1 2014-04-23 Nick Clifton <nickc@redhat.com>
3 * config/i386/cygwin.h (ENDFILE_SPEC): Include
4 default-manifest.o if it can be found in the search path.
5 * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
7 2014-04-23 Terry Guo <terry.guo@arm.com>
9 * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
11 2014-04-23 Richard Biener <rguenther@suse.de>
14 * tree-inline.c (declare_return_variable): Use mark_addressable.
16 2014-04-23 Richard Biener <rguenther@suse.de>
19 * loop-init.c (loop_optimizer_init): Make sure to apply
20 LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
22 2014-04-22 Jakub Jelinek <jakub@redhat.com>
25 * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
27 * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
28 if flag_sanitize_undefined_trap_on_error.
29 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
30 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
31 BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
32 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
33 BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
34 BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
35 BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
36 BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
37 BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
38 * ubsan.c (ubsan_instrument_unreachable): Return
39 __builtin_trap () if flag_sanitize_undefined_trap_on_error.
40 (ubsan_expand_null_ifn): Emit __builtin_trap ()
41 if flag_sanitize_undefined_trap_on_error and
42 __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
43 (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
44 instrument_bool_enum_load): Emit __builtin_trap () if
45 flag_sanitize_undefined_trap_on_error and
46 __builtin_handle_*_abort () if !flag_sanitize_recover.
47 * doc/invoke.texi (-fsanitize-recover,
48 -fsanitize-undefined-trap-on-error): Document.
50 2014-04-22 Christian Bruel <christian.bruel@st.com>
52 * config/sh/sh.md (mov<mode>): Replace movQIHI.
53 Force immediates to SImode.
55 2014-04-22 Sandra Loosemore <sandra@codesourcery.com>
57 * config/nios2/nios2.md (UNSPEC_ROUND): New.
59 * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
60 * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
61 * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
62 (nios2_fpu_insn): Add entry for round.
63 (N2FPU_NO_ERRNO_P): Define.
64 (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
66 * doc/invoke.texi (Nios II Options): Document -mcustom-round.
68 2014-04-22 Richard Henderson <rth@redhat.com>
70 * config/aarch64/aarch64 (addti3, subti3): New expanders.
71 (add<GPI>3_compare0): Remove leading * from name.
72 (add<GPI>3_carryin): Likewise.
73 (sub<GPI>3_compare0): Likewise.
74 (sub<GPI>3_carryin): Likewise.
75 (<su_optab>mulditi3): New expander.
76 (multi3): New expander.
77 (madd<GPI>): Remove leading * from name.
79 2014-04-22 Martin Jambor <mjambor@suse.cz>
81 * cgraphclones.c (cgraph_function_versioning): Copy
82 ipa_transforms_to_apply instead of asserting it is empty.
84 2014-04-22 H.J. Lu <hongjiu.lu@intel.com>
87 * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
88 on count_exp to get mode.
90 2014-04-22 Andrew Pinski <apinski@cavium.com>
92 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
94 * config/aarch64/aarch64.md (tlsie_small): Rename to ...
95 (tlsie_small_<mode>): this and handle PTR.
96 (tlsie_small_sidi): New pattern.
97 (tlsle_small): Change to an expand to handle ILP32.
98 (tlsle_small_<mode>): New pattern.
99 (tlsdesc_small): Rename to ...
100 (tlsdesc_small_<mode>): this and handle PTR.
102 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
104 * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
106 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
108 * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
109 (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
110 (aarch64_types_signed_poly_qualifiers): Likewise.
111 (aarch64_types_unsigned_signed_qualifiers): Likewise.
112 (aarch64_types_poly_signed_qualifiers): Likewise.
113 (TYPES_REINTERP_SS): Type macro added.
114 (TYPES_REINTERP_SU): Likewise.
115 (TYPES_REINTERP_SP): Likewise.
116 (TYPES_REINTERP_US): Likewise.
117 (TYPES_REINTERP_PS): Likewise.
118 (aarch64_fold_builtin): New expression folding added.
119 * config/aarch64/aarch64-simd-builtins.def (REINTERP):
120 Declarations removed.
121 (REINTERP_SS): Declarations added.
122 (REINTERP_US): Likewise.
123 (REINTERP_PS): Likewise.
124 (REINTERP_SU): Likewise.
125 (REINTERP_SP): Likewise.
126 * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
127 (vreinterpretq_p8_f64): Likewise.
128 (vreinterpret_p16_f64): Likewise.
129 (vreinterpretq_p16_f64): Likewise.
130 (vreinterpret_f32_f64): Likewise.
131 (vreinterpretq_f32_f64): Likewise.
132 (vreinterpret_f64_f32): Likewise.
133 (vreinterpret_f64_p8): Likewise.
134 (vreinterpret_f64_p16): Likewise.
135 (vreinterpret_f64_s8): Likewise.
136 (vreinterpret_f64_s16): Likewise.
137 (vreinterpret_f64_s32): Likewise.
138 (vreinterpret_f64_s64): Likewise.
139 (vreinterpret_f64_u8): Likewise.
140 (vreinterpret_f64_u16): Likewise.
141 (vreinterpret_f64_u32): Likewise.
142 (vreinterpret_f64_u64): Likewise.
143 (vreinterpretq_f64_f32): Likewise.
144 (vreinterpretq_f64_p8): Likewise.
145 (vreinterpretq_f64_p16): Likewise.
146 (vreinterpretq_f64_s8): Likewise.
147 (vreinterpretq_f64_s16): Likewise.
148 (vreinterpretq_f64_s32): Likewise.
149 (vreinterpretq_f64_s64): Likewise.
150 (vreinterpretq_f64_u8): Likewise.
151 (vreinterpretq_f64_u16): Likewise.
152 (vreinterpretq_f64_u32): Likewise.
153 (vreinterpretq_f64_u64): Likewise.
154 (vreinterpret_s64_f64): Likewise.
155 (vreinterpretq_s64_f64): Likewise.
156 (vreinterpret_u64_f64): Likewise.
157 (vreinterpretq_u64_f64): Likewise.
158 (vreinterpret_s8_f64): Likewise.
159 (vreinterpretq_s8_f64): Likewise.
160 (vreinterpret_s16_f64): Likewise.
161 (vreinterpretq_s16_f64): Likewise.
162 (vreinterpret_s32_f64): Likewise.
163 (vreinterpretq_s32_f64): Likewise.
164 (vreinterpret_u8_f64): Likewise.
165 (vreinterpretq_u8_f64): Likewise.
166 (vreinterpret_u16_f64): Likewise.
167 (vreinterpretq_u16_f64): Likewise.
168 (vreinterpret_u32_f64): Likewise.
169 (vreinterpretq_u32_f64): Likewise.
171 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
173 * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
174 * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
175 (vreinterpret_p8_s8): Likewise.
176 * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
177 (vreinterpret_p8_s16): Likewise.
178 (vreinterpret_p8_s32): Likewise.
179 (vreinterpret_p8_s64): Likewise.
180 (vreinterpret_p8_f32): Likewise.
181 (vreinterpret_p8_u8): Likewise.
182 (vreinterpret_p8_u16): Likewise.
183 (vreinterpret_p8_u32): Likewise.
184 (vreinterpret_p8_u64): Likewise.
185 (vreinterpret_p8_p16): Likewise.
186 (vreinterpretq_p8_s8): Likewise.
187 (vreinterpretq_p8_s16): Likewise.
188 (vreinterpretq_p8_s32): Likewise.
189 (vreinterpretq_p8_s64): Likewise.
190 (vreinterpretq_p8_f32): Likewise.
191 (vreinterpretq_p8_u8): Likewise.
192 (vreinterpretq_p8_u16): Likewise.
193 (vreinterpretq_p8_u32): Likewise.
194 (vreinterpretq_p8_u64): Likewise.
195 (vreinterpretq_p8_p16): Likewise.
196 (vreinterpret_p16_s8): Likewise.
197 (vreinterpret_p16_s16): Likewise.
198 (vreinterpret_p16_s32): Likewise.
199 (vreinterpret_p16_s64): Likewise.
200 (vreinterpret_p16_f32): Likewise.
201 (vreinterpret_p16_u8): Likewise.
202 (vreinterpret_p16_u16): Likewise.
203 (vreinterpret_p16_u32): Likewise.
204 (vreinterpret_p16_u64): Likewise.
205 (vreinterpret_p16_p8): Likewise.
206 (vreinterpretq_p16_s8): Likewise.
207 (vreinterpretq_p16_s16): Likewise.
208 (vreinterpretq_p16_s32): Likewise.
209 (vreinterpretq_p16_s64): Likewise.
210 (vreinterpretq_p16_f32): Likewise.
211 (vreinterpretq_p16_u8): Likewise.
212 (vreinterpretq_p16_u16): Likewise.
213 (vreinterpretq_p16_u32): Likewise.
214 (vreinterpretq_p16_u64): Likewise.
215 (vreinterpretq_p16_p8): Likewise.
216 (vreinterpret_f32_s8): Likewise.
217 (vreinterpret_f32_s16): Likewise.
218 (vreinterpret_f32_s32): Likewise.
219 (vreinterpret_f32_s64): Likewise.
220 (vreinterpret_f32_u8): Likewise.
221 (vreinterpret_f32_u16): Likewise.
222 (vreinterpret_f32_u32): Likewise.
223 (vreinterpret_f32_u64): Likewise.
224 (vreinterpret_f32_p8): Likewise.
225 (vreinterpret_f32_p16): Likewise.
226 (vreinterpretq_f32_s8): Likewise.
227 (vreinterpretq_f32_s16): Likewise.
228 (vreinterpretq_f32_s32): Likewise.
229 (vreinterpretq_f32_s64): Likewise.
230 (vreinterpretq_f32_u8): Likewise.
231 (vreinterpretq_f32_u16): Likewise.
232 (vreinterpretq_f32_u32): Likewise.
233 (vreinterpretq_f32_u64): Likewise.
234 (vreinterpretq_f32_p8): Likewise.
235 (vreinterpretq_f32_p16): Likewise.
236 (vreinterpret_s64_s8): Likewise.
237 (vreinterpret_s64_s16): Likewise.
238 (vreinterpret_s64_s32): Likewise.
239 (vreinterpret_s64_f32): Likewise.
240 (vreinterpret_s64_u8): Likewise.
241 (vreinterpret_s64_u16): Likewise.
242 (vreinterpret_s64_u32): Likewise.
243 (vreinterpret_s64_u64): Likewise.
244 (vreinterpret_s64_p8): Likewise.
245 (vreinterpret_s64_p16): Likewise.
246 (vreinterpretq_s64_s8): Likewise.
247 (vreinterpretq_s64_s16): Likewise.
248 (vreinterpretq_s64_s32): Likewise.
249 (vreinterpretq_s64_f32): Likewise.
250 (vreinterpretq_s64_u8): Likewise.
251 (vreinterpretq_s64_u16): Likewise.
252 (vreinterpretq_s64_u32): Likewise.
253 (vreinterpretq_s64_u64): Likewise.
254 (vreinterpretq_s64_p8): Likewise.
255 (vreinterpretq_s64_p16): Likewise.
256 (vreinterpret_u64_s8): Likewise.
257 (vreinterpret_u64_s16): Likewise.
258 (vreinterpret_u64_s32): Likewise.
259 (vreinterpret_u64_s64): Likewise.
260 (vreinterpret_u64_f32): Likewise.
261 (vreinterpret_u64_u8): Likewise.
262 (vreinterpret_u64_u16): Likewise.
263 (vreinterpret_u64_u32): Likewise.
264 (vreinterpret_u64_p8): Likewise.
265 (vreinterpret_u64_p16): Likewise.
266 (vreinterpretq_u64_s8): Likewise.
267 (vreinterpretq_u64_s16): Likewise.
268 (vreinterpretq_u64_s32): Likewise.
269 (vreinterpretq_u64_s64): Likewise.
270 (vreinterpretq_u64_f32): Likewise.
271 (vreinterpretq_u64_u8): Likewise.
272 (vreinterpretq_u64_u16): Likewise.
273 (vreinterpretq_u64_u32): Likewise.
274 (vreinterpretq_u64_p8): Likewise.
275 (vreinterpretq_u64_p16): Likewise.
276 (vreinterpret_s8_s16): Likewise.
277 (vreinterpret_s8_s32): Likewise.
278 (vreinterpret_s8_s64): Likewise.
279 (vreinterpret_s8_f32): Likewise.
280 (vreinterpret_s8_u8): Likewise.
281 (vreinterpret_s8_u16): Likewise.
282 (vreinterpret_s8_u32): Likewise.
283 (vreinterpret_s8_u64): Likewise.
284 (vreinterpret_s8_p8): Likewise.
285 (vreinterpret_s8_p16): Likewise.
286 (vreinterpretq_s8_s16): Likewise.
287 (vreinterpretq_s8_s32): Likewise.
288 (vreinterpretq_s8_s64): Likewise.
289 (vreinterpretq_s8_f32): Likewise.
290 (vreinterpretq_s8_u8): Likewise.
291 (vreinterpretq_s8_u16): Likewise.
292 (vreinterpretq_s8_u32): Likewise.
293 (vreinterpretq_s8_u64): Likewise.
294 (vreinterpretq_s8_p8): Likewise.
295 (vreinterpretq_s8_p16): Likewise.
296 (vreinterpret_s16_s8): Likewise.
297 (vreinterpret_s16_s32): Likewise.
298 (vreinterpret_s16_s64): Likewise.
299 (vreinterpret_s16_f32): Likewise.
300 (vreinterpret_s16_u8): Likewise.
301 (vreinterpret_s16_u16): Likewise.
302 (vreinterpret_s16_u32): Likewise.
303 (vreinterpret_s16_u64): Likewise.
304 (vreinterpret_s16_p8): Likewise.
305 (vreinterpret_s16_p16): Likewise.
306 (vreinterpretq_s16_s8): Likewise.
307 (vreinterpretq_s16_s32): Likewise.
308 (vreinterpretq_s16_s64): Likewise.
309 (vreinterpretq_s16_f32): Likewise.
310 (vreinterpretq_s16_u8): Likewise.
311 (vreinterpretq_s16_u16): Likewise.
312 (vreinterpretq_s16_u32): Likewise.
313 (vreinterpretq_s16_u64): Likewise.
314 (vreinterpretq_s16_p8): Likewise.
315 (vreinterpretq_s16_p16): Likewise.
316 (vreinterpret_s32_s8): Likewise.
317 (vreinterpret_s32_s16): Likewise.
318 (vreinterpret_s32_s64): Likewise.
319 (vreinterpret_s32_f32): Likewise.
320 (vreinterpret_s32_u8): Likewise.
321 (vreinterpret_s32_u16): Likewise.
322 (vreinterpret_s32_u32): Likewise.
323 (vreinterpret_s32_u64): Likewise.
324 (vreinterpret_s32_p8): Likewise.
325 (vreinterpret_s32_p16): Likewise.
326 (vreinterpretq_s32_s8): Likewise.
327 (vreinterpretq_s32_s16): Likewise.
328 (vreinterpretq_s32_s64): Likewise.
329 (vreinterpretq_s32_f32): Likewise.
330 (vreinterpretq_s32_u8): Likewise.
331 (vreinterpretq_s32_u16): Likewise.
332 (vreinterpretq_s32_u32): Likewise.
333 (vreinterpretq_s32_u64): Likewise.
334 (vreinterpretq_s32_p8): Likewise.
335 (vreinterpretq_s32_p16): Likewise.
336 (vreinterpret_u8_s8): Likewise.
337 (vreinterpret_u8_s16): Likewise.
338 (vreinterpret_u8_s32): Likewise.
339 (vreinterpret_u8_s64): Likewise.
340 (vreinterpret_u8_f32): Likewise.
341 (vreinterpret_u8_u16): Likewise.
342 (vreinterpret_u8_u32): Likewise.
343 (vreinterpret_u8_u64): Likewise.
344 (vreinterpret_u8_p8): Likewise.
345 (vreinterpret_u8_p16): Likewise.
346 (vreinterpretq_u8_s8): Likewise.
347 (vreinterpretq_u8_s16): Likewise.
348 (vreinterpretq_u8_s32): Likewise.
349 (vreinterpretq_u8_s64): Likewise.
350 (vreinterpretq_u8_f32): Likewise.
351 (vreinterpretq_u8_u16): Likewise.
352 (vreinterpretq_u8_u32): Likewise.
353 (vreinterpretq_u8_u64): Likewise.
354 (vreinterpretq_u8_p8): Likewise.
355 (vreinterpretq_u8_p16): Likewise.
356 (vreinterpret_u16_s8): Likewise.
357 (vreinterpret_u16_s16): Likewise.
358 (vreinterpret_u16_s32): Likewise.
359 (vreinterpret_u16_s64): Likewise.
360 (vreinterpret_u16_f32): Likewise.
361 (vreinterpret_u16_u8): Likewise.
362 (vreinterpret_u16_u32): Likewise.
363 (vreinterpret_u16_u64): Likewise.
364 (vreinterpret_u16_p8): Likewise.
365 (vreinterpret_u16_p16): Likewise.
366 (vreinterpretq_u16_s8): Likewise.
367 (vreinterpretq_u16_s16): Likewise.
368 (vreinterpretq_u16_s32): Likewise.
369 (vreinterpretq_u16_s64): Likewise.
370 (vreinterpretq_u16_f32): Likewise.
371 (vreinterpretq_u16_u8): Likewise.
372 (vreinterpretq_u16_u32): Likewise.
373 (vreinterpretq_u16_u64): Likewise.
374 (vreinterpretq_u16_p8): Likewise.
375 (vreinterpretq_u16_p16): Likewise.
376 (vreinterpret_u32_s8): Likewise.
377 (vreinterpret_u32_s16): Likewise.
378 (vreinterpret_u32_s32): Likewise.
379 (vreinterpret_u32_s64): Likewise.
380 (vreinterpret_u32_f32): Likewise.
381 (vreinterpret_u32_u8): Likewise.
382 (vreinterpret_u32_u16): Likewise.
383 (vreinterpret_u32_u64): Likewise.
384 (vreinterpret_u32_p8): Likewise.
385 (vreinterpret_u32_p16): Likewise.
386 (vreinterpretq_u32_s8): Likewise.
387 (vreinterpretq_u32_s16): Likewise.
388 (vreinterpretq_u32_s32): Likewise.
389 (vreinterpretq_u32_s64): Likewise.
390 (vreinterpretq_u32_f32): Likewise.
391 (vreinterpretq_u32_u8): Likewise.
392 (vreinterpretq_u32_u16): Likewise.
393 (vreinterpretq_u32_u64): Likewise.
394 (vreinterpretq_u32_p8): Likewise.
395 (vreinterpretq_u32_p16): Likewise.
397 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
399 * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
401 * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
404 * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
405 (vqnegd_s64): Likewise.
406 (vqabs_s64): Likewise.
407 (vqabsd_s64): Likewise.
409 2014-04-22 Richard Henderson <rth@redhat.com>
411 * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
412 computation to the top of the loop.
414 2014-04-22 Renlin <renlin.li@arm.com>
415 Jiong Wang <jiong.wang@arm.com>
417 * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
418 * config/aarch64/aarch64.c (aarch64_layout_frame)
419 (aarch64_initial_elimination_offset): Likewise.
421 2014-04-22 Marcus Shawcroft <marcus.shawcroft@arm.com>
423 * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
426 2014-04-22 Richard Sandiford <rdsandiford@googlemail.com>
428 * machmode.h (bitwise_mode_for_mode): Declare.
429 * stor-layout.h (bitwise_type_for_mode): Likewise.
430 * stor-layout.c (bitwise_mode_for_mode): New function.
431 (bitwise_type_for_mode): Likewise.
432 * builtins.c (fold_builtin_memory_op): Use it instead of
433 int_mode_for_mode and build_nonstandard_integer_type.
435 2014-04-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
437 * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
438 (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
439 (*-*-solaris2*): Simplify.
440 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
441 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
442 *-*-solaris2.9* handling.
444 * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
446 (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
447 (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
449 (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
450 * configure: Regenerate.
452 * config/i386/sol2-9.h: Remove.
454 * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
455 (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
456 Remove Solaris 9 references.
458 2014-04-22 Vidya Praveen <vidyapraveen@arm.com>
460 * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
461 (floatuns<GPI:mode><GPF:mode>2): Remove.
462 (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
463 and floatuns conversions.
464 (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
465 and floatuns conversions.
466 * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
467 (w1,w2): New mode attributes for inequal width conversions.
469 2014-04-22 Renlin Li <Renlin.Li@arm.com>
471 * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
472 the output asm format.
474 2014-04-22 James Greenhalgh <james.greenhalgh@arm.com>
476 * config/aarch64/aarch64-simd.md
477 (aarch64_cm<optab>di): Always split.
478 (*aarch64_cm<optab>di): New.
479 (aarch64_cmtstdi): Always split.
480 (*aarch64_cmtstdi): New.
482 2014-04-22 Jakub Jelinek <jakub@redhat.com>
484 PR tree-optimization/60823
485 * omp-low.c (ipa_simd_modify_function_body): Go through
486 all SSA_NAMEs and for those refering to vector arguments
487 which are going to be replaced adjust SSA_NAME_VAR and,
488 if it is a default definition, change it into a non-default
489 definition assigned at the beginning of function from new_decl.
490 (ipa_simd_modify_stmt_ops): Rewritten.
491 * tree-dfa.c (set_ssa_default_def): When removing default def,
492 check for NULL loc instead of NULL *loc.
494 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
496 * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
497 restrictions on core registers for DImode values in Thumb2.
499 2014-04-22 Ian Bolton <ian.bolton@arm.com>
501 * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
502 * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
504 2014-04-22 Ian Bolton <ian.bolton@arm.com>
506 * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
507 (*iordi_notzesidi_di): Likewise.
508 (*iordi_notsesidi_di): Likewise.
510 2014-04-22 Ian Bolton <ian.bolton@arm.com>
512 * config/arm/arm-protos.h (tune_params): New struct members.
513 * config/arm/arm.c: Initialise tune_params per processor.
514 (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
515 for speed, based on new tune_params.
517 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
519 * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
521 * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
523 * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
525 * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
526 * config/aarch64/arm_neon.h (vrnd_f64): Added.
527 (vrnda_f64): Likewise.
528 (vrndi_f64): Likewise.
529 (vrndm_f64): Likewise.
530 (vrndn_f64): Likewise.
531 (vrndp_f64): Likewise.
532 (vrndx_f64): Likewise.
534 2014-04-22 Zhenqiang Chen <zhenqiang.chen@linaro.org>
536 * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
537 GET_MODE_SIZE argument is enum machine_mode.
539 2014-04-22 Jakub Jelinek <jakub@redhat.com>
542 * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
543 value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
545 2014-04-22 Lin Zuojian <manjian2006@gmail.com>
548 * asan.c (asan_emit_stack_protection): Force the base to align to
549 appropriate bits if STRICT_ALIGNMENT. Set shadow_mem align to
550 appropriate bits if STRICT_ALIGNMENT.
551 * cfgexpand.c (expand_stack_vars): Set base_align appropriately
553 (expand_used_vars): Leave a space in the stack frame for alignment
556 2014-04-21 David Malcolm <dmalcolm@redhat.com>
558 * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
560 (gimple_store_p): Likewise.
561 (gimple_assign_load_p): Likewise.
562 (gimple_assign_cast_p): Likewise.
563 (gimple_clobber_p): Likewise.
565 * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
566 rather than a gimple.
567 (gimple_assign_cast_p): Likewise.
569 2014-04-21 Michael Meissner <meissner@linux.vnet.ibm.com>
572 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
573 If mode is DDmode and TARGET_E500_DOUBLE allow move.
575 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
576 more debug information for E500 if -mdebug=reg.
578 2014-04-21 Uros Bizjak <ubizjak@gmail.com>
581 * config/i386/i386.c (ix86_expand_builtin)
582 <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
583 register for target RTX.
584 <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
586 2014-04-18 Cong Hou <congh@google.com>
588 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
589 the widen-mult pattern by handling two operands with different sizes,
590 and operands whose size is smaller than half of the result type.
592 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
594 * ipa-inline.h (INLINE_HINT_known_hot): New hint.
595 * ipa-inline-analysis.c (dump_inline_hints): Dump it.
596 (do_estimate_edge_time): Compute it.
597 * ipa-inline.c (want_inline_small_function_p): Bypass
598 INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
600 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
602 * ipa-inline.c (spec_rem): New static variable.
603 (dump_overall_stats): New function.
604 (dump_inline_stats): New function.
606 2014-04-18 Richard Henderson <rth@redhat.com>
608 * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
609 to GET_MODE_SIZE, not a reg_class_t.
611 2014-04-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
613 * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
614 (vsx_xxmrglw_<mode>): Likewise.
616 2014-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
619 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
620 GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
621 (rs6000_init_hard_regno_mode_ok): Likewise.
623 2014-04-17 Jan Hubicka <hubicka@ucw.cz>
625 * ipa-inline.c (inline_small_functions): Account only non-cold
627 * doc/invoke.texi (inline-unit-growth): Update documentation.
629 2014-04-17 Pat Haugen <pthaugen@us.ibm.com>
631 * config/rs6000/rs6000.md (addti3, subti3): New.
633 2014-04-17 H.J. Lu <hongjiu.lu@intel.com>
636 * config/i386/i386.c (ix86_expand_clear): Remove outdated
637 comment. Check optimize_insn_for_size_p instead of
638 optimize_insn_for_speed_p.
640 2014-04-17 Martin Jambor <mjambor@suse.cz>
642 * gimple-iterator.c (gsi_start_edge): New function.
643 * gimple-iterator.h (gsi_start_edge): Declare.
644 * tree-sra.c (single_non_eh_succ): New function.
645 (disqualify_ops_if_throwing_stmt): Renamed to
646 disqualify_if_bad_bb_terminating_stmt. Allow throwing statements
647 having one non-EH successor BB.
648 (sra_modify_expr): If stmt ends bb, use single non-EH successor to
649 generate loads into replacements.
650 (sra_modify_assign): Likewise and and also use the simple path for
652 (sra_modify_function_body): Commit statements on edges.
654 2014-04-17 Richard Biener <rguenther@suse.de>
657 * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
658 comparison results and add clarifying comment.
660 2014-04-17 Jakub Jelinek <jakub@redhat.com>
662 * genmodes.c (struct mode_data): Add need_bytesize_adj field.
663 (blank_mode): Initialize it.
664 (emit_mode_size_inline, emit_mode_nunits_inline,
665 emit_mode_inner_inline): New functions.
666 (emit_insn_modes_h): Call them and surround their output with
667 #if GCC_VERSION >= 4001 ... #endif.
668 * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
669 For GCC_VERSION >= 4001 use mode_*_inline routines instead of
670 mode_* arrays if the argument is __builtin_constant_p.
671 * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
672 is enum machine_mode.
674 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
676 * passes.c (opt_pass::execute): Adjust.
677 (pass_manager::execute_pass_mode_switching): Likewise.
678 (early_local_passes::execute): Likewise.
679 (execute_one_pass): Pass cfun to the pass's execute method.
680 * tree-pass.h (opt_pass::execute): Add function * argument.
681 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
682 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
683 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
684 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
685 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
686 config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
687 except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.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-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
691 ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
692 lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
693 postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
694 reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
695 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
696 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
697 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
698 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
699 tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
700 tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
701 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
702 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
703 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
704 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
705 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
706 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
707 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
708 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
709 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
712 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
714 * passes.c (opt_pass::gate): Take function * argument.
715 (gate_all_early_local_passes): Merge into
716 (early_local_passes::gate): this.
717 (gate_all_early_optimizations): Merge into
718 (all_early_optimizations::gate): this.
719 (gate_all_optimizations): Mege into
720 (all_optimizations::gate): this.
721 (gate_all_optimizations_g): Merge into
722 (all_optimizations_g::gate): this.
723 (gate_rest_of_compilation): Mege into
724 (rest_of_compilation::gate): this.
725 (gate_postreload): Merge into
726 (postreload::gate): this.
727 (dump_one_pass): Pass cfun to the pass's gate method.
728 (execute_ipa_summary_passes): Likewise.
729 (execute_one_pass): Likewise.
730 (ipa_write_summaries_2): Likewise.
731 (ipa_write_optimization_summaries_1): Likewise.
732 (ipa_read_summaries_1): Likewise.
733 (ipa_read_optimization_summaries_1): Likewise.
734 (execute_ipa_stmt_fixups): Likewise.
735 * tree-pass.h (opt_pass::gate): Add function * argument.
736 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
737 combine-stack-adj.c, combine.c, compare-elim.c,
738 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
739 config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
740 config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
741 dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c, fwprop.c, gcse.c,
742 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
743 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
744 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
745 loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
746 omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
747 reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
748 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
749 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
750 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
751 tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
752 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
753 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
754 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
755 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
756 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
757 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
758 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
759 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
760 tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
761 tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
762 var-tracking.c, vtable-verify.c, web.c: Adjust.
764 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
766 * configure.ac: Check for -Woverloaded-virtual and enable it if found.
767 * configure: Regenerate.
769 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
771 * passes.c (dump_one_pass): don't check pass->has_gate.
772 (execute_ipa_summary_passes): Likewise.
773 (execute_one_pass): Likewise.
774 (ipa_write_summaries_2): Likewise.
775 (ipa_write_optimization_summaries_1): Likewise.
776 (ipa_read_optimization_summaries_1): Likewise.
777 (execute_ipa_stmt_fixups): Likewise.
778 * tree-pass.h (pass_data::has_gate): Remove.
779 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
780 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
781 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
782 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
783 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
784 config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
785 config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
786 dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
787 gimple-low.c, gimple-ssa-isolate-paths.c,
788 gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
789 ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
790 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
791 ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
792 modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
793 recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
794 reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
795 trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
796 tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
797 tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
798 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
799 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
800 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
801 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
802 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
803 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
804 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
805 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
806 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
807 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
808 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
809 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
812 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
814 * pass_manager.h (pass_manager::register_dump_files_1): Remove
816 * passes.c (pass_manager::register_dump_files_1): Merge into
817 (pass_manager::register_dump_files): this, and remove its handling of
818 properties since the pass always has the properties anyway.
819 (pass_manager::pass_manager): Adjust.
821 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
823 * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
824 * passes.c (pass_manager::register_dump_files_1): Remove dead code
825 dealing with properties.
826 (pass_manager::register_dump_files): Adjust.
828 2014-03-20 Mark Wielaard <mjw@redhat.com>
830 * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
831 then represent the bound as normal constant value.
833 2014-04-17 Jakub Jelinek <jakub@redhat.com>
836 Forward port from 4.8 branch
837 2013-07-19 Kirill Yukhin <kirill.yukhin@intel.com>
839 * config/i386/bmiintrin.h (_blsi_u32): New.
843 (_blsmsk_u32): Ditto.
844 (_blsmsk_u64): Ditto.
848 2014-04-17 Kito Cheng <kito@0xlab.org>
850 * gcc.c (used_arg): Prevent out of bound access for multilib_options.
852 2014-04-17 Richard Biener <rguenther@suse.de>
855 * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
856 boolean results for comparisons.
858 2014-04-17 Richard Biener <rguenther@suse.de>
860 PR tree-optimization/60836
861 * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
862 initial PHI args to be gimple values.
864 2014-04-17 Richard Biener <rguenther@suse.de>
866 PR tree-optimization/60841
867 * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
868 * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
869 of stmts to SLP build.
870 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
871 (vect_analyze_slp): Likewise.
872 (vect_analyze_slp_instance): Likewise.
873 (vect_build_slp_tree): Limit overall SLP tree growth.
874 * tree-vectorizer.h (vect_analyze_data_refs,
875 vect_analyze_slp): Adjust prototypes.
877 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
879 * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
882 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
884 * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
885 * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
886 * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
887 for TARGET_SLOW_PSHUFB
889 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
891 * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
892 * config/i386/i386.c (intel_cost): Ditto.
894 2014-04-17 Joey Ye <joey.ye@arm.com>
896 * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
898 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
900 * opts.c (common_handle_option): Disable -fipa-reference coorectly
903 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
905 * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
906 (type_all_derivations_known_p): New predicate.
907 (type_all_ctors_visible_p): New predicate.
908 (type_possibly_instantiated_p): New predicate.
909 (get_odr_type): Compute all_derivations_known.
910 (dump_odr_type): Dump the flag.
911 (maybe_record_type): Cleanup.
912 (record_target_from_binfo): Add bases_to_consider array;
913 record bases for types w/o instances and skip CXX destructor.
914 (possible_polymorphic_call_targets_1): Add bases_to_consider
915 and consider_construction parameters; check if type may have instance.
916 (get_polymorphic_call_info): Set maybe_in_construction to true
917 when we know nothing.
918 (record_targets_from_bases): Skip CXX destructors; they are
919 never called for types in construction.
920 (possible_polymorphic_call_targets): Do not record target when
921 type may not have instance.
923 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
926 * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
927 external aliases alive, too.
929 2014-04-16 Andrew Pinski <apinski@cavium.com>
931 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
934 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
936 * final.c (compute_alignments): Do not apply loop alignment to a block
937 falling through to the exit.
939 2014-04-16 Catherine Moore <clm@codesourcery.com>
941 * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
942 Adjust constraints for microMIPS store patterns.
944 2014-04-16 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
946 * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
948 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
950 * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
951 (append_use): Run at -O0.
952 (append_vdef): Likewise.
953 * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
954 * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
956 2014-04-16 Jakub Jelinek <jakub@redhat.com>
958 PR tree-optimization/60844
959 * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
960 (propagate_op_to_single_use, remove_visited_stmt_chain,
961 linearize_expr, repropagate_negates, reassociate_bb): Use it
962 instead of gsi_remove.
964 2014-04-16 Martin Jambor <mjambor@suse.cz>
966 * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
967 ipa_transforms_to_apply.
968 (cgraph_function_versioning): Assert that old_node has empty
969 ipa_transforms_to_apply.
970 * trans-mem.c (ipa_tm_create_version): Likewise.
971 * tree-inline.c (tree_function_versioning): Do not duplicate
972 ipa_transforms_to_apply.
974 2014-04-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
977 * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
979 Pass necessary as flags on 64-bit Solaris/x86.
980 Use lowercase relocs for x86_64-*-*.
981 * configure: Regenerate.
983 2014-04-15 Jan Hubicka <jh@suse.cz>
985 * ipa-devirt.c (referenced_from_vtable_p): New predicate.
986 (maybe_record_node, likely_target_p): Use it.
988 2014-04-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
991 Revert following patch
993 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
996 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
997 software floating point or no floating point registers, do not
998 allow any type in the FPRs. Eliminate a test for SPE SIMD types
999 in GPRs that occurs after we tested for GPRs that would never be
1002 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1003 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1004 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
1005 specifically allow DDmode, since that does not use the SPE SIMD
1008 2014-03-21 Mark Wielaard <mjw@redhat.com>
1010 * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
1011 as unsigned or int depending on type and value used.
1013 2014-04-15 Richard Biener <rguenther@suse.de>
1015 PR rtl-optimization/56965
1016 * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
1017 * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
1019 * alias.c (true_dependence_1): Do not call
1020 nonoverlapping_component_refs_p.
1021 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
1022 nonoverlapping_component_refs_p.
1023 (indirect_refs_may_alias_p): Likewise.
1025 2014-04-15 Teresa Johnson <tejohnson@google.com>
1027 * cfg.c (dump_bb_info): Fix flags check.
1028 * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
1030 2014-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1032 PR rtl-optimization/60663
1033 * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
1036 2014-04-15 Richard Biener <rguenther@suse.de>
1038 * lto-streamer.h (LTO_major_version): Bump to 4.
1040 2014-04-15 Richard Biener <rguenther@suse.de>
1042 * common.opt (lto_partition_model): New enum.
1043 (flto-partition=): Merge separate options with a single with argument,
1044 add -flto-partition=one support.
1045 * flag-types.h (enum lto_partition_model): Declare.
1046 * opts.c (finish_options): Remove duplicate -flto-partition=
1048 * lto-wrapper.c (run_gcc): Adjust.
1050 2014-04-15 Richard Biener <rguenther@suse.de>
1052 * alias.c (ncr_compar): New function.
1053 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
1055 2014-04-15 Richard Biener <rguenther@suse.de>
1057 * alias.c (record_component_aliases): Do not walk BINFOs.
1059 2014-04-15 Richard Biener <rguenther@suse.de>
1061 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1062 Add struct function argument and adjust.
1063 (find_func_aliases_for_call): Likewise.
1064 (find_func_aliases): Likewise.
1065 (find_func_clobbers): Likewise.
1066 (intra_create_variable_infos): Likewise.
1067 (compute_points_to_sets): Likewise.
1068 (ipa_pta_execute): Adjust. Do not push/pop cfun.
1070 2014-04-15 Richard Biener <rguenther@suse.de>
1072 * tree.c (iterative_hash_expr): Use enum tree_code_class
1073 to store TREE_CODE_CLASS.
1074 (tree_block): Likewise.
1075 (tree_set_block): Likewise.
1076 * tree.h (fold_build_pointer_plus_loc): Use
1077 convert_to_ptrofftype_loc.
1079 2014-04-15 Jakub Jelinek <jakub@redhat.com>
1082 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
1085 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
1087 * cfgloop.h (struct loop): Move force_vectorize down.
1088 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
1089 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
1090 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
1091 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
1092 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
1093 * tree-core.h (enum annot_expr_kind): Add new kind values.
1094 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
1095 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
1097 * tree.def (ANNOTATE_EXPR): Tweak comment.
1099 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1101 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
1104 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
1106 * tree.h (TYPE_IDENTIFIER): Declare.
1107 * tree.c (subrange_type_for_debug_p): Use it.
1108 * godump.c (go_format_type): Likewise.
1109 * dwarf2out.c (is_cxx_auto, modified_type_die,
1110 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
1111 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
1113 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1116 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
1118 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
1120 * config/i386/i386.c (examine_argument): Return bool. Return true if
1121 parameter should be passed in memory.
1122 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
1123 (construct_container): Update calls to examine_argument.
1124 (function_arg_advance_64): Ditto.
1125 (return_in_memory_32): Merge with ix86_return_in_memory.
1126 (return_in_memory_64): Ditto.
1127 (return_in_memory_ms_64): Ditto.
1129 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1131 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
1132 * coverage.c (coverage_compute_profile_id): Handle externally visible
1135 2014-04-14 Martin Jambor <mjambor@suse.cz>
1137 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
1138 DECL_DISREGARD_INLINE_LIMITS functions.
1140 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
1143 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
1145 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
1148 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
1149 optimize_insn_for_speed_p instead of
1150 optimize_function_for_speed_p.
1152 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
1154 * doc/invoke.texi (free): Document AArch64.
1156 2014-04-14 Richard Biener <rguenther@suse.de>
1158 PR tree-optimization/60042
1159 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
1160 (insert_into_preds_of_block): Do not prevent PHI insertion
1161 for REFERENCE exprs here ...
1162 (eliminate_dom_walker::before_dom_children): ... but prevent
1163 their use here under similar conditions when applied to the
1164 IL after PRE optimizations.
1166 2014-04-14 Richard Biener <rguenther@suse.de>
1168 * passes.def: Move early points-to after early SRA.
1170 2014-04-14 Richard Biener <rguenther@suse.de>
1172 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
1173 check for which sign-changes we allow when forwarding
1174 a converted value into a switch.
1176 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
1178 * stor-layout.c (place_field): Finalize non-constant offset for the
1181 2014-04-14 Richard Biener <rguenther@suse.de>
1183 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
1185 (expand_switch_using_bit_tests_p): Likewise.
1186 (process_switch): Compute and pass on speed_p based on the
1188 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
1189 optimize_bb_for_speed_p.
1191 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
1193 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
1194 * function.h (struct function): Rename has_force_vect_loops into
1195 has_force_vectorize_loops.
1196 * lto-streamer-in.c (input_cfg): Adjust for renaming.
1197 (input_struct_function_base): Likewise.
1198 * lto-streamer-out.c (output_cfg): Likewise.
1199 (output_struct_function_base): Likewise.
1200 * omp-low.c (expand_omp_simd): Likewise.
1201 * tree-cfg.c (move_sese_region_to_fn): Likewise.
1202 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1203 (version_loop_for_if_conversion): Likewise.
1204 (tree_if_conversion): Likewise.
1205 (main_tree_if_conversion): Likewise.
1206 (gate_tree_if_conversion): Likewise.
1207 * tree-inline.c (copy_loops): Likewise.
1208 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
1209 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
1210 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
1211 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
1212 * tree-vectorizer.c (vectorize_loops): Likewise.
1213 * tree-vectorizer.h (unlimited_cost_model): Likewise.
1215 2014-04-14 Richard Biener <rguenther@suse.de>
1218 * lto-streamer-out.c (wrap_refs): New function.
1219 (lto_output): Wrap symbol references in global initializes in
1220 type-preserving MEM_REFs.
1222 2014-04-14 Christian Bruel <christian.bruel@st.com>
1224 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
1226 2014-04-14 Christian Bruel <christian.bruel@st.com>
1228 * config/sh/sh.md (setmemqi): New expand pattern.
1229 * config/sh/sh.h (CLEAR_RATIO): Define.
1230 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
1231 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
1233 2014-04-14 Richard Biener <rguenther@suse.de>
1236 * fold-const.c (negate_expr_p): Don't negate directional rounding
1238 (fold_negate_expr): Likewise.
1240 2014-04-14 Richard Biener <rguenther@suse.de>
1242 PR tree-optimization/59817
1243 PR tree-optimization/60453
1244 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
1245 recursion to catch all CHRECs in the scalar evolution and restrict
1246 the predicate for the remains appropriately.
1248 2014-04-12 Catherine Moore <clm@codesourcery.com>
1250 * config/mips/constraints.md: Add new register constraint "kb".
1251 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
1252 (*movhi_internal): Likewise.
1253 (*movqi_internal): Likewise.
1254 * config/mips/mips.h (M16_STORE_REGS): New register class.
1255 (REG_CLASS_NAMES): Add M16_STORE_REGS.
1256 (REG_CLASS_CONTENTS): Likewise.
1257 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
1259 2014-04-11 Tobias Burnus <burnus@net-b.de>
1262 * doc/invoke.texi (-Wformat-signedness): Document it.
1263 (Wformat=2): Mention that this enables -Wformat-signedness.
1265 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1267 * common/config/epiphany/epiphany-common.c
1268 (epiphany_option_optimization_table): Enable section anchors by
1269 default at -O1 or higher.
1270 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
1271 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
1272 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
1273 carries no extra cost.
1274 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
1275 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
1276 * config/epiphany/predicates.md (memclob_operand): New predicate.
1277 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
1278 Use memclob_operand predicate and X constraint for operand 3.
1280 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1282 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
1283 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
1286 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1288 PR rtl-optimization/60651
1289 * mode-switching.c (optimize_mode_switching): Make sure to emit
1290 sets of a lower numbered entity before sets of a higher numbered
1291 entity to a mode of the same or lower priority.
1292 When creating a seginfo for a basic block that starts with a code
1293 label, move the insertion point past the code label.
1294 (new_seginfo): Document and enforce requirement that
1295 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
1296 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
1297 * doc/tm.texi: Regenerate.
1299 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
1302 * config/arc/arc.c (arc_save_restore): Fix assert typo.
1304 2013-04-11 Jakub Jelinek <jakub@redhat.com>
1306 * BASE-VER: Set to 4.10.0.
1308 2014-04-11 Tobias Burnus <burnus@net-b.de>
1311 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
1312 * doc/gcc.texi (Service): Update description in the @menu
1313 * doc/invoke.texi (Option Summary): Remove misplaced and
1316 2014-04-11 Steve Ellcey <sellcey@mips.com>
1317 Jakub Jelinek <jakub@redhat.com>
1320 * expr.c (convert_move): Use emit_store_flag_force instead of
1321 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
1324 2014-04-11 Richard Biener <rguenther@suse.de>
1327 * varasm.c (assemble_alias): Avoid endless error reporting
1328 recursion by setting TREE_ASM_WRITTEN.
1330 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1332 * config/s390/s390.md: Add a splitter for NOT rtx.
1334 2014-04-11 Jakub Jelinek <jakub@redhat.com>
1336 PR rtl-optimization/60663
1337 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
1339 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
1340 Jakub Jelinek <jakub@redhat.com>
1343 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
1344 flag from decl_node to node.
1346 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1349 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
1350 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
1351 ameliorating the cases where it can be.
1353 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
1356 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
1358 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1359 (loadsync_<mode>): Change mode.
1360 (load_quadpti, store_quadpti): New.
1361 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1362 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1363 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
1365 2014-04-09 Cong Hou <congh@google.com>
1368 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
1371 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1373 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
1374 instead of vnor to exploit possible fusion opportunity in the
1376 (altivec_expand_vec_perm_const_le): Likewise.
1378 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
1380 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1381 (loadsync_<mode>): Change mode.
1382 (load_quadpti, store_quadpti): New.
1383 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1384 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1386 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
1389 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
1390 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
1391 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
1393 2014-04-08 Richard Biener <rguenther@suse.de>
1396 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
1397 a 64bit widest int print double-int similar to on HWI64 hosts.
1399 2014-04-08 Richard Biener <rguenther@suse.de>
1401 PR tree-optimization/60785
1402 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
1403 default defs properly.
1405 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
1407 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
1408 (Weffc++): Likewise.
1410 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
1412 * ipa-devirt.c (maybe_record_node): When node is not recorded,
1413 set completep to false rather than true.
1415 2014-04-07 Douglas B Rupp <rupp@adacore.com>
1418 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
1419 ARM_TARGET2_DWARF_FORMAT.
1421 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
1424 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
1425 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
1428 2014-04-07 Richard Biener <rguenther@suse.de>
1430 PR tree-optimization/60766
1431 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
1432 (may_eliminate_iv): Convert cand_value_at result to desired type.
1434 2014-04-07 Jason Merrill <jason@redhat.com>
1437 * common.opt (-fno-gnu-unique): Add.
1438 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
1440 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1442 * haifa-sched.c: Fix outdated function reference and minor
1443 grammar errors in introductory comment.
1445 2014-04-07 Richard Biener <rguenther@suse.de>
1448 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
1450 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
1452 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
1455 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
1456 size accounting for thunks.
1457 (pa_asm_output_mi_thunk): Use final_start_function() and
1458 final_end_function() to output function start and end directives.
1460 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
1462 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
1463 device specific ISA/ feature information. Remove short_sp and
1464 errata_skip ds. Add avr_device_specific_features enum to have device
1466 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
1467 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
1468 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
1469 updated device specific info.
1470 * config/avr/avr-mcus.def: Merge device specific details to
1471 dev_attribute field.
1472 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
1474 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
1475 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
1476 assembler if RMW isa supported by current device.
1477 * config/avr/genmultilib.awk: Update as device info structure changed.
1478 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
1480 2014-04-04 Cong Hou <congh@google.com>
1482 PR tree-optimization/60656
1483 * tree-vect-stmts.c (supportable_widening_operation):
1484 Fix a bug that elements in a vector with vect_used_by_reduction
1485 property are incorrectly reordered when the operation on it is not
1486 consistant with the one in reduction operation.
1488 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
1490 PR rtl-optimization/60155
1491 * gcse.c (record_set_data): New function.
1492 (single_set_gcse): New function.
1493 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
1494 (hoist_code): Likewise.
1495 (get_pressure_class_and_nregs): Likewise.
1497 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
1499 * explow.c (probe_stack_range): Emit a final optimization blockage.
1501 2014-04-04 Anthony Green <green@moxielogic.com>
1503 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
1506 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
1509 * lto-cgraph.c (input_overwrite_node): Check that partitioning
1510 flags are set only during streaming.
1511 * ipa.c (process_references, walk_polymorphic_call_targets,
1512 symtab_remove_unreachable_nodes): Drop bodies of always inline
1513 after early inlining.
1514 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
1516 2014-04-04 Jakub Jelinek <jakub@redhat.com>
1517 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1520 * dwarf2out.c (const_ok_for_output_1): Reject expressions
1523 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1526 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
1528 (cortex_a53_fdivd): Likewise.
1530 2014-04-04 Martin Jambor <mjambor@suse.cz>
1533 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
1535 * cgraph.c (clone_of_p): Also return true if thunks match.
1536 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
1537 cgraph_function_or_thunk_node and an obsolete comment.
1538 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
1540 (build_function_decl_skip_args): Likewise.
1541 (set_new_clone_decl_and_node_flags): New function.
1542 (duplicate_thunk_for_node): Likewise.
1543 (redirect_edge_duplicating_thunks): Likewise.
1544 (cgraph_clone_node): New parameter args_to_skip, pass it to
1545 redirect_edge_duplicating_thunks which is called instead of
1546 cgraph_redirect_edge_callee.
1547 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
1548 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
1550 2014-04-04 Jeff Law <law@redhat.com>
1553 * config/arm/predicates.md (const_int_I_operand): New predicate.
1554 (const_int_M_operand): Similarly.
1555 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
1557 (insv_t2, extv_reg, extzv_t2): Likewise.
1558 (load_multiple_with_writeback): Similarly for const_int_I_operand.
1559 (pop_multiple_with_writeback_and_return): Likewise.
1560 (vfp_pop_multiple_with_writeback): Likewise
1562 2014-04-04 Richard Biener <rguenther@suse.de>
1565 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
1566 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
1568 * gimplify.h (gimple_add_tmp_var_fn): Declare.
1569 * gimplify.c (gimple_add_tmp_var_fn): New function.
1570 * gimple-expr.h (create_tmp_reg_fn): Declare.
1571 * gimple-expr.c (create_tmp_reg_fn): New function.
1572 * gimple-low.c (record_vars_into): Don't change cfun.
1573 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
1574 code generation without cfun.
1576 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
1579 * Makefile.in (install-driver): Fix shell scripting.
1581 2014-04-03 Cong Hou <congh@google.com>
1583 PR tree-optimization/60505
1584 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
1585 threshold of number of iterations below which no vectorization
1587 * tree-vect-loop.c (new_loop_vec_info):
1588 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
1589 * tree-vect-loop.c (vect_analyze_loop_operations):
1590 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
1591 * tree-vect-loop.c (vect_transform_loop):
1592 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
1593 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
1594 of iterations of the loop and see if we should build the epilogue.
1596 2014-04-03 Richard Biener <rguenther@suse.de>
1598 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
1599 (streamer_tree_cache_create): Adjust.
1600 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
1601 to allow optional nodes array.
1602 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
1603 (streamer_tree_cache_append): Likewise.
1604 (streamer_tree_cache_create): Create nodes array optionally
1605 as specified by parameter.
1606 * lto-streamer-out.c (create_output_block): Avoid maintaining
1607 the node array in the writer cache.
1608 (DFS_write_tree): Remove assertion.
1609 (produce_asm_for_decls): Free the out decl state hash table early.
1610 * lto-streamer-in.c (lto_data_in_create): Adjust for
1611 streamer_tree_cache_create prototype change.
1613 2014-04-03 Richard Biener <rguenther@suse.de>
1615 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
1616 set TREE_CHAIN to NULL_TREE.
1618 2014-04-03 Richard Biener <rguenther@suse.de>
1620 PR tree-optimization/60740
1621 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
1622 over all GIMPLE_COND operands.
1624 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
1626 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
1627 (Weffc++): Remove Scott's numbering, merge lists and reference
1630 2014-04-03 Nick Clifton <nickc@redhat.com>
1632 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
1635 2014-04-03 Martin Jambor <mjambor@suse.cz>
1637 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
1638 mention gcc_unreachable before failing.
1639 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
1642 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
1645 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
1646 inconsistent code and instead mark the context inconsistent.
1647 (possible_polymorphic_call_targets): For inconsistent contexts
1648 return empty complete list.
1650 2014-04-02 Anthony Green <green@moxielogic.com>
1652 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
1653 (extendqisi2, extendhisi2): Define.
1654 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
1655 (WCHAR_TYPE): Change to unsigned int.
1657 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1659 PR tree-optimization/60733
1660 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
1661 insertion point for PHI candidates to be the end of the feeding
1662 block for the PHI argument.
1664 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
1666 PR rtl-optimization/60650
1667 * lra-constraints.c (process_alt_operands): Decrease reject for
1668 earlyclobber matching.
1670 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1672 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
1674 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1676 * config/spu/spu.c (pad_bb): Do not crash when the last
1677 insn is CODE_FOR_blockage.
1679 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1681 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
1682 lies outside the target mode.
1684 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
1687 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1688 software floating point or no floating point registers, do not
1689 allow any type in the FPRs. Eliminate a test for SPE SIMD types
1690 in GPRs that occurs after we tested for GPRs that would never be
1693 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1694 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1695 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
1696 specifically allow DDmode, since that does not use the SPE SIMD
1699 2014-04-02 Richard Biener <rguenther@suse.de>
1702 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
1703 MODE_INTs. Properly use negv_optab.
1704 (expand_abs): Likewise.
1706 2014-04-02 Richard Biener <rguenther@suse.de>
1709 * Makefile.in (install-driver): Guard extra installs with special
1712 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1714 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1717 2014-04-01 Richard Henderson <rth@redhat.com>
1720 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
1721 alternative enabled before register allocation.
1723 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
1725 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
1726 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
1728 (nios2_large_got_address): Remove unneeded 'sym' parameter.
1729 (nios2_got_address): Update nios2_large_got_address call site.
1730 (nios2_delegitimize_address): New function.
1731 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
1732 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
1733 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
1735 2014-04-01 Martin Husemann <martin@duskware.de>
1737 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
1740 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
1742 PR rtl-optimization/60604
1743 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
1744 check from register_operand.
1745 (register_operand): Redefine in terms of general_operand.
1746 (nonmemory_operand): Use register_operand for the non-constant cases.
1748 2014-04-01 Richard Biener <rguenther@suse.de>
1750 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
1752 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
1754 * doc/invoke.texi (mapp-regs): Clarify.
1756 2014-03-31 Ulrich Drepper <drepper@gmail.com>
1758 * config/i386/avx512fintrin.h (__v32hi): Define type.
1759 (__v64qi): Likewise.
1760 (_mm512_set1_epi8): Define.
1761 (_mm512_set1_epi16): Define.
1762 (_mm512_set4_epi32): Define.
1763 (_mm512_set4_epi64): Define.
1764 (_mm512_set4_pd): Define.
1765 (_mm512_set4_ps): Define.
1766 (_mm512_setr4_epi64): Define.
1767 (_mm512_setr4_epi32): Define.
1768 (_mm512_setr4_pd): Define.
1769 (_mm512_setr4_ps): Define.
1770 (_mm512_setzero_epi32): Define.
1772 2014-03-31 Martin Jambor <mjambor@suse.cz>
1775 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
1776 callsite_arguments_match_p. Updated all callers. Also check types of
1777 corresponding formal parameters and actual arguments.
1778 (not_all_callers_have_enough_arguments_p) Renamed to
1779 some_callers_have_mismatched_arguments_p.
1781 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
1783 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
1785 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
1788 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
1791 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
1793 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
1794 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
1796 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
1797 Use FMAMODE_NOVF512 mode iterator.
1798 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
1799 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
1800 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
1801 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
1803 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
1804 Use VF_128_256 mode iterator.
1805 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
1808 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1810 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
1811 static chain if needed.
1813 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1816 * lra-constraints.c (index_part_to_reg): New.
1817 (process_address): Use it.
1819 2014-03-27 Jeff Law <law@redhat.com>
1820 Jakub Jelinek <jakub@redhat.com>
1823 * expr.c (do_tablejump): Use simplify_gen_binary rather than
1824 gen_rtx_{PLUS,MULT} to build up the address expression.
1826 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
1827 creating non-canonical RTL.
1829 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1832 * ipa-inline.c (want_inline_small_function_p): Short circuit large
1833 functions; reorganize to make cheap checks first.
1834 (inline_small_functions): Do not estimate growth when dumping;
1836 * ipa-inline.h (inline_summary): Add min_size.
1837 (growth_likely_positive): New function.
1838 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
1839 (set_cond_stmt_execution_predicate): Cleanup.
1840 (estimate_edge_size_and_time): Compute min_size.
1841 (estimate_calls_size_and_time): Likewise.
1842 (estimate_node_size_and_time): Likewise.
1843 (inline_update_overall_summary): Update min_size.
1844 (do_estimate_edge_time): Likewise.
1845 (do_estimate_edge_size): Update.
1846 (do_estimate_edge_hints): Update.
1847 (growth_likely_positive): New function.
1849 2014-03-28 Jakub Jelinek <jakub@redhat.com>
1852 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
1853 also if addr has VOIDmode.
1855 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1857 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
1858 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
1860 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
1861 instructions as well as AdvancedSIMD loads.
1863 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1865 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
1866 Use crypto_aese type.
1867 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
1868 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
1869 crypto_aese, crypto_aesmc. Move to types.md.
1870 * config/arm/types.md (crypto_aes): Split into crypto_aese,
1872 * config/arm/iterators.md (crypto_type): Likewise.
1874 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1876 * cgraph.c: Include expr.h and tree-dfa.h.
1877 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
1880 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1883 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
1884 regs from checking multi-reg pseudos.
1886 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1888 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
1890 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1892 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
1893 if it would clobber the stack pointer, even temporarily.
1895 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
1897 * mode-switching.c: Make small adjustments to the top comment.
1899 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1901 * config/rs6000/constraints.md (wD constraint): New constraint to
1902 match the constant integer to get the top DImode/DFmode out of a
1903 vector in a VSX register.
1905 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
1906 match the constant integer to get the top DImode/DFmode out of a
1907 vector in a VSX register.
1909 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
1912 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1915 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
1916 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
1918 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
1919 Optimize vec_extract of 64-bit values, where the value being
1920 extracted is in the top word, where we can use scalar
1921 instructions. Add direct move and store support. Combine the big
1922 endian/little endian vector select load support into a single insn.
1923 (vsx_extract_<mode>_internal1): Likewise.
1924 (vsx_extract_<mode>_internal2): Likewise.
1925 (vsx_extract_<mode>_load): Likewise.
1926 (vsx_extract_<mode>_store): Likewise.
1927 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
1928 combined into vsx_extract_<mode>_load.
1929 (vsx_extract_<mode>_one_le): Likewise.
1931 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
1932 define the top 64-bit vector element.
1934 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
1937 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1938 Document vec_vbpermq builtin.
1941 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
1942 enable use of xxsldwi and xxpermdi builtin functions.
1943 (vec_xxpermdi): Likewise.
1945 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1946 Document use of vec_xxsldwi and vec_xxpermdi builtins.
1948 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
1950 PR rtl-optimization/60650
1951 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
1953 (find_spills_for): New.
1954 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
1955 Spill all pseudos on the second iteration.
1957 2014-03-27 Marek Polacek <polacek@redhat.com>
1960 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
1963 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1965 * config/s390/s390.c (s390_can_use_return_insn): Check for
1966 call-saved FPRs on 31 bit.
1968 2014-03-27 Jakub Jelinek <jakub@redhat.com>
1971 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
1972 if they need regimplification, just drop them instead of
1973 calling gimple_regimplify_operands on them.
1975 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
1978 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
1979 (aarch64_frame_pointer_required): Adjust logic.
1980 (aarch64_can_eliminate): Adjust logic.
1981 (aarch64_override_options_after_change): Adjust logic.
1983 2014-03-27 Dehao Chen <dehao@google.com>
1985 * ipa-inline.c (early_inliner): Update node's inline info.
1987 2014-03-26 Dehao Chen <dehao@google.com>
1989 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
1990 compiler inserted conditional jumps for NAN float check.
1992 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1994 * ubsan.h (ubsan_create_data): Change second argument's type
1995 to const location_t *.
1996 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
1998 (ubsan_create_data): Change second argument to const location_t *PLOC.
1999 Create Loc field whenever PLOC is non-NULL.
2000 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
2001 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
2005 * real.c (real_to_integer2): Change type of low to UHWI.
2007 2014-03-26 Tobias Burnus <burnus@net-b.de>
2009 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
2010 (CILK_SELF_SPECS): New define.
2011 (driver_self_specs): Use it.
2013 2014-03-26 Richard Biener <rguenther@suse.de>
2015 * tree-pretty-print.c (percent_K_format): Implement special
2016 case for LTO and its stripped down BLOCK tree.
2018 2014-03-26 Jakub Jelinek <jakub@redhat.com>
2021 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
2023 * tree-vrp.c (simplify_internal_call_using_ranges): If only
2024 one range is range_int_cst_p, but not both, at least optimize
2025 addition/subtraction of 0 and multiplication by 0 or 1.
2026 * gimple-fold.c (gimple_fold_call): Fold
2027 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
2028 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
2029 INTEGER_CSTs, try to fold at least x * 0 and y - y.
2031 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
2033 PR rtl-optimization/60452
2034 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
2035 <case REG>: Return 1 for invalid offsets from the frame pointer.
2037 2014-03-26 Marek Polacek <polacek@redhat.com>
2040 * doc/extend.texi (C Extensions): Mention variable-length arrays in
2043 2014-03-26 Marek Polacek <polacek@redhat.com>
2046 * doc/extend.texi (Designated Inits): Describe what happens to omitted
2049 2014-03-26 Marek Polacek <polacek@redhat.com>
2052 * ira-color.c (update_conflict_hard_regno_costs): Perform the
2053 multiplication in unsigned type.
2055 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
2057 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
2059 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
2061 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
2063 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
2066 * cif-code.def (UNREACHABLE) New code.
2067 * ipa-inline.c (inline_small_functions): Skip edges to
2068 __builtlin_unreachable.
2069 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
2070 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
2071 predicate to __bulitin_unreachable.
2072 (set_cond_stmt_execution_predicate): Fix issue when
2073 invert_tree_comparison returns ERROR_MARK.
2074 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
2075 propagate to inline clones.
2076 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
2078 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
2079 * cgraphclones.c (cgraph_clone_node): If call destination is already
2080 ureachable, do not redirect it back.
2081 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
2084 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
2086 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
2087 Do not modify inline clones.
2089 2014-03-25 Jakub Jelinek <jakub@redhat.com>
2091 * config/i386/i386.md (general_sext_operand): New mode attr.
2092 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
2093 don't generate (sign_extend (const_int)).
2094 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
2095 operands[2]. Use We constraint instead of <i> and
2096 <general_sext_operand> predicate instead of <general_operand>.
2097 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
2098 * config/i386/constraints.md (We): New constraint.
2099 * config/i386/predicates.md (x86_64_sext_operand,
2100 sext_operand): New predicates.
2102 2014-03-25 Martin Jambor <mjambor@suse.cz>
2105 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
2106 inconsistent devirtualizations to __builtin_unreachable.
2108 2014-03-25 Marek Polacek <polacek@redhat.com>
2111 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
2113 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
2115 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
2116 order of elements for big-endian.
2118 2014-03-25 Richard Biener <rguenther@suse.de>
2121 * gimplify-me.c (gimple_regimplify_operands): Update the
2124 2014-03-25 Martin Jambor <mjambor@suse.cz>
2127 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
2128 (lto_output_varpool_node): Likewise.
2129 (input_overwrite_node): Likewise.
2130 (input_varpool_node): Likewise.
2132 2014-03-25 Richard Biener <rguenther@suse.de>
2134 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
2135 (run_gcc): Likewise.
2137 2014-03-25 Jakub Jelinek <jakub@redhat.com>
2139 * combine.c (simplify_compare_const): Add MODE argument.
2140 Handle mode_width 0 as very large mode_width.
2141 (try_combine, simplify_comparison): Adjust callers.
2143 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
2144 type to avoid signed integer overflow.
2145 * explow.c (plus_constant): Likewise.
2147 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
2149 * doc/generic.texi: Correct typos.
2151 2014-03-24 Tobias Burnus <burnus@net-b.de>
2153 * doc/invoke.texi (-flto): Expand section about
2154 using static libraries with LTO.
2156 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2158 PR rtl-optimization/60501
2159 * optabs.def (addptr3_optab): New optab.
2160 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
2161 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
2162 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
2164 * lra.c (emit_add3_insn): Use the addptr pattern if available.
2166 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
2168 2014-03-24 Ulrich Drepper <drepper@gmail.com>
2170 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
2173 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
2174 (_mm256_undefined_ps): Define.
2175 (_mm256_undefined_pd): Define.
2176 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
2177 (_mm_undefined_pd): Define.
2178 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
2179 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
2180 (_mm512_undefined_ps): Define.
2181 (_mm512_undefined_pd): Define.
2182 Use _mm*_undefined_*.
2183 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
2185 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
2187 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
2188 (lshr_simd): DI mode added.
2189 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
2190 (aarch64_ushr_simddi): Likewise.
2191 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
2192 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
2193 (vshrd_n_u64): Likewise.
2195 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2197 * Makefile.in (s-macro_list): Depend on cc1.
2199 2014-03-23 Teresa Johnson <tejohnson@google.com>
2201 * ipa-utils.c (ipa_print_order): Use specified dump file.
2203 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
2205 PR rtl-optimization/60601
2206 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
2208 * gcc.c (eval_spec_function): Initialize save_growing_value.
2210 2014-03-22 Jakub Jelinek <jakub@redhat.com>
2213 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
2214 code == MINUS_EXPR, never swap op0 with op1.
2216 * toplev.c (init_local_tick): Avoid signed integer multiplication
2218 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
2219 shift by first operand's bitsize.
2221 2014-03-21 Jakub Jelinek <jakub@redhat.com>
2224 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
2226 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
2227 TARGET_ISA_64BIT_P(x).
2229 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2231 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
2232 pattern for vector nor instead of subtract from splat(-1).
2233 (altivec_expand_vec_perm_const_le): Likewise.
2235 2014-03-21 Richard Henderson <rth@twiddle.net>
2238 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
2239 related insns after epilogue_completed.
2241 2014-03-21 Martin Jambor <mjambor@suse.cz>
2244 * cgraph.h (symtab_node): New flag body_removed.
2245 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
2246 when removing bodies.
2247 * symtab.c (dump_symtab_base): Dump body_removed flag.
2248 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
2249 had their bodies removed.
2251 2014-03-21 Martin Jambor <mjambor@suse.cz>
2254 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
2257 2014-03-21 Richard Biener <rguenther@suse.de>
2259 PR tree-optimization/60577
2260 * tree-core.h (struct tree_base): Document nothrow_flag use
2261 in VAR_DECL_NONALIASED.
2262 * tree.h (VAR_DECL_NONALIASED): New.
2263 (may_be_aliased): Adjust.
2264 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
2266 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2268 * expr.c (expand_expr_real_1): Remove outdated comment.
2270 2014-03-20 Jakub Jelinek <jakub@redhat.com>
2273 * ira.c (adjust_cleared_regs): Call copy_rtx on
2274 *reg_equiv[REGNO (loc)].src_p before passing it to
2275 simplify_replace_fn_rtx.
2278 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
2279 into CONST, put pic register as first operand of PLUS. Use
2280 gen_const_mem for both 32-bit and 64-bit PIC got loads.
2282 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2284 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
2286 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2288 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
2289 around for store forwarding issue in the FPU on the UT699.
2290 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
2291 loads and operations if -mfix-ut699 is specified.
2292 (divtf3_hq): Tweak attribute.
2293 (sqrttf2_hq): Likewise.
2295 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2297 * calls.c (store_one_arg): Remove incorrect const qualification on the
2298 type of the temporary.
2299 * cfgexpand.c (expand_return): Likewise.
2300 * expr.c (expand_constructor): Likewise.
2301 (expand_expr_real_1): Likewise.
2303 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
2305 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
2308 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
2311 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
2313 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
2315 * config/arm/aarch-common-protos.h
2316 (alu_cost_table): Fix spelling of "extend".
2317 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
2319 2014-03-19 Richard Biener <rguenther@suse.de>
2322 * tree-core.h (tree_type_common): Re-order pointer members
2323 to reduce recursion depth during GC walks.
2325 2014-03-19 Marek Polacek <polacek@redhat.com>
2328 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
2329 before accessing it.
2331 2014-03-19 Richard Biener <rguenther@suse.de>
2334 * lto-streamer-in.c (input_function): In WPA stage do not drop
2337 2014-03-19 Jakub Jelinek <jakub@redhat.com>
2339 PR tree-optimization/60559
2340 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
2341 with build_zero_cst assignment.
2343 2014-03-18 Kai Tietz <ktietz@redhat.com>
2345 PR rtl-optimization/56356
2346 * sdbout.c (sdbout_parms): Verify that parms'
2347 incoming argument is valid.
2348 (sdbout_reg_parms): Likewise.
2350 2014-03-18 Richard Henderson <rth@redhat.com>
2353 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
2354 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
2355 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
2357 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
2359 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
2360 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
2361 Italicize plugin event names in description. Explain that
2362 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
2363 Remind that no GCC functions should be called after PLUGIN_FINISH.
2364 Explain what pragmas with expansion are.
2366 2014-03-18 Martin Liska <mliska@suse.cz>
2368 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
2369 gimple call statement is update.
2370 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
2371 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
2373 2014-03-18 Jakub Jelinek <jakub@redhat.com>
2376 * ubsan.c (ubsan_instrument_unreachable): Call
2377 initialize_sanitizer_builtins.
2378 (ubsan_pass): Likewise.
2381 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
2382 varpool_finalize_decl instead of rest_of_decl_compilation.
2384 2014-03-18 Richard Biener <rguenther@suse.de>
2386 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
2387 by using bitmap_and_compl instead of bitmap_and_compl_into.
2388 (df_rd_transfer_function): Likewise.
2390 2014-03-18 Richard Biener <rguenther@suse.de>
2392 * doc/lto.texi (fresolution): Fix typo.
2394 2014-03-18 Richard Biener <rguenther@suse.de>
2396 * doc/invoke.texi (flto): Update for changes in 4.9.
2398 2014-03-18 Richard Biener <rguenther@suse.de>
2400 * doc/loop.texi: Remove section on the removed lambda framework.
2401 Update loop docs with recent changes in preserving loop structure.
2403 2014-03-18 Richard Biener <rguenther@suse.de>
2405 * doc/lto.texi (-fresolution): Document.
2407 2014-03-18 Richard Biener <rguenther@suse.de>
2409 * doc/contrib.texi: Adjust my name.
2411 2014-03-18 Jakub Jelinek <jakub@redhat.com>
2414 * internal-fn.c: Include diagnostic-core.h.
2415 (expand_BUILTIN_EXPECT): New function.
2416 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
2417 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
2418 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
2419 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
2421 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
2422 Revert 3 argument __builtin_expect code.
2423 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
2424 * gimple-fold.c (gimple_fold_call): Likewise.
2425 * tree.h (fold_builtin_expect): New prototype.
2426 * builtins.c (build_builtin_expect_predicate): Add predictor
2427 argument, if non-NULL, create 3 argument __builtin_expect.
2428 (fold_builtin_expect): No longer static. Add ARG2 argument,
2429 pass it through to build_builtin_expect_predicate.
2430 (fold_builtin_2): Adjust caller.
2431 (fold_builtin_3): Handle BUILT_IN_EXPECT.
2432 * internal-fn.def (BUILTIN_EXPECT): New.
2434 2014-03-18 Tobias Burnus <burnus@net-b.de>
2437 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
2438 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
2439 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
2441 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
2444 * predict.c (combine_predictions_for_bb): Fix up formatting.
2445 (expr_expected_value_1, expr_expected_value): Add predictor argument,
2446 fill what it points to if non-NULL.
2447 (tree_predict_by_opcode): Adjust caller, use the predictor.
2448 * predict.def (PRED_COMPARE_AND_SWAP): Add.
2450 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
2452 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
2453 proper constant for the store mode.
2455 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
2457 * symtab.c (change_decl_assembler_name): Fix transparent alias
2460 2014-03-16 Renlin Li <Renlin.Li@arm.com>
2462 * config/aarch64/aarch64.c: Correct the comments about the
2463 aarch64 stack layout.
2465 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
2467 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
2468 check for GF_OMP_FOR_KIND_FOR.
2470 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
2472 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
2473 ymm and zmm register names.
2475 2014-03-17 Jakub Jelinek <jakub@redhat.com>
2478 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
2479 note creation for the 2010-08-31 changes.
2481 2014-03-17 Marek Polacek <polacek@redhat.com>
2484 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
2485 as -fno-tree-loop-vectorize.
2486 (expand_omp_simd): Likewise.
2488 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
2490 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
2491 (eligible_for_call_delay): New prototype.
2492 * config/sparc/sparc.c (tls_call_delay): Rename into...
2493 (eligible_for_call_delay): ...this. Return false if the instruction
2494 cannot be put in the delay slot of a branch.
2495 (eligible_for_restore_insn): Simplify.
2496 (eligible_for_return_delay): Return false if the instruction cannot be
2497 put in the delay slot of a branch and simplify.
2498 (eligible_for_sibcall_delay): Return false if the instruction cannot be
2499 put in the delay slot of a branch.
2500 * config/sparc/sparc.md (fix_ut699): New attribute.
2501 (tls_call_delay): Delete.
2502 (in_call_delay): Reimplement.
2503 (eligible_for_sibcall_delay): Rename into...
2504 (in_sibcall_delay): ...this.
2505 (eligible_for_return_delay): Rename into...
2506 (in_return_delay): ...this.
2507 (in_branch_delay): Reimplement.
2508 (in_uncond_branch_delay): Delete.
2509 (in_annul_branch_delay): Delete.
2511 2014-03-14 Richard Henderson <rth@redhat.com>
2514 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
2515 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
2516 (*floathi<X87MODEF>2_i387_with_temp): Remove.
2517 (floathi splitters): Remove.
2518 (float<SWI48x>xf2): New pattern.
2519 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
2520 code that tried to handle DImode for 32-bit, but which was excluded
2521 by the pattern's condition. Drop allocation of stack temporary.
2522 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
2523 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
2524 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
2525 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
2526 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
2527 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
2528 (*float<SWI48><MODEF>2_sse_interunit): Remove.
2529 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
2530 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
2531 (*float<SWI48x><X87MODEF>2_i387): Remove.
2532 (all float _with_temp splitters): Remove.
2533 (*float<SWI48x><MODEF>2_i387): New pattern.
2534 (*float<SWI48><MODEF>2_sse): New pattern.
2535 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
2536 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
2538 2014-03-14 Jakub Jelinek <jakub@redhat.com>
2539 Marek Polacek <polacek@redhat.com>
2542 * common.opt (dump_base_name_prefixed): New Variable.
2543 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
2544 if x_dump_base_name_prefixed is already set, set it at the end.
2546 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
2548 PR rtl-optimization/60508
2549 * lra-constraints.c (get_reload_reg): Add new parameter
2551 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
2552 Pass the new parameter values.
2554 2014-03-14 Richard Biener <rguenther@suse.de>
2556 * common.opt: Revert unintented changes from r205065.
2559 2014-03-14 Richard Biener <rguenther@suse.de>
2562 * cfghooks.c (split_block): Properly adjust all loops the
2563 block was a latch of.
2565 2014-03-14 Martin Jambor <mjambor@suse.cz>
2568 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
2571 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
2574 * config/avr/avr.c (avr_set_current_function): Pass function name
2575 through default_strip_name_encoding before sanity checking instead
2576 of skipping the first char of the assembler name.
2578 2014-03-13 Richard Henderson <rth@redhat.com>
2581 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
2582 (ix86_force_to_memory, ix86_free_from_memory): Remove.
2583 * config/i386/i386-protos.h: Likewise.
2584 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
2585 in the expander instead of a splitter.
2586 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
2587 any possibility of requiring a memory.
2588 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
2589 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
2590 (fp branch splitters): Update for ix86_split_fp_branch.
2591 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
2592 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
2593 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
2594 (*fop_<MODEF>_2_i387): Remove f/r alternative.
2595 (*fop_<MODEF>_3_i387): Likewise.
2596 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
2597 (splitters for the fop_* register patterns): Remove.
2598 (fscalexf4_i387): Rename from *fscalexf4_i387.
2599 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
2601 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2603 PR tree-optimization/59779
2604 * tree-dfa.c (get_ref_base_and_extent): Use double_int
2605 type for bitsize and maxsize instead of HOST_WIDE_INT.
2607 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
2609 PR rtl-optimization/57320
2610 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
2611 the CFG after thread_prologue_and_epilogue_insns.
2613 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
2615 PR rtl-optimization/57189
2616 * lra-constraints.c (process_alt_operands): Disfavor spilling
2619 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
2621 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
2623 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2625 PR tree-optimization/59025
2627 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
2628 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
2630 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
2633 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
2634 calls of avr_out_plus_1.
2636 2014-03-13 Bin Cheng <bin.cheng@arm.com>
2638 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
2639 BB's single pred and update the father loop's latch info later.
2641 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2643 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
2647 (VEC_base): Likewise.
2648 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
2649 registers, we need to swap double words in little endian mode.
2651 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
2652 to be a container mode for 128-bit integer operations added in ISA
2653 2.07. Unlike TImode and PTImode, the preferred register set is
2654 the Altivec/VMX registers for the 128-bit operations.
2656 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
2658 (rs6000_split_128bit_ok_p): Likewise.
2660 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
2661 macros for creating ISA 2.07 normal and overloaded builtin
2662 functions with 3 arguments.
2663 (BU_P8V_OVERLOAD_3): Likewise.
2664 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
2665 for use as overloaded functions.
2666 (VPERM_1TI_UNS): Likewise.
2667 (VSEL_1TI): Likewise.
2668 (VSEL_1TI_UNS): Likewise.
2669 (ST_INTERNAL_1ti): Likewise.
2670 (LD_INTERNAL_1ti): Likewise.
2671 (XXSEL_1TI): Likewise.
2672 (XXSEL_1TI_UNS): Likewise.
2673 (VPERM_1TI): Likewise.
2674 (VPERM_1TI_UNS): Likewise.
2675 (XXPERMDI_1TI): Likewise.
2676 (SET_1TI): Likewise.
2677 (LXVD2X_V1TI): Likewise.
2678 (STXVD2X_V1TI): Likewise.
2679 (VEC_INIT_V1TI): Likewise.
2680 (VEC_SET_V1TI): Likewise.
2681 (VEC_EXT_V1TI): Likewise.
2682 (EQV_V1TI): Likewise.
2683 (NAND_V1TI): Likewise.
2684 (ORC_V1TI): Likewise.
2685 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
2686 added in ISA 2.07. Add both normal 'altivec' builtins, and the
2688 (VADDUQM): Likewise.
2689 (VSUBCUQ): Likewise.
2690 (VADDEUQM): Likewise.
2691 (VADDECUQ): Likewise.
2692 (VSUBEUQM): Likewise.
2693 (VSUBECUQ): Likewise.
2695 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
2696 __int128_t and __uint128_t types.
2697 (__uint128_type): Likewise.
2698 (altivec_categorize_keyword): Add support for vector __int128_t,
2699 vector __uint128_t, vector __int128, and vector unsigned __int128
2700 as a container type for TImode operations that need to be done in
2701 VSX/Altivec registers.
2702 (rs6000_macro_to_expand): Likewise.
2703 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
2704 to support 128-bit integer instructions vaddcuq, vadduqm,
2705 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
2706 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
2708 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
2709 for V1TImode, and set up preferences to use VSX/Altivec registers.
2710 Setup VSX reload handlers.
2711 (rs6000_debug_reg_global): Likewise.
2712 (rs6000_init_hard_regno_mode_ok): Likewise.
2713 (rs6000_preferred_simd_mode): Likewise.
2714 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
2715 (easy_altivec_constant): Likewise.
2716 (output_vec_const_move): Likewise.
2717 (rs6000_expand_vector_set): Convert V1TImode set and extract to
2719 (rs6000_expand_vector_extract): Likewise.
2720 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
2722 (rs6000_const_vec): Add support for V1TImode.
2723 (rs6000_emit_le_vsx_load): Swap double words when loading or
2724 storing TImode/V1TImode.
2725 (rs6000_emit_le_vsx_store): Likewise.
2726 (rs6000_emit_le_vsx_move): Likewise.
2727 (rs6000_emit_move): Add support for V1TImode.
2728 (altivec_expand_ld_builtin): Likewise.
2729 (altivec_expand_st_builtin): Likewise.
2730 (altivec_expand_vec_init_builtin): Likewise.
2731 (altivec_expand_builtin): Likewise.
2732 (rs6000_init_builtins): Add support for V1TImode type. Add
2733 support for ISA 2.07 128-bit integer builtins. Define type names
2734 for the VSX/Altivec vector types.
2735 (altivec_init_builtins): Add support for overloaded vector
2736 functions with V1TImode type.
2737 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
2738 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
2740 (rs6000_split_128bit_ok_p): Likewise.
2741 (rs6000_handle_altivec_attribute): Create V1TImode from vector
2742 __int128_t and vector __uint128_t.
2744 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
2745 and mode attributes.
2752 (VS_scalar): Likewise.
2753 (VS_double): Likewise.
2754 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
2756 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
2757 we support the ISA 2.07 128-bit integer arithmetic instructions.
2758 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
2759 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
2760 and TImode types for use with the builtin functions.
2761 (V1TI_type_node): Likewise.
2762 (unsigned_V1TI_type_node): Likewise.
2763 (intTI_type_internal_node): Likewise.
2764 (uintTI_type_internal_node): Likewise.
2766 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
2767 128-bit builtin functions.
2768 (UNSPEC_VADDEUQM): Likewise.
2769 (UNSPEC_VADDECUQ): Likewise.
2770 (UNSPEC_VSUBCUQ): Likewise.
2771 (UNSPEC_VSUBEUQM): Likewise.
2772 (UNSPEC_VSUBECUQ): Likewise.
2773 (VM): Add V1TImode to vector mode iterators.
2775 (VI_unit): Likewise.
2776 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
2777 (altivec_vaddcuq): Likewise.
2778 (altivec_vsubuqm): Likewise.
2779 (altivec_vsubcuq): Likewise.
2780 (altivec_vaddeuqm): Likewise.
2781 (altivec_vaddecuq): Likewise.
2782 (altivec_vsubeuqm): Likewise.
2783 (altivec_vsubecuq): Likewise.
2785 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
2787 (BOOL_128): Likewise.
2788 (BOOL_REGS_OUTPUT): Likewise.
2789 (BOOL_REGS_OP1): Likewise.
2790 (BOOL_REGS_OP2): Likewise.
2791 (BOOL_REGS_UNARY): Likewise.
2792 (BOOL_REGS_AND_CR0): Likewise.
2794 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
2795 128-bit integer builtin support.
2796 (vec_vadduqm): Likewise.
2797 (vec_vaddecuq): Likewise.
2798 (vec_vaddeuqm): Likewise.
2799 (vec_vsubecuq): Likewise.
2800 (vec_vsubeuqm): Likewise.
2801 (vec_vsubcuq): Likewise.
2802 (vec_vsubuqm): Likewise.
2804 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2805 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
2806 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
2807 128-bit integer add/subtract to ISA 2.07.
2809 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
2811 * config/arc/arc.c (arc_predicate_delay_insns):
2812 Fix third argument passed to conditionalize_nonjump.
2814 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
2816 * config/aarch64/aarch64-builtins.c
2817 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
2818 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
2819 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
2820 instead of __builtin_lfloor.
2821 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
2823 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2825 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
2826 (tree_ssa_ifcombine_bb_1): New function.
2827 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
2828 is an empty forwarder block to then_bb or vice versa and then_bb
2829 and else_bb are effectively swapped.
2831 2014-03-12 Christian Bruel <christian.bruel@st.com>
2834 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
2835 REG_CFA_DEF_CFA note.
2836 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
2837 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
2839 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
2841 PR tree-optimization/60454
2842 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
2844 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2846 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
2847 Do not define target_cpu_default2 to generic.
2848 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
2849 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
2850 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
2852 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2853 Marc Glisse <marc.glisse@inria.fr>
2855 PR tree-optimization/60502
2856 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
2857 instead of build_low_bits_mask.
2859 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2862 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
2863 if there are multiple uses, but op doesn't live on E edge.
2864 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
2865 clobber stmts before __builtin_unreachable.
2867 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
2869 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
2870 hard_frame_pointer_rtx.
2871 * cse.c (cse_insn): Remove volatile check.
2872 * cselib.c (cselib_process_insn): Likewise.
2873 * dse.c (scan_insn): Likewise.
2875 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2877 * config/arc/arc.c (conditionalize_nonjump): New function,
2879 (arc_ifcvt): ... this.
2880 (arc_predicate_delay_insns): Use it.
2882 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2884 * config/arc/predicates.md (extend_operand): During/after reload,
2885 allow const_int_operand.
2886 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
2887 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
2888 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
2890 (umulsi3_highpart_i): Likewise.
2892 2014-03-11 Richard Biener <rguenther@suse.de>
2894 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
2895 Add asserts to guard possible wrong-code bugs.
2897 2014-03-11 Richard Biener <rguenther@suse.de>
2899 PR tree-optimization/60429
2900 PR tree-optimization/60485
2901 * tree-ssa-structalias.c (set_union_with_increment): Properly
2902 take into account all fields that overlap the shifted vars.
2903 (do_sd_constraint): Likewise.
2904 (do_ds_constraint): Likewise.
2905 (get_constraint_for_ptr_offset): Likewise.
2907 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
2909 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
2910 (nios2_compute_frame_layout):
2911 Add calculation of cfun->machine->fp_save_offset.
2912 (nios2_expand_prologue): Correct setting of frame pointer register
2914 (nios2_expand_epilogue): Update recovery of stack pointer from
2915 frame pointer accordingly.
2916 (nios2_initial_elimination_offset): Update calculation of offset
2917 for eliminating to HARD_FRAME_POINTER_REGNUM.
2919 2014-03-10 Jakub Jelinek <jakub@redhat.com>
2922 * ipa.c (symtab_remove_unreachable_nodes): Don't call
2923 cgraph_get_create_node on VAR_DECLs.
2925 2014-03-10 Richard Biener <rguenther@suse.de>
2928 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
2930 2014-03-08 Douglas B Rupp <rupp@gnat.com>
2932 * config/vms/vms.opt (vms_float_format): New variable.
2934 2014-03-08 Tobias Burnus <burnus@net-b.de>
2936 * doc/invoke.texi (-fcilkplus): Update implementation status.
2938 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
2939 Richard Biener <rguenther@suse.de>
2941 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
2942 consistently accross all TUs.
2943 (run_gcc): Enable -fshort-double automatically at link at link-time
2944 and disallow override.
2946 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
2949 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
2950 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
2951 if they can't be used.
2953 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2955 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
2956 for Solaris 11/x86 ld.
2957 * configure: Regenerate.
2959 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2961 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
2962 (LIB_TLS_SPEC): Save as ld_tls_libs.
2963 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
2964 (HAVE_AS_IX86_TLSLDM): New test.
2965 * configure, config.in: Regenerate.
2966 * config/i386/i386.c (legitimize_tls_address): Fall back to
2967 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
2968 cannot support TLS_MODEL_LOCAL_DYNAMIC.
2969 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
2970 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
2972 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
2974 * common.opt (fira-loop-pressure): Mark as optimization.
2976 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
2978 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
2979 an OpenMP mappable type.
2981 2014-03-06 Matthias Klose <doko@ubuntu.com>
2983 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
2984 MULTILIB_OSDIRNAMES is not defined.
2986 2014-03-06 Jakub Jelinek <jakub@redhat.com>
2987 Meador Inge <meadori@codesourcery.com>
2990 * config/arm/arm.c (arm_tls_symbol_p): Remove.
2991 (arm_legitimize_address): Call legitimize_tls_address for any
2992 arm_tls_referenced_p expression, handle constant addend. Call it
2993 before testing for !TARGET_ARM.
2994 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
2996 2014-03-06 Richard Biener <rguenther@suse.de>
3002 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
3004 * tree-streamer.c (record_common_node): Assert we don't record
3005 nodes with type double.
3006 (preload_common_node): Skip type double, complex double and double
3007 pointer since it is now frontend dependent due to fshort-double option.
3009 2014-03-06 Richard Biener <rguenther@suse.de>
3011 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
3012 or -fno-lto is specified and the linker has full plugin support.
3013 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
3014 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
3015 * lto-wrapper.c (merge_and_complain): Merge compile-time
3016 optimization levels.
3017 (run_gcc): And pass it through to the link options.
3019 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
3023 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3025 * cselib.c (remove_useless_values): Skip to avoid quadratic
3026 behavior if the condition moved from...
3027 (cselib_process_insn): ... here holds.
3029 2014-03-05 Jakub Jelinek <jakub@redhat.com>
3032 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
3033 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
3036 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
3037 (TM_H): Add x86-tune.def.
3039 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3041 * config/aarch64/aarch64.c (generic_tunings):
3042 Use cortexa57_extra_costs.
3044 2014-03-05 Jakub Jelinek <jakub@redhat.com>
3047 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
3048 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
3049 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
3052 2014-03-04 Heiher <r@hev.cc>
3054 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
3055 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
3057 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
3059 * config/i386/predicates.md (const2356_operand): Change to ...
3060 (const2367_operand): ... this.
3061 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
3063 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3064 (*avx512pf_scatterpf<mode>sf): Ditto.
3065 (avx512pf_scatterpf<mode>df): Ditto.
3066 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3067 (*avx512pf_scatterpf<mode>df): Ditto.
3068 * config/i386/i386.c (ix86_expand_builtin): Update
3069 incorrect hint operand error message.
3071 2014-03-04 Richard Biener <rguenther@suse.de>
3073 * lto-section-in.c (lto_get_section_data): Fix const cast.
3075 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
3077 * tree-streamer.c (record_common_node): Assert we don't record
3078 nodes with type double.
3079 (preload_common_node): Skip type double, complex double and double
3080 pointer since it is now frontend dependent due to fshort-double option.
3082 2014-03-04 Richard Biener <rguenther@suse.de>
3085 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
3086 (lto_input_toplevel_asms): Likewise.
3087 * lto-section-in.c (lto_get_section_data): Instead do it here
3090 2014-03-04 Richard Biener <rguenther@suse.de>
3092 PR tree-optimization/60382
3093 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
3094 dead PHIs a reduction.
3096 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
3098 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
3100 (_mm_prefetch): Move out of GCC target("sse") pragma.
3101 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
3102 GCC target("prfchw") pragma.
3103 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
3105 * config/i386/i386.c (ix86_option_override_internal): Enable
3106 -mprfchw with -mprefetchwt1.
3108 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
3110 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
3113 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
3115 * opts.h (CL_PCH_IGNORE): Define.
3116 * targhooks.c (option_affects_pch_p):
3117 Return false for options that have CL_PCH_IGNORE set.
3118 * opt-functions.awk: Process PchIgnore.
3119 * doc/options.texi: Document PchIgnore.
3121 * config/arc/arc.opt (misize): Add PchIgnore property.
3123 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3125 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
3126 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
3127 constraint on constants to permit them being loaded into
3128 GENERAL_REGS or BASE_REGS.
3130 2014-03-03 Nick Clifton <nickc@redhat.com>
3132 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
3133 anti-cacnonical alternatives.
3134 (negandhi3_real): New pattern.
3135 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
3137 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3139 * config/avr/avr-mcus.def: Remove atxmega16x1.
3140 * config/avr/avr-tables.opt: Regenerate.
3141 * config/avr/t-multilib: Regenerate.
3142 * doc/avr-mmcu.texi: Regenerate.
3144 2014-03-03 Tobias Grosser <tobias@grosser.es>
3145 Mircea Namolaru <mircea.namolaru@inria.fr>
3147 PR tree-optimization/58028
3148 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
3151 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3153 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
3154 not handled by recognizers.
3156 2014-03-03 Jakub Jelinek <jakub@redhat.com>
3159 * function.c (expand_function_end): Don't emit
3160 clobber_return_register sequence if clobber_after is a BARRIER.
3161 * cfgexpand.c (construct_exit_block): Append instructions before
3162 return_label to prev_bb.
3164 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3166 * config/rs6000/constraints.md: Document reserved use of "wc".
3168 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
3171 * ipa.c (function_and_variable_visibility): When dissolving comdat
3172 group, also set all symbols to local.
3174 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
3179 2013-12-14 Jan Hubicka <jh@suse.cz>
3181 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
3183 2014-03-02 Jon Beniston <jon@beniston.com>
3189 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
3190 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
3191 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
3192 (simple_return, *simple_return): New patterns
3193 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
3194 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
3196 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
3198 * dwarf2out.c (gen_subprogram_die): Tidy.
3200 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
3203 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
3204 (*mov_t_msb_neg_negc): ... this new insn.
3206 2014-02-28 Jason Merrill <jason@redhat.com>
3209 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
3212 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
3215 * dwarf2out.c (decltype_auto_die): New static.
3216 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
3217 (gen_type_die_with_usage): Handle 'decltype(auto)'.
3218 (is_cxx_auto): Likewise.
3220 2014-02-28 Ian Bolton <ian.bolton@arm.com>
3222 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
3223 we are not using general regs only.
3225 2014-02-28 Richard Biener <rguenther@suse.de>
3228 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
3229 previous fix and only allow to remove trivial pre-headers
3230 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
3231 (remove_forwarder_block): Properly update the latch of a loop.
3233 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3236 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
3237 (cselib_preserved_hash_table): New.
3238 (preserve_constants_and_equivs): Move preserved vals to it.
3239 (cselib_find_slot): Look it up first.
3240 (cselib_init): Initialize it.
3241 (cselib_finish): Release it.
3242 (dump_cselib_table): Dump it.
3244 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3247 * cselib.c (remove_useless_values): Skip to avoid quadratic
3248 behavior if the condition moved from...
3249 (cselib_process_insn): ... here holds.
3251 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3254 * var-tracking.c (vt_initialize): Apply the same condition to
3255 preserve the CFA base value.
3257 2014-02-28 Joey Ye <joey.ye@arm.com>
3260 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
3261 if reload in progress or completed.
3263 2014-02-28 Tobias Burnus <burnus@net-b.de>
3266 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
3269 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
3271 * doc/tm.texi.in (Condition Code Status): Update documention for
3272 relative locations of cc0-setter and cc0-user.
3274 2014-02-27 Jeff Law <law@redhat.com>
3276 PR rtl-optimization/52714
3277 * combine.c (try_combine): When splitting an unrecognized PARALLEL
3278 into two independent simple sets, if I3 is a jump, ensure the
3279 pattern we place into I3 is a (set (pc) ...).
3281 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
3282 Jeff Law <law@redhat.com>
3284 PR rtl-optimization/49847
3285 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
3286 are in different blocks.
3287 * doc/tm.texi (Condition Code Status): Update documention for
3288 relative locations of cc0-setter and cc0-user.
3290 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
3293 * lra.c (lra_emit_add): Check SUBREG too.
3295 2014-02-27 Andreas Schwab <schwab@suse.de>
3297 * config/m68k/m68k.c (m68k_option_override): Disable
3298 -flive-range-shrinkage for classic m68k.
3299 (m68k_override_options_after_change): Likewise.
3301 2014-02-27 Marek Polacek <polacek@redhat.com>
3304 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
3305 -Wmaybe-uninitialized.
3307 2014-02-27 Alan Modra <amodra@gmail.com>
3310 * reload1.c (emit_input_reload_insns): When reload_override_in,
3311 set old to rl->in_reg when rl->in_reg is a subreg.
3313 2014-02-26 Richard Biener <rguenther@suse.de>
3316 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
3318 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
3320 * common/config/i386/predicates.md (const1256_operand): Remove.
3321 (const2356_operand): New.
3322 (const_1_to_2_operand): Remove.
3323 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3324 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3325 (*avx512pf_gatherpf<mode>sf): Ditto.
3326 (avx512pf_gatherpf<mode>df): Ditto.
3327 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3328 (*avx512pf_gatherpf<mode>df): Ditto.
3329 (avx512pf_scatterpf<mode>sf): Ditto.
3330 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3331 (*avx512pf_scatterpf<mode>sf): Ditto.
3332 (avx512pf_scatterpf<mode>df): Ditto.
3333 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3334 (*avx512pf_scatterpf<mode>df): Ditto.
3335 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
3337 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
3339 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
3340 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
3341 (_mm512_mask_testn_epi64_mask): Move to ...
3342 * config/i386/avx512cdintrin.h: Here.
3343 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
3344 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
3345 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
3346 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
3347 TARGET_AVX512F from TARGET_AVX512CD.
3349 2014-02-26 Richard Biener <rguenther@suse.de>
3352 * ipa.c (walk_polymorphic_call_targets): Properly guard
3353 call to inline_update_overall_summary.
3355 2014-02-26 Bin Cheng <bin.cheng@arm.com>
3358 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
3359 and latches only if requested. Fix latch if it is removed.
3360 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
3361 LOOPS_HAVE_PREHEADERS.
3363 2014-02-25 Andrew Pinski <apinski@cavium.com>
3365 * builtins.c (expand_builtin_thread_pointer): Create a new target
3366 when the target is NULL.
3368 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
3370 PR rtl-optimization/60317
3371 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3372 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3373 * lra-assigns.c: Include params.h.
3374 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
3375 other reload pseudos considerations.
3377 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3379 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
3380 to use canonical form for nor<mode>3.
3382 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3385 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
3388 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
3390 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
3391 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
3392 (ix86_handle_option): Handle OPT_mprefetchwt1.
3393 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
3394 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3396 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3397 OPTION_MASK_ISA_PREFETCHWT1.
3398 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
3399 (PTA_PREFETCHWT1): New.
3400 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
3401 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
3402 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
3403 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
3404 (*prefetch_avx512pf_<mode>_: Change into ...
3405 (*prefetch_prefetchwt1_<mode>: This.
3406 * config/i386/i386.opt (mprefetchwt1): New.
3407 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
3408 (_mm_prefetch): Handle intent to write.
3409 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
3411 2014-02-25 Richard Biener <rguenther@suse.de>
3414 * emit-rtl.c (mem_attrs_htab): Remove.
3415 (mem_attrs_htab_hash): Likewise.
3416 (mem_attrs_htab_eq): Likewise.
3417 (set_mem_attrs): Always allocate new mem-attrs when something changed.
3418 (init_emit_once): Do not allocate mem_attrs_htab.
3420 2014-02-25 Richard Biener <rguenther@suse.de>
3423 * lto-opts.c (lto_write_options): Output non-explicit conservative
3424 -fwrapv, -fno-trapv and -fno-strict-overflow.
3425 * lto-wrapper.c (merge_and_complain): Handle merging those options.
3426 (run_gcc): And pass them through.
3428 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3430 * sel-sched.c (calculate_new_fences): New parameter ptime.
3431 Calculate it as a maximum over all fence cycles.
3432 (sel_sched_region_2): Adjust the call to calculate_new_fences.
3433 Print the final schedule timing when sched_verbose.
3435 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3437 PR rtl-optimization/60292
3438 * sel-sched.c (fill_vec_av_set): Do not reset target availability
3439 bit fot the fence instruction.
3441 2014-02-24 Alangi Derick <alangiderick@gmail.com>
3443 * calls.h: Fix typo in comment.
3445 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
3447 * config/pa/pa.c (pa_output_move_double): Don't valididate when
3448 adjusting offsetable addresses.
3450 2014-02-24 Guozhi Wei <carrot@google.com>
3452 * sparseset.h (sparseset_pop): Fix the wrong index.
3454 2014-02-24 Walter Lee <walt@tilera.com>
3456 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
3457 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
3459 * common/config/tilegx/tilegx-common.c
3460 (TARGET_DEFAULT_TARGET_FLAGS): Define.
3461 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
3463 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
3464 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
3465 (tilegx_gimplify_va_arg_expr): Handle big endian.
3466 (tilegx_expand_unaligned_load): Ditto.
3467 (tilegx_expand_unaligned_store): Ditto.
3468 (TARGET_RETURN_IN_MSB): New.
3469 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
3470 (TARGET_ENDIAN_DEFAULT): New.
3471 (TARGET_BIG_ENDIAN): Handle big endian.
3472 (BYTES_BIG_ENDIAN): Ditto.
3473 (WORDS_BIG_ENDIAN): Ditto.
3474 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
3477 * config/tilegx/tilegx.md (extv): Handle big endian.
3479 (insn_st<n>): Ditto.
3480 (insn_st<n>_add<bitsuffix>): Ditto.
3481 (insn_stnt<n>): Ditto.
3482 (insn_stnt<n>_add<bitsuffix>):Ditto.
3483 (vec_interleave_highv8qi): Handle big endian.
3484 (vec_interleave_highv8qi_be): New.
3485 (vec_interleave_highv8qi_le): New.
3486 (insn_v1int_h): Handle big endian.
3487 (vec_interleave_lowv8qi): Handle big endian.
3488 (vec_interleave_lowv8qi_be): New.
3489 (vec_interleave_lowv8qi_le): New.
3490 (insn_v1int_l): Handle big endian.
3491 (vec_interleave_highv4hi): Handle big endian.
3492 (vec_interleave_highv4hi_be): New.
3493 (vec_interleave_highv4hi_le): New.
3494 (insn_v2int_h): Handle big endian.
3495 (vec_interleave_lowv4hi): Handle big endian.
3496 (vec_interleave_lowv4hi_be): New.
3497 (vec_interleave_lowv4hi_le): New.
3498 (insn_v2int_l): Handle big endian.
3499 (vec_interleave_highv2si): Handle big endian.
3500 (vec_interleave_highv2si_be): New.
3501 (vec_interleave_highv2si_le): New.
3502 (insn_v4int_h): Handle big endian.
3503 (vec_interleave_lowv2si): Handle big endian.
3504 (vec_interleave_lowv2si_be): New.
3505 (vec_interleave_lowv2si_le): New.
3506 (insn_v4int_l): Handle big endian.
3507 * config/tilegx/tilegx.opt (mbig-endian): New option.
3508 (mlittle-endian): New option.
3509 * doc/install.texi: Document tilegxbe-linux.
3510 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
3512 2014-02-24 Martin Jambor <mjambor@suse.cz>
3515 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
3516 there are no parameter descriptors.
3518 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
3520 PR rtl-optimization/60268
3521 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
3522 initialization to ...
3523 (sched_rgn_init): ... here.
3524 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
3526 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3528 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
3531 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3533 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
3536 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3538 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
3539 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
3541 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3543 * config/microblaze/predicates.md: Add cmp_op predicate.
3544 * config/microblaze/microblaze.md: Add branch_compare instruction
3545 which uses cmp_op predicate and emits cmp insn before branch.
3546 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
3547 to microblaze_expand_conditional_branch and consolidate logic.
3548 (microblaze_expand_conditional_branch): emit branch_compare
3549 insn instead of handling cmp op separate from branch insn.
3551 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3553 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
3556 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3558 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
3559 define_insn with define_expand and new define_insn
3560 *altivec_lve<VI_char>x_internal.
3561 (altivec_stve<VI_char>x): Replace define_insn with define_expand
3562 and new define_insn *altivec_stve<VI_char>x_internal.
3563 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
3565 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
3567 (altivec_expand_stvex_be): New function.
3569 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
3571 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
3572 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
3573 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
3574 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
3576 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
3579 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
3580 instead of emit_move_insn.
3582 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3584 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
3586 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
3587 gen_altivec_vsumsws.
3589 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3591 * config/rs6000/altivec.md (altivec_lvxl): Rename as
3592 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
3593 (altivec_lvxl_<mode>): New define_expand incorporating
3594 -maltivec=be semantics where needed.
3595 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
3596 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
3597 semantics where needed.
3598 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
3599 (altivec_stvx_<mode>): New define_expand incorporating
3600 -maltivec=be semantics where needed.
3601 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
3602 VM2 iterator instead of V4SI.
3603 (altivec_stvxl_<mode>): New define_expand incorporating
3604 -maltivec=be semantics where needed.
3605 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
3606 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
3607 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
3608 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
3609 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
3610 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
3611 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
3612 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
3613 ALTIVEC_BUILTIN_STVXL.
3614 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
3615 (altivec_expand_stvx_be): Likewise.
3616 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
3617 (altivec_expand_lvx_be): Likewise.
3618 (altivec_expand_stvx_be): Likewise.
3619 (altivec_expand_builtin): Add cases for
3620 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
3621 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
3622 (altivec_init_builtins): Add definitions for
3623 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
3624 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
3626 2014-02-21 Catherine Moore <clm@codesourcery.com>
3628 * doc/invoke.texi (mvirt, mno-virt): Document.
3629 * config/mips/mips.opt (mvirt): New option.
3630 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
3632 2014-02-21 Richard Biener <rguenther@suse.de>
3634 PR tree-optimization/60276
3635 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
3636 (STMT_VINFO_MIN_NEG_DIST): New macro.
3637 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
3638 STMT_VINFO_MIN_NEG_DIST.
3639 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
3640 made for negative dependence distances still hold.
3642 2014-02-21 Richard Biener <rguenther@suse.de>
3645 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
3646 DECL_INITIAL for globals not in the current function context.
3648 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3650 PR tree-optimization/56490
3651 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
3652 * tree-ssa-uninit.c: Include params.h.
3653 (compute_control_dep_chain): Add num_calls argument, return false
3654 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
3655 num_calls to recursive call.
3656 (find_predicates): Change dep_chain into normal array,
3657 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
3658 variable and adjust compute_control_dep_chain caller.
3659 (find_def_preds): Likewise.
3661 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
3663 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
3664 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
3666 2014-02-21 Nick Clifton <nickc@redhat.com>
3668 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
3669 (pushhi1): Likewise.
3670 (popqi1): Add mode to pre_dec.
3673 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3675 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
3676 mode for mask of V8SFmode permutation.
3678 2014-02-20 Richard Henderson <rth@redhat.com>
3681 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
3682 a new pseudo for OLDVAL.
3684 2014-02-20 Jakub Jelinek <jakub@redhat.com>
3687 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
3688 gen_reg_rtx if d->testing_p.
3689 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
3690 if d->testing_p and we will certainly return true.
3691 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
3694 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
3696 * emit-rtl.c (gen_reg_rtx): Assert that
3697 crtl->emit.regno_pointer_align_length is non-zero.
3699 2014-02-20 Richard Henderson <rth@redhat.com>
3702 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
3703 on failure the store back into EXPECT.
3705 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
3706 Sandra Loosemore <sandra@codesourcery.com>
3708 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
3709 * config/nios2/nios2.c (nios2_function_profiler): Add
3710 -fPIC (flag_pic == 2) support.
3711 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
3712 (nios2_large_offset_p): New function.
3713 (nios2_unspec_reloc_p): Move up position, update to use
3714 nios2_large_offset_p.
3715 (nios2_unspec_address): Remove function.
3716 (nios2_unspec_offset): New function.
3717 (nios2_large_got_address): New function.
3718 (nios2_got_address): Add large offset support.
3719 (nios2_legitimize_tls_address): Update usage of removed and new
3721 (nios2_symbol_binds_local_p): New function.
3722 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
3723 (nios2_legitimize_address): Update to use nios2_large_offset_p.
3724 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
3725 (nios2_print_operand): Merge H/L processing, add hiadj/lo
3726 processing for (const (unspec ...)).
3727 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
3729 2014-02-20 Richard Biener <rguenther@suse.de>
3731 * tree-cfg.c (replace_uses_by): Mark altered BBs before
3732 doing the substitution.
3733 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
3735 2014-02-20 Martin Jambor <mjambor@suse.cz>
3738 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
3739 info when checking whether lattices are bottom.
3741 2014-02-20 Richard Biener <rguenther@suse.de>
3744 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
3747 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
3750 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
3751 parameter specifying the scaling.
3752 (inline_call): Update.
3753 (want_inline_recursively): Guard division by zero.
3754 (recursive_inlining): Update.
3755 * ipa-inline.h (clone_inlined_nodes): Update.
3757 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3760 * config/i386/i386.c (classify_argument): Pass structures of size
3761 64 bytes or less in register.
3763 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3764 Kirill Yukhin <kirill.yukhin@intel.com>
3766 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
3767 (_mm_rcp28_round_ss): Ditto.
3768 (_mm_rsqrt28_round_sd): Ditto.
3769 (_mm_rsqrt28_round_ss): Ditto.
3770 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
3771 (_mm_rcp14_round_ss): Ditto.
3772 (_mm_rsqrt14_round_sd): Ditto.
3773 (_mm_rsqrt14_round_ss): Ditto.
3774 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
3775 the first input operand, get rid of match_dup.
3776 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
3778 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3780 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
3781 operand as the first input operand, set type attribute.
3782 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3784 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
3785 operand as the first input operand, set type attribute.
3787 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3789 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
3792 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
3795 * config/i386/i386.c (construct_container): Remove TFmode check
3796 for X86_64_INTEGER_CLASS.
3798 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
3801 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
3802 only when -Wpsabi is enabled.
3804 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
3807 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
3808 passing arrays in registers are the same as for structs, so remove the
3809 special case for them.
3811 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
3813 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
3814 destination type, extract only the valid bits if the source type is not
3815 integral and has a different mode.
3817 2014-02-19 Richard Biener <rguenther@suse.de>
3820 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
3823 2014-02-19 Richard Biener <rguenther@suse.de>
3826 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
3827 (ipa_modify_call_arguments): Emit an argument load explicitely and
3828 preserve virtual SSA form there and for the replacement call.
3829 Do not update SSA form nor free dominance info.
3831 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3833 * ipa.c (function_and_variable_visibility): Also clear WEAK
3834 flag when disolving COMDAT_GROUP.
3836 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3838 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
3839 * ipa-prop.c (ipa_set_jf_known_type): Return early when
3841 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
3842 do more sanity checks.
3843 (detect_type_change): Return true when giving up early.
3844 (compute_complex_assign_jump_func): Fix type parameter of
3845 ipa_set_ancestor_jf.
3846 (compute_complex_ancestor_jump_func): Likewise.
3847 (update_jump_functions_after_inlining): Fix updating of
3849 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
3851 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3853 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
3854 inline clones when edge disappears.
3856 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
3859 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
3860 Split 64-bit moves into 2 patterns. Do not allow the use of
3861 direct move for TDmode in little endian, since the decimal value
3862 has little endian bytes within a word, but the 64-bit pieces are
3863 ordered in a big endian fashion, and normal subreg's of TDmode are
3865 (mov<mode>_64bit_dm): Likewise.
3866 (movtd_64bit_nodm): Likewise.
3868 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3870 PR tree-optimization/60174
3871 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
3872 statement of an SSA_NAME that occurs in an abnormal PHI node.
3874 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3877 * final.c (SEEN_BB): Remove.
3878 (SEEN_NOTE, SEEN_EMITTED): Renumber.
3879 (final_scan_insn): Don't force_source_line on second
3880 NOTE_INSN_BASIC_BLOCK.
3882 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
3885 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
3886 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
3887 (type_natural_mode): Warn ABI change when %zmm register is not
3888 available for AVX512F vector value passing.
3890 2014-02-18 Kai Tietz <ktietz@redhat.com>
3893 * config/i386/i386.c (ix86_expand_prologue): Use value in
3894 rax register as displacement when restoring %r10 or %rax.
3895 Fix wrong offset when restoring both registers.
3897 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3899 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
3900 assertion with conditional return.
3902 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3903 Uros Bizjak <ubizjak@gmail.com>
3906 * config/i386/driver-i386.c (host_detect_local_cpu): If
3907 YMM state is not saved by the OS, also clear has_f16c. Move
3908 CPUID 0x80000001 handling before YMM state saving checking.
3910 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
3912 PR rtl-optimization/58960
3913 * haifa-sched.c (alloc_global_sched_pressure_data): New,
3914 factored out from ...
3915 (sched_init): ... here.
3916 (free_global_sched_pressure_data): New, factored out from ...
3917 (sched_finish): ... here.
3918 * sched-int.h (free_global_sched_pressure_data): Declare.
3919 * sched-rgn.c (nr_regions_initial): New static global.
3920 (haifa_find_rgns): Initialize it.
3921 (schedule_region): Disable sched-pressure for the newly
3924 2014-02-17 Richard Biener <rguenther@suse.de>
3926 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
3927 release SSA defs of pattern stmts.
3929 2014-02-17 Richard Biener <rguenther@suse.de>
3931 * tree-inline.c (expand_call_inline): Release the virtual
3932 operand defined by the call we are about to inline.
3934 2014-02-17 Richard Biener <rguenther@suse.de>
3936 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
3938 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
3939 Ilya Tocar <ilya.tocar@intel.com>
3941 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
3942 arguments order in builtin.
3943 (_mm512_permutexvar_epi64): Ditto.
3944 (_mm512_mask_permutexvar_epi64): Ditto
3945 (_mm512_maskz_permutexvar_epi32): Ditto
3946 (_mm512_permutexvar_epi32): Ditto
3947 (_mm512_mask_permutexvar_epi32): Ditto
3949 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3951 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
3952 (p8_vmrgow): Likewise.
3954 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3956 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
3959 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3962 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
3963 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
3964 into 64-bit and 32-bit moves. On 64-bit moves, add support for
3965 using direct move instructions on ISA 2.07. Also adjust
3966 instruction length for 64-bit.
3967 (mov<mode>_64bit, TFmode/TDmode): Likewise.
3968 (mov<mode>_32bit, TFmode/TDmode): Likewise.
3970 2014-02-15 Alan Modra <amodra@gmail.com>
3974 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
3975 find_replacement on parts of insn rtl that might be reloaded.
3977 2014-02-15 Richard Biener <rguenther@suse.de>
3979 PR tree-optimization/60183
3980 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
3981 (tree_ssa_phiprop): Calculate and free post-dominators.
3983 2014-02-14 Jeff Law <law@redhat.com>
3985 PR rtl-optimization/60131
3986 * ree.c (get_extended_src_reg): New function.
3987 (combine_reaching_defs): Use it rather than assuming location of REG.
3988 (find_and_remove_re): Verify first operand of extension is
3989 a REG before adding the insns to the copy list.
3991 2014-02-14 Roland McGrath <mcgrathr@google.com>
3993 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
3994 * configure: Regenerated.
3995 * config.in: Regenerated.
3996 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
3997 instead of ASM_SHORT.
3999 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
4000 Richard Earnshaw <rearnsha@arm.com>
4002 PR rtl-optimization/59535
4003 * lra-constraints.c (process_alt_operands): Encourage alternative
4004 when unassigned pseudo class is superset of the alternative class.
4005 (inherit_reload_reg): Don't inherit when optimizing for code size.
4006 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
4007 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
4008 modes not less than 4 for Thumb1.
4010 2014-02-14 Kyle McMartin <kyle@redhat.com>
4013 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
4015 2014-02-14 Richard Biener <rguenther@suse.de>
4017 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
4018 (get_frame_arg): Drop the assert with langhook types_compatible_p.
4019 Do not strip INDIRECT_REFs.
4021 2014-02-14 Richard Biener <rguenther@suse.de>
4024 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
4025 DECL_FUNCTION_SPECIFIC_TARGET.
4026 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
4027 * tree-streamer-out.c (pack_ts_target_option): Remove.
4028 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
4029 (write_ts_function_decl_tree_pointers): Do not stream
4030 DECL_FUNCTION_SPECIFIC_TARGET.
4031 * tree-streamer-in.c (unpack_ts_target_option): Remove.
4032 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
4033 (lto_input_ts_function_decl_tree_pointers): Do not stream
4034 DECL_FUNCTION_SPECIFIC_TARGET.
4036 2014-02-14 Jakub Jelinek <jakub@redhat.com>
4038 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
4039 (get_initial_def_for_induction, vectorizable_induction): Ignore
4040 debug stmts when looking for exit_phi.
4041 (vectorizable_live_operation): Fix up condition.
4043 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
4045 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
4046 nreverse() because it changes the content of original tree list.
4048 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
4050 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
4051 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
4053 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
4055 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
4056 GNU coding standards.
4058 2014-02-13 Jakub Jelinek <jakub@redhat.com>
4061 * dwarf2out.c (gen_subprogram_die): Don't call
4062 add_calling_convention_attribute if subr_die is old_die.
4064 2014-02-13 Sharad Singhai <singhai@google.com>
4066 * doc/optinfo.texi: Fix order of nodes.
4068 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
4070 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
4071 operands[2], not operands[3].
4073 2014-02-13 Richard Biener <rguenther@suse.de>
4076 * doc/install.texi (ISL): Update recommended version to 0.12.2,
4077 mention the possibility of an in-tree build.
4078 (CLooG): Update recommended version to 0.18.1, mention the
4079 possibility of an in-tree build and clarify that the ISL
4080 bundled with CLooG does not work.
4082 2014-02-13 Jakub Jelinek <jakub@redhat.com>
4085 * expr.c (compress_float_constant): If x is a hard register,
4086 extend into a pseudo and then move to x.
4088 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
4090 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
4091 caused by bad second argument to warning_at() with -mhotpatch and
4092 nested functions (e.g. with gfortran).
4094 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
4096 * opts.c (option_name): Remove "enabled by default" rider.
4098 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
4100 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
4102 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
4103 Uros Bizjak <ubizjak@gmail.com>
4106 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
4107 * configure: Regenerated.
4109 2014-02-12 Richard Biener <rguenther@suse.de>
4111 * vec.c (vec_prefix::calculate_allocation): Move as
4112 inline variant to vec.h.
4113 (vec_prefix::calculate_allocation_1): New out-of-line version.
4114 * vec.h (vec_prefix::calculate_allocation_1): Declare.
4115 (vec_prefix::m_has_auto_buf): Rename to ...
4116 (vec_prefix::m_using_auto_storage): ... this.
4117 (vec_prefix::calculate_allocation): Inline the easy cases
4118 and dispatch to calculate_allocation_1 which doesn't need the
4120 (va_heap::reserve): Use gcc_checking_assert.
4121 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
4122 m_using_auto_storage.
4123 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
4125 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
4126 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
4127 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
4129 2014-02-12 Richard Biener <rguenther@suse.de>
4131 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
4132 when we found a dependence.
4134 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
4136 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
4138 (maybe_fold_stmt): ... into this new function.
4139 * omp-low.c (lower_omp): Update comment.
4141 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
4144 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
4147 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
4149 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
4150 identifiers in comments.
4151 (cortexa53_extra_costs): Likewise.
4152 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
4153 (cortexa7_extra_costs): Likewise.
4154 (cortexa12_extra_costs): Likewise.
4155 (cortexa15_extra_costs): Likewise.
4156 (v7m_extra_costs): Likewise.
4158 2014-02-12 Richard Biener <rguenther@suse.de>
4161 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
4162 of posix_memalign being successful.
4163 (lower_stmt): Restrict lowering of posix_memalign to when
4164 -ftree-bit-ccp is enabled.
4166 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
4168 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
4170 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
4172 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
4174 PR rtl-optimization/60116
4175 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
4176 other_insn once the combination has been validated.
4178 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
4181 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
4183 * ipa-devirt.c: Include demangle.h
4184 (odr_violation_reported): New static variable.
4185 (add_type_duplicate): Update odr_violations.
4186 (maybe_record_node): Add completep parameter; update it.
4187 (record_target_from_binfo): Add COMPLETEP parameter;
4188 update it as needed.
4189 (possible_polymorphic_call_targets_1): Likewise.
4190 (struct polymorphic_call_target_d): Add nonconstruction_targets;
4191 rename FINAL to COMPLETE.
4192 (record_targets_from_bases): Sanity check we found the binfo;
4193 fix COMPLETEP updating.
4194 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
4195 parameter, fix computing of COMPLETEP.
4196 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
4197 at LTO time do demangling.
4198 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
4199 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
4201 (gimple_get_virt_method_for_binfo): Likewise.
4202 * gimple-fold.h (gimple_get_virt_method_for_binfo,
4203 gimple_get_virt_method_for_vtable): Update prototypes.
4205 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
4208 * genautomata.c (add_presence_absence): Fix typo with
4209 {final_}presence_list.
4211 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4214 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
4215 for VSX/Altivec vectors that land in GPR registers.
4217 2014-02-11 Richard Henderson <rth@redhat.com>
4218 Jakub Jelinek <jakub@redhat.com>
4221 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
4222 around drhs if type conversion to lacc->type is not useless.
4224 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4226 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
4228 (cortex-a57.cortex-a53): Likewise.
4229 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
4231 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4233 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
4236 2014-02-11 Renlin Li <Renlin.Li@arm.com>
4238 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
4239 add_options_for_arm_vfp3.
4241 2014-02-11 Jeff Law <law@redhat.com>
4244 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
4245 object with an undesirable mode.
4247 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4250 * config/i386/sol2-9.h: New file.
4251 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
4252 *-*-solaris2.9*): Use it.
4254 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
4256 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
4257 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
4259 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
4261 * config/microblaze/microblaze.c: Extend mcpu version format
4263 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
4265 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
4267 2014-02-10 Richard Henderson <rth@redhat.com>
4270 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
4271 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
4272 ms-abi vs -mno-accumulate-outgoing-args.
4273 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
4274 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
4277 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
4280 * cfgexpand.c (expand_asm_operands): Attach source location to
4281 ASM_INPUT rtx objects.
4282 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
4284 2014-02-10 Nick Clifton <nickc@redhat.com>
4286 * config/mn10300/mn10300.c (popcount): New function.
4287 (mn10300_expand_prologue): Include saved registers in stack usage
4290 2014-02-10 Jeff Law <law@redhat.com>
4293 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
4294 when changing the SET_DEST of a prior insn to avoid an input reload.
4296 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
4298 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
4299 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
4300 -mcall-openbsd, or -mcall-linux.
4301 (CC1_ENDIAN_BIG_SPEC): Remove.
4302 (CC1_ENDIAN_LITTLE_SPEC): Remove.
4303 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4304 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
4305 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
4306 and %cc1_endian_default.
4307 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4309 2014-02-10 Richard Biener <rguenther@suse.de>
4311 PR tree-optimization/60115
4312 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
4313 MEM_REF handling. Properly verify that the accesses are not
4314 out of the objects bound.
4316 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4318 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
4321 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
4323 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
4324 proper constants and fix formatting.
4325 (possible_polymorphic_call_targets): Fix formatting.
4327 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
4328 Ilya Tocar <ilya.tocar@intel.com>
4330 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
4331 (_mm512_loadu_epi32): Renamed into...
4332 (_mm512_loadu_si512): This.
4333 (_mm512_storeu_epi32): Renamed into...
4334 (_mm512_storeu_si512): This.
4335 (_mm512_maskz_ceil_ps): Removed.
4336 (_mm512_maskz_ceil_pd): Ditto.
4337 (_mm512_maskz_floor_ps): Ditto.
4338 (_mm512_maskz_floor_pd): Ditto.
4339 (_mm512_floor_round_ps): Ditto.
4340 (_mm512_floor_round_pd): Ditto.
4341 (_mm512_ceil_round_ps): Ditto.
4342 (_mm512_ceil_round_pd): Ditto.
4343 (_mm512_mask_floor_round_ps): Ditto.
4344 (_mm512_mask_floor_round_pd): Ditto.
4345 (_mm512_mask_ceil_round_ps): Ditto.
4346 (_mm512_mask_ceil_round_pd): Ditto.
4347 (_mm512_maskz_floor_round_ps): Ditto.
4348 (_mm512_maskz_floor_round_pd): Ditto.
4349 (_mm512_maskz_ceil_round_ps): Ditto.
4350 (_mm512_maskz_ceil_round_pd): Ditto.
4351 (_mm512_expand_pd): Ditto.
4352 (_mm512_expand_ps): Ditto.
4353 * config/i386/i386.c (ix86_builtins): Remove
4354 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
4355 (bdesc_args): Ditto.
4356 * config/i386/predicates.md (const1256_operand): New.
4357 (const_1_to_2_operand): Ditto.
4358 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4359 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4360 (*avx512pf_gatherpf<mode>sf): Ditto.
4361 (avx512pf_gatherpf<mode>df): Ditto.
4362 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4363 (*avx512pf_gatherpf<mode>df): Ditto.
4364 (avx512pf_scatterpf<mode>sf): Ditto.
4365 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4366 (*avx512pf_scatterpf<mode>sf): Ditto.
4367 (avx512pf_scatterpf<mode>df): Ditto.
4368 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4369 (*avx512pf_scatterpf<mode>df): Ditto.
4370 (avx512f_expand<mode>): Removed.
4371 (<shift_insn><mode>3<mask_name>): Change predicate type.
4373 2014-02-08 Jakub Jelinek <jakub@redhat.com>
4375 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
4376 not at the end of datarefs vector use ordered_remove to avoid
4377 reordering datarefs vector.
4380 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
4381 mark local addressable non-static vars as GOVD_PRIVATE
4382 instead of GOVD_LOCAL.
4383 * omp-low.c (lower_omp_for): Move gimple_bind_vars
4384 and BLOCK_VARS of gimple_bind_block to new_stmt rather
4388 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
4389 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
4390 assume_aligned or alloc_align attributes.
4391 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
4392 arguments. Handle also assume_aligned and alloc_align attributes.
4393 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
4394 calls to functions with assume_aligned or alloc_align attributes.
4395 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
4397 2014-02-08 Terry Guo <terry.guo@arm.com>
4399 * doc/invoke.texi: Document ARM -march=armv7e-m.
4401 2014-02-08 Jakub Jelinek <jakub@redhat.com>
4403 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
4404 flag on __cilkrts_rethrow builtin.
4407 * ipa-cp.c (determine_versionability): Fail at -O0
4408 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
4409 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
4412 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4415 * tree-inline.c (copy_forbidden): Fail for
4416 __attribute__((optimize (0))) functions.
4418 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4420 * varpool.c: Include pointer-set.h.
4421 (varpool_remove_unreferenced_decls): Variables in other partitions
4422 will not be output; be however careful to not lose information
4425 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4427 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
4428 lookup in the vtable constructor.
4430 2014-02-07 Jeff Law <law@redhat.com>
4433 * config/m68k/m68k.md (ashldi_extsi): Turn into a
4434 define_insn_and_split.
4436 * ipa-inline.c (inline_small_functions): Fix typos.
4438 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4440 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
4441 (s390_can_use_return_insn): Declare.
4442 * config/s390/s390.h (EPILOGUE_USES): Define.
4443 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
4445 (s390_chunkify_start): Handle return JUMP_LABELs.
4446 (s390_early_mach): Emit a main_pool instruction on the entry edge.
4447 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
4448 (s390_can_use_return_insn): New functions.
4449 (s390_fix_long_loop_prediction): Handle conditional returns.
4450 (TARGET_SET_UP_BY_PROLOGUE): Define.
4451 * config/s390/s390.md (ANY_RETURN): New code iterator.
4452 (*creturn, *csimple_return, return, simple_return): New patterns.
4454 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4456 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
4457 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
4458 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
4459 REG_CFA_RESTORE list when deciding not to restore a register.
4461 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4463 * config/s390/s390.c: Include tree-pass.h and context.h.
4464 (s390_early_mach): New function, split out from...
4465 (s390_emit_prologue): ...here.
4466 (pass_data_s390_early_mach): New pass structure.
4467 (pass_s390_early_mach): New class.
4468 (s390_option_override): Create and register early_mach pass.
4469 Move to end of file.
4471 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4473 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
4474 to match for the exit block.
4476 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4478 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
4479 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
4480 Reject misaligned operands.
4482 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4484 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
4486 2014-02-07 Richard Biener <rguenther@suse.de>
4489 * gimple-low.c (lower_builtin_posix_memalign): New function.
4490 (lower_stmt): Call it to lower posix_memalign in a way
4491 to make alignment info accessible.
4493 2014-02-07 Jakub Jelinek <jakub@redhat.com>
4496 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
4497 __builtin_setjmp_receiver.
4499 2014-02-07 Richard Biener <rguenther@suse.de>
4502 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
4503 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
4504 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4505 Handle BUILT_IN_POSIX_MEMALIGN.
4506 (find_func_clobbers): Likewise.
4507 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4508 (call_may_clobber_ref_p_1): Likewise.
4510 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4513 * ipa-devirt.c (record_target_from_binfo): Remove overactive
4516 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4519 * lto-cgraph.c (lto_output_node): Use
4520 symtab_get_symbol_partitioning_class.
4521 (lto_output_varpool_node): likewise.
4522 (symtab_get_symbol_partitioning_class): Move here from
4524 * cgraph.h (symbol_partitioning_class): Likewise.
4525 (symtab_get_symbol_partitioning_class): Declare.
4527 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4529 * ggc.h (ggc_internal_cleared_alloc): New macro.
4530 * vec.h (vec_safe_copy): Handle memory stats.
4531 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
4532 * target-globals.c (save_target_globals): Likewise.
4534 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4537 * expr.c (emit_move_resolve_push): Export; be bit more selective
4538 on when to clear alias set.
4539 * expr.h (emit_move_resolve_push): Declare.
4540 * function.h (struct function): Add tail_call_marked.
4541 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
4542 * config/i386/i386-protos.h (ix86_expand_push): Remove.
4543 * config/i386/i386.md (TImode move expander): De not call
4545 (FP push expanders): Preserve memory attributes.
4546 * config/i386/sse.md (push<mode>1): Remove.
4547 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
4548 (ix86_expand_push): Remove.
4549 * config/i386/mmx.md (push<mode>1): Remove.
4551 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4553 PR rtl-optimization/60030
4554 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
4555 lopart with paradoxical subreg before shifting it up by hprec.
4557 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4559 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
4560 Remove extra newline at end of file.
4561 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
4562 (arm_issue_rate): Handle cortexa57.
4563 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
4564 (cortex-a57.cortex-a53): Likewise.
4566 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4569 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
4570 don't record in REG_FRAME_RELATED_EXPR registers not set in that
4572 (arm_expand_prologue): Adjust all callers.
4573 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
4574 info, registers also at the lowest numbered registers side. Use
4575 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
4579 * var-tracking.c (adjust_mems): Before adding a SET to
4580 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
4582 2014-02-06 Alan Modra <amodra@gmail.com>
4585 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
4586 change SDmode to DDmode when lra_in_progress.
4588 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4591 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
4592 free_data_ref on the dr first, and before goto again also set dr
4593 to the next dr. For simd_lane_access, free old datarefs[i] before
4594 overwriting it. For get_vectype_for_scalar_type failure, don't
4595 free_data_ref if simd_lane_access.
4597 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
4600 * tree.h (opts_for_fn): New inline function.
4601 (opt_for_fn): Define.
4602 * config/i386/i386.c (ix86_function_regparm): Use
4603 opt_for_fn (decl, optimize) instead of optimize.
4605 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
4607 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
4608 for SYMBOL_REF in large memory model.
4610 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4612 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
4614 (cortex-a57): Likewise.
4615 (cortex-a57.cortex-a53): Likewise.
4617 2014-02-06 Yury Gribov <y.gribov@samsung.com>
4618 Kugan Vivekanandarajah <kuganv@linaro.org>
4620 * config/arm/arm.c (arm_vector_alignment_reachable): Check
4622 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
4624 2014-02-06 Richard Biener <rguenther@suse.de>
4626 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
4627 set_loop_copy and initialize_original_copy_tables.
4629 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
4631 * config/aarch64/aarch64-simd.md
4632 (aarch64_ashr_simddi): Change QI to SI.
4634 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4635 Jakub Jelinek <jakub@redhat.com>
4638 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
4641 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4643 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
4644 CODE_FOR_altivec_vpku[hw]um to
4645 CODE_FOR_altivec_vpku[hw]um_direct.
4646 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
4647 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
4648 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
4649 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
4651 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4653 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
4654 generation for -maltivec=be.
4655 (altivec_vsumsws): Simplify redundant test.
4657 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4659 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
4660 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
4661 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
4662 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
4663 gen_altivec_vpkuwum.
4664 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
4666 (altivec_vpks<VI_char>ss): Likewise.
4667 (altivec_vpks<VI_char>us): Likewise.
4668 (altivec_vpku<VI_char>us): Likewise.
4669 (altivec_vpku<VI_char>um): Likewise.
4670 (altivec_vpku<VI_char>um_direct): New (copy of
4671 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
4673 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
4674 target is little endian and -maltivec=be is not specified.
4675 (*altivec_vupkhs<VU_char>_direct): New (copy of
4676 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
4677 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
4678 target is little endian and -maltivec=be is not specified.
4679 (*altivec_vupkls<VU_char>_direct): New (copy of
4680 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
4681 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
4682 little endian and -maltivec=be is not specified.
4683 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
4684 little endian and -maltivec=be is not specified.
4686 2014-02-05 Richard Henderson <rth@redhat.com>
4689 * combine-stack-adj.c: Revert r206943.
4690 * sched-int.h (struct deps_desc): Add last_args_size.
4691 * sched-deps.c (init_deps): Initialize it.
4692 (sched_analyze_insn): Add OUTPUT dependencies between insns that
4693 contain REG_ARGS_SIZE notes.
4695 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4697 * lto-cgraph.c (asm_nodes_output): Make global.
4698 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
4699 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
4700 (driver_handle_option): Handle OPT_fwpa.
4702 2014-02-05 Jakub Jelinek <jakub@redhat.com>
4705 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
4706 a comment typo and formatting issue. If odr_hash hasn't been
4707 created, return vNULL and set *completep to false.
4710 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
4711 bb with no successors.
4713 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
4716 * doc/invoke.texi (-march): Clarify documentation for ARM.
4720 2014-02-05 Richard Biener <rguenther@suse.de>
4722 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
4723 when not vectorizing because of too many alias checks.
4724 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
4725 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
4727 2014-02-05 Nick Clifton <nickc@redhat.com>
4729 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
4730 accept extended registers in any mode when compiling for the MN10300.
4732 2014-02-05 Yury Gribov <y.gribov@samsung.com>
4734 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
4735 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
4736 sanitization attributes.
4737 (can_inline_edge_p): Likewise.
4738 (sanitize_attrs_match_for_inline_p): New function.
4740 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4742 * ipa-prop.c (detect_type_change): Shor circuit testing of
4743 type changes on THIS pointer.
4745 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
4748 * config/pa/pa.c (legitimize_tls_address): Return original address
4749 if not passed a SYMBOL_REF rtx.
4750 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
4752 (pa_emit_move_sequence): Simplify TLS source operands.
4753 (pa_legitimate_constant_p): Reject all TLS constants.
4754 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
4755 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
4757 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4759 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
4760 groups when we know they are controlled by LTO.
4761 * varasm.c (default_binds_local_p_1): If object is in other partition,
4762 it will be resolved locally.
4764 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4766 * config/host-linux.c (linux_gt_pch_use_address): Don't
4767 use SSIZE_MAX because it is not always defined.
4769 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
4772 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
4773 threshold for pseudo splitting.
4774 (update_ebb_live_info): Process call argument hard registers and
4775 hard registers from insn definition too.
4776 (max_small_class_regs_num): New constant.
4777 (inherit_in_ebb): Update live hard regs through EBBs. Update
4778 reloads_num only for small register classes. Don't split for
4781 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
4784 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
4787 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4789 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
4792 2014-02-04 Marek Polacek <polacek@redhat.com>
4794 * gdbinit.in (pel): Define.
4796 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4798 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
4801 2014-02-04 Richard Biener <rguenther@suse.de>
4804 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
4805 in function context local.
4806 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
4807 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
4808 similar to LTO_imported_decl_ref.
4810 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4812 PR tree-optimization/60002
4813 * cgraphclones.c (build_function_decl_skip_args): Clear
4816 PR tree-optimization/60023
4817 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
4818 false to gsi_replace.
4819 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
4820 has been in some EH region and vec_stmt could throw, add
4821 vec_stmt into the same EH region.
4822 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
4823 has no lhs, ignore it.
4824 * internal-fn.c (expand_MASK_LOAD): Likewise.
4827 * tree-inline.c (copy_forbidden): Fail for
4828 __attribute__((optimize (0))) functions.
4831 * omp-low.c (simd_clone_struct_copy): If from->inbranch
4832 is set, copy one less argument.
4833 (expand_simd_clones): Don't subtract clone_info->inbranch
4834 from simd_clone_struct_alloc argument.
4836 PR rtl-optimization/57915
4837 * recog.c (simplify_while_replacing): If all unary/binary/relational
4838 operation arguments are constant, attempt to simplify those.
4841 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
4842 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
4844 2014-02-04 Richard Biener <rguenther@suse.de>
4846 PR tree-optimization/60012
4847 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
4848 TBAA disambiguation to all DDRs.
4850 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4853 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
4854 (LINK_SPEC): Use it for -shared, -shared-libgcc.
4856 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4859 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
4861 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4863 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
4864 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
4866 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4869 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
4870 to figure out targets of polymorphic calls with known decl.
4871 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4872 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
4873 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
4874 (get_polymorphic_call_info): ... here.
4875 (get_polymorphic_call_info_from_invariant): New function.
4877 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4879 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
4880 lookup via vtable pointer; check for type consistency
4881 and turn inconsitent facts into UNREACHABLE.
4882 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4883 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
4884 type inconsistent querries; return UNREACHABLE instead.
4886 2014-02-03 Richard Henderson <rth@twiddle.net>
4889 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
4890 already processed this node.
4891 (normalize_one_pred_1): Pass along mark_set.
4892 (normalize_one_pred): Create and destroy a pointer_set_t.
4893 (normalize_one_pred_chain): Likewise.
4895 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
4897 PR gcov-profile/58602
4898 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
4900 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4903 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
4904 -fno-devirtualize; try to devirtualize by the knowledge of
4905 virtual table pointer given by aggregate propagation.
4906 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4907 (ipa_print_node_jump_functions): Dump also offset that
4908 is relevant for polymorphic calls.
4909 (determine_known_aggregate_parts): Add arg_type parameter; use it
4910 instead of determining the type from pointer type.
4911 (ipa_compute_jump_functions_for_edge): Update call of
4912 determine_known_aggregate_parts.
4913 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
4914 (gimple_get_virt_method_for_binfo): ... here; simplify using
4915 vtable_pointer_value_to_vtable.
4916 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
4917 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
4918 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
4919 (vtable_pointer_value_to_vtable): Break out from ...; handle also
4921 (vtable_pointer_value_to_binfo): ... here.
4922 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
4924 2014-02-03 Teresa Johnson <tejohnson@google.com>
4926 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
4927 redef of outer loop index variable.
4929 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
4933 * doc/extend.texi (Function Attributes): Typo.
4935 2014-02-03 Cong Hou <congh@google.com>
4937 PR tree-optimization/60000
4938 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
4939 if the vectorized statement is a store. A store statement can only
4940 appear at the end of pattern statements.
4942 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4944 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
4945 (ix86_option_override_internal): Default long double to 64-bit for
4946 32-bit Bionic and to 128-bit for 64-bit Bionic.
4948 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
4949 TARGET_LONG_DOUBLE_128 is true.
4950 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
4952 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
4953 (mlong-double-64): Negate -mlong-double-128.
4954 (mlong-double-128): New option.
4956 * config/i386/i386-c.c (ix86_target_macros): Define
4957 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
4959 * doc/invoke.texi: Document -mlong-double-128.
4961 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4963 PR rtl-optimization/60024
4964 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
4966 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
4968 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
4970 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4972 PR rtl-optimization/57662
4973 * sel-sched.c (code_motion_path_driver): Do not mark already not
4974 existing blocks in the visiting bitmap.
4976 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4978 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
4979 on the insn being emitted.
4981 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
4982 Will Deacon <will.deacon@arm.com>
4984 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
4986 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4988 * config/arm/arm-tables.opt: Regenerate.
4990 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4992 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
4993 for vector types other than V16QImode.
4994 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
4995 define_expand, and call altivec_expand_vec_perm_le when producing
4996 code with little endian element order.
4997 (*altivec_vperm_<mode>_internal): New insn having previous
4998 behavior of altivec_vperm_<mode>.
4999 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
5000 altivec_expand_vec_perm_le when producing code with little endian
5002 (*altivec_vperm_<mode>_uns_internal): New insn having previous
5003 behavior of altivec_vperm_<mode>_uns.
5005 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5007 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
5008 (altivec_vsumsws): Add handling for -maltivec=be with a little
5010 (altivec_vsumsws_direct): New.
5011 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
5012 gen_altivec_vsumsws.
5014 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
5016 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
5017 vtable_pointer_value_to_binfo): New functions.
5018 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
5019 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
5021 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
5023 * config/nios2/nios2.md (load_got_register): Initialize GOT
5024 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
5025 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
5027 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
5029 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
5030 preserverd by passthrough, do not propagate the type.
5032 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
5034 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
5035 (mips_atomic_assign_expand_fenv): New function.
5036 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
5038 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
5040 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
5041 (__builtin_mips_set_fcsr): Likewise.
5042 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
5043 MIPS_USI_FTYPE_VOID.
5044 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
5045 (mips16_expand_set_fcsr): Likewise.
5046 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
5047 (mips16_set_fcsr_stub): Likewise.
5048 (mips16_get_fcsr_one_only_stub): New class.
5049 (mips16_set_fcsr_one_only_stub): Likewise.
5050 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
5051 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
5052 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
5053 (hard_float): New availability predicate.
5054 (mips_builtins): Add get_fcsr and set_fcsr.
5055 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
5056 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
5057 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
5058 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
5059 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
5062 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
5064 * config/mips/mips.c (mips_one_only_stub): New class.
5065 (mips_need_mips16_rdhwr_p): Replace with...
5066 (mips16_rdhwr_stub): ...this new variable.
5067 (mips16_stub_call_address): New function.
5068 (mips16_rdhwr_one_only_stub): New class.
5069 (mips_expand_thread_pointer): Use mips16_stub_call_address.
5070 (mips_output_mips16_rdhwr): Delete.
5071 (mips_finish_stub): New function.
5072 (mips_code_end): Use it to handle rdhwr stubs.
5074 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
5077 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
5078 when calculating size of integer atomic types.
5080 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
5082 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
5084 2014-02-01 Jakub Jelinek <jakub@redhat.com>
5086 PR tree-optimization/60003
5087 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
5088 * profile.c (branch_prob): Use gimple_call_builtin_p
5089 to check for BUILT_IN_SETJMP_RECEIVER.
5090 * tree-inline.c (copy_bb): Call notice_special_calls.
5092 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
5095 * lra-constraints.c (process_alt_operands): Update reload_sum only
5098 2014-01-31 Richard Henderson <rth@redhat.com>
5101 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
5102 until after else_eh is processed.
5104 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
5106 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
5107 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
5108 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
5109 in smmintrin.h, remove them.
5110 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
5111 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
5112 * config/i386/i386.md (ROUND_SAE): Fix value.
5113 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
5114 (const48_operand): New.
5115 * config/i386/subst.md (round), (round_expand): Use
5116 const_4_or_8_to_11_operand.
5117 (round_saeonly), (round_saeonly_expand): Use const48_operand.
5119 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
5121 * config/i386/constraints.md (Yk): Swap meaning with k.
5122 * config/i386/i386.md (movhi_internal): Change Yk to k.
5123 (movqi_internal): Ditto.
5124 (*k<logic><mode>): Ditto.
5127 (kandn<mode>): Ditto.
5128 (*<code>hi_1): Ditto.
5129 (*<code>qi_1): Ditto.
5130 (kxnor<mode>): Ditto.
5131 (kortestzhi): Ditto.
5132 (kortestchi): Ditto.
5134 (*one_cmplhi2_1): Ditto.
5135 (*one_cmplqi2_1): Ditto.
5136 * config/i386/sse.md (): Change k to Yk.
5137 (avx512f_load<mode>_mask): Ditto.
5138 (avx512f_blendm<mode>): Ditto.
5139 (avx512f_store<mode>_mask): Ditto.
5140 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
5141 (avx512f_storedqu<mode>_mask): Ditto.
5142 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
5144 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
5145 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
5146 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
5147 (avx512f_maskcmp<mode>3): Ditto.
5148 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
5149 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
5150 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
5151 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
5152 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
5153 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
5154 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
5155 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
5156 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
5157 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
5158 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
5159 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
5160 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
5161 (vec_extract_lo_<mode>_maskm): Ditto.
5162 (vec_extract_hi_<mode>_maskm): Ditto.
5163 (avx512f_vternlog<mode>_mask): Ditto.
5164 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
5165 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
5166 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
5167 (avx512f_<code>v8div16qi2_mask): Ditto.
5168 (avx512f_<code>v8div16qi2_mask_store): Ditto.
5169 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
5170 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
5171 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
5172 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
5173 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5174 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5175 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5176 (*avx512pf_scatterpf<mode>df_mask): Ditto.
5177 (avx512cd_maskb_vec_dupv8di): Ditto.
5178 (avx512cd_maskw_vec_dupv16si): Ditto.
5179 (avx512f_vpermi2var<mode>3_maskz): Ditto.
5180 (avx512f_vpermi2var<mode>3_mask): Ditto.
5181 (avx512f_vpermi2var<mode>3_mask): Ditto.
5182 (avx512f_vpermt2var<mode>3_maskz): Ditto.
5183 (*avx512f_gathersi<mode>): Ditto.
5184 (*avx512f_gathersi<mode>_2): Ditto.
5185 (*avx512f_gatherdi<mode>): Ditto.
5186 (*avx512f_gatherdi<mode>_2): Ditto.
5187 (*avx512f_scattersi<mode>): Ditto.
5188 (*avx512f_scatterdi<mode>): Ditto.
5189 (avx512f_compress<mode>_mask): Ditto.
5190 (avx512f_compressstore<mode>_mask): Ditto.
5191 (avx512f_expand<mode>_mask): Ditto.
5192 * config/i386/subst.md (mask): Change k to Yk.
5193 (mask_scalar_merge): Ditto.
5196 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
5198 * doc/extend.texi (Vector Extensions): Document ?: in C++.
5200 2014-01-31 Richard Biener <rguenther@suse.de>
5203 * builtins.c (fold_builtin_memory_op): Make sure to not
5204 use a floating-point mode or a boolean or enumeral type for
5207 2014-01-30 DJ Delorie <dj@redhat.com>
5209 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
5210 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
5211 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
5212 whenever main() has an epilogue.
5214 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5216 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
5217 unused variable "field".
5218 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
5219 (vsx_mergeh_<mode>): Likewise.
5220 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
5221 (altivec_vmrghh): Likewise.
5222 (altivec_vmrghw): Likewise.
5223 (altivec_vmrglb): Likewise.
5224 (altivec_vmrglh): Likewise.
5225 (altivec_vmrglw): Likewise.
5226 (altivec_vspltb): Add missing uses.
5227 (altivec_vsplth): Likewise.
5228 (altivec_vspltw): Likewise.
5229 (altivec_vspltsf): Likewise.
5231 2014-01-30 Jakub Jelinek <jakub@redhat.com>
5234 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
5235 frame related instructions.
5237 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
5239 PR rtl-optimization/59959
5240 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
5241 any reload of register whose subreg is invalid.
5243 2014-01-30 Jakub Jelinek <jakub@redhat.com>
5245 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
5246 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
5247 Add missing return type - void.
5249 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5251 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
5252 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
5253 remove element index adjustment for endian (now handled in vsx.md
5255 (altivec_expand_vec_perm_const): Use
5256 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
5257 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
5258 (vsx_xxspltw_<mode>): Adjust element index for little endian.
5259 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
5260 define_expand and a new define_insn *altivec_vspltb_internal;
5261 adjust for -maltivec=be on a little endian target.
5262 (altivec_vspltb_direct): New.
5263 (altivec_vsplth): Divide into a define_expand and a new
5264 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
5265 little endian target.
5266 (altivec_vsplth_direct): New.
5267 (altivec_vspltw): Divide into a define_expand and a new
5268 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
5269 little endian target.
5270 (altivec_vspltw_direct): New.
5271 (altivec_vspltsf): Divide into a define_expand and a new
5272 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
5273 a little endian target.
5275 2014-01-30 Richard Biener <rguenther@suse.de>
5277 PR tree-optimization/59993
5278 * tree-ssa-forwprop.c (associate_pointerplus): Check we
5279 can propagate form the earlier stmt and avoid the transform
5280 when the intermediate result is needed.
5282 2014-01-30 Alangi Derick <alangiderick@gmail.com>
5284 * README.Portability: Fix typo.
5286 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
5288 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
5289 comparison_operator with ordered_comparison_operator.
5291 2014-01-30 Nick Clifton <nickc@redhat.com>
5293 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
5294 Rename to mn10300_store_multiple_regs.
5295 * config/mn10300/mn10300.c: Likewise.
5296 * config/mn10300/mn10300.md (store_movm): Fix typo: call
5297 store_multiple_regs.
5298 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
5299 Call mn10300_store_multiple_regs.
5301 2014-01-30 Nick Clifton <nickc@redhat.com>
5302 DJ Delorie <dj@redhat.com>
5304 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
5305 %fp 2 to keep registers after it properly word-aligned.
5306 (rl78_alloc_physical_registers_umul): Handle the case where both
5307 input operands are the same.
5309 2014-01-30 Richard Biener <rguenther@suse.de>
5311 PR tree-optimization/59903
5312 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
5315 2014-01-30 Jason Merrill <jason@redhat.com>
5318 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
5321 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
5323 2014-01-30 Richard Biener <rguenther@suse.de>
5325 PR tree-optimization/59951
5326 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
5328 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
5331 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
5332 SFmode to DFmode case.
5334 2014-01-29 DJ Delorie <dj@redhat.com>
5336 * config/msp430/msp430.opt (-minrt): New.
5337 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
5339 (ENDFILE_SPEC): Likewise.
5341 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
5343 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
5344 (estimate_function_body_sizes): Use it.
5346 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
5349 * dwarf2out.c (is_cxx_auto): New.
5350 (is_base_type): Use it.
5351 (gen_type_die_with_usage): Likewise.
5353 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5355 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
5356 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
5357 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
5358 -maltivec=be with LE targets.
5359 (vsx_mergeh_<mode>): Likewise.
5360 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
5361 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
5362 (altivec_vmrghb): Replace with define_expand and new
5363 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
5364 (altivec_vmrghb_direct): New define_insn.
5365 (altivec_vmrghh): Replace with define_expand and new
5366 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
5367 (altivec_vmrghh_direct): New define_insn.
5368 (altivec_vmrghw): Replace with define_expand and new
5369 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
5370 (altivec_vmrghw_direct): New define_insn.
5371 (*altivec_vmrghsf): Adjust for endianness.
5372 (altivec_vmrglb): Replace with define_expand and new
5373 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
5374 (altivec_vmrglb_direct): New define_insn.
5375 (altivec_vmrglh): Replace with define_expand and new
5376 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
5377 (altivec_vmrglh_direct): New define_insn.
5378 (altivec_vmrglw): Replace with define_expand and new
5379 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
5380 (altivec_vmrglw_direct): New define_insn.
5381 (*altivec_vmrglsf): Adjust for endianness.
5382 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5383 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5384 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5385 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5386 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5387 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5388 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5389 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5391 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
5393 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
5394 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
5397 2014-01-29 Richard Biener <rguenther@suse.de>
5399 PR tree-optimization/58742
5400 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
5401 associate_pointerplus_align.
5402 (associate_pointerplus_diff): New function.
5403 (associate_pointerplus): Likewise. Call associate_pointerplus_align
5404 and associate_pointerplus_diff.
5406 2014-01-29 Richard Biener <rguenther@suse.de>
5408 * lto-streamer.h (LTO_major_version): Bump to 3.
5409 (LTO_minor_version): Reset to 0.
5411 2014-01-29 Renlin Li <Renlin.Li@arm.com>
5413 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
5414 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
5415 (arm_file_start): Generate correct asm header for armv7ve.
5416 * config/arm/bpabi.h: Add multilib support for armv7ve.
5417 * config/arm/driver-arm.c: Change the architectures of cortex-a7
5418 and cortex-a15 to armv7ve.
5419 * config/arm/t-aprofile: Add multilib support for armv7ve.
5420 * doc/invoke.texi: Document -march=armv7ve.
5422 2014-01-29 Richard Biener <rguenther@suse.de>
5424 PR tree-optimization/58742
5425 * tree-ssa-forwprop.c (associate_plusminus): Return true
5426 if we changed sth, defer EH cleanup to ...
5427 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
5428 (simplify_mult): New function.
5430 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5433 PR tree-optimization/59920
5434 * tree.c (build_common_builtin_nodes): Remove
5435 __builtin_setjmp_dispatcher initialization.
5436 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
5437 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
5438 instead of gsi_after_labels + manually skipping debug stmts.
5439 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
5440 ignore bbs with IFN_ABNORMAL_DISPATCHER.
5441 * tree-inline.c (copy_edges_for_bb): Remove
5442 can_make_abnormal_goto argument, instead add abnormal_goto_dest
5443 argument. Ignore computed_goto_p stmts. Don't call
5444 make_abnormal_goto_edges. If a call might need abnormal edges
5445 for non-local gotos, see if it already has an edge to
5446 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
5447 with true argument, don't do anything then, otherwise add
5448 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
5449 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
5451 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
5452 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
5453 (lower_stmt): Don't set data->calls_builtin_setjmp.
5454 (lower_builtin_setjmp): Adjust comment.
5455 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
5456 * tree-cfg.c (found_computed_goto): Remove.
5457 (factor_computed_gotos): Remove.
5458 (make_goto_expr_edges): Return bool, true for computed gotos.
5459 Don't call make_abnormal_goto_edges.
5460 (build_gimple_cfg): Don't set found_computed_goto, don't call
5461 factor_computed_gotos.
5462 (computed_goto_p): No longer static.
5463 (make_blocks): Don't set found_computed_goto.
5464 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
5465 (make_edges): If make_goto_expr_edges returns true, push bb
5466 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
5467 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
5468 vector. Record mapping between bbs and OpenMP regions if there
5469 are any, adjust make_gimple_omp_edges caller. Call
5470 handle_abnormal_edges.
5471 (make_abnormal_goto_edges): Remove.
5472 * tree-cfg.h (make_abnormal_goto_edges): Remove.
5473 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
5474 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
5475 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
5476 * internal-fn.def (ABNORMAL_DISPATCHER): New.
5477 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
5478 filling *region also set *region_idx to (*region)->entry->index.
5481 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
5482 For REGs set ORIGINAL_REGNO.
5484 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
5486 * doc/md.texi: Mention that a target shouldn't implement
5487 vec_widen_(s|u)mul_even/odd pair if it is less efficient
5490 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5492 PR tree-optimization/59594
5493 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
5494 a copy of the datarefs vector rather than the vector itself.
5496 2014-01-28 Jason Merrill <jason@redhat.com>
5499 * dwarf2out.c (auto_die): New static.
5500 (gen_type_die_with_usage): Handle C++1y 'auto'.
5501 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
5504 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
5507 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
5508 (SPEC_X32): Likewise.
5509 (SPEC_64): Likewise.
5510 * config/i386/i386.c (ix86_option_override_internal): Turn off
5511 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
5513 (x86_file_start): Output .code16gcc for TARGET_16BIT.
5514 * config/i386/i386.h (TARGET_16BIT): New macro.
5515 (TARGET_16BIT_P): Likewise.
5516 * config/i386/i386.opt: Add m16.
5517 * doc/invoke.texi: Document -m16.
5519 2014-01-28 Jakub Jelinek <jakub@redhat.com>
5521 PR preprocessor/59935
5522 * input.c (location_get_source_line): Bail out on when line number
5523 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
5525 2014-01-28 Richard Biener <rguenther@suse.de>
5527 PR tree-optimization/58742
5528 * tree-ssa-forwprop.c (associate_plusminus): Handle
5529 pointer subtraction of the form (T)(P + A) - (T)P.
5531 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5533 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
5536 2014-01-28 Richard Biener <rguenther@suse.de>
5539 2014-01-28 Richard Biener <rguenther@suse.de>
5541 PR rtl-optimization/45364
5542 PR rtl-optimization/59890
5543 * var-tracking.c (local_get_addr_clear_given_value): Handle
5544 already cleared slot.
5545 (val_reset): Handle not allocated local_get_addr_cache.
5546 (vt_find_locations): Use post-order on the inverted CFG.
5548 2014-01-28 Richard Biener <rguenther@suse.de>
5550 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
5552 2014-01-28 Richard Biener <rguenther@suse.de>
5554 PR rtl-optimization/45364
5555 PR rtl-optimization/59890
5556 * var-tracking.c (local_get_addr_clear_given_value): Handle
5557 already cleared slot.
5558 (val_reset): Handle not allocated local_get_addr_cache.
5559 (vt_find_locations): Use post-order on the inverted CFG.
5561 2014-01-28 Alan Modra <amodra@gmail.com>
5563 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
5564 * configure.ac <recursive call for build != host>: Define
5565 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
5566 and LD_FOR_BUILD too.
5567 * configure: Regenerate.
5569 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
5571 * config/i386/i386.c (get_builtin_code_for_version): Separate
5572 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
5573 Broadwell from Haswell.
5575 2014-01-27 Steve Ellcey <sellcey@mips.com>
5577 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
5578 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
5579 * config/mips/mips.c (mips_option_override): Change setting
5581 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
5582 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
5583 Change from Mask to Var.
5585 2014-01-27 Jeff Law <law@redhat.com>
5587 * ipa-inline.c (inline_small_functions): Fix typo.
5589 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
5591 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
5592 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
5593 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
5594 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
5595 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
5596 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
5597 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
5598 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
5599 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
5600 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
5601 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
5602 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
5603 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
5604 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
5605 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
5606 (_mm512_storeu_epi64): Ditto.
5607 (_mm512_cmpge_epi32_mask): Ditto.
5608 (_mm512_cmpge_epu32_mask): Ditto.
5609 (_mm512_cmpge_epi64_mask): Ditto.
5610 (_mm512_cmpge_epu64_mask): Ditto.
5611 (_mm512_cmple_epi32_mask): Ditto.
5612 (_mm512_cmple_epu32_mask): Ditto.
5613 (_mm512_cmple_epi64_mask): Ditto.
5614 (_mm512_cmple_epu64_mask): Ditto.
5615 (_mm512_cmplt_epi32_mask): Ditto.
5616 (_mm512_cmplt_epu32_mask): Ditto.
5617 (_mm512_cmplt_epi64_mask): Ditto.
5618 (_mm512_cmplt_epu64_mask): Ditto.
5619 (_mm512_cmpneq_epi32_mask): Ditto.
5620 (_mm512_cmpneq_epu32_mask): Ditto.
5621 (_mm512_cmpneq_epi64_mask): Ditto.
5622 (_mm512_cmpneq_epu64_mask): Ditto.
5623 (_mm512_expand_pd): Ditto.
5624 (_mm512_expand_ps): Ditto.
5625 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
5626 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
5627 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
5628 * config/i386/i386.c (ix86_builtins): Add
5629 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
5630 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
5631 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
5632 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
5633 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
5634 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
5635 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
5636 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
5637 IX86_BUILTIN_PMOVUSQW512_MEM.
5638 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
5639 __builtin_ia32_pmovsqd512mem_mask,
5640 __builtin_ia32_pmovqd512mem_mask,
5641 __builtin_ia32_pmovusqw512mem_mask,
5642 __builtin_ia32_pmovsqw512mem_mask,
5643 __builtin_ia32_pmovqw512mem_mask,
5644 __builtin_ia32_pmovusdw512mem_mask,
5645 __builtin_ia32_pmovsdw512mem_mask,
5646 __builtin_ia32_pmovdw512mem_mask,
5647 __builtin_ia32_pmovqb512mem_mask,
5648 __builtin_ia32_pmovusqb512mem_mask,
5649 __builtin_ia32_pmovsqb512mem_mask,
5650 __builtin_ia32_pmovusdb512mem_mask,
5651 __builtin_ia32_pmovsdb512mem_mask,
5652 __builtin_ia32_pmovdb512mem_mask.
5653 (bdesc_args): Add __builtin_ia32_expanddf512,
5654 __builtin_ia32_expandsf512.
5655 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
5656 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
5657 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
5658 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
5659 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
5660 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
5661 (avx512f_<code>v8div16qi2_mask_store): This.
5662 (avx512f_expand<mode>): New.
5664 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
5666 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
5668 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5669 (_mm512_prefetch_i32scatter_pd): Ditto.
5670 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5671 (_mm512_prefetch_i64scatter_pd): Ditto.
5672 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5673 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
5674 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5675 (_mm512_prefetch_i32scatter_ps): Ditto.
5676 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5677 (_mm512_prefetch_i64scatter_ps): Ditto.
5678 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5679 * config/i386/i386-builtin-types.def: Define
5680 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
5681 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
5682 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
5683 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
5684 IX86_BUILTIN_SCATTERPFQPD.
5685 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
5686 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
5687 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
5688 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
5689 __builtin_ia32_scatterpfqps.
5690 (ix86_expand_builtin): Expand new built-ins.
5691 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
5692 fix memory access data type.
5693 (*avx512pf_gatherpf<mode>_mask): Ditto.
5694 (*avx512pf_gatherpf<mode>): Ditto.
5695 (avx512pf_scatterpf<mode>): Ditto.
5696 (*avx512pf_scatterpf<mode>_mask): Ditto.
5697 (*avx512pf_scatterpf<mode>): Ditto.
5698 (GATHER_SCATTER_SF_MEM_MODE): New.
5699 (avx512pf_gatherpf<mode>df): Ditto.
5700 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5701 (*avx512pf_scatterpf<mode>df): Ditto.
5703 2014-01-27 Jakub Jelinek <jakub@redhat.com>
5706 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
5707 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
5710 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5712 * common/config/arm/arm-common.c
5713 (arm_rewrite_mcpu): Handle multiple names.
5715 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5717 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5719 * gimple-builder.h (create_gimple_tmp): Delete.
5721 2014-01-27 Christian Bruel <christian.bruel@st.com>
5723 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
5726 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
5728 * config/pa/pa.md (call): Generate indirect long calls to non-local
5729 functions when outputing 32-bit code.
5730 (call_value): Likewise except for special call to buggy powf function.
5732 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
5733 portable runtime and PIC indirect calls.
5734 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
5735 and PIC call sequences. Use ldo instead of blr to set return register
5736 in PIC call sequence.
5738 2014-01-25 Walter Lee <walt@tilera.com>
5740 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
5741 avoid clobbering a live register.
5743 2014-01-25 Walter Lee <walt@tilera.com>
5745 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
5746 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
5747 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
5748 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
5750 2014-01-25 Walter Lee <walt@tilera.com>
5752 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
5753 arguments on even registers.
5754 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
5756 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
5757 (BIGGEST_ALIGNMENT): Ditto.
5758 (BIGGEST_FIELD_ALIGNMENT): Ditto.
5760 2014-01-25 Walter Lee <walt@tilera.com>
5762 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
5763 insns before bundling.
5764 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
5766 2014-01-25 Walter Lee <walt@tilera.com>
5768 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
5769 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
5770 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
5772 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5774 * config/mips/constraints.md (kl): Delete.
5775 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
5776 define expands, using...
5777 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
5778 instructions for MIPS16.
5779 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
5780 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
5782 2014-01-25 Walter Lee <walt@tilera.com>
5784 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
5788 2014-01-25 Walter Lee <walt@tilera.com>
5790 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
5791 (TARGET_EXPAND_TO_RTL_HOOK): Define.
5793 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5795 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
5798 2014-01-25 Jakub Jelinek <jakub@redhat.com>
5800 * print-rtl.c (in_call_function_usage): New var.
5801 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
5802 EXPR_LIST mode as mode and not as reg note name.
5805 * cfgloopmanip.c (copy_loop_info): If
5806 loop->warned_aggressive_loop_optimizations, make sure
5807 the flag is set in target loop too.
5809 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
5811 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
5813 * builtins.def: Likewise.
5814 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
5815 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
5816 * ira.c (ira_setup_eliminable_regset): Likewise.
5817 * omp-low.c (gate_expand_omp): Likewise.
5818 (execute_lower_omp): Likewise.
5819 (diagnose_sb_0): Likewise.
5820 (gate_diagnose_omp_blocks): Likewise.
5821 (simd_clone_clauses_extract): Likewise.
5824 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5826 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
5827 correction for little endian...
5828 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
5831 2014-01-24 Jeff Law <law@redhat.com>
5833 PR tree-optimization/59919
5834 * tree-vrp.c (find_assert_locations_1): Do not register asserts
5835 for non-returning calls.
5837 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
5839 * common/config/aarch64/aarch64-common.c
5840 (aarch64_rewrite_mcpu): Handle multiple names.
5841 * config/aarch64/aarch64.h
5842 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5844 2014-01-24 Dodji Seketeli <dodji@redhat.com>
5846 * input.c (add_file_to_cache_tab): Handle the case where fopen
5849 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
5852 * config/i386/i386.md (pushsf splitter): Get stack adjustment
5853 from push operand if code of push isn't PRE_DEC.
5855 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
5858 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
5859 -mquad-memory-atomic. Update -mquad-memory documentation to say
5860 it is only used for non-atomic loads/stores.
5862 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
5863 -mquad-memory or -mquad-memory-atomic switches.
5865 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
5866 -mquad-memory-atomic to ISA 2.07 support.
5868 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
5869 to separate support of normal quad word memory operations (ldq, stq)
5870 from the atomic quad word memory operations.
5872 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5873 support to separate non-atomic quad word operations from atomic
5874 quad word operations. Disable non-atomic quad word operations in
5875 little endian mode so that we don't have to swap words after the
5876 load and before the store.
5877 (quad_load_store_p): Add comment about atomic quad word support.
5878 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
5879 options printed with -mdebug=reg.
5881 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
5882 -mquad-memory-atomic as the test for whether we have quad word
5883 atomic instructions.
5884 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
5885 or -mp8-vector are used, allow byte/half-word atomic operations.
5887 * config/rs6000/sync.md (load_lockedti): Insure that the address
5888 is a proper indexed or indirect address for the lqarx instruction.
5889 On little endian systems, swap the hi/lo registers after the lqarx
5891 (load_lockedpti): Use indexed_or_indirect_operand predicate to
5892 insure the address is valid for the lqarx instruction.
5893 (store_conditionalti): Insure that the address is a proper indexed
5894 or indirect address for the stqcrx. instruction. On little endian
5895 systems, swap the hi/lo registers before doing the stqcrx.
5897 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
5898 insure the address is valid for the stqcrx. instruction.
5900 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
5901 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
5902 type of quad memory support is available.
5904 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
5907 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
5908 there is a danger of looping.
5910 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
5912 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5913 force flag_ira_loop_pressure if set via command line.
5915 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5917 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
5918 (ashr_simd): New builtin handling DI mode.
5919 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
5920 (aarch64_sshr_simddi): New match pattern.
5921 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
5922 (vshrd_n_s64): Likewise.
5923 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
5925 2014-01-23 Nick Clifton <nickc@redhat.com>
5927 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
5928 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
5929 favour of mcu specific scripts.
5930 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
5933 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5934 Alex Velenko <Alex.Velenko@arm.com>
5936 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
5937 (vaddv_s16): Likewise.
5938 (vaddv_s32): Likewise.
5939 (vaddv_u8): Likewise.
5940 (vaddv_u16): Likewise.
5941 (vaddv_u32): Likewise.
5942 (vaddvq_s8): Likewise.
5943 (vaddvq_s16): Likewise.
5944 (vaddvq_s32): Likewise.
5945 (vaddvq_s64): Likewise.
5946 (vaddvq_u8): Likewise.
5947 (vaddvq_u16): Likewise.
5948 (vaddvq_u32): Likewise.
5949 (vaddvq_u64): Likewise.
5950 (vaddv_f32): Likewise.
5951 (vaddvq_f32): Likewise.
5952 (vaddvq_f64): Likewise.
5953 (vmaxv_f32): Likewise.
5954 (vmaxv_s8): Likewise.
5955 (vmaxv_s16): Likewise.
5956 (vmaxv_s32): Likewise.
5957 (vmaxv_u8): Likewise.
5958 (vmaxv_u16): Likewise.
5959 (vmaxv_u32): Likewise.
5960 (vmaxvq_f32): Likewise.
5961 (vmaxvq_f64): Likewise.
5962 (vmaxvq_s8): Likewise.
5963 (vmaxvq_s16): Likewise.
5964 (vmaxvq_s32): Likewise.
5965 (vmaxvq_u8): Likewise.
5966 (vmaxvq_u16): Likewise.
5967 (vmaxvq_u32): Likewise.
5968 (vmaxnmv_f32): Likewise.
5969 (vmaxnmvq_f32): Likewise.
5970 (vmaxnmvq_f64): Likewise.
5971 (vminv_f32): Likewise.
5972 (vminv_s8): Likewise.
5973 (vminv_s16): Likewise.
5974 (vminv_s32): Likewise.
5975 (vminv_u8): Likewise.
5976 (vminv_u16): Likewise.
5977 (vminv_u32): Likewise.
5978 (vminvq_f32): Likewise.
5979 (vminvq_f64): Likewise.
5980 (vminvq_s8): Likewise.
5981 (vminvq_s16): Likewise.
5982 (vminvq_s32): Likewise.
5983 (vminvq_u8): Likewise.
5984 (vminvq_u16): Likewise.
5985 (vminvq_u32): Likewise.
5986 (vminnmv_f32): Likewise.
5987 (vminnmvq_f32): Likewise.
5988 (vminnmvq_f64): Likewise.
5990 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5992 * config/aarch64/aarch64-simd.md
5993 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
5994 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
5995 (*aarch64_mul3_elt<mode>): Likewise.
5996 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
5997 (*aarch64_mul3_elt_to_64v2df): Likewise.
5998 (*aarch64_mla_elt<mode>): Likewise.
5999 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
6000 (*aarch64_mls_elt<mode>): Likewise.
6001 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
6002 (*aarch64_fma4_elt<mode>): Likewise.
6003 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
6004 (*aarch64_fma4_elt_to_64v2df): Likewise.
6005 (*aarch64_fnma4_elt<mode>): Likewise.
6006 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6007 (*aarch64_fnma4_elt_to_64v2df): Likewise.
6008 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
6009 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
6010 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6011 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6012 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6013 (aarch64_sqdmull_lane<mode>_internal): Likewise.
6014 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6016 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
6018 * config/aarch64/aarch64-simd.md
6019 (aarch64_be_checked_get_lane<mode>): New define_expand.
6020 * config/aarch64/aarch64-simd-builtins.def
6021 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
6022 New builtin definition.
6023 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
6024 Use new safe be builtin.
6026 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
6028 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
6030 (aarch64_be_st1<mode>): Likewise.
6031 (aarch_ld1<VALL:mode>): Define_expand modified.
6032 (aarch_st1<VALL:mode>): Likewise.
6033 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
6034 (UNSPEC_ST1): Likewise.
6036 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
6038 * config/microblaze/microblaze.md: Add trap insn and attribute
6040 2014-01-23 Dodji Seketeli <dodji@redhat.com>
6042 PR preprocessor/58580
6043 * input.h (location_get_source_line): Take an additional line_size
6045 (void diagnostics_file_cache_fini): Declare new function.
6046 * input.c (struct fcache): New type.
6047 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
6048 New static constants.
6049 (diagnostic_file_cache_init, total_lines_num)
6050 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
6051 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
6052 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
6053 (get_next_line, read_next_line, goto_next_line, read_line_num):
6054 New static function definitions.
6055 (diagnostic_file_cache_fini): New function.
6056 (location_get_source_line): Take an additional output line_len
6057 parameter. Re-write using lookup_or_add_file_to_cache_tab and
6059 * diagnostic.c (diagnostic_finish): Call
6060 diagnostic_file_cache_fini.
6061 (adjust_line): Take an additional input parameter for the length
6062 of the line, rather than calculating it with strlen.
6063 (diagnostic_show_locus): Adjust the use of
6064 location_get_source_line and adjust_line with respect to their new
6065 signature. While displaying a line now, do not stop at the first
6066 null byte. Rather, display the zero byte as a space and keep
6067 going until we reach the size of the line.
6068 * Makefile.in: Add vec.o to OBJS-libcommon
6070 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
6071 Ilya Tocar <ilya.tocar@intel.com>
6073 * config/i386/avx512fintrin.h (_mm512_kmov): New.
6074 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
6075 (__builtin_ia32_kmov16): Ditto.
6076 * config/i386/i386.md (UNSPEC_KMOV): New.
6079 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
6081 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
6082 (_mm512_storeu_si512): Ditto.
6084 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
6087 * rtl.h (get_referenced_operands): Declare.
6088 * recog.c (get_referenced_operands): New function.
6089 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
6090 operands have been referenced when recording LO_SUM references.
6092 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
6094 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
6096 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
6098 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
6099 Enable for generic and recent AMD targets.
6101 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
6103 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
6104 ARG_SIZE note when adjustment was eliminated.
6106 2014-01-22 Jeff Law <law@redhat.com>
6108 PR tree-optimization/59597
6109 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
6110 in file. Accept new argument REGISTERING and use it to modify
6111 dump output appropriately.
6112 (register_jump_thread): Corresponding changes.
6113 (mark_threaded_blocks): Reinstate code to cancel unprofitable
6114 thread paths involving joiner blocks. Add code to dump cancelled
6115 jump threading paths.
6117 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
6119 PR rtl-optimization/59477
6120 * lra-constraints.c (inherit_in_ebb): Process call for living hard
6121 regs. Update reloads_num and potential_reload_hard_regs for all insns.
6123 2014-01-22 Tom Tromey <tromey@redhat.com>
6125 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
6127 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
6129 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
6131 PR rtl-optimization/59896
6132 * lra-constraints.c (process_alt_operands): Check unused note for
6133 matched operands of insn with no output reloads.
6135 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
6137 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
6138 (mips_move_from_gpr_cost): Likewise.
6140 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
6142 PR rtl-optimization/59858
6143 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
6144 ira_class_hard_regs_num.
6145 (process_alt_operands): Increase reject for dying matched operand.
6147 2014-01-21 Jakub Jelinek <jakub@redhat.com>
6150 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
6151 smaller than size, perform several stores or loads and stores
6152 at dst + count - size to store or copy all of size bytes, rather
6153 than just last modesize bytes.
6155 2014-01-20 DJ Delorie <dj@redhat.com>
6157 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
6158 that CLOBBERs are REGs before propogating their values.
6160 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
6163 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
6164 (cgraph_inline_failed_type): New function.
6165 * cgraph.h (DEFCIFCODE): Add type.
6166 (cgraph_inline_failed_type_t): New enum.
6167 (cgraph_inline_failed_type): New prototype.
6168 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
6169 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
6170 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
6171 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
6172 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
6173 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
6174 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
6175 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
6176 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
6177 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
6178 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
6179 OPTIMIZATION_MISMATCH.
6180 * tree-inline.c (expand_call_inline): Emit errors during
6181 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
6183 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
6186 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
6187 mode attribute in insn output.
6189 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
6191 * output.h (output_constant): Delete.
6192 * varasm.c (output_constant): Make private.
6194 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
6196 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
6198 2014-01-20 Jakub Jelinek <jakub@redhat.com>
6201 * tree.h (fold_builtin_strcat): New prototype.
6202 * builtins.c (fold_builtin_strcat): No longer static. Add len
6203 argument, if non-NULL, don't call c_strlen. Optimize
6204 directly into __builtin_memcpy instead of __builtin_strcpy.
6205 (fold_builtin_2): Adjust fold_builtin_strcat caller.
6206 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
6208 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
6210 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6211 for SImode_address_operand operands, having only a REG argument.
6213 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
6215 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
6216 loader name using mbig-endian.
6217 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
6219 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
6221 * doc/invoke.texi (-march): Clarify documentation for AArch64.
6225 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
6227 * config/aarch64/aarch64-protos.h
6228 (aarch64_cannot_change_mode_class_ptr): Declare.
6229 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
6230 aarch64_cannot_change_mode_class_ptr): New.
6231 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
6232 backend hook aarch64_cannot_change_mode_class.
6234 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
6236 * common/config/aarch64/aarch64-common.c
6237 (aarch64_handle_option): Don't handle any option order logic here.
6238 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
6239 selected_cpu, warn on architecture version mismatch.
6240 (aarch64_override_options): Fix parsing order for option strings.
6242 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6243 Iain Sandoe <iain@codesourcery.com>
6246 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
6247 warning. Amend comment to reflect current functionality.
6249 2014-01-20 Richard Biener <rguenther@suse.de>
6252 * builtins.c (fold_builtin_strcat): Remove case better handled
6253 by tree-ssa-strlen.c.
6255 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
6257 * config/aarch64/aarch64.opt
6258 (mcpu, march, mtune): Make case-insensitive.
6260 2014-01-20 Jakub Jelinek <jakub@redhat.com>
6263 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6264 if operands[1] is a REG or ZERO_EXTEND of a REG.
6266 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
6268 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
6270 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
6272 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
6273 long non-pic millicode calls.
6275 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6277 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
6279 2014-01-19 Kito Cheng <kito@0xlab.org>
6281 * builtins.c (expand_movstr): Check movstr expand done or fail.
6283 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6284 H.J. Lu <hongjiu.lu@intel.com>
6287 * config/i386/i386.md (*lea<mode>): Zero-extend return register
6288 to DImode for zero-extended addresses.
6290 2014-01-19 Jakub Jelinek <jakub@redhat.com>
6292 PR rtl-optimization/57763
6293 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
6294 on the new indirect jump_insn and increment LABEL_NUSES (label).
6296 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
6300 * config.gcc (x86_archs): New variable.
6301 (x86_64_archs): Likewise.
6302 (x86_cpus): Likewise.
6303 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
6304 --with-arch/--with-cpu= options.
6305 Support --with-arch=/--with-cpu={nehalem,westmere,
6306 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
6308 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6310 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
6311 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
6313 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6315 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
6317 2014-01-18 Jakub Jelinek <jakub@redhat.com>
6320 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
6321 clear cpp_get_options (parse_in)->warn_unused_macros for
6322 ix86_target_macros_internal with cpp_define.
6324 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
6326 * jump.c (delete_related_insns): Keep (use (insn))s.
6327 * reorg.c (redundant_insn): Check for barriers too.
6329 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6331 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
6333 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
6335 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
6336 call to $$dyncall when TARGET_LONG_CALLS is true.
6338 2014-01-17 Jeff Law <law@redhat.com>
6340 * ree.c (combine_set_extension): Temporarily disable test for
6341 changing number of hard registers.
6343 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
6346 * ipa-inline-analysis.c (inline_free_summary):
6347 Do not free summary of aliases.
6349 2014-01-17 Jakub Jelinek <jakub@redhat.com>
6352 * gimplify.c (gimplify_expr): Use create_tmp_var
6353 instead of create_tmp_var_raw. If cond doesn't have
6354 integral type, don't add the IFN_ANNOTATE builtin at all.
6356 2014-01-17 Martin Jambor <mjambor@suse.cz>
6359 * ipa-cp.c (prev_edge_clone): New variable.
6360 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
6361 Also resize prev_edge_clone vector.
6362 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
6363 (ipcp_edge_removal_hook): New function.
6364 (ipcp_driver): Register ipcp_edge_removal_hook.
6366 2014-01-17 Andrew Pinski <apinski@cavium.com>
6367 Steve Ellcey <sellcey@mips.com>
6370 * config/mips/mips.c (mips_print_operand): Check operand mode instead
6373 2014-01-17 Jeff Law <law@redhat.com>
6376 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
6377 so that pass_ccp runs first.
6379 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6381 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
6382 (ix86_adjust_cost): Use !TARGET_XXX.
6383 (do_reorder_for_imul): Likewise.
6384 (swap_top_of_ready_list): Likewise.
6385 (ix86_sched_reorder): Likewise.
6387 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6389 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6390 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
6391 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
6392 (intel_memset): New. Duplicate slm_memset.
6393 (intel_cost): New. Duplicate slm_cost.
6394 (m_INTEL): New macro.
6395 (processor_target_table): Add "intel".
6396 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
6397 with PROCESSOR_INTEL for "intel".
6398 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
6399 PROCESSOR_SILVERMONT.
6400 (ix86_issue_rate): Likewise.
6401 (ix86_adjust_cost): Likewise.
6402 (ia32_multipass_dfa_lookahead): Likewise.
6403 (swap_top_of_ready_list): Likewise.
6404 (ix86_sched_reorder): Likewise.
6405 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
6406 instead of TARGET_OPT_AGU.
6407 * config/i386/i386.h (TARGET_INTEL): New.
6408 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
6409 (processor_type): Add PROCESSOR_INTEL.
6410 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
6411 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
6413 2014-01-17 Marek Polacek <polacek@redhat.com>
6416 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
6419 2014-01-17 Richard Biener <rguenther@suse.de>
6421 PR tree-optimization/46590
6422 * opts.c (default_options_table): Add entries for
6423 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
6424 all enabled at -O1 but not for -Og.
6425 * common.opt (fbranch-count-reg): Remove Init(1).
6426 (fmove-loop-invariants): Likewise.
6427 (ftree-pta): Likewise.
6429 2014-01-17 Jakub Jelinek <jakub@redhat.com>
6431 * config/i386/i386.c (ix86_data_alignment): For compatibility with
6432 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
6433 decls to at least the GCC 4.8 used alignments.
6436 * tree-nested.c (convert_nonlocal_reference_stmt,
6437 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
6438 of GIMPLE_BIND stmts, adjust associated decls.
6440 2014-01-17 Richard Biener <rguenther@suse.de>
6442 PR tree-optimization/46590
6443 * vec.h (vec<>::bseach): New member function implementing
6444 binary search according to C89 bsearch.
6445 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
6446 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
6447 bitmap pointer again. Make accesses_in_loop a flat array.
6448 (mem_ref_obstack): New global.
6449 (outermost_indep_loop): Adjust for mem_ref->stored changes.
6450 (mark_ref_stored): Likewise.
6451 (ref_indep_loop_p_2): Likewise.
6452 (set_ref_stored_in_loop): New helper function.
6453 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
6454 (memref_free): Adjust.
6455 (record_mem_ref_loc): Simplify.
6456 (gather_mem_refs_stmt): Adjust.
6457 (sort_locs_in_loop_postorder_cmp): New function.
6458 (analyze_memory_references): Sort accesses_in_loop after
6459 loop postorder number.
6460 (find_ref_loc_in_loop_cmp): New function.
6461 (for_all_locs_in_loop): Find relevant cluster of locs in
6462 accesses_in_loop and iterate without recursion.
6463 (execute_sm): Avoid uninit warning.
6464 (struct ref_always_accessed): Simplify.
6465 (ref_always_accessed::operator ()): Likewise.
6466 (ref_always_accessed_p): Likewise.
6467 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
6468 loop postorder numbers here.
6469 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
6472 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
6475 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
6476 on decls for which assemble_alias has been called.
6478 2014-01-17 Nick Clifton <nickc@redhat.com>
6480 * config/msp430/msp430.opt: (mcpu): New option.
6481 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
6482 (msp430_option_override): Parse target_cpu. If the MCU name
6483 matches a generic string, clear target_mcu.
6484 (msp430_attr): Allow numeric interrupt values up to 63.
6485 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
6486 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
6488 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
6490 * config/msp430/msp430.md (popm): Use %J rather than %I.
6491 (addsi3): Use msp430_nonimmediate_operand for operand 2.
6492 (addhi_cy_i): Use immediate_operand for operand 2.
6493 * doc/invoke.texi: Document -mcpu option.
6495 2014-01-17 Richard Biener <rguenther@suse.de>
6497 PR rtl-optimization/38518
6498 * df.h (df_analyze_loop): Declare.
6499 * df-core.c: Include cfgloop.h.
6500 (df_analyze_1): Split out main part of df_analyze.
6501 (df_analyze): Adjust.
6502 (loop_inverted_post_order_compute): New function.
6503 (loop_post_order_compute): Likewise.
6504 (df_analyze_loop): New function avoiding whole-function
6506 * loop-invariant.c (find_defs): Use df_analyze_loop.
6507 (find_invariants): Adjust.
6508 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
6510 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
6512 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
6513 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
6515 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
6517 * ipa-ref.c (ipa_remove_stmt_references): Fix references
6518 traversal when removing references.
6520 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
6523 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
6525 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
6528 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
6529 pushing a reload for an autoinc when we had previously reloaded an
6530 inner part of the address.
6532 2014-01-16 Jakub Jelinek <jakub@redhat.com>
6534 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
6536 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
6537 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
6538 when not giving up or versioning for alias only because of
6540 (vect_analyze_data_ref_dependences): Set to true.
6541 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
6543 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
6544 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
6548 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
6551 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
6552 operand 0 predicate for gathers, use a new pseudo as subtarget.
6554 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6557 * lra-constraints.c (process_alt_operands): Add printing debug info.
6558 Check absence of input/output reloads for matched operands too.
6560 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6562 PR rtl-optimization/59835
6563 * ira.c (ira_init_register_move_cost): Increase cost for
6566 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
6568 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
6570 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
6573 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
6574 non-register objects. Use gen_(high/low)part more consistently.
6577 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
6580 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
6581 endian support, remove tests for WORDS_BIG_ENDIAN.
6582 (p8_mfvsrd_3_<mode>): Likewise.
6583 (reload_gpr_from_vsx<mode>): Likewise.
6584 (reload_gpr_from_vsxsf): Likewise.
6585 (p8_mfvsrd_4_disf): Likewise.
6587 2014-01-16 Richard Biener <rguenther@suse.de>
6589 PR rtl-optimization/46590
6590 * lcm.c (compute_antinout_edge): Use postorder iteration.
6591 (compute_laterin): Use inverted postorder iteration.
6593 2014-01-16 Nick Clifton <nickc@redhat.com>
6596 * varasm.c (output_constant): Return the number of bytes actually
6598 (output_constructor_array_range): Update the field size with the
6599 number of bytes emitted by output_constant.
6600 (output_constructor_regular_field): Likewise. Also do not
6601 complain if the total number of bytes emitted is now greater
6602 than the expected fieldpos.
6603 * output.h (output_constant): Update prototype and descriptive comment.
6605 2014-01-16 Marek Polacek <polacek@redhat.com>
6608 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
6609 it is error_mark_node.
6611 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
6613 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
6614 VALID_AVX256_REG_OR_OI_MODE.
6616 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
6618 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
6619 current procedure should be profiled.
6621 2014-01-15 Andrew Pinski <apinski@cavium.com>
6623 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
6624 of moving from/to the STACK_REG register class.
6626 2014-01-15 Richard Henderson <rth@redhat.com>
6629 * reginfo.c (global_regs_decl): Globalize.
6630 * rtl.h (global_regs_decl): Declare.
6631 * ira.c (do_reload): Diagnose frame_pointer_needed and it
6632 reserved via global_regs.
6634 2014-01-15 Teresa Johnson <tejohnson@google.com>
6636 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
6638 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6640 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
6641 and vmulosh rather than call gen_vec_widen_smult_*.
6642 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
6643 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
6644 (vec_widen_smult_even_v16qi): Likewise.
6645 (vec_widen_umult_even_v8hi): Likewise.
6646 (vec_widen_smult_even_v8hi): Likewise.
6647 (vec_widen_umult_odd_v16qi): Likewise.
6648 (vec_widen_smult_odd_v16qi): Likewise.
6649 (vec_widen_umult_odd_v8hi): Likewise.
6650 (vec_widen_smult_odd_v8hi): Likewise.
6651 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
6652 vmuloub rather than call gen_vec_widen_umult_*.
6653 (vec_widen_umult_lo_v16qi): Likewise.
6654 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
6655 vmulosb rather than call gen_vec_widen_smult_*.
6656 (vec_widen_smult_lo_v16qi): Likewise.
6657 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
6658 rather than call gen_vec_widen_umult_*.
6659 (vec_widen_umult_lo_v8hi): Likewise.
6660 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
6661 rather than call gen_vec_widen_smult_*.
6662 (vec_widen_smult_lo_v8hi): Likewise.
6664 2014-01-15 Jeff Law <law@redhat.com>
6666 PR tree-optimization/59747
6667 * ree.c (find_and_remove_re): Properly handle case where a second
6668 eliminated extension requires widening a copy created for elimination
6669 of a prior extension.
6670 (combine_set_extension): Ensure that the number of hard regs needed
6671 for a destination register does not change when we widen it.
6673 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
6675 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
6676 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
6677 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
6678 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
6679 (avr-*-rtems*): Likewise.
6680 (bfin*-rtems*): Likewise.
6681 (moxie-*-rtems*): Likewise.
6682 (h8300-*-rtems*): Likewise.
6683 (i[34567]86-*-rtems*): Likewise.
6684 (lm32-*-rtems*): Likewise.
6685 (m32r-*-rtems*): Likewise.
6686 (m68k-*-rtems*): Likewise.
6687 (microblaze*-*-rtems*): Likewise.
6688 (mips*-*-rtems*): Likewise.
6689 (powerpc-*-rtems*): Likewise.
6690 (sh-*-rtems*): Likewise.
6691 (sparc-*-rtems*): Likewise.
6692 (sparc64-*-rtems*): Likewise.
6693 (v850-*-rtems*): Likewise.
6694 (m32c-*-rtems*): Likewise.
6696 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
6698 PR rtl-optimization/59511
6699 * ira.c (ira_init_register_move_cost): Use memory costs for some
6700 cases of register move cost calculations.
6701 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
6702 instead of BB frequency.
6703 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
6704 * lra-assigns.c (find_hard_regno_for): Ditto.
6706 2014-01-15 Richard Biener <rguenther@suse.de>
6708 PR tree-optimization/59822
6709 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
6710 (vectorizable_load): Use it to hoist defs of uses of invariant
6711 loads out of the loop.
6713 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
6714 Kugan Vivekanandarajah <kuganv@linaro.org>
6717 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
6720 2014-01-15 Richard Biener <rguenther@suse.de>
6722 PR rtl-optimization/59802
6723 * lcm.c (compute_available): Use inverted postorder to seed
6724 the initial worklist.
6726 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6729 * config/s390/s390.c (s390_preferred_reload_class): Don't return
6730 ADDR_REGS for invalid symrefs in non-PIC code.
6732 2014-01-15 Jakub Jelinek <jakub@redhat.com>
6735 * builtins.c (determine_block_size): Initialize *probable_max_size
6736 even if len_rtx is CONST_INT.
6738 2014-01-14 Andrew Pinski <apinski@cavium.com>
6740 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
6741 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
6742 (cortexa53_tunings): Likewise.
6743 (aarch64_sched_issue_rate): New function.
6744 (TARGET_SCHED_ISSUE_RATE): Define.
6746 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6748 * ira-costs.c (find_costs_and_classes): Add missed
6749 ira_init_register_move_cost_if_necessary.
6751 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6754 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
6756 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
6759 * config/i386/i386.c (type_natural_mode): Add a bool parameter
6760 to indicate if type is used for function return value. Warn ABI
6761 change if the vector mode isn't available for function return value.
6762 (ix86_function_arg_advance): Pass false to type_natural_mode.
6763 (ix86_function_arg): Likewise.
6764 (ix86_gimplify_va_arg): Likewise.
6765 (function_arg_32): Don't warn ABI change.
6766 (ix86_function_value): Pass true to type_natural_mode.
6767 (ix86_return_in_memory): Likewise.
6768 (ix86_struct_value_rtx): Removed.
6769 (TARGET_STRUCT_VALUE_RTX): Likewise.
6771 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
6773 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
6774 converting a conditional jump into a conditional return.
6776 2014-01-14 Richard Biener <rguenther@suse.de>
6778 PR tree-optimization/58921
6779 PR tree-optimization/59006
6780 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
6781 hoisting invariant stmts.
6782 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
6783 invariant loads on the preheader edge if possible.
6785 2014-01-14 Joey Ye <joey.ye@arm.com>
6787 * doc/plugin.texi (Building GCC plugins): Update to C++.
6789 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
6791 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
6792 (_mm_rcp28_round_ss): Ditto.
6793 (_mm_rsqrt28_round_sd): Ditto.
6794 (_mm_rsqrt28_round_ss): Ditto.
6795 (_mm_rcp28_sd): Ditto.
6796 (_mm_rcp28_ss): Ditto.
6797 (_mm_rsqrt28_sd): Ditto.
6798 (_mm_rsqrt28_ss): Ditto.
6799 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
6800 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
6801 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
6802 (IX86_BUILTIN_RCP28SD): Ditto.
6803 (IX86_BUILTIN_RCP28SS): Ditto.
6804 (IX86_BUILTIN_RSQRT28SD): Ditto.
6805 (IX86_BUILTIN_RSQRT28SS): Ditto.
6806 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
6807 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
6808 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
6809 (ix86_expand_special_args_builtin): Expand new FTYPE.
6810 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
6811 (srcp14<mode>): Make insn unary.
6812 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
6813 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
6814 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
6815 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
6816 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
6817 Fix rounding: make it SAE only.
6818 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6820 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6822 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
6823 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
6824 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
6825 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
6826 (round_saeonly_mask_scalar_operand4): Ditto.
6827 (round_saeonly_mask_scalar_op3): Ditto.
6828 (round_saeonly_mask_scalar_op4): Ditto.
6830 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6832 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6833 Implement -maltivec=be for vec_insert and vec_extract.
6835 2014-01-10 DJ Delorie <dj@redhat.com>
6837 * config/msp430/msp430.md (call_internal): Don't allow memory
6838 references with SP as the base register.
6839 (call_value_internal): Likewise.
6840 * config/msp430/constraints.md (Yc): New. For memory references
6841 that don't use SP as a base register.
6843 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
6844 "an integer without a # prefix"
6845 * config/msp430/msp430.md (epilogue_helper): Use it.
6847 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6850 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
6851 AVX512F gather builtins.
6852 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
6853 on gather decls with INTEGER_TYPE masktype.
6854 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
6855 directly into the builtin rather than hoisting it before loop.
6857 PR tree-optimization/59387
6858 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
6859 (scev_const_prop): If folded_casts and type has undefined overflow,
6860 use force_gimple_operand instead of force_gimple_operand_gsi and
6861 for each added stmt if it is assign with
6862 arith_code_with_undefined_signed_overflow, call
6863 rewrite_to_defined_overflow.
6864 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
6865 gimple-fold.h instead.
6866 (arith_code_with_undefined_signed_overflow,
6867 rewrite_to_defined_overflow): Moved to ...
6868 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
6869 rewrite_to_defined_overflow): ... here. No longer static.
6870 Include gimplify-me.h.
6871 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
6872 rewrite_to_defined_overflow): New prototypes.
6874 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6876 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
6878 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
6880 * builtins.c (get_object_alignment_2): Minor tweak.
6881 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
6883 2014-01-13 Christian Bruel <christian.bruel@st.com>
6885 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
6886 optimized non constant lengths.
6888 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6891 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
6892 load as __atomic_load_N if possible.
6894 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6896 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
6898 (rs6000_expand_builtin): Adjust call.
6900 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6903 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
6904 * config/rs6000/rs6000.c: Include target-globals.h.
6905 (rs6000_set_current_function): Instead of doing target_reinit
6906 unconditionally, use save_target_globals_default_opts and
6907 restore_target_globals.
6909 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
6911 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
6912 (rs6000_expand_builtin): Handle mffs and mtfsf.
6913 (rs6000_init_builtins): Define mffs and mtfsf.
6914 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
6915 (rs6000_mffs): New pattern.
6916 (rs6000_mtfsf): New pattern.
6918 2014-01-11 Bin Cheng <bin.cheng@arm.com>
6920 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
6921 Start narrowing with START. Apply candidate-use pair
6922 and check overall cost in narrowing.
6923 (iv_ca_prune): Pass new argument.
6925 2014-01-10 Jeff Law <law@redhat.com>
6928 * ree.c (combine_reaching_defs): Ensure the defining statement
6929 occurs before the extension when optimizing extensions with
6930 different source and destination hard registers.
6932 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6935 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
6936 vtables into the type inheritance graph.
6938 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6940 PR rtl-optimization/59754
6941 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
6942 modes in the REGNO != REGNO case.
6944 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6946 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
6948 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6950 PR tree-optimization/59745
6951 * tree-predcom.c (tree_predictive_commoning_loop): Call
6952 free_affine_expand_cache if giving up because components is NULL.
6954 * target-globals.c (save_target_globals): Allocate < 4KB structs using
6955 GC in payload of target_globals struct instead of allocating them on
6956 the heap and the larger structs separately using GC.
6957 * target-globals.h (struct target_globals): Make regs, hard_regs,
6958 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
6959 of GTY((skip)) and change type to void *.
6960 (reset_target_globals): Cast loads from those fields to corresponding
6963 2014-01-10 Steve Ellcey <sellcey@mips.com>
6966 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
6967 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
6968 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
6970 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
6973 * aarch64-modes.def (CC_Zmode): New flags mode.
6974 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
6975 represents an equality.
6976 (aarch64_get_condition_code): Handle CC_Zmode.
6977 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
6979 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6981 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
6982 extraction in good case.
6984 2014-01-10 Richard Biener <rguenther@suse.de>
6986 PR tree-optimization/59374
6987 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
6988 checking after SLP discovery. Mark stmts not participating
6989 in any SLP instance properly.
6991 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6993 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
6994 when handling a SET rtx.
6996 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6998 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
6999 (cortex-a57): Likewise.
7000 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
7002 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7004 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
7005 non-iwmmxt builtins.
7007 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
7011 * ipa-devirt.c record_target_from_binfo): Take as argument
7012 stack of binfos and lookup matching one for virtual inheritance.
7013 (possible_polymorphic_call_targets_1): Update.
7015 2014-01-10 Huacai Chen <chenhc@lemote.com>
7017 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
7018 kernel strings for Loongson-2E/2F/3A.
7020 2014-01-10 Jakub Jelinek <jakub@redhat.com>
7023 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
7024 is_gimple_call before calling gimple_call_internal_p.
7026 2014-01-09 Steve Ellcey <sellcey@mips.com>
7028 * Makefile.in (TREE_FLOW_H): Remove.
7029 (TREE_SSA_H): Add file names from tree-flow.h.
7030 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
7031 * tree.h: Remove tree-flow.h reference.
7032 * hash-table.h: Remove tree-flow.h reference.
7033 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
7034 reference with tree-ssa-loop.h.
7036 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7038 * doc/invoke.texi: Add -maltivec={be,le} options, and document
7039 default element-order behavior for -maltivec.
7040 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
7041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
7042 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
7043 when targeting big endian, at least for now.
7044 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
7046 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7049 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
7050 var satisfies use_register_for_decl, just take into account type
7051 alignment, rather than decl alignment.
7053 PR tree-optimization/59622
7054 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
7055 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
7056 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
7057 Don't devirtualize for inplace at all. For targets.length () == 1,
7058 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
7060 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
7062 * config/i386/i386.md (cpu): Remove the unused btver1.
7064 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
7066 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
7068 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7071 * tree-core.h (struct target_globals): New forward declaration.
7072 (struct tree_target_option): Add globals field.
7073 * tree.h (TREE_TARGET_GLOBALS): Define.
7074 (prepare_target_option_nodes_for_pch): New prototype.
7075 * target-globals.h (struct target_globals): Define even if
7077 * tree.c (prepare_target_option_node_for_pch,
7078 prepare_target_option_nodes_for_pch): New functions.
7079 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
7080 * config/i386/i386.c: Include target-globals.h.
7081 (ix86_set_current_function): Instead of doing target_reinit
7082 unconditionally, use save_target_globals_default_opts and
7083 restore_target_globals.
7085 2014-01-09 Richard Biener <rguenther@suse.de>
7087 PR tree-optimization/59715
7088 * tree-cfg.h (split_critical_edges): Declare.
7089 * tree-cfg.c (split_critical_edges): Export.
7090 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
7092 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
7094 * cfgexpand.c (expand_stack_vars): Optionally disable
7095 asan stack protection.
7096 (expand_used_vars): Likewise.
7097 (partition_stack_vars): Likewise.
7098 * asan.c (asan_emit_stack_protection): Optionally disable
7099 after return stack usage.
7100 (instrument_derefs): Optionally disable memory access instrumentation.
7101 (instrument_builtin_call): Likewise.
7102 (instrument_strlen_call): Likewise.
7103 (asan_protect_global): Optionally disable global variables protection.
7104 * doc/invoke.texi: Added doc for new options.
7105 * params.def: Added new options.
7106 * params.h: Likewise.
7108 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7110 PR rtl-optimization/59724
7111 * ifcvt.c (cond_exec_process_if_block): Don't call
7112 flow_find_head_matching_sequence with 0 longest_match.
7113 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
7114 non-active insns if !stop_after.
7115 (try_head_merge_bb): Revert 2014-01-07 changes.
7117 2014-01-08 Jeff Law <law@redhat.com>
7119 * ree.c (get_sub_rtx): New function, extracted from...
7120 (merge_def_and_ext): Here.
7121 (combine_reaching_defs): Use get_sub_rtx.
7123 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
7125 * cgraph.h (varpool_variable_node): Do not choke on null node.
7127 2014-01-08 Catherine Moore <clm@codesourcery.com>
7129 * config/mips/mips.md (simple_return): Attempt to use JRC
7131 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
7133 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
7135 PR rtl-optimization/59137
7136 * reorg.c (steal_delay_list_from_target): Call update_block for
7138 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
7140 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7142 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
7143 two duplicate entries.
7145 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
7148 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
7150 * config/mips/mips.c (mips_truncated_op_cost): New function.
7151 (mips_rtx_costs): Adjust test for BADDU.
7152 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
7154 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
7156 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
7157 (*baddu_si): ...this new pattern.
7159 2014-01-08 Jakub Jelinek <jakub@redhat.com>
7162 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
7164 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
7167 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
7169 (expand_expr, expand_normal): Adjust.
7170 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
7171 inner_reference_p. Use inner_reference_p to expand inner references.
7172 (store_expr): Adjust.
7173 * cfgexpand.c (expand_call_stmt): Adjust.
7175 2014-01-08 Rong Xu <xur@google.com>
7177 * gcov-io.c (gcov_var): Move from gcov-io.h.
7178 (gcov_position): Ditto.
7179 (gcov_is_error): Ditto.
7180 (gcov_rewrite): Ditto.
7181 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
7182 only part to libgcc/libgcov.h.
7184 2014-01-08 Marek Polacek <polacek@redhat.com>
7187 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
7189 2014-01-08 Marek Polacek <polacek@redhat.com>
7192 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
7194 2014-01-08 Jakub Jelinek <jakub@redhat.com>
7196 PR rtl-optimization/59649
7197 * stor-layout.c (get_mode_bounds): For BImode return
7198 0 and STORE_FLAG_VALUE.
7200 2014-01-08 Richard Biener <rguenther@suse.de>
7203 * gimple.h (is_gimple_builtin_call): Remove.
7204 (gimple_builtin_call_types_compatible_p): New.
7205 (gimple_call_builtin_p): New overload.
7206 * gimple.c (is_gimple_builtin_call): Remove.
7207 (validate_call): Rename to ...
7208 (gimple_builtin_call_types_compatible_p): ... this and export. Also
7210 (validate_type): New static function.
7211 (gimple_call_builtin_p): New overload and adjust.
7212 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
7213 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
7214 (gimple_fold_stmt_to_constant_1): Likewise.
7215 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
7217 2014-01-08 Richard Biener <rguenther@suse.de>
7220 * gimplify.c (gimplify_expr): Gimplify register-register type
7221 VIEW_CONVERT_EXPRs to separate stmts.
7223 2014-01-07 Jeff Law <law@redhat.com>
7226 * ree.c (combine_set_extension): Handle case where source
7227 and destination registers in an extension insn are different.
7228 (combine_reaching_defs): Allow source and destination registers
7229 in extension to be different under limited circumstances.
7230 (add_removable_extension): Remove restriction that the
7231 source and destination registers in the extension are the same.
7232 (find_and_remove_re): Emit a copy from the extension's
7233 destination to its source after the defining insn if
7234 the source and destination registers are different.
7237 * ifcvt.c (merge_if_block): If we are merging a block with more than
7238 one successor with a block with no successors, remove any BARRIER
7239 after the second block.
7241 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
7243 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
7245 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
7248 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
7249 for 14-bit register offsets when INT14_OK_STRICT is false.
7251 2014-01-07 Roland Stigge <stigge@antcom.de>
7252 Michael Meissner <meissner@linux.vnet.ibm.com>
7255 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
7256 Only check TFmode for SPE constants. Don't check TImode or TDmode.
7258 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
7260 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
7263 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
7265 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
7266 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
7267 rtx is const0_rtx or not.
7269 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
7272 * target-globals.c (save_target_globals): Remove this_fn_optab
7274 * toplev.c: Include optabs.h.
7275 (target_reinit): Temporarily restore the global options if another
7276 set of options are in force.
7278 2014-01-07 Jakub Jelinek <jakub@redhat.com>
7280 PR rtl-optimization/58668
7281 * cfgcleanup.c (flow_find_cross_jump): Don't count
7282 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
7283 to determine what is counted.
7284 (flow_find_head_matching_sequence): Use active_insn_p to determine
7286 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
7288 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
7289 determine what is counted.
7291 PR tree-optimization/59643
7292 * tree-predcom.c (split_data_refs_to_components): If one dr is
7293 read and one write, determine_offset fails and the write isn't
7294 in the bad component, just put the read into the bad component.
7296 2014-01-07 Mike Stump <mikestump@comcast.net>
7297 Jakub Jelinek <jakub@redhat.com>
7300 * tree-core.h (struct tree_optimization_option): Change optabs
7301 type from unsigned char * to void *.
7302 * optabs.c (init_tree_optimization_optabs): Adjust
7303 TREE_OPTIMIZATION_OPTABS initialization.
7305 2014-01-06 Jakub Jelinek <jakub@redhat.com>
7308 * config/i386/i386.h (struct machine_function): Add
7309 no_drap_save_restore field.
7310 * config/i386/i386.c (ix86_save_reg): Use
7311 !cfun->machine->no_drap_save_restore instead of
7312 crtl->stack_realign_needed.
7313 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
7314 this function clears frame_pointer_needed. Set
7315 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
7316 and DRAP reg is needed.
7318 2014-01-06 Marek Polacek <polacek@redhat.com>
7321 * doc/implement-c.texi: Mention that other integer types are
7322 permitted as bit-field types in strictly conforming mode.
7324 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
7326 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
7329 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
7331 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
7333 2014-01-06 Martin Jambor <mjambor@suse.cz>
7336 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
7338 * ipa-prop.c (ipa_print_node_params): Fix indentation.
7340 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
7344 * var-tracking.c (add_stores): Preserve the value of the source even if
7345 we don't record the store.
7347 2014-01-06 Terry Guo <terry.guo@arm.com>
7349 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
7351 2014-01-05 Iain Sandoe <iain@codesourcery.com>
7354 * config/darwin.c (darwin_function_section): Adjust return values to
7355 correspond to optimisation changes made in r206070.
7357 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
7359 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
7360 from prefetch_block tune setting.
7361 (nocona_cost): Correct size of prefetch block to 64.
7363 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
7365 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
7366 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
7367 used to save the static chain register in the computation of the offset
7368 from which the FP registers need to be restored.
7370 2014-01-04 Jakub Jelinek <jakub@redhat.com>
7372 PR tree-optimization/59519
7373 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
7374 ICE if get_current_def (current_new_name) is already non-NULL, as long
7375 as it is a phi result of some other phi in *new_exit_bb that has
7378 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
7379 or vmovdqu* for misaligned_operand.
7380 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
7381 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
7382 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
7383 aligned_mem for AVX512F masked aligned load and store builtins and for
7386 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
7388 PR tree-optimization/59651
7389 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
7390 Address range for negative step should be added by TYPE_SIZE_UNIT.
7392 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
7394 * config/m68k/m68k.c (handle_move_double): Handle pushes with
7395 overlapping registers also for registers other than the stack pointer.
7397 2014-01-03 Marek Polacek <polacek@redhat.com>
7400 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
7403 2014-01-03 Jakub Jelinek <jakub@redhat.com>
7406 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
7409 * config/i386/i386.md (MODE_SIZE): New mode attribute.
7410 (push splitter): Use <P:MODE_SIZE> instead of
7411 GET_MODE_SIZE (<P:MODE>mode).
7412 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
7413 (mov -1, reg peephole2): Likewise.
7414 * config/i386/sse.md (*mov<mode>_internal,
7415 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
7416 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
7417 *<code><mode>3, *andnot<mode>3<mask_name>,
7418 <mask_codefor><code><mode>3<mask_name>): Likewise.
7419 * config/i386/subst.md (mask_mode512bit_condition,
7420 sd_mask_mode512bit_condition): Likewise.
7422 2014-01-02 Xinliang David Li <davidxl@google.com>
7424 PR tree-optimization/59303
7425 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
7426 (dump_predicates): Better output format.
7427 (pred_equal_p): New function.
7428 (is_neq_relop_p): Ditto.
7429 (is_neq_zero_form_p): Ditto.
7430 (pred_expr_equal_p): Ditto.
7431 (pred_neg_p): Ditto.
7432 (simplify_pred): Ditto.
7433 (simplify_preds_2): Ditto.
7434 (simplify_preds_3): Ditto.
7435 (simplify_preds_4): Ditto.
7436 (simplify_preds): Ditto.
7438 (push_to_worklist): Ditto.
7439 (get_pred_info_from_cmp): Ditto.
7440 (is_degenerated_phi): Ditto.
7441 (normalize_one_pred_1): Ditto.
7442 (normalize_one_pred): Ditto.
7443 (normalize_one_pred_chain): Ditto.
7444 (normalize_preds): Ditto.
7445 (normalize_cond_1): Remove function.
7446 (normalize_cond): Ditto.
7447 (is_gcond_subset_of): Ditto.
7448 (is_subset_of_any): Ditto.
7449 (is_or_set_subset_of): Ditto.
7450 (is_and_set_subset_of): Ditto.
7451 (is_norm_cond_subset_of): Ditto.
7452 (pred_chain_length_cmp): Ditto.
7453 (convert_control_dep_chain_into_preds): Type change.
7454 (find_predicates): Ditto.
7455 (find_def_preds): Ditto.
7456 (destroy_predicates_vecs): Ditto.
7457 (find_matching_predicates_in_rest_chains): Ditto.
7458 (use_pred_not_overlap_with_undef_path_pred): Ditto.
7459 (is_pred_expr_subset): Ditto.
7460 (is_pred_chain_subset_of): Ditto.
7461 (is_included_in): Ditto.
7462 (is_superset_of): Ditto.
7464 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7466 Update copyright years.
7468 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7470 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
7471 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
7472 config/arc/arc.md, config/arc/arc.opt,
7473 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
7474 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
7475 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
7476 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
7477 config/linux-protos.h, config/linux.c, config/winnt-c.c,
7478 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
7479 vtable-verify.c, vtable-verify.h: Use the standard form for the
7482 2014-01-02 Tobias Burnus <burnus@net-b.de>
7484 * gcc.c (process_command): Update copyright notice dates.
7485 * gcov-dump.c: Ditto.
7487 * doc/cpp.texi: Bump @copying's copyright year.
7488 * doc/cppinternals.texi: Ditto.
7489 * doc/gcc.texi: Ditto.
7490 * doc/gccint.texi: Ditto.
7491 * doc/gcov.texi: Ditto.
7492 * doc/install.texi: Ditto.
7493 * doc/invoke.texi: Ditto.
7495 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
7497 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
7499 2014-01-01 Jakub Jelinek <jakub@redhat.com>
7501 * config/i386/sse.md (*mov<mode>_internal): Guard
7502 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
7504 PR rtl-optimization/59647
7505 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
7506 new_rtx into UNSIGNED_FLOAT rtxes.
7508 Copyright (C) 2014 Free Software Foundation, Inc.
7510 Copying and distribution of this file, with or without modification,
7511 are permitted in any medium without royalty provided the copyright
7512 notice and this notice are preserved.