1 2014-04-23 Richard Biener <rguenther@suse.de>
3 * Makefile.in (OBJS): Remove loop-unswitch.o.
4 * loop-unswitch.c: Delete.
5 * tree-pass.h (make_pass_rtl_unswitch): Remove.
6 * passes.def (pass_rtl_unswitch): Likewise.
7 * loop-init.c (gate_rtl_unswitch): Likewise.
8 (rtl_unswitch): Likewise.
9 (pass_data_rtl_unswitch): Likewise.
10 (pass_rtl_unswitch): Likewise.
11 (make_pass_rtl_unswitch): Likewise.
12 * rtl.h (reversed_condition): Likewise.
13 (compare_and_jump_seq): Likewise.
14 * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
16 * loop-unroll.c (compare_and_jump_seq): Likewise.
18 2014-04-23 Richard Biener <rguenther@suse.de>
20 PR tree-optimization/60903
21 * tree-ssa-loop-im.c (analyze_memory_references): Remove
23 (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
24 loop flags to newly created BBs and edges.
26 2014-04-23 Nick Clifton <nickc@redhat.com>
28 * config/msp430/msp430.c (msp430_handle_option): Move function
30 (msp430_option_override): Simplify mcu and mcpu option handling.
31 (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult. Add
32 support for -mhwmult command line option.
33 (has_32bit_hwmult): Rename to use_32bit_hwmult. Add support for
34 -mhwmult command line option.
35 (msp430_hwmult_enabled): Delete.
36 (msp43o_output_labelref): Add support for -mhwmult command line
38 * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
39 (umulsidi3): Likewise.
40 * config/msp430/msp430.opt (mmcu): Add Report attribute.
41 (mcpu, mlarge, msmall): Likewise.
42 (mhwmult): New option.
43 * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
45 (msp430_is_f5_mcu): Remove prototype.
46 (msp430_use_f5_series_hwmult): Add prototype.
47 * config/msp430/msp430-opts.h: New file.
48 * common/config/msp430: New directory.
49 * common/config/msp430/msp430-common.c: New file.
50 * config.gcc (msp430): Remove target_has_targetm_common.
51 * doc/invoke.texi: Document -mhwmult command line option.
53 2014-04-23 Nick Clifton <nickc@redhat.com>
55 * config/i386/cygwin.h (ENDFILE_SPEC): Include
56 default-manifest.o if it can be found in the search path.
57 * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
59 2014-04-23 Terry Guo <terry.guo@arm.com>
61 * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
63 2014-04-23 Richard Biener <rguenther@suse.de>
66 * tree-inline.c (declare_return_variable): Use mark_addressable.
68 2014-04-23 Richard Biener <rguenther@suse.de>
71 * loop-init.c (loop_optimizer_init): Make sure to apply
72 LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
74 2014-04-22 Jakub Jelinek <jakub@redhat.com>
77 * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
79 * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
80 if flag_sanitize_undefined_trap_on_error.
81 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
82 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
83 BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
84 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
85 BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
86 BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
87 BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
88 BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
89 BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
90 * ubsan.c (ubsan_instrument_unreachable): Return
91 __builtin_trap () if flag_sanitize_undefined_trap_on_error.
92 (ubsan_expand_null_ifn): Emit __builtin_trap ()
93 if flag_sanitize_undefined_trap_on_error and
94 __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
95 (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
96 instrument_bool_enum_load): Emit __builtin_trap () if
97 flag_sanitize_undefined_trap_on_error and
98 __builtin_handle_*_abort () if !flag_sanitize_recover.
99 * doc/invoke.texi (-fsanitize-recover,
100 -fsanitize-undefined-trap-on-error): Document.
102 2014-04-22 Christian Bruel <christian.bruel@st.com>
104 * config/sh/sh.md (mov<mode>): Replace movQIHI.
105 Force immediates to SImode.
107 2014-04-22 Sandra Loosemore <sandra@codesourcery.com>
109 * config/nios2/nios2.md (UNSPEC_ROUND): New.
111 * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
112 * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
113 * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
114 (nios2_fpu_insn): Add entry for round.
115 (N2FPU_NO_ERRNO_P): Define.
116 (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
118 * doc/invoke.texi (Nios II Options): Document -mcustom-round.
120 2014-04-22 Richard Henderson <rth@redhat.com>
122 * config/aarch64/aarch64 (addti3, subti3): New expanders.
123 (add<GPI>3_compare0): Remove leading * from name.
124 (add<GPI>3_carryin): Likewise.
125 (sub<GPI>3_compare0): Likewise.
126 (sub<GPI>3_carryin): Likewise.
127 (<su_optab>mulditi3): New expander.
128 (multi3): New expander.
129 (madd<GPI>): Remove leading * from name.
131 2014-04-22 Martin Jambor <mjambor@suse.cz>
133 * cgraphclones.c (cgraph_function_versioning): Copy
134 ipa_transforms_to_apply instead of asserting it is empty.
136 2014-04-22 H.J. Lu <hongjiu.lu@intel.com>
139 * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
140 on count_exp to get mode.
142 2014-04-22 Andrew Pinski <apinski@cavium.com>
144 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
145 Handle TLS for ILP32.
146 * config/aarch64/aarch64.md (tlsie_small): Rename to ...
147 (tlsie_small_<mode>): this and handle PTR.
148 (tlsie_small_sidi): New pattern.
149 (tlsle_small): Change to an expand to handle ILP32.
150 (tlsle_small_<mode>): New pattern.
151 (tlsdesc_small): Rename to ...
152 (tlsdesc_small_<mode>): this and handle PTR.
154 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
156 * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
158 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
160 * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
161 (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
162 (aarch64_types_signed_poly_qualifiers): Likewise.
163 (aarch64_types_unsigned_signed_qualifiers): Likewise.
164 (aarch64_types_poly_signed_qualifiers): Likewise.
165 (TYPES_REINTERP_SS): Type macro added.
166 (TYPES_REINTERP_SU): Likewise.
167 (TYPES_REINTERP_SP): Likewise.
168 (TYPES_REINTERP_US): Likewise.
169 (TYPES_REINTERP_PS): Likewise.
170 (aarch64_fold_builtin): New expression folding added.
171 * config/aarch64/aarch64-simd-builtins.def (REINTERP):
172 Declarations removed.
173 (REINTERP_SS): Declarations added.
174 (REINTERP_US): Likewise.
175 (REINTERP_PS): Likewise.
176 (REINTERP_SU): Likewise.
177 (REINTERP_SP): Likewise.
178 * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
179 (vreinterpretq_p8_f64): Likewise.
180 (vreinterpret_p16_f64): Likewise.
181 (vreinterpretq_p16_f64): Likewise.
182 (vreinterpret_f32_f64): Likewise.
183 (vreinterpretq_f32_f64): Likewise.
184 (vreinterpret_f64_f32): Likewise.
185 (vreinterpret_f64_p8): Likewise.
186 (vreinterpret_f64_p16): Likewise.
187 (vreinterpret_f64_s8): Likewise.
188 (vreinterpret_f64_s16): Likewise.
189 (vreinterpret_f64_s32): Likewise.
190 (vreinterpret_f64_s64): Likewise.
191 (vreinterpret_f64_u8): Likewise.
192 (vreinterpret_f64_u16): Likewise.
193 (vreinterpret_f64_u32): Likewise.
194 (vreinterpret_f64_u64): Likewise.
195 (vreinterpretq_f64_f32): Likewise.
196 (vreinterpretq_f64_p8): Likewise.
197 (vreinterpretq_f64_p16): Likewise.
198 (vreinterpretq_f64_s8): Likewise.
199 (vreinterpretq_f64_s16): Likewise.
200 (vreinterpretq_f64_s32): Likewise.
201 (vreinterpretq_f64_s64): Likewise.
202 (vreinterpretq_f64_u8): Likewise.
203 (vreinterpretq_f64_u16): Likewise.
204 (vreinterpretq_f64_u32): Likewise.
205 (vreinterpretq_f64_u64): Likewise.
206 (vreinterpret_s64_f64): Likewise.
207 (vreinterpretq_s64_f64): Likewise.
208 (vreinterpret_u64_f64): Likewise.
209 (vreinterpretq_u64_f64): Likewise.
210 (vreinterpret_s8_f64): Likewise.
211 (vreinterpretq_s8_f64): Likewise.
212 (vreinterpret_s16_f64): Likewise.
213 (vreinterpretq_s16_f64): Likewise.
214 (vreinterpret_s32_f64): Likewise.
215 (vreinterpretq_s32_f64): Likewise.
216 (vreinterpret_u8_f64): Likewise.
217 (vreinterpretq_u8_f64): Likewise.
218 (vreinterpret_u16_f64): Likewise.
219 (vreinterpretq_u16_f64): Likewise.
220 (vreinterpret_u32_f64): Likewise.
221 (vreinterpretq_u32_f64): Likewise.
223 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
225 * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
226 * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
227 (vreinterpret_p8_s8): Likewise.
228 * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
229 (vreinterpret_p8_s16): Likewise.
230 (vreinterpret_p8_s32): Likewise.
231 (vreinterpret_p8_s64): Likewise.
232 (vreinterpret_p8_f32): Likewise.
233 (vreinterpret_p8_u8): Likewise.
234 (vreinterpret_p8_u16): Likewise.
235 (vreinterpret_p8_u32): Likewise.
236 (vreinterpret_p8_u64): Likewise.
237 (vreinterpret_p8_p16): Likewise.
238 (vreinterpretq_p8_s8): Likewise.
239 (vreinterpretq_p8_s16): Likewise.
240 (vreinterpretq_p8_s32): Likewise.
241 (vreinterpretq_p8_s64): Likewise.
242 (vreinterpretq_p8_f32): Likewise.
243 (vreinterpretq_p8_u8): Likewise.
244 (vreinterpretq_p8_u16): Likewise.
245 (vreinterpretq_p8_u32): Likewise.
246 (vreinterpretq_p8_u64): Likewise.
247 (vreinterpretq_p8_p16): Likewise.
248 (vreinterpret_p16_s8): Likewise.
249 (vreinterpret_p16_s16): Likewise.
250 (vreinterpret_p16_s32): Likewise.
251 (vreinterpret_p16_s64): Likewise.
252 (vreinterpret_p16_f32): Likewise.
253 (vreinterpret_p16_u8): Likewise.
254 (vreinterpret_p16_u16): Likewise.
255 (vreinterpret_p16_u32): Likewise.
256 (vreinterpret_p16_u64): Likewise.
257 (vreinterpret_p16_p8): Likewise.
258 (vreinterpretq_p16_s8): Likewise.
259 (vreinterpretq_p16_s16): Likewise.
260 (vreinterpretq_p16_s32): Likewise.
261 (vreinterpretq_p16_s64): Likewise.
262 (vreinterpretq_p16_f32): Likewise.
263 (vreinterpretq_p16_u8): Likewise.
264 (vreinterpretq_p16_u16): Likewise.
265 (vreinterpretq_p16_u32): Likewise.
266 (vreinterpretq_p16_u64): Likewise.
267 (vreinterpretq_p16_p8): Likewise.
268 (vreinterpret_f32_s8): Likewise.
269 (vreinterpret_f32_s16): Likewise.
270 (vreinterpret_f32_s32): Likewise.
271 (vreinterpret_f32_s64): Likewise.
272 (vreinterpret_f32_u8): Likewise.
273 (vreinterpret_f32_u16): Likewise.
274 (vreinterpret_f32_u32): Likewise.
275 (vreinterpret_f32_u64): Likewise.
276 (vreinterpret_f32_p8): Likewise.
277 (vreinterpret_f32_p16): Likewise.
278 (vreinterpretq_f32_s8): Likewise.
279 (vreinterpretq_f32_s16): Likewise.
280 (vreinterpretq_f32_s32): Likewise.
281 (vreinterpretq_f32_s64): Likewise.
282 (vreinterpretq_f32_u8): Likewise.
283 (vreinterpretq_f32_u16): Likewise.
284 (vreinterpretq_f32_u32): Likewise.
285 (vreinterpretq_f32_u64): Likewise.
286 (vreinterpretq_f32_p8): Likewise.
287 (vreinterpretq_f32_p16): Likewise.
288 (vreinterpret_s64_s8): Likewise.
289 (vreinterpret_s64_s16): Likewise.
290 (vreinterpret_s64_s32): Likewise.
291 (vreinterpret_s64_f32): Likewise.
292 (vreinterpret_s64_u8): Likewise.
293 (vreinterpret_s64_u16): Likewise.
294 (vreinterpret_s64_u32): Likewise.
295 (vreinterpret_s64_u64): Likewise.
296 (vreinterpret_s64_p8): Likewise.
297 (vreinterpret_s64_p16): Likewise.
298 (vreinterpretq_s64_s8): Likewise.
299 (vreinterpretq_s64_s16): Likewise.
300 (vreinterpretq_s64_s32): Likewise.
301 (vreinterpretq_s64_f32): Likewise.
302 (vreinterpretq_s64_u8): Likewise.
303 (vreinterpretq_s64_u16): Likewise.
304 (vreinterpretq_s64_u32): Likewise.
305 (vreinterpretq_s64_u64): Likewise.
306 (vreinterpretq_s64_p8): Likewise.
307 (vreinterpretq_s64_p16): Likewise.
308 (vreinterpret_u64_s8): Likewise.
309 (vreinterpret_u64_s16): Likewise.
310 (vreinterpret_u64_s32): Likewise.
311 (vreinterpret_u64_s64): Likewise.
312 (vreinterpret_u64_f32): Likewise.
313 (vreinterpret_u64_u8): Likewise.
314 (vreinterpret_u64_u16): Likewise.
315 (vreinterpret_u64_u32): Likewise.
316 (vreinterpret_u64_p8): Likewise.
317 (vreinterpret_u64_p16): Likewise.
318 (vreinterpretq_u64_s8): Likewise.
319 (vreinterpretq_u64_s16): Likewise.
320 (vreinterpretq_u64_s32): Likewise.
321 (vreinterpretq_u64_s64): Likewise.
322 (vreinterpretq_u64_f32): Likewise.
323 (vreinterpretq_u64_u8): Likewise.
324 (vreinterpretq_u64_u16): Likewise.
325 (vreinterpretq_u64_u32): Likewise.
326 (vreinterpretq_u64_p8): Likewise.
327 (vreinterpretq_u64_p16): Likewise.
328 (vreinterpret_s8_s16): Likewise.
329 (vreinterpret_s8_s32): Likewise.
330 (vreinterpret_s8_s64): Likewise.
331 (vreinterpret_s8_f32): Likewise.
332 (vreinterpret_s8_u8): Likewise.
333 (vreinterpret_s8_u16): Likewise.
334 (vreinterpret_s8_u32): Likewise.
335 (vreinterpret_s8_u64): Likewise.
336 (vreinterpret_s8_p8): Likewise.
337 (vreinterpret_s8_p16): Likewise.
338 (vreinterpretq_s8_s16): Likewise.
339 (vreinterpretq_s8_s32): Likewise.
340 (vreinterpretq_s8_s64): Likewise.
341 (vreinterpretq_s8_f32): Likewise.
342 (vreinterpretq_s8_u8): Likewise.
343 (vreinterpretq_s8_u16): Likewise.
344 (vreinterpretq_s8_u32): Likewise.
345 (vreinterpretq_s8_u64): Likewise.
346 (vreinterpretq_s8_p8): Likewise.
347 (vreinterpretq_s8_p16): Likewise.
348 (vreinterpret_s16_s8): Likewise.
349 (vreinterpret_s16_s32): Likewise.
350 (vreinterpret_s16_s64): Likewise.
351 (vreinterpret_s16_f32): Likewise.
352 (vreinterpret_s16_u8): Likewise.
353 (vreinterpret_s16_u16): Likewise.
354 (vreinterpret_s16_u32): Likewise.
355 (vreinterpret_s16_u64): Likewise.
356 (vreinterpret_s16_p8): Likewise.
357 (vreinterpret_s16_p16): Likewise.
358 (vreinterpretq_s16_s8): Likewise.
359 (vreinterpretq_s16_s32): Likewise.
360 (vreinterpretq_s16_s64): Likewise.
361 (vreinterpretq_s16_f32): Likewise.
362 (vreinterpretq_s16_u8): Likewise.
363 (vreinterpretq_s16_u16): Likewise.
364 (vreinterpretq_s16_u32): Likewise.
365 (vreinterpretq_s16_u64): Likewise.
366 (vreinterpretq_s16_p8): Likewise.
367 (vreinterpretq_s16_p16): Likewise.
368 (vreinterpret_s32_s8): Likewise.
369 (vreinterpret_s32_s16): Likewise.
370 (vreinterpret_s32_s64): Likewise.
371 (vreinterpret_s32_f32): Likewise.
372 (vreinterpret_s32_u8): Likewise.
373 (vreinterpret_s32_u16): Likewise.
374 (vreinterpret_s32_u32): Likewise.
375 (vreinterpret_s32_u64): Likewise.
376 (vreinterpret_s32_p8): Likewise.
377 (vreinterpret_s32_p16): Likewise.
378 (vreinterpretq_s32_s8): Likewise.
379 (vreinterpretq_s32_s16): Likewise.
380 (vreinterpretq_s32_s64): Likewise.
381 (vreinterpretq_s32_f32): Likewise.
382 (vreinterpretq_s32_u8): Likewise.
383 (vreinterpretq_s32_u16): Likewise.
384 (vreinterpretq_s32_u32): Likewise.
385 (vreinterpretq_s32_u64): Likewise.
386 (vreinterpretq_s32_p8): Likewise.
387 (vreinterpretq_s32_p16): Likewise.
388 (vreinterpret_u8_s8): Likewise.
389 (vreinterpret_u8_s16): Likewise.
390 (vreinterpret_u8_s32): Likewise.
391 (vreinterpret_u8_s64): Likewise.
392 (vreinterpret_u8_f32): Likewise.
393 (vreinterpret_u8_u16): Likewise.
394 (vreinterpret_u8_u32): Likewise.
395 (vreinterpret_u8_u64): Likewise.
396 (vreinterpret_u8_p8): Likewise.
397 (vreinterpret_u8_p16): Likewise.
398 (vreinterpretq_u8_s8): Likewise.
399 (vreinterpretq_u8_s16): Likewise.
400 (vreinterpretq_u8_s32): Likewise.
401 (vreinterpretq_u8_s64): Likewise.
402 (vreinterpretq_u8_f32): Likewise.
403 (vreinterpretq_u8_u16): Likewise.
404 (vreinterpretq_u8_u32): Likewise.
405 (vreinterpretq_u8_u64): Likewise.
406 (vreinterpretq_u8_p8): Likewise.
407 (vreinterpretq_u8_p16): Likewise.
408 (vreinterpret_u16_s8): Likewise.
409 (vreinterpret_u16_s16): Likewise.
410 (vreinterpret_u16_s32): Likewise.
411 (vreinterpret_u16_s64): Likewise.
412 (vreinterpret_u16_f32): Likewise.
413 (vreinterpret_u16_u8): Likewise.
414 (vreinterpret_u16_u32): Likewise.
415 (vreinterpret_u16_u64): Likewise.
416 (vreinterpret_u16_p8): Likewise.
417 (vreinterpret_u16_p16): Likewise.
418 (vreinterpretq_u16_s8): Likewise.
419 (vreinterpretq_u16_s16): Likewise.
420 (vreinterpretq_u16_s32): Likewise.
421 (vreinterpretq_u16_s64): Likewise.
422 (vreinterpretq_u16_f32): Likewise.
423 (vreinterpretq_u16_u8): Likewise.
424 (vreinterpretq_u16_u32): Likewise.
425 (vreinterpretq_u16_u64): Likewise.
426 (vreinterpretq_u16_p8): Likewise.
427 (vreinterpretq_u16_p16): Likewise.
428 (vreinterpret_u32_s8): Likewise.
429 (vreinterpret_u32_s16): Likewise.
430 (vreinterpret_u32_s32): Likewise.
431 (vreinterpret_u32_s64): Likewise.
432 (vreinterpret_u32_f32): Likewise.
433 (vreinterpret_u32_u8): Likewise.
434 (vreinterpret_u32_u16): Likewise.
435 (vreinterpret_u32_u64): Likewise.
436 (vreinterpret_u32_p8): Likewise.
437 (vreinterpret_u32_p16): Likewise.
438 (vreinterpretq_u32_s8): Likewise.
439 (vreinterpretq_u32_s16): Likewise.
440 (vreinterpretq_u32_s32): Likewise.
441 (vreinterpretq_u32_s64): Likewise.
442 (vreinterpretq_u32_f32): Likewise.
443 (vreinterpretq_u32_u8): Likewise.
444 (vreinterpretq_u32_u16): Likewise.
445 (vreinterpretq_u32_u64): Likewise.
446 (vreinterpretq_u32_p8): Likewise.
447 (vreinterpretq_u32_p16): Likewise.
449 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
451 * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
453 * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
456 * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
457 (vqnegd_s64): Likewise.
458 (vqabs_s64): Likewise.
459 (vqabsd_s64): Likewise.
461 2014-04-22 Richard Henderson <rth@redhat.com>
463 * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
464 computation to the top of the loop.
466 2014-04-22 Renlin <renlin.li@arm.com>
467 Jiong Wang <jiong.wang@arm.com>
469 * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
470 * config/aarch64/aarch64.c (aarch64_layout_frame)
471 (aarch64_initial_elimination_offset): Likewise.
473 2014-04-22 Marcus Shawcroft <marcus.shawcroft@arm.com>
475 * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
478 2014-04-22 Richard Sandiford <rdsandiford@googlemail.com>
480 * machmode.h (bitwise_mode_for_mode): Declare.
481 * stor-layout.h (bitwise_type_for_mode): Likewise.
482 * stor-layout.c (bitwise_mode_for_mode): New function.
483 (bitwise_type_for_mode): Likewise.
484 * builtins.c (fold_builtin_memory_op): Use it instead of
485 int_mode_for_mode and build_nonstandard_integer_type.
487 2014-04-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
489 * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
490 (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
491 (*-*-solaris2*): Simplify.
492 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
493 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
494 *-*-solaris2.9* handling.
496 * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
498 (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
499 (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
501 (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
502 * configure: Regenerate.
504 * config/i386/sol2-9.h: Remove.
506 * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
507 (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
508 Remove Solaris 9 references.
510 2014-04-22 Vidya Praveen <vidyapraveen@arm.com>
512 * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
513 (floatuns<GPI:mode><GPF:mode>2): Remove.
514 (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
515 and floatuns conversions.
516 (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
517 and floatuns conversions.
518 * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
519 (w1,w2): New mode attributes for inequal width conversions.
521 2014-04-22 Renlin Li <Renlin.Li@arm.com>
523 * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
524 the output asm format.
526 2014-04-22 James Greenhalgh <james.greenhalgh@arm.com>
528 * config/aarch64/aarch64-simd.md
529 (aarch64_cm<optab>di): Always split.
530 (*aarch64_cm<optab>di): New.
531 (aarch64_cmtstdi): Always split.
532 (*aarch64_cmtstdi): New.
534 2014-04-22 Jakub Jelinek <jakub@redhat.com>
536 PR tree-optimization/60823
537 * omp-low.c (ipa_simd_modify_function_body): Go through
538 all SSA_NAMEs and for those refering to vector arguments
539 which are going to be replaced adjust SSA_NAME_VAR and,
540 if it is a default definition, change it into a non-default
541 definition assigned at the beginning of function from new_decl.
542 (ipa_simd_modify_stmt_ops): Rewritten.
543 * tree-dfa.c (set_ssa_default_def): When removing default def,
544 check for NULL loc instead of NULL *loc.
546 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
548 * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
549 restrictions on core registers for DImode values in Thumb2.
551 2014-04-22 Ian Bolton <ian.bolton@arm.com>
553 * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
554 * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
556 2014-04-22 Ian Bolton <ian.bolton@arm.com>
558 * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
559 (*iordi_notzesidi_di): Likewise.
560 (*iordi_notsesidi_di): Likewise.
562 2014-04-22 Ian Bolton <ian.bolton@arm.com>
564 * config/arm/arm-protos.h (tune_params): New struct members.
565 * config/arm/arm.c: Initialise tune_params per processor.
566 (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
567 for speed, based on new tune_params.
569 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
571 * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
573 * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
575 * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
577 * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
578 * config/aarch64/arm_neon.h (vrnd_f64): Added.
579 (vrnda_f64): Likewise.
580 (vrndi_f64): Likewise.
581 (vrndm_f64): Likewise.
582 (vrndn_f64): Likewise.
583 (vrndp_f64): Likewise.
584 (vrndx_f64): Likewise.
586 2014-04-22 Zhenqiang Chen <zhenqiang.chen@linaro.org>
588 * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
589 GET_MODE_SIZE argument is enum machine_mode.
591 2014-04-22 Jakub Jelinek <jakub@redhat.com>
594 * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
595 value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
597 2014-04-22 Lin Zuojian <manjian2006@gmail.com>
600 * asan.c (asan_emit_stack_protection): Force the base to align to
601 appropriate bits if STRICT_ALIGNMENT. Set shadow_mem align to
602 appropriate bits if STRICT_ALIGNMENT.
603 * cfgexpand.c (expand_stack_vars): Set base_align appropriately
605 (expand_used_vars): Leave a space in the stack frame for alignment
608 2014-04-21 David Malcolm <dmalcolm@redhat.com>
610 * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
612 (gimple_store_p): Likewise.
613 (gimple_assign_load_p): Likewise.
614 (gimple_assign_cast_p): Likewise.
615 (gimple_clobber_p): Likewise.
617 * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
618 rather than a gimple.
619 (gimple_assign_cast_p): Likewise.
621 2014-04-21 Michael Meissner <meissner@linux.vnet.ibm.com>
624 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
625 If mode is DDmode and TARGET_E500_DOUBLE allow move.
627 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
628 more debug information for E500 if -mdebug=reg.
630 2014-04-21 Uros Bizjak <ubizjak@gmail.com>
633 * config/i386/i386.c (ix86_expand_builtin)
634 <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
635 register for target RTX.
636 <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
638 2014-04-18 Cong Hou <congh@google.com>
640 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
641 the widen-mult pattern by handling two operands with different sizes,
642 and operands whose size is smaller than half of the result type.
644 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
646 * ipa-inline.h (INLINE_HINT_known_hot): New hint.
647 * ipa-inline-analysis.c (dump_inline_hints): Dump it.
648 (do_estimate_edge_time): Compute it.
649 * ipa-inline.c (want_inline_small_function_p): Bypass
650 INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
652 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
654 * ipa-inline.c (spec_rem): New static variable.
655 (dump_overall_stats): New function.
656 (dump_inline_stats): New function.
658 2014-04-18 Richard Henderson <rth@redhat.com>
660 * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
661 to GET_MODE_SIZE, not a reg_class_t.
663 2014-04-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
665 * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
666 (vsx_xxmrglw_<mode>): Likewise.
668 2014-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
671 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
672 GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
673 (rs6000_init_hard_regno_mode_ok): Likewise.
675 2014-04-17 Jan Hubicka <hubicka@ucw.cz>
677 * ipa-inline.c (inline_small_functions): Account only non-cold
679 * doc/invoke.texi (inline-unit-growth): Update documentation.
681 2014-04-17 Pat Haugen <pthaugen@us.ibm.com>
683 * config/rs6000/rs6000.md (addti3, subti3): New.
685 2014-04-17 H.J. Lu <hongjiu.lu@intel.com>
688 * config/i386/i386.c (ix86_expand_clear): Remove outdated
689 comment. Check optimize_insn_for_size_p instead of
690 optimize_insn_for_speed_p.
692 2014-04-17 Martin Jambor <mjambor@suse.cz>
694 * gimple-iterator.c (gsi_start_edge): New function.
695 * gimple-iterator.h (gsi_start_edge): Declare.
696 * tree-sra.c (single_non_eh_succ): New function.
697 (disqualify_ops_if_throwing_stmt): Renamed to
698 disqualify_if_bad_bb_terminating_stmt. Allow throwing statements
699 having one non-EH successor BB.
700 (sra_modify_expr): If stmt ends bb, use single non-EH successor to
701 generate loads into replacements.
702 (sra_modify_assign): Likewise and and also use the simple path for
704 (sra_modify_function_body): Commit statements on edges.
706 2014-04-17 Richard Biener <rguenther@suse.de>
709 * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
710 comparison results and add clarifying comment.
712 2014-04-17 Jakub Jelinek <jakub@redhat.com>
714 * genmodes.c (struct mode_data): Add need_bytesize_adj field.
715 (blank_mode): Initialize it.
716 (emit_mode_size_inline, emit_mode_nunits_inline,
717 emit_mode_inner_inline): New functions.
718 (emit_insn_modes_h): Call them and surround their output with
719 #if GCC_VERSION >= 4001 ... #endif.
720 * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
721 For GCC_VERSION >= 4001 use mode_*_inline routines instead of
722 mode_* arrays if the argument is __builtin_constant_p.
723 * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
724 is enum machine_mode.
726 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
728 * passes.c (opt_pass::execute): Adjust.
729 (pass_manager::execute_pass_mode_switching): Likewise.
730 (early_local_passes::execute): Likewise.
731 (execute_one_pass): Pass cfun to the pass's execute method.
732 * tree-pass.h (opt_pass::execute): Add function * argument.
733 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
734 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
735 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
736 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
737 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
738 config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
739 except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
740 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
741 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
742 ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
743 ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
744 lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
745 postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
746 reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
747 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
748 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
749 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
750 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
751 tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
752 tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
753 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
754 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
755 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
756 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
757 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
758 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
759 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
760 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
761 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
764 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
766 * passes.c (opt_pass::gate): Take function * argument.
767 (gate_all_early_local_passes): Merge into
768 (early_local_passes::gate): this.
769 (gate_all_early_optimizations): Merge into
770 (all_early_optimizations::gate): this.
771 (gate_all_optimizations): Mege into
772 (all_optimizations::gate): this.
773 (gate_all_optimizations_g): Merge into
774 (all_optimizations_g::gate): this.
775 (gate_rest_of_compilation): Mege into
776 (rest_of_compilation::gate): this.
777 (gate_postreload): Merge into
778 (postreload::gate): this.
779 (dump_one_pass): Pass cfun to the pass's gate method.
780 (execute_ipa_summary_passes): Likewise.
781 (execute_one_pass): Likewise.
782 (ipa_write_summaries_2): Likewise.
783 (ipa_write_optimization_summaries_1): Likewise.
784 (ipa_read_summaries_1): Likewise.
785 (ipa_read_optimization_summaries_1): Likewise.
786 (execute_ipa_stmt_fixups): Likewise.
787 * tree-pass.h (opt_pass::gate): Add function * argument.
788 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
789 combine-stack-adj.c, combine.c, compare-elim.c,
790 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
791 config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
792 config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
793 dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c, fwprop.c, gcse.c,
794 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
795 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
796 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
797 loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
798 omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
799 reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
800 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
801 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
802 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
803 tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
804 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
805 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
806 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
807 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
808 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
809 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
810 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
811 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
812 tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
813 tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
814 var-tracking.c, vtable-verify.c, web.c: Adjust.
816 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
818 * configure.ac: Check for -Woverloaded-virtual and enable it if found.
819 * configure: Regenerate.
821 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
823 * passes.c (dump_one_pass): don't check pass->has_gate.
824 (execute_ipa_summary_passes): Likewise.
825 (execute_one_pass): Likewise.
826 (ipa_write_summaries_2): Likewise.
827 (ipa_write_optimization_summaries_1): Likewise.
828 (ipa_read_optimization_summaries_1): Likewise.
829 (execute_ipa_stmt_fixups): Likewise.
830 * tree-pass.h (pass_data::has_gate): Remove.
831 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
832 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
833 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
834 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
835 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
836 config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
837 config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
838 dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
839 gimple-low.c, gimple-ssa-isolate-paths.c,
840 gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
841 ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
842 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
843 ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
844 modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
845 recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
846 reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
847 trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
848 tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
849 tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
850 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
851 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
852 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
853 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
854 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
855 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
856 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
857 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
858 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
859 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
860 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
861 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
864 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
866 * pass_manager.h (pass_manager::register_dump_files_1): Remove
868 * passes.c (pass_manager::register_dump_files_1): Merge into
869 (pass_manager::register_dump_files): this, and remove its handling of
870 properties since the pass always has the properties anyway.
871 (pass_manager::pass_manager): Adjust.
873 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
875 * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
876 * passes.c (pass_manager::register_dump_files_1): Remove dead code
877 dealing with properties.
878 (pass_manager::register_dump_files): Adjust.
880 2014-03-20 Mark Wielaard <mjw@redhat.com>
882 * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
883 then represent the bound as normal constant value.
885 2014-04-17 Jakub Jelinek <jakub@redhat.com>
888 Forward port from 4.8 branch
889 2013-07-19 Kirill Yukhin <kirill.yukhin@intel.com>
891 * config/i386/bmiintrin.h (_blsi_u32): New.
895 (_blsmsk_u32): Ditto.
896 (_blsmsk_u64): Ditto.
900 2014-04-17 Kito Cheng <kito@0xlab.org>
902 * gcc.c (used_arg): Prevent out of bound access for multilib_options.
904 2014-04-17 Richard Biener <rguenther@suse.de>
907 * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
908 boolean results for comparisons.
910 2014-04-17 Richard Biener <rguenther@suse.de>
912 PR tree-optimization/60836
913 * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
914 initial PHI args to be gimple values.
916 2014-04-17 Richard Biener <rguenther@suse.de>
918 PR tree-optimization/60841
919 * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
920 * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
921 of stmts to SLP build.
922 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
923 (vect_analyze_slp): Likewise.
924 (vect_analyze_slp_instance): Likewise.
925 (vect_build_slp_tree): Limit overall SLP tree growth.
926 * tree-vectorizer.h (vect_analyze_data_refs,
927 vect_analyze_slp): Adjust prototypes.
929 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
931 * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
934 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
936 * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
937 * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
938 * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
939 for TARGET_SLOW_PSHUFB
941 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
943 * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
944 * config/i386/i386.c (intel_cost): Ditto.
946 2014-04-17 Joey Ye <joey.ye@arm.com>
948 * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
950 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
952 * opts.c (common_handle_option): Disable -fipa-reference coorectly
955 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
957 * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
958 (type_all_derivations_known_p): New predicate.
959 (type_all_ctors_visible_p): New predicate.
960 (type_possibly_instantiated_p): New predicate.
961 (get_odr_type): Compute all_derivations_known.
962 (dump_odr_type): Dump the flag.
963 (maybe_record_type): Cleanup.
964 (record_target_from_binfo): Add bases_to_consider array;
965 record bases for types w/o instances and skip CXX destructor.
966 (possible_polymorphic_call_targets_1): Add bases_to_consider
967 and consider_construction parameters; check if type may have instance.
968 (get_polymorphic_call_info): Set maybe_in_construction to true
969 when we know nothing.
970 (record_targets_from_bases): Skip CXX destructors; they are
971 never called for types in construction.
972 (possible_polymorphic_call_targets): Do not record target when
973 type may not have instance.
975 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
978 * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
979 external aliases alive, too.
981 2014-04-16 Andrew Pinski <apinski@cavium.com>
983 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
986 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
988 * final.c (compute_alignments): Do not apply loop alignment to a block
989 falling through to the exit.
991 2014-04-16 Catherine Moore <clm@codesourcery.com>
993 * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
994 Adjust constraints for microMIPS store patterns.
996 2014-04-16 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
998 * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
1000 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
1002 * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
1003 (append_use): Run at -O0.
1004 (append_vdef): Likewise.
1005 * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
1006 * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
1008 2014-04-16 Jakub Jelinek <jakub@redhat.com>
1010 PR tree-optimization/60844
1011 * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
1012 (propagate_op_to_single_use, remove_visited_stmt_chain,
1013 linearize_expr, repropagate_negates, reassociate_bb): Use it
1014 instead of gsi_remove.
1016 2014-04-16 Martin Jambor <mjambor@suse.cz>
1018 * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
1019 ipa_transforms_to_apply.
1020 (cgraph_function_versioning): Assert that old_node has empty
1021 ipa_transforms_to_apply.
1022 * trans-mem.c (ipa_tm_create_version): Likewise.
1023 * tree-inline.c (tree_function_versioning): Do not duplicate
1024 ipa_transforms_to_apply.
1026 2014-04-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1029 * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
1031 Pass necessary as flags on 64-bit Solaris/x86.
1032 Use lowercase relocs for x86_64-*-*.
1033 * configure: Regenerate.
1035 2014-04-15 Jan Hubicka <jh@suse.cz>
1037 * ipa-devirt.c (referenced_from_vtable_p): New predicate.
1038 (maybe_record_node, likely_target_p): Use it.
1040 2014-04-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1043 Revert following patch
1045 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
1048 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1049 software floating point or no floating point registers, do not
1050 allow any type in the FPRs. Eliminate a test for SPE SIMD types
1051 in GPRs that occurs after we tested for GPRs that would never be
1054 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1055 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1056 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
1057 specifically allow DDmode, since that does not use the SPE SIMD
1060 2014-03-21 Mark Wielaard <mjw@redhat.com>
1062 * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
1063 as unsigned or int depending on type and value used.
1065 2014-04-15 Richard Biener <rguenther@suse.de>
1067 PR rtl-optimization/56965
1068 * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
1069 * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
1071 * alias.c (true_dependence_1): Do not call
1072 nonoverlapping_component_refs_p.
1073 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
1074 nonoverlapping_component_refs_p.
1075 (indirect_refs_may_alias_p): Likewise.
1077 2014-04-15 Teresa Johnson <tejohnson@google.com>
1079 * cfg.c (dump_bb_info): Fix flags check.
1080 * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
1082 2014-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1084 PR rtl-optimization/60663
1085 * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
1088 2014-04-15 Richard Biener <rguenther@suse.de>
1090 * lto-streamer.h (LTO_major_version): Bump to 4.
1092 2014-04-15 Richard Biener <rguenther@suse.de>
1094 * common.opt (lto_partition_model): New enum.
1095 (flto-partition=): Merge separate options with a single with argument,
1096 add -flto-partition=one support.
1097 * flag-types.h (enum lto_partition_model): Declare.
1098 * opts.c (finish_options): Remove duplicate -flto-partition=
1100 * lto-wrapper.c (run_gcc): Adjust.
1102 2014-04-15 Richard Biener <rguenther@suse.de>
1104 * alias.c (ncr_compar): New function.
1105 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
1107 2014-04-15 Richard Biener <rguenther@suse.de>
1109 * alias.c (record_component_aliases): Do not walk BINFOs.
1111 2014-04-15 Richard Biener <rguenther@suse.de>
1113 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1114 Add struct function argument and adjust.
1115 (find_func_aliases_for_call): Likewise.
1116 (find_func_aliases): Likewise.
1117 (find_func_clobbers): Likewise.
1118 (intra_create_variable_infos): Likewise.
1119 (compute_points_to_sets): Likewise.
1120 (ipa_pta_execute): Adjust. Do not push/pop cfun.
1122 2014-04-15 Richard Biener <rguenther@suse.de>
1124 * tree.c (iterative_hash_expr): Use enum tree_code_class
1125 to store TREE_CODE_CLASS.
1126 (tree_block): Likewise.
1127 (tree_set_block): Likewise.
1128 * tree.h (fold_build_pointer_plus_loc): Use
1129 convert_to_ptrofftype_loc.
1131 2014-04-15 Jakub Jelinek <jakub@redhat.com>
1134 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
1137 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
1139 * cfgloop.h (struct loop): Move force_vectorize down.
1140 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
1141 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
1142 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
1143 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
1144 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
1145 * tree-core.h (enum annot_expr_kind): Add new kind values.
1146 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
1147 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
1149 * tree.def (ANNOTATE_EXPR): Tweak comment.
1151 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1153 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
1156 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
1158 * tree.h (TYPE_IDENTIFIER): Declare.
1159 * tree.c (subrange_type_for_debug_p): Use it.
1160 * godump.c (go_format_type): Likewise.
1161 * dwarf2out.c (is_cxx_auto, modified_type_die,
1162 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
1163 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
1165 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1168 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
1170 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
1172 * config/i386/i386.c (examine_argument): Return bool. Return true if
1173 parameter should be passed in memory.
1174 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
1175 (construct_container): Update calls to examine_argument.
1176 (function_arg_advance_64): Ditto.
1177 (return_in_memory_32): Merge with ix86_return_in_memory.
1178 (return_in_memory_64): Ditto.
1179 (return_in_memory_ms_64): Ditto.
1181 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
1183 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
1184 * coverage.c (coverage_compute_profile_id): Handle externally visible
1187 2014-04-14 Martin Jambor <mjambor@suse.cz>
1189 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
1190 DECL_DISREGARD_INLINE_LIMITS functions.
1192 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
1195 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
1197 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
1200 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
1201 optimize_insn_for_speed_p instead of
1202 optimize_function_for_speed_p.
1204 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
1206 * doc/invoke.texi (free): Document AArch64.
1208 2014-04-14 Richard Biener <rguenther@suse.de>
1210 PR tree-optimization/60042
1211 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
1212 (insert_into_preds_of_block): Do not prevent PHI insertion
1213 for REFERENCE exprs here ...
1214 (eliminate_dom_walker::before_dom_children): ... but prevent
1215 their use here under similar conditions when applied to the
1216 IL after PRE optimizations.
1218 2014-04-14 Richard Biener <rguenther@suse.de>
1220 * passes.def: Move early points-to after early SRA.
1222 2014-04-14 Richard Biener <rguenther@suse.de>
1224 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
1225 check for which sign-changes we allow when forwarding
1226 a converted value into a switch.
1228 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
1230 * stor-layout.c (place_field): Finalize non-constant offset for the
1233 2014-04-14 Richard Biener <rguenther@suse.de>
1235 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
1237 (expand_switch_using_bit_tests_p): Likewise.
1238 (process_switch): Compute and pass on speed_p based on the
1240 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
1241 optimize_bb_for_speed_p.
1243 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
1245 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
1246 * function.h (struct function): Rename has_force_vect_loops into
1247 has_force_vectorize_loops.
1248 * lto-streamer-in.c (input_cfg): Adjust for renaming.
1249 (input_struct_function_base): Likewise.
1250 * lto-streamer-out.c (output_cfg): Likewise.
1251 (output_struct_function_base): Likewise.
1252 * omp-low.c (expand_omp_simd): Likewise.
1253 * tree-cfg.c (move_sese_region_to_fn): Likewise.
1254 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1255 (version_loop_for_if_conversion): Likewise.
1256 (tree_if_conversion): Likewise.
1257 (main_tree_if_conversion): Likewise.
1258 (gate_tree_if_conversion): Likewise.
1259 * tree-inline.c (copy_loops): Likewise.
1260 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
1261 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
1262 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
1263 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
1264 * tree-vectorizer.c (vectorize_loops): Likewise.
1265 * tree-vectorizer.h (unlimited_cost_model): Likewise.
1267 2014-04-14 Richard Biener <rguenther@suse.de>
1270 * lto-streamer-out.c (wrap_refs): New function.
1271 (lto_output): Wrap symbol references in global initializes in
1272 type-preserving MEM_REFs.
1274 2014-04-14 Christian Bruel <christian.bruel@st.com>
1276 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
1278 2014-04-14 Christian Bruel <christian.bruel@st.com>
1280 * config/sh/sh.md (setmemqi): New expand pattern.
1281 * config/sh/sh.h (CLEAR_RATIO): Define.
1282 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
1283 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
1285 2014-04-14 Richard Biener <rguenther@suse.de>
1288 * fold-const.c (negate_expr_p): Don't negate directional rounding
1290 (fold_negate_expr): Likewise.
1292 2014-04-14 Richard Biener <rguenther@suse.de>
1294 PR tree-optimization/59817
1295 PR tree-optimization/60453
1296 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
1297 recursion to catch all CHRECs in the scalar evolution and restrict
1298 the predicate for the remains appropriately.
1300 2014-04-12 Catherine Moore <clm@codesourcery.com>
1302 * config/mips/constraints.md: Add new register constraint "kb".
1303 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
1304 (*movhi_internal): Likewise.
1305 (*movqi_internal): Likewise.
1306 * config/mips/mips.h (M16_STORE_REGS): New register class.
1307 (REG_CLASS_NAMES): Add M16_STORE_REGS.
1308 (REG_CLASS_CONTENTS): Likewise.
1309 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
1311 2014-04-11 Tobias Burnus <burnus@net-b.de>
1314 * doc/invoke.texi (-Wformat-signedness): Document it.
1315 (Wformat=2): Mention that this enables -Wformat-signedness.
1317 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1319 * common/config/epiphany/epiphany-common.c
1320 (epiphany_option_optimization_table): Enable section anchors by
1321 default at -O1 or higher.
1322 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
1323 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
1324 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
1325 carries no extra cost.
1326 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
1327 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
1328 * config/epiphany/predicates.md (memclob_operand): New predicate.
1329 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
1330 Use memclob_operand predicate and X constraint for operand 3.
1332 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1334 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
1335 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
1338 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
1340 PR rtl-optimization/60651
1341 * mode-switching.c (optimize_mode_switching): Make sure to emit
1342 sets of a lower numbered entity before sets of a higher numbered
1343 entity to a mode of the same or lower priority.
1344 When creating a seginfo for a basic block that starts with a code
1345 label, move the insertion point past the code label.
1346 (new_seginfo): Document and enforce requirement that
1347 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
1348 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
1349 * doc/tm.texi: Regenerate.
1351 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
1354 * config/arc/arc.c (arc_save_restore): Fix assert typo.
1356 2013-04-11 Jakub Jelinek <jakub@redhat.com>
1358 * BASE-VER: Set to 4.10.0.
1360 2014-04-11 Tobias Burnus <burnus@net-b.de>
1363 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
1364 * doc/gcc.texi (Service): Update description in the @menu
1365 * doc/invoke.texi (Option Summary): Remove misplaced and
1368 2014-04-11 Steve Ellcey <sellcey@mips.com>
1369 Jakub Jelinek <jakub@redhat.com>
1372 * expr.c (convert_move): Use emit_store_flag_force instead of
1373 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
1376 2014-04-11 Richard Biener <rguenther@suse.de>
1379 * varasm.c (assemble_alias): Avoid endless error reporting
1380 recursion by setting TREE_ASM_WRITTEN.
1382 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1384 * config/s390/s390.md: Add a splitter for NOT rtx.
1386 2014-04-11 Jakub Jelinek <jakub@redhat.com>
1388 PR rtl-optimization/60663
1389 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
1391 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
1392 Jakub Jelinek <jakub@redhat.com>
1395 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
1396 flag from decl_node to node.
1398 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1401 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
1402 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
1403 ameliorating the cases where it can be.
1405 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
1408 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
1410 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1411 (loadsync_<mode>): Change mode.
1412 (load_quadpti, store_quadpti): New.
1413 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1414 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1415 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
1417 2014-04-09 Cong Hou <congh@google.com>
1420 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
1423 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1425 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
1426 instead of vnor to exploit possible fusion opportunity in the
1428 (altivec_expand_vec_perm_const_le): Likewise.
1430 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
1432 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
1433 (loadsync_<mode>): Change mode.
1434 (load_quadpti, store_quadpti): New.
1435 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
1436 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
1438 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
1441 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
1442 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
1443 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
1445 2014-04-08 Richard Biener <rguenther@suse.de>
1448 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
1449 a 64bit widest int print double-int similar to on HWI64 hosts.
1451 2014-04-08 Richard Biener <rguenther@suse.de>
1453 PR tree-optimization/60785
1454 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
1455 default defs properly.
1457 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
1459 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
1460 (Weffc++): Likewise.
1462 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
1464 * ipa-devirt.c (maybe_record_node): When node is not recorded,
1465 set completep to false rather than true.
1467 2014-04-07 Douglas B Rupp <rupp@adacore.com>
1470 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
1471 ARM_TARGET2_DWARF_FORMAT.
1473 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
1476 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
1477 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
1480 2014-04-07 Richard Biener <rguenther@suse.de>
1482 PR tree-optimization/60766
1483 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
1484 (may_eliminate_iv): Convert cand_value_at result to desired type.
1486 2014-04-07 Jason Merrill <jason@redhat.com>
1489 * common.opt (-fno-gnu-unique): Add.
1490 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
1492 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1494 * haifa-sched.c: Fix outdated function reference and minor
1495 grammar errors in introductory comment.
1497 2014-04-07 Richard Biener <rguenther@suse.de>
1500 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
1502 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
1504 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
1507 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
1508 size accounting for thunks.
1509 (pa_asm_output_mi_thunk): Use final_start_function() and
1510 final_end_function() to output function start and end directives.
1512 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
1514 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
1515 device specific ISA/ feature information. Remove short_sp and
1516 errata_skip ds. Add avr_device_specific_features enum to have device
1518 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
1519 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
1520 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
1521 updated device specific info.
1522 * config/avr/avr-mcus.def: Merge device specific details to
1523 dev_attribute field.
1524 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
1526 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
1527 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
1528 assembler if RMW isa supported by current device.
1529 * config/avr/genmultilib.awk: Update as device info structure changed.
1530 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
1532 2014-04-04 Cong Hou <congh@google.com>
1534 PR tree-optimization/60656
1535 * tree-vect-stmts.c (supportable_widening_operation):
1536 Fix a bug that elements in a vector with vect_used_by_reduction
1537 property are incorrectly reordered when the operation on it is not
1538 consistant with the one in reduction operation.
1540 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
1542 PR rtl-optimization/60155
1543 * gcse.c (record_set_data): New function.
1544 (single_set_gcse): New function.
1545 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
1546 (hoist_code): Likewise.
1547 (get_pressure_class_and_nregs): Likewise.
1549 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
1551 * explow.c (probe_stack_range): Emit a final optimization blockage.
1553 2014-04-04 Anthony Green <green@moxielogic.com>
1555 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
1558 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
1561 * lto-cgraph.c (input_overwrite_node): Check that partitioning
1562 flags are set only during streaming.
1563 * ipa.c (process_references, walk_polymorphic_call_targets,
1564 symtab_remove_unreachable_nodes): Drop bodies of always inline
1565 after early inlining.
1566 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
1568 2014-04-04 Jakub Jelinek <jakub@redhat.com>
1569 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1572 * dwarf2out.c (const_ok_for_output_1): Reject expressions
1575 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1578 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
1580 (cortex_a53_fdivd): Likewise.
1582 2014-04-04 Martin Jambor <mjambor@suse.cz>
1585 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
1587 * cgraph.c (clone_of_p): Also return true if thunks match.
1588 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
1589 cgraph_function_or_thunk_node and an obsolete comment.
1590 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
1592 (build_function_decl_skip_args): Likewise.
1593 (set_new_clone_decl_and_node_flags): New function.
1594 (duplicate_thunk_for_node): Likewise.
1595 (redirect_edge_duplicating_thunks): Likewise.
1596 (cgraph_clone_node): New parameter args_to_skip, pass it to
1597 redirect_edge_duplicating_thunks which is called instead of
1598 cgraph_redirect_edge_callee.
1599 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
1600 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
1602 2014-04-04 Jeff Law <law@redhat.com>
1605 * config/arm/predicates.md (const_int_I_operand): New predicate.
1606 (const_int_M_operand): Similarly.
1607 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
1609 (insv_t2, extv_reg, extzv_t2): Likewise.
1610 (load_multiple_with_writeback): Similarly for const_int_I_operand.
1611 (pop_multiple_with_writeback_and_return): Likewise.
1612 (vfp_pop_multiple_with_writeback): Likewise
1614 2014-04-04 Richard Biener <rguenther@suse.de>
1617 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
1618 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
1620 * gimplify.h (gimple_add_tmp_var_fn): Declare.
1621 * gimplify.c (gimple_add_tmp_var_fn): New function.
1622 * gimple-expr.h (create_tmp_reg_fn): Declare.
1623 * gimple-expr.c (create_tmp_reg_fn): New function.
1624 * gimple-low.c (record_vars_into): Don't change cfun.
1625 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
1626 code generation without cfun.
1628 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
1631 * Makefile.in (install-driver): Fix shell scripting.
1633 2014-04-03 Cong Hou <congh@google.com>
1635 PR tree-optimization/60505
1636 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
1637 threshold of number of iterations below which no vectorization
1639 * tree-vect-loop.c (new_loop_vec_info):
1640 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
1641 * tree-vect-loop.c (vect_analyze_loop_operations):
1642 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
1643 * tree-vect-loop.c (vect_transform_loop):
1644 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
1645 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
1646 of iterations of the loop and see if we should build the epilogue.
1648 2014-04-03 Richard Biener <rguenther@suse.de>
1650 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
1651 (streamer_tree_cache_create): Adjust.
1652 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
1653 to allow optional nodes array.
1654 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
1655 (streamer_tree_cache_append): Likewise.
1656 (streamer_tree_cache_create): Create nodes array optionally
1657 as specified by parameter.
1658 * lto-streamer-out.c (create_output_block): Avoid maintaining
1659 the node array in the writer cache.
1660 (DFS_write_tree): Remove assertion.
1661 (produce_asm_for_decls): Free the out decl state hash table early.
1662 * lto-streamer-in.c (lto_data_in_create): Adjust for
1663 streamer_tree_cache_create prototype change.
1665 2014-04-03 Richard Biener <rguenther@suse.de>
1667 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
1668 set TREE_CHAIN to NULL_TREE.
1670 2014-04-03 Richard Biener <rguenther@suse.de>
1672 PR tree-optimization/60740
1673 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
1674 over all GIMPLE_COND operands.
1676 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
1678 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
1679 (Weffc++): Remove Scott's numbering, merge lists and reference
1682 2014-04-03 Nick Clifton <nickc@redhat.com>
1684 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
1687 2014-04-03 Martin Jambor <mjambor@suse.cz>
1689 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
1690 mention gcc_unreachable before failing.
1691 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
1694 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
1697 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
1698 inconsistent code and instead mark the context inconsistent.
1699 (possible_polymorphic_call_targets): For inconsistent contexts
1700 return empty complete list.
1702 2014-04-02 Anthony Green <green@moxielogic.com>
1704 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
1705 (extendqisi2, extendhisi2): Define.
1706 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
1707 (WCHAR_TYPE): Change to unsigned int.
1709 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1711 PR tree-optimization/60733
1712 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
1713 insertion point for PHI candidates to be the end of the feeding
1714 block for the PHI argument.
1716 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
1718 PR rtl-optimization/60650
1719 * lra-constraints.c (process_alt_operands): Decrease reject for
1720 earlyclobber matching.
1722 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1724 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
1726 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1728 * config/spu/spu.c (pad_bb): Do not crash when the last
1729 insn is CODE_FOR_blockage.
1731 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1733 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
1734 lies outside the target mode.
1736 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
1739 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1740 software floating point or no floating point registers, do not
1741 allow any type in the FPRs. Eliminate a test for SPE SIMD types
1742 in GPRs that occurs after we tested for GPRs that would never be
1745 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1746 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1747 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
1748 specifically allow DDmode, since that does not use the SPE SIMD
1751 2014-04-02 Richard Biener <rguenther@suse.de>
1754 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
1755 MODE_INTs. Properly use negv_optab.
1756 (expand_abs): Likewise.
1758 2014-04-02 Richard Biener <rguenther@suse.de>
1761 * Makefile.in (install-driver): Guard extra installs with special
1764 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1766 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1769 2014-04-01 Richard Henderson <rth@redhat.com>
1772 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
1773 alternative enabled before register allocation.
1775 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
1777 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
1778 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
1780 (nios2_large_got_address): Remove unneeded 'sym' parameter.
1781 (nios2_got_address): Update nios2_large_got_address call site.
1782 (nios2_delegitimize_address): New function.
1783 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
1784 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
1785 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
1787 2014-04-01 Martin Husemann <martin@duskware.de>
1789 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
1792 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
1794 PR rtl-optimization/60604
1795 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
1796 check from register_operand.
1797 (register_operand): Redefine in terms of general_operand.
1798 (nonmemory_operand): Use register_operand for the non-constant cases.
1800 2014-04-01 Richard Biener <rguenther@suse.de>
1802 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
1804 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
1806 * doc/invoke.texi (mapp-regs): Clarify.
1808 2014-03-31 Ulrich Drepper <drepper@gmail.com>
1810 * config/i386/avx512fintrin.h (__v32hi): Define type.
1811 (__v64qi): Likewise.
1812 (_mm512_set1_epi8): Define.
1813 (_mm512_set1_epi16): Define.
1814 (_mm512_set4_epi32): Define.
1815 (_mm512_set4_epi64): Define.
1816 (_mm512_set4_pd): Define.
1817 (_mm512_set4_ps): Define.
1818 (_mm512_setr4_epi64): Define.
1819 (_mm512_setr4_epi32): Define.
1820 (_mm512_setr4_pd): Define.
1821 (_mm512_setr4_ps): Define.
1822 (_mm512_setzero_epi32): Define.
1824 2014-03-31 Martin Jambor <mjambor@suse.cz>
1827 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
1828 callsite_arguments_match_p. Updated all callers. Also check types of
1829 corresponding formal parameters and actual arguments.
1830 (not_all_callers_have_enough_arguments_p) Renamed to
1831 some_callers_have_mismatched_arguments_p.
1833 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
1835 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
1837 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
1840 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
1843 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
1845 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
1846 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
1848 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
1849 Use FMAMODE_NOVF512 mode iterator.
1850 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
1851 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
1852 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
1853 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
1855 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
1856 Use VF_128_256 mode iterator.
1857 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
1860 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1862 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
1863 static chain if needed.
1865 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1868 * lra-constraints.c (index_part_to_reg): New.
1869 (process_address): Use it.
1871 2014-03-27 Jeff Law <law@redhat.com>
1872 Jakub Jelinek <jakub@redhat.com>
1875 * expr.c (do_tablejump): Use simplify_gen_binary rather than
1876 gen_rtx_{PLUS,MULT} to build up the address expression.
1878 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
1879 creating non-canonical RTL.
1881 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1884 * ipa-inline.c (want_inline_small_function_p): Short circuit large
1885 functions; reorganize to make cheap checks first.
1886 (inline_small_functions): Do not estimate growth when dumping;
1888 * ipa-inline.h (inline_summary): Add min_size.
1889 (growth_likely_positive): New function.
1890 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
1891 (set_cond_stmt_execution_predicate): Cleanup.
1892 (estimate_edge_size_and_time): Compute min_size.
1893 (estimate_calls_size_and_time): Likewise.
1894 (estimate_node_size_and_time): Likewise.
1895 (inline_update_overall_summary): Update min_size.
1896 (do_estimate_edge_time): Likewise.
1897 (do_estimate_edge_size): Update.
1898 (do_estimate_edge_hints): Update.
1899 (growth_likely_positive): New function.
1901 2014-03-28 Jakub Jelinek <jakub@redhat.com>
1904 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
1905 also if addr has VOIDmode.
1907 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1909 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
1910 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
1912 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
1913 instructions as well as AdvancedSIMD loads.
1915 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1917 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
1918 Use crypto_aese type.
1919 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
1920 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
1921 crypto_aese, crypto_aesmc. Move to types.md.
1922 * config/arm/types.md (crypto_aes): Split into crypto_aese,
1924 * config/arm/iterators.md (crypto_type): Likewise.
1926 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1928 * cgraph.c: Include expr.h and tree-dfa.h.
1929 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
1932 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1935 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
1936 regs from checking multi-reg pseudos.
1938 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1940 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
1942 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1944 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
1945 if it would clobber the stack pointer, even temporarily.
1947 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
1949 * mode-switching.c: Make small adjustments to the top comment.
1951 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1953 * config/rs6000/constraints.md (wD constraint): New constraint to
1954 match the constant integer to get the top DImode/DFmode out of a
1955 vector in a VSX register.
1957 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
1958 match the constant integer to get the top DImode/DFmode out of a
1959 vector in a VSX register.
1961 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
1964 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1967 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
1968 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
1970 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
1971 Optimize vec_extract of 64-bit values, where the value being
1972 extracted is in the top word, where we can use scalar
1973 instructions. Add direct move and store support. Combine the big
1974 endian/little endian vector select load support into a single insn.
1975 (vsx_extract_<mode>_internal1): Likewise.
1976 (vsx_extract_<mode>_internal2): Likewise.
1977 (vsx_extract_<mode>_load): Likewise.
1978 (vsx_extract_<mode>_store): Likewise.
1979 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
1980 combined into vsx_extract_<mode>_load.
1981 (vsx_extract_<mode>_one_le): Likewise.
1983 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
1984 define the top 64-bit vector element.
1986 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
1989 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1990 Document vec_vbpermq builtin.
1993 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
1994 enable use of xxsldwi and xxpermdi builtin functions.
1995 (vec_xxpermdi): Likewise.
1997 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1998 Document use of vec_xxsldwi and vec_xxpermdi builtins.
2000 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
2002 PR rtl-optimization/60650
2003 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
2005 (find_spills_for): New.
2006 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
2007 Spill all pseudos on the second iteration.
2009 2014-03-27 Marek Polacek <polacek@redhat.com>
2012 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
2015 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2017 * config/s390/s390.c (s390_can_use_return_insn): Check for
2018 call-saved FPRs on 31 bit.
2020 2014-03-27 Jakub Jelinek <jakub@redhat.com>
2023 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
2024 if they need regimplification, just drop them instead of
2025 calling gimple_regimplify_operands on them.
2027 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
2030 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
2031 (aarch64_frame_pointer_required): Adjust logic.
2032 (aarch64_can_eliminate): Adjust logic.
2033 (aarch64_override_options_after_change): Adjust logic.
2035 2014-03-27 Dehao Chen <dehao@google.com>
2037 * ipa-inline.c (early_inliner): Update node's inline info.
2039 2014-03-26 Dehao Chen <dehao@google.com>
2041 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
2042 compiler inserted conditional jumps for NAN float check.
2044 2014-03-26 Jakub Jelinek <jakub@redhat.com>
2046 * ubsan.h (ubsan_create_data): Change second argument's type
2047 to const location_t *.
2048 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
2050 (ubsan_create_data): Change second argument to const location_t *PLOC.
2051 Create Loc field whenever PLOC is non-NULL.
2052 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
2053 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
2057 * real.c (real_to_integer2): Change type of low to UHWI.
2059 2014-03-26 Tobias Burnus <burnus@net-b.de>
2061 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
2062 (CILK_SELF_SPECS): New define.
2063 (driver_self_specs): Use it.
2065 2014-03-26 Richard Biener <rguenther@suse.de>
2067 * tree-pretty-print.c (percent_K_format): Implement special
2068 case for LTO and its stripped down BLOCK tree.
2070 2014-03-26 Jakub Jelinek <jakub@redhat.com>
2073 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
2075 * tree-vrp.c (simplify_internal_call_using_ranges): If only
2076 one range is range_int_cst_p, but not both, at least optimize
2077 addition/subtraction of 0 and multiplication by 0 or 1.
2078 * gimple-fold.c (gimple_fold_call): Fold
2079 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
2080 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
2081 INTEGER_CSTs, try to fold at least x * 0 and y - y.
2083 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
2085 PR rtl-optimization/60452
2086 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
2087 <case REG>: Return 1 for invalid offsets from the frame pointer.
2089 2014-03-26 Marek Polacek <polacek@redhat.com>
2092 * doc/extend.texi (C Extensions): Mention variable-length arrays in
2095 2014-03-26 Marek Polacek <polacek@redhat.com>
2098 * doc/extend.texi (Designated Inits): Describe what happens to omitted
2101 2014-03-26 Marek Polacek <polacek@redhat.com>
2104 * ira-color.c (update_conflict_hard_regno_costs): Perform the
2105 multiplication in unsigned type.
2107 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
2109 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
2111 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
2113 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
2115 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
2118 * cif-code.def (UNREACHABLE) New code.
2119 * ipa-inline.c (inline_small_functions): Skip edges to
2120 __builtlin_unreachable.
2121 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
2122 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
2123 predicate to __bulitin_unreachable.
2124 (set_cond_stmt_execution_predicate): Fix issue when
2125 invert_tree_comparison returns ERROR_MARK.
2126 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
2127 propagate to inline clones.
2128 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
2130 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
2131 * cgraphclones.c (cgraph_clone_node): If call destination is already
2132 ureachable, do not redirect it back.
2133 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
2136 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
2138 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
2139 Do not modify inline clones.
2141 2014-03-25 Jakub Jelinek <jakub@redhat.com>
2143 * config/i386/i386.md (general_sext_operand): New mode attr.
2144 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
2145 don't generate (sign_extend (const_int)).
2146 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
2147 operands[2]. Use We constraint instead of <i> and
2148 <general_sext_operand> predicate instead of <general_operand>.
2149 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
2150 * config/i386/constraints.md (We): New constraint.
2151 * config/i386/predicates.md (x86_64_sext_operand,
2152 sext_operand): New predicates.
2154 2014-03-25 Martin Jambor <mjambor@suse.cz>
2157 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
2158 inconsistent devirtualizations to __builtin_unreachable.
2160 2014-03-25 Marek Polacek <polacek@redhat.com>
2163 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
2165 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
2167 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
2168 order of elements for big-endian.
2170 2014-03-25 Richard Biener <rguenther@suse.de>
2173 * gimplify-me.c (gimple_regimplify_operands): Update the
2176 2014-03-25 Martin Jambor <mjambor@suse.cz>
2179 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
2180 (lto_output_varpool_node): Likewise.
2181 (input_overwrite_node): Likewise.
2182 (input_varpool_node): Likewise.
2184 2014-03-25 Richard Biener <rguenther@suse.de>
2186 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
2187 (run_gcc): Likewise.
2189 2014-03-25 Jakub Jelinek <jakub@redhat.com>
2191 * combine.c (simplify_compare_const): Add MODE argument.
2192 Handle mode_width 0 as very large mode_width.
2193 (try_combine, simplify_comparison): Adjust callers.
2195 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
2196 type to avoid signed integer overflow.
2197 * explow.c (plus_constant): Likewise.
2199 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
2201 * doc/generic.texi: Correct typos.
2203 2014-03-24 Tobias Burnus <burnus@net-b.de>
2205 * doc/invoke.texi (-flto): Expand section about
2206 using static libraries with LTO.
2208 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2210 PR rtl-optimization/60501
2211 * optabs.def (addptr3_optab): New optab.
2212 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
2213 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
2214 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
2216 * lra.c (emit_add3_insn): Use the addptr pattern if available.
2218 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
2220 2014-03-24 Ulrich Drepper <drepper@gmail.com>
2222 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
2225 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
2226 (_mm256_undefined_ps): Define.
2227 (_mm256_undefined_pd): Define.
2228 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
2229 (_mm_undefined_pd): Define.
2230 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
2231 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
2232 (_mm512_undefined_ps): Define.
2233 (_mm512_undefined_pd): Define.
2234 Use _mm*_undefined_*.
2235 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
2237 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
2239 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
2240 (lshr_simd): DI mode added.
2241 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
2242 (aarch64_ushr_simddi): Likewise.
2243 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
2244 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
2245 (vshrd_n_u64): Likewise.
2247 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2249 * Makefile.in (s-macro_list): Depend on cc1.
2251 2014-03-23 Teresa Johnson <tejohnson@google.com>
2253 * ipa-utils.c (ipa_print_order): Use specified dump file.
2255 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
2257 PR rtl-optimization/60601
2258 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
2260 * gcc.c (eval_spec_function): Initialize save_growing_value.
2262 2014-03-22 Jakub Jelinek <jakub@redhat.com>
2265 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
2266 code == MINUS_EXPR, never swap op0 with op1.
2268 * toplev.c (init_local_tick): Avoid signed integer multiplication
2270 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
2271 shift by first operand's bitsize.
2273 2014-03-21 Jakub Jelinek <jakub@redhat.com>
2276 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
2278 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
2279 TARGET_ISA_64BIT_P(x).
2281 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2283 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
2284 pattern for vector nor instead of subtract from splat(-1).
2285 (altivec_expand_vec_perm_const_le): Likewise.
2287 2014-03-21 Richard Henderson <rth@twiddle.net>
2290 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
2291 related insns after epilogue_completed.
2293 2014-03-21 Martin Jambor <mjambor@suse.cz>
2296 * cgraph.h (symtab_node): New flag body_removed.
2297 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
2298 when removing bodies.
2299 * symtab.c (dump_symtab_base): Dump body_removed flag.
2300 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
2301 had their bodies removed.
2303 2014-03-21 Martin Jambor <mjambor@suse.cz>
2306 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
2309 2014-03-21 Richard Biener <rguenther@suse.de>
2311 PR tree-optimization/60577
2312 * tree-core.h (struct tree_base): Document nothrow_flag use
2313 in VAR_DECL_NONALIASED.
2314 * tree.h (VAR_DECL_NONALIASED): New.
2315 (may_be_aliased): Adjust.
2316 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
2318 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2320 * expr.c (expand_expr_real_1): Remove outdated comment.
2322 2014-03-20 Jakub Jelinek <jakub@redhat.com>
2325 * ira.c (adjust_cleared_regs): Call copy_rtx on
2326 *reg_equiv[REGNO (loc)].src_p before passing it to
2327 simplify_replace_fn_rtx.
2330 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
2331 into CONST, put pic register as first operand of PLUS. Use
2332 gen_const_mem for both 32-bit and 64-bit PIC got loads.
2334 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2336 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
2338 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2340 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
2341 around for store forwarding issue in the FPU on the UT699.
2342 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
2343 loads and operations if -mfix-ut699 is specified.
2344 (divtf3_hq): Tweak attribute.
2345 (sqrttf2_hq): Likewise.
2347 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
2349 * calls.c (store_one_arg): Remove incorrect const qualification on the
2350 type of the temporary.
2351 * cfgexpand.c (expand_return): Likewise.
2352 * expr.c (expand_constructor): Likewise.
2353 (expand_expr_real_1): Likewise.
2355 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
2357 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
2360 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
2363 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
2365 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
2367 * config/arm/aarch-common-protos.h
2368 (alu_cost_table): Fix spelling of "extend".
2369 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
2371 2014-03-19 Richard Biener <rguenther@suse.de>
2374 * tree-core.h (tree_type_common): Re-order pointer members
2375 to reduce recursion depth during GC walks.
2377 2014-03-19 Marek Polacek <polacek@redhat.com>
2380 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
2381 before accessing it.
2383 2014-03-19 Richard Biener <rguenther@suse.de>
2386 * lto-streamer-in.c (input_function): In WPA stage do not drop
2389 2014-03-19 Jakub Jelinek <jakub@redhat.com>
2391 PR tree-optimization/60559
2392 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
2393 with build_zero_cst assignment.
2395 2014-03-18 Kai Tietz <ktietz@redhat.com>
2397 PR rtl-optimization/56356
2398 * sdbout.c (sdbout_parms): Verify that parms'
2399 incoming argument is valid.
2400 (sdbout_reg_parms): Likewise.
2402 2014-03-18 Richard Henderson <rth@redhat.com>
2405 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
2406 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
2407 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
2409 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
2411 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
2412 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
2413 Italicize plugin event names in description. Explain that
2414 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
2415 Remind that no GCC functions should be called after PLUGIN_FINISH.
2416 Explain what pragmas with expansion are.
2418 2014-03-18 Martin Liska <mliska@suse.cz>
2420 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
2421 gimple call statement is update.
2422 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
2423 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
2425 2014-03-18 Jakub Jelinek <jakub@redhat.com>
2428 * ubsan.c (ubsan_instrument_unreachable): Call
2429 initialize_sanitizer_builtins.
2430 (ubsan_pass): Likewise.
2433 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
2434 varpool_finalize_decl instead of rest_of_decl_compilation.
2436 2014-03-18 Richard Biener <rguenther@suse.de>
2438 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
2439 by using bitmap_and_compl instead of bitmap_and_compl_into.
2440 (df_rd_transfer_function): Likewise.
2442 2014-03-18 Richard Biener <rguenther@suse.de>
2444 * doc/lto.texi (fresolution): Fix typo.
2446 2014-03-18 Richard Biener <rguenther@suse.de>
2448 * doc/invoke.texi (flto): Update for changes in 4.9.
2450 2014-03-18 Richard Biener <rguenther@suse.de>
2452 * doc/loop.texi: Remove section on the removed lambda framework.
2453 Update loop docs with recent changes in preserving loop structure.
2455 2014-03-18 Richard Biener <rguenther@suse.de>
2457 * doc/lto.texi (-fresolution): Document.
2459 2014-03-18 Richard Biener <rguenther@suse.de>
2461 * doc/contrib.texi: Adjust my name.
2463 2014-03-18 Jakub Jelinek <jakub@redhat.com>
2466 * internal-fn.c: Include diagnostic-core.h.
2467 (expand_BUILTIN_EXPECT): New function.
2468 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
2469 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
2470 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
2471 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
2473 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
2474 Revert 3 argument __builtin_expect code.
2475 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
2476 * gimple-fold.c (gimple_fold_call): Likewise.
2477 * tree.h (fold_builtin_expect): New prototype.
2478 * builtins.c (build_builtin_expect_predicate): Add predictor
2479 argument, if non-NULL, create 3 argument __builtin_expect.
2480 (fold_builtin_expect): No longer static. Add ARG2 argument,
2481 pass it through to build_builtin_expect_predicate.
2482 (fold_builtin_2): Adjust caller.
2483 (fold_builtin_3): Handle BUILT_IN_EXPECT.
2484 * internal-fn.def (BUILTIN_EXPECT): New.
2486 2014-03-18 Tobias Burnus <burnus@net-b.de>
2489 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
2490 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
2491 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
2493 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
2496 * predict.c (combine_predictions_for_bb): Fix up formatting.
2497 (expr_expected_value_1, expr_expected_value): Add predictor argument,
2498 fill what it points to if non-NULL.
2499 (tree_predict_by_opcode): Adjust caller, use the predictor.
2500 * predict.def (PRED_COMPARE_AND_SWAP): Add.
2502 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
2504 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
2505 proper constant for the store mode.
2507 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
2509 * symtab.c (change_decl_assembler_name): Fix transparent alias
2512 2014-03-16 Renlin Li <Renlin.Li@arm.com>
2514 * config/aarch64/aarch64.c: Correct the comments about the
2515 aarch64 stack layout.
2517 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
2519 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
2520 check for GF_OMP_FOR_KIND_FOR.
2522 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
2524 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
2525 ymm and zmm register names.
2527 2014-03-17 Jakub Jelinek <jakub@redhat.com>
2530 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
2531 note creation for the 2010-08-31 changes.
2533 2014-03-17 Marek Polacek <polacek@redhat.com>
2536 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
2537 as -fno-tree-loop-vectorize.
2538 (expand_omp_simd): Likewise.
2540 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
2542 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
2543 (eligible_for_call_delay): New prototype.
2544 * config/sparc/sparc.c (tls_call_delay): Rename into...
2545 (eligible_for_call_delay): ...this. Return false if the instruction
2546 cannot be put in the delay slot of a branch.
2547 (eligible_for_restore_insn): Simplify.
2548 (eligible_for_return_delay): Return false if the instruction cannot be
2549 put in the delay slot of a branch and simplify.
2550 (eligible_for_sibcall_delay): Return false if the instruction cannot be
2551 put in the delay slot of a branch.
2552 * config/sparc/sparc.md (fix_ut699): New attribute.
2553 (tls_call_delay): Delete.
2554 (in_call_delay): Reimplement.
2555 (eligible_for_sibcall_delay): Rename into...
2556 (in_sibcall_delay): ...this.
2557 (eligible_for_return_delay): Rename into...
2558 (in_return_delay): ...this.
2559 (in_branch_delay): Reimplement.
2560 (in_uncond_branch_delay): Delete.
2561 (in_annul_branch_delay): Delete.
2563 2014-03-14 Richard Henderson <rth@redhat.com>
2566 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
2567 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
2568 (*floathi<X87MODEF>2_i387_with_temp): Remove.
2569 (floathi splitters): Remove.
2570 (float<SWI48x>xf2): New pattern.
2571 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
2572 code that tried to handle DImode for 32-bit, but which was excluded
2573 by the pattern's condition. Drop allocation of stack temporary.
2574 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
2575 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
2576 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
2577 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
2578 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
2579 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
2580 (*float<SWI48><MODEF>2_sse_interunit): Remove.
2581 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
2582 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
2583 (*float<SWI48x><X87MODEF>2_i387): Remove.
2584 (all float _with_temp splitters): Remove.
2585 (*float<SWI48x><MODEF>2_i387): New pattern.
2586 (*float<SWI48><MODEF>2_sse): New pattern.
2587 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
2588 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
2590 2014-03-14 Jakub Jelinek <jakub@redhat.com>
2591 Marek Polacek <polacek@redhat.com>
2594 * common.opt (dump_base_name_prefixed): New Variable.
2595 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
2596 if x_dump_base_name_prefixed is already set, set it at the end.
2598 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
2600 PR rtl-optimization/60508
2601 * lra-constraints.c (get_reload_reg): Add new parameter
2603 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
2604 Pass the new parameter values.
2606 2014-03-14 Richard Biener <rguenther@suse.de>
2608 * common.opt: Revert unintented changes from r205065.
2611 2014-03-14 Richard Biener <rguenther@suse.de>
2614 * cfghooks.c (split_block): Properly adjust all loops the
2615 block was a latch of.
2617 2014-03-14 Martin Jambor <mjambor@suse.cz>
2620 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
2623 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
2626 * config/avr/avr.c (avr_set_current_function): Pass function name
2627 through default_strip_name_encoding before sanity checking instead
2628 of skipping the first char of the assembler name.
2630 2014-03-13 Richard Henderson <rth@redhat.com>
2633 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
2634 (ix86_force_to_memory, ix86_free_from_memory): Remove.
2635 * config/i386/i386-protos.h: Likewise.
2636 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
2637 in the expander instead of a splitter.
2638 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
2639 any possibility of requiring a memory.
2640 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
2641 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
2642 (fp branch splitters): Update for ix86_split_fp_branch.
2643 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
2644 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
2645 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
2646 (*fop_<MODEF>_2_i387): Remove f/r alternative.
2647 (*fop_<MODEF>_3_i387): Likewise.
2648 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
2649 (splitters for the fop_* register patterns): Remove.
2650 (fscalexf4_i387): Rename from *fscalexf4_i387.
2651 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
2653 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2655 PR tree-optimization/59779
2656 * tree-dfa.c (get_ref_base_and_extent): Use double_int
2657 type for bitsize and maxsize instead of HOST_WIDE_INT.
2659 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
2661 PR rtl-optimization/57320
2662 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
2663 the CFG after thread_prologue_and_epilogue_insns.
2665 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
2667 PR rtl-optimization/57189
2668 * lra-constraints.c (process_alt_operands): Disfavor spilling
2671 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
2673 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
2675 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2677 PR tree-optimization/59025
2679 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
2680 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
2682 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
2685 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
2686 calls of avr_out_plus_1.
2688 2014-03-13 Bin Cheng <bin.cheng@arm.com>
2690 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
2691 BB's single pred and update the father loop's latch info later.
2693 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2695 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
2699 (VEC_base): Likewise.
2700 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
2701 registers, we need to swap double words in little endian mode.
2703 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
2704 to be a container mode for 128-bit integer operations added in ISA
2705 2.07. Unlike TImode and PTImode, the preferred register set is
2706 the Altivec/VMX registers for the 128-bit operations.
2708 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
2710 (rs6000_split_128bit_ok_p): Likewise.
2712 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
2713 macros for creating ISA 2.07 normal and overloaded builtin
2714 functions with 3 arguments.
2715 (BU_P8V_OVERLOAD_3): Likewise.
2716 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
2717 for use as overloaded functions.
2718 (VPERM_1TI_UNS): Likewise.
2719 (VSEL_1TI): Likewise.
2720 (VSEL_1TI_UNS): Likewise.
2721 (ST_INTERNAL_1ti): Likewise.
2722 (LD_INTERNAL_1ti): Likewise.
2723 (XXSEL_1TI): Likewise.
2724 (XXSEL_1TI_UNS): Likewise.
2725 (VPERM_1TI): Likewise.
2726 (VPERM_1TI_UNS): Likewise.
2727 (XXPERMDI_1TI): Likewise.
2728 (SET_1TI): Likewise.
2729 (LXVD2X_V1TI): Likewise.
2730 (STXVD2X_V1TI): Likewise.
2731 (VEC_INIT_V1TI): Likewise.
2732 (VEC_SET_V1TI): Likewise.
2733 (VEC_EXT_V1TI): Likewise.
2734 (EQV_V1TI): Likewise.
2735 (NAND_V1TI): Likewise.
2736 (ORC_V1TI): Likewise.
2737 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
2738 added in ISA 2.07. Add both normal 'altivec' builtins, and the
2740 (VADDUQM): Likewise.
2741 (VSUBCUQ): Likewise.
2742 (VADDEUQM): Likewise.
2743 (VADDECUQ): Likewise.
2744 (VSUBEUQM): Likewise.
2745 (VSUBECUQ): Likewise.
2747 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
2748 __int128_t and __uint128_t types.
2749 (__uint128_type): Likewise.
2750 (altivec_categorize_keyword): Add support for vector __int128_t,
2751 vector __uint128_t, vector __int128, and vector unsigned __int128
2752 as a container type for TImode operations that need to be done in
2753 VSX/Altivec registers.
2754 (rs6000_macro_to_expand): Likewise.
2755 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
2756 to support 128-bit integer instructions vaddcuq, vadduqm,
2757 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
2758 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
2760 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
2761 for V1TImode, and set up preferences to use VSX/Altivec registers.
2762 Setup VSX reload handlers.
2763 (rs6000_debug_reg_global): Likewise.
2764 (rs6000_init_hard_regno_mode_ok): Likewise.
2765 (rs6000_preferred_simd_mode): Likewise.
2766 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
2767 (easy_altivec_constant): Likewise.
2768 (output_vec_const_move): Likewise.
2769 (rs6000_expand_vector_set): Convert V1TImode set and extract to
2771 (rs6000_expand_vector_extract): Likewise.
2772 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
2774 (rs6000_const_vec): Add support for V1TImode.
2775 (rs6000_emit_le_vsx_load): Swap double words when loading or
2776 storing TImode/V1TImode.
2777 (rs6000_emit_le_vsx_store): Likewise.
2778 (rs6000_emit_le_vsx_move): Likewise.
2779 (rs6000_emit_move): Add support for V1TImode.
2780 (altivec_expand_ld_builtin): Likewise.
2781 (altivec_expand_st_builtin): Likewise.
2782 (altivec_expand_vec_init_builtin): Likewise.
2783 (altivec_expand_builtin): Likewise.
2784 (rs6000_init_builtins): Add support for V1TImode type. Add
2785 support for ISA 2.07 128-bit integer builtins. Define type names
2786 for the VSX/Altivec vector types.
2787 (altivec_init_builtins): Add support for overloaded vector
2788 functions with V1TImode type.
2789 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
2790 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
2792 (rs6000_split_128bit_ok_p): Likewise.
2793 (rs6000_handle_altivec_attribute): Create V1TImode from vector
2794 __int128_t and vector __uint128_t.
2796 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
2797 and mode attributes.
2804 (VS_scalar): Likewise.
2805 (VS_double): Likewise.
2806 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
2808 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
2809 we support the ISA 2.07 128-bit integer arithmetic instructions.
2810 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
2811 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
2812 and TImode types for use with the builtin functions.
2813 (V1TI_type_node): Likewise.
2814 (unsigned_V1TI_type_node): Likewise.
2815 (intTI_type_internal_node): Likewise.
2816 (uintTI_type_internal_node): Likewise.
2818 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
2819 128-bit builtin functions.
2820 (UNSPEC_VADDEUQM): Likewise.
2821 (UNSPEC_VADDECUQ): Likewise.
2822 (UNSPEC_VSUBCUQ): Likewise.
2823 (UNSPEC_VSUBEUQM): Likewise.
2824 (UNSPEC_VSUBECUQ): Likewise.
2825 (VM): Add V1TImode to vector mode iterators.
2827 (VI_unit): Likewise.
2828 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
2829 (altivec_vaddcuq): Likewise.
2830 (altivec_vsubuqm): Likewise.
2831 (altivec_vsubcuq): Likewise.
2832 (altivec_vaddeuqm): Likewise.
2833 (altivec_vaddecuq): Likewise.
2834 (altivec_vsubeuqm): Likewise.
2835 (altivec_vsubecuq): Likewise.
2837 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
2839 (BOOL_128): Likewise.
2840 (BOOL_REGS_OUTPUT): Likewise.
2841 (BOOL_REGS_OP1): Likewise.
2842 (BOOL_REGS_OP2): Likewise.
2843 (BOOL_REGS_UNARY): Likewise.
2844 (BOOL_REGS_AND_CR0): Likewise.
2846 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
2847 128-bit integer builtin support.
2848 (vec_vadduqm): Likewise.
2849 (vec_vaddecuq): Likewise.
2850 (vec_vaddeuqm): Likewise.
2851 (vec_vsubecuq): Likewise.
2852 (vec_vsubeuqm): Likewise.
2853 (vec_vsubcuq): Likewise.
2854 (vec_vsubuqm): Likewise.
2856 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2857 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
2858 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
2859 128-bit integer add/subtract to ISA 2.07.
2861 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
2863 * config/arc/arc.c (arc_predicate_delay_insns):
2864 Fix third argument passed to conditionalize_nonjump.
2866 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
2868 * config/aarch64/aarch64-builtins.c
2869 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
2870 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
2871 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
2872 instead of __builtin_lfloor.
2873 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
2875 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2877 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
2878 (tree_ssa_ifcombine_bb_1): New function.
2879 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
2880 is an empty forwarder block to then_bb or vice versa and then_bb
2881 and else_bb are effectively swapped.
2883 2014-03-12 Christian Bruel <christian.bruel@st.com>
2886 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
2887 REG_CFA_DEF_CFA note.
2888 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
2889 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
2891 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
2893 PR tree-optimization/60454
2894 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
2896 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2898 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
2899 Do not define target_cpu_default2 to generic.
2900 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
2901 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
2902 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
2904 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2905 Marc Glisse <marc.glisse@inria.fr>
2907 PR tree-optimization/60502
2908 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
2909 instead of build_low_bits_mask.
2911 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2914 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
2915 if there are multiple uses, but op doesn't live on E edge.
2916 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
2917 clobber stmts before __builtin_unreachable.
2919 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
2921 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
2922 hard_frame_pointer_rtx.
2923 * cse.c (cse_insn): Remove volatile check.
2924 * cselib.c (cselib_process_insn): Likewise.
2925 * dse.c (scan_insn): Likewise.
2927 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2929 * config/arc/arc.c (conditionalize_nonjump): New function,
2931 (arc_ifcvt): ... this.
2932 (arc_predicate_delay_insns): Use it.
2934 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2936 * config/arc/predicates.md (extend_operand): During/after reload,
2937 allow const_int_operand.
2938 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
2939 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
2940 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
2942 (umulsi3_highpart_i): Likewise.
2944 2014-03-11 Richard Biener <rguenther@suse.de>
2946 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
2947 Add asserts to guard possible wrong-code bugs.
2949 2014-03-11 Richard Biener <rguenther@suse.de>
2951 PR tree-optimization/60429
2952 PR tree-optimization/60485
2953 * tree-ssa-structalias.c (set_union_with_increment): Properly
2954 take into account all fields that overlap the shifted vars.
2955 (do_sd_constraint): Likewise.
2956 (do_ds_constraint): Likewise.
2957 (get_constraint_for_ptr_offset): Likewise.
2959 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
2961 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
2962 (nios2_compute_frame_layout):
2963 Add calculation of cfun->machine->fp_save_offset.
2964 (nios2_expand_prologue): Correct setting of frame pointer register
2966 (nios2_expand_epilogue): Update recovery of stack pointer from
2967 frame pointer accordingly.
2968 (nios2_initial_elimination_offset): Update calculation of offset
2969 for eliminating to HARD_FRAME_POINTER_REGNUM.
2971 2014-03-10 Jakub Jelinek <jakub@redhat.com>
2974 * ipa.c (symtab_remove_unreachable_nodes): Don't call
2975 cgraph_get_create_node on VAR_DECLs.
2977 2014-03-10 Richard Biener <rguenther@suse.de>
2980 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
2982 2014-03-08 Douglas B Rupp <rupp@gnat.com>
2984 * config/vms/vms.opt (vms_float_format): New variable.
2986 2014-03-08 Tobias Burnus <burnus@net-b.de>
2988 * doc/invoke.texi (-fcilkplus): Update implementation status.
2990 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
2991 Richard Biener <rguenther@suse.de>
2993 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
2994 consistently accross all TUs.
2995 (run_gcc): Enable -fshort-double automatically at link at link-time
2996 and disallow override.
2998 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
3001 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
3002 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
3003 if they can't be used.
3005 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3007 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
3008 for Solaris 11/x86 ld.
3009 * configure: Regenerate.
3011 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3013 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
3014 (LIB_TLS_SPEC): Save as ld_tls_libs.
3015 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
3016 (HAVE_AS_IX86_TLSLDM): New test.
3017 * configure, config.in: Regenerate.
3018 * config/i386/i386.c (legitimize_tls_address): Fall back to
3019 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
3020 cannot support TLS_MODEL_LOCAL_DYNAMIC.
3021 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
3022 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
3024 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
3026 * common.opt (fira-loop-pressure): Mark as optimization.
3028 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
3030 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
3031 an OpenMP mappable type.
3033 2014-03-06 Matthias Klose <doko@ubuntu.com>
3035 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
3036 MULTILIB_OSDIRNAMES is not defined.
3038 2014-03-06 Jakub Jelinek <jakub@redhat.com>
3039 Meador Inge <meadori@codesourcery.com>
3042 * config/arm/arm.c (arm_tls_symbol_p): Remove.
3043 (arm_legitimize_address): Call legitimize_tls_address for any
3044 arm_tls_referenced_p expression, handle constant addend. Call it
3045 before testing for !TARGET_ARM.
3046 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
3048 2014-03-06 Richard Biener <rguenther@suse.de>
3054 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
3056 * tree-streamer.c (record_common_node): Assert we don't record
3057 nodes with type double.
3058 (preload_common_node): Skip type double, complex double and double
3059 pointer since it is now frontend dependent due to fshort-double option.
3061 2014-03-06 Richard Biener <rguenther@suse.de>
3063 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
3064 or -fno-lto is specified and the linker has full plugin support.
3065 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
3066 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
3067 * lto-wrapper.c (merge_and_complain): Merge compile-time
3068 optimization levels.
3069 (run_gcc): And pass it through to the link options.
3071 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
3075 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3077 * cselib.c (remove_useless_values): Skip to avoid quadratic
3078 behavior if the condition moved from...
3079 (cselib_process_insn): ... here holds.
3081 2014-03-05 Jakub Jelinek <jakub@redhat.com>
3084 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
3085 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
3088 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
3089 (TM_H): Add x86-tune.def.
3091 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3093 * config/aarch64/aarch64.c (generic_tunings):
3094 Use cortexa57_extra_costs.
3096 2014-03-05 Jakub Jelinek <jakub@redhat.com>
3099 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
3100 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
3101 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
3104 2014-03-04 Heiher <r@hev.cc>
3106 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
3107 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
3109 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
3111 * config/i386/predicates.md (const2356_operand): Change to ...
3112 (const2367_operand): ... this.
3113 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
3115 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3116 (*avx512pf_scatterpf<mode>sf): Ditto.
3117 (avx512pf_scatterpf<mode>df): Ditto.
3118 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3119 (*avx512pf_scatterpf<mode>df): Ditto.
3120 * config/i386/i386.c (ix86_expand_builtin): Update
3121 incorrect hint operand error message.
3123 2014-03-04 Richard Biener <rguenther@suse.de>
3125 * lto-section-in.c (lto_get_section_data): Fix const cast.
3127 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
3129 * tree-streamer.c (record_common_node): Assert we don't record
3130 nodes with type double.
3131 (preload_common_node): Skip type double, complex double and double
3132 pointer since it is now frontend dependent due to fshort-double option.
3134 2014-03-04 Richard Biener <rguenther@suse.de>
3137 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
3138 (lto_input_toplevel_asms): Likewise.
3139 * lto-section-in.c (lto_get_section_data): Instead do it here
3142 2014-03-04 Richard Biener <rguenther@suse.de>
3144 PR tree-optimization/60382
3145 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
3146 dead PHIs a reduction.
3148 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
3150 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
3152 (_mm_prefetch): Move out of GCC target("sse") pragma.
3153 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
3154 GCC target("prfchw") pragma.
3155 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
3157 * config/i386/i386.c (ix86_option_override_internal): Enable
3158 -mprfchw with -mprefetchwt1.
3160 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
3162 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
3165 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
3167 * opts.h (CL_PCH_IGNORE): Define.
3168 * targhooks.c (option_affects_pch_p):
3169 Return false for options that have CL_PCH_IGNORE set.
3170 * opt-functions.awk: Process PchIgnore.
3171 * doc/options.texi: Document PchIgnore.
3173 * config/arc/arc.opt (misize): Add PchIgnore property.
3175 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3177 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
3178 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
3179 constraint on constants to permit them being loaded into
3180 GENERAL_REGS or BASE_REGS.
3182 2014-03-03 Nick Clifton <nickc@redhat.com>
3184 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
3185 anti-cacnonical alternatives.
3186 (negandhi3_real): New pattern.
3187 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
3189 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3191 * config/avr/avr-mcus.def: Remove atxmega16x1.
3192 * config/avr/avr-tables.opt: Regenerate.
3193 * config/avr/t-multilib: Regenerate.
3194 * doc/avr-mmcu.texi: Regenerate.
3196 2014-03-03 Tobias Grosser <tobias@grosser.es>
3197 Mircea Namolaru <mircea.namolaru@inria.fr>
3199 PR tree-optimization/58028
3200 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
3203 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3205 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
3206 not handled by recognizers.
3208 2014-03-03 Jakub Jelinek <jakub@redhat.com>
3211 * function.c (expand_function_end): Don't emit
3212 clobber_return_register sequence if clobber_after is a BARRIER.
3213 * cfgexpand.c (construct_exit_block): Append instructions before
3214 return_label to prev_bb.
3216 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3218 * config/rs6000/constraints.md: Document reserved use of "wc".
3220 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
3223 * ipa.c (function_and_variable_visibility): When dissolving comdat
3224 group, also set all symbols to local.
3226 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
3231 2013-12-14 Jan Hubicka <jh@suse.cz>
3233 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
3235 2014-03-02 Jon Beniston <jon@beniston.com>
3241 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
3242 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
3243 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
3244 (simple_return, *simple_return): New patterns
3245 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
3246 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
3248 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
3250 * dwarf2out.c (gen_subprogram_die): Tidy.
3252 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
3255 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
3256 (*mov_t_msb_neg_negc): ... this new insn.
3258 2014-02-28 Jason Merrill <jason@redhat.com>
3261 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
3264 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
3267 * dwarf2out.c (decltype_auto_die): New static.
3268 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
3269 (gen_type_die_with_usage): Handle 'decltype(auto)'.
3270 (is_cxx_auto): Likewise.
3272 2014-02-28 Ian Bolton <ian.bolton@arm.com>
3274 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
3275 we are not using general regs only.
3277 2014-02-28 Richard Biener <rguenther@suse.de>
3280 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
3281 previous fix and only allow to remove trivial pre-headers
3282 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
3283 (remove_forwarder_block): Properly update the latch of a loop.
3285 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3288 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
3289 (cselib_preserved_hash_table): New.
3290 (preserve_constants_and_equivs): Move preserved vals to it.
3291 (cselib_find_slot): Look it up first.
3292 (cselib_init): Initialize it.
3293 (cselib_finish): Release it.
3294 (dump_cselib_table): Dump it.
3296 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3299 * cselib.c (remove_useless_values): Skip to avoid quadratic
3300 behavior if the condition moved from...
3301 (cselib_process_insn): ... here holds.
3303 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
3306 * var-tracking.c (vt_initialize): Apply the same condition to
3307 preserve the CFA base value.
3309 2014-02-28 Joey Ye <joey.ye@arm.com>
3312 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
3313 if reload in progress or completed.
3315 2014-02-28 Tobias Burnus <burnus@net-b.de>
3318 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
3321 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
3323 * doc/tm.texi.in (Condition Code Status): Update documention for
3324 relative locations of cc0-setter and cc0-user.
3326 2014-02-27 Jeff Law <law@redhat.com>
3328 PR rtl-optimization/52714
3329 * combine.c (try_combine): When splitting an unrecognized PARALLEL
3330 into two independent simple sets, if I3 is a jump, ensure the
3331 pattern we place into I3 is a (set (pc) ...).
3333 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
3334 Jeff Law <law@redhat.com>
3336 PR rtl-optimization/49847
3337 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
3338 are in different blocks.
3339 * doc/tm.texi (Condition Code Status): Update documention for
3340 relative locations of cc0-setter and cc0-user.
3342 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
3345 * lra.c (lra_emit_add): Check SUBREG too.
3347 2014-02-27 Andreas Schwab <schwab@suse.de>
3349 * config/m68k/m68k.c (m68k_option_override): Disable
3350 -flive-range-shrinkage for classic m68k.
3351 (m68k_override_options_after_change): Likewise.
3353 2014-02-27 Marek Polacek <polacek@redhat.com>
3356 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
3357 -Wmaybe-uninitialized.
3359 2014-02-27 Alan Modra <amodra@gmail.com>
3362 * reload1.c (emit_input_reload_insns): When reload_override_in,
3363 set old to rl->in_reg when rl->in_reg is a subreg.
3365 2014-02-26 Richard Biener <rguenther@suse.de>
3368 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
3370 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
3372 * common/config/i386/predicates.md (const1256_operand): Remove.
3373 (const2356_operand): New.
3374 (const_1_to_2_operand): Remove.
3375 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3376 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3377 (*avx512pf_gatherpf<mode>sf): Ditto.
3378 (avx512pf_gatherpf<mode>df): Ditto.
3379 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3380 (*avx512pf_gatherpf<mode>df): Ditto.
3381 (avx512pf_scatterpf<mode>sf): Ditto.
3382 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3383 (*avx512pf_scatterpf<mode>sf): Ditto.
3384 (avx512pf_scatterpf<mode>df): Ditto.
3385 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3386 (*avx512pf_scatterpf<mode>df): Ditto.
3387 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
3389 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
3391 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
3392 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
3393 (_mm512_mask_testn_epi64_mask): Move to ...
3394 * config/i386/avx512cdintrin.h: Here.
3395 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
3396 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
3397 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
3398 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
3399 TARGET_AVX512F from TARGET_AVX512CD.
3401 2014-02-26 Richard Biener <rguenther@suse.de>
3404 * ipa.c (walk_polymorphic_call_targets): Properly guard
3405 call to inline_update_overall_summary.
3407 2014-02-26 Bin Cheng <bin.cheng@arm.com>
3410 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
3411 and latches only if requested. Fix latch if it is removed.
3412 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
3413 LOOPS_HAVE_PREHEADERS.
3415 2014-02-25 Andrew Pinski <apinski@cavium.com>
3417 * builtins.c (expand_builtin_thread_pointer): Create a new target
3418 when the target is NULL.
3420 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
3422 PR rtl-optimization/60317
3423 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3424 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
3425 * lra-assigns.c: Include params.h.
3426 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
3427 other reload pseudos considerations.
3429 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3431 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
3432 to use canonical form for nor<mode>3.
3434 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3437 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
3440 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
3442 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
3443 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
3444 (ix86_handle_option): Handle OPT_mprefetchwt1.
3445 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
3446 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3448 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3449 OPTION_MASK_ISA_PREFETCHWT1.
3450 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
3451 (PTA_PREFETCHWT1): New.
3452 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
3453 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
3454 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
3455 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
3456 (*prefetch_avx512pf_<mode>_: Change into ...
3457 (*prefetch_prefetchwt1_<mode>: This.
3458 * config/i386/i386.opt (mprefetchwt1): New.
3459 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
3460 (_mm_prefetch): Handle intent to write.
3461 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
3463 2014-02-25 Richard Biener <rguenther@suse.de>
3466 * emit-rtl.c (mem_attrs_htab): Remove.
3467 (mem_attrs_htab_hash): Likewise.
3468 (mem_attrs_htab_eq): Likewise.
3469 (set_mem_attrs): Always allocate new mem-attrs when something changed.
3470 (init_emit_once): Do not allocate mem_attrs_htab.
3472 2014-02-25 Richard Biener <rguenther@suse.de>
3475 * lto-opts.c (lto_write_options): Output non-explicit conservative
3476 -fwrapv, -fno-trapv and -fno-strict-overflow.
3477 * lto-wrapper.c (merge_and_complain): Handle merging those options.
3478 (run_gcc): And pass them through.
3480 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3482 * sel-sched.c (calculate_new_fences): New parameter ptime.
3483 Calculate it as a maximum over all fence cycles.
3484 (sel_sched_region_2): Adjust the call to calculate_new_fences.
3485 Print the final schedule timing when sched_verbose.
3487 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3489 PR rtl-optimization/60292
3490 * sel-sched.c (fill_vec_av_set): Do not reset target availability
3491 bit fot the fence instruction.
3493 2014-02-24 Alangi Derick <alangiderick@gmail.com>
3495 * calls.h: Fix typo in comment.
3497 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
3499 * config/pa/pa.c (pa_output_move_double): Don't valididate when
3500 adjusting offsetable addresses.
3502 2014-02-24 Guozhi Wei <carrot@google.com>
3504 * sparseset.h (sparseset_pop): Fix the wrong index.
3506 2014-02-24 Walter Lee <walt@tilera.com>
3508 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
3509 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
3511 * common/config/tilegx/tilegx-common.c
3512 (TARGET_DEFAULT_TARGET_FLAGS): Define.
3513 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
3515 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
3516 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
3517 (tilegx_gimplify_va_arg_expr): Handle big endian.
3518 (tilegx_expand_unaligned_load): Ditto.
3519 (tilegx_expand_unaligned_store): Ditto.
3520 (TARGET_RETURN_IN_MSB): New.
3521 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
3522 (TARGET_ENDIAN_DEFAULT): New.
3523 (TARGET_BIG_ENDIAN): Handle big endian.
3524 (BYTES_BIG_ENDIAN): Ditto.
3525 (WORDS_BIG_ENDIAN): Ditto.
3526 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
3529 * config/tilegx/tilegx.md (extv): Handle big endian.
3531 (insn_st<n>): Ditto.
3532 (insn_st<n>_add<bitsuffix>): Ditto.
3533 (insn_stnt<n>): Ditto.
3534 (insn_stnt<n>_add<bitsuffix>):Ditto.
3535 (vec_interleave_highv8qi): Handle big endian.
3536 (vec_interleave_highv8qi_be): New.
3537 (vec_interleave_highv8qi_le): New.
3538 (insn_v1int_h): Handle big endian.
3539 (vec_interleave_lowv8qi): Handle big endian.
3540 (vec_interleave_lowv8qi_be): New.
3541 (vec_interleave_lowv8qi_le): New.
3542 (insn_v1int_l): Handle big endian.
3543 (vec_interleave_highv4hi): Handle big endian.
3544 (vec_interleave_highv4hi_be): New.
3545 (vec_interleave_highv4hi_le): New.
3546 (insn_v2int_h): Handle big endian.
3547 (vec_interleave_lowv4hi): Handle big endian.
3548 (vec_interleave_lowv4hi_be): New.
3549 (vec_interleave_lowv4hi_le): New.
3550 (insn_v2int_l): Handle big endian.
3551 (vec_interleave_highv2si): Handle big endian.
3552 (vec_interleave_highv2si_be): New.
3553 (vec_interleave_highv2si_le): New.
3554 (insn_v4int_h): Handle big endian.
3555 (vec_interleave_lowv2si): Handle big endian.
3556 (vec_interleave_lowv2si_be): New.
3557 (vec_interleave_lowv2si_le): New.
3558 (insn_v4int_l): Handle big endian.
3559 * config/tilegx/tilegx.opt (mbig-endian): New option.
3560 (mlittle-endian): New option.
3561 * doc/install.texi: Document tilegxbe-linux.
3562 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
3564 2014-02-24 Martin Jambor <mjambor@suse.cz>
3567 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
3568 there are no parameter descriptors.
3570 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
3572 PR rtl-optimization/60268
3573 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
3574 initialization to ...
3575 (sched_rgn_init): ... here.
3576 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
3578 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3580 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
3583 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3585 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
3588 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3590 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
3591 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
3593 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3595 * config/microblaze/predicates.md: Add cmp_op predicate.
3596 * config/microblaze/microblaze.md: Add branch_compare instruction
3597 which uses cmp_op predicate and emits cmp insn before branch.
3598 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
3599 to microblaze_expand_conditional_branch and consolidate logic.
3600 (microblaze_expand_conditional_branch): emit branch_compare
3601 insn instead of handling cmp op separate from branch insn.
3603 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3605 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
3608 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3610 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
3611 define_insn with define_expand and new define_insn
3612 *altivec_lve<VI_char>x_internal.
3613 (altivec_stve<VI_char>x): Replace define_insn with define_expand
3614 and new define_insn *altivec_stve<VI_char>x_internal.
3615 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
3617 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
3619 (altivec_expand_stvex_be): New function.
3621 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
3623 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
3624 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
3625 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
3626 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
3628 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
3631 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
3632 instead of emit_move_insn.
3634 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3636 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
3638 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
3639 gen_altivec_vsumsws.
3641 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3643 * config/rs6000/altivec.md (altivec_lvxl): Rename as
3644 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
3645 (altivec_lvxl_<mode>): New define_expand incorporating
3646 -maltivec=be semantics where needed.
3647 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
3648 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
3649 semantics where needed.
3650 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
3651 (altivec_stvx_<mode>): New define_expand incorporating
3652 -maltivec=be semantics where needed.
3653 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
3654 VM2 iterator instead of V4SI.
3655 (altivec_stvxl_<mode>): New define_expand incorporating
3656 -maltivec=be semantics where needed.
3657 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
3658 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
3659 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
3660 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
3661 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
3662 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
3663 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
3664 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
3665 ALTIVEC_BUILTIN_STVXL.
3666 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
3667 (altivec_expand_stvx_be): Likewise.
3668 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
3669 (altivec_expand_lvx_be): Likewise.
3670 (altivec_expand_stvx_be): Likewise.
3671 (altivec_expand_builtin): Add cases for
3672 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
3673 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
3674 (altivec_init_builtins): Add definitions for
3675 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
3676 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
3678 2014-02-21 Catherine Moore <clm@codesourcery.com>
3680 * doc/invoke.texi (mvirt, mno-virt): Document.
3681 * config/mips/mips.opt (mvirt): New option.
3682 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
3684 2014-02-21 Richard Biener <rguenther@suse.de>
3686 PR tree-optimization/60276
3687 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
3688 (STMT_VINFO_MIN_NEG_DIST): New macro.
3689 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
3690 STMT_VINFO_MIN_NEG_DIST.
3691 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
3692 made for negative dependence distances still hold.
3694 2014-02-21 Richard Biener <rguenther@suse.de>
3697 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
3698 DECL_INITIAL for globals not in the current function context.
3700 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3702 PR tree-optimization/56490
3703 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
3704 * tree-ssa-uninit.c: Include params.h.
3705 (compute_control_dep_chain): Add num_calls argument, return false
3706 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
3707 num_calls to recursive call.
3708 (find_predicates): Change dep_chain into normal array,
3709 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
3710 variable and adjust compute_control_dep_chain caller.
3711 (find_def_preds): Likewise.
3713 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
3715 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
3716 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
3718 2014-02-21 Nick Clifton <nickc@redhat.com>
3720 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
3721 (pushhi1): Likewise.
3722 (popqi1): Add mode to pre_dec.
3725 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3727 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
3728 mode for mask of V8SFmode permutation.
3730 2014-02-20 Richard Henderson <rth@redhat.com>
3733 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
3734 a new pseudo for OLDVAL.
3736 2014-02-20 Jakub Jelinek <jakub@redhat.com>
3739 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
3740 gen_reg_rtx if d->testing_p.
3741 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
3742 if d->testing_p and we will certainly return true.
3743 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
3746 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
3748 * emit-rtl.c (gen_reg_rtx): Assert that
3749 crtl->emit.regno_pointer_align_length is non-zero.
3751 2014-02-20 Richard Henderson <rth@redhat.com>
3754 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
3755 on failure the store back into EXPECT.
3757 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
3758 Sandra Loosemore <sandra@codesourcery.com>
3760 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
3761 * config/nios2/nios2.c (nios2_function_profiler): Add
3762 -fPIC (flag_pic == 2) support.
3763 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
3764 (nios2_large_offset_p): New function.
3765 (nios2_unspec_reloc_p): Move up position, update to use
3766 nios2_large_offset_p.
3767 (nios2_unspec_address): Remove function.
3768 (nios2_unspec_offset): New function.
3769 (nios2_large_got_address): New function.
3770 (nios2_got_address): Add large offset support.
3771 (nios2_legitimize_tls_address): Update usage of removed and new
3773 (nios2_symbol_binds_local_p): New function.
3774 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
3775 (nios2_legitimize_address): Update to use nios2_large_offset_p.
3776 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
3777 (nios2_print_operand): Merge H/L processing, add hiadj/lo
3778 processing for (const (unspec ...)).
3779 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
3781 2014-02-20 Richard Biener <rguenther@suse.de>
3783 * tree-cfg.c (replace_uses_by): Mark altered BBs before
3784 doing the substitution.
3785 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
3787 2014-02-20 Martin Jambor <mjambor@suse.cz>
3790 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
3791 info when checking whether lattices are bottom.
3793 2014-02-20 Richard Biener <rguenther@suse.de>
3796 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
3799 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
3802 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
3803 parameter specifying the scaling.
3804 (inline_call): Update.
3805 (want_inline_recursively): Guard division by zero.
3806 (recursive_inlining): Update.
3807 * ipa-inline.h (clone_inlined_nodes): Update.
3809 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3812 * config/i386/i386.c (classify_argument): Pass structures of size
3813 64 bytes or less in register.
3815 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3816 Kirill Yukhin <kirill.yukhin@intel.com>
3818 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
3819 (_mm_rcp28_round_ss): Ditto.
3820 (_mm_rsqrt28_round_sd): Ditto.
3821 (_mm_rsqrt28_round_ss): Ditto.
3822 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
3823 (_mm_rcp14_round_ss): Ditto.
3824 (_mm_rsqrt14_round_sd): Ditto.
3825 (_mm_rsqrt14_round_ss): Ditto.
3826 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
3827 the first input operand, get rid of match_dup.
3828 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
3830 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3832 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
3833 operand as the first input operand, set type attribute.
3834 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3836 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
3837 operand as the first input operand, set type attribute.
3839 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3841 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
3844 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
3847 * config/i386/i386.c (construct_container): Remove TFmode check
3848 for X86_64_INTEGER_CLASS.
3850 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
3853 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
3854 only when -Wpsabi is enabled.
3856 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
3859 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
3860 passing arrays in registers are the same as for structs, so remove the
3861 special case for them.
3863 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
3865 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
3866 destination type, extract only the valid bits if the source type is not
3867 integral and has a different mode.
3869 2014-02-19 Richard Biener <rguenther@suse.de>
3872 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
3875 2014-02-19 Richard Biener <rguenther@suse.de>
3878 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
3879 (ipa_modify_call_arguments): Emit an argument load explicitely and
3880 preserve virtual SSA form there and for the replacement call.
3881 Do not update SSA form nor free dominance info.
3883 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3885 * ipa.c (function_and_variable_visibility): Also clear WEAK
3886 flag when disolving COMDAT_GROUP.
3888 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3890 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
3891 * ipa-prop.c (ipa_set_jf_known_type): Return early when
3893 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
3894 do more sanity checks.
3895 (detect_type_change): Return true when giving up early.
3896 (compute_complex_assign_jump_func): Fix type parameter of
3897 ipa_set_ancestor_jf.
3898 (compute_complex_ancestor_jump_func): Likewise.
3899 (update_jump_functions_after_inlining): Fix updating of
3901 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
3903 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3905 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
3906 inline clones when edge disappears.
3908 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
3911 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
3912 Split 64-bit moves into 2 patterns. Do not allow the use of
3913 direct move for TDmode in little endian, since the decimal value
3914 has little endian bytes within a word, but the 64-bit pieces are
3915 ordered in a big endian fashion, and normal subreg's of TDmode are
3917 (mov<mode>_64bit_dm): Likewise.
3918 (movtd_64bit_nodm): Likewise.
3920 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3922 PR tree-optimization/60174
3923 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
3924 statement of an SSA_NAME that occurs in an abnormal PHI node.
3926 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3929 * final.c (SEEN_BB): Remove.
3930 (SEEN_NOTE, SEEN_EMITTED): Renumber.
3931 (final_scan_insn): Don't force_source_line on second
3932 NOTE_INSN_BASIC_BLOCK.
3934 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
3937 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
3938 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
3939 (type_natural_mode): Warn ABI change when %zmm register is not
3940 available for AVX512F vector value passing.
3942 2014-02-18 Kai Tietz <ktietz@redhat.com>
3945 * config/i386/i386.c (ix86_expand_prologue): Use value in
3946 rax register as displacement when restoring %r10 or %rax.
3947 Fix wrong offset when restoring both registers.
3949 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3951 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
3952 assertion with conditional return.
3954 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3955 Uros Bizjak <ubizjak@gmail.com>
3958 * config/i386/driver-i386.c (host_detect_local_cpu): If
3959 YMM state is not saved by the OS, also clear has_f16c. Move
3960 CPUID 0x80000001 handling before YMM state saving checking.
3962 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
3964 PR rtl-optimization/58960
3965 * haifa-sched.c (alloc_global_sched_pressure_data): New,
3966 factored out from ...
3967 (sched_init): ... here.
3968 (free_global_sched_pressure_data): New, factored out from ...
3969 (sched_finish): ... here.
3970 * sched-int.h (free_global_sched_pressure_data): Declare.
3971 * sched-rgn.c (nr_regions_initial): New static global.
3972 (haifa_find_rgns): Initialize it.
3973 (schedule_region): Disable sched-pressure for the newly
3976 2014-02-17 Richard Biener <rguenther@suse.de>
3978 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
3979 release SSA defs of pattern stmts.
3981 2014-02-17 Richard Biener <rguenther@suse.de>
3983 * tree-inline.c (expand_call_inline): Release the virtual
3984 operand defined by the call we are about to inline.
3986 2014-02-17 Richard Biener <rguenther@suse.de>
3988 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
3990 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
3991 Ilya Tocar <ilya.tocar@intel.com>
3993 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
3994 arguments order in builtin.
3995 (_mm512_permutexvar_epi64): Ditto.
3996 (_mm512_mask_permutexvar_epi64): Ditto
3997 (_mm512_maskz_permutexvar_epi32): Ditto
3998 (_mm512_permutexvar_epi32): Ditto
3999 (_mm512_mask_permutexvar_epi32): Ditto
4001 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4003 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
4004 (p8_vmrgow): Likewise.
4006 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4008 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
4011 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4014 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
4015 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
4016 into 64-bit and 32-bit moves. On 64-bit moves, add support for
4017 using direct move instructions on ISA 2.07. Also adjust
4018 instruction length for 64-bit.
4019 (mov<mode>_64bit, TFmode/TDmode): Likewise.
4020 (mov<mode>_32bit, TFmode/TDmode): Likewise.
4022 2014-02-15 Alan Modra <amodra@gmail.com>
4026 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
4027 find_replacement on parts of insn rtl that might be reloaded.
4029 2014-02-15 Richard Biener <rguenther@suse.de>
4031 PR tree-optimization/60183
4032 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
4033 (tree_ssa_phiprop): Calculate and free post-dominators.
4035 2014-02-14 Jeff Law <law@redhat.com>
4037 PR rtl-optimization/60131
4038 * ree.c (get_extended_src_reg): New function.
4039 (combine_reaching_defs): Use it rather than assuming location of REG.
4040 (find_and_remove_re): Verify first operand of extension is
4041 a REG before adding the insns to the copy list.
4043 2014-02-14 Roland McGrath <mcgrathr@google.com>
4045 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
4046 * configure: Regenerated.
4047 * config.in: Regenerated.
4048 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
4049 instead of ASM_SHORT.
4051 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
4052 Richard Earnshaw <rearnsha@arm.com>
4054 PR rtl-optimization/59535
4055 * lra-constraints.c (process_alt_operands): Encourage alternative
4056 when unassigned pseudo class is superset of the alternative class.
4057 (inherit_reload_reg): Don't inherit when optimizing for code size.
4058 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
4059 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
4060 modes not less than 4 for Thumb1.
4062 2014-02-14 Kyle McMartin <kyle@redhat.com>
4065 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
4067 2014-02-14 Richard Biener <rguenther@suse.de>
4069 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
4070 (get_frame_arg): Drop the assert with langhook types_compatible_p.
4071 Do not strip INDIRECT_REFs.
4073 2014-02-14 Richard Biener <rguenther@suse.de>
4076 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
4077 DECL_FUNCTION_SPECIFIC_TARGET.
4078 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
4079 * tree-streamer-out.c (pack_ts_target_option): Remove.
4080 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
4081 (write_ts_function_decl_tree_pointers): Do not stream
4082 DECL_FUNCTION_SPECIFIC_TARGET.
4083 * tree-streamer-in.c (unpack_ts_target_option): Remove.
4084 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
4085 (lto_input_ts_function_decl_tree_pointers): Do not stream
4086 DECL_FUNCTION_SPECIFIC_TARGET.
4088 2014-02-14 Jakub Jelinek <jakub@redhat.com>
4090 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
4091 (get_initial_def_for_induction, vectorizable_induction): Ignore
4092 debug stmts when looking for exit_phi.
4093 (vectorizable_live_operation): Fix up condition.
4095 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
4097 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
4098 nreverse() because it changes the content of original tree list.
4100 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
4102 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
4103 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
4105 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
4107 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
4108 GNU coding standards.
4110 2014-02-13 Jakub Jelinek <jakub@redhat.com>
4113 * dwarf2out.c (gen_subprogram_die): Don't call
4114 add_calling_convention_attribute if subr_die is old_die.
4116 2014-02-13 Sharad Singhai <singhai@google.com>
4118 * doc/optinfo.texi: Fix order of nodes.
4120 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
4122 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
4123 operands[2], not operands[3].
4125 2014-02-13 Richard Biener <rguenther@suse.de>
4128 * doc/install.texi (ISL): Update recommended version to 0.12.2,
4129 mention the possibility of an in-tree build.
4130 (CLooG): Update recommended version to 0.18.1, mention the
4131 possibility of an in-tree build and clarify that the ISL
4132 bundled with CLooG does not work.
4134 2014-02-13 Jakub Jelinek <jakub@redhat.com>
4137 * expr.c (compress_float_constant): If x is a hard register,
4138 extend into a pseudo and then move to x.
4140 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
4142 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
4143 caused by bad second argument to warning_at() with -mhotpatch and
4144 nested functions (e.g. with gfortran).
4146 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
4148 * opts.c (option_name): Remove "enabled by default" rider.
4150 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
4152 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
4154 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
4155 Uros Bizjak <ubizjak@gmail.com>
4158 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
4159 * configure: Regenerated.
4161 2014-02-12 Richard Biener <rguenther@suse.de>
4163 * vec.c (vec_prefix::calculate_allocation): Move as
4164 inline variant to vec.h.
4165 (vec_prefix::calculate_allocation_1): New out-of-line version.
4166 * vec.h (vec_prefix::calculate_allocation_1): Declare.
4167 (vec_prefix::m_has_auto_buf): Rename to ...
4168 (vec_prefix::m_using_auto_storage): ... this.
4169 (vec_prefix::calculate_allocation): Inline the easy cases
4170 and dispatch to calculate_allocation_1 which doesn't need the
4172 (va_heap::reserve): Use gcc_checking_assert.
4173 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
4174 m_using_auto_storage.
4175 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
4177 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
4178 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
4179 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
4181 2014-02-12 Richard Biener <rguenther@suse.de>
4183 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
4184 when we found a dependence.
4186 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
4188 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
4190 (maybe_fold_stmt): ... into this new function.
4191 * omp-low.c (lower_omp): Update comment.
4193 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
4196 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
4199 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
4201 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
4202 identifiers in comments.
4203 (cortexa53_extra_costs): Likewise.
4204 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
4205 (cortexa7_extra_costs): Likewise.
4206 (cortexa12_extra_costs): Likewise.
4207 (cortexa15_extra_costs): Likewise.
4208 (v7m_extra_costs): Likewise.
4210 2014-02-12 Richard Biener <rguenther@suse.de>
4213 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
4214 of posix_memalign being successful.
4215 (lower_stmt): Restrict lowering of posix_memalign to when
4216 -ftree-bit-ccp is enabled.
4218 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
4220 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
4222 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
4224 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
4226 PR rtl-optimization/60116
4227 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
4228 other_insn once the combination has been validated.
4230 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
4233 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
4235 * ipa-devirt.c: Include demangle.h
4236 (odr_violation_reported): New static variable.
4237 (add_type_duplicate): Update odr_violations.
4238 (maybe_record_node): Add completep parameter; update it.
4239 (record_target_from_binfo): Add COMPLETEP parameter;
4240 update it as needed.
4241 (possible_polymorphic_call_targets_1): Likewise.
4242 (struct polymorphic_call_target_d): Add nonconstruction_targets;
4243 rename FINAL to COMPLETE.
4244 (record_targets_from_bases): Sanity check we found the binfo;
4245 fix COMPLETEP updating.
4246 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
4247 parameter, fix computing of COMPLETEP.
4248 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
4249 at LTO time do demangling.
4250 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
4251 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
4253 (gimple_get_virt_method_for_binfo): Likewise.
4254 * gimple-fold.h (gimple_get_virt_method_for_binfo,
4255 gimple_get_virt_method_for_vtable): Update prototypes.
4257 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
4260 * genautomata.c (add_presence_absence): Fix typo with
4261 {final_}presence_list.
4263 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
4266 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
4267 for VSX/Altivec vectors that land in GPR registers.
4269 2014-02-11 Richard Henderson <rth@redhat.com>
4270 Jakub Jelinek <jakub@redhat.com>
4273 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
4274 around drhs if type conversion to lacc->type is not useless.
4276 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4278 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
4280 (cortex-a57.cortex-a53): Likewise.
4281 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
4283 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4285 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
4288 2014-02-11 Renlin Li <Renlin.Li@arm.com>
4290 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
4291 add_options_for_arm_vfp3.
4293 2014-02-11 Jeff Law <law@redhat.com>
4296 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
4297 object with an undesirable mode.
4299 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4302 * config/i386/sol2-9.h: New file.
4303 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
4304 *-*-solaris2.9*): Use it.
4306 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
4308 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
4309 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
4311 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
4313 * config/microblaze/microblaze.c: Extend mcpu version format
4315 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
4317 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
4319 2014-02-10 Richard Henderson <rth@redhat.com>
4322 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
4323 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
4324 ms-abi vs -mno-accumulate-outgoing-args.
4325 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
4326 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
4329 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
4332 * cfgexpand.c (expand_asm_operands): Attach source location to
4333 ASM_INPUT rtx objects.
4334 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
4336 2014-02-10 Nick Clifton <nickc@redhat.com>
4338 * config/mn10300/mn10300.c (popcount): New function.
4339 (mn10300_expand_prologue): Include saved registers in stack usage
4342 2014-02-10 Jeff Law <law@redhat.com>
4345 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
4346 when changing the SET_DEST of a prior insn to avoid an input reload.
4348 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
4350 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
4351 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
4352 -mcall-openbsd, or -mcall-linux.
4353 (CC1_ENDIAN_BIG_SPEC): Remove.
4354 (CC1_ENDIAN_LITTLE_SPEC): Remove.
4355 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4356 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
4357 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
4358 and %cc1_endian_default.
4359 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
4361 2014-02-10 Richard Biener <rguenther@suse.de>
4363 PR tree-optimization/60115
4364 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
4365 MEM_REF handling. Properly verify that the accesses are not
4366 out of the objects bound.
4368 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4370 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
4373 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
4375 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
4376 proper constants and fix formatting.
4377 (possible_polymorphic_call_targets): Fix formatting.
4379 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
4380 Ilya Tocar <ilya.tocar@intel.com>
4382 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
4383 (_mm512_loadu_epi32): Renamed into...
4384 (_mm512_loadu_si512): This.
4385 (_mm512_storeu_epi32): Renamed into...
4386 (_mm512_storeu_si512): This.
4387 (_mm512_maskz_ceil_ps): Removed.
4388 (_mm512_maskz_ceil_pd): Ditto.
4389 (_mm512_maskz_floor_ps): Ditto.
4390 (_mm512_maskz_floor_pd): Ditto.
4391 (_mm512_floor_round_ps): Ditto.
4392 (_mm512_floor_round_pd): Ditto.
4393 (_mm512_ceil_round_ps): Ditto.
4394 (_mm512_ceil_round_pd): Ditto.
4395 (_mm512_mask_floor_round_ps): Ditto.
4396 (_mm512_mask_floor_round_pd): Ditto.
4397 (_mm512_mask_ceil_round_ps): Ditto.
4398 (_mm512_mask_ceil_round_pd): Ditto.
4399 (_mm512_maskz_floor_round_ps): Ditto.
4400 (_mm512_maskz_floor_round_pd): Ditto.
4401 (_mm512_maskz_ceil_round_ps): Ditto.
4402 (_mm512_maskz_ceil_round_pd): Ditto.
4403 (_mm512_expand_pd): Ditto.
4404 (_mm512_expand_ps): Ditto.
4405 * config/i386/i386.c (ix86_builtins): Remove
4406 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
4407 (bdesc_args): Ditto.
4408 * config/i386/predicates.md (const1256_operand): New.
4409 (const_1_to_2_operand): Ditto.
4410 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4411 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4412 (*avx512pf_gatherpf<mode>sf): Ditto.
4413 (avx512pf_gatherpf<mode>df): Ditto.
4414 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4415 (*avx512pf_gatherpf<mode>df): Ditto.
4416 (avx512pf_scatterpf<mode>sf): Ditto.
4417 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4418 (*avx512pf_scatterpf<mode>sf): Ditto.
4419 (avx512pf_scatterpf<mode>df): Ditto.
4420 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4421 (*avx512pf_scatterpf<mode>df): Ditto.
4422 (avx512f_expand<mode>): Removed.
4423 (<shift_insn><mode>3<mask_name>): Change predicate type.
4425 2014-02-08 Jakub Jelinek <jakub@redhat.com>
4427 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
4428 not at the end of datarefs vector use ordered_remove to avoid
4429 reordering datarefs vector.
4432 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
4433 mark local addressable non-static vars as GOVD_PRIVATE
4434 instead of GOVD_LOCAL.
4435 * omp-low.c (lower_omp_for): Move gimple_bind_vars
4436 and BLOCK_VARS of gimple_bind_block to new_stmt rather
4440 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
4441 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
4442 assume_aligned or alloc_align attributes.
4443 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
4444 arguments. Handle also assume_aligned and alloc_align attributes.
4445 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
4446 calls to functions with assume_aligned or alloc_align attributes.
4447 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
4449 2014-02-08 Terry Guo <terry.guo@arm.com>
4451 * doc/invoke.texi: Document ARM -march=armv7e-m.
4453 2014-02-08 Jakub Jelinek <jakub@redhat.com>
4455 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
4456 flag on __cilkrts_rethrow builtin.
4459 * ipa-cp.c (determine_versionability): Fail at -O0
4460 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
4461 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
4464 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4467 * tree-inline.c (copy_forbidden): Fail for
4468 __attribute__((optimize (0))) functions.
4470 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4472 * varpool.c: Include pointer-set.h.
4473 (varpool_remove_unreferenced_decls): Variables in other partitions
4474 will not be output; be however careful to not lose information
4477 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4479 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
4480 lookup in the vtable constructor.
4482 2014-02-07 Jeff Law <law@redhat.com>
4485 * config/m68k/m68k.md (ashldi_extsi): Turn into a
4486 define_insn_and_split.
4488 * ipa-inline.c (inline_small_functions): Fix typos.
4490 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4492 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
4493 (s390_can_use_return_insn): Declare.
4494 * config/s390/s390.h (EPILOGUE_USES): Define.
4495 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
4497 (s390_chunkify_start): Handle return JUMP_LABELs.
4498 (s390_early_mach): Emit a main_pool instruction on the entry edge.
4499 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
4500 (s390_can_use_return_insn): New functions.
4501 (s390_fix_long_loop_prediction): Handle conditional returns.
4502 (TARGET_SET_UP_BY_PROLOGUE): Define.
4503 * config/s390/s390.md (ANY_RETURN): New code iterator.
4504 (*creturn, *csimple_return, return, simple_return): New patterns.
4506 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4508 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
4509 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
4510 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
4511 REG_CFA_RESTORE list when deciding not to restore a register.
4513 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4515 * config/s390/s390.c: Include tree-pass.h and context.h.
4516 (s390_early_mach): New function, split out from...
4517 (s390_emit_prologue): ...here.
4518 (pass_data_s390_early_mach): New pass structure.
4519 (pass_s390_early_mach): New class.
4520 (s390_option_override): Create and register early_mach pass.
4521 Move to end of file.
4523 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4525 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
4526 to match for the exit block.
4528 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4530 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
4531 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
4532 Reject misaligned operands.
4534 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4536 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
4538 2014-02-07 Richard Biener <rguenther@suse.de>
4541 * gimple-low.c (lower_builtin_posix_memalign): New function.
4542 (lower_stmt): Call it to lower posix_memalign in a way
4543 to make alignment info accessible.
4545 2014-02-07 Jakub Jelinek <jakub@redhat.com>
4548 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
4549 __builtin_setjmp_receiver.
4551 2014-02-07 Richard Biener <rguenther@suse.de>
4554 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
4555 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
4556 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4557 Handle BUILT_IN_POSIX_MEMALIGN.
4558 (find_func_clobbers): Likewise.
4559 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4560 (call_may_clobber_ref_p_1): Likewise.
4562 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4565 * ipa-devirt.c (record_target_from_binfo): Remove overactive
4568 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4571 * lto-cgraph.c (lto_output_node): Use
4572 symtab_get_symbol_partitioning_class.
4573 (lto_output_varpool_node): likewise.
4574 (symtab_get_symbol_partitioning_class): Move here from
4576 * cgraph.h (symbol_partitioning_class): Likewise.
4577 (symtab_get_symbol_partitioning_class): Declare.
4579 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4581 * ggc.h (ggc_internal_cleared_alloc): New macro.
4582 * vec.h (vec_safe_copy): Handle memory stats.
4583 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
4584 * target-globals.c (save_target_globals): Likewise.
4586 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4589 * expr.c (emit_move_resolve_push): Export; be bit more selective
4590 on when to clear alias set.
4591 * expr.h (emit_move_resolve_push): Declare.
4592 * function.h (struct function): Add tail_call_marked.
4593 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
4594 * config/i386/i386-protos.h (ix86_expand_push): Remove.
4595 * config/i386/i386.md (TImode move expander): De not call
4597 (FP push expanders): Preserve memory attributes.
4598 * config/i386/sse.md (push<mode>1): Remove.
4599 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
4600 (ix86_expand_push): Remove.
4601 * config/i386/mmx.md (push<mode>1): Remove.
4603 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4605 PR rtl-optimization/60030
4606 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
4607 lopart with paradoxical subreg before shifting it up by hprec.
4609 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4611 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
4612 Remove extra newline at end of file.
4613 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
4614 (arm_issue_rate): Handle cortexa57.
4615 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
4616 (cortex-a57.cortex-a53): Likewise.
4618 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4621 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
4622 don't record in REG_FRAME_RELATED_EXPR registers not set in that
4624 (arm_expand_prologue): Adjust all callers.
4625 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
4626 info, registers also at the lowest numbered registers side. Use
4627 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
4631 * var-tracking.c (adjust_mems): Before adding a SET to
4632 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
4634 2014-02-06 Alan Modra <amodra@gmail.com>
4637 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
4638 change SDmode to DDmode when lra_in_progress.
4640 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4643 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
4644 free_data_ref on the dr first, and before goto again also set dr
4645 to the next dr. For simd_lane_access, free old datarefs[i] before
4646 overwriting it. For get_vectype_for_scalar_type failure, don't
4647 free_data_ref if simd_lane_access.
4649 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
4652 * tree.h (opts_for_fn): New inline function.
4653 (opt_for_fn): Define.
4654 * config/i386/i386.c (ix86_function_regparm): Use
4655 opt_for_fn (decl, optimize) instead of optimize.
4657 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
4659 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
4660 for SYMBOL_REF in large memory model.
4662 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4664 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
4666 (cortex-a57): Likewise.
4667 (cortex-a57.cortex-a53): Likewise.
4669 2014-02-06 Yury Gribov <y.gribov@samsung.com>
4670 Kugan Vivekanandarajah <kuganv@linaro.org>
4672 * config/arm/arm.c (arm_vector_alignment_reachable): Check
4674 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
4676 2014-02-06 Richard Biener <rguenther@suse.de>
4678 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
4679 set_loop_copy and initialize_original_copy_tables.
4681 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
4683 * config/aarch64/aarch64-simd.md
4684 (aarch64_ashr_simddi): Change QI to SI.
4686 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4687 Jakub Jelinek <jakub@redhat.com>
4690 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
4693 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4695 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
4696 CODE_FOR_altivec_vpku[hw]um to
4697 CODE_FOR_altivec_vpku[hw]um_direct.
4698 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
4699 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
4700 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
4701 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
4703 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4705 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
4706 generation for -maltivec=be.
4707 (altivec_vsumsws): Simplify redundant test.
4709 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4711 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
4712 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
4713 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
4714 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
4715 gen_altivec_vpkuwum.
4716 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
4718 (altivec_vpks<VI_char>ss): Likewise.
4719 (altivec_vpks<VI_char>us): Likewise.
4720 (altivec_vpku<VI_char>us): Likewise.
4721 (altivec_vpku<VI_char>um): Likewise.
4722 (altivec_vpku<VI_char>um_direct): New (copy of
4723 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
4725 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
4726 target is little endian and -maltivec=be is not specified.
4727 (*altivec_vupkhs<VU_char>_direct): New (copy of
4728 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
4729 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
4730 target is little endian and -maltivec=be is not specified.
4731 (*altivec_vupkls<VU_char>_direct): New (copy of
4732 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
4733 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
4734 little endian and -maltivec=be is not specified.
4735 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
4736 little endian and -maltivec=be is not specified.
4738 2014-02-05 Richard Henderson <rth@redhat.com>
4741 * combine-stack-adj.c: Revert r206943.
4742 * sched-int.h (struct deps_desc): Add last_args_size.
4743 * sched-deps.c (init_deps): Initialize it.
4744 (sched_analyze_insn): Add OUTPUT dependencies between insns that
4745 contain REG_ARGS_SIZE notes.
4747 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4749 * lto-cgraph.c (asm_nodes_output): Make global.
4750 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
4751 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
4752 (driver_handle_option): Handle OPT_fwpa.
4754 2014-02-05 Jakub Jelinek <jakub@redhat.com>
4757 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
4758 a comment typo and formatting issue. If odr_hash hasn't been
4759 created, return vNULL and set *completep to false.
4762 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
4763 bb with no successors.
4765 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
4768 * doc/invoke.texi (-march): Clarify documentation for ARM.
4772 2014-02-05 Richard Biener <rguenther@suse.de>
4774 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
4775 when not vectorizing because of too many alias checks.
4776 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
4777 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
4779 2014-02-05 Nick Clifton <nickc@redhat.com>
4781 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
4782 accept extended registers in any mode when compiling for the MN10300.
4784 2014-02-05 Yury Gribov <y.gribov@samsung.com>
4786 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
4787 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
4788 sanitization attributes.
4789 (can_inline_edge_p): Likewise.
4790 (sanitize_attrs_match_for_inline_p): New function.
4792 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4794 * ipa-prop.c (detect_type_change): Shor circuit testing of
4795 type changes on THIS pointer.
4797 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
4800 * config/pa/pa.c (legitimize_tls_address): Return original address
4801 if not passed a SYMBOL_REF rtx.
4802 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
4804 (pa_emit_move_sequence): Simplify TLS source operands.
4805 (pa_legitimate_constant_p): Reject all TLS constants.
4806 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
4807 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
4809 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4811 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
4812 groups when we know they are controlled by LTO.
4813 * varasm.c (default_binds_local_p_1): If object is in other partition,
4814 it will be resolved locally.
4816 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4818 * config/host-linux.c (linux_gt_pch_use_address): Don't
4819 use SSIZE_MAX because it is not always defined.
4821 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
4824 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
4825 threshold for pseudo splitting.
4826 (update_ebb_live_info): Process call argument hard registers and
4827 hard registers from insn definition too.
4828 (max_small_class_regs_num): New constant.
4829 (inherit_in_ebb): Update live hard regs through EBBs. Update
4830 reloads_num only for small register classes. Don't split for
4833 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
4836 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
4839 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4841 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
4844 2014-02-04 Marek Polacek <polacek@redhat.com>
4846 * gdbinit.in (pel): Define.
4848 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4850 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
4853 2014-02-04 Richard Biener <rguenther@suse.de>
4856 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
4857 in function context local.
4858 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
4859 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
4860 similar to LTO_imported_decl_ref.
4862 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4864 PR tree-optimization/60002
4865 * cgraphclones.c (build_function_decl_skip_args): Clear
4868 PR tree-optimization/60023
4869 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
4870 false to gsi_replace.
4871 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
4872 has been in some EH region and vec_stmt could throw, add
4873 vec_stmt into the same EH region.
4874 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
4875 has no lhs, ignore it.
4876 * internal-fn.c (expand_MASK_LOAD): Likewise.
4879 * tree-inline.c (copy_forbidden): Fail for
4880 __attribute__((optimize (0))) functions.
4883 * omp-low.c (simd_clone_struct_copy): If from->inbranch
4884 is set, copy one less argument.
4885 (expand_simd_clones): Don't subtract clone_info->inbranch
4886 from simd_clone_struct_alloc argument.
4888 PR rtl-optimization/57915
4889 * recog.c (simplify_while_replacing): If all unary/binary/relational
4890 operation arguments are constant, attempt to simplify those.
4893 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
4894 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
4896 2014-02-04 Richard Biener <rguenther@suse.de>
4898 PR tree-optimization/60012
4899 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
4900 TBAA disambiguation to all DDRs.
4902 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4905 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
4906 (LINK_SPEC): Use it for -shared, -shared-libgcc.
4908 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4911 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
4913 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4915 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
4916 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
4918 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4921 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
4922 to figure out targets of polymorphic calls with known decl.
4923 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4924 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
4925 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
4926 (get_polymorphic_call_info): ... here.
4927 (get_polymorphic_call_info_from_invariant): New function.
4929 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4931 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
4932 lookup via vtable pointer; check for type consistency
4933 and turn inconsitent facts into UNREACHABLE.
4934 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4935 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
4936 type inconsistent querries; return UNREACHABLE instead.
4938 2014-02-03 Richard Henderson <rth@twiddle.net>
4941 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
4942 already processed this node.
4943 (normalize_one_pred_1): Pass along mark_set.
4944 (normalize_one_pred): Create and destroy a pointer_set_t.
4945 (normalize_one_pred_chain): Likewise.
4947 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
4949 PR gcov-profile/58602
4950 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
4952 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4955 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
4956 -fno-devirtualize; try to devirtualize by the knowledge of
4957 virtual table pointer given by aggregate propagation.
4958 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4959 (ipa_print_node_jump_functions): Dump also offset that
4960 is relevant for polymorphic calls.
4961 (determine_known_aggregate_parts): Add arg_type parameter; use it
4962 instead of determining the type from pointer type.
4963 (ipa_compute_jump_functions_for_edge): Update call of
4964 determine_known_aggregate_parts.
4965 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
4966 (gimple_get_virt_method_for_binfo): ... here; simplify using
4967 vtable_pointer_value_to_vtable.
4968 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
4969 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
4970 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
4971 (vtable_pointer_value_to_vtable): Break out from ...; handle also
4973 (vtable_pointer_value_to_binfo): ... here.
4974 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
4976 2014-02-03 Teresa Johnson <tejohnson@google.com>
4978 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
4979 redef of outer loop index variable.
4981 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
4985 * doc/extend.texi (Function Attributes): Typo.
4987 2014-02-03 Cong Hou <congh@google.com>
4989 PR tree-optimization/60000
4990 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
4991 if the vectorized statement is a store. A store statement can only
4992 appear at the end of pattern statements.
4994 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4996 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
4997 (ix86_option_override_internal): Default long double to 64-bit for
4998 32-bit Bionic and to 128-bit for 64-bit Bionic.
5000 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
5001 TARGET_LONG_DOUBLE_128 is true.
5002 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
5004 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
5005 (mlong-double-64): Negate -mlong-double-128.
5006 (mlong-double-128): New option.
5008 * config/i386/i386-c.c (ix86_target_macros): Define
5009 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
5011 * doc/invoke.texi: Document -mlong-double-128.
5013 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
5015 PR rtl-optimization/60024
5016 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
5018 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
5020 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
5022 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
5024 PR rtl-optimization/57662
5025 * sel-sched.c (code_motion_path_driver): Do not mark already not
5026 existing blocks in the visiting bitmap.
5028 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
5030 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
5031 on the insn being emitted.
5033 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
5034 Will Deacon <will.deacon@arm.com>
5036 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
5038 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5040 * config/arm/arm-tables.opt: Regenerate.
5042 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5044 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
5045 for vector types other than V16QImode.
5046 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
5047 define_expand, and call altivec_expand_vec_perm_le when producing
5048 code with little endian element order.
5049 (*altivec_vperm_<mode>_internal): New insn having previous
5050 behavior of altivec_vperm_<mode>.
5051 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
5052 altivec_expand_vec_perm_le when producing code with little endian
5054 (*altivec_vperm_<mode>_uns_internal): New insn having previous
5055 behavior of altivec_vperm_<mode>_uns.
5057 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5059 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
5060 (altivec_vsumsws): Add handling for -maltivec=be with a little
5062 (altivec_vsumsws_direct): New.
5063 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
5064 gen_altivec_vsumsws.
5066 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
5068 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
5069 vtable_pointer_value_to_binfo): New functions.
5070 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
5071 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
5073 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
5075 * config/nios2/nios2.md (load_got_register): Initialize GOT
5076 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
5077 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
5079 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
5081 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
5082 preserverd by passthrough, do not propagate the type.
5084 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
5086 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
5087 (mips_atomic_assign_expand_fenv): New function.
5088 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
5090 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
5092 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
5093 (__builtin_mips_set_fcsr): Likewise.
5094 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
5095 MIPS_USI_FTYPE_VOID.
5096 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
5097 (mips16_expand_set_fcsr): Likewise.
5098 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
5099 (mips16_set_fcsr_stub): Likewise.
5100 (mips16_get_fcsr_one_only_stub): New class.
5101 (mips16_set_fcsr_one_only_stub): Likewise.
5102 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
5103 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
5104 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
5105 (hard_float): New availability predicate.
5106 (mips_builtins): Add get_fcsr and set_fcsr.
5107 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
5108 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
5109 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
5110 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
5111 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
5114 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
5116 * config/mips/mips.c (mips_one_only_stub): New class.
5117 (mips_need_mips16_rdhwr_p): Replace with...
5118 (mips16_rdhwr_stub): ...this new variable.
5119 (mips16_stub_call_address): New function.
5120 (mips16_rdhwr_one_only_stub): New class.
5121 (mips_expand_thread_pointer): Use mips16_stub_call_address.
5122 (mips_output_mips16_rdhwr): Delete.
5123 (mips_finish_stub): New function.
5124 (mips_code_end): Use it to handle rdhwr stubs.
5126 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
5129 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
5130 when calculating size of integer atomic types.
5132 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
5134 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
5136 2014-02-01 Jakub Jelinek <jakub@redhat.com>
5138 PR tree-optimization/60003
5139 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
5140 * profile.c (branch_prob): Use gimple_call_builtin_p
5141 to check for BUILT_IN_SETJMP_RECEIVER.
5142 * tree-inline.c (copy_bb): Call notice_special_calls.
5144 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
5147 * lra-constraints.c (process_alt_operands): Update reload_sum only
5150 2014-01-31 Richard Henderson <rth@redhat.com>
5153 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
5154 until after else_eh is processed.
5156 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
5158 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
5159 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
5160 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
5161 in smmintrin.h, remove them.
5162 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
5163 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
5164 * config/i386/i386.md (ROUND_SAE): Fix value.
5165 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
5166 (const48_operand): New.
5167 * config/i386/subst.md (round), (round_expand): Use
5168 const_4_or_8_to_11_operand.
5169 (round_saeonly), (round_saeonly_expand): Use const48_operand.
5171 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
5173 * config/i386/constraints.md (Yk): Swap meaning with k.
5174 * config/i386/i386.md (movhi_internal): Change Yk to k.
5175 (movqi_internal): Ditto.
5176 (*k<logic><mode>): Ditto.
5179 (kandn<mode>): Ditto.
5180 (*<code>hi_1): Ditto.
5181 (*<code>qi_1): Ditto.
5182 (kxnor<mode>): Ditto.
5183 (kortestzhi): Ditto.
5184 (kortestchi): Ditto.
5186 (*one_cmplhi2_1): Ditto.
5187 (*one_cmplqi2_1): Ditto.
5188 * config/i386/sse.md (): Change k to Yk.
5189 (avx512f_load<mode>_mask): Ditto.
5190 (avx512f_blendm<mode>): Ditto.
5191 (avx512f_store<mode>_mask): Ditto.
5192 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
5193 (avx512f_storedqu<mode>_mask): Ditto.
5194 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
5196 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
5197 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
5198 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
5199 (avx512f_maskcmp<mode>3): Ditto.
5200 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
5201 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
5202 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
5203 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
5204 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
5205 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
5206 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
5207 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
5208 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
5209 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
5210 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
5211 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
5212 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
5213 (vec_extract_lo_<mode>_maskm): Ditto.
5214 (vec_extract_hi_<mode>_maskm): Ditto.
5215 (avx512f_vternlog<mode>_mask): Ditto.
5216 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
5217 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
5218 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
5219 (avx512f_<code>v8div16qi2_mask): Ditto.
5220 (avx512f_<code>v8div16qi2_mask_store): Ditto.
5221 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
5222 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
5223 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
5224 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
5225 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5226 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5227 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5228 (*avx512pf_scatterpf<mode>df_mask): Ditto.
5229 (avx512cd_maskb_vec_dupv8di): Ditto.
5230 (avx512cd_maskw_vec_dupv16si): Ditto.
5231 (avx512f_vpermi2var<mode>3_maskz): Ditto.
5232 (avx512f_vpermi2var<mode>3_mask): Ditto.
5233 (avx512f_vpermi2var<mode>3_mask): Ditto.
5234 (avx512f_vpermt2var<mode>3_maskz): Ditto.
5235 (*avx512f_gathersi<mode>): Ditto.
5236 (*avx512f_gathersi<mode>_2): Ditto.
5237 (*avx512f_gatherdi<mode>): Ditto.
5238 (*avx512f_gatherdi<mode>_2): Ditto.
5239 (*avx512f_scattersi<mode>): Ditto.
5240 (*avx512f_scatterdi<mode>): Ditto.
5241 (avx512f_compress<mode>_mask): Ditto.
5242 (avx512f_compressstore<mode>_mask): Ditto.
5243 (avx512f_expand<mode>_mask): Ditto.
5244 * config/i386/subst.md (mask): Change k to Yk.
5245 (mask_scalar_merge): Ditto.
5248 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
5250 * doc/extend.texi (Vector Extensions): Document ?: in C++.
5252 2014-01-31 Richard Biener <rguenther@suse.de>
5255 * builtins.c (fold_builtin_memory_op): Make sure to not
5256 use a floating-point mode or a boolean or enumeral type for
5259 2014-01-30 DJ Delorie <dj@redhat.com>
5261 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
5262 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
5263 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
5264 whenever main() has an epilogue.
5266 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5268 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
5269 unused variable "field".
5270 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
5271 (vsx_mergeh_<mode>): Likewise.
5272 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
5273 (altivec_vmrghh): Likewise.
5274 (altivec_vmrghw): Likewise.
5275 (altivec_vmrglb): Likewise.
5276 (altivec_vmrglh): Likewise.
5277 (altivec_vmrglw): Likewise.
5278 (altivec_vspltb): Add missing uses.
5279 (altivec_vsplth): Likewise.
5280 (altivec_vspltw): Likewise.
5281 (altivec_vspltsf): Likewise.
5283 2014-01-30 Jakub Jelinek <jakub@redhat.com>
5286 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
5287 frame related instructions.
5289 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
5291 PR rtl-optimization/59959
5292 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
5293 any reload of register whose subreg is invalid.
5295 2014-01-30 Jakub Jelinek <jakub@redhat.com>
5297 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
5298 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
5299 Add missing return type - void.
5301 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5303 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
5304 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
5305 remove element index adjustment for endian (now handled in vsx.md
5307 (altivec_expand_vec_perm_const): Use
5308 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
5309 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
5310 (vsx_xxspltw_<mode>): Adjust element index for little endian.
5311 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
5312 define_expand and a new define_insn *altivec_vspltb_internal;
5313 adjust for -maltivec=be on a little endian target.
5314 (altivec_vspltb_direct): New.
5315 (altivec_vsplth): Divide into a define_expand and a new
5316 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
5317 little endian target.
5318 (altivec_vsplth_direct): New.
5319 (altivec_vspltw): Divide into a define_expand and a new
5320 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
5321 little endian target.
5322 (altivec_vspltw_direct): New.
5323 (altivec_vspltsf): Divide into a define_expand and a new
5324 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
5325 a little endian target.
5327 2014-01-30 Richard Biener <rguenther@suse.de>
5329 PR tree-optimization/59993
5330 * tree-ssa-forwprop.c (associate_pointerplus): Check we
5331 can propagate form the earlier stmt and avoid the transform
5332 when the intermediate result is needed.
5334 2014-01-30 Alangi Derick <alangiderick@gmail.com>
5336 * README.Portability: Fix typo.
5338 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
5340 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
5341 comparison_operator with ordered_comparison_operator.
5343 2014-01-30 Nick Clifton <nickc@redhat.com>
5345 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
5346 Rename to mn10300_store_multiple_regs.
5347 * config/mn10300/mn10300.c: Likewise.
5348 * config/mn10300/mn10300.md (store_movm): Fix typo: call
5349 store_multiple_regs.
5350 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
5351 Call mn10300_store_multiple_regs.
5353 2014-01-30 Nick Clifton <nickc@redhat.com>
5354 DJ Delorie <dj@redhat.com>
5356 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
5357 %fp 2 to keep registers after it properly word-aligned.
5358 (rl78_alloc_physical_registers_umul): Handle the case where both
5359 input operands are the same.
5361 2014-01-30 Richard Biener <rguenther@suse.de>
5363 PR tree-optimization/59903
5364 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
5367 2014-01-30 Jason Merrill <jason@redhat.com>
5370 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
5373 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
5375 2014-01-30 Richard Biener <rguenther@suse.de>
5377 PR tree-optimization/59951
5378 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
5380 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
5383 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
5384 SFmode to DFmode case.
5386 2014-01-29 DJ Delorie <dj@redhat.com>
5388 * config/msp430/msp430.opt (-minrt): New.
5389 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
5391 (ENDFILE_SPEC): Likewise.
5393 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
5395 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
5396 (estimate_function_body_sizes): Use it.
5398 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
5401 * dwarf2out.c (is_cxx_auto): New.
5402 (is_base_type): Use it.
5403 (gen_type_die_with_usage): Likewise.
5405 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5407 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
5408 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
5409 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
5410 -maltivec=be with LE targets.
5411 (vsx_mergeh_<mode>): Likewise.
5412 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
5413 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
5414 (altivec_vmrghb): Replace with define_expand and new
5415 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
5416 (altivec_vmrghb_direct): New define_insn.
5417 (altivec_vmrghh): Replace with define_expand and new
5418 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
5419 (altivec_vmrghh_direct): New define_insn.
5420 (altivec_vmrghw): Replace with define_expand and new
5421 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
5422 (altivec_vmrghw_direct): New define_insn.
5423 (*altivec_vmrghsf): Adjust for endianness.
5424 (altivec_vmrglb): Replace with define_expand and new
5425 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
5426 (altivec_vmrglb_direct): New define_insn.
5427 (altivec_vmrglh): Replace with define_expand and new
5428 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
5429 (altivec_vmrglh_direct): New define_insn.
5430 (altivec_vmrglw): Replace with define_expand and new
5431 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
5432 (altivec_vmrglw_direct): New define_insn.
5433 (*altivec_vmrglsf): Adjust for endianness.
5434 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5435 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5436 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
5437 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
5438 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5439 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5440 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
5441 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
5443 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
5445 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
5446 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
5449 2014-01-29 Richard Biener <rguenther@suse.de>
5451 PR tree-optimization/58742
5452 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
5453 associate_pointerplus_align.
5454 (associate_pointerplus_diff): New function.
5455 (associate_pointerplus): Likewise. Call associate_pointerplus_align
5456 and associate_pointerplus_diff.
5458 2014-01-29 Richard Biener <rguenther@suse.de>
5460 * lto-streamer.h (LTO_major_version): Bump to 3.
5461 (LTO_minor_version): Reset to 0.
5463 2014-01-29 Renlin Li <Renlin.Li@arm.com>
5465 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
5466 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
5467 (arm_file_start): Generate correct asm header for armv7ve.
5468 * config/arm/bpabi.h: Add multilib support for armv7ve.
5469 * config/arm/driver-arm.c: Change the architectures of cortex-a7
5470 and cortex-a15 to armv7ve.
5471 * config/arm/t-aprofile: Add multilib support for armv7ve.
5472 * doc/invoke.texi: Document -march=armv7ve.
5474 2014-01-29 Richard Biener <rguenther@suse.de>
5476 PR tree-optimization/58742
5477 * tree-ssa-forwprop.c (associate_plusminus): Return true
5478 if we changed sth, defer EH cleanup to ...
5479 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
5480 (simplify_mult): New function.
5482 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5485 PR tree-optimization/59920
5486 * tree.c (build_common_builtin_nodes): Remove
5487 __builtin_setjmp_dispatcher initialization.
5488 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
5489 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
5490 instead of gsi_after_labels + manually skipping debug stmts.
5491 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
5492 ignore bbs with IFN_ABNORMAL_DISPATCHER.
5493 * tree-inline.c (copy_edges_for_bb): Remove
5494 can_make_abnormal_goto argument, instead add abnormal_goto_dest
5495 argument. Ignore computed_goto_p stmts. Don't call
5496 make_abnormal_goto_edges. If a call might need abnormal edges
5497 for non-local gotos, see if it already has an edge to
5498 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
5499 with true argument, don't do anything then, otherwise add
5500 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
5501 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
5503 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
5504 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
5505 (lower_stmt): Don't set data->calls_builtin_setjmp.
5506 (lower_builtin_setjmp): Adjust comment.
5507 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
5508 * tree-cfg.c (found_computed_goto): Remove.
5509 (factor_computed_gotos): Remove.
5510 (make_goto_expr_edges): Return bool, true for computed gotos.
5511 Don't call make_abnormal_goto_edges.
5512 (build_gimple_cfg): Don't set found_computed_goto, don't call
5513 factor_computed_gotos.
5514 (computed_goto_p): No longer static.
5515 (make_blocks): Don't set found_computed_goto.
5516 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
5517 (make_edges): If make_goto_expr_edges returns true, push bb
5518 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
5519 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
5520 vector. Record mapping between bbs and OpenMP regions if there
5521 are any, adjust make_gimple_omp_edges caller. Call
5522 handle_abnormal_edges.
5523 (make_abnormal_goto_edges): Remove.
5524 * tree-cfg.h (make_abnormal_goto_edges): Remove.
5525 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
5526 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
5527 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
5528 * internal-fn.def (ABNORMAL_DISPATCHER): New.
5529 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
5530 filling *region also set *region_idx to (*region)->entry->index.
5533 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
5534 For REGs set ORIGINAL_REGNO.
5536 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
5538 * doc/md.texi: Mention that a target shouldn't implement
5539 vec_widen_(s|u)mul_even/odd pair if it is less efficient
5542 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5544 PR tree-optimization/59594
5545 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
5546 a copy of the datarefs vector rather than the vector itself.
5548 2014-01-28 Jason Merrill <jason@redhat.com>
5551 * dwarf2out.c (auto_die): New static.
5552 (gen_type_die_with_usage): Handle C++1y 'auto'.
5553 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
5556 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
5559 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
5560 (SPEC_X32): Likewise.
5561 (SPEC_64): Likewise.
5562 * config/i386/i386.c (ix86_option_override_internal): Turn off
5563 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
5565 (x86_file_start): Output .code16gcc for TARGET_16BIT.
5566 * config/i386/i386.h (TARGET_16BIT): New macro.
5567 (TARGET_16BIT_P): Likewise.
5568 * config/i386/i386.opt: Add m16.
5569 * doc/invoke.texi: Document -m16.
5571 2014-01-28 Jakub Jelinek <jakub@redhat.com>
5573 PR preprocessor/59935
5574 * input.c (location_get_source_line): Bail out on when line number
5575 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
5577 2014-01-28 Richard Biener <rguenther@suse.de>
5579 PR tree-optimization/58742
5580 * tree-ssa-forwprop.c (associate_plusminus): Handle
5581 pointer subtraction of the form (T)(P + A) - (T)P.
5583 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5585 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
5588 2014-01-28 Richard Biener <rguenther@suse.de>
5591 2014-01-28 Richard Biener <rguenther@suse.de>
5593 PR rtl-optimization/45364
5594 PR rtl-optimization/59890
5595 * var-tracking.c (local_get_addr_clear_given_value): Handle
5596 already cleared slot.
5597 (val_reset): Handle not allocated local_get_addr_cache.
5598 (vt_find_locations): Use post-order on the inverted CFG.
5600 2014-01-28 Richard Biener <rguenther@suse.de>
5602 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
5604 2014-01-28 Richard Biener <rguenther@suse.de>
5606 PR rtl-optimization/45364
5607 PR rtl-optimization/59890
5608 * var-tracking.c (local_get_addr_clear_given_value): Handle
5609 already cleared slot.
5610 (val_reset): Handle not allocated local_get_addr_cache.
5611 (vt_find_locations): Use post-order on the inverted CFG.
5613 2014-01-28 Alan Modra <amodra@gmail.com>
5615 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
5616 * configure.ac <recursive call for build != host>: Define
5617 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
5618 and LD_FOR_BUILD too.
5619 * configure: Regenerate.
5621 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
5623 * config/i386/i386.c (get_builtin_code_for_version): Separate
5624 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
5625 Broadwell from Haswell.
5627 2014-01-27 Steve Ellcey <sellcey@mips.com>
5629 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
5630 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
5631 * config/mips/mips.c (mips_option_override): Change setting
5633 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
5634 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
5635 Change from Mask to Var.
5637 2014-01-27 Jeff Law <law@redhat.com>
5639 * ipa-inline.c (inline_small_functions): Fix typo.
5641 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
5643 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
5644 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
5645 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
5646 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
5647 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
5648 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
5649 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
5650 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
5651 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
5652 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
5653 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
5654 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
5655 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
5656 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
5657 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
5658 (_mm512_storeu_epi64): Ditto.
5659 (_mm512_cmpge_epi32_mask): Ditto.
5660 (_mm512_cmpge_epu32_mask): Ditto.
5661 (_mm512_cmpge_epi64_mask): Ditto.
5662 (_mm512_cmpge_epu64_mask): Ditto.
5663 (_mm512_cmple_epi32_mask): Ditto.
5664 (_mm512_cmple_epu32_mask): Ditto.
5665 (_mm512_cmple_epi64_mask): Ditto.
5666 (_mm512_cmple_epu64_mask): Ditto.
5667 (_mm512_cmplt_epi32_mask): Ditto.
5668 (_mm512_cmplt_epu32_mask): Ditto.
5669 (_mm512_cmplt_epi64_mask): Ditto.
5670 (_mm512_cmplt_epu64_mask): Ditto.
5671 (_mm512_cmpneq_epi32_mask): Ditto.
5672 (_mm512_cmpneq_epu32_mask): Ditto.
5673 (_mm512_cmpneq_epi64_mask): Ditto.
5674 (_mm512_cmpneq_epu64_mask): Ditto.
5675 (_mm512_expand_pd): Ditto.
5676 (_mm512_expand_ps): Ditto.
5677 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
5678 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
5679 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
5680 * config/i386/i386.c (ix86_builtins): Add
5681 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
5682 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
5683 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
5684 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
5685 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
5686 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
5687 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
5688 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
5689 IX86_BUILTIN_PMOVUSQW512_MEM.
5690 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
5691 __builtin_ia32_pmovsqd512mem_mask,
5692 __builtin_ia32_pmovqd512mem_mask,
5693 __builtin_ia32_pmovusqw512mem_mask,
5694 __builtin_ia32_pmovsqw512mem_mask,
5695 __builtin_ia32_pmovqw512mem_mask,
5696 __builtin_ia32_pmovusdw512mem_mask,
5697 __builtin_ia32_pmovsdw512mem_mask,
5698 __builtin_ia32_pmovdw512mem_mask,
5699 __builtin_ia32_pmovqb512mem_mask,
5700 __builtin_ia32_pmovusqb512mem_mask,
5701 __builtin_ia32_pmovsqb512mem_mask,
5702 __builtin_ia32_pmovusdb512mem_mask,
5703 __builtin_ia32_pmovsdb512mem_mask,
5704 __builtin_ia32_pmovdb512mem_mask.
5705 (bdesc_args): Add __builtin_ia32_expanddf512,
5706 __builtin_ia32_expandsf512.
5707 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
5708 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
5709 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
5710 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
5711 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
5712 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
5713 (avx512f_<code>v8div16qi2_mask_store): This.
5714 (avx512f_expand<mode>): New.
5716 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
5718 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
5720 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5721 (_mm512_prefetch_i32scatter_pd): Ditto.
5722 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5723 (_mm512_prefetch_i64scatter_pd): Ditto.
5724 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5725 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
5726 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5727 (_mm512_prefetch_i32scatter_ps): Ditto.
5728 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5729 (_mm512_prefetch_i64scatter_ps): Ditto.
5730 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5731 * config/i386/i386-builtin-types.def: Define
5732 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
5733 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
5734 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
5735 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
5736 IX86_BUILTIN_SCATTERPFQPD.
5737 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
5738 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
5739 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
5740 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
5741 __builtin_ia32_scatterpfqps.
5742 (ix86_expand_builtin): Expand new built-ins.
5743 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
5744 fix memory access data type.
5745 (*avx512pf_gatherpf<mode>_mask): Ditto.
5746 (*avx512pf_gatherpf<mode>): Ditto.
5747 (avx512pf_scatterpf<mode>): Ditto.
5748 (*avx512pf_scatterpf<mode>_mask): Ditto.
5749 (*avx512pf_scatterpf<mode>): Ditto.
5750 (GATHER_SCATTER_SF_MEM_MODE): New.
5751 (avx512pf_gatherpf<mode>df): Ditto.
5752 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5753 (*avx512pf_scatterpf<mode>df): Ditto.
5755 2014-01-27 Jakub Jelinek <jakub@redhat.com>
5758 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
5759 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
5762 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5764 * common/config/arm/arm-common.c
5765 (arm_rewrite_mcpu): Handle multiple names.
5767 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5769 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5771 * gimple-builder.h (create_gimple_tmp): Delete.
5773 2014-01-27 Christian Bruel <christian.bruel@st.com>
5775 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
5778 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
5780 * config/pa/pa.md (call): Generate indirect long calls to non-local
5781 functions when outputing 32-bit code.
5782 (call_value): Likewise except for special call to buggy powf function.
5784 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
5785 portable runtime and PIC indirect calls.
5786 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
5787 and PIC call sequences. Use ldo instead of blr to set return register
5788 in PIC call sequence.
5790 2014-01-25 Walter Lee <walt@tilera.com>
5792 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
5793 avoid clobbering a live register.
5795 2014-01-25 Walter Lee <walt@tilera.com>
5797 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
5798 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
5799 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
5800 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
5802 2014-01-25 Walter Lee <walt@tilera.com>
5804 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
5805 arguments on even registers.
5806 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
5808 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
5809 (BIGGEST_ALIGNMENT): Ditto.
5810 (BIGGEST_FIELD_ALIGNMENT): Ditto.
5812 2014-01-25 Walter Lee <walt@tilera.com>
5814 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
5815 insns before bundling.
5816 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
5818 2014-01-25 Walter Lee <walt@tilera.com>
5820 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
5821 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
5822 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
5824 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5826 * config/mips/constraints.md (kl): Delete.
5827 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
5828 define expands, using...
5829 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
5830 instructions for MIPS16.
5831 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
5832 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
5834 2014-01-25 Walter Lee <walt@tilera.com>
5836 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
5840 2014-01-25 Walter Lee <walt@tilera.com>
5842 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
5843 (TARGET_EXPAND_TO_RTL_HOOK): Define.
5845 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5847 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
5850 2014-01-25 Jakub Jelinek <jakub@redhat.com>
5852 * print-rtl.c (in_call_function_usage): New var.
5853 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
5854 EXPR_LIST mode as mode and not as reg note name.
5857 * cfgloopmanip.c (copy_loop_info): If
5858 loop->warned_aggressive_loop_optimizations, make sure
5859 the flag is set in target loop too.
5861 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
5863 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
5865 * builtins.def: Likewise.
5866 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
5867 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
5868 * ira.c (ira_setup_eliminable_regset): Likewise.
5869 * omp-low.c (gate_expand_omp): Likewise.
5870 (execute_lower_omp): Likewise.
5871 (diagnose_sb_0): Likewise.
5872 (gate_diagnose_omp_blocks): Likewise.
5873 (simd_clone_clauses_extract): Likewise.
5876 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5878 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
5879 correction for little endian...
5880 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
5883 2014-01-24 Jeff Law <law@redhat.com>
5885 PR tree-optimization/59919
5886 * tree-vrp.c (find_assert_locations_1): Do not register asserts
5887 for non-returning calls.
5889 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
5891 * common/config/aarch64/aarch64-common.c
5892 (aarch64_rewrite_mcpu): Handle multiple names.
5893 * config/aarch64/aarch64.h
5894 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5896 2014-01-24 Dodji Seketeli <dodji@redhat.com>
5898 * input.c (add_file_to_cache_tab): Handle the case where fopen
5901 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
5904 * config/i386/i386.md (pushsf splitter): Get stack adjustment
5905 from push operand if code of push isn't PRE_DEC.
5907 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
5910 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
5911 -mquad-memory-atomic. Update -mquad-memory documentation to say
5912 it is only used for non-atomic loads/stores.
5914 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
5915 -mquad-memory or -mquad-memory-atomic switches.
5917 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
5918 -mquad-memory-atomic to ISA 2.07 support.
5920 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
5921 to separate support of normal quad word memory operations (ldq, stq)
5922 from the atomic quad word memory operations.
5924 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5925 support to separate non-atomic quad word operations from atomic
5926 quad word operations. Disable non-atomic quad word operations in
5927 little endian mode so that we don't have to swap words after the
5928 load and before the store.
5929 (quad_load_store_p): Add comment about atomic quad word support.
5930 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
5931 options printed with -mdebug=reg.
5933 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
5934 -mquad-memory-atomic as the test for whether we have quad word
5935 atomic instructions.
5936 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
5937 or -mp8-vector are used, allow byte/half-word atomic operations.
5939 * config/rs6000/sync.md (load_lockedti): Insure that the address
5940 is a proper indexed or indirect address for the lqarx instruction.
5941 On little endian systems, swap the hi/lo registers after the lqarx
5943 (load_lockedpti): Use indexed_or_indirect_operand predicate to
5944 insure the address is valid for the lqarx instruction.
5945 (store_conditionalti): Insure that the address is a proper indexed
5946 or indirect address for the stqcrx. instruction. On little endian
5947 systems, swap the hi/lo registers before doing the stqcrx.
5949 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
5950 insure the address is valid for the stqcrx. instruction.
5952 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
5953 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
5954 type of quad memory support is available.
5956 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
5959 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
5960 there is a danger of looping.
5962 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
5964 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5965 force flag_ira_loop_pressure if set via command line.
5967 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5969 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
5970 (ashr_simd): New builtin handling DI mode.
5971 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
5972 (aarch64_sshr_simddi): New match pattern.
5973 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
5974 (vshrd_n_s64): Likewise.
5975 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
5977 2014-01-23 Nick Clifton <nickc@redhat.com>
5979 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
5980 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
5981 favour of mcu specific scripts.
5982 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
5985 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5986 Alex Velenko <Alex.Velenko@arm.com>
5988 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
5989 (vaddv_s16): Likewise.
5990 (vaddv_s32): Likewise.
5991 (vaddv_u8): Likewise.
5992 (vaddv_u16): Likewise.
5993 (vaddv_u32): Likewise.
5994 (vaddvq_s8): Likewise.
5995 (vaddvq_s16): Likewise.
5996 (vaddvq_s32): Likewise.
5997 (vaddvq_s64): Likewise.
5998 (vaddvq_u8): Likewise.
5999 (vaddvq_u16): Likewise.
6000 (vaddvq_u32): Likewise.
6001 (vaddvq_u64): Likewise.
6002 (vaddv_f32): Likewise.
6003 (vaddvq_f32): Likewise.
6004 (vaddvq_f64): Likewise.
6005 (vmaxv_f32): Likewise.
6006 (vmaxv_s8): Likewise.
6007 (vmaxv_s16): Likewise.
6008 (vmaxv_s32): Likewise.
6009 (vmaxv_u8): Likewise.
6010 (vmaxv_u16): Likewise.
6011 (vmaxv_u32): Likewise.
6012 (vmaxvq_f32): Likewise.
6013 (vmaxvq_f64): Likewise.
6014 (vmaxvq_s8): Likewise.
6015 (vmaxvq_s16): Likewise.
6016 (vmaxvq_s32): Likewise.
6017 (vmaxvq_u8): Likewise.
6018 (vmaxvq_u16): Likewise.
6019 (vmaxvq_u32): Likewise.
6020 (vmaxnmv_f32): Likewise.
6021 (vmaxnmvq_f32): Likewise.
6022 (vmaxnmvq_f64): Likewise.
6023 (vminv_f32): Likewise.
6024 (vminv_s8): Likewise.
6025 (vminv_s16): Likewise.
6026 (vminv_s32): Likewise.
6027 (vminv_u8): Likewise.
6028 (vminv_u16): Likewise.
6029 (vminv_u32): Likewise.
6030 (vminvq_f32): Likewise.
6031 (vminvq_f64): Likewise.
6032 (vminvq_s8): Likewise.
6033 (vminvq_s16): Likewise.
6034 (vminvq_s32): Likewise.
6035 (vminvq_u8): Likewise.
6036 (vminvq_u16): Likewise.
6037 (vminvq_u32): Likewise.
6038 (vminnmv_f32): Likewise.
6039 (vminnmvq_f32): Likewise.
6040 (vminnmvq_f64): Likewise.
6042 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
6044 * config/aarch64/aarch64-simd.md
6045 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
6046 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
6047 (*aarch64_mul3_elt<mode>): Likewise.
6048 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
6049 (*aarch64_mul3_elt_to_64v2df): Likewise.
6050 (*aarch64_mla_elt<mode>): Likewise.
6051 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
6052 (*aarch64_mls_elt<mode>): Likewise.
6053 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
6054 (*aarch64_fma4_elt<mode>): Likewise.
6055 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
6056 (*aarch64_fma4_elt_to_64v2df): Likewise.
6057 (*aarch64_fnma4_elt<mode>): Likewise.
6058 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6059 (*aarch64_fnma4_elt_to_64v2df): Likewise.
6060 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
6061 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
6062 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6063 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6064 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6065 (aarch64_sqdmull_lane<mode>_internal): Likewise.
6066 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6068 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
6070 * config/aarch64/aarch64-simd.md
6071 (aarch64_be_checked_get_lane<mode>): New define_expand.
6072 * config/aarch64/aarch64-simd-builtins.def
6073 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
6074 New builtin definition.
6075 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
6076 Use new safe be builtin.
6078 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
6080 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
6082 (aarch64_be_st1<mode>): Likewise.
6083 (aarch_ld1<VALL:mode>): Define_expand modified.
6084 (aarch_st1<VALL:mode>): Likewise.
6085 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
6086 (UNSPEC_ST1): Likewise.
6088 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
6090 * config/microblaze/microblaze.md: Add trap insn and attribute
6092 2014-01-23 Dodji Seketeli <dodji@redhat.com>
6094 PR preprocessor/58580
6095 * input.h (location_get_source_line): Take an additional line_size
6097 (void diagnostics_file_cache_fini): Declare new function.
6098 * input.c (struct fcache): New type.
6099 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
6100 New static constants.
6101 (diagnostic_file_cache_init, total_lines_num)
6102 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
6103 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
6104 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
6105 (get_next_line, read_next_line, goto_next_line, read_line_num):
6106 New static function definitions.
6107 (diagnostic_file_cache_fini): New function.
6108 (location_get_source_line): Take an additional output line_len
6109 parameter. Re-write using lookup_or_add_file_to_cache_tab and
6111 * diagnostic.c (diagnostic_finish): Call
6112 diagnostic_file_cache_fini.
6113 (adjust_line): Take an additional input parameter for the length
6114 of the line, rather than calculating it with strlen.
6115 (diagnostic_show_locus): Adjust the use of
6116 location_get_source_line and adjust_line with respect to their new
6117 signature. While displaying a line now, do not stop at the first
6118 null byte. Rather, display the zero byte as a space and keep
6119 going until we reach the size of the line.
6120 * Makefile.in: Add vec.o to OBJS-libcommon
6122 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
6123 Ilya Tocar <ilya.tocar@intel.com>
6125 * config/i386/avx512fintrin.h (_mm512_kmov): New.
6126 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
6127 (__builtin_ia32_kmov16): Ditto.
6128 * config/i386/i386.md (UNSPEC_KMOV): New.
6131 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
6133 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
6134 (_mm512_storeu_si512): Ditto.
6136 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
6139 * rtl.h (get_referenced_operands): Declare.
6140 * recog.c (get_referenced_operands): New function.
6141 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
6142 operands have been referenced when recording LO_SUM references.
6144 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
6146 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
6148 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
6150 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
6151 Enable for generic and recent AMD targets.
6153 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
6155 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
6156 ARG_SIZE note when adjustment was eliminated.
6158 2014-01-22 Jeff Law <law@redhat.com>
6160 PR tree-optimization/59597
6161 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
6162 in file. Accept new argument REGISTERING and use it to modify
6163 dump output appropriately.
6164 (register_jump_thread): Corresponding changes.
6165 (mark_threaded_blocks): Reinstate code to cancel unprofitable
6166 thread paths involving joiner blocks. Add code to dump cancelled
6167 jump threading paths.
6169 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
6171 PR rtl-optimization/59477
6172 * lra-constraints.c (inherit_in_ebb): Process call for living hard
6173 regs. Update reloads_num and potential_reload_hard_regs for all insns.
6175 2014-01-22 Tom Tromey <tromey@redhat.com>
6177 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
6179 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
6181 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
6183 PR rtl-optimization/59896
6184 * lra-constraints.c (process_alt_operands): Check unused note for
6185 matched operands of insn with no output reloads.
6187 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
6189 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
6190 (mips_move_from_gpr_cost): Likewise.
6192 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
6194 PR rtl-optimization/59858
6195 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
6196 ira_class_hard_regs_num.
6197 (process_alt_operands): Increase reject for dying matched operand.
6199 2014-01-21 Jakub Jelinek <jakub@redhat.com>
6202 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
6203 smaller than size, perform several stores or loads and stores
6204 at dst + count - size to store or copy all of size bytes, rather
6205 than just last modesize bytes.
6207 2014-01-20 DJ Delorie <dj@redhat.com>
6209 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
6210 that CLOBBERs are REGs before propogating their values.
6212 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
6215 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
6216 (cgraph_inline_failed_type): New function.
6217 * cgraph.h (DEFCIFCODE): Add type.
6218 (cgraph_inline_failed_type_t): New enum.
6219 (cgraph_inline_failed_type): New prototype.
6220 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
6221 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
6222 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
6223 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
6224 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
6225 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
6226 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
6227 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
6228 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
6229 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
6230 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
6231 OPTIMIZATION_MISMATCH.
6232 * tree-inline.c (expand_call_inline): Emit errors during
6233 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
6235 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
6238 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
6239 mode attribute in insn output.
6241 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
6243 * output.h (output_constant): Delete.
6244 * varasm.c (output_constant): Make private.
6246 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
6248 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
6250 2014-01-20 Jakub Jelinek <jakub@redhat.com>
6253 * tree.h (fold_builtin_strcat): New prototype.
6254 * builtins.c (fold_builtin_strcat): No longer static. Add len
6255 argument, if non-NULL, don't call c_strlen. Optimize
6256 directly into __builtin_memcpy instead of __builtin_strcpy.
6257 (fold_builtin_2): Adjust fold_builtin_strcat caller.
6258 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
6260 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
6262 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6263 for SImode_address_operand operands, having only a REG argument.
6265 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
6267 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
6268 loader name using mbig-endian.
6269 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
6271 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
6273 * doc/invoke.texi (-march): Clarify documentation for AArch64.
6277 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
6279 * config/aarch64/aarch64-protos.h
6280 (aarch64_cannot_change_mode_class_ptr): Declare.
6281 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
6282 aarch64_cannot_change_mode_class_ptr): New.
6283 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
6284 backend hook aarch64_cannot_change_mode_class.
6286 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
6288 * common/config/aarch64/aarch64-common.c
6289 (aarch64_handle_option): Don't handle any option order logic here.
6290 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
6291 selected_cpu, warn on architecture version mismatch.
6292 (aarch64_override_options): Fix parsing order for option strings.
6294 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6295 Iain Sandoe <iain@codesourcery.com>
6298 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
6299 warning. Amend comment to reflect current functionality.
6301 2014-01-20 Richard Biener <rguenther@suse.de>
6304 * builtins.c (fold_builtin_strcat): Remove case better handled
6305 by tree-ssa-strlen.c.
6307 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
6309 * config/aarch64/aarch64.opt
6310 (mcpu, march, mtune): Make case-insensitive.
6312 2014-01-20 Jakub Jelinek <jakub@redhat.com>
6315 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
6316 if operands[1] is a REG or ZERO_EXTEND of a REG.
6318 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
6320 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
6322 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
6324 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
6325 long non-pic millicode calls.
6327 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6329 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
6331 2014-01-19 Kito Cheng <kito@0xlab.org>
6333 * builtins.c (expand_movstr): Check movstr expand done or fail.
6335 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6336 H.J. Lu <hongjiu.lu@intel.com>
6339 * config/i386/i386.md (*lea<mode>): Zero-extend return register
6340 to DImode for zero-extended addresses.
6342 2014-01-19 Jakub Jelinek <jakub@redhat.com>
6344 PR rtl-optimization/57763
6345 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
6346 on the new indirect jump_insn and increment LABEL_NUSES (label).
6348 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
6352 * config.gcc (x86_archs): New variable.
6353 (x86_64_archs): Likewise.
6354 (x86_cpus): Likewise.
6355 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
6356 --with-arch/--with-cpu= options.
6357 Support --with-arch=/--with-cpu={nehalem,westmere,
6358 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
6360 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6362 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
6363 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
6365 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
6367 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
6369 2014-01-18 Jakub Jelinek <jakub@redhat.com>
6372 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
6373 clear cpp_get_options (parse_in)->warn_unused_macros for
6374 ix86_target_macros_internal with cpp_define.
6376 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
6378 * jump.c (delete_related_insns): Keep (use (insn))s.
6379 * reorg.c (redundant_insn): Check for barriers too.
6381 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6383 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
6385 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
6387 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
6388 call to $$dyncall when TARGET_LONG_CALLS is true.
6390 2014-01-17 Jeff Law <law@redhat.com>
6392 * ree.c (combine_set_extension): Temporarily disable test for
6393 changing number of hard registers.
6395 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
6398 * ipa-inline-analysis.c (inline_free_summary):
6399 Do not free summary of aliases.
6401 2014-01-17 Jakub Jelinek <jakub@redhat.com>
6404 * gimplify.c (gimplify_expr): Use create_tmp_var
6405 instead of create_tmp_var_raw. If cond doesn't have
6406 integral type, don't add the IFN_ANNOTATE builtin at all.
6408 2014-01-17 Martin Jambor <mjambor@suse.cz>
6411 * ipa-cp.c (prev_edge_clone): New variable.
6412 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
6413 Also resize prev_edge_clone vector.
6414 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
6415 (ipcp_edge_removal_hook): New function.
6416 (ipcp_driver): Register ipcp_edge_removal_hook.
6418 2014-01-17 Andrew Pinski <apinski@cavium.com>
6419 Steve Ellcey <sellcey@mips.com>
6422 * config/mips/mips.c (mips_print_operand): Check operand mode instead
6425 2014-01-17 Jeff Law <law@redhat.com>
6428 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
6429 so that pass_ccp runs first.
6431 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6433 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
6434 (ix86_adjust_cost): Use !TARGET_XXX.
6435 (do_reorder_for_imul): Likewise.
6436 (swap_top_of_ready_list): Likewise.
6437 (ix86_sched_reorder): Likewise.
6439 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
6441 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6442 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
6443 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
6444 (intel_memset): New. Duplicate slm_memset.
6445 (intel_cost): New. Duplicate slm_cost.
6446 (m_INTEL): New macro.
6447 (processor_target_table): Add "intel".
6448 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
6449 with PROCESSOR_INTEL for "intel".
6450 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
6451 PROCESSOR_SILVERMONT.
6452 (ix86_issue_rate): Likewise.
6453 (ix86_adjust_cost): Likewise.
6454 (ia32_multipass_dfa_lookahead): Likewise.
6455 (swap_top_of_ready_list): Likewise.
6456 (ix86_sched_reorder): Likewise.
6457 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
6458 instead of TARGET_OPT_AGU.
6459 * config/i386/i386.h (TARGET_INTEL): New.
6460 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
6461 (processor_type): Add PROCESSOR_INTEL.
6462 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
6463 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
6465 2014-01-17 Marek Polacek <polacek@redhat.com>
6468 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
6471 2014-01-17 Richard Biener <rguenther@suse.de>
6473 PR tree-optimization/46590
6474 * opts.c (default_options_table): Add entries for
6475 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
6476 all enabled at -O1 but not for -Og.
6477 * common.opt (fbranch-count-reg): Remove Init(1).
6478 (fmove-loop-invariants): Likewise.
6479 (ftree-pta): Likewise.
6481 2014-01-17 Jakub Jelinek <jakub@redhat.com>
6483 * config/i386/i386.c (ix86_data_alignment): For compatibility with
6484 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
6485 decls to at least the GCC 4.8 used alignments.
6488 * tree-nested.c (convert_nonlocal_reference_stmt,
6489 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
6490 of GIMPLE_BIND stmts, adjust associated decls.
6492 2014-01-17 Richard Biener <rguenther@suse.de>
6494 PR tree-optimization/46590
6495 * vec.h (vec<>::bseach): New member function implementing
6496 binary search according to C89 bsearch.
6497 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
6498 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
6499 bitmap pointer again. Make accesses_in_loop a flat array.
6500 (mem_ref_obstack): New global.
6501 (outermost_indep_loop): Adjust for mem_ref->stored changes.
6502 (mark_ref_stored): Likewise.
6503 (ref_indep_loop_p_2): Likewise.
6504 (set_ref_stored_in_loop): New helper function.
6505 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
6506 (memref_free): Adjust.
6507 (record_mem_ref_loc): Simplify.
6508 (gather_mem_refs_stmt): Adjust.
6509 (sort_locs_in_loop_postorder_cmp): New function.
6510 (analyze_memory_references): Sort accesses_in_loop after
6511 loop postorder number.
6512 (find_ref_loc_in_loop_cmp): New function.
6513 (for_all_locs_in_loop): Find relevant cluster of locs in
6514 accesses_in_loop and iterate without recursion.
6515 (execute_sm): Avoid uninit warning.
6516 (struct ref_always_accessed): Simplify.
6517 (ref_always_accessed::operator ()): Likewise.
6518 (ref_always_accessed_p): Likewise.
6519 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
6520 loop postorder numbers here.
6521 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
6524 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
6527 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
6528 on decls for which assemble_alias has been called.
6530 2014-01-17 Nick Clifton <nickc@redhat.com>
6532 * config/msp430/msp430.opt: (mcpu): New option.
6533 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
6534 (msp430_option_override): Parse target_cpu. If the MCU name
6535 matches a generic string, clear target_mcu.
6536 (msp430_attr): Allow numeric interrupt values up to 63.
6537 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
6538 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
6540 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
6542 * config/msp430/msp430.md (popm): Use %J rather than %I.
6543 (addsi3): Use msp430_nonimmediate_operand for operand 2.
6544 (addhi_cy_i): Use immediate_operand for operand 2.
6545 * doc/invoke.texi: Document -mcpu option.
6547 2014-01-17 Richard Biener <rguenther@suse.de>
6549 PR rtl-optimization/38518
6550 * df.h (df_analyze_loop): Declare.
6551 * df-core.c: Include cfgloop.h.
6552 (df_analyze_1): Split out main part of df_analyze.
6553 (df_analyze): Adjust.
6554 (loop_inverted_post_order_compute): New function.
6555 (loop_post_order_compute): Likewise.
6556 (df_analyze_loop): New function avoiding whole-function
6558 * loop-invariant.c (find_defs): Use df_analyze_loop.
6559 (find_invariants): Adjust.
6560 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
6562 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
6564 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
6565 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
6567 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
6569 * ipa-ref.c (ipa_remove_stmt_references): Fix references
6570 traversal when removing references.
6572 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
6575 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
6577 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
6580 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
6581 pushing a reload for an autoinc when we had previously reloaded an
6582 inner part of the address.
6584 2014-01-16 Jakub Jelinek <jakub@redhat.com>
6586 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
6588 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
6589 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
6590 when not giving up or versioning for alias only because of
6592 (vect_analyze_data_ref_dependences): Set to true.
6593 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
6595 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
6596 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
6600 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
6603 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
6604 operand 0 predicate for gathers, use a new pseudo as subtarget.
6606 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6609 * lra-constraints.c (process_alt_operands): Add printing debug info.
6610 Check absence of input/output reloads for matched operands too.
6612 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6614 PR rtl-optimization/59835
6615 * ira.c (ira_init_register_move_cost): Increase cost for
6618 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
6620 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
6622 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
6625 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
6626 non-register objects. Use gen_(high/low)part more consistently.
6629 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
6632 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
6633 endian support, remove tests for WORDS_BIG_ENDIAN.
6634 (p8_mfvsrd_3_<mode>): Likewise.
6635 (reload_gpr_from_vsx<mode>): Likewise.
6636 (reload_gpr_from_vsxsf): Likewise.
6637 (p8_mfvsrd_4_disf): Likewise.
6639 2014-01-16 Richard Biener <rguenther@suse.de>
6641 PR rtl-optimization/46590
6642 * lcm.c (compute_antinout_edge): Use postorder iteration.
6643 (compute_laterin): Use inverted postorder iteration.
6645 2014-01-16 Nick Clifton <nickc@redhat.com>
6648 * varasm.c (output_constant): Return the number of bytes actually
6650 (output_constructor_array_range): Update the field size with the
6651 number of bytes emitted by output_constant.
6652 (output_constructor_regular_field): Likewise. Also do not
6653 complain if the total number of bytes emitted is now greater
6654 than the expected fieldpos.
6655 * output.h (output_constant): Update prototype and descriptive comment.
6657 2014-01-16 Marek Polacek <polacek@redhat.com>
6660 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
6661 it is error_mark_node.
6663 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
6665 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
6666 VALID_AVX256_REG_OR_OI_MODE.
6668 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
6670 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
6671 current procedure should be profiled.
6673 2014-01-15 Andrew Pinski <apinski@cavium.com>
6675 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
6676 of moving from/to the STACK_REG register class.
6678 2014-01-15 Richard Henderson <rth@redhat.com>
6681 * reginfo.c (global_regs_decl): Globalize.
6682 * rtl.h (global_regs_decl): Declare.
6683 * ira.c (do_reload): Diagnose frame_pointer_needed and it
6684 reserved via global_regs.
6686 2014-01-15 Teresa Johnson <tejohnson@google.com>
6688 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
6690 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6692 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
6693 and vmulosh rather than call gen_vec_widen_smult_*.
6694 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
6695 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
6696 (vec_widen_smult_even_v16qi): Likewise.
6697 (vec_widen_umult_even_v8hi): Likewise.
6698 (vec_widen_smult_even_v8hi): Likewise.
6699 (vec_widen_umult_odd_v16qi): Likewise.
6700 (vec_widen_smult_odd_v16qi): Likewise.
6701 (vec_widen_umult_odd_v8hi): Likewise.
6702 (vec_widen_smult_odd_v8hi): Likewise.
6703 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
6704 vmuloub rather than call gen_vec_widen_umult_*.
6705 (vec_widen_umult_lo_v16qi): Likewise.
6706 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
6707 vmulosb rather than call gen_vec_widen_smult_*.
6708 (vec_widen_smult_lo_v16qi): Likewise.
6709 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
6710 rather than call gen_vec_widen_umult_*.
6711 (vec_widen_umult_lo_v8hi): Likewise.
6712 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
6713 rather than call gen_vec_widen_smult_*.
6714 (vec_widen_smult_lo_v8hi): Likewise.
6716 2014-01-15 Jeff Law <law@redhat.com>
6718 PR tree-optimization/59747
6719 * ree.c (find_and_remove_re): Properly handle case where a second
6720 eliminated extension requires widening a copy created for elimination
6721 of a prior extension.
6722 (combine_set_extension): Ensure that the number of hard regs needed
6723 for a destination register does not change when we widen it.
6725 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
6727 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
6728 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
6729 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
6730 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
6731 (avr-*-rtems*): Likewise.
6732 (bfin*-rtems*): Likewise.
6733 (moxie-*-rtems*): Likewise.
6734 (h8300-*-rtems*): Likewise.
6735 (i[34567]86-*-rtems*): Likewise.
6736 (lm32-*-rtems*): Likewise.
6737 (m32r-*-rtems*): Likewise.
6738 (m68k-*-rtems*): Likewise.
6739 (microblaze*-*-rtems*): Likewise.
6740 (mips*-*-rtems*): Likewise.
6741 (powerpc-*-rtems*): Likewise.
6742 (sh-*-rtems*): Likewise.
6743 (sparc-*-rtems*): Likewise.
6744 (sparc64-*-rtems*): Likewise.
6745 (v850-*-rtems*): Likewise.
6746 (m32c-*-rtems*): Likewise.
6748 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
6750 PR rtl-optimization/59511
6751 * ira.c (ira_init_register_move_cost): Use memory costs for some
6752 cases of register move cost calculations.
6753 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
6754 instead of BB frequency.
6755 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
6756 * lra-assigns.c (find_hard_regno_for): Ditto.
6758 2014-01-15 Richard Biener <rguenther@suse.de>
6760 PR tree-optimization/59822
6761 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
6762 (vectorizable_load): Use it to hoist defs of uses of invariant
6763 loads out of the loop.
6765 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
6766 Kugan Vivekanandarajah <kuganv@linaro.org>
6769 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
6772 2014-01-15 Richard Biener <rguenther@suse.de>
6774 PR rtl-optimization/59802
6775 * lcm.c (compute_available): Use inverted postorder to seed
6776 the initial worklist.
6778 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6781 * config/s390/s390.c (s390_preferred_reload_class): Don't return
6782 ADDR_REGS for invalid symrefs in non-PIC code.
6784 2014-01-15 Jakub Jelinek <jakub@redhat.com>
6787 * builtins.c (determine_block_size): Initialize *probable_max_size
6788 even if len_rtx is CONST_INT.
6790 2014-01-14 Andrew Pinski <apinski@cavium.com>
6792 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
6793 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
6794 (cortexa53_tunings): Likewise.
6795 (aarch64_sched_issue_rate): New function.
6796 (TARGET_SCHED_ISSUE_RATE): Define.
6798 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6800 * ira-costs.c (find_costs_and_classes): Add missed
6801 ira_init_register_move_cost_if_necessary.
6803 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6806 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
6808 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
6811 * config/i386/i386.c (type_natural_mode): Add a bool parameter
6812 to indicate if type is used for function return value. Warn ABI
6813 change if the vector mode isn't available for function return value.
6814 (ix86_function_arg_advance): Pass false to type_natural_mode.
6815 (ix86_function_arg): Likewise.
6816 (ix86_gimplify_va_arg): Likewise.
6817 (function_arg_32): Don't warn ABI change.
6818 (ix86_function_value): Pass true to type_natural_mode.
6819 (ix86_return_in_memory): Likewise.
6820 (ix86_struct_value_rtx): Removed.
6821 (TARGET_STRUCT_VALUE_RTX): Likewise.
6823 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
6825 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
6826 converting a conditional jump into a conditional return.
6828 2014-01-14 Richard Biener <rguenther@suse.de>
6830 PR tree-optimization/58921
6831 PR tree-optimization/59006
6832 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
6833 hoisting invariant stmts.
6834 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
6835 invariant loads on the preheader edge if possible.
6837 2014-01-14 Joey Ye <joey.ye@arm.com>
6839 * doc/plugin.texi (Building GCC plugins): Update to C++.
6841 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
6843 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
6844 (_mm_rcp28_round_ss): Ditto.
6845 (_mm_rsqrt28_round_sd): Ditto.
6846 (_mm_rsqrt28_round_ss): Ditto.
6847 (_mm_rcp28_sd): Ditto.
6848 (_mm_rcp28_ss): Ditto.
6849 (_mm_rsqrt28_sd): Ditto.
6850 (_mm_rsqrt28_ss): Ditto.
6851 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
6852 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
6853 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
6854 (IX86_BUILTIN_RCP28SD): Ditto.
6855 (IX86_BUILTIN_RCP28SS): Ditto.
6856 (IX86_BUILTIN_RSQRT28SD): Ditto.
6857 (IX86_BUILTIN_RSQRT28SS): Ditto.
6858 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
6859 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
6860 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
6861 (ix86_expand_special_args_builtin): Expand new FTYPE.
6862 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
6863 (srcp14<mode>): Make insn unary.
6864 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
6865 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
6866 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
6867 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
6868 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
6869 Fix rounding: make it SAE only.
6870 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6872 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6874 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
6875 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
6876 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
6877 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
6878 (round_saeonly_mask_scalar_operand4): Ditto.
6879 (round_saeonly_mask_scalar_op3): Ditto.
6880 (round_saeonly_mask_scalar_op4): Ditto.
6882 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6884 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6885 Implement -maltivec=be for vec_insert and vec_extract.
6887 2014-01-10 DJ Delorie <dj@redhat.com>
6889 * config/msp430/msp430.md (call_internal): Don't allow memory
6890 references with SP as the base register.
6891 (call_value_internal): Likewise.
6892 * config/msp430/constraints.md (Yc): New. For memory references
6893 that don't use SP as a base register.
6895 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
6896 "an integer without a # prefix"
6897 * config/msp430/msp430.md (epilogue_helper): Use it.
6899 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6902 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
6903 AVX512F gather builtins.
6904 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
6905 on gather decls with INTEGER_TYPE masktype.
6906 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
6907 directly into the builtin rather than hoisting it before loop.
6909 PR tree-optimization/59387
6910 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
6911 (scev_const_prop): If folded_casts and type has undefined overflow,
6912 use force_gimple_operand instead of force_gimple_operand_gsi and
6913 for each added stmt if it is assign with
6914 arith_code_with_undefined_signed_overflow, call
6915 rewrite_to_defined_overflow.
6916 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
6917 gimple-fold.h instead.
6918 (arith_code_with_undefined_signed_overflow,
6919 rewrite_to_defined_overflow): Moved to ...
6920 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
6921 rewrite_to_defined_overflow): ... here. No longer static.
6922 Include gimplify-me.h.
6923 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
6924 rewrite_to_defined_overflow): New prototypes.
6926 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6928 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
6930 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
6932 * builtins.c (get_object_alignment_2): Minor tweak.
6933 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
6935 2014-01-13 Christian Bruel <christian.bruel@st.com>
6937 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
6938 optimized non constant lengths.
6940 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6943 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
6944 load as __atomic_load_N if possible.
6946 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6948 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
6950 (rs6000_expand_builtin): Adjust call.
6952 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6955 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
6956 * config/rs6000/rs6000.c: Include target-globals.h.
6957 (rs6000_set_current_function): Instead of doing target_reinit
6958 unconditionally, use save_target_globals_default_opts and
6959 restore_target_globals.
6961 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
6963 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
6964 (rs6000_expand_builtin): Handle mffs and mtfsf.
6965 (rs6000_init_builtins): Define mffs and mtfsf.
6966 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
6967 (rs6000_mffs): New pattern.
6968 (rs6000_mtfsf): New pattern.
6970 2014-01-11 Bin Cheng <bin.cheng@arm.com>
6972 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
6973 Start narrowing with START. Apply candidate-use pair
6974 and check overall cost in narrowing.
6975 (iv_ca_prune): Pass new argument.
6977 2014-01-10 Jeff Law <law@redhat.com>
6980 * ree.c (combine_reaching_defs): Ensure the defining statement
6981 occurs before the extension when optimizing extensions with
6982 different source and destination hard registers.
6984 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6987 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
6988 vtables into the type inheritance graph.
6990 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6992 PR rtl-optimization/59754
6993 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
6994 modes in the REGNO != REGNO case.
6996 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6998 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
7000 2014-01-10 Jakub Jelinek <jakub@redhat.com>
7002 PR tree-optimization/59745
7003 * tree-predcom.c (tree_predictive_commoning_loop): Call
7004 free_affine_expand_cache if giving up because components is NULL.
7006 * target-globals.c (save_target_globals): Allocate < 4KB structs using
7007 GC in payload of target_globals struct instead of allocating them on
7008 the heap and the larger structs separately using GC.
7009 * target-globals.h (struct target_globals): Make regs, hard_regs,
7010 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
7011 of GTY((skip)) and change type to void *.
7012 (reset_target_globals): Cast loads from those fields to corresponding
7015 2014-01-10 Steve Ellcey <sellcey@mips.com>
7018 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
7019 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
7020 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
7022 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
7025 * aarch64-modes.def (CC_Zmode): New flags mode.
7026 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
7027 represents an equality.
7028 (aarch64_get_condition_code): Handle CC_Zmode.
7029 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
7031 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7033 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
7034 extraction in good case.
7036 2014-01-10 Richard Biener <rguenther@suse.de>
7038 PR tree-optimization/59374
7039 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
7040 checking after SLP discovery. Mark stmts not participating
7041 in any SLP instance properly.
7043 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7045 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
7046 when handling a SET rtx.
7048 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7050 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
7051 (cortex-a57): Likewise.
7052 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
7054 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7056 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
7057 non-iwmmxt builtins.
7059 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
7063 * ipa-devirt.c record_target_from_binfo): Take as argument
7064 stack of binfos and lookup matching one for virtual inheritance.
7065 (possible_polymorphic_call_targets_1): Update.
7067 2014-01-10 Huacai Chen <chenhc@lemote.com>
7069 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
7070 kernel strings for Loongson-2E/2F/3A.
7072 2014-01-10 Jakub Jelinek <jakub@redhat.com>
7075 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
7076 is_gimple_call before calling gimple_call_internal_p.
7078 2014-01-09 Steve Ellcey <sellcey@mips.com>
7080 * Makefile.in (TREE_FLOW_H): Remove.
7081 (TREE_SSA_H): Add file names from tree-flow.h.
7082 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
7083 * tree.h: Remove tree-flow.h reference.
7084 * hash-table.h: Remove tree-flow.h reference.
7085 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
7086 reference with tree-ssa-loop.h.
7088 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7090 * doc/invoke.texi: Add -maltivec={be,le} options, and document
7091 default element-order behavior for -maltivec.
7092 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
7093 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
7094 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
7095 when targeting big endian, at least for now.
7096 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
7098 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7101 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
7102 var satisfies use_register_for_decl, just take into account type
7103 alignment, rather than decl alignment.
7105 PR tree-optimization/59622
7106 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
7107 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
7108 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
7109 Don't devirtualize for inplace at all. For targets.length () == 1,
7110 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
7112 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
7114 * config/i386/i386.md (cpu): Remove the unused btver1.
7116 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
7118 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
7120 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7123 * tree-core.h (struct target_globals): New forward declaration.
7124 (struct tree_target_option): Add globals field.
7125 * tree.h (TREE_TARGET_GLOBALS): Define.
7126 (prepare_target_option_nodes_for_pch): New prototype.
7127 * target-globals.h (struct target_globals): Define even if
7129 * tree.c (prepare_target_option_node_for_pch,
7130 prepare_target_option_nodes_for_pch): New functions.
7131 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
7132 * config/i386/i386.c: Include target-globals.h.
7133 (ix86_set_current_function): Instead of doing target_reinit
7134 unconditionally, use save_target_globals_default_opts and
7135 restore_target_globals.
7137 2014-01-09 Richard Biener <rguenther@suse.de>
7139 PR tree-optimization/59715
7140 * tree-cfg.h (split_critical_edges): Declare.
7141 * tree-cfg.c (split_critical_edges): Export.
7142 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
7144 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
7146 * cfgexpand.c (expand_stack_vars): Optionally disable
7147 asan stack protection.
7148 (expand_used_vars): Likewise.
7149 (partition_stack_vars): Likewise.
7150 * asan.c (asan_emit_stack_protection): Optionally disable
7151 after return stack usage.
7152 (instrument_derefs): Optionally disable memory access instrumentation.
7153 (instrument_builtin_call): Likewise.
7154 (instrument_strlen_call): Likewise.
7155 (asan_protect_global): Optionally disable global variables protection.
7156 * doc/invoke.texi: Added doc for new options.
7157 * params.def: Added new options.
7158 * params.h: Likewise.
7160 2014-01-09 Jakub Jelinek <jakub@redhat.com>
7162 PR rtl-optimization/59724
7163 * ifcvt.c (cond_exec_process_if_block): Don't call
7164 flow_find_head_matching_sequence with 0 longest_match.
7165 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
7166 non-active insns if !stop_after.
7167 (try_head_merge_bb): Revert 2014-01-07 changes.
7169 2014-01-08 Jeff Law <law@redhat.com>
7171 * ree.c (get_sub_rtx): New function, extracted from...
7172 (merge_def_and_ext): Here.
7173 (combine_reaching_defs): Use get_sub_rtx.
7175 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
7177 * cgraph.h (varpool_variable_node): Do not choke on null node.
7179 2014-01-08 Catherine Moore <clm@codesourcery.com>
7181 * config/mips/mips.md (simple_return): Attempt to use JRC
7183 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
7185 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
7187 PR rtl-optimization/59137
7188 * reorg.c (steal_delay_list_from_target): Call update_block for
7190 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
7192 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7194 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
7195 two duplicate entries.
7197 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
7200 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
7202 * config/mips/mips.c (mips_truncated_op_cost): New function.
7203 (mips_rtx_costs): Adjust test for BADDU.
7204 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
7206 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
7208 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
7209 (*baddu_si): ...this new pattern.
7211 2014-01-08 Jakub Jelinek <jakub@redhat.com>
7214 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
7216 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
7219 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
7221 (expand_expr, expand_normal): Adjust.
7222 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
7223 inner_reference_p. Use inner_reference_p to expand inner references.
7224 (store_expr): Adjust.
7225 * cfgexpand.c (expand_call_stmt): Adjust.
7227 2014-01-08 Rong Xu <xur@google.com>
7229 * gcov-io.c (gcov_var): Move from gcov-io.h.
7230 (gcov_position): Ditto.
7231 (gcov_is_error): Ditto.
7232 (gcov_rewrite): Ditto.
7233 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
7234 only part to libgcc/libgcov.h.
7236 2014-01-08 Marek Polacek <polacek@redhat.com>
7239 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
7241 2014-01-08 Marek Polacek <polacek@redhat.com>
7244 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
7246 2014-01-08 Jakub Jelinek <jakub@redhat.com>
7248 PR rtl-optimization/59649
7249 * stor-layout.c (get_mode_bounds): For BImode return
7250 0 and STORE_FLAG_VALUE.
7252 2014-01-08 Richard Biener <rguenther@suse.de>
7255 * gimple.h (is_gimple_builtin_call): Remove.
7256 (gimple_builtin_call_types_compatible_p): New.
7257 (gimple_call_builtin_p): New overload.
7258 * gimple.c (is_gimple_builtin_call): Remove.
7259 (validate_call): Rename to ...
7260 (gimple_builtin_call_types_compatible_p): ... this and export. Also
7262 (validate_type): New static function.
7263 (gimple_call_builtin_p): New overload and adjust.
7264 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
7265 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
7266 (gimple_fold_stmt_to_constant_1): Likewise.
7267 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
7269 2014-01-08 Richard Biener <rguenther@suse.de>
7272 * gimplify.c (gimplify_expr): Gimplify register-register type
7273 VIEW_CONVERT_EXPRs to separate stmts.
7275 2014-01-07 Jeff Law <law@redhat.com>
7278 * ree.c (combine_set_extension): Handle case where source
7279 and destination registers in an extension insn are different.
7280 (combine_reaching_defs): Allow source and destination registers
7281 in extension to be different under limited circumstances.
7282 (add_removable_extension): Remove restriction that the
7283 source and destination registers in the extension are the same.
7284 (find_and_remove_re): Emit a copy from the extension's
7285 destination to its source after the defining insn if
7286 the source and destination registers are different.
7289 * ifcvt.c (merge_if_block): If we are merging a block with more than
7290 one successor with a block with no successors, remove any BARRIER
7291 after the second block.
7293 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
7295 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
7297 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
7300 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
7301 for 14-bit register offsets when INT14_OK_STRICT is false.
7303 2014-01-07 Roland Stigge <stigge@antcom.de>
7304 Michael Meissner <meissner@linux.vnet.ibm.com>
7307 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
7308 Only check TFmode for SPE constants. Don't check TImode or TDmode.
7310 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
7312 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
7315 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
7317 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
7318 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
7319 rtx is const0_rtx or not.
7321 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
7324 * target-globals.c (save_target_globals): Remove this_fn_optab
7326 * toplev.c: Include optabs.h.
7327 (target_reinit): Temporarily restore the global options if another
7328 set of options are in force.
7330 2014-01-07 Jakub Jelinek <jakub@redhat.com>
7332 PR rtl-optimization/58668
7333 * cfgcleanup.c (flow_find_cross_jump): Don't count
7334 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
7335 to determine what is counted.
7336 (flow_find_head_matching_sequence): Use active_insn_p to determine
7338 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
7340 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
7341 determine what is counted.
7343 PR tree-optimization/59643
7344 * tree-predcom.c (split_data_refs_to_components): If one dr is
7345 read and one write, determine_offset fails and the write isn't
7346 in the bad component, just put the read into the bad component.
7348 2014-01-07 Mike Stump <mikestump@comcast.net>
7349 Jakub Jelinek <jakub@redhat.com>
7352 * tree-core.h (struct tree_optimization_option): Change optabs
7353 type from unsigned char * to void *.
7354 * optabs.c (init_tree_optimization_optabs): Adjust
7355 TREE_OPTIMIZATION_OPTABS initialization.
7357 2014-01-06 Jakub Jelinek <jakub@redhat.com>
7360 * config/i386/i386.h (struct machine_function): Add
7361 no_drap_save_restore field.
7362 * config/i386/i386.c (ix86_save_reg): Use
7363 !cfun->machine->no_drap_save_restore instead of
7364 crtl->stack_realign_needed.
7365 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
7366 this function clears frame_pointer_needed. Set
7367 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
7368 and DRAP reg is needed.
7370 2014-01-06 Marek Polacek <polacek@redhat.com>
7373 * doc/implement-c.texi: Mention that other integer types are
7374 permitted as bit-field types in strictly conforming mode.
7376 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
7378 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
7381 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
7383 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
7385 2014-01-06 Martin Jambor <mjambor@suse.cz>
7388 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
7390 * ipa-prop.c (ipa_print_node_params): Fix indentation.
7392 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
7396 * var-tracking.c (add_stores): Preserve the value of the source even if
7397 we don't record the store.
7399 2014-01-06 Terry Guo <terry.guo@arm.com>
7401 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
7403 2014-01-05 Iain Sandoe <iain@codesourcery.com>
7406 * config/darwin.c (darwin_function_section): Adjust return values to
7407 correspond to optimisation changes made in r206070.
7409 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
7411 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
7412 from prefetch_block tune setting.
7413 (nocona_cost): Correct size of prefetch block to 64.
7415 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
7417 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
7418 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
7419 used to save the static chain register in the computation of the offset
7420 from which the FP registers need to be restored.
7422 2014-01-04 Jakub Jelinek <jakub@redhat.com>
7424 PR tree-optimization/59519
7425 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
7426 ICE if get_current_def (current_new_name) is already non-NULL, as long
7427 as it is a phi result of some other phi in *new_exit_bb that has
7430 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
7431 or vmovdqu* for misaligned_operand.
7432 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
7433 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
7434 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
7435 aligned_mem for AVX512F masked aligned load and store builtins and for
7438 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
7440 PR tree-optimization/59651
7441 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
7442 Address range for negative step should be added by TYPE_SIZE_UNIT.
7444 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
7446 * config/m68k/m68k.c (handle_move_double): Handle pushes with
7447 overlapping registers also for registers other than the stack pointer.
7449 2014-01-03 Marek Polacek <polacek@redhat.com>
7452 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
7455 2014-01-03 Jakub Jelinek <jakub@redhat.com>
7458 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
7461 * config/i386/i386.md (MODE_SIZE): New mode attribute.
7462 (push splitter): Use <P:MODE_SIZE> instead of
7463 GET_MODE_SIZE (<P:MODE>mode).
7464 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
7465 (mov -1, reg peephole2): Likewise.
7466 * config/i386/sse.md (*mov<mode>_internal,
7467 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
7468 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
7469 *<code><mode>3, *andnot<mode>3<mask_name>,
7470 <mask_codefor><code><mode>3<mask_name>): Likewise.
7471 * config/i386/subst.md (mask_mode512bit_condition,
7472 sd_mask_mode512bit_condition): Likewise.
7474 2014-01-02 Xinliang David Li <davidxl@google.com>
7476 PR tree-optimization/59303
7477 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
7478 (dump_predicates): Better output format.
7479 (pred_equal_p): New function.
7480 (is_neq_relop_p): Ditto.
7481 (is_neq_zero_form_p): Ditto.
7482 (pred_expr_equal_p): Ditto.
7483 (pred_neg_p): Ditto.
7484 (simplify_pred): Ditto.
7485 (simplify_preds_2): Ditto.
7486 (simplify_preds_3): Ditto.
7487 (simplify_preds_4): Ditto.
7488 (simplify_preds): Ditto.
7490 (push_to_worklist): Ditto.
7491 (get_pred_info_from_cmp): Ditto.
7492 (is_degenerated_phi): Ditto.
7493 (normalize_one_pred_1): Ditto.
7494 (normalize_one_pred): Ditto.
7495 (normalize_one_pred_chain): Ditto.
7496 (normalize_preds): Ditto.
7497 (normalize_cond_1): Remove function.
7498 (normalize_cond): Ditto.
7499 (is_gcond_subset_of): Ditto.
7500 (is_subset_of_any): Ditto.
7501 (is_or_set_subset_of): Ditto.
7502 (is_and_set_subset_of): Ditto.
7503 (is_norm_cond_subset_of): Ditto.
7504 (pred_chain_length_cmp): Ditto.
7505 (convert_control_dep_chain_into_preds): Type change.
7506 (find_predicates): Ditto.
7507 (find_def_preds): Ditto.
7508 (destroy_predicates_vecs): Ditto.
7509 (find_matching_predicates_in_rest_chains): Ditto.
7510 (use_pred_not_overlap_with_undef_path_pred): Ditto.
7511 (is_pred_expr_subset): Ditto.
7512 (is_pred_chain_subset_of): Ditto.
7513 (is_included_in): Ditto.
7514 (is_superset_of): Ditto.
7516 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7518 Update copyright years.
7520 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7522 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
7523 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
7524 config/arc/arc.md, config/arc/arc.opt,
7525 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
7526 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
7527 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
7528 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
7529 config/linux-protos.h, config/linux.c, config/winnt-c.c,
7530 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
7531 vtable-verify.c, vtable-verify.h: Use the standard form for the
7534 2014-01-02 Tobias Burnus <burnus@net-b.de>
7536 * gcc.c (process_command): Update copyright notice dates.
7537 * gcov-dump.c: Ditto.
7539 * doc/cpp.texi: Bump @copying's copyright year.
7540 * doc/cppinternals.texi: Ditto.
7541 * doc/gcc.texi: Ditto.
7542 * doc/gccint.texi: Ditto.
7543 * doc/gcov.texi: Ditto.
7544 * doc/install.texi: Ditto.
7545 * doc/invoke.texi: Ditto.
7547 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
7549 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
7551 2014-01-01 Jakub Jelinek <jakub@redhat.com>
7553 * config/i386/sse.md (*mov<mode>_internal): Guard
7554 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
7556 PR rtl-optimization/59647
7557 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
7558 new_rtx into UNSIGNED_FLOAT rtxes.
7560 Copyright (C) 2014 Free Software Foundation, Inc.
7562 Copying and distribution of this file, with or without modification,
7563 are permitted in any medium without royalty provided the copyright
7564 notice and this notice are preserved.