 Douglas B Rupp <rupp@adacore.com>
1364 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
1365 ARM_TARGET2_DWARF_FORMAT.
1367 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
1370 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
1371 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
1374 2014-04-07 Richard Biener <rguenther@suse.de>
1376 PR tree-optimization/60766
1377 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
1378 (may_eliminate_iv): Convert cand_value_at result to desired type.
1380 2014-04-07 Jason Merrill <jason@redhat.com>
1383 * common.opt (-fno-gnu-unique): Add.
1384 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
1386 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1388 * haifa-sched.c: Fix outdated function reference and minor
1389 grammar errors in introductory comment.
1391 2014-04-07 Richard Biener <rguenther@suse.de>
1394 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
1396 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
1398 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
1401 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
1402 size accounting for thunks.
1403 (pa_asm_output_mi_thunk): Use final_start_function() and
1404 final_end_function() to output function start and end directives.
1406 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
1408 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
1409 device specific ISA/ feature information. Remove short_sp and
1410 errata_skip ds. Add avr_device_specific_features enum to have device
1412 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
1413 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
1414 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
1415 updated device specific info.
1416 * config/avr/avr-mcus.def: Merge device specific details to
1417 dev_attribute field.
1418 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
1420 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
1421 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
1422 assembler if RMW isa supported by current device.
1423 * config/avr/genmultilib.awk: Update as device info structure changed.
1424 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
1426 2014-04-04 Cong Hou <congh@google.com>
1428 PR tree-optimization/60656
1429 * tree-vect-stmts.c (supportable_widening_operation):
1430 Fix a bug that elements in a vector with vect_used_by_reduction
1431 property are incorrectly reordered when the operation on it is not
1432 consistant with the one in reduction operation.
1434 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
1436 PR rtl-optimization/60155
1437 * gcse.c (record_set_data): New function.
1438 (single_set_gcse): New function.
1439 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
1440 (hoist_code): Likewise.
1441 (get_pressure_class_and_nregs): Likewise.
1443 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
1445 * explow.c (probe_stack_range): Emit a final optimization blockage.
1447 2014-04-04 Anthony Green <green@moxielogic.com>
1449 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
1452 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
1455 * lto-cgraph.c (input_overwrite_node): Check that partitioning
1456 flags are set only during streaming.
1457 * ipa.c (process_references, walk_polymorphic_call_targets,
1458 symtab_remove_unreachable_nodes): Drop bodies of always inline
1459 after early inlining.
1460 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
1462 2014-04-04 Jakub Jelinek <jakub@redhat.com>
1463 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1466 * dwarf2out.c (const_ok_for_output_1): Reject expressions
1469 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1472 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
1474 (cortex_a53_fdivd): Likewise.
1476 2014-04-04 Martin Jambor <mjambor@suse.cz>
1479 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
1481 * cgraph.c (clone_of_p): Also return true if thunks match.
1482 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
1483 cgraph_function_or_thunk_node and an obsolete comment.
1484 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
1486 (build_function_decl_skip_args): Likewise.
1487 (set_new_clone_decl_and_node_flags): New function.
1488 (duplicate_thunk_for_node): Likewise.
1489 (redirect_edge_duplicating_thunks): Likewise.
1490 (cgraph_clone_node): New parameter args_to_skip, pass it to
1491 redirect_edge_duplicating_thunks which is called instead of
1492 cgraph_redirect_edge_callee.
1493 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
1494 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
1496 2014-04-04 Jeff Law <law@redhat.com>
1499 * config/arm/predicates.md (const_int_I_operand): New predicate.
1500 (const_int_M_operand): Similarly.
1501 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
1503 (insv_t2, extv_reg, extzv_t2): Likewise.
1504 (load_multiple_with_writeback): Similarly for const_int_I_operand.
1505 (pop_multiple_with_writeback_and_return): Likewise.
1506 (vfp_pop_multiple_with_writeback): Likewise
1508 2014-04-04 Richard Biener <rguenther@suse.de>
1511 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
1512 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
1514 * gimplify.h (gimple_add_tmp_var_fn): Declare.
1515 * gimplify.c (gimple_add_tmp_var_fn): New function.
1516 * gimple-expr.h (create_tmp_reg_fn): Declare.
1517 * gimple-expr.c (create_tmp_reg_fn): New function.
1518 * gimple-low.c (record_vars_into): Don't change cfun.
1519 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
1520 code generation without cfun.
1522 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
1525 * Makefile.in (install-driver): Fix shell scripting.
1527 2014-04-03 Cong Hou <congh@google.com>
1529 PR tree-optimization/60505
1530 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
1531 threshold of number of iterations below which no vectorization
1533 * tree-vect-loop.c (new_loop_vec_info):
1534 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
1535 * tree-vect-loop.c (vect_analyze_loop_operations):
1536 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
1537 * tree-vect-loop.c (vect_transform_loop):
1538 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
1539 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
1540 of iterations of the loop and see if we should build the epilogue.
1542 2014-04-03 Richard Biener <rguenther@suse.de>
1544 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
1545 (streamer_tree_cache_create): Adjust.
1546 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
1547 to allow optional nodes array.
1548 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
1549 (streamer_tree_cache_append): Likewise.
1550 (streamer_tree_cache_create): Create nodes array optionally
1551 as specified by parameter.
1552 * lto-streamer-out.c (create_output_block): Avoid maintaining
1553 the node array in the writer cache.
1554 (DFS_write_tree): Remove assertion.
1555 (produce_asm_for_decls): Free the out decl state hash table early.
1556 * lto-streamer-in.c (lto_data_in_create): Adjust for
1557 streamer_tree_cache_create prototype change.
1559 2014-04-03 Richard Biener <rguenther@suse.de>
1561 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
1562 set TREE_CHAIN to NULL_TREE.
1564 2014-04-03 Richard Biener <rguenther@suse.de>
1566 PR tree-optimization/60740
1567 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
1568 over all GIMPLE_COND operands.
1570 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
1572 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
1573 (Weffc++): Remove Scott's numbering, merge lists and reference
1576 2014-04-03 Nick Clifton <nickc@redhat.com>
1578 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
1581 2014-04-03 Martin Jambor <mjambor@suse.cz>
1583 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
1584 mention gcc_unreachable before failing.
1585 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
1588 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
1591 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
1592 inconsistent code and instead mark the context inconsistent.
1593 (possible_polymorphic_call_targets): For inconsistent contexts
1594 return empty complete list.
1596 2014-04-02 Anthony Green <green@moxielogic.com>
1598 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
1599 (extendqisi2, extendhisi2): Define.
1600 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
1601 (WCHAR_TYPE): Change to unsigned int.
1603 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1605 PR tree-optimization/60733
1606 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
1607 insertion point for PHI candidates to be the end of the feeding
1608 block for the PHI argument.
1610 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
1612 PR rtl-optimization/60650
1613 * lra-constraints.c (process_alt_operands): Decrease reject for
1614 earlyclobber matching.
1616 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1618 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
1620 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1622 * config/spu/spu.c (pad_bb): Do not crash when the last
1623 insn is CODE_FOR_blockage.
1625 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1627 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
1628 lies outside the target mode.
1630 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
1633 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1634 software floating point or no floating point registers, do not
1635 allow any type in the FPRs. Eliminate a test for SPE SIMD types
1636 in GPRs that occurs after we tested for GPRs that would never be
1639 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1640 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1641 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
1642 specifically allow DDmode, since that does not use the SPE SIMD
1645 2014-04-02 Richard Biener <rguenther@suse.de>
1648 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
1649 MODE_INTs. Properly use negv_optab.
1650 (expand_abs): Likewise.
1652 2014-04-02 Richard Biener <rguenther@suse.de>
1655 * Makefile.in (install-driver): Guard extra installs with special
1658 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1660 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1663 2014-04-01 Richard Henderson <rth@redhat.com>
1666 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
1667 alternative enabled before register allocation.
1669 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
1671 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
1672 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
1674 (nios2_large_got_address): Remove unneeded 'sym' parameter.
1675 (nios2_got_address): Update nios2_large_got_address call site.
1676 (nios2_delegitimize_address): New function.
1677 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
1678 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
1679 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
1681 2014-04-01 Martin Husemann <martin@duskware.de>
1683 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
1686 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
1688 PR rtl-optimization/60604
1689 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
1690 check from register_operand.
1691 (register_operand): Redefine in terms of general_operand.
1692 (nonmemory_operand): Use register_operand for the non-constant cases.
1694 2014-04-01 Richard Biener <rguenther@suse.de>
1696 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
1698 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
1700 * doc/invoke.texi (mapp-regs): Clarify.
1702 2014-03-31 Ulrich Drepper <drepper@gmail.com>
1704 * config/i386/avx512fintrin.h (__v32hi): Define type.
1705 (__v64qi): Likewise.
1706 (_mm512_set1_epi8): Define.
1707 (_mm512_set1_epi16): Define.
1708 (_mm512_set4_epi32): Define.
1709 (_mm512_set4_epi64): Define.
1710 (_mm512_set4_pd): Define.
1711 (_mm512_set4_ps): Define.
1712 (_mm512_setr4_epi64): Define.
1713 (_mm512_setr4_epi32): Define.
1714 (_mm512_setr4_pd): Define.
1715 (_mm512_setr4_ps): Define.
1716 (_mm512_setzero_epi32): Define.
1718 2014-03-31 Martin Jambor <mjambor@suse.cz>
1721 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
1722 callsite_arguments_match_p. Updated all callers. Also check types of
1723 corresponding formal parameters and actual arguments.
1724 (not_all_callers_have_enough_arguments_p) Renamed to
1725 some_callers_have_mismatched_arguments_p.
1727 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
1729 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
1731 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
1734 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
1737 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
1739 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
1740 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
1742 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
1743 Use FMAMODE_NOVF512 mode iterator.
1744 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
1745 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
1746 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
1747 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
1749 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
1750 Use VF_128_256 mode iterator.
1751 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
1754 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1756 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
1757 static chain if needed.
1759 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1762 * lra-constraints.c (index_part_to_reg): New.
1763 (process_address): Use it.
1765 2014-03-27 Jeff Law <law@redhat.com>
1766 Jakub Jelinek <jakub@redhat.com>
1769 * expr.c (do_tablejump): Use simplify_gen_binary rather than
1770 gen_rtx_{PLUS,MULT} to build up the address expression.
1772 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
1773 creating non-canonical RTL.
1775 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1778 * ipa-inline.c (want_inline_small_function_p): Short circuit large
1779 functions; reorganize to make cheap checks first.
1780 (inline_small_functions): Do not estimate growth when dumping;
1782 * ipa-inline.h (inline_summary): Add min_size.
1783 (growth_likely_positive): New function.
1784 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
1785 (set_cond_stmt_execution_predicate): Cleanup.
1786 (estimate_edge_size_and_time): Compute min_size.
1787 (estimate_calls_size_and_time): Likewise.
1788 (estimate_node_size_and_time): Likewise.
1789 (inline_update_overall_summary): Update min_size.
1790 (do_estimate_edge_time): Likewise.
1791 (do_estimate_edge_size): Update.
1792 (do_estimate_edge_hints): Update.
1793 (growth_likely_positive): New function.
1795 2014-03-28 Jakub Jelinek <jakub@redhat.com>
1798 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
1799 also if addr has VOIDmode.
1801 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1803 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
1804 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
1806 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
1807 instructions as well as AdvancedSIMD loads.
1809 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1811 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
1812 Use crypto_aese type.
1813 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
1814 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
1815 crypto_aese, crypto_aesmc. Move to types.md.
1816 * config/arm/types.md (crypto_aes): Split into crypto_aese,
1818 * config/arm/iterators.md (crypto_type): Likewise.
1820 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1822 * cgraph.c: Include expr.h and tree-dfa.h.
1823 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
1826 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1829 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
1830 regs from checking multi-reg pseudos.
1832 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1834 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
1836 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1838 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
1839 if it would clobber the stack pointer, even temporarily.
1841 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
1843 * mode-switching.c: Make small adjustments to the top comment.
1845 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1847 * config/rs6000/constraints.md (wD constraint): New constraint to
1848 match the constant integer to get the top DImode/DFmode out of a
1849 vector in a VSX register.
1851 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
1852 match the constant integer to get the top DImode/DFmode out of a
1853 vector in a VSX register.
1855 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
1858 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1861 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
1862 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
1864 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
1865 Optimize vec_extract of 64-bit values, where the value being
1866 extracted is in the top word, where we can use scalar
1867 instructions. Add direct move and store support. Combine the big
1868 endian/little endian vector select load support into a single insn.
1869 (vsx_extract_<mode>_internal1): Likewise.
1870 (vsx_extract_<mode>_internal2): Likewise.
1871 (vsx_extract_<mode>_load): Likewise.
1872 (vsx_extract_<mode>_store): Likewise.
1873 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
1874 combined into vsx_extract_<mode>_load.
1875 (vsx_extract_<mode>_one_le): Likewise.
1877 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
1878 define the top 64-bit vector element.
1880 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
1883 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1884 Document vec_vbpermq builtin.
1887 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
1888 enable use of xxsldwi and xxpermdi builtin functions.
1889 (vec_xxpermdi): Likewise.
1891 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1892 Document use of vec_xxsldwi and vec_xxpermdi builtins.
1894 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
1896 PR rtl-optimization/60650
1897 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
1899 (find_spills_for): New.
1900 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
1901 Spill all pseudos on the second iteration.
1903 2014-03-27 Marek Polacek <polacek@redhat.com>
1906 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
1909 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1911 * config/s390/s390.c (s390_can_use_return_insn): Check for
1912 call-saved FPRs on 31 bit.
1914 2014-03-27 Jakub Jelinek <jakub@redhat.com>
1917 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
1918 if they need regimplification, just drop them instead of
1919 calling gimple_regimplify_operands on them.
1921 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
1924 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
1925 (aarch64_frame_pointer_required): Adjust logic.
1926 (aarch64_can_eliminate): Adjust logic.
1927 (aarch64_override_options_after_change): Adjust logic.
1929 2014-03-27 Dehao Chen <dehao@google.com>
1931 * ipa-inline.c (early_inliner): Update node's inline info.
1933 2014-03-26 Dehao Chen <dehao@google.com>
1935 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
1936 compiler inserted conditional jumps for NAN float check.
1938 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1940 * ubsan.h (ubsan_create_data): Change second argument's type
1941 to const location_t *.
1942 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
1944 (ubsan_create_data): Change second argument to const location_t *PLOC.
1945 Create Loc field whenever PLOC is non-NULL.
1946 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
1947 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
1951 * real.c (real_to_integer2): Change type of low to UHWI.
1953 2014-03-26 Tobias Burnus <burnus@net-b.de>
1955 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
1956 (CILK_SELF_SPECS): New define.
1957 (driver_self_specs): Use it.
1959 2014-03-26 Richard Biener <rguenther@suse.de>
1961 * tree-pretty-print.c (percent_K_format): Implement special
1962 case for LTO and its stripped down BLOCK tree.
1964 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1967 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
1969 * tree-vrp.c (simplify_internal_call_using_ranges): If only
1970 one range is range_int_cst_p, but not both, at least optimize
1971 addition/subtraction of 0 and multiplication by 0 or 1.
1972 * gimple-fold.c (gimple_fold_call): Fold
1973 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
1974 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
1975 INTEGER_CSTs, try to fold at least x * 0 and y - y.
1977 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
1979 PR rtl-optimization/60452
1980 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
1981 <case REG>: Return 1 for invalid offsets from the frame pointer.
1983 2014-03-26 Marek Polacek <polacek@redhat.com>
1986 * doc/extend.texi (C Extensions): Mention variable-length arrays in
1989 2014-03-26 Marek Polacek <polacek@redhat.com>
1992 * doc/extend.texi (Designated Inits): Describe what happens to omitted
1995 2014-03-26 Marek Polacek <polacek@redhat.com>
1998 * ira-color.c (update_conflict_hard_regno_costs): Perform the
1999 multiplication in unsigned type.
2001 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
2003 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
2005 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
2007 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
2009 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
2012 * cif-code.def (UNREACHABLE) New code.
2013 * ipa-inline.c (inline_small_functions): Skip edges to
2014 __builtlin_unreachable.
2015 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
2016 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
2017 predicate to __bulitin_unreachable.
2018 (set_cond_stmt_execution_predicate): Fix issue when
2019 invert_tree_comparison returns ERROR_MARK.
2020 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
2021 propagate to inline clones.
2022 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
2024 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
2025 * cgraphclones.c (cgraph_clone_node): If call destination is already
2026 ureachable, do not redirect it back.
2027 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
2030 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
2032 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
2033 Do not modify inline clones.
2035 2014-03-25 Jakub Jelinek <jakub@redhat.com>
2037 * config/i386/i386.md (general_sext_operand): New mode attr.
2038 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
2039 don't generate (sign_extend (const_int)).
2040 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
2041 operands[2]. Use We constraint instead of <i> and
2042 <general_sext_operand> predicate instead of <general_operand>.
2043 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
2044 * config/i386/constraints.md (We): New constraint.
2045 * config/i386/predicates.md (x86_64_sext_operand,
2046 sext_operand): New predicates.
2048 2014-03-25 Martin Jambor <mjambor@suse.cz>
2051 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
2052 inconsistent devirtualizations to __builtin_unreachable.
2054 2014-03-25 Marek Polacek <polacek@redhat.com>
2057 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
2059 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
2061 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
2062 order of elements for big-endian.
2064 2014-03-25 Richard Biener <rguenther@suse.de>
2067 * gimplify-me.c (gimple_regimplify_operands): Update the
2070 2014-03-25 Martin Jambor <mjambor@suse.cz>
2073 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
2074 (lto_output_varpool_node): Likewise.
2075 (input_overwrite_node): Likewise.
2076 (input_varpool_node): Likewise.
2078 2014-03-25 Richard Biener <rguenther@suse.de>
2080 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
2081 (run_gcc): Likewise.
2083 2014-03-25 Jakub Jelinek <jakub@redhat.com>
2085 * combine.c (simplify_compare_const): Add MODE argument.
2086 Handle mode_width 0 as very large mode_width.
2087 (try_combine, simplify_comparison): Adjust callers.
2089 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
2090 type to avoid signed integer overflow.
2091 * explow.c (plus_constant): Likewise.
2093 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
2095 * doc/generic.texi: Correct typos.
2097 2014-03-24 Tobias Burnus <burnus@net-b.de>
2099 * doc/invoke.texi (-flto): Expand section about
2100 using static libraries with LTO.
2102 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2104 PR rtl-optimization/60501
2105 * optabs.def (addptr3_optab): New optab.
2106 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
2107 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
2108 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
2110 * lra.c (emit_add3_insn): Use the addptr pattern if available.
2112 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
2114 2014-03-24 Ulrich Drepper <drepper@gmail.com>
2116 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
2119 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
2120 (_mm256_undefined_ps): Define.
2121 (_mm256_undefined_pd): Define.
2122 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
2123 (_mm_undefined_pd): Define.
2124 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
2125 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
2126 (_mm512_undefined_ps): Define.
2127 (_mm512_undefined_pd): Define.
2128 Use _mm*_undefined_*.
2129 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
2131 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
2133 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
2134 (lshr_simd): DI mode added.
2135 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
2136 (aarch64_ushr_simddi): Likewise.
2137 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
2138 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
2139 (vshrd_n_u64): Likewise.
2141 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2143 * Makefile.in (s-macro_list): Depend on cc1.
2145 2014-03-23 Teresa Johnson <tejohnson@google.com>
2147 * ipa-utils.c (ipa_print_order): Use specified dump file.
2149 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
2151 PR rtl-optimization/60601
2152 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
2154 * gcc.c (eval_spec_function): Initialize save_growing_value.
2156 2014-03-22 Jakub Jelinek <jakub@redhat.com>
2159 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
2160 code == MINUS_EXPR, never swap op0 with op1.
2162 * toplev.c (init_local_tick): Avoid signed integer multiplication
2164 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
2165 shift by first operand's bitsize.
2167 2014-03-21 Jakub Jelinek <jakub@redhat.com>
2170 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
2172 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
2173 TARGET_ISA_64BIT_P(x).
2175 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2177 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
2178 pattern for vector nor instead of subtract from splat(-1).
2179 (altivec_expand_vec_perm_const_le): Likewise.
2181 2014-03-21 Richard Henderson <rth@twiddle.net>
2184 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
2185 related insns after epilogue_completed.
2187 2014-03-21 Martin Jambor <mjambor@suse.cz>
2190 * cgraph.h (symtab_node): New flag body_removed.
2191 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
2192 when removing bodies.
2193 * symtab.c (dump_symtab_base): Dump body_removed flag.
2194 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
2195 had their bodies removed.
2197 2014-03-21 Martin Jambor <mjambor@suse.cz>
2200 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
2203 2014-03-21 Richard Biener <rguenther@suse.de>
2205 PR tree-optimization/60577
2206 * tree-core.h (struct tree_base): Document nothrow_flag use
2207 in VAR_DECL_NONALIASED.
2208 * tree.h (VAR_DECL_NONALIASED): New.
2209 (may_be_aliased): Adjust.
2210 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
2212 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2214 * expr.c (expand_expr_real_1): Remove outdated comment.
2216 2014-03-20 Jakub Jelinek <jakub@redhat.com>
2219 * ira.c (adjust_cleared_regs): Call copy_rtx on
2220 *reg_equiv[REGNO (loc)].src_p before passing it to
2221 simplify_replace_fn_rtx.
2224 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
2225 into CONST, put pic register as first operand of PLUS. Use
2226 gen_const_mem for both 32-bit and 64-bit PIC got loads.
2228 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2230 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
2232 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2234 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
2235 around for store forwarding issue in the FPU on the UT699.
2236 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
2237 loads and operations if -mfix-ut699 is specified.
2238 (divtf3_hq): Tweak attribute.
2239 (sqrttf2_hq): Likewise.
2241 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2243 * calls.c (store_one_arg): Remove incorrect const qualification on the
2244 type of the temporary.
2245 * cfgexpand.c (expand_return): Likewise.
2246 * expr.c (expand_constructor): Likewise.
2247 (expand_expr_real_1): Likewise.
2249 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
2251 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
2254 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
2257 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
2259 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
2261 * config/arm/aarch-common-protos.h
2262 (alu_cost_table): Fix spelling of "extend".
2263 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
2265 2014-03-19 Richard Biener <rguenther@suse.de>
2268 * tree-core.h (tree_type_common): Re-order pointer members
2269 to reduce recursion depth during GC walks.
2271 2014-03-19 Marek Polacek <polacek@redhat.com>
2274 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
2275 before accessing it.
2277 2014-03-19 Richard Biener <rguenther@suse.de>
2280 * lto-streamer-in.c (input_function): In WPA stage do not drop
2283 2014-03-19 Jakub Jelinek <jakub@redhat.com>
2285 PR tree-optimization/60559
2286 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
2287 with build_zero_cst assignment.
2289 2014-03-18 Kai Tietz <ktietz@redhat.com>
2291 PR rtl-optimization/56356
2292 * sdbout.c (sdbout_parms): Verify that parms'
2293 incoming argument is valid.
2294 (sdbout_reg_parms): Likewise.
2296 2014-03-18 Richard Henderson <rth@redhat.com>
2299 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
2300 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
2301 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
2303 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
2305 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
2306 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
2307 Italicize plugin event names in description. Explain that
2308 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
2309 Remind that no GCC functions should be called after PLUGIN_FINISH.
2310 Explain what pragmas with expansion are.
2312 2014-03-18 Martin Liska <mliska@suse.cz>
2314 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
2315 gimple call statement is update.
2316 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
2317 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
2319 2014-03-18 Jakub Jelinek <jakub@redhat.com>
2322 * ubsan.c (ubsan_instrument_unreachable): Call
2323 initialize_sanitizer_builtins.
2324 (ubsan_pass): Likewise.
2327 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
2328 varpool_finalize_decl instead of rest_of_decl_compilation.
2330 2014-03-18 Richard Biener <rguenther@suse.de>
2332 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
2333 by using bitmap_and_compl instead of bitmap_and_compl_into.
2334 (df_rd_transfer_function): Likewise.
2336 2014-03-18 Richard Biener <rguenther@suse.de>
2338 * doc/lto.texi (fresolution): Fix typo.
2340 2014-03-18 Richard Biener <rguenther@suse.de>
2342 * doc/invoke.texi (flto): Update for changes in 4.9.
2344 2014-03-18 Richard Biener <rguenther@suse.de>
2346 * doc/loop.texi: Remove section on the removed lambda framework.
2347 Update loop docs with recent changes in preserving loop structure.
2349 2014-03-18 Richard Biener <rguenther@suse.de>
2351 * doc/lto.texi (-fresolution): Document.
2353 2014-03-18 Richard Biener <rguenther@suse.de>
2355 * doc/contrib.texi: Adjust my name.
2357 2014-03-18 Jakub Jelinek <jakub@redhat.com>
2360 * internal-fn.c: Include diagnostic-core.h.
2361 (expand_BUILTIN_EXPECT): New function.
2362 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
2363 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
2364 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
2365 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
2367 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
2368 Revert 3 argument __builtin_expect code.
2369 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
2370 * gimple-fold.c (gimple_fold_call): Likewise.
2371 * tree.h (fold_builtin_expect): New prototype.
2372 * builtins.c (build_builtin_expect_predicate): Add predictor
2373 argument, if non-NULL, create 3 argument __builtin_expect.
2374 (fold_builtin_expect): No longer static. Add ARG2 argument,
2375 pass it through to build_builtin_expect_predicate.
2376 (fold_builtin_2): Adjust caller.
2377 (fold_builtin_3): Handle BUILT_IN_EXPECT.
2378 * internal-fn.def (BUILTIN_EXPECT): New.
2380 2014-03-18 Tobias Burnus <burnus@net-b.de>
2383 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
2384 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
2385 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
2387 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
2390 * predict.c (combine_predictions_for_bb): Fix up formatting.
2391 (expr_expected_value_1, expr_expected_value): Add predictor argument,
2392 fill what it points to if non-NULL.
2393 (tree_predict_by_opcode): Adjust caller, use the predictor.
2394 * predict.def (PRED_COMPARE_AND_SWAP): Add.
2396 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
2398 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
2399 proper constant for the store mode.
2401 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
2403 * symtab.c (change_decl_assembler_name): Fix transparent alias
2406 2014-03-16 Renlin Li <Renlin.Li@arm.com>
2408 * config/aarch64/aarch64.c: Correct the comments about the
2409 aarch64 stack layout.
2411 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
2413 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
2414 check for GF_OMP_FOR_KIND_FOR.
2416 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
2418 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
2419 ymm and zmm register names.
2421 2014-03-17 Jakub Jelinek <jakub@redhat.com>
2424 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
2425 note creation for the 2010-08-31 changes.
2427 2014-03-17 Marek Polacek <polacek@redhat.com>
2430 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
2431 as -fno-tree-loop-vectorize.
2432 (expand_omp_simd): Likewise.
2434 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
2436 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
2437 (eligible_for_call_delay): New prototype.
2438 * config/sparc/sparc.c (tls_call_delay): Rename into...
2439 (eligible_for_call_delay): ...this. Return false if the instruction
2440 cannot be put in the delay slot of a branch.
2441 (eligible_for_restore_insn): Simplify.
2442 (eligible_for_return_delay): Return false if the instruction cannot be
2443 put in the delay slot of a branch and simplify.
2444 (eligible_for_sibcall_delay): Return false if the instruction cannot be
2445 put in the delay slot of a branch.
2446 * config/sparc/sparc.md (fix_ut699): New attribute.
2447 (tls_call_delay): Delete.
2448 (in_call_delay): Reimplement.
2449 (eligible_for_sibcall_delay): Rename into...
2450 (in_sibcall_delay): ...this.
2451 (eligible_for_return_delay): Rename into...
2452 (in_return_delay): ...this.
2453 (in_branch_delay): Reimplement.
2454 (in_uncond_branch_delay): Delete.
2455 (in_annul_branch_delay): Delete.
2457 2014-03-14 Richard Henderson <rth@redhat.com>
2460 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
2461 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
2462 (*floathi<X87MODEF>2_i387_with_temp): Remove.
2463 (floathi splitters): Remove.
2464 (float<SWI48x>xf2): New pattern.
2465 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
2466 code that tried to handle DImode for 32-bit, but which was excluded
2467 by the pattern's condition. Drop allocation of stack temporary.
2468 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
2469 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
2470 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
2471 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
2472 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
2473 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
2474 (*float<SWI48><MODEF>2_sse_interunit): Remove.
2475 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
2476 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
2477 (*float<SWI48x><X87MODEF>2_i387): Remove.
2478 (all float _with_temp splitters): Remove.
2479 (*float<SWI48x><MODEF>2_i387): New pattern.
2480 (*float<SWI48><MODEF>2_sse): New pattern.
2481 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
2482 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
2484 2014-03-14 Jakub Jelinek <jakub@redhat.com>
2485 Marek Polacek <polacek@redhat.com>
2488 * common.opt (dump_base_name_prefixed): New Variable.
2489 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
2490 if x_dump_base_name_prefixed is already set, set it at the end.
2492 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
2494 PR rtl-optimization/60508
2495 * lra-constraints.c (get_reload_reg): Add new parameter
2497 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
2498 Pass the new parameter values.
2500 2014-03-14 Richard Biener <rguenther@suse.de>
2502 * common.opt: Revert unintented changes from r205065.
2505 2014-03-14 Richard Biener <rguenther@suse.de>
2508 * cfghooks.c (split_block): Properly adjust all loops the
2509 block was a latch of.
2511 2014-03-14 Martin Jambor <mjambor@suse.cz>
2514 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
2517 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
2520 * config/avr/avr.c (avr_set_current_function): Pass function name
2521 through default_strip_name_encoding before sanity checking instead
2522 of skipping the first char of the assembler name.
2524 2014-03-13 Richard Henderson <rth@redhat.com>
2527 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
2528 (ix86_force_to_memory, ix86_free_from_memory): Remove.
2529 * config/i386/i386-protos.h: Likewise.
2530 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
2531 in the expander instead of a splitter.
2532 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
2533 any possibility of requiring a memory.
2534 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
2535 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
2536 (fp branch splitters): Update for ix86_split_fp_branch.
2537 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
2538 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
2539 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
2540 (*fop_<MODEF>_2_i387): Remove f/r alternative.
2541 (*fop_<MODEF>_3_i387): Likewise.
2542 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
2543 (splitters for the fop_* register patterns): Remove.
2544 (fscalexf4_i387): Rename from *fscalexf4_i387.
2545 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
2547 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2549 PR tree-optimization/59779
2550 * tree-dfa.c (get_ref_base_and_extent): Use double_int
2551 type for bitsize and maxsize instead of HOST_WIDE_INT.
2553 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
2555 PR rtl-optimization/57320
2556 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
2557 the CFG after thread_prologue_and_epilogue_insns.
2559 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
2561 PR rtl-optimization/57189
2562 * lra-constraints.c (process_alt_operands): Disfavor spilling
2565 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
2567 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
2569 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2571 PR tree-optimization/59025
2573 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
2574 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
2576 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
2579 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
2580 calls of avr_out_plus_1.
2582 2014-03-13 Bin Cheng <bin.cheng@arm.com>
2584 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
2585 BB's single pred and update the father loop's latch info later.
2587 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2589 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
2593 (VEC_base): Likewise.
2594 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
2595 registers, we need to swap double words in little endian mode.
2597 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
2598 to be a container mode for 128-bit integer operations added in ISA
2599 2.07. Unlike TImode and PTImode, the preferred register set is
2600 the Altivec/VMX registers for the 128-bit operations.
2602 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
2604 (rs6000_split_128bit_ok_p): Likewise.
2606 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
2607 macros for creating ISA 2.07 normal and overloaded builtin
2608 functions with 3 arguments.
2609 (BU_P8V_OVERLOAD_3): Likewise.
2610 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
2611 for use as overloaded functions.
2612 (VPERM_1TI_UNS): Likewise.
2613 (VSEL_1TI): Likewise.
2614 (VSEL_1TI_UNS): Likewise.
2615 (ST_INTERNAL_1ti): Likewise.
2616 (LD_INTERNAL_1ti): Likewise.
2617 (XXSEL_1TI): Likewise.
2618 (XXSEL_1TI_UNS): Likewise.
2619 (VPERM_1TI): Likewise.
2620 (VPERM_1TI_UNS): Likewise.
2621 (XXPERMDI_1TI): Likewise.
2622 (SET_1TI): Likewise.
2623 (LXVD2X_V1TI): Likewise.
2624 (STXVD2X_V1TI): Likewise.
2625 (VEC_INIT_V1TI): Likewise.
2626 (VEC_SET_V1TI): Likewise.
2627 (VEC_EXT_V1TI): Likewise.
2628 (EQV_V1TI): Likewise.
2629 (NAND_V1TI): Likewise.
2630 (ORC_V1TI): Likewise.
2631 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
2632 added in ISA 2.07. Add both normal 'altivec' builtins, and the
2634 (VADDUQM): Likewise.
2635 (VSUBCUQ): Likewise.
2636 (VADDEUQM): Likewise.
2637 (VADDECUQ): Likewise.
2638 (VSUBEUQM): Likewise.
2639 (VSUBECUQ): Likewise.
2641 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
2642 __int128_t and __uint128_t types.
2643 (__uint128_type): Likewise.
2644 (altivec_categorize_keyword): Add support for vector __int128_t,
2645 vector __uint128_t, vector __int128, and vector unsigned __int128
2646 as a container type for TImode operations that need to be done in
2647 VSX/Altivec registers.
2648 (rs6000_macro_to_expand): Likewise.
2649 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
2650 to support 128-bit integer instructions vaddcuq, vadduqm,
2651 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
2652 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
2654 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
2655 for V1TImode, and set up preferences to use VSX/Altivec registers.
2656 Setup VSX reload handlers.
2657 (rs6000_debug_reg_global): Likewise.
2658 (rs6000_init_hard_regno_mode_ok): Likewise.
2659 (rs6000_preferred_simd_mode): Likewise.
2660 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
2661 (easy_altivec_constant): Likewise.
2662 (output_vec_const_move): Likewise.
2663 (rs6000_expand_vector_set): Convert V1TImode set and extract to
2665 (rs6000_expand_vector_extract): Likewise.
2666 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
2668 (rs6000_const_vec): Add support for V1TImode.
2669 (rs6000_emit_le_vsx_load): Swap double words when loading or
2670 storing TImode/V1TImode.
2671 (rs6000_emit_le_vsx_store): Likewise.
2672 (rs6000_emit_le_vsx_move): Likewise.
2673 (rs6000_emit_move): Add support for V1TImode.
2674 (altivec_expand_ld_builtin): Likewise.
2675 (altivec_expand_st_builtin): Likewise.
2676 (altivec_expand_vec_init_builtin): Likewise.
2677 (altivec_expand_builtin): Likewise.
2678 (rs6000_init_builtins): Add support for V1TImode type. Add
2679 support for ISA 2.07 128-bit integer builtins. Define type names
2680 for the VSX/Altivec vector types.
2681 (altivec_init_builtins): Add support for overloaded vector
2682 functions with V1TImode type.
2683 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
2684 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
2686 (rs6000_split_128bit_ok_p): Likewise.
2687 (rs6000_handle_altivec_attribute): Create V1TImode from vector
2688 __int128_t and vector __uint128_t.
2690 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
2691 and mode attributes.
2698 (VS_scalar): Likewise.
2699 (VS_double): Likewise.
2700 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
2702 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
2703 we support the ISA 2.07 128-bit integer arithmetic instructions.
2704 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
2705 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
2706 and TImode types for use with the builtin functions.
2707 (V1TI_type_node): Likewise.
2708 (unsigned_V1TI_type_node): Likewise.
2709 (intTI_type_internal_node): Likewise.
2710 (uintTI_type_internal_node): Likewise.
2712 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
2713 128-bit builtin functions.
2714 (UNSPEC_VADDEUQM): Likewise.
2715 (UNSPEC_VADDECUQ): Likewise.
2716 (UNSPEC_VSUBCUQ): Likewise.
2717 (UNSPEC_VSUBEUQM): Likewise.
2718 (UNSPEC_VSUBECUQ): Likewise.
2719 (VM): Add V1TImode to vector mode iterators.
2721 (VI_unit): Likewise.
2722 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
2723 (altivec_vaddcuq): Likewise.
2724 (altivec_vsubuqm): Likewise.
2725 (altivec_vsubcuq): Likewise.
2726 (altivec_vaddeuqm): Likewise.
2727 (altivec_vaddecuq): Likewise.
2728 (altivec_vsubeuqm): Likewise.
2729 (altivec_vsubecuq): Likewise.
2731 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
2733 (BOOL_128): Likewise.
2734 (BOOL_REGS_OUTPUT): Likewise.
2735 (BOOL_REGS_OP1): Likewise.
2736 (BOOL_REGS_OP2): Likewise.
2737 (BOOL_REGS_UNARY): Likewise.
2738 (BOOL_REGS_AND_CR0): Likewise.
2740 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
2741 128-bit integer builtin support.
2742 (vec_vadduqm): Likewise.
2743 (vec_vaddecuq): Likewise.
2744 (vec_vaddeuqm): Likewise.
2745 (vec_vsubecuq): Likewise.
2746 (vec_vsubeuqm): Likewise.
2747 (vec_vsubcuq): Likewise.
2748 (vec_vsubuqm): Likewise.
2750 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2751 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
2752 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
2753 128-bit integer add/subtract to ISA 2.07.
2755 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
2757 * config/arc/arc.c (arc_predicate_delay_insns):
2758 Fix third argument passed to conditionalize_nonjump.
2760 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
2762 * config/aarch64/aarch64-builtins.c
2763 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
2764 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
2765 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
2766 instead of __builtin_lfloor.
2767 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
2769 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2771 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
2772 (tree_ssa_ifcombine_bb_1): New function.
2773 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
2774 is an empty forwarder block to then_bb or vice versa and then_bb
2775 and else_bb are effectively swapped.
2777 2014-03-12 Christian Bruel <christian.bruel@st.com>
2780 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
2781 REG_CFA_DEF_CFA note.
2782 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
2783 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
2785 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
2787 PR tree-optimization/60454
2788 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
2790 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2792 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
2793 Do not define target_cpu_default2 to generic.
2794 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
2795 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
2796 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
2798 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2799 Marc Glisse <marc.glisse@inria.fr>
2801 PR tree-optimization/60502
2802 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
2803 instead of build_low_bits_mask.
2805 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2808 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
2809 if there are multiple uses, but op doesn't live on E edge.
2810 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
2811 clobber stmts before __builtin_unreachable.
2813 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
2815 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
2816 hard_frame_pointer_rtx.
2817 * cse.c (cse_insn): Remove volatile check.
2818 * cselib.c (cselib_process_insn): Likewise.
2819 * dse.c (scan_insn): Likewise.
2821 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2823 * config/arc/arc.c (conditionalize_nonjump): New function,
2825 (arc_ifcvt): ... this.
2826 (arc_predicate_delay_insns): Use it.
2828 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2830 * config/arc/predicates.md (extend_operand): During/after reload,
2831 allow const_int_operand.
2832 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
2833 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
2834 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
2836 (umulsi3_highpart_i): Likewise.
2838 2014-03-11 Richard Biener <rguenther@suse.de>
2840 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
2841 Add asserts to guard possible wrong-code bugs.
2843 2014-03-11 Richard Biener <rguenther@suse.de>
2845 PR tree-optimization/60429
2846 PR tree-optimization/60485
2847 * tree-ssa-structalias.c (set_union_with_increment): Properly
2848 take into account all fields that overlap the shifted vars.
2849 (do_sd_constraint): Likewise.
2850 (do_ds_constraint): Likewise.
2851 (get_constraint_for_ptr_offset): Likewise.
2853 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
2855 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
2856 (nios2_compute_frame_layout):
2857 Add calculation of cfun->machine->fp_save_offset.
2858 (nios2_expand_prologue): Correct setting of frame pointer register
2860 (nios2_expand_epilogue): Update recovery of stack pointer from
2861 frame pointer accordingly.
2862 (nios2_initial_elimination_offset): Update calculation of offset
2863 for eliminating to HARD_FRAME_POINTER_REGNUM.
2865 2014-03-10 Jakub Jelinek <jakub@redhat.com>
2868 * ipa.c (symtab_remove_unreachable_nodes): Don't call
2869 cgraph_get_create_node on VAR_DECLs.
2871 2014-03-10 Richard Biener <rguenther@suse.de>
2874 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
2876 2014-03-08 Douglas B Rupp <rupp@gnat.com>
2878 * config/vms/vms.opt (vms_float_format): New variable.
2880 2014-03-08 Tobias Burnus <burnus@net-b.de>
2882 * doc/invoke.texi (-fcilkplus): Update implementation status.
2884 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
2885 Richard Biener <rguenther@suse.de>
2887 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
2888 consistently accross all TUs.
2889 (run_gcc): Enable -fshort-double automatically at link at link-time
2890 and disallow override.
2892 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
2895 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
2896 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
2897 if they can't be used.
2899 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2901 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
2902 for Solaris 11/x86 ld.
2903 * configure: Regenerate.
2905 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2907 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
2908 (LIB_TLS_SPEC): Save as ld_tls_libs.
2909 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
2910 (HAVE_AS_IX86_TLSLDM): New test.
2911 * configure, config.in: Regenerate.
2912 * config/i386/i386.c (legitimize_tls_address): Fall back to
2913 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
2914 cannot support TLS_MODEL_LOCAL_DYNAMIC.
2915 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
2916 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
2918 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
2920 * common.opt (fira-loop-pressure): Mark as optimization.
2922 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
2924 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
2925 an OpenMP mappable type.
2927 2014-03-06 Matthias Klose <doko@ubuntu.com>
2929 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
2930 MULTILIB_OSDIRNAMES is not defined.
2932 2014-03-06 Jakub Jelinek <jakub@redhat.com>
2933 Meador Inge <meadori@codesourcery.com>
2936 * config/arm/arm.c (arm_tls_symbol_p): Remove.
2937 (arm_legitimize_address): Call legitimize_tls_address for any
2938 arm_tls_referenced_p expression, handle constant addend. Call it
2939 before testing for !TARGET_ARM.
2940 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
2942 2014-03-06 Richard Biener <rguenther@suse.de>
2948 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2950 * tree-streamer.c (record_common_node): Assert we don't record
2951 nodes with type double.
2952 (preload_common_node): Skip type double, complex double and double
2953 pointer since it is now frontend dependent due to fshort-double option.
2955 2014-03-06 Richard Biener <rguenther@suse.de>
2957 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
2958 or -fno-lto is specified and the linker has full plugin support.
2959 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
2960 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
2961 * lto-wrapper.c (merge_and_complain): Merge compile-time
2962 optimization levels.
2963 (run_gcc): And pass it through to the link options.
2965 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
2969 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2971 * cselib.c (remove_useless_values): Skip to avoid quadratic
2972 behavior if the condition moved from...
2973 (cselib_process_insn): ... here holds.
2975 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2978 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
2979 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
2982 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
2983 (TM_H): Add x86-tune.def.
2985 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2987 * config/aarch64/aarch64.c (generic_tunings):
2988 Use cortexa57_extra_costs.
2990 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2993 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
2994 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
2995 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
2998 2014-03-04 Heiher <r@hev.cc>
3000 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
3001 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
3003 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
3005 * config/i386/predicates.md (const2356_operand): Change to ...
3006 (const2367_operand): ... this.
3007 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
3009 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3010 (*avx512pf_scatterpf<mode>sf): Ditto.
3011 (avx512pf_scatterpf<mode>df): Ditto.
3012 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3013 (*avx512pf_scatterpf<mode>df): Ditto.
3014 * config/i386/i386.c (ix86_expand_builtin): Update
3015 incorrect hint operand error message.
3017 2014-03-04 Richard Biener <rguenther@suse.de>
3019 * lto-section-in.c (lto_get_section_data): Fix const cast.
3021 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
3023 * tree-streamer.c (record_common_node): Assert we don't record
3024 nodes with type double.
3025 (preload_common_node): Skip type double, complex double and double
3026 pointer since it is now frontend dependent due to fshort-double option.
3028 2014-03-04 Richard Biener <rguenther@suse.de>
3031 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
3032 (lto_input_toplevel_asms): Likewise.
3033 * lto-section-in.c (lto_get_section_data): Instead do it here
3036 2014-03-04 Richard Biener <rguenther@suse.de>
3038 PR tree-optimization/60382
3039 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
3040 dead PHIs a reduction.
3042 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
3044 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
3046 (_mm_prefetch): Move out of GCC target("sse") pragma.
3047 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
3048 GCC target("prfchw") pragma.
3049 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
3051 * config/i386/i386.c (ix86_option_override_internal): Enable
3052 -mprfchw with -mprefetchwt1.
3054 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
3056 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
3059 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
3061 * opts.h (CL_PCH_IGNORE): Define.
3062 * targhooks.c (option_affects_pch_p):
3063 Return false for options that have CL_PCH_IGNORE set.
3064 * opt-functions.awk: Process PchIgnore.
3065 * doc/options.texi: Document PchIgnore.
3067 * config/arc/arc.opt (misize): Add PchIgnore property.
3069 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3071 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
3072 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
3073 constraint on constants to permit them being loaded into
3074 GENERAL_REGS or BASE_REGS.
3076 2014-03-03 Nick Clifton <nickc@redhat.com>
3078 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
3079 anti-cacnonical alternatives.
3080 (negandhi3_real): New pattern.
3081 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
3083 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3085 * config/avr/avr-mcus.def: Remove atxmega16x1.
3086 * config/avr/avr-tables.opt: Regenerate.
3087 * config/avr/t-multilib: Regenerate.
3088 * doc/avr-mmcu.texi: Regenerate.
3090 2014-03-03 Tobias Grosser <tobias@grosser.es>
3091 Mircea Namolaru <mircea.namolaru@inria.fr>
3093 PR tree-optimization/58028
3094 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
3097 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3099 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
3100 not handled by recognizers.
3102 2014-03-03 Jakub Jelinek <jakub@redhat.com>
3105 * function.c (expand_function_end): Don't emit
3106 clobber_return_register sequence if clobber_after is a BARRIER.
3107 * cfgexpand.c (construct_exit_block): Append instructions before
3108 return_label to prev_bb.
3110 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3112 * config/rs6000/constraints.md: Document reserved use of "wc".
3114 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
3117 * ipa.c (function_and_variable_visibility): When dissolving comdat
3118 group, also set all symbols to local.
3120 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
3125 2013-12-14 Jan Hubicka <jh@suse.cz>
3127 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
3129 2014-03-02 Jon Beniston <jon@beniston.com>
3135 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
3136 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
3137 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
3138 (simple_return, *simple_return): New patterns
3139 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
3140 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
3142 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
3144 * dwarf2out.c (gen_subprogram_die): Tidy.
3146 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
3149 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
3150 (*mov_t_msb_neg_negc): ... this new insn.
3152 2014-02-28 Jason Merrill <jason@redhat.com>
3155 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
3158 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
3161 * dwarf2out.c (decltype_auto_die): New static.
3162 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
3163 (gen_type_die_with_usage): Handle 'decltype(auto)'.
3164 (is_cxx_auto): Likewise.
3166 2014-02-28 Ian Bolton <ian.bolton@arm.com>
3168 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
3169 we are not using general regs only.
3171 2014-02-28 Richard Biener <rguenther@suse.de>
3174 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
3175 previous fix and only allow to remove trivial pre-headers
3176 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
3177 (remove_forwarder_block): Properly update the latch of a loop.
3179 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3182 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
3183 (cselib_preserved_hash_table): New.
3184 (preserve_constants_and_equivs): Move preserved vals to it.
3185 (cselib_find_slot): Look it up first.
3186 (cselib_init): Initialize it.
3187 (cselib_finish): Release it.
3188 (dump_cselib_table): Dump it.
3190 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3193 * cselib.c (remove_useless_values): Skip to avoid quadratic
3194 behavior if the condition moved from...
3195 (cselib_process_insn): ... here holds.
3197 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3200 * var-tracking.c (vt_initialize): Apply the same condition to
3201 preserve the CFA base value.
3203 2014-02-28 Joey Ye <joey.ye@arm.com>
3206 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
3207 if reload in progress or completed.
3209 2014-02-28 Tobias Burnus <burnus@net-b.de>
3212 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
3215 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
3217 * doc/tm.texi.in (Condition Code Status): Update documention for
3218 relative locations of cc0-setter and cc0-user.
3220 2014-02-27 Jeff Law <law@redhat.com>
3222 PR rtl-optimization/52714
3223 * combine.c (try_combine): When splitting an unrecognized PARALLEL
3224 into two independent simple sets, if I3 is a jump, ensure the
3225 pattern we place into I3 is a (set (pc) ...).
3227 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
3228 Jeff Law <law@redhat.com>
3230 PR rtl-optimization/49847
3231 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
3232 are in different blocks.
3233 * doc/tm.texi (Condition Code Status): Update documention for
3234 relative locations of cc0-setter and cc0-user.
3236 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
3239 * lra.c (lra_emit_add): Check SUBREG too.
3241 2014-02-27 Andreas Schwab <schwab@suse.de>
3243 * config/m68k/m68k.c (m68k_option_override): Disable
3244 -flive-range-shrinkage for classic m68k.
3245 (m68k_override_options_after_change): Likewise.
3247 2014-02-27 Marek Polacek <polacek@redhat.com>
3250 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
3251 -Wmaybe-uninitialized.
3253 2014-02-27 Alan Modra <amodra@gmail.com>
3256 * reload1.c (emit_input_reload_insns): When reload_override_in,
3257 set old to rl->in_reg when rl->in_reg is a subreg.
3259 2014-02-26 Richard Biener <rguenther@suse.de>
3262 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
3264 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
3266 * common/config/i386/predicates.md (const1256_operand): Remove.
3267 (const2356_operand): New.
3268 (const_1_to_2_operand): Remove.
3269 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3270 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3271 (*avx512pf_gatherpf<mode>sf): Ditto.
3272 (avx512pf_gatherpf<mode>df): Ditto.
3273 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3274 (*avx512pf_gatherpf<mode>df): Ditto.
3275 (avx512pf_scatterpf<mode>sf): Ditto.
3276 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3277 (*avx512pf_scatterpf<mode>sf): Ditto.
3278 (avx512pf_scatterpf<mode>df): Ditto.
3279 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3280 (*avx512pf_scatterpf<mode>df): Ditto.
3281 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
3283 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
3285 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
3286 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
3287 (_mm512_mask_testn_epi64_mask): Move to ...
3288 * config/i386/avx512cdintrin.h: Here.
3289 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
3290 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
3291 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
3292 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
3293 TARGET_AVX512F from TARGET_AVX512CD.
3295 2014-02-26 Richard Biener <rguenther@suse.de>
3298 * ipa.c (walk_polymorphic_call_targets): Properly guard
3299 call to inline_update_overall_summary.
3301 2014-02-26 Bin Cheng <bin.cheng@arm.com>
3304 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
3305 and latches only if requested. Fix latch if it is removed.
3306 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
3307 LOOPS_HAVE_PREHEADERS.
3309 2014-02-25 Andrew Pinski <apinski@cavium.com>
3311 * builtins.c (expand_builtin_thread_pointer): Create a new target
3312 when the target is NULL.
3314 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
3316 PR rtl-optimization/60317
3317 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3318 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3319 * lra-assigns.c: Include params.h.
3320 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
3321 other reload pseudos considerations.
3323 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3325 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
3326 to use canonical form for nor<mode>3.
3328 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3331 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
3334 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
3336 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
3337 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
3338 (ix86_handle_option): Handle OPT_mprefetchwt1.
3339 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
3340 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3342 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3343 OPTION_MASK_ISA_PREFETCHWT1.
3344 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
3345 (PTA_PREFETCHWT1): New.
3346 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
3347 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
3348 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
3349 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
3350 (*prefetch_avx512pf_<mode>_: Change into ...
3351 (*prefetch_prefetchwt1_<mode>: This.
3352 * config/i386/i386.opt (mprefetchwt1): New.
3353 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
3354 (_mm_prefetch): Handle intent to write.
3355 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
3357 2014-02-25 Richard Biener <rguenther@suse.de>
3360 * emit-rtl.c (mem_attrs_htab): Remove.
3361 (mem_attrs_htab_hash): Likewise.
3362 (mem_attrs_htab_eq): Likewise.
3363 (set_mem_attrs): Always allocate new mem-attrs when something changed.
3364 (init_emit_once): Do not allocate mem_attrs_htab.
3366 2014-02-25 Richard Biener <rguenther@suse.de>
3369 * lto-opts.c (lto_write_options): Output non-explicit conservative
3370 -fwrapv, -fno-trapv and -fno-strict-overflow.
3371 * lto-wrapper.c (merge_and_complain): Handle merging those options.
3372 (run_gcc): And pass them through.
3374 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3376 * sel-sched.c (calculate_new_fences): New parameter ptime.
3377 Calculate it as a maximum over all fence cycles.
3378 (sel_sched_region_2): Adjust the call to calculate_new_fences.
3379 Print the final schedule timing when sched_verbose.
3381 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3383 PR rtl-optimization/60292
3384 * sel-sched.c (fill_vec_av_set): Do not reset target availability
3385 bit fot the fence instruction.
3387 2014-02-24 Alangi Derick <alangiderick@gmail.com>
3389 * calls.h: Fix typo in comment.
3391 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
3393 * config/pa/pa.c (pa_output_move_double): Don't valididate when
3394 adjusting offsetable addresses.
3396 2014-02-24 Guozhi Wei <carrot@google.com>
3398 * sparseset.h (sparseset_pop): Fix the wrong index.
3400 2014-02-24 Walter Lee <walt@tilera.com>
3402 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
3403 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
3405 * common/config/tilegx/tilegx-common.c
3406 (TARGET_DEFAULT_TARGET_FLAGS): Define.
3407 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
3409 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
3410 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
3411 (tilegx_gimplify_va_arg_expr): Handle big endian.
3412 (tilegx_expand_unaligned_load): Ditto.
3413 (tilegx_expand_unaligned_store): Ditto.
3414 (TARGET_RETURN_IN_MSB): New.
3415 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
3416 (TARGET_ENDIAN_DEFAULT): New.
3417 (TARGET_BIG_ENDIAN): Handle big endian.
3418 (BYTES_BIG_ENDIAN): Ditto.
3419 (WORDS_BIG_ENDIAN): Ditto.
3420 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
3423 * config/tilegx/tilegx.md (extv): Handle big endian.
3425 (insn_st<n>): Ditto.
3426 (insn_st<n>_add<bitsuffix>): Ditto.
3427 (insn_stnt<n>): Ditto.
3428 (insn_stnt<n>_add<bitsuffix>):Ditto.
3429 (vec_interleave_highv8qi): Handle big endian.
3430 (vec_interleave_highv8qi_be): New.
3431 (vec_interleave_highv8qi_le): New.
3432 (insn_v1int_h): Handle big endian.
3433 (vec_interleave_lowv8qi): Handle big endian.
3434 (vec_interleave_lowv8qi_be): New.
3435 (vec_interleave_lowv8qi_le): New.
3436 (insn_v1int_l): Handle big endian.
3437 (vec_interleave_highv4hi): Handle big endian.
3438 (vec_interleave_highv4hi_be): New.
3439 (vec_interleave_highv4hi_le): New.
3440 (insn_v2int_h): Handle big endian.
3441 (vec_interleave_lowv4hi): Handle big endian.
3442 (vec_interleave_lowv4hi_be): New.
3443 (vec_interleave_lowv4hi_le): New.
3444 (insn_v2int_l): Handle big endian.
3445 (vec_interleave_highv2si): Handle big endian.
3446 (vec_interleave_highv2si_be): New.
3447 (vec_interleave_highv2si_le): New.
3448 (insn_v4int_h): Handle big endian.
3449 (vec_interleave_lowv2si): Handle big endian.
3450 (vec_interleave_lowv2si_be): New.
3451 (vec_interleave_lowv2si_le): New.
3452 (insn_v4int_l): Handle big endian.
3453 * config/tilegx/tilegx.opt (mbig-endian): New option.
3454 (mlittle-endian): New option.
3455 * doc/install.texi: Document tilegxbe-linux.
3456 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
3458 2014-02-24 Martin Jambor <mjambor@suse.cz>
3461 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
3462 there are no parameter descriptors.
3464 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
3466 PR rtl-optimization/60268
3467 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
3468 initialization to ...
3469 (sched_rgn_init): ... here.
3470 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
3472 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3474 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
3477 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3479 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
3482 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3484 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
3485 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
3487 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3489 * config/microblaze/predicates.md: Add cmp_op predicate.
3490 * config/microblaze/microblaze.md: Add branch_compare instruction
3491 which uses cmp_op predicate and emits cmp insn before branch.
3492 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
3493 to microblaze_expand_conditional_branch and consolidate logic.
3494 (microblaze_expand_conditional_branch): emit branch_compare
3495 insn instead of handling cmp op separate from branch insn.
3497 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3499 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
3502 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3504 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
3505 define_insn with define_expand and new define_insn
3506 *altivec_lve<VI_char>x_internal.
3507 (altivec_stve<VI_char>x): Replace define_insn with define_expand
3508 and new define_insn *altivec_stve<VI_char>x_internal.
3509 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
3511 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
3513 (altivec_expand_stvex_be): New function.
3515 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
3517 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
3518 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
3519 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
3520 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
3522 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
3525 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
3526 instead of emit_move_insn.
3528 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3530 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
3532 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
3533 gen_altivec_vsumsws.
3535 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3537 * config/rs6000/altivec.md (altivec_lvxl): Rename as
3538 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
3539 (altivec_lvxl_<mode>): New define_expand incorporating
3540 -maltivec=be semantics where needed.
3541 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
3542 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
3543 semantics where needed.
3544 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
3545 (altivec_stvx_<mode>): New define_expand incorporating
3546 -maltivec=be semantics where needed.
3547 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
3548 VM2 iterator instead of V4SI.
3549 (altivec_stvxl_<mode>): New define_expand incorporating
3550 -maltivec=be semantics where needed.
3551 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
3552 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
3553 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
3554 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
3555 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
3556 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
3557 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
3558 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
3559 ALTIVEC_BUILTIN_STVXL.
3560 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
3561 (altivec_expand_stvx_be): Likewise.
3562 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
3563 (altivec_expand_lvx_be): Likewise.
3564 (altivec_expand_stvx_be): Likewise.
3565 (altivec_expand_builtin): Add cases for
3566 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
3567 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
3568 (altivec_init_builtins): Add definitions for
3569 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
3570 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
3572 2014-02-21 Catherine Moore <clm@codesourcery.com>
3574 * doc/invoke.texi (mvirt, mno-virt): Document.
3575 * config/mips/mips.opt (mvirt): New option.
3576 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
3578 2014-02-21 Richard Biener <rguenther@suse.de>
3580 PR tree-optimization/60276
3581 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
3582 (STMT_VINFO_MIN_NEG_DIST): New macro.
3583 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
3584 STMT_VINFO_MIN_NEG_DIST.
3585 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
3586 made for negative dependence distances still hold.
3588 2014-02-21 Richard Biener <rguenther@suse.de>
3591 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
3592 DECL_INITIAL for globals not in the current function context.
3594 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3596 PR tree-optimization/56490
3597 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
3598 * tree-ssa-uninit.c: Include params.h.
3599 (compute_control_dep_chain): Add num_calls argument, return false
3600 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
3601 num_calls to recursive call.
3602 (find_predicates): Change dep_chain into normal array,
3603 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
3604 variable and adjust compute_control_dep_chain caller.
3605 (find_def_preds): Likewise.
3607 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
3609 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
3610 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
3612 2014-02-21 Nick Clifton <nickc@redhat.com>
3614 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
3615 (pushhi1): Likewise.
3616 (popqi1): Add mode to pre_dec.
3619 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3621 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
3622 mode for mask of V8SFmode permutation.
3624 2014-02-20 Richard Henderson <rth@redhat.com>
3627 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
3628 a new pseudo for OLDVAL.
3630 2014-02-20 Jakub Jelinek <jakub@redhat.com>
3633 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
3634 gen_reg_rtx if d->testing_p.
3635 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
3636 if d->testing_p and we will certainly return true.
3637 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
3640 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
3642 * emit-rtl.c (gen_reg_rtx): Assert that
3643 crtl->emit.regno_pointer_align_length is non-zero.
3645 2014-02-20 Richard Henderson <rth@redhat.com>
3648 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
3649 on failure the store back into EXPECT.
3651 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
3652 Sandra Loosemore <sandra@codesourcery.com>
3654 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
3655 * config/nios2/nios2.c (nios2_function_profiler): Add
3656 -fPIC (flag_pic == 2) support.
3657 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
3658 (nios2_large_offset_p): New function.
3659 (nios2_unspec_reloc_p): Move up position, update to use
3660 nios2_large_offset_p.
3661 (nios2_unspec_address): Remove function.
3662 (nios2_unspec_offset): New function.
3663 (nios2_large_got_address): New function.
3664 (nios2_got_address): Add large offset support.
3665 (nios2_legitimize_tls_address): Update usage of removed and new
3667 (nios2_symbol_binds_local_p): New function.
3668 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
3669 (nios2_legitimize_address): Update to use nios2_large_offset_p.
3670 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
3671 (nios2_print_operand): Merge H/L processing, add hiadj/lo
3672 processing for (const (unspec ...)).
3673 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
3675 2014-02-20 Richard Biener <rguenther@suse.de>
3677 * tree-cfg.c (replace_uses_by): Mark altered BBs before
3678 doing the substitution.
3679 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
3681 2014-02-20 Martin Jambor <mjambor@suse.cz>
3684 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
3685 info when checking whether lattices are bottom.
3687 2014-02-20 Richard Biener <rguenther@suse.de>
3690 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
3693 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
3696 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
3697 parameter specifying the scaling.
3698 (inline_call): Update.
3699 (want_inline_recursively): Guard division by zero.
3700 (recursive_inlining): Update.
3701 * ipa-inline.h (clone_inlined_nodes): Update.
3703 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3706 * config/i386/i386.c (classify_argument): Pass structures of size
3707 64 bytes or less in register.
3709 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3710 Kirill Yukhin <kirill.yukhin@intel.com>
3712 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
3713 (_mm_rcp28_round_ss): Ditto.
3714 (_mm_rsqrt28_round_sd): Ditto.
3715 (_mm_rsqrt28_round_ss): Ditto.
3716 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
3717 (_mm_rcp14_round_ss): Ditto.
3718 (_mm_rsqrt14_round_sd): Ditto.
3719 (_mm_rsqrt14_round_ss): Ditto.
3720 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
3721 the first input operand, get rid of match_dup.
3722 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
3724 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3726 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
3727 operand as the first input operand, set type attribute.
3728 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3730 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
3731 operand as the first input operand, set type attribute.
3733 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3735 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
3738 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
3741 * config/i386/i386.c (construct_container): Remove TFmode check
3742 for X86_64_INTEGER_CLASS.
3744 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
3747 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
3748 only when -Wpsabi is enabled.
3750 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
3753 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
3754 passing arrays in registers are the same as for structs, so remove the
3755 special case for them.
3757 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
3759 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
3760 destination type, extract only the valid bits if the source type is not
3761 integral and has a different mode.
3763 2014-02-19 Richard Biener <rguenther@suse.de>
3766 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
3769 2014-02-19 Richard Biener <rguenther@suse.de>
3772 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
3773 (ipa_modify_call_arguments): Emit an argument load explicitely and
3774 preserve virtual SSA form there and for the replacement call.
3775 Do not update SSA form nor free dominance info.
3777 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3779 * ipa.c (function_and_variable_visibility): Also clear WEAK
3780 flag when disolving COMDAT_GROUP.
3782 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3784 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
3785 * ipa-prop.c (ipa_set_jf_known_type): Return early when
3787 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
3788 do more sanity checks.
3789 (detect_type_change): Return true when giving up early.
3790 (compute_complex_assign_jump_func): Fix type parameter of
3791 ipa_set_ancestor_jf.
3792 (compute_complex_ancestor_jump_func): Likewise.
3793 (update_jump_functions_after_inlining): Fix updating of
3795 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
3797 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3799 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
3800 inline clones when edge disappears.
3802 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
3805 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
3806 Split 64-bit moves into 2 patterns. Do not allow the use of
3807 direct move for TDmode in little endian, since the decimal value
3808 has little endian bytes within a word, but the 64-bit pieces are
3809 ordered in a big endian fashion, and normal subreg's of TDmode are
3811 (mov<mode>_64bit_dm): Likewise.
3812 (movtd_64bit_nodm): Likewise.
3814 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3816 PR tree-optimization/60174
3817 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
3818 statement of an SSA_NAME that occurs in an abnormal PHI node.
3820 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3823 * final.c (SEEN_BB): Remove.
3824 (SEEN_NOTE, SEEN_EMITTED): Renumber.
3825 (final_scan_insn): Don't force_source_line on second
3826 NOTE_INSN_BASIC_BLOCK.
3828 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
3831 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
3832 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
3833 (type_natural_mode): Warn ABI change when %zmm register is not
3834 available for AVX512F vector value passing.
3836 2014-02-18 Kai Tietz <ktietz@redhat.com>
3839 * config/i386/i386.c (ix86_expand_prologue): Use value in
3840 rax register as displacement when restoring %r10 or %rax.
3841 Fix wrong offset when restoring both registers.
3843 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3845 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
3846 assertion with conditional return.
3848 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3849 Uros Bizjak <ubizjak@gmail.com>
3852 * config/i386/driver-i386.c (host_detect_local_cpu): If
3853 YMM state is not saved by the OS, also clear has_f16c. Move
3854 CPUID 0x80000001 handling before YMM state saving checking.
3856 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
3858 PR rtl-optimization/58960
3859 * haifa-sched.c (alloc_global_sched_pressure_data): New,
3860 factored out from ...
3861 (sched_init): ... here.
3862 (free_global_sched_pressure_data): New, factored out from ...
3863 (sched_finish): ... here.
3864 * sched-int.h (free_global_sched_pressure_data): Declare.
3865 * sched-rgn.c (nr_regions_initial): New static global.
3866 (haifa_find_rgns): Initialize it.
3867 (schedule_region): Disable sched-pressure for the newly
3870 2014-02-17 Richard Biener <rguenther@suse.de>
3872 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
3873 release SSA defs of pattern stmts.
3875 2014-02-17 Richard Biener <rguenther@suse.de>
3877 * tree-inline.c (expand_call_inline): Release the virtual
3878 operand defined by the call we are about to inline.
3880 2014-02-17 Richard Biener <rguenther@suse.de>
3882 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
3884 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
3885 Ilya Tocar <ilya.tocar@intel.com>
3887 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
3888 arguments order in builtin.
3889 (_mm512_permutexvar_epi64): Ditto.
3890 (_mm512_mask_permutexvar_epi64): Ditto
3891 (_mm512_maskz_permutexvar_epi32): Ditto
3892 (_mm512_permutexvar_epi32): Ditto
3893 (_mm512_mask_permutexvar_epi32): Ditto
3895 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3897 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
3898 (p8_vmrgow): Likewise.
3900 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3902 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
3905 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3908 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
3909 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
3910 into 64-bit and 32-bit moves. On 64-bit moves, add support for
3911 using direct move instructions on ISA 2.07. Also adjust
3912 instruction length for 64-bit.
3913 (mov<mode>_64bit, TFmode/TDmode): Likewise.
3914 (mov<mode>_32bit, TFmode/TDmode): Likewise.
3916 2014-02-15 Alan Modra <amodra@gmail.com>
3920 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
3921 find_replacement on parts of insn rtl that might be reloaded.
3923 2014-02-15 Richard Biener <rguenther@suse.de>
3925 PR tree-optimization/60183
3926 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
3927 (tree_ssa_phiprop): Calculate and free post-dominators.
3929 2014-02-14 Jeff Law <law@redhat.com>
3931 PR rtl-optimization/60131
3932 * ree.c (get_extended_src_reg): New function.
3933 (combine_reaching_defs): Use it rather than assuming location of REG.
3934 (find_and_remove_re): Verify first operand of extension is
3935 a REG before adding the insns to the copy list.
3937 2014-02-14 Roland McGrath <mcgrathr@google.com>
3939 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
3940 * configure: Regenerated.
3941 * config.in: Regenerated.
3942 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
3943 instead of ASM_SHORT.
3945 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
3946 Richard Earnshaw <rearnsha@arm.com>
3948 PR rtl-optimization/59535
3949 * lra-constraints.c (process_alt_operands): Encourage alternative
3950 when unassigned pseudo class is superset of the alternative class.
3951 (inherit_reload_reg): Don't inherit when optimizing for code size.
3952 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
3953 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
3954 modes not less than 4 for Thumb1.
3956 2014-02-14 Kyle McMartin <kyle@redhat.com>
3959 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
3961 2014-02-14 Richard Biener <rguenther@suse.de>
3963 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
3964 (get_frame_arg): Drop the assert with langhook types_compatible_p.
3965 Do not strip INDIRECT_REFs.
3967 2014-02-14 Richard Biener <rguenther@suse.de>
3970 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
3971 DECL_FUNCTION_SPECIFIC_TARGET.
3972 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
3973 * tree-streamer-out.c (pack_ts_target_option): Remove.
3974 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
3975 (write_ts_function_decl_tree_pointers): Do not stream
3976 DECL_FUNCTION_SPECIFIC_TARGET.
3977 * tree-streamer-in.c (unpack_ts_target_option): Remove.
3978 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
3979 (lto_input_ts_function_decl_tree_pointers): Do not stream
3980 DECL_FUNCTION_SPECIFIC_TARGET.
3982 2014-02-14 Jakub Jelinek <jakub@redhat.com>
3984 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
3985 (get_initial_def_for_induction, vectorizable_induction): Ignore
3986 debug stmts when looking for exit_phi.
3987 (vectorizable_live_operation): Fix up condition.
3989 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3991 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
3992 nreverse() because it changes the content of original tree list.
3994 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3996 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
3997 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
3999 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
4001 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
4002 GNU coding standards.
4004 2014-02-13 Jakub Jelinek <jakub@redhat.com>
4007 * dwarf2out.c (gen_subprogram_die): Don't call
4008 add_calling_convention_attribute if subr_die is old_die.
4010 2014-02-13 Sharad Singhai <singhai@google.com>
4012 * doc/optinfo.texi: Fix order of nodes.
4014 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
4016 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
4017 operands[2], not operands[3].
4019 2014-02-13 Richard Biener <rguenther@suse.de>
4022 * doc/install.texi (ISL): Update recommended version to 0.12.2,
4023 mention the possibility of an in-tree build.
4024 (CLooG): Update recommended version to 0.18.1, mention the
4025 possibility of an in-tree build and clarify that the ISL
4026 bundled with CLooG does not work.
4028 2014-02-13 Jakub Jelinek <jakub@redhat.com>
4031 * expr.c (compress_float_constant): If x is a hard register,
4032 extend into a pseudo and then move to x.
4034 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
4036 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
4037 caused by bad second argument to warning_at() with -mhotpatch and
4038 nested functions (e.g. with gfortran).
4040 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
4042 * opts.c (option_name): Remove "enabled by default" rider.
4044 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
4046 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
4048 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
4049 Uros Bizjak <ubizjak@gmail.com>
4052 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
4053 * configure: Regenerated.
4055 2014-02-12 Richard Biener <rguenther@suse.de>
4057 * vec.c (vec_prefix::calculate_allocation): Move as
4058 inline variant to vec.h.
4059 (vec_prefix::calculate_allocation_1): New out-of-line version.
4060 * vec.h (vec_prefix::calculate_allocation_1): Declare.
4061 (vec_prefix::m_has_auto_buf): Rename to ...
4062 (vec_prefix::m_using_auto_storage): ... this.
4063 (vec_prefix::calculate_allocation): Inline the easy cases
4064 and dispatch to calculate_allocation_1 which doesn't need the
4066 (va_heap::reserve): Use gcc_checking_assert.
4067 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
4068 m_using_auto_storage.
4069 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
4071 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
4072 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
4073 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
4075 2014-02-12 Richard Biener <rguenther@suse.de>
4077 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
4078 when we found a dependence.
4080 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
4082 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
4084 (maybe_fold_stmt): ... into this new function.
4085 * omp-low.c (lower_omp): Update comment.
4087 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
4090 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
4093 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
4095 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
4096 identifiers in comments.
4097 (cortexa53_extra_costs): Likewise.
4098 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
4099 (cortexa7_extra_costs): Likewise.
4100 (cortexa12_extra_costs): Likewise.
4101 (cortexa15_extra_costs): Likewise.
4102 (v7m_extra_costs): Likewise.
4104 2014-02-12 Richard Biener <rguenther@suse.de>
4107 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
4108 of posix_memalign being successful.
4109 (lower_stmt): Restrict lowering of posix_memalign to when
4110 -ftree-bit-ccp is enabled.
4112 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
4114 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
4116 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
4118 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
4120 PR rtl-optimization/60116
4121 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
4122 other_insn once the combination has been validated.
4124 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
4127 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
4129 * ipa-devirt.c: Include demangle.h
4130 (odr_violation_reported): New static variable.
4131 (add_type_duplicate): Update odr_violations.
4132 (maybe_record_node): Add completep parameter; update it.
4133 (record_target_from_binfo): Add COMPLETEP parameter;
4134 update it as needed.
4135 (possible_polymorphic_call_targets_1): Likewise.
4136 (struct polymorphic_call_target_d): Add nonconstruction_targets;
4137 rename FINAL to COMPLETE.
4138 (record_targets_from_bases): Sanity check we found the binfo;
4139 fix COMPLETEP updating.
4140 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
4141 parameter, fix computing of COMPLETEP.
4142 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
4143 at LTO time do demangling.
4144 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
4145 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
4147 (gimple_get_virt_method_for_binfo): Likewise.
4148 * gimple-fold.h (gimple_get_virt_method_for_binfo,
4149 gimple_get_virt_method_for_vtable): Update prototypes.
4151 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
4154 * genautomata.c (add_presence_absence): Fix typo with
4155 {final_}presence_list.
4157 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4160 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
4161 for VSX/Altivec vectors that land in GPR registers.
4163 2014-02-11 Richard Henderson <rth@redhat.com>
4164 Jakub Jelinek <jakub@redhat.com>
4167 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
4168 around drhs if type conversion to lacc->type is not useless.
4170 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4172 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
4174 (cortex-a57.cortex-a53): Likewise.
4175 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
4177 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4179 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
4182 2014-02-11 Renlin Li <Renlin.Li@arm.com>
4184 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
4185 add_options_for_arm_vfp3.
4187 2014-02-11 Jeff Law <law@redhat.com>
4190 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
4191 object with an undesirable mode.
4193 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4196 * config/i386/sol2-9.h: New file.
4197 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
4198 *-*-solaris2.9*): Use it.
4200 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
4202 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
4203 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
4205 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
4207 * config/microblaze/microblaze.c: Extend mcpu version format
4209 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
4211 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
4213 2014-02-10 Richard Henderson <rth@redhat.com>
4216 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
4217 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
4218 ms-abi vs -mno-accumulate-outgoing-args.
4219 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
4220 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
4223 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
4226 * cfgexpand.c (expand_asm_operands): Attach source location to
4227 ASM_INPUT rtx objects.
4228 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
4230 2014-02-10 Nick Clifton <nickc@redhat.com>
4232 * config/mn10300/mn10300.c (popcount): New function.
4233 (mn10300_expand_prologue): Include saved registers in stack usage
4236 2014-02-10 Jeff Law <law@redhat.com>
4239 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
4240 when changing the SET_DEST of a prior insn to avoid an input reload.
4242 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
4244 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
4245 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
4246 -mcall-openbsd, or -mcall-linux.
4247 (CC1_ENDIAN_BIG_SPEC): Remove.
4248 (CC1_ENDIAN_LITTLE_SPEC): Remove.
4249 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4250 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
4251 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
4252 and %cc1_endian_default.
4253 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4255 2014-02-10 Richard Biener <rguenther@suse.de>
4257 PR tree-optimization/60115
4258 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
4259 MEM_REF handling. Properly verify that the accesses are not
4260 out of the objects bound.
4262 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4264 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
4267 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
4269 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
4270 proper constants and fix formatting.
4271 (possible_polymorphic_call_targets): Fix formatting.
4273 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
4274 Ilya Tocar <ilya.tocar@intel.com>
4276 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
4277 (_mm512_loadu_epi32): Renamed into...
4278 (_mm512_loadu_si512): This.
4279 (_mm512_storeu_epi32): Renamed into...
4280 (_mm512_storeu_si512): This.
4281 (_mm512_maskz_ceil_ps): Removed.
4282 (_mm512_maskz_ceil_pd): Ditto.
4283 (_mm512_maskz_floor_ps): Ditto.
4284 (_mm512_maskz_floor_pd): Ditto.
4285 (_mm512_floor_round_ps): Ditto.
4286 (_mm512_floor_round_pd): Ditto.
4287 (_mm512_ceil_round_ps): Ditto.
4288 (_mm512_ceil_round_pd): Ditto.
4289 (_mm512_mask_floor_round_ps): Ditto.
4290 (_mm512_mask_floor_round_pd): Ditto.
4291 (_mm512_mask_ceil_round_ps): Ditto.
4292 (_mm512_mask_ceil_round_pd): Ditto.
4293 (_mm512_maskz_floor_round_ps): Ditto.
4294 (_mm512_maskz_floor_round_pd): Ditto.
4295 (_mm512_maskz_ceil_round_ps): Ditto.
4296 (_mm512_maskz_ceil_round_pd): Ditto.
4297 (_mm512_expand_pd): Ditto.
4298 (_mm512_expand_ps): Ditto.
4299 * config/i386/i386.c (ix86_builtins): Remove
4300 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
4301 (bdesc_args): Ditto.
4302 * config/i386/predicates.md (const1256_operand): New.
4303 (const_1_to_2_operand): Ditto.
4304 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4305 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4306 (*avx512pf_gatherpf<mode>sf): Ditto.
4307 (avx512pf_gatherpf<mode>df): Ditto.
4308 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4309 (*avx512pf_gatherpf<mode>df): Ditto.
4310 (avx512pf_scatterpf<mode>sf): Ditto.
4311 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4312 (*avx512pf_scatterpf<mode>sf): Ditto.
4313 (avx512pf_scatterpf<mode>df): Ditto.
4314 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4315 (*avx512pf_scatterpf<mode>df): Ditto.
4316 (avx512f_expand<mode>): Removed.
4317 (<shift_insn><mode>3<mask_name>): Change predicate type.
4319 2014-02-08 Jakub Jelinek <jakub@redhat.com>
4321 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
4322 not at the end of datarefs vector use ordered_remove to avoid
4323 reordering datarefs vector.
4326 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
4327 mark local addressable non-static vars as GOVD_PRIVATE
4328 instead of GOVD_LOCAL.
4329 * omp-low.c (lower_omp_for): Move gimple_bind_vars
4330 and BLOCK_VARS of gimple_bind_block to new_stmt rather
4334 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
4335 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
4336 assume_aligned or alloc_align attributes.
4337 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
4338 arguments. Handle also assume_aligned and alloc_align attributes.
4339 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
4340 calls to functions with assume_aligned or alloc_align attributes.
4341 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
4343 2014-02-08 Terry Guo <terry.guo@arm.com>
4345 * doc/invoke.texi: Document ARM -march=armv7e-m.
4347 2014-02-08 Jakub Jelinek <jakub@redhat.com>
4349 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
4350 flag on __cilkrts_rethrow builtin.
4353 * ipa-cp.c (determine_versionability): Fail at -O0
4354 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
4355 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
4358 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4361 * tree-inline.c (copy_forbidden): Fail for
4362 __attribute__((optimize (0))) functions.
4364 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4366 * varpool.c: Include pointer-set.h.
4367 (varpool_remove_unreferenced_decls): Variables in other partitions
4368 will not be output; be however careful to not lose information
4371 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4373 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
4374 lookup in the vtable constructor.
4376 2014-02-07 Jeff Law <law@redhat.com>
4379 * config/m68k/m68k.md (ashldi_extsi): Turn into a
4380 define_insn_and_split.
4382 * ipa-inline.c (inline_small_functions): Fix typos.
4384 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4386 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
4387 (s390_can_use_return_insn): Declare.
4388 * config/s390/s390.h (EPILOGUE_USES): Define.
4389 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
4391 (s390_chunkify_start): Handle return JUMP_LABELs.
4392 (s390_early_mach): Emit a main_pool instruction on the entry edge.
4393 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
4394 (s390_can_use_return_insn): New functions.
4395 (s390_fix_long_loop_prediction): Handle conditional returns.
4396 (TARGET_SET_UP_BY_PROLOGUE): Define.
4397 * config/s390/s390.md (ANY_RETURN): New code iterator.
4398 (*creturn, *csimple_return, return, simple_return): New patterns.
4400 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4402 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
4403 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
4404 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
4405 REG_CFA_RESTORE list when deciding not to restore a register.
4407 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4409 * config/s390/s390.c: Include tree-pass.h and context.h.
4410 (s390_early_mach): New function, split out from...
4411 (s390_emit_prologue): ...here.
4412 (pass_data_s390_early_mach): New pass structure.
4413 (pass_s390_early_mach): New class.
4414 (s390_option_override): Create and register early_mach pass.
4415 Move to end of file.
4417 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4419 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
4420 to match for the exit block.
4422 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4424 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
4425 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
4426 Reject misaligned operands.
4428 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4430 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
4432 2014-02-07 Richard Biener <rguenther@suse.de>
4435 * gimple-low.c (lower_builtin_posix_memalign): New function.
4436 (lower_stmt): Call it to lower posix_memalign in a way
4437 to make alignment info accessible.
4439 2014-02-07 Jakub Jelinek <jakub@redhat.com>
4442 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
4443 __builtin_setjmp_receiver.
4445 2014-02-07 Richard Biener <rguenther@suse.de>
4448 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
4449 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
4450 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4451 Handle BUILT_IN_POSIX_MEMALIGN.
4452 (find_func_clobbers): Likewise.
4453 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4454 (call_may_clobber_ref_p_1): Likewise.
4456 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4459 * ipa-devirt.c (record_target_from_binfo): Remove overactive
4462 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4465 * lto-cgraph.c (lto_output_node): Use
4466 symtab_get_symbol_partitioning_class.
4467 (lto_output_varpool_node): likewise.
4468 (symtab_get_symbol_partitioning_class): Move here from
4470 * cgraph.h (symbol_partitioning_class): Likewise.
4471 (symtab_get_symbol_partitioning_class): Declare.
4473 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4475 * ggc.h (ggc_internal_cleared_alloc): New macro.
4476 * vec.h (vec_safe_copy): Handle memory stats.
4477 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
4478 * target-globals.c (save_target_globals): Likewise.
4480 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4483 * expr.c (emit_move_resolve_push): Export; be bit more selective
4484 on when to clear alias set.
4485 * expr.h (emit_move_resolve_push): Declare.
4486 * function.h (struct function): Add tail_call_marked.
4487 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
4488 * config/i386/i386-protos.h (ix86_expand_push): Remove.
4489 * config/i386/i386.md (TImode move expander): De not call
4491 (FP push expanders): Preserve memory attributes.
4492 * config/i386/sse.md (push<mode>1): Remove.
4493 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
4494 (ix86_expand_push): Remove.
4495 * config/i386/mmx.md (push<mode>1): Remove.
4497 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4499 PR rtl-optimization/60030
4500 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
4501 lopart with paradoxical subreg before shifting it up by hprec.
4503 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4505 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
4506 Remove extra newline at end of file.
4507 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
4508 (arm_issue_rate): Handle cortexa57.
4509 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
4510 (cortex-a57.cortex-a53): Likewise.
4512 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4515 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
4516 don't record in REG_FRAME_RELATED_EXPR registers not set in that
4518 (arm_expand_prologue): Adjust all callers.
4519 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
4520 info, registers also at the lowest numbered registers side. Use
4521 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
4525 * var-tracking.c (adjust_mems): Before adding a SET to
4526 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
4528 2014-02-06 Alan Modra <amodra@gmail.com>
4531 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
4532 change SDmode to DDmode when lra_in_progress.
4534 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4537 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
4538 free_data_ref on the dr first, and before goto again also set dr
4539 to the next dr. For simd_lane_access, free old datarefs[i] before
4540 overwriting it. For get_vectype_for_scalar_type failure, don't
4541 free_data_ref if simd_lane_access.
4543 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
4546 * tree.h (opts_for_fn): New inline function.
4547 (opt_for_fn): Define.
4548 * config/i386/i386.c (ix86_function_regparm): Use
4549 opt_for_fn (decl, optimize) instead of optimize.
4551 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
4553 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
4554 for SYMBOL_REF in large memory model.
4556 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4558 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
4560 (cortex-a57): Likewise.
4561 (cortex-a57.cortex-a53): Likewise.
4563 2014-02-06 Yury Gribov <y.gribov@samsung.com>
4564 Kugan Vivekanandarajah <kuganv@linaro.org>
4566 * config/arm/arm.c (arm_vector_alignment_reachable): Check
4568 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
4570 2014-02-06 Richard Biener <rguenther@suse.de>
4572 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
4573 set_loop_copy and initialize_original_copy_tables.
4575 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
4577 * config/aarch64/aarch64-simd.md
4578 (aarch64_ashr_simddi): Change QI to SI.
4580 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4581 Jakub Jelinek <jakub@redhat.com>
4584 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
4587 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4589 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
4590 CODE_FOR_altivec_vpku[hw]um to
4591 CODE_FOR_altivec_vpku[hw]um_direct.
4592 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
4593 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
4594 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
4595 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
4597 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4599 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
4600 generation for -maltivec=be.
4601 (altivec_vsumsws): Simplify redundant test.
4603 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4605 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
4606 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
4607 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
4608 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
4609 gen_altivec_vpkuwum.
4610 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
4612 (altivec_vpks<VI_char>ss): Likewise.
4613 (altivec_vpks<VI_char>us): Likewise.
4614 (altivec_vpku<VI_char>us): Likewise.
4615 (altivec_vpku<VI_char>um): Likewise.
4616 (altivec_vpku<VI_char>um_direct): New (copy of
4617 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
4619 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
4620 target is little endian and -maltivec=be is not specified.
4621 (*altivec_vupkhs<VU_char>_direct): New (copy of
4622 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
4623 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
4624 target is little endian and -maltivec=be is not specified.
4625 (*altivec_vupkls<VU_char>_direct): New (copy of
4626 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
4627 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
4628 little endian and -maltivec=be is not specified.
4629 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
4630 little endian and -maltivec=be is not specified.
4632 2014-02-05 Richard Henderson <rth@redhat.com>
4635 * combine-stack-adj.c: Revert r206943.
4636 * sched-int.h (struct deps_desc): Add last_args_size.
4637 * sched-deps.c (init_deps): Initialize it.
4638 (sched_analyze_insn): Add OUTPUT dependencies between insns that
4639 contain REG_ARGS_SIZE notes.
4641 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4643 * lto-cgraph.c (asm_nodes_output): Make global.
4644 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
4645 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
4646 (driver_handle_option): Handle OPT_fwpa.
4648 2014-02-05 Jakub Jelinek <jakub@redhat.com>
4651 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
4652 a comment typo and formatting issue. If odr_hash hasn't been
4653 created, return vNULL and set *completep to false.
4656 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
4657 bb with no successors.
4659 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
4662 * doc/invoke.texi (-march): Clarify documentation for ARM.
4666 2014-02-05 Richard Biener <rguenther@suse.de>
4668 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
4669 when not vectorizing because of too many alias checks.
4670 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
4671 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
4673 2014-02-05 Nick Clifton <nickc@redhat.com>
4675 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
4676 accept extended registers in any mode when compiling for the MN10300.
4678 2014-02-05 Yury Gribov <y.gribov@samsung.com>
4680 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
4681 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
4682 sanitization attributes.
4683 (can_inline_edge_p): Likewise.
4684 (sanitize_attrs_match_for_inline_p): New function.
4686 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4688 * ipa-prop.c (detect_type_change): Shor circuit testing of
4689 type changes on THIS pointer.
4691 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
4694 * config/pa/pa.c (legitimize_tls_address): Return original address
4695 if not passed a SYMBOL_REF rtx.
4696 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
4698 (pa_emit_move_sequence): Simplify TLS source operands.
4699 (pa_legitimate_constant_p): Reject all TLS constants.
4700 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
4701 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
4703 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4705 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
4706 groups when we know they are controlled by LTO.
4707 * varasm.c (default_binds_local_p_1): If object is in other partition,
4708 it will be resolved locally.
4710 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4712 * config/host-linux.c (linux_gt_pch_use_address): Don't
4713 use SSIZE_MAX because it is not always defined.
4715 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
4718 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
4719 threshold for pseudo splitting.
4720 (update_ebb_live_info): Process call argument hard registers and
4721 hard registers from insn definition too.
4722 (max_small_class_regs_num): New constant.
4723 (inherit_in_ebb): Update live hard regs through EBBs. Update
4724 reloads_num only for small register classes. Don't split for
4727 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
4730 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
4733 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4735 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
4738 2014-02-04 Marek Polacek <polacek@redhat.com>
4740 * gdbinit.in (pel): Define.
4742 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4744 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
4747 2014-02-04 Richard Biener <rguenther@suse.de>
4750 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
4751 in function context local.
4752 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
4753 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
4754 similar to LTO_imported_decl_ref.
4756 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4758 PR tree-optimization/60002
4759 * cgraphclones.c (build_function_decl_skip_args): Clear
4762 PR tree-optimization/60023
4763 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
4764 false to gsi_replace.
4765 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
4766 has been in some EH region and vec_stmt could throw, add
4767 vec_stmt into the same EH region.
4768 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
4769 has no lhs, ignore it.
4770 * internal-fn.c (expand_MASK_LOAD): Likewise.
4773 * tree-inline.c (copy_forbidden): Fail for
4774 __attribute__((optimize (0))) functions.
4777 * omp-low.c (simd_clone_struct_copy): If from->inbranch
4778 is set, copy one less argument.
4779 (expand_simd_clones): Don't subtract clone_info->inbranch
4780 from simd_clone_struct_alloc argument.
4782 PR rtl-optimization/57915
4783 * recog.c (simplify_while_replacing): If all unary/binary/relational
4784 operation arguments are constant, attempt to simplify those.
4787 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
4788 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
4790 2014-02-04 Richard Biener <rguenther@suse.de>
4792 PR tree-optimization/60012
4793 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
4794 TBAA disambiguation to all DDRs.
4796 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4799 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
4800 (LINK_SPEC): Use it for -shared, -shared-libgcc.
4802 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4805 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
4807 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4809 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
4810 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
4812 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4815 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
4816 to figure out targets of polymorphic calls with known decl.
4817 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4818 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
4819 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
4820 (get_polymorphic_call_info): ... here.
4821 (get_polymorphic_call_info_from_invariant): New function.
4823 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4825 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
4826 lookup via vtable pointer; check for type consistency
4827 and turn inconsitent facts into UNREACHABLE.
4828 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4829 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
4830 type inconsistent querries; return UNREACHABLE instead.
4832 2014-02-03 Richard Henderson <rth@twiddle.net>
4835 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
4836 already processed this node.
4837 (normalize_one_pred_1): Pass along mark_set.
4838 (normalize_one_pred): Create and destroy a pointer_set_t.
4839 (normalize_one_pred_chain): Likewise.
4841 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
4843 PR gcov-profile/58602
4844 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
4846 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4849 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
4850 -fno-devirtualize; try to devirtualize by the knowledge of
4851 virtual table pointer given by aggregate propagation.
4852 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4853 (ipa_print_node_jump_functions): Dump also offset that
4854 is relevant for polymorphic calls.
4855 (determine_known_aggregate_parts): Add arg_type parameter; use it
4856 instead of determining the type from pointer type.
4857 (ipa_compute_jump_functions_for_edge): Update call of
4858 determine_known_aggregate_parts.
4859 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
4860 (gimple_get_virt_method_for_binfo): ... here; simplify using
4861 vtable_pointer_value_to_vtable.
4862 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
4863 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
4864 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
4865 (vtable_pointer_value_to_vtable): Break out from ...; handle also
4867 (vtable_pointer_value_to_binfo): ... here.
4868 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
4870 2014-02-03 Teresa Johnson <tejohnson@google.com>
4872 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
4873 redef of outer loop index variable.
4875 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
4879 * doc/extend.texi (Function Attributes): Typo.
4881 2014-02-03 Cong Hou <congh@google.com>
4883 PR tree-optimization/60000
4884 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
4885 if the vectorized statement is a store. A store statement can only
4886 appear at the end of pattern statements.
4888 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4890 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
4891 (ix86_option_override_internal): Default long double to 64-bit for
4892 32-bit Bionic and to 128-bit for 64-bit Bionic.
4894 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
4895 TARGET_LONG_DOUBLE_128 is true.
4896 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
4898 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
4899 (mlong-double-64): Negate -mlong-double-128.
4900 (mlong-double-128): New option.
4902 * config/i386/i386-c.c (ix86_target_macros): Define
4903 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
4905 * doc/invoke.texi: Document -mlong-double-128.
4907 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4909 PR rtl-optimization/60024
4910 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
4912 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
4914 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
4916 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4918 PR rtl-optimization/57662
4919 * sel-sched.c (code_motion_path_driver): Do not mark already not
4920 existing blocks in the visiting bitmap.
4922 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4924 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
4925 on the insn being emitted.
4927 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
4928 Will Deacon <will.deacon@arm.com>
4930 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
4932 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4934 * config/arm/arm-tables.opt: Regenerate.
4936 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4938 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
4939 for vector types other than V16QImode.
4940 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
4941 define_expand, and call altivec_expand_vec_perm_le when producing
4942 code with little endian element order.
4943 (*altivec_vperm_<mode>_internal): New insn having previous
4944 behavior of altivec_vperm_<mode>.
4945 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
4946 altivec_expand_vec_perm_le when producing code with little endian
4948 (*altivec_vperm_<mode>_uns_internal): New insn having previous
4949 behavior of altivec_vperm_<mode>_uns.
4951 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4953 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
4954 (altivec_vsumsws): Add handling for -maltivec=be with a little
4956 (altivec_vsumsws_direct): New.
4957 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
4958 gen_altivec_vsumsws.
4960 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
4962 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
4963 vtable_pointer_value_to_binfo): New functions.
4964 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
4965 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
4967 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
4969 * config/nios2/nios2.md (load_got_register): Initialize GOT
4970 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
4971 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
4973 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
4975 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
4976 preserverd by passthrough, do not propagate the type.
4978 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4980 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
4981 (mips_atomic_assign_expand_fenv): New function.
4982 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
4984 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4986 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
4987 (__builtin_mips_set_fcsr): Likewise.
4988 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
4989 MIPS_USI_FTYPE_VOID.
4990 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
4991 (mips16_expand_set_fcsr): Likewise.
4992 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
4993 (mips16_set_fcsr_stub): Likewise.
4994 (mips16_get_fcsr_one_only_stub): New class.
4995 (mips16_set_fcsr_one_only_stub): Likewise.
4996 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
4997 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
4998 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
4999 (hard_float): New availability predicate.
5000 (mips_builtins): Add get_fcsr and set_fcsr.
5001 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
5002 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
5003 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
5004 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
5005 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
5008 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
5010 * config/mips/mips.c (mips_one_only_stub): New class.
5011 (mips_need_mips16_rdhwr_p): Replace with...
5012 (mips16_rdhwr_stub): ...this new variable.
5013 (mips16_stub_call_address): New function.
5014 (mips16_rdhwr_one_only_stub): New class.
5015 (mips_expand_thread_pointer): Use mips16_stub_call_address.
5016 (mips_output_mips16_rdhwr): Delete.
5017 (mips_finish_stub): New function.
5018 (mips_code_end): Use it to handle rdhwr stubs.
5020 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
5023 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
5024 when calculating size of integer atomic types.
5026 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
5028 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
5030 2014-02-01 Jakub Jelinek <jakub@redhat.com>
5032 PR tree-optimization/60003
5033 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
5034 * profile.c (branch_prob): Use gimple_call_builtin_p
5035 to check for BUILT_IN_SETJMP_RECEIVER.
5036 * tree-inline.c (copy_bb): Call notice_special_calls.
5038 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
5041 * lra-constraints.c (process_alt_operands): Update reload_sum only
5044 2014-01-31 Richard Henderson <rth@redhat.com>
5047 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
5048 until after else_eh is processed.
5050 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
5052 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
5053 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
5054 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
5055 in smmintrin.h, remove them.
5056 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
5057 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
5058 * config/i386/i386.md (ROUND_SAE): Fix value.
5059 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
5060 (const48_operand): New.
5061 * config/i386/subst.md (round), (round_expand): Use
5062 const_4_or_8_to_11_operand.
5063 (round_saeonly), (round_saeonly_expand): Use const48_operand.
5065 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
5067 * config/i386/constraints.md (Yk): Swap meaning with k.
5068 * config/i386/i386.md (movhi_internal): Change Yk to k.
5069 (movqi_internal): Ditto.
5070 (*k<logic><mode>): Ditto.
5073 (kandn<mode>): Ditto.
5074 (*<code>hi_1): Ditto.
5075 (*<code>qi_1): Ditto.
5076 (kxnor<mode>): Ditto.
5077 (kortestzhi): Ditto.
5078 (kortestchi): Ditto.
5080 (*one_cmplhi2_1): Ditto.
5081 (*one_cmplqi2_1): Ditto.
5082 * config/i386/sse.md (): Change k to Yk.
5083 (avx512f_load<mode>_mask): Ditto.
5084 (avx512f_blendm<mode>): Ditto.
5085 (avx512f_store<mode>_mask): Ditto.
5086 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
5087 (avx512f_storedqu<mode>_mask): Ditto.
5088 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
5090 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
5091 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
5092 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
5093 (avx512f_maskcmp<mode>3): Ditto.
5094 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
5095 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
5096 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
5097 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
5098 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
5099 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
5100 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
5101 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
5102 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
5103 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
5104 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
5105 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
5106 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
5107 (vec_extract_lo_<mode>_maskm): Ditto.
5108 (vec_extract_hi_<mode>_maskm): Ditto.
5109 (avx512f_vternlog<mode>_mask): Ditto.
5110 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
5111 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
5112 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
5113 (avx512f_<code>v8div16qi2_mask): Ditto.
5114 (avx512f_<code>v8div16qi2_mask_store): Ditto.
5115 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
5116 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
5117 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
5118 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
5119 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5120 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5121 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5122 (*avx512pf_scatterpf<mode>df_mask): Ditto.
5123 (avx512cd_maskb_vec_dupv8di): Ditto.
5124 (avx512cd_maskw_vec_dupv16si): Ditto.
5125 (avx512f_vpermi2var<mode>3_maskz): Ditto.
5126 (avx512f_vpermi2var<mode>3_mask): Ditto.
5127 (avx512f_vpermi2var<mode>3_mask): Ditto.
5128 (avx512f_vpermt2var<mode>3_maskz): Ditto.
5129 (*avx512f_gathersi<mode>): Ditto.
5130 (*avx512f_gathersi<mode>_2): Ditto.
5131 (*avx512f_gatherdi<mode>): Ditto.
5132 (*avx512f_gatherdi<mode>_2): Ditto.
5133 (*avx512f_scattersi<mode>): Ditto.
5134 (*avx512f_scatterdi<mode>): Ditto.
5135 (avx512f_compress<mode>_mask): Ditto.
5136 (avx512f_compressstore<mode>_mask): Ditto.
5137 (avx512f_expand<mode>_mask): Ditto.
5138 * config/i386/subst.md (mask): Change k to Yk.
5139 (mask_scalar_merge): Ditto.
5142 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
5144 * doc/extend.texi (Vector Extensions): Document ?: in C++.
5146 2014-01-31 Richard Biener <rguenther@suse.de>
5149 * builtins.c (fold_builtin_memory_op): Make sure to not
5150 use a floating-point mode or a boolean or enumeral type for
5153 2014-01-30 DJ Delorie <dj@redhat.com>
5155 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
5156 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
5157 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
5158 whenever main() has an epilogue.
5160 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5162 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
5163 unused variable "field".
5164 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
5165 (vsx_mergeh_<mode>): Likewise.
5166 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
5167 (altivec_vmrghh): Likewise.
5168 (altivec_vmrghw): Likewise.
5169 (altivec_vmrglb): Likewise.
5170 (altivec_vmrglh): Likewise.
5171 (altivec_vmrglw): Likewise.
5172 (altivec_vspltb): Add missing uses.
5173 (altivec_vsplth): Likewise.
5174 (altivec_vspltw): Likewise.
5175 (altivec_vspltsf): Likewise.
5177 2014-01-30 Jakub Jelinek <jakub@redhat.com>
5180 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
5181 frame related instructions.
5183 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
5185 PR rtl-optimization/59959
5186 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
5187 any reload of register whose subreg is invalid.
5189 2014-01-30 Jakub Jelinek <jakub@redhat.com>
5191 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
5192 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
5193 Add missing return type - void.
5195 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5197 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
5198 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
5199 remove element index adjustment for endian (now handled in vsx.md
5201 (altivec_expand_vec_perm_const): Use
5202 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
5203 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
5204 (vsx_xxspltw_<mode>): Adjust element index for little endian.
5205 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
5206 define_expand and a new define_insn *altivec_vspltb_internal;
5207 adjust for -maltivec=be on a little endian target.
5208 (altivec_vspltb_direct): New.
5209 (altivec_vsplth): Divide into a define_expand and a new
5210 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
5211 little endian target.
5212 (altivec_vsplth_direct): New.
5213 (altivec_vspltw): Divide into a define_expand and a new
5214 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
5215 little endian target.
5216 (altivec_vspltw_direct): New.
5217 (altivec_vspltsf): Divide into a define_expand and a new
5218 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
5219 a little endian target.
5221 2014-01-30 Richard Biener <rguenther@suse.de>
5223 PR tree-optimization/59993
5224 * tree-ssa-forwprop.c (associate_pointerplus): Check we
5225 can propagate form the earlier stmt and avoid the transform
5226 when the intermediate result is needed.
5228 2014-01-30 Alangi Derick <alangiderick@gmail.com>
5230 * README.Portability: Fix typo.
5232 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
5234 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
5235 comparison_operator with ordered_comparison_operator.
5237 2014-01-30 Nick Clifton <nickc@redhat.com>
5239 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
5240 Rename to mn10300_store_multiple_regs.
5241 * config/mn10300/mn10300.c: Likewise.
5242 * config/mn10300/mn10300.md (store_movm): Fix typo: call
5243 store_multiple_regs.
5244 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
5245 Call mn10300_store_multiple_regs.
5247 2014-01-30 Nick Clifton <nickc@redhat.com>
5248 DJ Delorie <dj@redhat.com>
5250 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
5251 %fp 2 to keep registers after it properly word-aligned.
5252 (rl78_alloc_physical_registers_umul): Handle the case where both
5253 input operands are the same.
5255 2014-01-30 Richard Biener <rguenther@suse.de>
5257 PR tree-optimization/59903
5258 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
5261 2014-01-30 Jason Merrill <jason@redhat.com>
5264 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
5267 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
5269 2014-01-30 Richard Biener <rguenther@suse.de>
5271 PR tree-optimization/59951
5272 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
5274 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
5277 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
5278 SFmode to DFmode case.
5280 2014-01-29 DJ Delorie <dj@redhat.com>
5282 * config/msp430/msp430.opt (-minrt): New.
5283 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
5285 (ENDFILE_SPEC): Likewise.
5287 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
5289 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
5290 (estimate_function_body_sizes): Use it.
5292 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
5295 * dwarf2out.c (is_cxx_auto): New.
5296 (is_base_type): Use it.
5297 (gen_type_die_with_usage): Likewise.
5299 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5301 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
5302 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
5303 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
5304 -maltivec=be with LE targets.
5305 (vsx_mergeh_<mode>): Likewise.
5306 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
5307 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
5308 (altivec_vmrghb): Replace with define_expand and new
5309 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
5310 (altivec_vmrghb_direct): New define_insn.
5311 (altivec_vmrghh): Replace with define_expand and new
5312 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
5313 (altivec_vmrghh_direct): New define_insn.
5314 (altivec_vmrghw): Replace with define_expand and new
5315 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
5316 (altivec_vmrghw_direct): New define_insn.
5317 (*altivec_vmrghsf): Adjust for endianness.
5318 (altivec_vmrglb): Replace with define_expand and new
5319 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
5320 (altivec_vmrglb_direct): New define_insn.
5321 (altivec_vmrglh): Replace with define_expand and new
5322 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
5323 (altivec_vmrglh_direct): New define_insn.
5324 (altivec_vmrglw): Replace with define_expand and new
5325 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
5326 (altivec_vmrglw_direct): New define_insn.
5327 (*altivec_vmrglsf): Adjust for endianness.
5328 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5329 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5330 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5331 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5332 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5333 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5334 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5335 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5337 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
5339 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
5340 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
5343 2014-01-29 Richard Biener <rguenther@suse.de>
5345 PR tree-optimization/58742
5346 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
5347 associate_pointerplus_align.
5348 (associate_pointerplus_diff): New function.
5349 (associate_pointerplus): Likewise. Call associate_pointerplus_align
5350 and associate_pointerplus_diff.
5352 2014-01-29 Richard Biener <rguenther@suse.de>
5354 * lto-streamer.h (LTO_major_version): Bump to 3.
5355 (LTO_minor_version): Reset to 0.
5357 2014-01-29 Renlin Li <Renlin.Li@arm.com>
5359 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
5360 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
5361 (arm_file_start): Generate correct asm header for armv7ve.
5362 * config/arm/bpabi.h: Add multilib support for armv7ve.
5363 * config/arm/driver-arm.c: Change the architectures of cortex-a7
5364 and cortex-a15 to armv7ve.
5365 * config/arm/t-aprofile: Add multilib support for armv7ve.
5366 * doc/invoke.texi: Document -march=armv7ve.
5368 2014-01-29 Richard Biener <rguenther@suse.de>
5370 PR tree-optimization/58742
5371 * tree-ssa-forwprop.c (associate_plusminus): Return true
5372 if we changed sth, defer EH cleanup to ...
5373 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
5374 (simplify_mult): New function.
5376 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5379 PR tree-optimization/59920
5380 * tree.c (build_common_builtin_nodes): Remove
5381 __builtin_setjmp_dispatcher initialization.
5382 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
5383 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
5384 instead of gsi_after_labels + manually skipping debug stmts.
5385 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
5386 ignore bbs with IFN_ABNORMAL_DISPATCHER.
5387 * tree-inline.c (copy_edges_for_bb): Remove
5388 can_make_abnormal_goto argument, instead add abnormal_goto_dest
5389 argument. Ignore computed_goto_p stmts. Don't call
5390 make_abnormal_goto_edges. If a call might need abnormal edges
5391 for non-local gotos, see if it already has an edge to
5392 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
5393 with true argument, don't do anything then, otherwise add
5394 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
5395 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
5397 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
5398 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
5399 (lower_stmt): Don't set data->calls_builtin_setjmp.
5400 (lower_builtin_setjmp): Adjust comment.
5401 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
5402 * tree-cfg.c (found_computed_goto): Remove.
5403 (factor_computed_gotos): Remove.
5404 (make_goto_expr_edges): Return bool, true for computed gotos.
5405 Don't call make_abnormal_goto_edges.
5406 (build_gimple_cfg): Don't set found_computed_goto, don't call
5407 factor_computed_gotos.
5408 (computed_goto_p): No longer static.
5409 (make_blocks): Don't set found_computed_goto.
5410 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
5411 (make_edges): If make_goto_expr_edges returns true, push bb
5412 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
5413 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
5414 vector. Record mapping between bbs and OpenMP regions if there
5415 are any, adjust make_gimple_omp_edges caller. Call
5416 handle_abnormal_edges.
5417 (make_abnormal_goto_edges): Remove.
5418 * tree-cfg.h (make_abnormal_goto_edges): Remove.
5419 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
5420 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
5421 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
5422 * internal-fn.def (ABNORMAL_DISPATCHER): New.
5423 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
5424 filling *region also set *region_idx to (*region)->entry->index.
5427 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
5428 For REGs set ORIGINAL_REGNO.
5430 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
5432 * doc/md.texi: Mention that a target shouldn't implement
5433 vec_widen_(s|u)mul_even/odd pair if it is less efficient
5436 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5438 PR tree-optimization/59594
5439 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
5440 a copy of the datarefs vector rather than the vector itself.
5442 2014-01-28 Jason Merrill <jason@redhat.com>
5445 * dwarf2out.c (auto_die): New static.
5446 (gen_type_die_with_usage): Handle C++1y 'auto'.
5447 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
5450 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
5453 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
5454 (SPEC_X32): Likewise.
5455 (SPEC_64): Likewise.
5456 * config/i386/i386.c (ix86_option_override_internal): Turn off
5457 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
5459 (x86_file_start): Output .code16gcc for TARGET_16BIT.
5460 * config/i386/i386.h (TARGET_16BIT): New macro.
5461 (TARGET_16BIT_P): Likewise.
5462 * config/i386/i386.opt: Add m16.
5463 * doc/invoke.texi: Document -m16.
5465 2014-01-28 Jakub Jelinek <jakub@redhat.com>
5467 PR preprocessor/59935
5468 * input.c (location_get_source_line): Bail out on when line number
5469 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
5471 2014-01-28 Richard Biener <rguenther@suse.de>
5473 PR tree-optimization/58742
5474 * tree-ssa-forwprop.c (associate_plusminus): Handle
5475 pointer subtraction of the form (T)(P + A) - (T)P.
5477 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5479 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
5482 2014-01-28 Richard Biener <rguenther@suse.de>
5485 2014-01-28 Richard Biener <rguenther@suse.de>
5487 PR rtl-optimization/45364
5488 PR rtl-optimization/59890
5489 * var-tracking.c (local_get_addr_clear_given_value): Handle
5490 already cleared slot.
5491 (val_reset): Handle not allocated local_get_addr_cache.
5492 (vt_find_locations): Use post-order on the inverted CFG.
5494 2014-01-28 Richard Biener <rguenther@suse.de>
5496 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
5498 2014-01-28 Richard Biener <rguenther@suse.de>
5500 PR rtl-optimization/45364
5501 PR rtl-optimization/59890
5502 * var-tracking.c (local_get_addr_clear_given_value): Handle
5503 already cleared slot.
5504 (val_reset): Handle not allocated local_get_addr_cache.
5505 (vt_find_locations): Use post-order on the inverted CFG.
5507 2014-01-28 Alan Modra <amodra@gmail.com>
5509 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
5510 * configure.ac <recursive call for build != host>: Define
5511 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
5512 and LD_FOR_BUILD too.
5513 * configure: Regenerate.
5515 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
5517 * config/i386/i386.c (get_builtin_code_for_version): Separate
5518 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
5519 Broadwell from Haswell.
5521 2014-01-27 Steve Ellcey <sellcey@mips.com>
5523 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
5524 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
5525 * config/mips/mips.c (mips_option_override): Change setting
5527 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
5528 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
5529 Change from Mask to Var.
5531 2014-01-27 Jeff Law <law@redhat.com>
5533 * ipa-inline.c (inline_small_functions): Fix typo.
5535 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
5537 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
5538 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
5539 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
5540 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
5541 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
5542 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
5543 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
5544 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
5545 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
5546 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
5547 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
5548 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
5549 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
5550 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
5551 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
5552 (_mm512_storeu_epi64): Ditto.
5553 (_mm512_cmpge_epi32_mask): Ditto.
5554 (_mm512_cmpge_epu32_mask): Ditto.
5555 (_mm512_cmpge_epi64_mask): Ditto.
5556 (_mm512_cmpge_epu64_mask): Ditto.
5557 (_mm512_cmple_epi32_mask): Ditto.
5558 (_mm512_cmple_epu32_mask): Ditto.
5559 (_mm512_cmple_epi64_mask): Ditto.
5560 (_mm512_cmple_epu64_mask): Ditto.
5561 (_mm512_cmplt_epi32_mask): Ditto.
5562 (_mm512_cmplt_epu32_mask): Ditto.
5563 (_mm512_cmplt_epi64_mask): Ditto.
5564 (_mm512_cmplt_epu64_mask): Ditto.
5565 (_mm512_cmpneq_epi32_mask): Ditto.
5566 (_mm512_cmpneq_epu32_mask): Ditto.
5567 (_mm512_cmpneq_epi64_mask): Ditto.
5568 (_mm512_cmpneq_epu64_mask): Ditto.
5569 (_mm512_expand_pd): Ditto.
5570 (_mm512_expand_ps): Ditto.
5571 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
5572 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
5573 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
5574 * config/i386/i386.c (ix86_builtins): Add
5575 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
5576 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
5577 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
5578 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
5579 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
5580 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
5581 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
5582 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
5583 IX86_BUILTIN_PMOVUSQW512_MEM.
5584 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
5585 __builtin_ia32_pmovsqd512mem_mask,
5586 __builtin_ia32_pmovqd512mem_mask,
5587 __builtin_ia32_pmovusqw512mem_mask,
5588 __builtin_ia32_pmovsqw512mem_mask,
5589 __builtin_ia32_pmovqw512mem_mask,
5590 __builtin_ia32_pmovusdw512mem_mask,
5591 __builtin_ia32_pmovsdw512mem_mask,
5592 __builtin_ia32_pmovdw512mem_mask,
5593 __builtin_ia32_pmovqb512mem_mask,
5594 __builtin_ia32_pmovusqb512mem_mask,
5595 __builtin_ia32_pmovsqb512mem_mask,
5596 __builtin_ia32_pmovusdb512mem_mask,
5597 __builtin_ia32_pmovsdb512mem_mask,
5598 __builtin_ia32_pmovdb512mem_mask.
5599 (bdesc_args): Add __builtin_ia32_expanddf512,
5600 __builtin_ia32_expandsf512.
5601 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
5602 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
5603 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
5604 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
5605 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
5606 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
5607 (avx512f_<code>v8div16qi2_mask_store): This.
5608 (avx512f_expand<mode>): New.
5610 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
5612 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
5614 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5615 (_mm512_prefetch_i32scatter_pd): Ditto.
5616 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5617 (_mm512_prefetch_i64scatter_pd): Ditto.
5618 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5619 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
5620 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5621 (_mm512_prefetch_i32scatter_ps): Ditto.
5622 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5623 (_mm512_prefetch_i64scatter_ps): Ditto.
5624 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5625 * config/i386/i386-builtin-types.def: Define
5626 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
5627 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
5628 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
5629 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
5630 IX86_BUILTIN_SCATTERPFQPD.
5631 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
5632 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
5633 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
5634 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
5635 __builtin_ia32_scatterpfqps.
5636 (ix86_expand_builtin): Expand new built-ins.
5637 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
5638 fix memory access data type.
5639 (*avx512pf_gatherpf<mode>_mask): Ditto.
5640 (*avx512pf_gatherpf<mode>): Ditto.
5641 (avx512pf_scatterpf<mode>): Ditto.
5642 (*avx512pf_scatterpf<mode>_mask): Ditto.
5643 (*avx512pf_scatterpf<mode>): Ditto.
5644 (GATHER_SCATTER_SF_MEM_MODE): New.
5645 (avx512pf_gatherpf<mode>df): Ditto.
5646 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5647 (*avx512pf_scatterpf<mode>df): Ditto.
5649 2014-01-27 Jakub Jelinek <jakub@redhat.com>
5652 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
5653 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
5656 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5658 * common/config/arm/arm-common.c
5659 (arm_rewrite_mcpu): Handle multiple names.
5661 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5663 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5665 * gimple-builder.h (create_gimple_tmp): Delete.
5667 2014-01-27 Christian Bruel <christian.bruel@st.com>
5669 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
5672 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
5674 * config/pa/pa.md (call): Generate indirect long calls to non-local
5675 functions when outputing 32-bit code.
5676 (call_value): Likewise except for special call to buggy powf function.
5678 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
5679 portable runtime and PIC indirect calls.
5680 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
5681 and PIC call sequences. Use ldo instead of blr to set return register
5682 in PIC call sequence.
5684 2014-01-25 Walter Lee <walt@tilera.com>
5686 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
5687 avoid clobbering a live register.
5689 2014-01-25 Walter Lee <walt@tilera.com>
5691 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
5692 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
5693 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
5694 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
5696 2014-01-25 Walter Lee <walt@tilera.com>
5698 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
5699 arguments on even registers.
5700 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
5702 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
5703 (BIGGEST_ALIGNMENT): Ditto.
5704 (BIGGEST_FIELD_ALIGNMENT): Ditto.
5706 2014-01-25 Walter Lee <walt@tilera.com>
5708 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
5709 insns before bundling.
5710 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
5712 2014-01-25 Walter Lee <walt@tilera.com>
5714 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
5715 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
5716 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
5718 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5720 * config/mips/constraints.md (kl): Delete.
5721 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
5722 define expands, using...
5723 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
5724 instructions for MIPS16.
5725 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
5726 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
5728 2014-01-25 Walter Lee <walt@tilera.com>
5730 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
5734 2014-01-25 Walter Lee <walt@tilera.com>
5736 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
5737 (TARGET_EXPAND_TO_RTL_HOOK): Define.
5739 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5741 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
5744 2014-01-25 Jakub Jelinek <jakub@redhat.com>
5746 * print-rtl.c (in_call_function_usage): New var.
5747 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
5748 EXPR_LIST mode as mode and not as reg note name.
5751 * cfgloopmanip.c (copy_loop_info): If
5752 loop->warned_aggressive_loop_optimizations, make sure
5753 the flag is set in target loop too.
5755 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
5757 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
5759 * builtins.def: Likewise.
5760 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
5761 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
5762 * ira.c (ira_setup_eliminable_regset): Likewise.
5763 * omp-low.c (gate_expand_omp): Likewise.
5764 (execute_lower_omp): Likewise.
5765 (diagnose_sb_0): Likewise.
5766 (gate_diagnose_omp_blocks): Likewise.
5767 (simd_clone_clauses_extract): Likewise.
5770 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5772 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
5773 correction for little endian...
5774 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
5777 2014-01-24 Jeff Law <law@redhat.com>
5779 PR tree-optimization/59919
5780 * tree-vrp.c (find_assert_locations_1): Do not register asserts
5781 for non-returning calls.
5783 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
5785 * common/config/aarch64/aarch64-common.c
5786 (aarch64_rewrite_mcpu): Handle multiple names.
5787 * config/aarch64/aarch64.h
5788 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5790 2014-01-24 Dodji Seketeli <dodji@redhat.com>
5792 * input.c (add_file_to_cache_tab): Handle the case where fopen
5795 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
5798 * config/i386/i386.md (pushsf splitter): Get stack adjustment
5799 from push operand if code of push isn't PRE_DEC.
5801 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
5804 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
5805 -mquad-memory-atomic. Update -mquad-memory documentation to say
5806 it is only used for non-atomic loads/stores.
5808 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
5809 -mquad-memory or -mquad-memory-atomic switches.
5811 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
5812 -mquad-memory-atomic to ISA 2.07 support.
5814 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
5815 to separate support of normal quad word memory operations (ldq, stq)
5816 from the atomic quad word memory operations.
5818 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5819 support to separate non-atomic quad word operations from atomic
5820 quad word operations. Disable non-atomic quad word operations in
5821 little endian mode so that we don't have to swap words after the
5822 load and before the store.
5823 (quad_load_store_p): Add comment about atomic quad word support.
5824 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
5825 options printed with -mdebug=reg.
5827 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
5828 -mquad-memory-atomic as the test for whether we have quad word
5829 atomic instructions.
5830 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
5831 or -mp8-vector are used, allow byte/half-word atomic operations.
5833 * config/rs6000/sync.md (load_lockedti): Insure that the address
5834 is a proper indexed or indirect address for the lqarx instruction.
5835 On little endian systems, swap the hi/lo registers after the lqarx
5837 (load_lockedpti): Use indexed_or_indirect_operand predicate to
5838 insure the address is valid for the lqarx instruction.
5839 (store_conditionalti): Insure that the address is a proper indexed
5840 or indirect address for the stqcrx. instruction. On little endian
5841 systems, swap the hi/lo registers before doing the stqcrx.
5843 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
5844 insure the address is valid for the stqcrx. instruction.
5846 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
5847 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
5848 type of quad memory support is available.
5850 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
5853 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
5854 there is a danger of looping.
5856 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
5858 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5859 force flag_ira_loop_pressure if set via command line.
5861 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5863 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
5864 (ashr_simd): New builtin handling DI mode.
5865 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
5866 (aarch64_sshr_simddi): New match pattern.
5867 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
5868 (vshrd_n_s64): Likewise.
5869 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
5871 2014-01-23 Nick Clifton <nickc@redhat.com>
5873 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
5874 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
5875 favour of mcu specific scripts.
5876 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
5879 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5880 Alex Velenko <Alex.Velenko@arm.com>
5882 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
5883 (vaddv_s16): Likewise.
5884 (vaddv_s32): Likewise.
5885 (vaddv_u8): Likewise.
5886 (vaddv_u16): Likewise.
5887 (vaddv_u32): Likewise.
5888 (vaddvq_s8): Likewise.
5889 (vaddvq_s16): Likewise.
5890 (vaddvq_s32): Likewise.
5891 (vaddvq_s64): Likewise.
5892 (vaddvq_u8): Likewise.
5893 (vaddvq_u16): Likewise.
5894 (vaddvq_u32): Likewise.
5895 (vaddvq_u64): Likewise.
5896 (vaddv_f32): Likewise.
5897 (vaddvq_f32): Likewise.
5898 (vaddvq_f64): Likewise.
5899 (vmaxv_f32): Likewise.
5900 (vmaxv_s8): Likewise.
5901 (vmaxv_s16): Likewise.
5902 (vmaxv_s32): Likewise.
5903 (vmaxv_u8): Likewise.
5904 (vmaxv_u16): Likewise.
5905 (vmaxv_u32): Likewise.
5906 (vmaxvq_f32): Likewise.
5907 (vmaxvq_f64): Likewise.
5908 (vmaxvq_s8): Likewise.
5909 (vmaxvq_s16): Likewise.
5910 (vmaxvq_s32): Likewise.
5911 (vmaxvq_u8): Likewise.
5912 (vmaxvq_u16): Likewise.
5913 (vmaxvq_u32): Likewise.
5914 (vmaxnmv_f32): Likewise.
5915 (vmaxnmvq_f32): Likewise.
5916 (vmaxnmvq_f64): Likewise.
5917 (vminv_f32): Likewise.
5918 (vminv_s8): Likewise.
5919 (vminv_s16): Likewise.
5920 (vminv_s32): Likewise.
5921 (vminv_u8): Likewise.
5922 (vminv_u16): Likewise.
5923 (vminv_u32): Likewise.
5924 (vminvq_f32): Likewise.
5925 (vminvq_f64): Likewise.
5926 (vminvq_s8): Likewise.
5927 (vminvq_s16): Likewise.
5928 (vminvq_s32): Likewise.
5929 (vminvq_u8): Likewise.
5930 (vminvq_u16): Likewise.
5931 (vminvq_u32): Likewise.
5932 (vminnmv_f32): Likewise.
5933 (vminnmvq_f32): Likewise.
5934 (vminnmvq_f64): Likewise.
5936 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5938 * config/aarch64/aarch64-simd.md
5939 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
5940 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
5941 (*aarch64_mul3_elt<mode>): Likewise.
5942 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
5943 (*aarch64_mul3_elt_to_64v2df): Likewise.
5944 (*aarch64_mla_elt<mode>): Likewise.
5945 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5946 (*aarch64_mls_elt<mode>): Likewise.
5947 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5948 (*aarch64_fma4_elt<mode>): Likewise.
5949 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5950 (*aarch64_fma4_elt_to_64v2df): Likewise.
5951 (*aarch64_fnma4_elt<mode>): Likewise.
5952 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5953 (*aarch64_fnma4_elt_to_64v2df): Likewise.
5954 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
5955 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
5956 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
5957 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
5958 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5959 (aarch64_sqdmull_lane<mode>_internal): Likewise.
5960 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5962 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
5964 * config/aarch64/aarch64-simd.md
5965 (aarch64_be_checked_get_lane<mode>): New define_expand.
5966 * config/aarch64/aarch64-simd-builtins.def
5967 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
5968 New builtin definition.
5969 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
5970 Use new safe be builtin.
5972 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5974 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
5976 (aarch64_be_st1<mode>): Likewise.
5977 (aarch_ld1<VALL:mode>): Define_expand modified.
5978 (aarch_st1<VALL:mode>): Likewise.
5979 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
5980 (UNSPEC_ST1): Likewise.
5982 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
5984 * config/microblaze/microblaze.md: Add trap insn and attribute
5986 2014-01-23 Dodji Seketeli <dodji@redhat.com>
5988 PR preprocessor/58580
5989 * input.h (location_get_source_line): Take an additional line_size
5991 (void diagnostics_file_cache_fini): Declare new function.
5992 * input.c (struct fcache): New type.
5993 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
5994 New static constants.
5995 (diagnostic_file_cache_init, total_lines_num)
5996 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
5997 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
5998 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
5999 (get_next_line, read_next_line, goto_next_line, read_line_num):
6000 New static function definitions.
6001 (diagnostic_file_cache_fini): New function.
6002 (location_get_source_line): Take an additional output line_len
6003 parameter. Re-write using lookup_or_add_file_to_cache_tab and
6005 * diagnostic.c (diagnostic_finish): Call
6006 diagnostic_file_cache_fini.
6007 (adjust_line): Take an additional input parameter for the length
6008 of the line, rather than calculating it with strlen.
6009 (diagnostic_show_locus): Adjust the use of
6010 location_get_source_line and adjust_line with respect to their new
6011 signature. While displaying a line now, do not stop at the first
6012 null byte. Rather, display the zero byte as a space and keep
6013 going until we reach the size of the line.
6014 * Makefile.in: Add vec.o to OBJS-libcommon
6016 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
6017 Ilya Tocar <ilya.tocar@intel.com>
6019 * config/i386/avx512fintrin.h (_mm512_kmov): New.
6020 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
6021 (__builtin_ia32_kmov16): Ditto.
6022 * config/i386/i386.md (UNSPEC_KMOV): New.
6025 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
6027 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
6028 (_mm512_storeu_si512): Ditto.
6030 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
6033 * rtl.h (get_referenced_operands): Declare.
6034 * recog.c (get_referenced_operands): New function.
6035 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
6036 operands have been referenced when recording LO_SUM references.
6038 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
6040 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
6042 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
6044 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
6045 Enable for generic and recent AMD targets.
6047 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
6049 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
6050 ARG_SIZE note when adjustment was eliminated.
6052 2014-01-22 Jeff Law <law@redhat.com>
6054 PR tree-optimization/59597
6055 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
6056 in file. Accept new argument REGISTERING and use it to modify
6057 dump output appropriately.
6058 (register_jump_thread): Corresponding changes.
6059 (mark_threaded_blocks): Reinstate code to cancel unprofitable
6060 thread paths involving joiner blocks. Add code to dump cancelled
6061 jump threading paths.
6063 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
6065 PR rtl-optimization/59477
6066 * lra-constraints.c (inherit_in_ebb): Process call for living hard
6067 regs. Update reloads_num and potential_reload_hard_regs for all insns.
6069 2014-01-22 Tom Tromey <tromey@redhat.com>
6071 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
6073 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
6075 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
6077 PR rtl-optimization/59896
6078 * lra-constraints.c (process_alt_operands): Check unused note for
6079 matched operands of insn with no output reloads.
6081 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
6083 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
6084 (mips_move_from_gpr_cost): Likewise.
6086 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
6088 PR rtl-optimization/59858
6089 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
6090 ira_class_hard_regs_num.
6091 (process_alt_operands): Increase reject for dying matched operand.
6093 2014-01-21 Jakub Jelinek <jakub@redhat.com>
6096 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
6097 smaller than size, perform several stores or loads and stores
6098 at dst + count - size to store or copy all of size bytes, rather
6099 than just last modesize bytes.
6101 2014-01-20 DJ Delorie <dj@redhat.com>
6103 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
6104 that CLOBBERs are REGs before propogating their values.
6106 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
6109 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
6110 (cgraph_inline_failed_type): New function.
6111 * cgraph.h (DEFCIFCODE): Add type.
6112 (cgraph_inline_failed_type_t): New enum.
6113 (cgraph_inline_failed_type): New prototype.
6114 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
6115 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
6116 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
6117 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
6118 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
6119 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
6120 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
6121 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
6122 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
6123 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
6124 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
6125 OPTIMIZATION_MISMATCH.
6126 * tree-inline.c (expand_call_inline): Emit errors during
6127 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
6129 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
6132 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
6133 mode attribute in insn output.
6135 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
6137 * output.h (output_constant): Delete.
6138 * varasm.c (output_constant): Make private.
6140 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
6142 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
6144 2014-01-20 Jakub Jelinek <jakub@redhat.com>
6147 * tree.h (fold_builtin_strcat): New prototype.
6148 * builtins.c (fold_builtin_strcat): No longer static. Add len
6149 argument, if non-NULL, don't call c_strlen. Optimize
6150 directly into __builtin_memcpy instead of __builtin_strcpy.
6151 (fold_builtin_2): Adjust fold_builtin_strcat caller.
6152 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
6154 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
6156 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6157 for SImode_address_operand operands, having only a REG argument.
6159 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
6161 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
6162 loader name using mbig-endian.
6163 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
6165 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
6167 * doc/invoke.texi (-march): Clarify documentation for AArch64.
6171 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
6173 * config/aarch64/aarch64-protos.h
6174 (aarch64_cannot_change_mode_class_ptr): Declare.
6175 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
6176 aarch64_cannot_change_mode_class_ptr): New.
6177 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
6178 backend hook aarch64_cannot_change_mode_class.
6180 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
6182 * common/config/aarch64/aarch64-common.c
6183 (aarch64_handle_option): Don't handle any option order logic here.
6184 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
6185 selected_cpu, warn on architecture version mismatch.
6186 (aarch64_override_options): Fix parsing order for option strings.
6188 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6189 Iain Sandoe <iain@codesourcery.com>
6192 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
6193 warning. Amend comment to reflect current functionality.
6195 2014-01-20 Richard Biener <rguenther@suse.de>
6198 * builtins.c (fold_builtin_strcat): Remove case better handled
6199 by tree-ssa-strlen.c.
6201 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
6203 * config/aarch64/aarch64.opt
6204 (mcpu, march, mtune): Make case-insensitive.
6206 2014-01-20 Jakub Jelinek <jakub@redhat.com>
6209 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6210 if operands[1] is a REG or ZERO_EXTEND of a REG.
6212 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
6214 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
6216 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
6218 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
6219 long non-pic millicode calls.
6221 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6223 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
6225 2014-01-19 Kito Cheng <kito@0xlab.org>
6227 * builtins.c (expand_movstr): Check movstr expand done or fail.
6229 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6230 H.J. Lu <hongjiu.lu@intel.com>
6233 * config/i386/i386.md (*lea<mode>): Zero-extend return register
6234 to DImode for zero-extended addresses.
6236 2014-01-19 Jakub Jelinek <jakub@redhat.com>
6238 PR rtl-optimization/57763
6239 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
6240 on the new indirect jump_insn and increment LABEL_NUSES (label).
6242 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
6246 * config.gcc (x86_archs): New variable.
6247 (x86_64_archs): Likewise.
6248 (x86_cpus): Likewise.
6249 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
6250 --with-arch/--with-cpu= options.
6251 Support --with-arch=/--with-cpu={nehalem,westmere,
6252 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
6254 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6256 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
6257 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
6259 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6261 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
6263 2014-01-18 Jakub Jelinek <jakub@redhat.com>
6266 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
6267 clear cpp_get_options (parse_in)->warn_unused_macros for
6268 ix86_target_macros_internal with cpp_define.
6270 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
6272 * jump.c (delete_related_insns): Keep (use (insn))s.
6273 * reorg.c (redundant_insn): Check for barriers too.
6275 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6277 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
6279 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
6281 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
6282 call to $$dyncall when TARGET_LONG_CALLS is true.
6284 2014-01-17 Jeff Law <law@redhat.com>
6286 * ree.c (combine_set_extension): Temporarily disable test for
6287 changing number of hard registers.
6289 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
6292 * ipa-inline-analysis.c (inline_free_summary):
6293 Do not free summary of aliases.
6295 2014-01-17 Jakub Jelinek <jakub@redhat.com>
6298 * gimplify.c (gimplify_expr): Use create_tmp_var
6299 instead of create_tmp_var_raw. If cond doesn't have
6300 integral type, don't add the IFN_ANNOTATE builtin at all.
6302 2014-01-17 Martin Jambor <mjambor@suse.cz>
6305 * ipa-cp.c (prev_edge_clone): New variable.
6306 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
6307 Also resize prev_edge_clone vector.
6308 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
6309 (ipcp_edge_removal_hook): New function.
6310 (ipcp_driver): Register ipcp_edge_removal_hook.
6312 2014-01-17 Andrew Pinski <apinski@cavium.com>
6313 Steve Ellcey <sellcey@mips.com>
6316 * config/mips/mips.c (mips_print_operand): Check operand mode instead
6319 2014-01-17 Jeff Law <law@redhat.com>
6322 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
6323 so that pass_ccp runs first.
6325 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6327 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
6328 (ix86_adjust_cost): Use !TARGET_XXX.
6329 (do_reorder_for_imul): Likewise.
6330 (swap_top_of_ready_list): Likewise.
6331 (ix86_sched_reorder): Likewise.
6333 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6335 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6336 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
6337 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
6338 (intel_memset): New. Duplicate slm_memset.
6339 (intel_cost): New. Duplicate slm_cost.
6340 (m_INTEL): New macro.
6341 (processor_target_table): Add "intel".
6342 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
6343 with PROCESSOR_INTEL for "intel".
6344 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
6345 PROCESSOR_SILVERMONT.
6346 (ix86_issue_rate): Likewise.
6347 (ix86_adjust_cost): Likewise.
6348 (ia32_multipass_dfa_lookahead): Likewise.
6349 (swap_top_of_ready_list): Likewise.
6350 (ix86_sched_reorder): Likewise.
6351 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
6352 instead of TARGET_OPT_AGU.
6353 * config/i386/i386.h (TARGET_INTEL): New.
6354 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
6355 (processor_type): Add PROCESSOR_INTEL.
6356 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
6357 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
6359 2014-01-17 Marek Polacek <polacek@redhat.com>
6362 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
6365 2014-01-17 Richard Biener <rguenther@suse.de>
6367 PR tree-optimization/46590
6368 * opts.c (default_options_table): Add entries for
6369 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
6370 all enabled at -O1 but not for -Og.
6371 * common.opt (fbranch-count-reg): Remove Init(1).
6372 (fmove-loop-invariants): Likewise.
6373 (ftree-pta): Likewise.
6375 2014-01-17 Jakub Jelinek <jakub@redhat.com>
6377 * config/i386/i386.c (ix86_data_alignment): For compatibility with
6378 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
6379 decls to at least the GCC 4.8 used alignments.
6382 * tree-nested.c (convert_nonlocal_reference_stmt,
6383 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
6384 of GIMPLE_BIND stmts, adjust associated decls.
6386 2014-01-17 Richard Biener <rguenther@suse.de>
6388 PR tree-optimization/46590
6389 * vec.h (vec<>::bseach): New member function implementing
6390 binary search according to C89 bsearch.
6391 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
6392 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
6393 bitmap pointer again. Make accesses_in_loop a flat array.
6394 (mem_ref_obstack): New global.
6395 (outermost_indep_loop): Adjust for mem_ref->stored changes.
6396 (mark_ref_stored): Likewise.
6397 (ref_indep_loop_p_2): Likewise.
6398 (set_ref_stored_in_loop): New helper function.
6399 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
6400 (memref_free): Adjust.
6401 (record_mem_ref_loc): Simplify.
6402 (gather_mem_refs_stmt): Adjust.
6403 (sort_locs_in_loop_postorder_cmp): New function.
6404 (analyze_memory_references): Sort accesses_in_loop after
6405 loop postorder number.
6406 (find_ref_loc_in_loop_cmp): New function.
6407 (for_all_locs_in_loop): Find relevant cluster of locs in
6408 accesses_in_loop and iterate without recursion.
6409 (execute_sm): Avoid uninit warning.
6410 (struct ref_always_accessed): Simplify.
6411 (ref_always_accessed::operator ()): Likewise.
6412 (ref_always_accessed_p): Likewise.
6413 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
6414 loop postorder numbers here.
6415 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
6418 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
6421 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
6422 on decls for which assemble_alias has been called.
6424 2014-01-17 Nick Clifton <nickc@redhat.com>
6426 * config/msp430/msp430.opt: (mcpu): New option.
6427 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
6428 (msp430_option_override): Parse target_cpu. If the MCU name
6429 matches a generic string, clear target_mcu.
6430 (msp430_attr): Allow numeric interrupt values up to 63.
6431 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
6432 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
6434 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
6436 * config/msp430/msp430.md (popm): Use %J rather than %I.
6437 (addsi3): Use msp430_nonimmediate_operand for operand 2.
6438 (addhi_cy_i): Use immediate_operand for operand 2.
6439 * doc/invoke.texi: Document -mcpu option.
6441 2014-01-17 Richard Biener <rguenther@suse.de>
6443 PR rtl-optimization/38518
6444 * df.h (df_analyze_loop): Declare.
6445 * df-core.c: Include cfgloop.h.
6446 (df_analyze_1): Split out main part of df_analyze.
6447 (df_analyze): Adjust.
6448 (loop_inverted_post_order_compute): New function.
6449 (loop_post_order_compute): Likewise.
6450 (df_analyze_loop): New function avoiding whole-function
6452 * loop-invariant.c (find_defs): Use df_analyze_loop.
6453 (find_invariants): Adjust.
6454 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
6456 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
6458 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
6459 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
6461 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
6463 * ipa-ref.c (ipa_remove_stmt_references): Fix references
6464 traversal when removing references.
6466 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
6469 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
6471 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
6474 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
6475 pushing a reload for an autoinc when we had previously reloaded an
6476 inner part of the address.
6478 2014-01-16 Jakub Jelinek <jakub@redhat.com>
6480 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
6482 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
6483 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
6484 when not giving up or versioning for alias only because of
6486 (vect_analyze_data_ref_dependences): Set to true.
6487 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
6489 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
6490 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
6494 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
6497 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
6498 operand 0 predicate for gathers, use a new pseudo as subtarget.
6500 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6503 * lra-constraints.c (process_alt_operands): Add printing debug info.
6504 Check absence of input/output reloads for matched operands too.
6506 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6508 PR rtl-optimization/59835
6509 * ira.c (ira_init_register_move_cost): Increase cost for
6512 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
6514 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
6516 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
6519 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
6520 non-register objects. Use gen_(high/low)part more consistently.
6523 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
6526 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
6527 endian support, remove tests for WORDS_BIG_ENDIAN.
6528 (p8_mfvsrd_3_<mode>): Likewise.
6529 (reload_gpr_from_vsx<mode>): Likewise.
6530 (reload_gpr_from_vsxsf): Likewise.
6531 (p8_mfvsrd_4_disf): Likewise.
6533 2014-01-16 Richard Biener <rguenther@suse.de>
6535 PR rtl-optimization/46590
6536 * lcm.c (compute_antinout_edge): Use postorder iteration.
6537 (compute_laterin): Use inverted postorder iteration.
6539 2014-01-16 Nick Clifton <nickc@redhat.com>
6542 * varasm.c (output_constant): Return the number of bytes actually
6544 (output_constructor_array_range): Update the field size with the
6545 number of bytes emitted by output_constant.
6546 (output_constructor_regular_field): Likewise. Also do not
6547 complain if the total number of bytes emitted is now greater
6548 than the expected fieldpos.
6549 * output.h (output_constant): Update prototype and descriptive comment.
6551 2014-01-16 Marek Polacek <polacek@redhat.com>
6554 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
6555 it is error_mark_node.
6557 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
6559 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
6560 VALID_AVX256_REG_OR_OI_MODE.
6562 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
6564 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
6565 current procedure should be profiled.
6567 2014-01-15 Andrew Pinski <apinski@cavium.com>
6569 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
6570 of moving from/to the STACK_REG register class.
6572 2014-01-15 Richard Henderson <rth@redhat.com>
6575 * reginfo.c (global_regs_decl): Globalize.
6576 * rtl.h (global_regs_decl): Declare.
6577 * ira.c (do_reload): Diagnose frame_pointer_needed and it
6578 reserved via global_regs.
6580 2014-01-15 Teresa Johnson <tejohnson@google.com>
6582 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
6584 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6586 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
6587 and vmulosh rather than call gen_vec_widen_smult_*.
6588 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
6589 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
6590 (vec_widen_smult_even_v16qi): Likewise.
6591 (vec_widen_umult_even_v8hi): Likewise.
6592 (vec_widen_smult_even_v8hi): Likewise.
6593 (vec_widen_umult_odd_v16qi): Likewise.
6594 (vec_widen_smult_odd_v16qi): Likewise.
6595 (vec_widen_umult_odd_v8hi): Likewise.
6596 (vec_widen_smult_odd_v8hi): Likewise.
6597 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
6598 vmuloub rather than call gen_vec_widen_umult_*.
6599 (vec_widen_umult_lo_v16qi): Likewise.
6600 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
6601 vmulosb rather than call gen_vec_widen_smult_*.
6602 (vec_widen_smult_lo_v16qi): Likewise.
6603 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
6604 rather than call gen_vec_widen_umult_*.
6605 (vec_widen_umult_lo_v8hi): Likewise.
6606 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
6607 rather than call gen_vec_widen_smult_*.
6608 (vec_widen_smult_lo_v8hi): Likewise.
6610 2014-01-15 Jeff Law <law@redhat.com>
6612 PR tree-optimization/59747
6613 * ree.c (find_and_remove_re): Properly handle case where a second
6614 eliminated extension requires widening a copy created for elimination
6615 of a prior extension.
6616 (combine_set_extension): Ensure that the number of hard regs needed
6617 for a destination register does not change when we widen it.
6619 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
6621 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
6622 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
6623 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
6624 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
6625 (avr-*-rtems*): Likewise.
6626 (bfin*-rtems*): Likewise.
6627 (moxie-*-rtems*): Likewise.
6628 (h8300-*-rtems*): Likewise.
6629 (i[34567]86-*-rtems*): Likewise.
6630 (lm32-*-rtems*): Likewise.
6631 (m32r-*-rtems*): Likewise.
6632 (m68k-*-rtems*): Likewise.
6633 (microblaze*-*-rtems*): Likewise.
6634 (mips*-*-rtems*): Likewise.
6635 (powerpc-*-rtems*): Likewise.
6636 (sh-*-rtems*): Likewise.
6637 (sparc-*-rtems*): Likewise.
6638 (sparc64-*-rtems*): Likewise.
6639 (v850-*-rtems*): Likewise.
6640 (m32c-*-rtems*): Likewise.
6642 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
6644 PR rtl-optimization/59511
6645 * ira.c (ira_init_register_move_cost): Use memory costs for some
6646 cases of register move cost calculations.
6647 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
6648 instead of BB frequency.
6649 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
6650 * lra-assigns.c (find_hard_regno_for): Ditto.
6652 2014-01-15 Richard Biener <rguenther@suse.de>
6654 PR tree-optimization/59822
6655 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
6656 (vectorizable_load): Use it to hoist defs of uses of invariant
6657 loads out of the loop.
6659 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
6660 Kugan Vivekanandarajah <kuganv@linaro.org>
6663 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
6666 2014-01-15 Richard Biener <rguenther@suse.de>
6668 PR rtl-optimization/59802
6669 * lcm.c (compute_available): Use inverted postorder to seed
6670 the initial worklist.
6672 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6675 * config/s390/s390.c (s390_preferred_reload_class): Don't return
6676 ADDR_REGS for invalid symrefs in non-PIC code.
6678 2014-01-15 Jakub Jelinek <jakub@redhat.com>
6681 * builtins.c (determine_block_size): Initialize *probable_max_size
6682 even if len_rtx is CONST_INT.
6684 2014-01-14 Andrew Pinski <apinski@cavium.com>
6686 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
6687 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
6688 (cortexa53_tunings): Likewise.
6689 (aarch64_sched_issue_rate): New function.
6690 (TARGET_SCHED_ISSUE_RATE): Define.
6692 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6694 * ira-costs.c (find_costs_and_classes): Add missed
6695 ira_init_register_move_cost_if_necessary.
6697 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6700 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
6702 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
6705 * config/i386/i386.c (type_natural_mode): Add a bool parameter
6706 to indicate if type is used for function return value. Warn ABI
6707 change if the vector mode isn't available for function return value.
6708 (ix86_function_arg_advance): Pass false to type_natural_mode.
6709 (ix86_function_arg): Likewise.
6710 (ix86_gimplify_va_arg): Likewise.
6711 (function_arg_32): Don't warn ABI change.
6712 (ix86_function_value): Pass true to type_natural_mode.
6713 (ix86_return_in_memory): Likewise.
6714 (ix86_struct_value_rtx): Removed.
6715 (TARGET_STRUCT_VALUE_RTX): Likewise.
6717 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
6719 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
6720 converting a conditional jump into a conditional return.
6722 2014-01-14 Richard Biener <rguenther@suse.de>
6724 PR tree-optimization/58921
6725 PR tree-optimization/59006
6726 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
6727 hoisting invariant stmts.
6728 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
6729 invariant loads on the preheader edge if possible.
6731 2014-01-14 Joey Ye <joey.ye@arm.com>
6733 * doc/plugin.texi (Building GCC plugins): Update to C++.
6735 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
6737 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
6738 (_mm_rcp28_round_ss): Ditto.
6739 (_mm_rsqrt28_round_sd): Ditto.
6740 (_mm_rsqrt28_round_ss): Ditto.
6741 (_mm_rcp28_sd): Ditto.
6742 (_mm_rcp28_ss): Ditto.
6743 (_mm_rsqrt28_sd): Ditto.
6744 (_mm_rsqrt28_ss): Ditto.
6745 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
6746 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
6747 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
6748 (IX86_BUILTIN_RCP28SD): Ditto.
6749 (IX86_BUILTIN_RCP28SS): Ditto.
6750 (IX86_BUILTIN_RSQRT28SD): Ditto.
6751 (IX86_BUILTIN_RSQRT28SS): Ditto.
6752 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
6753 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
6754 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
6755 (ix86_expand_special_args_builtin): Expand new FTYPE.
6756 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
6757 (srcp14<mode>): Make insn unary.
6758 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
6759 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
6760 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
6761 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
6762 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
6763 Fix rounding: make it SAE only.
6764 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6766 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6768 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
6769 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
6770 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
6771 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
6772 (round_saeonly_mask_scalar_operand4): Ditto.
6773 (round_saeonly_mask_scalar_op3): Ditto.
6774 (round_saeonly_mask_scalar_op4): Ditto.
6776 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6778 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6779 Implement -maltivec=be for vec_insert and vec_extract.
6781 2014-01-10 DJ Delorie <dj@redhat.com>
6783 * config/msp430/msp430.md (call_internal): Don't allow memory
6784 references with SP as the base register.
6785 (call_value_internal): Likewise.
6786 * config/msp430/constraints.md (Yc): New. For memory references
6787 that don't use SP as a base register.
6789 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
6790 "an integer without a # prefix"
6791 * config/msp430/msp430.md (epilogue_helper): Use it.
6793 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6796 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
6797 AVX512F gather builtins.
6798 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
6799 on gather decls with INTEGER_TYPE masktype.
6800 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
6801 directly into the builtin rather than hoisting it before loop.
6803 PR tree-optimization/59387
6804 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
6805 (scev_const_prop): If folded_casts and type has undefined overflow,
6806 use force_gimple_operand instead of force_gimple_operand_gsi and
6807 for each added stmt if it is assign with
6808 arith_code_with_undefined_signed_overflow, call
6809 rewrite_to_defined_overflow.
6810 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
6811 gimple-fold.h instead.
6812 (arith_code_with_undefined_signed_overflow,
6813 rewrite_to_defined_overflow): Moved to ...
6814 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
6815 rewrite_to_defined_overflow): ... here. No longer static.
6816 Include gimplify-me.h.
6817 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
6818 rewrite_to_defined_overflow): New prototypes.
6820 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6822 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
6824 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
6826 * builtins.c (get_object_alignment_2): Minor tweak.
6827 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
6829 2014-01-13 Christian Bruel <christian.bruel@st.com>
6831 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
6832 optimized non constant lengths.
6834 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6837 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
6838 load as __atomic_load_N if possible.
6840 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6842 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
6844 (rs6000_expand_builtin): Adjust call.
6846 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6849 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
6850 * config/rs6000/rs6000.c: Include target-globals.h.
6851 (rs6000_set_current_function): Instead of doing target_reinit
6852 unconditionally, use save_target_globals_default_opts and
6853 restore_target_globals.
6855 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
6857 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
6858 (rs6000_expand_builtin): Handle mffs and mtfsf.
6859 (rs6000_init_builtins): Define mffs and mtfsf.
6860 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
6861 (rs6000_mffs): New pattern.
6862 (rs6000_mtfsf): New pattern.
6864 2014-01-11 Bin Cheng <bin.cheng@arm.com>
6866 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
6867 Start narrowing with START. Apply candidate-use pair
6868 and check overall cost in narrowing.
6869 (iv_ca_prune): Pass new argument.
6871 2014-01-10 Jeff Law <law@redhat.com>
6874 * ree.c (combine_reaching_defs): Ensure the defining statement
6875 occurs before the extension when optimizing extensions with
6876 different source and destination hard registers.
6878 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6881 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
6882 vtables into the type inheritance graph.
6884 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6886 PR rtl-optimization/59754
6887 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
6888 modes in the REGNO != REGNO case.
6890 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6892 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
6894 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6896 PR tree-optimization/59745
6897 * tree-predcom.c (tree_predictive_commoning_loop): Call
6898 free_affine_expand_cache if giving up because components is NULL.
6900 * target-globals.c (save_target_globals): Allocate < 4KB structs using
6901 GC in payload of target_globals struct instead of allocating them on
6902 the heap and the larger structs separately using GC.
6903 * target-globals.h (struct target_globals): Make regs, hard_regs,
6904 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
6905 of GTY((skip)) and change type to void *.
6906 (reset_target_globals): Cast loads from those fields to corresponding
6909 2014-01-10 Steve Ellcey <sellcey@mips.com>
6912 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
6913 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
6914 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
6916 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
6919 * aarch64-modes.def (CC_Zmode): New flags mode.
6920 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
6921 represents an equality.
6922 (aarch64_get_condition_code): Handle CC_Zmode.
6923 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
6925 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6927 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
6928 extraction in good case.
6930 2014-01-10 Richard Biener <rguenther@suse.de>
6932 PR tree-optimization/59374
6933 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
6934 checking after SLP discovery. Mark stmts not participating
6935 in any SLP instance properly.
6937 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6939 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
6940 when handling a SET rtx.
6942 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6944 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
6945 (cortex-a57): Likewise.
6946 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
6948 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6950 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
6951 non-iwmmxt builtins.
6953 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6957 * ipa-devirt.c record_target_from_binfo): Take as argument
6958 stack of binfos and lookup matching one for virtual inheritance.
6959 (possible_polymorphic_call_targets_1): Update.
6961 2014-01-10 Huacai Chen <chenhc@lemote.com>
6963 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
6964 kernel strings for Loongson-2E/2F/3A.
6966 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6969 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
6970 is_gimple_call before calling gimple_call_internal_p.
6972 2014-01-09 Steve Ellcey <sellcey@mips.com>
6974 * Makefile.in (TREE_FLOW_H): Remove.
6975 (TREE_SSA_H): Add file names from tree-flow.h.
6976 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
6977 * tree.h: Remove tree-flow.h reference.
6978 * hash-table.h: Remove tree-flow.h reference.
6979 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
6980 reference with tree-ssa-loop.h.
6982 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6984 * doc/invoke.texi: Add -maltivec={be,le} options, and document
6985 default element-order behavior for -maltivec.
6986 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
6987 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
6988 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
6989 when targeting big endian, at least for now.
6990 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
6992 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6995 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
6996 var satisfies use_register_for_decl, just take into account type
6997 alignment, rather than decl alignment.
6999 PR tree-optimization/59622
7000 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
7001 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
7002 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
7003 Don't devirtualize for inplace at all. For targets.length () == 1,
7004 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
7006 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
7008 * config/i386/i386.md (cpu): Remove the unused btver1.
7010 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
7012 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
7014 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7017 * tree-core.h (struct target_globals): New forward declaration.
7018 (struct tree_target_option): Add globals field.
7019 * tree.h (TREE_TARGET_GLOBALS): Define.
7020 (prepare_target_option_nodes_for_pch): New prototype.
7021 * target-globals.h (struct target_globals): Define even if
7023 * tree.c (prepare_target_option_node_for_pch,
7024 prepare_target_option_nodes_for_pch): New functions.
7025 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
7026 * config/i386/i386.c: Include target-globals.h.
7027 (ix86_set_current_function): Instead of doing target_reinit
7028 unconditionally, use save_target_globals_default_opts and
7029 restore_target_globals.
7031 2014-01-09 Richard Biener <rguenther@suse.de>
7033 PR tree-optimization/59715
7034 * tree-cfg.h (split_critical_edges): Declare.
7035 * tree-cfg.c (split_critical_edges): Export.
7036 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
7038 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
7040 * cfgexpand.c (expand_stack_vars): Optionally disable
7041 asan stack protection.
7042 (expand_used_vars): Likewise.
7043 (partition_stack_vars): Likewise.
7044 * asan.c (asan_emit_stack_protection): Optionally disable
7045 after return stack usage.
7046 (instrument_derefs): Optionally disable memory access instrumentation.
7047 (instrument_builtin_call): Likewise.
7048 (instrument_strlen_call): Likewise.
7049 (asan_protect_global): Optionally disable global variables protection.
7050 * doc/invoke.texi: Added doc for new options.
7051 * params.def: Added new options.
7052 * params.h: Likewise.
7054 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7056 PR rtl-optimization/59724
7057 * ifcvt.c (cond_exec_process_if_block): Don't call
7058 flow_find_head_matching_sequence with 0 longest_match.
7059 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
7060 non-active insns if !stop_after.
7061 (try_head_merge_bb): Revert 2014-01-07 changes.
7063 2014-01-08 Jeff Law <law@redhat.com>
7065 * ree.c (get_sub_rtx): New function, extracted from...
7066 (merge_def_and_ext): Here.
7067 (combine_reaching_defs): Use get_sub_rtx.
7069 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
7071 * cgraph.h (varpool_variable_node): Do not choke on null node.
7073 2014-01-08 Catherine Moore <clm@codesourcery.com>
7075 * config/mips/mips.md (simple_return): Attempt to use JRC
7077 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
7079 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
7081 PR rtl-optimization/59137
7082 * reorg.c (steal_delay_list_from_target): Call update_block for
7084 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
7086 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7088 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
7089 two duplicate entries.
7091 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
7094 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
7096 * config/mips/mips.c (mips_truncated_op_cost): New function.
7097 (mips_rtx_costs): Adjust test for BADDU.
7098 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
7100 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
7102 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
7103 (*baddu_si): ...this new pattern.
7105 2014-01-08 Jakub Jelinek <jakub@redhat.com>
7108 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
7110 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
7113 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
7115 (expand_expr, expand_normal): Adjust.
7116 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
7117 inner_reference_p. Use inner_reference_p to expand inner references.
7118 (store_expr): Adjust.
7119 * cfgexpand.c (expand_call_stmt): Adjust.
7121 2014-01-08 Rong Xu <xur@google.com>
7123 * gcov-io.c (gcov_var): Move from gcov-io.h.
7124 (gcov_position): Ditto.
7125 (gcov_is_error): Ditto.
7126 (gcov_rewrite): Ditto.
7127 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
7128 only part to libgcc/libgcov.h.
7130 2014-01-08 Marek Polacek <polacek@redhat.com>
7133 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
7135 2014-01-08 Marek Polacek <polacek@redhat.com>
7138 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
7140 2014-01-08 Jakub Jelinek <jakub@redhat.com>
7142 PR rtl-optimization/59649
7143 * stor-layout.c (get_mode_bounds): For BImode return
7144 0 and STORE_FLAG_VALUE.
7146 2014-01-08 Richard Biener <rguenther@suse.de>
7149 * gimple.h (is_gimple_builtin_call): Remove.
7150 (gimple_builtin_call_types_compatible_p): New.
7151 (gimple_call_builtin_p): New overload.
7152 * gimple.c (is_gimple_builtin_call): Remove.
7153 (validate_call): Rename to ...
7154 (gimple_builtin_call_types_compatible_p): ... this and export. Also
7156 (validate_type): New static function.
7157 (gimple_call_builtin_p): New overload and adjust.
7158 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
7159 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
7160 (gimple_fold_stmt_to_constant_1): Likewise.
7161 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
7163 2014-01-08 Richard Biener <rguenther@suse.de>
7166 * gimplify.c (gimplify_expr): Gimplify register-register type
7167 VIEW_CONVERT_EXPRs to separate stmts.
7169 2014-01-07 Jeff Law <law@redhat.com>
7172 * ree.c (combine_set_extension): Handle case where source
7173 and destination registers in an extension insn are different.
7174 (combine_reaching_defs): Allow source and destination registers
7175 in extension to be different under limited circumstances.
7176 (add_removable_extension): Remove restriction that the
7177 source and destination registers in the extension are the same.
7178 (find_and_remove_re): Emit a copy from the extension's
7179 destination to its source after the defining insn if
7180 the source and destination registers are different.
7183 * ifcvt.c (merge_if_block): If we are merging a block with more than
7184 one successor with a block with no successors, remove any BARRIER
7185 after the second block.
7187 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
7189 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
7191 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
7194 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
7195 for 14-bit register offsets when INT14_OK_STRICT is false.
7197 2014-01-07 Roland Stigge <stigge@antcom.de>
7198 Michael Meissner <meissner@linux.vnet.ibm.com>
7201 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
7202 Only check TFmode for SPE constants. Don't check TImode or TDmode.
7204 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
7206 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
7209 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
7211 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
7212 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
7213 rtx is const0_rtx or not.
7215 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
7218 * target-globals.c (save_target_globals): Remove this_fn_optab
7220 * toplev.c: Include optabs.h.
7221 (target_reinit): Temporarily restore the global options if another
7222 set of options are in force.
7224 2014-01-07 Jakub Jelinek <jakub@redhat.com>
7226 PR rtl-optimization/58668
7227 * cfgcleanup.c (flow_find_cross_jump): Don't count
7228 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
7229 to determine what is counted.
7230 (flow_find_head_matching_sequence): Use active_insn_p to determine
7232 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
7234 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
7235 determine what is counted.
7237 PR tree-optimization/59643
7238 * tree-predcom.c (split_data_refs_to_components): If one dr is
7239 read and one write, determine_offset fails and the write isn't
7240 in the bad component, just put the read into the bad component.
7242 2014-01-07 Mike Stump <mikestump@comcast.net>
7243 Jakub Jelinek <jakub@redhat.com>
7246 * tree-core.h (struct tree_optimization_option): Change optabs
7247 type from unsigned char * to void *.
7248 * optabs.c (init_tree_optimization_optabs): Adjust
7249 TREE_OPTIMIZATION_OPTABS initialization.
7251 2014-01-06 Jakub Jelinek <jakub@redhat.com>
7254 * config/i386/i386.h (struct machine_function): Add
7255 no_drap_save_restore field.
7256 * config/i386/i386.c (ix86_save_reg): Use
7257 !cfun->machine->no_drap_save_restore instead of
7258 crtl->stack_realign_needed.
7259 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
7260 this function clears frame_pointer_needed. Set
7261 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
7262 and DRAP reg is needed.
7264 2014-01-06 Marek Polacek <polacek@redhat.com>
7267 * doc/implement-c.texi: Mention that other integer types are
7268 permitted as bit-field types in strictly conforming mode.
7270 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
7272 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
7275 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
7277 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
7279 2014-01-06 Martin Jambor <mjambor@suse.cz>
7282 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
7284 * ipa-prop.c (ipa_print_node_params): Fix indentation.
7286 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
7290 * var-tracking.c (add_stores): Preserve the value of the source even if
7291 we don't record the store.
7293 2014-01-06 Terry Guo <terry.guo@arm.com>
7295 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
7297 2014-01-05 Iain Sandoe <iain@codesourcery.com>
7300 * config/darwin.c (darwin_function_section): Adjust return values to
7301 correspond to optimisation changes made in r206070.
7303 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
7305 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
7306 from prefetch_block tune setting.
7307 (nocona_cost): Correct size of prefetch block to 64.
7309 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
7311 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
7312 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
7313 used to save the static chain register in the computation of the offset
7314 from which the FP registers need to be restored.
7316 2014-01-04 Jakub Jelinek <jakub@redhat.com>
7318 PR tree-optimization/59519
7319 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
7320 ICE if get_current_def (current_new_name) is already non-NULL, as long
7321 as it is a phi result of some other phi in *new_exit_bb that has
7324 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
7325 or vmovdqu* for misaligned_operand.
7326 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
7327 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
7328 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
7329 aligned_mem for AVX512F masked aligned load and store builtins and for
7332 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
7334 PR tree-optimization/59651
7335 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
7336 Address range for negative step should be added by TYPE_SIZE_UNIT.
7338 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
7340 * config/m68k/m68k.c (handle_move_double): Handle pushes with
7341 overlapping registers also for registers other than the stack pointer.
7343 2014-01-03 Marek Polacek <polacek@redhat.com>
7346 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
7349 2014-01-03 Jakub Jelinek <jakub@redhat.com>
7352 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
7355 * config/i386/i386.md (MODE_SIZE): New mode attribute.
7356 (push splitter): Use <P:MODE_SIZE> instead of
7357 GET_MODE_SIZE (<P:MODE>mode).
7358 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
7359 (mov -1, reg peephole2): Likewise.
7360 * config/i386/sse.md (*mov<mode>_internal,
7361 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
7362 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
7363 *<code><mode>3, *andnot<mode>3<mask_name>,
7364 <mask_codefor><code><mode>3<mask_name>): Likewise.
7365 * config/i386/subst.md (mask_mode512bit_condition,
7366 sd_mask_mode512bit_condition): Likewise.
7368 2014-01-02 Xinliang David Li <davidxl@google.com>
7370 PR tree-optimization/59303
7371 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
7372 (dump_predicates): Better output format.
7373 (pred_equal_p): New function.
7374 (is_neq_relop_p): Ditto.
7375 (is_neq_zero_form_p): Ditto.
7376 (pred_expr_equal_p): Ditto.
7377 (pred_neg_p): Ditto.
7378 (simplify_pred): Ditto.
7379 (simplify_preds_2): Ditto.
7380 (simplify_preds_3): Ditto.
7381 (simplify_preds_4): Ditto.
7382 (simplify_preds): Ditto.
7384 (push_to_worklist): Ditto.
7385 (get_pred_info_from_cmp): Ditto.
7386 (is_degenerated_phi): Ditto.
7387 (normalize_one_pred_1): Ditto.
7388 (normalize_one_pred): Ditto.
7389 (normalize_one_pred_chain): Ditto.
7390 (normalize_preds): Ditto.
7391 (normalize_cond_1): Remove function.
7392 (normalize_cond): Ditto.
7393 (is_gcond_subset_of): Ditto.
7394 (is_subset_of_any): Ditto.
7395 (is_or_set_subset_of): Ditto.
7396 (is_and_set_subset_of): Ditto.
7397 (is_norm_cond_subset_of): Ditto.
7398 (pred_chain_length_cmp): Ditto.
7399 (convert_control_dep_chain_into_preds): Type change.
7400 (find_predicates): Ditto.
7401 (find_def_preds): Ditto.
7402 (destroy_predicates_vecs): Ditto.
7403 (find_matching_predicates_in_rest_chains): Ditto.
7404 (use_pred_not_overlap_with_undef_path_pred): Ditto.
7405 (is_pred_expr_subset): Ditto.
7406 (is_pred_chain_subset_of): Ditto.
7407 (is_included_in): Ditto.
7408 (is_superset_of): Ditto.
7410 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7412 Update copyright years.
7414 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7416 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
7417 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
7418 config/arc/arc.md, config/arc/arc.opt,
7419 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
7420 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
7421 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
7422 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
7423 config/linux-protos.h, config/linux.c, config/winnt-c.c,
7424 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
7425 vtable-verify.c, vtable-verify.h: Use the standard form for the
7428 2014-01-02 Tobias Burnus <burnus@net-b.de>
7430 * gcc.c (process_command): Update copyright notice dates.
7431 * gcov-dump.c: Ditto.
7433 * doc/cpp.texi: Bump @copying's copyright year.
7434 * doc/cppinternals.texi: Ditto.
7435 * doc/gcc.texi: Ditto.
7436 * doc/gccint.texi: Ditto.
7437 * doc/gcov.texi: Ditto.
7438 * doc/install.texi: Ditto.
7439 * doc/invoke.texi: Ditto.
7441 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
7443 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
7445 2014-01-01 Jakub Jelinek <jakub@redhat.com>
7447 * config/i386/sse.md (*mov<mode>_internal): Guard
7448 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
7450 PR rtl-optimization/59647
7451 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
7452 new_rtx into UNSIGNED_FLOAT rtxes.
7454 Copyright (C) 2014 Free Software Foundation, Inc.
7456 Copying and distribution of this file, with or without modification,
7457 are permitted in any medium without royalty provided the copyright
7458 notice and this notice are preserved.